Using Linux, 5th Edition
Uploaded by
MSkiles26Using Linux, 5th Edition
Uploaded by
MSkiles26Contents at a Glance
Introduction
I Installing Linux
1 Understanding Linux 13
2 Linux Installation Overview 31
3 Installing Red Hat Linux 61
4 Installing Caldera Op"enLinux 93
5 Installing Debian Linux 119
6 Adding Sound Cards and Other Multimedia
Hardware 145
7
Using
Upgrading and Installing Software 163
II System Administration
8 Understanding System Administration 183
9 Using the vi Editor 203
10 Booting and Shutting Down 233
11 Managing User Accounts 249
12
Linux
Backing Up Data 259
13 Improving System Security 275
14 Configuring the Linux Kernel 295
15 Linux on PowerPC Platforms 307
III Working with Linux
16 Understanding Linux Shells 317
17
Fifth Edition 18
Managing Multiple Processes 365
Printing 391
19 Understanding the File and Directory System 407
20 Managing File Systems 439
21 Managing NFS 463
22 Managing NIS and LDAP 475
23 Using Samba 485
IV Using X Windows
24 Installing the X Window System 503
25 Using the X Window System 529
26 Working with KDE 557
27 Working with GNOME 577
V Network Administration
Jack Tackett, Jr.
28 Understanding the TCP/IP Protocol Suite 597
Steve Burnett
29 Configuring a TCP/IP Network 619
30 IP Firewalling and Masquerading 635
Rob Napier
31 Connecting to the Internet 663
VI Using the Internet
Jeff Tranter
32 Accessing the Network with telnet, ftp, and
the r- Commands 683
33 Surfing the Internet with the World Wide Web 701
34 Using Electronic Mail 717
35 Surviving Usenet News 741
VII Setting Up Linux Internet Servers
36 Getting Started with Apache 759
37 Configuring an FTP Server 781
38 Configuring Domain Name Service (DNS) 793
39 Configuring Email 809
40 Configuring a Usenet News Service 825
VIII Appendixes
A Division of Macmillan Publishing USA
A Sources of Information 835
201 W. 103rd Street B The Linux How-To Index 843
Indianapolis, Indiana 46290
C The GNU General Public License 861
D The Open Source Definition 871
Index 875
Special Edition Using Linux, Fifth Edition
Publisher
Copyright # 2000 by Que 1 Corporation Dean Miller
Executive Editor
All rights reserved. No part of this book shall be
Jeff Koch
reproduced, stored in a retrieval system, or transmitted by
any means, electronic, mechanical, photocopying, record- Acquisitions Editor
ing, or otherwise, without written permission from the Gretchen Ganser
publisher. No patent liability is assumed with respect to the
Development Editor
use of the information contained herein. Although every
Maureen A. McDaniel
precaution has been taken in the preparation of this book,
the publisher and author assume no responsibility for errors
or omissions. Neither is any liability assumed for damages
resulting from the use of the information contained herein.
International Standard Book Number: 0-7897-2180-5
Managing Editor
Lisa Wilson
Library of Congress Catalog Card Number: 99-63851
Senior Editor
Printed in the United States of America
Susan Ross Moore
First Printing: October 1999
Copy Editor
Chuck Hutchinson
01 00 99 4 3 2 1
Indexer
Trademarks Erika Millen
All terms mentioned in this book that are known to be
Proofreaders
trademarks or service marks have been appropriately Christina Smith
capitalized. Que Corporation cannot attest to the accuracy Megan Wade
of this information. Use of a term in this book should not be
Technical Editor
regarded as affecting the validity of any trademark or service John Scroggins
mark.
Team Coordinator
This publication was produced using the Advent 3B2 Cindy Teeters
Publishing System. Software Development Specialist
Michael Hunter
Warning and Disclaimer Interior Designers
Louisa Klucznik
Every effort has been made to make this book as complete
Ruth Lewis
and as accurate as possible, but no warranty or fitness is
implied. The information provided is on an ``as is'' basis. Cover Designer
Dan Armstrong
The authors and the publisher shall have neither liability
nor responsibility to any person or entity with respect to any Copy Writer
loss or damages arising from the information contained in Eric Borgert
this book or from the use of the CDs or programs
3B2 Design
accompanying it. Daniela Raderstorf
3B2 Layout
Susan Geiselman
Daniela Raderstorf
Contents Memory Needs 34
Disk Drives and Space Requirements 34
Swap Space 35
I Installing Linux Monitor Requirements 35
CD-ROMs 37
37
1 Understanding Linux 13 Network Access
Miscellaneous Hardware 38
What Is Linux? 14
Compiling Necessary Information 39
Why Use Linux? 15
Starting the Installation Process 39
Linux Distributions 16
Understanding the Various Installation
Advantages of Using Linux 17 Methods 40
Creating the Boot and Supplemental
Open Systems Portability 18
Applications 18
Disks 44
Advantages for Computer Professionals
Partitioning Your Hard Drive 44
18
Installing the System 45
Education 19
Hackers 20
Maneuvering Through Linux 45
Entering Commands 45
Disadvantages of Using Linux 20
Recalling Command History 45
Lack of Technical Support 20
Making Selections 46
Hardware Problems 21
Completing Commands 46
Inability to Use Current Software 22
Lack of Experience 23
Managing Users 47
Logging In and Out 47
Overcoming the Disadvantages 23
Using Basic Commands 48
Disappearing Disadvantages 24
Getting Help for Commands with man
The Commercial Side of Linux 24 48
Using Directory-Manipulation
Commercial Programs from Red Hat 24
Commercial Programs from Caldera 25
Commands 48
Commercial Programs from Debian 25
Using File-Manipulation Commands 50
A Brief History of Linux 25 Dealing with DOS Files Under Linux 52
AT&T 25
Shutting Down Linux 54
BSD 26
USL 26
Troubleshooting 55
XENIX, SunOS, and AIX 27
27
3 Installing Red Hat Linux 61
Linux
Who Owns Linux? 28
Starting the Installation Process 62
2 Linux Installation Overview 31 Installing from Floppies or CD-ROM 63
Understanding Linux's Hardware Understanding the Various Installation
Requirements 32 Methods 64
The System's CPU 33 Compiling Necessary Information 65
The System's Bus 33
iv Special Edition Using Linux, Fifth Edition
Creating the Boot, Supplemental, and Formatting the Partition 107
Rescue Disks 66 Using the Linux fdisk Program 107
Adding the Necessary Partitions 109
Installing the Linux System 67 Creating the Swap Partition 111
Using the Linux fdisk Program 72
Adding the Necessary Partition 75 Installing the Linux Software System 111
Creating the Swap Partition 77
Using Disk Druid 78 Configuring Your System 112
Installing the Software Components 80
Installing LILO 114
Configuring Your System 82 Uninstalling LILO 114
Configuring Your Network 84 Going Back to the Beginning 115
Configuring the TCP/IP Network 84
Troubleshooting 115
Configuring the Clock 85
Selecting the Services to Start
on Reboot 85 5 Installing Debian Linux 119
Selecting Your Root Password 87
About Debian 120
Installing LILO 87
What You Need to Install Debian 120
Troubleshooting 88
Preparing to Install Debian 121
Going Back to the Beginning 89
Partitioning Your Hard Drive 124
Case Study: Installing Red Hat Linux on
Why Use Partitions? 125
DEC Alphas 89
Naming Partitions 125
Using Supported Alpha Hardware 90
Explaining Partitions 126
Creating the Boot and Root Disks 90
Using FDISK 126
Installing the Main Red Hat
Repartitioning 128
Distribution 91
Installing Debian 131
4 Installing Caldera OpenLinux 93 Installing from CD-ROM 131
Installing from DOS 131
What You Need to Install OpenLinux 94
Creating the Rescue and
Installation Methods 94
Driver Floppies 132
Making the Preparations 95
Installing the System Files 134
Using the Linux fdisk Program 135
Preparing the Installation Floppies 96
Adding the Necessary Partitions 138
Creating the Installation and Modules
Configuring a Swap Partition 140
Disks 97
Initializing Linux Partitions 140
Installing Linux 100 Configuring the System 140
Using a Previous Configuration 100 Configuring the Network 141
Configuring LISA 100
Installing the Base System 141
Probing for Hardware 101
Preparing the Hard Disks 103
Configuring the Base System 142
Repartitioning Your DOS Drive 104
Deleting Partitions 105
Adding Partitions 106
Contents v
6 Adding Sound Cards and Other Installing Non-Linux Software 175
Multimedia Hardware 145 Deciphering Software Package Formats
175
Sound Cards 146 Installing the Software 175
A Little History 146 Reviewing File Permissions 177
Sound Drivers 147 Solving Problems 178
Sound Card Technology 148 Removing Applications 178
Collecting Hardware Information 149
Configuration Methods 150 Case Study: Upgrading Your Kernel 178
Testing Your Sound Card 153
Audio CDs 155 II System Administration
Installation 156
Doing More with CD Audio 156 8 Understanding System Administration 183
Joysticks 157 Understanding the Importance of Proper
Loading the Kernel Joystick
Administration 184
Modules 157
Understanding Multiuser Concepts 185
Other Multimedia Devices 158
Understanding Centralized-Processing
Multimedia Applications 158 Systems 186
Elements of the Centralized-Processing
Information Resources 159
Model 187
Troubleshooting 160
Understanding Distributed-Processing
Systems 188
7 Upgrading and Installing Software 163 Elements of the Distributed-Processing
Model 189
Understanding Key Terms Used in This
Topologies 190
Chapter 164
Understanding the Client/Server Model
Understanding the Politics
192
of Upgrading 165
Performing Administration in a Networked
Installing Software 166
Environment 192
Understanding the System
Administrator's Job 167 Defining the Role of the Network
Administrator 192
Using the Red Hat Package Manager 167
Understanding Hardware and Software
Locating Packages 168
Issues 193
Installing Packages with RPM 169
Performing Common Networking
Uninstalling Packages with RPM 170
Administrative Tasks 194
Updating Packages with RPM 171
Monitoring the System 196
Querying Packages with RPM 172
Coping with Software Upgrades 198
Verifying Packages with RPM 173
Training the Administrator 198
Using the Debian Package Management
Troubleshooting the Network 200
System 174
vi Special Edition Using Linux, Fifth Edition
9 Using the vi Editor 203 11 Managing User Accounts 249
Why vi? 204 Working with Users 250
What Is vi? 205 Adding a User 250
Understanding the Editing Process 206 Using the adduser Command 251
Setting User Passwords 252
Using vi 207 Removing a User 253
Looking at vi's Two Modes 208
Creating Your First vi File 208 Working with Groups 253
Starting vi by Using an Existing File 209 Adding a Group 254
Exiting vi 210 Deleting a Group 254
Undoing a Command 212
Writing Files and Saving the Buffer 213 Managing Home Directories 254
Positioning the Cursor 215
Web-Based Administration 255
Adding Text 217
Deleting Text 220
Project: Using Userconf 255
Searching 221
Adding a User with Userconf 255
Changing and Replacing Text 222
Adding a User with Userconf 257
Copying, Cutting, and Pasting 223
Deleting a User with Userconf 257
Repeating Commands 225
vi Command Summary 226 12 Backing Up Data 259
Setting the vi Environment 228 Considering Backup Issues 260
Using set to See and Set Options 229
Considering Backup Tips 261
Setting the showmode Option 229
Setting Toggle Options 230
Planning a Backup Schedule 262
Changing Options for Every vi
Session 230 Performing Backups and Restoring
Files 263
Troubleshooting 231
Using tar 265
Using cpio 268
10 Booting and Shutting Down 233
Using taper 269
Understanding the Boot Process 234
Using dump 270
Booting Linux from a Floppy 240
Case Study: Copying Files 272
Booting from a Boot Manager 241 Using rdist 272
Using wget 273
Understanding LILO, the Linux Loader
242
Configuring LILO 242
13 Improving System Security 275
Using LILO 243
Handling Physical Security 276
Shutting Down Linux 243
Dealing with Password Security 277
Troubleshooting Startup and Shutdown
Developing Login Security 278
245
Accounts Without Passwords 279
Unused Accounts 279
Default Accounts 279
Contents vii
Guest Accounts 279 Compiling the New Kernel 301
Command Accounts 280
Group Accounts 280 Project: Building a Modularized Kernel 302
Working with Kernel Modules 303
Handling File Security 281 Restarting kerneld 304
Permissions 281
SUID and SGID Programs 282 15 Linux on PowerPC Platforms 307
Avoiding Social Engineering Threats 282 Linux for the PowerPC 308
Recording Use of the su Command 283 MkLinux 308
Developing a Secure System 284 Yellow Dog Linux 309
Security Threats 284
Controlling the Root 284 LinuxPPC 310
Controlling Modems and Crackers 285
Preventing Idle Terminals 285 SheepShaver: Use Linux and Have Your
Enforcing Security 285 Macintosh Too 312
Handling Security Breaches 286
Project: Adding Linux to a Macintosh
Performing Backups 287
Environment with Netatalk 312
PAM: The Pluggable Authentication
Modules Architecture 287
Understanding PAM Configuration III Working with Linux
Files 288
Required, Requisite, and Optional: 16 Understanding Linux Shells 317
Module Order and Necessity 288
Logging In 318
Shadow Passwords: What Good Are They?
289 Understanding Shells 319
The /etc/password and /etc/shadow Looking at Different Shells 319
Files 289 Configuring Your Login Environment
Adding, Changing, and Deleting Users 322
with Shadowed Passwords 290 Understanding Processes 331
Project: Establishing Security Procedures Understanding Shell Command
292 Parsing 332
Using Commands, Flags, and
14 Configuring the Linux Kernel 295 Parameters 332
Performing Filename Matching 334
What Is the Kernel? 296 Connecting Processes with Pipes 337
Redirecting Input and Output 337
Preparing to Build a New Kernel 296 Substituting Shell Variables 339
Substituting Command Results 339
Configuring a New Kernel 297 Regular Expressions 340
The Interactive Text-Based Program
Understanding Command Groups,
298 Subshells, and Other Commands 342
Using the Menu-Based Program 300
Using the X Windows System-Based Doing Background Processing 343
Program 300 Arranging for Processes to Run in the
Background 343
viii Special Edition Using Linux, Fifth Edition
Using the nohup Command 344 Controlling Multiple Processes 383
Using the cron Daemon 344 Using nohup with Background
Processes 383
Understanding Command Feedback 346 Scheduling the Priority of Commands
with nice 384
Editing and Aliasing Shell Commands 346
Scheduling the Priority of Running
Editing Commands 347
Processes with renice 385
Viewing Command History 347
Terminating Processes with kill 386
Aliasing Commands 347
Using kill to Send Signals to
Completing Commands 348
Processes 389
Adding Text with Cut and Paste 348
Troubleshooting 390
Working with Shell Scripts 348
350
Writing Programs with the Shell
Programming with Control Structures
18 Printing 391
355
Selecting a Printer to Work with Linux 392
Customizing Linux Shells 361
Knowing What You Need to Configure
Exporting Variables to the New Shell
Printers 392
362
Defining Command Aliases 363 Knowing How Printing Works Under
Linux 393
Troubleshooting 364
Understanding the Important Programs for
17 Managing Multiple Processes 365 Printing 394
The lpd Daemon 394
Understanding Multitasking 366 The lpr Command 395
The lpq Command 395
Initiating Multiple Processes 368
The lprm Command 395
Starting Multiple Processes 368
The lpc Command 395
Starting a Background Process 368
Using Pipes to Start Multiple Processes Understanding the Important
369 Directories 396
Using the Scheduling Commands 370 Understanding the Important Files 397
Running Commands at Specified Times
with at 370 Understanding the /etc/printcap File 397
Running Long Tasks with batch 372 Understanding the Fields in /etc/
Scheduling Commands with cron and printcap 398
crontab 373 Setting thePRINTER Environment
Variable 400
Reporting On and Monitoring the
Multitasking Environment 376 Creating a Test printcap Entry 401
Finding Out Who's on the System with
who 376 Putting It All Together 401
Reporting On the Status of Processes
Configuring Red Hat Printers 402
with ps 379
Troubleshooting 405
Contents ix
19 Understanding the File and Directory Mounting File Systems at
System 407 Boot Time 446
Unmounting File Systems 448
Understanding Filenames and
Pathnames 408 Maintaining File Systems 449
File Types 410
Ordinary Files 410
Using the fsck Command 449
Directory Files 411
Creating and Formatting File Systems 451
Directories and Physical Disks 411
Using fdisk to Create Disk
Links 412
Partitions 451
Special Files 413
Using mkfs to Build a File System 458
File Permissions 414
Project: Using Swap Files and
Linux Standard Directories418
Partitions 459
Classic UNIX Directories 418
Creating a Swap Partition 459
Linux Directories 420
Creating a Swap File 460
Managing Files and Directories 420
21 Managing NFS 463
Listing Files 421
Understanding the Network File
Organizing Files 424 System 464
Copying Files 425 Installing NFS 464
Moving and Renaming Files 426 Exporting an NFS File System 468
Removing Files or Directories 426 Understanding the /etc/exports File 469
Viewing the Contents of a File 428 Mounting NFS File Systems 471
Using cat to View a File 429 Mounting NFS File Systems via
Using more to View a File 429 /etc/fstab 471
Using less to View a File 429 Mounting NFS File Systems
Searching Through a File and Escaping Interactively 472
to the Shell 430
Viewing Files in Other Forms 430 Troubleshooting 472
Searching for Files 432 22 Managing NIS and LDAP 475
Changing File Time and Date Stamps 434 What Is NIS? 476
NIS Members 476
Compressing Files 435 NIS Files, Maps, and Domains 476
NIS Security Concerns 481
Case Study: Undeleting Files 436
LDAP: What Is It, and Why Is It Better
20 Managing File Systems 439 Than NIS? 481
Understanding File Systems 440 Project: Installing an LDAP Server 482
Downloading and Installing
Mounting and Unmounting
OpenLDAP 482
File Systems 444
Mounting File Systems Interactively 445
x Special Edition Using Linux, Fifth Edition
Populating the Database 483 Running XFree86 in Probe-Only
Configuring a Client 484 Mode 523
23 Using Samba 485 Troubleshooting 524
Case Study: The X Window System Across
What Is Samba? 486
a Network 525
Installing Samba 487
25 Using the X Window System 529
Configuring Samba on Linux 487
The [global] Section 491 Navigating the X Window System 530
The [homes] Section 492 Getting Focus 530
The [printers] Section 493 Using Menus 531
Sharing Directories 493 Using Virtual Terminals in XFree86 531
Testing the smb.conf File 494
Using Window Managers for Linux 531
Running the Samba Server 494 twm 532
fvwm 533
Using smbclient 495 fvwm95 535
AfterStep 535
Case Study: OpenLinux and swat 497
Window Maker 535
Running swat from inetd 498
Blackbox 536
Running swat via the Web 498
Enlightenment 536
kwm 536
IV Using X Windows Choosing a Window Manager 536
24 Installing the X Window System 503 Themes 537
What Is the X Window System? 504 Display Managers and Logging In 538
xdm 538
Understanding the X Window System 504 gdm 539
What Is a Client/Server System? 506 kdm 539
Output Capabilities 507
User Interface Capabilities 507 Choosing Your Display Manager 539
Input Capabilities 508
XFree86 Startup 540
Installing the XFree86 System 508
Using X Applications 541
Ensuring Hardware Support for
xterm 541
XFree86 508
xv 544
Understanding the XFree86 RPMs 510
xcalc 545
Installing XFree86 for Red Hat 6.0 512
The GIMP 548
Configuring XFree86 514 Seyon 549
Using Xconfigurator 514 xlock 551
Using XF86Setup 515
Troubleshooting 551
Running the SuperProbe Program 515
Understanding the XF86Config
Project: Making Yourself at Home with
Sections 516 fvwm2 552
Running the xf86config Program 523 Choosing fvwm2 552
Contents xi
Preparing to Customize fvwm2 553 Navigating GNOME 582
Customizing the Menu 553 The Panel 582
Specifying Start-Up Programs 554 The Root Menu 585
Further Customizations 555 The Enlightenment Menu 585
Midnight Commander File Manager
26 Working with KDE 557 586
Tear-Off Menus 587
What Is KDE? 558
Configuring GNOME 587
Installing KDE 560 Making GNOME Useful 587
Installing KDE for Red Hat 6.0 560 Making It Pretty 590
Using the switchdesk Tool for Themes 591
Red Hat 6.0 561 Getting More Desktop 592
Choosing KDE Without switchdesk 562
The Pros and Cons of GNOME 592
Navigating KDE 562 Resources 593
The Panel 562 Performance 593
The Taskbar 566 Configuration 593
The Root Menu 567 Integration 593
Trash 568 Stability 593
Templates 568 Final Words 594
Autostart 568
KDE File Manager 568
V Network Administration
Configuring KDE 570
570
Templates
Bookmarks 571
28 Understanding the TCP/IP
Making It Pretty 571
Protocol Suite 597
Using Themes in KDE 573
The History of TCP/IP 598
Getting More Desktop 573
Internet Terminology 598
The Pros and Cons of KDE 574
Resources 574 The Open Systems Interconnection
Performance 574 Model 600
Configuration 574
Integration 575 The TCP/IP Protocol Stack 603
Stability 575
Final Analysis 575 IP Addresses 604
IP Address Classes 604
27 Working with GNOME 577 Network Naming 606
NIC Naming Tree 607
What Is GNOME? 578
Subnetworks and Subnet Masks 608
Installing GNOME 580
Using the switchdesk Tool for Red Hat
Routing 610
6.0 581 Routing Information Protocol (RIP) 610
Choosing GNOME Without
Network Segmentation 611
switchdesk 581
Internet Network Setup 611
Understanding the Types of
Connections 612
xii Special Edition Using Linux, Fifth Edition
Choosing a Networking Configuration ipchains General 646
614 Simple Firewall Policies 651
Understanding Network Configuration Implementing the Policies 653
Guidelines 615 Testing the Policies 654
Using Routers, Bridges, and Switches
616 Monitoring 654
Troubleshooting 617 Under Attack 655
655
29 Configuring a TCP/IP Network 619 Network Security Policy
Configuring IP Masquerading 655
Understanding the TCP/IP Configuration
Required Kernel Components 657
Files 620
Setting Up 658
The /etc/hosts File 620
The /etc/networks File 622 Troubleshooting IP Masquerading 660
Initializing Ethernet Interfaces 622
Using ifconfig to Inspect a Network
31 Connecting to the Internet 663
Interface 624
Understanding the Requirements for SLIP
Configuring the Software Loopback
and PPP 664
Interface 624
Configuring a Network Interface 625 Using dip to Automate SLIP Operations
Configuring Parallel IP Interfaces 625 664
Using dip in Command Mode 665
Understanding TCP/IP Routing 626
Using dip with Static IP Addresses 667
Deciding On a Routing Policy 626
Using dip with Dynamic IP
Using the /sbin/route Program 626
Addresses 669
Project: Monitoring a TCP/IP Network
Using diplogin to Provide SLIP
with netstat 630
Service 670
Displaying Active Network
Creating SLIP Accounts 670
Connections 631
Using the /etc/diphosts File 670
Examining the Kernel Routing
Table 633 Using PPP 671
Displaying Network Interface
Automating PPP Links with pppd and
Statistics 634 chat 671
Providing PPP Service 675
30 IP Firewalling and Masquerading 635 Keeping Your PPP Link Secure 676
IP Firewalling and Masquerading 636 Project: Configuring PPP with KDE 677
Introduction to Firewalls 636
Packet Filters636 VI Using the Internet
Proxy Firewalls 637
Which to Use? 638
Physical Configurations 639
32 Accessing the Network with telnet, ftp,
and the r- Commands 683
Port Forwarding 644
Using telnet to Access Remote
A Simple Packet Filtering Firewall 645 Computers 684
Planning 645 telnet Command Summary 684
Contents xiii
Sample telnet Session 685 34 Using Electronic Mail 717
Using ftp for Remote File Transfer 686 Understanding Email 718
Anonymous FTP 686
ftp Command Summary 687 Sending Email with mail 720
A Sample FTP Session 692 Writing a Message While Sending Email
A Sample Anonymous FTP Session 693 721
Canceling a Message 721
Using the r- Commands 695 Sending a Prepared Message 722
The rlogin Command 695 Sending the Result of a Command or
The rsh Command 696 Program by Email 723
The rcp Command 697
Reading Your Mail 723
The ssh Command 698 Using mail to Read Mail 723
Reading Email from Other Files 725
Troubleshooting 699 Sending Mail While Reading 726
33 Surfing the Internet with the World Wide Printing Mail Messages 726
Web 701
Getting Help with mail 727
Introducing the World Wide Web 702
Understanding the Web's Structure 702 Saving Email to Files with mail 728
Understanding URLs 703
Deleting and Undeleting Messages with
Searching the Web 704
mail 729
Using FTP with a Web Browser 706
Replying to Email with mail 729
Using Archie with a Web Browser 708
Routing Mail to Others 731
Using telnet with a Web Browser 709 Forwarding Messages 731
Sending a Copy with mail 732
Using gopher with a Web Browser 710 Using Aliases and Mailing Lists 733
Accessing Usenet News with a Web Customizing Your mail Environment 734
Browser 711
Quitting the mail Program 736
Getting on Mailing Lists 711 Quitting and Saving Changes 736
Finding Mailing Lists 712 Quitting and Not Saving Changes 737
Using Mailing Lists 712
Using the elm Mailer 737
majordomo 712 Starting elm 737
Using elm Commands 738
Using Wide Area Information Servers
(WAIS) 714 Project: Using the Mutt Email Client 739
Where to Get Mutt 740
Case Study: A Mailing List Subscription For More Information on Mutt 740
Process 714
35 Surviving Usenet News 741
What Is Usenet News? 742
A Usenet Glossary 742
xiv Special Edition Using Linux, Fifth Edition
A Brief History 744 Using .htaccess Files 776
How Usenet Is Structured 745 Understanding Security Issues 777
Group Hierarchies 745 CGI Issues 777
News Distributions 746 Trust Issues with Server-Side Includes
778
No Central Authority 747 Symbolic Links 778
Publicly Writable Spaces 779
Usenet Culture 747
Lack of Visual Reference 748 Project: Adding Customized Error
Newsgroup Culture 748 Messages 780
749
Reading and Posting News
Subscribing to Newsgroups 749
37 Configuring FTP Servers 781
Reading News 750
Installing WU-FTP 782
Replying via Email 750
Posting an Article 751 Anonymous FTP 782
Netiquette on Usenet 752 Configuring wu-ftpd 789
Project: Using the rn Newsreader 753 Project: Setting Up an Anonymous FTP
Server 790
Setting Permissions 790
VII Setting Up Linux Internet Servers Password and Group Files 791
36 Getting Started with Apache 759 38 Configuring Domain Name Service (DNS)
793
Installing Apache 760
How The Net Began 794
Establishing the File Hierarchy 760
Introducing DNS 794
Basic Configuration 761
httpd.conf 762 Configuring the Resolver 795
srm.conf 764 The /etc/host.conf File 796
access.conf 764 The /etc/resolv.conf File 797
User Directories 765
Using the named Daemon to Set Up the
Starting Apache 765 Server 798
The named.boot File 798
Debugging Server Startup 766 Database Files and Resource
Server Startup Error Messages 766 Records 800
Initial Server Startup Error
The named.hosts File 803
Messages 767 The named.rev File 805
The named.ca File 806
Secure Transactions with SSL 768
Troubleshooting 807
Special Modules 768
Server-Side Includes 768
Cookies 771
Configurable Logging 772
Host-Based Access Control 775
Contents xv
39 Configuring Email 809 Linux FTP Sites 840
An Overview of Electronic Mail 810 For Linux Developers 841
History and General Concepts 810
The Shared-File Messaging Model 810 B The Linux How-To Index 843
The Client/Server Messaging
Model 811 What Are Linux How-Tos? 844
MUAs, MTAs, and MDAs 811
The IETF Requests for Comment 812 Where Can I Get Linux How-Tos? 844
Internet Protocols 814 How-To Translations 844
Mail Message Formatting 815
How-To Index 845
sendmail 816 Mini-How-To Index 851
sendmail's History 817 Special How-To Index 857
Unmaintained How-Tos and
sendmail's Architecture 817
sendmail Rulesets 821 Mini-How-Tos 857
Project: Creating a Mailing List with
Writing and Submitting a How-To 858
sendmail and majordomo 822
Copyright 859
40 Configuring a Usenet News Service 825 C The GNU General Public License 861
A Usenet Primer 826
The GNU License 863
History and Origins of Usenet 826
Usenet Structure 827 Preamble 863
Usenet Servers 828 GNU General Public License Terms and
Conditions for Copying, Distribution, and
Configuring Usenet Clients 829
Modification 864
TIN and NN 829
How to Apply These Terms to Your New
Project: Configuring Pine as a
Programs 868
Newsreader 831
D The Open Source Definition 871
VIII Appendixes
The Open Source Definition 872
A Sources of Information 835 Index 875
Linux Web Sites 836
Usenet Newsgroups 837
Online Documents 839
Linux How-Tos 839
man Pages 840
Magazines 840
About the Authors
Jack Tackett, Jr. is the Senior System and Network Administrator for Railinc's Netredi
E-Commerce project, serving over 16 million transactions a month to North America's rail
industry. He oversees the daily administration for the site's servers and network conntections
across North America. He has been a system administrator for a variety of other operating
systemsÐfrom Unisys Mainframes to UNIX and Windows NT. Jack's books include the
bestselling Special Edition Using Linux (Editions 1-4), Red Hat Unleashed, Using Visual C++, and
The Visual C++ Construction Kit. Jack has been quoted in various news media including the
Wall Street Journal, PC Magazine, and The LA Times. He invites your comments at
Steve Burnett is an information technology consultant in Research Triangle Park, NC. His
most recent degree was a M.S. in Technical Communication. His professional interests have
centered around systems administration, integration, and interoperability. Any comments
may be sent to him at [email protected].
Rob Napier ([email protected]) is an engineer with Cisco Systems in Research
Triangle Park, NC. He has been working with Linux since 1995 and is currently an officer
for the Triangle Linux Users Group ( www.trilug.org). He is the author of the ``Finding the
Answers in Linux'' mini-HOWTO ( http://www.employees.org/~rnapier/
linuxanswers.html).
Jeff Tranter started using Linux in 1992. He is the author of the book Linux Multimedia
Guide, co-author of Tcl/Tk in a Nutshell, maintains the Linux CD-ROM and Sound
HOWTO documents, and has written some small Linux utilities and a number of magazine
articles. In spite of all that, he still considers playing with Linux a hobby and works by day as a
software developer for a large telecommunications company. He lives in Kanata, Ontario,
sometimes called Canada's Silicon Valley North, with his wife and two teenagers.
Dedication
To Peggy, Mary Louise, and CarolynÐLove always...
ÐJack
To Merrie.
ÐSteve
Acknowledgments
From Jack Tackett:
First, I want to thank all the readers of the first four editions of this book for their patronage
and for their helpful and insightful comments. Your comments are important and make a
difference, please keep them coming. You have made this a better project!
I want to say thank you for the tremendous efforts put forth by the Linux developers scattered
across the globe. I also want to acknowledge the fine contributions begun by Linus Torvalds
and continued by so many others around the worldÐthanks for creating Linux and breathing
life into such a monumental effort! Also, thanks to Matt Welsh, et al., for the work on the
Linux Documentation project.
Next, I want to express my regards for the people at the Que Continuum. I especially want to
thank Gretchen, Maureen, Chuck, Susan, Jeff, and John for their help in getting this project
off the ground and finished.
To Steve Burnett for once again lending his immense talents to getting this project
completed. Also thanks to Paul Barrett for his initial help with the research that eventually led
to this book.
To David Fugate of Waterside, the greatest agent in the world. Thanks for all your help,
David!
To my friends Paul Barrett, Keith E. Bugg, Gregg and Beckie Field, Dave and Lola Gunter,
Israel Janovich, Dianna Smith, Kell and Joy Wilson, Dennis and Carolyn Golden, Adrian
Polimico, Mark and Jennifer Shaw, and Joe Williams: Thanks for the memories! Thanks to
my combined familyÐthe Tacketts and the MartinsÐfor their support in all my endeavors.
Also, a big thank you to the best cousins in the worldÐBill and Hope Tackett, Jr.
I'd like to thank Dr. Joe Daugherty of the University of North Carolina-Asheville for all his
support. Also thanks to Myrtice Trent of the Blue Ridge Technical Community College.
Thanks for the help and encouragement you both provided.
Next a great big thank you to my co-workers at RailincÐthanks to everyone!
Finally, to my wife Peggy, who has yet again put up with me spending endless hours at the
computer writing yet another computer book. Thanks, sweetheart, and I love you!
xviii Chapter 0
From Steve Burnett:
To the Linux development community at large: your continued efforts have made enormous
improvements to the usage and capabilities of the operating system and applications. In the
short time since Jack and I wrote the previous edition of this book, Linux has grown so much
we felt a new edition was necessary.
Locally, the members of the recently formed Triangle Linux User Group have collectively
been of great help with their activity and enthusiasm. For more specific thanks: Jack, nice to
work together again. My thanks to Rob Napier for contributing on this edition, and to David
Fugate for his usual coordination. Of course, thanks to Merrie for tolerating my extra work.
Introduction
Linux is no longer a ``not ready for prime time'' operating system! That's what we wrote in the
beginning of the fourth edition of this book, and a few months later Linux exploded onto the
mainstream consciousness. Major news media from the Wall Street Journal to CNN seemed
to discover Linux overnight (probably not as a complete result of the fourth edition though
<g>). Today Linux is a viable alternative to the high-priced, closed-ended operating systems
of the past. Linux is one of two operating systems gaining market share, and it's gaining share
more quickly than the other one, Windows NT.
Many commercial uses of Linux abound, ranging from being used to create many of the
fantastic special effects for James Cameron's Oscar-winning mega hit Titanic; to being used as
the operating system for new network computers; to being used to create inexpensive,
massively parallel supercomputers as in the Beowulf project. Many large-scale IS departments
are no longer afraid to adopt Linux into their operations, even though many had Linux
systems clandestinely running services before the media attention! As Linux evolves, you have
to keep up with the changes; that's why you will find plenty of new material in this fifth
edition of Que's popular Special Edition Using Linux.
Many chapters have been rewritten to highlight the most available distributionsÐRed Hat,
Caldera, and Debian. In addition, the book contains updated and expanded coverage of X
Windows and two of the more popular window managers, KDE and GNOME. We've
rearranged several chapters to provide a better flow and ease of use in finding information
when you need it. Also, we've considered your suggestions and have revamped the installation
chapters and provided more troubleshooting tips throughout the book.
If you're just tuning in, though, you might want to know just what the heck Linux is. In 1991,
Linus Torvalds, then a 23-year-old college student, began a personal project to expand the
Minix operating system into a full-fledged clone of the UNIX operating system that was so
popular on college campuses. The project is still evolving: Linux is continuously updated and
expanded by literally hundreds of people around the world.
Therefore, Linux is a unique animal in the computer (r)evolution. It isn't a commercial
product backed by a huge corporation; rather, it's an operating system born of frustration and
built by a ragtag team of computer enthusiasts around the world. This team used Internet
resources to communicate and build the operating system named Linux.
xx Introduction
Chapter 0 Introduction
But don't think Linux is just a hobby for hackers around the worldÐit's not! Plenty of
commercial products are being written specifically for Linux. In fact, several companies are
porting their UNIX-based applications, such as Corel's WordPerfect, to Linux. As a matter
of fact, Corel has adopted the Debian distribution of Linux for its new Network Computer
and has released the resulting software to the development community. Companies such as
IBM, Oracle, and Dell are actively supporting Linux and the Open Source community.
Although estimating the total number of Linux users or installations in the world is
difficult, a working estimate shows between 5,000,000 and 10,500,000 active Linux
users worldwide. For more information, check out the following URL:
http://www.redhat.com/redhat/linuxmarket.html
If you don't understand what a uniform resource locator (URL) is or how to use it,
don't despair! This book will help you learn how to make your way around the
Internet using Linux.
? See ``Understanding URLs,'' p. 703
Many Fortune 500 companies use Linux for internal projects and mission-critical
applications. And recently, large companies such as IBM, Oracle, and Corel began to embrace
the concept of open solutions by releasing their own software into the development
community, just as Linus Torvalds and others released their software to the world.
Plenty of free applications and utilities are also available for Linux. Since the inception of
Linux, almost the entire GNU library of utilities has been ported to Linux, and the X
Windows GUI systemÐso popular on UNIX-type workstationsÐalso has been ported.
GNU (a recursive acronym for GNU's Not UNIX ) is a project started by one man to make
software available to anyone who wants access. The GNU General Public License in
Appendix C describes the philosophy under which Linux and many other fine software
packages are distributed. The accompanying CD-ROMs contain many of these packages.
This book provides you with enough information to use and enjoy Linux. The accompanying
CD-ROMs contain the Red Hat 6.0, Caldera OpenLinux 2.2 Lite, and Debian 2.1
distributions, each of which uses the 2.2 Linux kernel.
At this point, the first order of business is to help you pronounce the word Linux. To most
Americans, the pronunciation is LEN-nucks, with the short i sound. The official
pronunciation is LIH-nucks.
Who Should Use This Book? 1
ON THE WEB
You can hear Linus pronounce Linux in English at the following URL:
ftp://ftp.linux.org/pub/kernel/SillySounds/english.au
Who Should Use This Book?
Anyone interested in the Linux phenomenon can use this book as a guide to installing,
configuring, and using Linux. Linux is often called a UNIX clone, but it's actually a POSIX-
compliant multiuser, multitasking operating system for Intel 386 and later processors.
(POSIX is an international standard for operating systems and software detailing
interoperability standards.) Linux doesn't require MS-DOS or Windows to operate; in fact,
Linux can replace those programs on your computer.
Because Linux is still evolving, it's imperative that you understand the possibility of losing
existing data on your system. Do not install Linux without first backing up your system.You might
need to repartition your hard drive to make room for this new operating system, although you
can install Linux on top of MS-DOS or repartition your hard drive without losing data. If you
take the proper precautions, you can easily install and enjoy Linux.
The most current version of Linux is always available on the Internet from the sources
listed in Appendix A, ``Sources of Information.'' The accompanying CD-ROMs contain
the latest possible versions of Linux, but due to the rapid development of this popular
operating system and the chaotic process in which it's developed, providing the latest
and greatest versions on a CD-ROM is impossible. In fact, although all efforts are
made to keep the book and CD-ROMs in sync, that also is nearly impossible. Unlike
commercial software, which changes infrequently and under controlled conditions,
Linux and related software are perpetually dynamic.
Because Linux is similar to UNIX, many of the operations and procedures necessary for using
Linux also apply to many UNIX systems. By learning to use Linux, you also learn how to use
most UNIX systems.
UNIX has evolved over the years to become the premier operating system used by hundreds
of thousands of people throughout the world. This isn't an accident. Earlier versions of
UNIX were harder to manipulate than other operating systems, but despite this fact, UNIX
managed to amass a distinguished following in academic and scientific circles. These
professionals realized not only what a powerful, flexible, and manageable operating system
UNIX is, but also its potential to be the best operating system ever. Their efforts have
culminated in the UNIX of today with its marvelous utilities bundled with the newest
communications capabilities and graphical user interfaces (GUIs).
The UNIX of today promises again to revolutionize the personal computer industry and
perhaps redirect the industry's growth. UNIX has evolved from a minicomputer operating
2 Introduction
Chapter 0 Introduction
system to one that crosses all hardware platforms. We have no reason to think that this evolution
will stop. UNIX may well become the standard for what most users dream ofÐcomplete
standardization and compatibility of all computer systems, regardless of size or power.
UNIX comes in several flavors from a variety of vendors, including versions for the Intel PC
platforms, but most of these versions cost big bucks. Linux provides a relatively
inexpensiveÐfree if you have access to the InternetÐsolution to learning about UNIX-type
procedures and commands, understanding the X Windows GUI, and accessing the Internet
via Linux.
Who Should Not Use This Book?
If you are a Linux kernel hacker or a UNIX guru, this book may not be your cup of tea. This
book is a great resource if you want to know more about Linux and UNIX but have never
been involved with either operating system.
However, if you know how to install Linux and maneuver around in UNIX, you might still
find this book of use, particularly if you are only a UNIX user and have never had the chance
to perform system administration tasks. Several sections of the book explain the finer points
of system administration and how to maintain a Linux/UNIX system. Typically, a normal
UNIX user is never allowed to perform these system administration tasks, but with Linux,
you become king of the hill and ruler of the system, free to do whatever you want to do!
Now, if you don't have a clue what MS-DOS is or what a floppy disk looks like, you might
want to brush up on some computer basics before tackling Linux. Linux isn't for the faint of
heart; you must have some understanding of how a computer works. If the thought of
repartitioning or reformatting your hard drive sends shivers down your spine, you probably
should put off learning Linux for a while until you become more comfortable with your
computer system.
Hardware Needed to Use This Book
Most of Linux has been written across the Internet by computer hackers(not crackers, but
people who truly enjoy writing software that accomplishes something). Thus, the hardware
supported by Linux is the hardware owned by the various hackers.
Many hardware manufacturers are accepting Linux as a valid market and are beginning to
write drivers for their hardware. They are also providing hardware specifications to the world
so that Linux developers can write software to work with the hardware. Many companies are
also farming out work to Linux developers specifically to write drivers for their hardware.
These companies then release the code into the community under GNU guidelines. This is a
dramatic change from a few years ago when many manufacturers withheld information for
proprietary and competitive reasons.
Table I.1 provides a brief list of the supported hardware. If you don't have the correct
hardware, it's unlikely that you'll be able to boot Linux and productively use the system.
Hardware Needed to Use This Book 3
Forewarned is forearmed! If you are unsure about your hardware, then you can check out the
Linux Hardware Database on the Web at the following site:
http://ldh.datapower.com.
Red Hat also maintains a list of supported hardware at the following site:
http://www.redhat.com/corp/support/hardware/intel/60/rh6.0-hcl-i.ld.html
Our Linux Environment
We feel it's only fair to let you know what types of systems we used to create this book. The test machine was a
Pentium II 233-based system with 64MB of RAM and an Adaptec SCSI controller, a 6 gigabyte Segate SCSI drive,
an NE2000 PCI Ethernet card, a 24x ATAPI CD-ROM, and a Matrox Mystique video card. The name server used
in our network is a no-name 486dx100 system with IDE drives and 32MB of memory running Red Hat Linux.
This machine is also the main sendmail server for the site. The main Web server is a Digital Equipment Alpha
(multia) also running Red Hat Linux. The entire site is connected to the Internet via an ISDN line using an
Ascend Pipeline 75 router.
Table I.1 A Brief List of Hardware Supported by Linux
Item Description
CPU Intel 386 and later (and compatibles), DEC Alpha, Sun Sparcs, and
PowerMacs.
Bus ISA, EISA, VESA local bus, and PCI; the MicroChannel bus isn't
fully supported yet.
RAM Minimum of 2MB of RAM; 4MB is recommended.
Hard drive controller AT standard hard drive controller; Linux supports MFM, RLL,
ESDI, and IDE controllers. Linux also supports several popular
SCSI drive and CD-ROM drive controllers.
Disk space Minimum of 20MB; 80MB is recommended.
Monitor Linux supports Hercules, CGA, EGA, VGA, and SVGA video
cards and systems; X Windows has other requirements detailed in
Chapter 24, ``Installing the X Windows System.''
Mouse Any standard serial mouse (for example, Logitech, Microsoft, or
Mouse Systems) or bus mouse from Microsoft, Logitech, or
ATIXL.
CD-ROM drive Any CD-ROM drive that uses a true SCSI interface works; some
proprietary CD-ROM drives such as the SoundBlaster series are
also supported. CD-ROM drives known to work with Linux
include NEC CDR-74, Sony CDU-45, Sony CDU-31a, Mitsumi
CD-ROMs, and Texel DM-3042.
Tape drive Any SCSI tape drive works; other drives hosted from a floppy
controller may also be supported. Now, the Colorado Jumbo 120
and 250 using the QIC 80 format are supported.
Printer If you can access your parallel printer from MS-DOS, you should
be able to access it from Linux; some fancy features might not be
accessible.
4 Introduction
Chapter 0 Introduction
Table I.1 A Brief List of Hardware Supported by Linux
Item Description
Ethernet card If you have access to an Ethernet network, Linux supports several
standard Ethernet cards for accessing your network. Cards
supported include 3Com's 3C503, 3C509, and 3C503/16; Novell's
NE1000 and NE2000; and Western Digital's WD8003 and
WD8013.
ON THE WEB
The following Web site provides more information on MicroChannel bus support:
http://glycerine.itsmm.uni.edu/mca/
How to Use This Book
You might prefer to read this book from cover to cover. The information progresses from
simple to complex as you read through the various sections and their chapters. Because the
information is separated into seven parts and five appendixes, each with its own particular
emphasis, you can choose to read only those parts that appeal to your immediate needs.
Don't, however, let your immediate needs deter you from eventually giving attention to each
chapter. Whenever you have the time, you can find a wealth of information in them all!
Part I: Installing Linux
Part I, ``Installing Linux,'' provides a detailed overview of the Linux system as well as
instructions to get Linux up and running. It consists of seven chapters:
& Chapter 1, ``Understanding Linux,'' introduces the Linux operating system and
provides a general overview of the various components that make up the Linux system
and various distributions.
& Chapter 2, ``Linux Installation Overview,'' provides a general overview of installing
various Linux distributions, with specific emphasis on supported hardware and
potential problems and their resolutions.
& Chapter 3, ``Installing Red Hat Linux,'' gives detailed instructions for installing the
version of Red Hat provided on the accompanying Red Hat CD-ROM.
& Chapter 4, ``Installing Caldera OpenLinux,'' gives detailed instructions for installing
the version of OpenLinux provided on the accompanying CD-ROM.
& Chapter 5, ``Installing Debian Linux,'' provides a basic introduction to installing the
version of GNU/Debian Linux provided on the accompanying CD-ROM.
How to Use This Book 5
& Chapter 6, ``Adding Sound Cards and Other Hardware,'' provides instructions on
adding sound to your Linux system as well as other hardware.
& Chapter 7, ``Upgrading and Installing Software,'' provides you with the information
needed to install new software using the Red Hat Package Management system (RPM)
and the Debian Package Management system. The chapter also covers installing
software from the Internet and tells you how to patch existing programs.
Part II: System Administration
Part II, ``System Administration,'' provides basic information on configuring and managing a
typical Linux installation. The following eight chapters cover this topic:
& Chapter 8, ``Understanding System Administration,'' provides a brief background of
the processes and procedures needed to configure and maintain a Linux system.
& Chapter 9, ``Using the vi Editor,'' instructs you how to use UNIX's visual editor.
Although vi isn't the most productive editor in the world, every Linux/UNIX system
has it, and sometimes it's the only editor available for use.
& Chapter 10, ``Booting and Shutting Down,'' details the various actions that happen
when you boot up or shut down a Linux system, and it explains why you can't simply
switch off the power supply. This chapter contains a complete description of the files
Linux uses to boot.
& Chapter 11, ``Managing User Accounts,'' shows you how to add, delete, and manage
user accounts on your machine.
& Chapter 12, ``Backing Up Data,'' explains the necessity of backing up your data, as
well as the procedures needed to back up your Linux system.
& Chapter 13, ``Improving System Security,'' gives you a brief overview of system
security on Linux systems and then explains the procedures needed to maintain a
reasonably secure system.
& Chapter 14, ``Configuring the Linux Kernel,'' illustrates how to configure a kernel, no
matter what distribution you are using, for your hardware.
& Chapter 15, ``Working with the Linux PPC Kernel,'' illustrates how to configure a
kernel for the Power PC chip found in a variety of computers such as Apple Macintosh
systems.
Part III: Working with Linux
Part III, ``Working with Linux,'' increases your skill at working with the Linux command-line
tools and utilities. This section also provides detailed knowledge of how to be more
productive with various Linux features. Everything you learn in these eight chapters can be
transferred easily to other UNIX-type systems:
6 Introduction
Chapter 0 Introduction
& Chapter 16, ``Understanding Linux Shells,'' introduces you to the magical world of
Linux shells, the powerful capabilities that exist through the use of shell scripting, and
the different shells you might encounter with different versions of Linux.
& Chapter 17, ``Managing Multiple Processes,'' explores the capabilities of Linux when
you run more than one process at a time. You learn how to initiate and manage
multiple processes, as well as how to control and stop them.
& Chapter 18, ``Printing,'' covers all the printing basics, from issuing print commands
and checking printer status to canceling print jobs and dealing with common printing
problems.
& Chapter 19, ``Understanding the File and Directory System,'' provides an overview of
file permissions, users, and file types.
& Chapter 20, ``Managing File Systems,'' provides an overview of creating, mounting,
and using a file system under Linux.
& Chapter 21, ``Managing NFS and the Automounter,'' provides a detailed explanation
of Network File System (NFS) services and how to configure Linux to use NFS with
other systems.
& Chapter 22, ``Managing NIS and LDAP,'' provides an overview of Network
Information Services (NIS) and how to configure Linux to use NIS with other systems
in your network.
& Chapter 23, ``Using Samba,'' provides a detailed explanation of Samba and how to
configure Linux to use Samba with other Linux systems, as well as with NT systems.
Part IV: Using X Windows
Part IV, ``Using X Windows,'' provides a greater understanding of the procedures and
processes necessary to install, configure, and use the X Windows system on Linux. This
section also provides an overview of the top two window managers for X KDE and
GNOME.
& Chapter 24, ``Installing the X Windows System,'' provides you with the necessary
information to get the X Windows system up and running under Linux. Under Linux,
the X Windows system is called XFree86 and is similar to other GUI environments
such as Microsoft Windows or the OS/2 Workplace Shell.
& Chapter 25, ``Using X Windows,'' provides you with information necessary to use the
X Windows system under Linux.
& Chapter 26, ``Working with KDE,'' provides you with information necessary to use
the KDE window manager with X under Linux. KDE is typically the default window
manager for Debian Distributions.
& Chapter 27, ``Working with GNOME,'' provides you with information necessary to
use the GNOME window manager with X under Linux. GNOME is installed by
default on Linux distributions based on Red Hat.
How to Use This Book 7
Part V: Network Administration
Part V, ``Network Administration,'' provides a greater understanding of the procedures and
processes necessary to administer a robust Linux system.
& Chapter 28, ``Understanding the TCP/IP Protocol Suite,'' provides an overview of the
network transport protocol suite in use today on the Internet.
& Chapter 29, ``Configuring a TCP/IP Network,'' shows you how to set up and
configure TCP/IP on Linux.
& Chapter 30, ``Configuring IP Masquerade,'' gives you the basics of IP masquerade.
This chapter covers compiling, installing, and configuring the necessary files to create
a firewall and basic administration of the system.
& Chapter 31, ``Connecting to the Internet,'' illustrates how to configure and use
Point-to-Point Protocol (PPP) lines to connect with the Internet.
Part VI: Using the Internet
The four chapters in Part VI, ``Using the Internet,'' provide a basic overview of the Internet.
& Chapter 32, ``Accessing the Network with telnet, ftp, and the r- Commands,''
provides you with information on how to use various programs such as Telnet and
FTP to access information around the world.
& Chapter 33, ``Surfing the Internet with the World Wide Web,'' gives you an overview
of using various Linux utilities to search for and retrieve information from the
Internet, with emphasis on the Web.
& Chapter 34, ``Using Electronic Mail,'' gives you an overview of electronic mail (email)
and how to use it in Linux.
& Chapter 35, ``Surviving Usenet News,'' provides you with an explanation of Usenet
newsgroups, as well as instructions for accessing this global community of
newsgroups.
Part VII: Setting Up Linux Internet Servers
Part VII, ``Setting Up Linux Internet Servers,'' provides detailed information on setting up
and running various Internet servers on Linux. Five chapters make up this part:
& Chapter 36, ``Getting Started with Apache,'' gives you the basics to get started with
Apache. It covers compiling and installing Apache, and it gives you the basic
configuration options.
& Chapter 37, ``Configuring an FTP Server,'' discusses the major configuration options
for creating an anonymous FTP server.
& Chapter 38, ``Configuring Domain Name Service (DNS),'' provides you with the
necessary information to get your system up and running with Domain Name Service
(DNS).
8 Introduction
Chapter 0 Introduction
& Chapter 39, ``Configuring Email,'' provides you with the necessary information to get
your email system up and running with sendmail.
& Chapter 40, ``Configuring a Usenet News Service,'' provides you with the necessary
information to set up Usenet news on your system.
Part VIII: Appendixes
The appendixes provide supplementary information on installing and using Linux, as well as
licensing information for using Linux. The book contains the following five appendixes:
& Appendix A, ``Sources of Information,'' provides you with a detailed listing of books,
magazines, Usenet newsgroups, and FTP sites dealing with Linux. Also, you get a
brief glimpse of the myriad resources available to you as a Linux user.
& Appendix B, ``The Linux How-To Index,'' provides a list of all the main and mini-
How-Tos available. How-Tos provide information on how to accomplish a specific
task with Linux. This How-To comes directly from the Internet.
& Appendix C, ``The GNU General Public License,'' is the verbatim license for using
GNU applications. It describes your responsibilities when modifying, distributing, or
using GNU programs.
& Appendix D, ``The Open Source Definition,'' is the verbatim definition of the Open
Source Movement as defined at www.opensource.org.
& Appendix E, ``What's on the CD-ROM?'' discusses the contents of the three CD-
ROMs included with this book.
Conventions Used in This Book
This book uses several special conventions that you need to become familiar with. These
conventions are listed here for your reference:
& Linux is a case-sensitiveoperating system; that means when this book instructs you to
type something at a command or shell prompt, you must type exactly what appears in
the book, exactly as it is capitalized.
& This book uses a monospaced typeface for Linux commands to set them off from
standard text.
& If you're instructed to type something, what you are to type also appears in monospace
text. For example, if the book gives the instruction
Enter cat
you must press the letters c, a, and t and then press the Enter key.
Conventions Used in This Book 9
& Keys are sometimes pressed in combination; when this is the case, the keys are
presented like this:
Ctrl+h
This example implies that you must press and hold the Ctrl key, press the h key,
and then release both keys.
This book uses a convention for key names that may vary from what you are
accustomed to. To avoid confusion in the case-sensitive UNIX environment, this book
uses lowercase letters to refer to keys when uppercase letters may be the norm. For
example, this book uses the form Ctrl+c instead of the form Ctrl+C (the latter form
may make some readers wonder whether they should press Ctrl and Shift and c).
& Some sample listings show a portion of the screen after you type a specific command.
These listings show the command prompt or shell promptÐusually a dollar sign
$
( )Ðfollowed by what you type in bold monospace. Don't type the dollar sign when
you follow the example on your own system. Consider this example:
$ lp report.txt &
3146
$
You should type only what appears in bold on the first line (that is, type lp report.txt
& and then press Enter). The rest of the listing shows Linux's response to the
command.
& When discussing the syntax of a Linux command, this book uses some special
formatting to distinguish between the required portions and the variable portions.
Consider the following example:
lp filename
In this syntax, the filename portion of the command is a variable; that is, it changes
depending on what file you actually want the lp command to work with. The lp is
required because it's the actual command name. Variable information is presented in
italic monospace; information that must be typed exactly is not in italic.
& In some cases, command information is optional; that is, it's not required for the
command to work. Square brackets ( []) enclose optional parts of the command syntax.
Consider the following example:
lp filename [device1] [abc]
Here, lp is the command name and is neither optional nor variable. The device1
parameter is both variable and optional (it is in italic and enclosed in square brackets);
this means that you can type any device name in place of device1 (without the
brackets), or you can type nothing at all for that parameter. The abc parameter is
optional (you don't have to use it if you don't want to), but it's not variable; if you use
it, you must type it exactly as it appears in the bookÐagain, without the brackets.
10 Chapter 0 Introduction
& Tips, notes, and cautions appear throughout the book in special formats to make the
information they contain easy to locate. Longer discussions not integral to the flow of
the chapter are set aside as sidebars with their own headings.
& The book also contains many cross-references to appropriate topics throughout the
book. A typical cross-reference appears as follows:
? See ``Using X Windows,'' p. xxx
PART
I
Installing Linux
1 Understanding Linux 13
2 Linux Installation Overview 31
3 Installing Red Had Linux 61
4 Installing Caldera OpenLinux 93
5 Installing Debian Linux 119
6 Adding Sound Cards and Other Multimedia Hardware 145
7 Upgrading and Installing Software 163
CHAPTER
1
Understanding Linux
In this chapter by Jack Tackett, Jr.
What Is Linux? 14
Why Use Linux? 15
Linux Distributions 16
Advantages of Using Linux 17
Disadvantages of Using Linux 20
Overcoming the Disadvantages 23
Disappearing Disadvantages 24
The Commercial Side of Linux 24
A Brief History of Linux 25
Who Owns Linux? 28
14 Chapter 1 Understanding Linux
What Is Linux?
To understand Linux, you must first understand the question, ``What is UNIX?'' UNIX is
arguably the most versatile and popular operating system found today on scientific and high-
end workstations.
Linux is a project initiated to create a working version of UNIX on Intel-based machines,
more commonly referred to as IBM PC-compatible computers that most people are familiar
with.
The Linux operating system has been designed and built by hundreds of programmers
scattered around the world. The goal has been to create a UNIX clone, free of any
commercially copyrighted software, which the entire world can use.
Actually, Linux started out as a hobby of Linus Torvalds while he was a student at the
University of Helsinki in Finland. He wanted to create a replacement for the Minix operating
system, a UNIX-like operating system available for Intel-based PCs.
Many of the terms used within the chapter are discussed later, so don't worry if some
of them are unfamiliar to you now.
Linux is basically a UNIX clone, which means that with Linux you get many of the
advantages of UNIX. Linux multitasking is fully preemptive, meaning that you can run
multiple programs at the same time, and each program seems to process continuously. Other
systems, such as Microsoft Windows 3.1, allow you to run multiple programs, but when you
switch from one program to another, the first program typically stops running. Microsoft's
Windows 98 and Windows NT are more like Linux because they allow preemptive
multitasking. Linux allows you to start a file transfer, print a document, copy a floppy, use a
CD-ROM, and play a gameÐall at the same time.
Linux is fully multiuser capable, which means that more than one person can log in to and use
the system at the same time. Although the multiuser feature may not be very useful at home,
it gives many people in a corporate or university setting access to the same resources at the
same time yet eliminates the need to duplicate expensive machines. Even at home, you'll find
the capability to log in to separate accounts on what are called virtual terminals very useful.
Also from home, you could provide your own personal online service by using Linux and
several modems.
? See ``Managing Users,'' p. 47
Linux is freeÐor nearly so. In fact, for the price of this book, you've received three fully
functioning distributions of Linux (Red Hat Linux, Caldera OpenLinux, and Debian) on the
accompanying CD-ROMs. Everything you need to get Linux up and running is provided on
the CD-ROMs, including hundreds of applications.
Why Use Linux? 15
Linux provides a learning opportunity unparalleled today. Here, you have a complete
Part
working operating system, including source code, with which to play and learn what makes it
tick. Learning what makes Linux tick is something you can't do in a typical UNIX I
environment, and it's definitely something you can't do with a commercial operating system
Ch
because no vendor is willing to give away the source code.
1
Finally, Linux gives you a chance to reliveÐor perhaps experience for the first timeÐthe
chaos of the early PC revolution. In the mid-1970s, computers were the provinces of large
organizations, such as government, big business, and universities. The ordinary person had
no access to these marvels. But with the introduction of the microprocessor and the first
personal computers, things changed. At first, PCs were the province of the hackersÐdedicated
computer enthusiastsÐwho hacked the early systems because those systems could do very
little in the way of productive work. But as the hackers experimented and became
entrepreneurs, and as the capabilities of PCs increased, PCs became commonplace.
The term hacker has unfortunately taken on a negative connotation in today's society.
See the section ``Hackers'' later in this chapter for more details on hackers versus
crackers.
The same is true today of system software (that is, operating systems). Linux represents a
breakaway from a system controlled by large organizations that stifle creativity and
enhancements in the name of market share.
Why Use Linux?
You'll want to use Linux because it's the only operating system today that's freely available to
provide multitasking and multiprocessing capabilities for multiple users on IBM PC-
compatible hardware platforms. No other operating system gives you these same features
with the power that Linux enjoys. Linux also separates you from the marketing whims of the
various commercial providers. You aren't locked into upgrading every few years and paying
outrageous sums to update all your applications. Many applications for Linux are freely
available on the Internet, just as the source code to Linux itself is available on the Internet.
Thus, you have access to the source code to modify and expand the operating system to your
needsÐsomething you can't do with commercial operating systems such as Windows NT,
Windows 98, MS-DOS, and OS/2.
Freedom from commercial vendors is also a potential downside to using Linux. Because no
single commercial vendor supports Linux, getting help isn't just a phone call away. Linux can
be finicky and may or may not run properly on a wide range of hardware. The potential to
damage or delete data files residing on your system also exists because Linux is constantly
changing and doesn't go through a rigorous testing process before it's released.
16 Chapter 1 Understanding Linux
Linux isn't a toy; it's a system designed to give users the feeling of tinkering with a new
project, just like in the beginning of the PC revolution. However, Linux is relatively stable on
many systems and presents you with an inexpensive opportunity to learn and use one of the
most popular operating systems in the world todayÐUNIX. Many CD-ROM vendors and
software companies, such as Red Hat and Caldera, now support the Linux operating system.
Linux is an alternative to other UNIX systems and can be used in place of those sometimes-
expensive systems. If you program on UNIX systems at work, for example, you might want a
UNIX-like system at home. Are you a system administrator of a UNIX system at work? If so,
you can perform some of your duties from home by using Linux. Or do you not have a clue as
to what UNIX is? Well, then, Linux provides a low-cost introduction to UNIX.
Linux also provides you with easy access to the Internet and the rest of the information
superhighway.
Linux Distributions
Linux is distributed by many different organizations, each of which provides a unique
collection of programs along with the core group of files that constitute a Linux release. The
current release of Linux on the accompanying CD-ROMs is kernel version 2.2.5-15. This
distribution may also contain experimental kernels with drivers for unique hardware. Under
Red Hat, the kernels are part of the Red Hat Package Management (RPM) system and are
installed as part of the system. Caldera's OpenLinux follows the same scheme because it is
based on the Red Hat distribution.
Luckily for you, by having bought this book, you've made the decision of which distribution
to use rather easy. The three CDs accompanying this book offer complete versions of Red
Hat's, Debian's, and Caldera's distributions (the companies' Internet versions, not the ones
sold commercially). The following lists many distributions that are available on the Net and
from various CD-ROM vendors:
& MCC Interim Linux
& Linux Mandrake
& LinuxPPC
& LinuxWare
& Slackware Linux
& S.u.S.E. Linux
& TurboLinux
& Debian Linux
& Yggdrasil Plug-and-Play Linux CD-ROM and the Linux Bible
Advantages of Using Linux 17
& Caldera (this vendor uses Red Hat's distribution as the base for its distribution, then
Part
adds additional features such as Novell and Windows support along with customer
support) I
& Red Hat Ch
The ``Distribution How-To'' also provides an exhaustive list of Linux distributions. You'll
1
learn later in this chapter how to access the various How-Tos that accompany each Linux
release.
Tip #1 from
Jack After installing Linux, you can find the distribution How-To in the directory
/usr/doc/HOW_TO/.
Before installation, you can find the file in the following directories:
. Red Hat: < cdrom-drive letter >:doc/HOWTO/other-formats/html/
Distribution-HOWTO.html
. cdrom-drive letter>:col/doc/HOWTO/other-formats/
Caldera: <
html/Distribution-HOWTO.html
. Debian: not available from the CD-ROM.
. On the web at http://www.linux.org/help/howto.html.
Advantages of Using Linux
Using Linux has many advantages. Of the many operating systems available today, Linux is
the most popular free system that's widely available. For the IBM PC, Linux provides a
complete system with built-in multiuser and multitasking capabilities that take advantage of
the entire processing power of your 386 and higher computer systems.
Linux comes with a complete implementation of the TCP/IP networking protocol. With
Linux, you can connect to the Internet and the vast wealth of information it contains. Linux
also provides a complete email system to send messages back and forth through cyberspace.
Linux also has a complete graphical user interface (GUI), XFree86, that's based on the
popular X Window System. XFree86 is a complete implementation of the X Window System
that can be distributed free of charge with Linux. XFree86 provides the common GUI
elements you find on other commercial GUI platforms, such as Windows and OS/2.
Today, all these features are available for Linux and are basically free. All you have to pay is
the price for acquiring the programs from the Internet or via mail order (available from
several different vendors). Of course, because you've purchased this book, you already have
the entire Linux system on the accompanying CD-ROMs.
18 Chapter 1 Understanding Linux
Open Systems Portability
In the never-ending quest for standardization, many organizations have taken a renewed
interest in the direction in which operating systems are developing. UNIX hasn't gone
unnoticed. The drive to standardize UNIX stems from the many UNIX variants now
available. You'll learn more about how those variants were developed in the following section.
Efforts have been made to combine, collate, and otherwise absorb all versions of UNIX into a
single all-encompassing version of the operating system. Initially, the effort met with guarded
enthusiasm, and some effort was expended on coming to terms with blending the different
versions. As with many noble efforts, this one was doomed to failure because developers
weren't willing to sacrifice part of what they had already invested in their particular versions.
(Sadly enough, many developers still feel that way.)
However, the continued existence of UNIX varieties isn't necessarily cause for alarm. Despite
the different varieties, all are still inherently superior to all other operating systems available
today because each contains the same elements described in the preceding pages.
Portability is merely the capability to transport an operating system from one platform to
another so that it still performs the way it should. UNIX is indeed a portable operating
system. Initially, UNIX could operate on only one specific platformÐthe DEC PDP-7
minicomputer. Today, the many UNIX variants can operate in any environment and on any
platform, from laptops to mainframes.
Portability provides the means for different computer platforms running UNIX to
communicate accurately and effectively with any of the other platforms. These systems can
communicate without the addition of special high-priced after-market communications
interfaces. No other operating system in existence can make this claim.
Applications
Although using an operating system is sometimes fun in and of itself, it isn't the reason most
people use a computer. Most people need to do productive work with their computers. Linux
has literally thousands of applications available today, including programs for spreadsheets,
databases, word processing, application development in a variety of computer languages, and
telecommunications packages to get you online. Linux also comes with a wide range of
games, both text- and graphics-based. When you need a break from the drudgery of the daily
grind, Linux can provide a few minutes (or hours) of relaxation.
Advantages for Computer Professionals
If you're a computer professional, Linux provides a wealth of tools for program development.
It includes compilers for many of the top computer programming languages today, such as C,
C++, and Smalltalk. If you don't like those languages, Linux provides you with tools, such as
Flex and Bison, that you can use to build your own computer languages. These tools come
with the CD-ROMs accompanying this book, but their commercial counterparts can cost
several hundred dollars each. If you want to learn one of the aforementioned languages but
Advantages of Using Linux 19
don't want to spend hundreds of dollars for another compiler, Linux and its development tools are
for you. Part
Linux also allows you to communicate with your company's office systems. And if you're a
I
UNIX system administrator, Linux can help you perform your duties from home. Although Ch
working from home is booming as more companies realize the productivity increases and
1
savings to overhead, perhaps some day you too can use Linux to do your job at home and then
only occasionally visit the office for personal meetings.
Two of the industry's buzzwords are open systems and interoperability, both of which refer to
the capability of many different systems to communicate with one another. Most open
systems specifications require Portable Operating System Interface (POSIX) compliance,
which means some form of UNIX. Linux meets those standards today. In fact, Linux was
designed for source-code portability, so if you have a corporate program running on one
version of UNIX, you should be able to port that system relatively quickly to a system
running Linux.
Corporations are insisting on these types of open systems so that they aren't locked into using
any one vendor. Remember the old adage ``Don't put all your eggs in one basket''?
Corporations today are becoming leery of systems controlled by single companies because
those in control can dictate how the software behaves and what hardware systems the software
supports. If that company chooses a direction that's not good for your corporation, tough
luck. You're stuck with that company's decision whether you like it or not. With UNIX/
Linux and open systems, however, you're in control of your own destiny, and many of today's
top companiesÐIBM, Dell, and NortelÐare embracing this new technology. Why? Because
if the operating system doesn't have a feature they need, they can find plenty of consultants
who can make the necessary changes, which is possible because you have the source code to
your operating system.
Education
Students, note that Linux provides you with editors to write your assignments and spell
checkers to proof those assignments. With Linux, you should be able to log in to your
school's computer network. Of course, with access to the Internet, you also have an instant
tap into the limitless wealth of information there. You also have access to thousands of experts
in a wide variety of subjects who can answer your questions. Linux can be useful, even if your
major isn't computer science.
Linux provides such advantages for so little because of the spirit and philosophy of the
community that built and continues to build it. Linux is a great experiment that took on a life
of its own. Literally hundreds of computer hackers from around the world contributed to its
development. Linus Torvalds first developed what became Linux for himself and later
released his brainchild to the world under the GNU copyleft (as opposed to a copyright).
? See ``The GNU General Public License,'' p. 861
20 Chapter 1 Understanding Linux
Hackers
At the basic level, Linux is a system built by and for hackers. The popular definition of hacker
has a negative connotation in today's society, but computer hackers aren't criminals by their
definition of the word. Their definition deals with how they approach any activity in lifeÐnot
just when dealing with computers. Hackers feel a certain depth of commitment and an
enhanced level of excitement at hacking a system. Hacking basically means learning all there is
to know about a system, becoming immersed in the system to the point of distraction, and
being able to fix the system if it breaks.
Hackers basically want to know how a system they find interesting works. Most are not
interested in making money or seeking revenge, although certain hackers do cross that line to
become what the hacker community calls crackers. Computer hackers become outraged when
they're compared with these vandals and criminals the popular media now call hackers (instead
of crackers). We hope that Linux gives you a feeling of what it's like to be a hacker, and ideally,
you will avoid becoming a cracker.
If you're simply the curious type and want to learn more about UNIX, Linux is for you. In it,
you'll find a fully functional version of UNIX to which you have free, unrestricted
accessÐsomething you seldom find in the real world. Most UNIX users are given accounts
on UNIX machines that grant them only limited rights and privileges, and in such cases,
normal users can't use or experiment with certain UNIX/Linux commands. But this
environment isn't conducive to learning all about UNIX. With Linux, however, you have
complete run of the place and can do what you want whenever you want. Of course, with this
great power comes great responsibility: You must learn how to manage a real UNIX system,
which can be fun in and of itself.
Disadvantages of Using Linux
Linux has many advantages, and disadvantages. No operating system can meet the needs of
everyone, including Linux. The following section outlines some of those disadvantages.
Lack of Technical Support
Perhaps the biggest disadvantage of using Linux is the fact that no single corporate entity is in
charge of its development. If something goes wrong or you have a problem, no toll-free
technical support numbers are available for you to call for help. But when you really think
about it, do such numbers provide real support for current commercial systems? How often
are you referred elsewhereÐproviding you get through to tech supportÐto have your
question answered? How many times are you asked to post a question on an online service to
get help? Well, with Linux, although no tech support number is available, literally thousands
of users in the online communities can help answer your questions. (See Appendix A,
``Sources of Information,'' for places to go for help.) Also, many providers, particularly Red
Hat and Global Knowledge, now offer certifications in installing and maintaining Linux
systems.
Disadvantages of Using Linux 21
In the past, having no source of technical support was a problem with Linux, no doubt about
Part
it. The same was true of Linux applications. Although a few commercial programs are
available for Linux, most programs are developed by small groups and then posted to the I
world via the Internet. Many developers, however, do help out with questions. Today, most
Ch
of the major software application vendors are rushing to port their products to
1
LinuxÐamong them Oracle, Corel, and SAS.
Many commercial companies are now building Linux applications that they sell. For
people to use their applications, though, these companies typically provide a free
copy of a Linux distribution along with their product and thus supply technical
support for that version of Linux.
Hardware Problems
Other disadvantages are that Linux can be hard to install, and it doesn't work on all hardware
platforms. Unlike in a commercial program development operation, for which a cohesive
group spends months building and testing a program against a variety of conditions and
hardware, Linux developers are scattered across the globe. No formal quality-assurance
program exists. Developers release their programs when they feel like releasing them. Also,
the hardware supported by Linux depends on the hardware each developer owns while
writing that portion of the code. Thus, Linux doesn't work with all the hardware available for
PCs today.
Of course as many software vendors rush to support Linux, so too have hardware vendors.
IBM is the biggest proponent of Linux. It goes out of its way to ensure hardware from various
vendors will work on Linux (and other Unix OS) in their Netfinity server line. In fact IBM
adopted Apache as its Web server.
If your system doesn't have the hardware supported by Linux, you'll have problems
installing and running the system. Chapter 3, ``Installing Red Hat Linux,'' Chapter 4,
``Installing Caldera OpenLinux,'' and Chapter 5, ``Installing Debian Linux,'' provide
details on the hardware you need to use Linux. You can also search the Linux
Hardware database online at http://lhd.datapower.com.
If you have the hardware that's supported, chances are you'll have no problem installing and
using Linux. If you don't have the necessary hardwareÐwellÐLinux developers expect you to
fix it. After all, it is a hacker's system.
22 Chapter 1 Understanding Linux
Inability to Use Current Software
Another disadvantage is that your current applications for such operating systems as DOS
and OS/2 more than likely won't work under Linux. Fortunately, those other systems can
coexistwith Linux, overview of; thus, although you can't use both operating systems at the
same time, you can leave Linux and boot the other operating system to use your applications
there.
Work is in progress on Linux emulators that run DOS and Windows programs, as well as the
Executor project to run Macintosh programs under Linux. Red Hat and Caldera both
support 16-bit DOS programs. The WINE emulator is still under development.
Tip #2 from
Jack To use DOS under Red Hat Linux, simply type the command DOS:
[root@test]#dos
You can then run any available 16-bit DOS programs.
Also, Caldera, Inc. has ported Sun's Windows Applications Binary Interface (WABI) product
to Linux. WABI allows Windows 3.1 applications to run under X on Linux. Caldera sells this
product, unlike many Linux applications, along with several other Linux applications.
However, Caldera provides the Red Hat distribution of Linux free of charge to run the
applications the company sells.
To install Linux, you typically have to repartition your hard driveÐalthough repartitioning is
not always necessary. Repartitioning means erasing part of your drive, which wipes out your
programs and data on that drive. Currently, you cannot install Linux safely without
repartitioning. If you plan to install Linux, you should back up your disk first (making two or
three backups is safest). Also, you might not have enough hard disk space to install Linux and
keep your other software on the same disk, in which case you have to decide what goes and
what stays. No matter what, you have to back up your system, repartition the drive, restore
your old software, and then install Linux, which can be a time-consuming and error-prone
process.
Some alternatives to repartitioning your hard drive do exist. You can share space with
Linux and DOS, or you can use a program that repartitions your drive without erasing
files. These alternatives do work, but you still face the possibility of losing data while
installing the system. Also, by repartitioning, you gain improved performance and
better control over the amount of disk space used for Linux.
The amount of disk space you need to run Linux depends on the various applications
you plan to install. You should have at least 700MB free on the drive where you want
to install Linux, in addition to the programs and data you want to keep from your
Overcoming the Disadvantages 23
other operating systems. If you have 1.6GB free, you should have more than enough
space for a full installation of Linux. Part
I
Ch
1
Lack of Experience
Finally, unless already a UNIX guru, you must learn how to manage a Linux system. Unlike
DOS, Windows, and OS/2, Linux and UNIX need to be managed. The manager, usually
called the system administrator, or sys admin, is responsible for maintaining the system. The
system administrator is responsible for performing such duties as adding and deleting user
accounts, backing up the system on a regular basis, installing new software, configuring the
system, and fixing things when they go wrong (which happens even on commercial versions of
UNIX in use every day). Because UNIX doesn't run perfectly 100 percent of the time, the
system administrator must maintain the system. This need for maintenance presents a great
opportunity for you to learn how to be a system administrator on a UNIX system.
Tip #3 from
Jack There are many training programs now available, so if your company is leery of using
Linux, just show them support and training is available (and that you of course would
be the perfect person to attend the training).
See the following Web sites for more information:
http://www.redhat.com/products/training.html
http://db.globalknowledge.com/catalog/course.asp?course=6900
? See ``Understanding Centralized-Processing Systems,'' p. 186
Overcoming the Disadvantages
At first, you might think that using Linux puts you all alone in the world, making you survive
by yourself. This notion is partially true because Linux started life as a hacker's system, and
hackers like to tinker and fix systems themselves. But today, because the popularity of Linux
has grown, many sources of help are available.
Thousands of pages of documentation are provided with most distributions of Linux. You can
find this information in the /doc directory on the accompanying CD-ROM.
In addition, several magazines are devoted to Linux, and you can find plenty of online sources
of information and online users willing to help with your questions. If you work for a
commercial entity and need a professional contractor, these contractors also are available.
After you install Linux, you can also find a wealth of online help providing information on
almost every Linux command and program available. Check out Appendix A, ``Sources of
Information,'' to see that you're not alone.
24 Chapter 1 Understanding Linux
Disappearing Disadvantages
Although all the disadvantages discussed in the preceding sections still exist, many are slowly
disappearing as new companies come into existence to build on Linux and offer new
solutions. Two such companies are Red Hat and Caldera. We chose Red Hat as the primary
distribution for this book because of its ease of use and installation. Caldera also uses the Red
Hat package management system for installation and its own programs, LISA and Lizard, to
install the distribution. Both Red Hat and Caldera provide online, fax, and email-based
technical support for their products and their versions of Linux.
The Commercial Side of Linux
Linux has become an enterprise-ready operating system. Many companies use Linux as an
inexpensive Web server for their intranets. Linux is also used for various network applications
(such as DNS), for routing, and as firewalls. Also, many Internet service providers (ISPs) use
Linux as their main operating system.
Many commercial programs are also available for Linux, overview of; you can check them out
in the ``Commercial How-To.'' Other organizations, such as NASA and Digital Domain, use
Linux to render various images, including high-resolution planetary images (NASA) or
realistic special effects for movies such as Titanic (Digital Domain).
Tip #4 from
Jack After installing Linux, you can find the commercial How-To in the directory
/usr/doc/HOW_TO/.
Before installation, you can find the file in the following directories:
. cdrom-drive letter >:doc/HOWTO/other-formats/html/
Red Hat: <
Commercial-HOWTO.html
. Caldera: < cdrom-drive letter >:col/doc/HOWTO/other-formats/
html/Commercial-HOWTO.html
. Debian: not available from the CD-ROM.
. On the Web at http://www.linux.org/help/howto.html.
Commercial Programs from Red Hat
Although Red Hat released one of the most popular distributions of Linux, the company also
has produced several commercial programs. Red Hat has also created a Linux package
manager, called RPM (Red Hat Package Manager), which it released under the General
Public License (GPL) for other distributions to use.
A Brief History of Linux 25
Along with its GPL versions of Linux and RPM, Red Hat also provides an application
Part
framework called Applixware, which contains a word processor, a spreadsheet program, a
presentation graphics program, a mail tool, and various development tools. Red Hat also I
provides a commercial version of Motif for developing and running X under Linux.
Ch
1
Commercial Programs from Caldera
Caldera originally provided a networking-based distribution based on Red Hat and
technology from Novell, where many of Caldera's principals previously worked. Their
second-generation product, Caldera OpenLinux Base, is a low-cost UNIX-like operating
system based on the Linux 2.2 kernel and the OpenLinux distribution from Caldera. It
includes a graphical user interface capable of managing system and networked resources,
including client and server interaction with the Internet and all major networking systems.
The menu-driven installation is provided in multiple languages. Caldera OpenLinux Base is a
non-dedicated gateway and includes all Internet client, server, and router protocols and
services. OpenLinux Base also includes a commercial X server from MetroLink and a fully
licensed Linux version of the Netscape Navigator.
Caldera also provides Corel's WordPerfect for Linux, as well as an Internet office suite
containing a bundle of complete business applications. These commercial programs, as well
as dozens more, are available from Caldera.
Caldera also licensed and ported Sunsoft's WABI technology to allow end users to run the
most popular Windows 3.1 applications on Linux-based system software.
Commercial Programs from Debian
As for Debian, Corel recently selected the Debian distribution for their upcoming KDE/
WordPerfect offering.
A Brief History of Linux
The history of Linux is tied to the history of UNIX and, to a lesser extent, a program called
Minix. Minix was an operating system tutorial written by the well-known and respected
computer scientist Andrew Tannebaum. This operating system became popular on several
PC platforms, including MS-DOS-based PCs. But more on Minix laterÐfirst, a brief history
of UNIX.
Although AT&T created the UNIX operating system, many other companies and individuals
have tried to improve the basic idea over the years. The following sections examine a few of
the leading variants in use today.
AT&T
Ken Thompson (a computer programmer for AT&T Bell Laboratories) and a group of
people working under Ken's direction developed an operating system that was flexible and
completely compatible with programmers' varied needs. Legend tells that Ken, who had been
26 Chapter 1 Understanding Linux
using the MULTICS operating system, dubbed this new product UNIX as he joked with others on
his development team. He was lampooning the MULTICS multiuser operating system: UNIX
was derived from uni, meaning one or single, followed by the homophone X. Perhaps the greater
joke in this bit of folklore lies in the fact that MULTICS is remembered by few users today as a
viable multiuser operating system, whereas UNIX has become the de facto industry standard for
multiuser multitasking operating systems.
BSD
Berkeley Software Distribution (BSD), University of California at Berkeley, released its first
version of UNIX, based on AT&T's Version 7, in 1978. BSD UNIX, as it's known
throughout the industry, contained enhancements developed by the academic community at
Berkeley that were designed to make UNIX more user-friendly. The user-friendly
``improvements'' in BSD UNIX were an attempt to make UNIX appeal to casual users in
addition to the advanced programmers who liked its flexibility in conforming to their
changing demands. Despite being less than 100 percent compatible with AT&T's original
UNIX, BSD UNIX did accomplish its goals: The added features enticed casual users to use
UNIX.
BSD has become the academic UNIX standard. The original creators of BSD have since
released a version for the Intel platform called, appropriately enough, BSD. This version too
has a limited distribution on the Internet and via CD-ROM vendors. The authors also wrote
several articles a few years ago in the computer magazine Dr. Dobb's Journal, detailing the
design and implementation of BSD386 or FreeBSD. Today, BSDI, the commercial version of
FreeBSD, is another popular operating system similar to Linux.
USL
UNIX System Laboratories (USL) was an AT&T spin-off company that had been developing
the UNIX operating system since the early 1980s. Before Novell purchased it in 1993, USL
produced the source code for all UNIX System V derivatives in the industry. However, USL
itself didn't sell a shrink-wrapped product at that time.
USL's last release of UNIX was UNIX System V Release 4.2 (SVR4.2). SVR4.2 marked
USL's first entry into the off-the-shelf UNIX marketplace. In a joint venture with Novell,
which temporarily created a company called Univel, USL produced a shrink-wrapped version
of SVR4.2 called UnixWare. With Novell's purchase of USL, Novell shifted the focus of
USL from source-code producer to UnixWare producer. Novell has now sold its version of
UNIX to the Santa Cruz Operation (SCO).
Recently, SCO made a free single-user license available to the public for using SCO UNIX.
The program costs $19 for the distribution media, not unlike Linux. However, whereas SCO
provides a copy of its operating system, it doesn't provide the source code. Some in the Linux
community suspect Linux is giving the UNIX communityÐor at least the SCO
communityÐsome stiff competition.
A Brief History of Linux 27
XENIX, SunOS, and AIX Part
Microsoft developed its UNIX version, XENIX, in the late 1970s and early 1980s, during the
I
peak of the PC revolution. Processing power available in PCs began to rival that of existing
minicomputers. With the advent of Intel's 80386 microprocessor, it soon became evident that Ch
XENIX, which had been developed specifically for PCs, was no longer necessary. Microsoft 1
and AT&T merged XENIX and UNIX into a single operating system called System V/386
Release 3.2, which can operate on practically any common hardware configuration. XENIX is
still available today from Santa Cruz Operation (SCO), a co-developer with Microsoft, whose
efforts to promote XENIX in the PC market have made this version of UNIX one of the most
commercially successful.
Sun Microsystems has contributed greatly to UNIX marketability by promoting SunOS and
its associated workstations. Sun's work with UNIX produced a version based on BSD.
Interestingly enough, AT&T's SVR4 is compatible with BSD, tooÐno doubt an offshoot of
AT&T and Sun Microsystems' collaboration in UNIX System V Release 4.0.
IBM's venture into the world of UNIX yielded a product called Advanced Interactive
Executive (AIX). Although AIX isn't as well known as some other UNIX versions, AIX
performs well and has no problem holding its share of the operating system market. It's
perhaps the old belief that any UNIX version is an unfriendly, unforgiving operating system
that has kept AIX from gaining a better market reception.
Linux
As you learned earlier in this chapter, Linux is the brainchild of a computer science student
named Linus Torvalds. Linux began life as a hobby project in 1991 for Linus, who was then
23. He hoped to create a more robust version of UNIX for Minix users. Minix, as mentioned
earlier, is a program developed by computer science professor Andrew Tannebaum.
The Minix system was written to demonstrate several computer science concepts found in
operating systems. Torvalds incorporated these concepts into a standalone system that
mimics UNIX. The program was widely available to computer science students all over the
world and soon generated a wide following, including its own Usenet newsgroups. Linus
Torvalds set out to provide his fellow Minix users with a better platform that could run on the
widely available IBM PC. Linus targeted the emerging 386-based computers because of the
task-switching properties of the 80386 protected-mode interface.
What follows are some of the statements Linus made when announcing his Linux program:
``After that it was plain sailing: hairy coding still, but I had some devices, and
debugging was easier. I started using C at this stage, and it certainly speeds up
development. This is also when I started to get serious about my megalomaniac ideas
to make 'a better Minix than Minix.' I was hoping I'd be able to recompile gcc under
Linux some day...
``Two months for basic setup, but then only slightly longer until I had a disk driver
(seriously buggy, but it happened to work on my machine) and a small file system.
28 Chapter 1 Understanding Linux
That was about when I made 0.01 available [around late August of 1991]: it wasn't pretty, it
had no floppy driver, and it couldn't do much [of] anything. I don't think anybody ever
compiled that version. But by then I was hooked, and didn't want to stop until I could
chuck out Minix.''
These announcements are from the ``Linux Installation and Getting Started Guide,'' by
Matt Welsh (copyright 1992-94 by Matt Welsh, 205 Gray Street NE, Wilson, NC 27893,
[email protected]). They're used subject to section 3 of Matt's copyright.
You can obtain the complete ``Linux Installation and Getting Started Guide'' from the
Linux Documentation Project's various archives sites. You can find this book on
metalab.unc.edu in the directory /pub/Linux/docs/LDP/install-guide. For
information on how to access archives and download files, refer to Chapter 33,
``Surfing the Internet with the World Wide Web.''
In a later announcement, made in the comp.os.minix user group on October 5, 1991, Linus
introduced to the world Linux version 0.02, the first official version of Linux:
``Do you pine for the nice days of Minix 1.1, when men were men and wrote their own
device drivers? Are you without a nice project and just dying to cut your teeth on an
OS you can try to modify for your needs? Are you finding it frustrating when
everything works on Minix? No more all-nighters to get a nifty program working?
Then this post might be just for you.''
``As I mentioned a month ago, I'm working on a free version of a Minix look-alike for
AT-386 computers. It has finally reached the stage where it's even usable (though [it]
may not be depending on what you want), and I am willing to put out the sources for
wider distribution. It's just version 0.02, but I've successfully run bash, gcc, gnu-make,
gnu-sed, compress, and so forth under it.''
Who Owns Linux?
IBM owns the rights to OS/2, and Microsoft owns the rights to MS-DOS and MS Windows,
but who owns the rights to Linux? First and foremost, Linux isn't public domain software;
various components of Linux are copyrighted by many people. Linus Torvalds holds the
copyright to the basic Linux kernel. Red Hat, Inc. owns the rights to the Red Hat distribution
version. Caldera holds the rights to the Caldera OpenLinux product. The Debian
distribution is a creature of the Open Systems movement with Richard Stallman's group
embacing it as the first of many kernels as part of the GNU Hurd project, but single
corporation controls Debian. Many Linux utilities are under the GNU General Public
License (GPL). In fact, Linus and most Linux contributors have also placed their work under
the protection of the GNU GPL. You can find the license on each of the accompanying
CD-ROMs in the root directory in the file named copying.
Who Owns Linux? 29
This license is sometimes referred to as the GNU copyleft (a play on the word copyright). This
Part
license covers all the software produced by GNU (itself a play on wordsÐGNU's Not
UNIX) and the Free Software Foundation. The license allows programmers to create I
software for everyone. The basic premise behind GNU is that software should be available to
Ch
everyone, and that if someone wants to modify the program to his or her own ends, that
1
should be possible. The only caveat is that the modified code can't be restricted; others must
also have the right to the new code.
? See ``How to Apply These Terms to Your New Programs,'' p. 868
The GNU copyleft, or GPL, allows a program's creators to keep their legal copyright but
allows others to take, modify, and sell the resulting new program. However, in doing so, the
original programmers can't restrict any of these same rights to modify the program from the
people buying the software. If you sell the program as is or in a modified form, you must
provide the source code. That's why Linux comes with the complete License source code.
CHAPTER
2
Linux Installation Overview
In this chapter by Jack Tackett, Jr.
Understanding Linux's Hardware Requirements 32
Compiling Necessary Information 39
Starting the Installation Process 39
Understanding the Various Installation Methods 40
Partitioning Your Hard Drive 44
Maneuvering Through Linux 45
Managing Users 47
Using Basic Commands 48
Dealing with DOS Files Under Linux 52
Shutting Down Linux 54
Troubleshooting 55
32 Chapter 2 Linux Installation Overview
Understanding Linux's Hardware Requirements
This chapter gives you a basic overview of the information you need to install, configure, and
use almost any Linux distribution.
This book assumes that you have a working knowledge of DOS and of such tasks as
formatting your hard drive, working with partition tables, and determining sector
sizes. If this information sounds like a foreign language, check out Using MS-DOS 6.2,
Special Edition, or ask a computer guru buddy to help you through this information.
You're about to make major changes to your system, so be careful. It's a good idea to
have paper and pen nearby to take notes just in case something does go wrong;
besides, you'll need to jot down some numbers along the way.
To be able to install Linux successfully, you need supported hardware. Choosing the right
level of hardware for your Linux system depends on such factors as the number of users to be
supported and the types of applications to be run. All this information translates into
requirements for working memory, hard disk storage space, the types of terminals needed,
and so forth.
ON THE WEB
For current information on supported hardware, see the following Web sites.
For the Red Hat distribution, see
http://www.redhat.com/corp/support/hardware/intel/60/rh6.0-hcl-i.ld-1.html
For Caldera, see
http://www.calderasystems.com/products/openlinux/hardware.html
For Debian, see
http://www.debian.org/releases/slink/i386/
ch-hardware-req.en.html#s-hardware-supported
Most Linux systems today consist of PCs. These Linux installations are often for only a single
user, although they may also be tied into larger Linux or UNIX systems.
Linux is a constantly evolving system, and hardware support is sporadically updated.
The distributions on the accompanying CD-ROMs are relatively stable, but new
hardware support may have been provided by the time this book is printed and the
CD-ROM is created. Although many hardware components have clones or compatible
replacements, not all may work with Linux. If you do have the hardware described in
this chapter, the odds are excellent that Linux installs, boots, and operates properly. If
Understanding Linux's Hardware Requirements 33
you don't have the equipment listed, Linux may or may not operate properly, but the
odds are against you getting a system up and running.
If you're using a version of Linux in a single-user configuration (the most likely
configuration), you're the system administrator, also known as the superuser. It's your
responsibility to understand the system well enough to perform the administrative duties Part
required to keep it operating at an optimum level. These duties include keeping enough space
I
on the hard drive, backing up regularly, ensuring that all devices attached to the system have
the proper software drivers, installing and configuring software, and so forth. Ch
2
Tip #5 from
Jack As the system administrator, or superuser, you wield immense power on your system.
It is best if you log in as a regular user for your day-to-day work and log in as the
superuser, whose login name is root, only when you need to administer the system.
The level of hardware you need depends heavily on the hardware used by the myriad people
who programmed the Linux system. Unlike commercial software developers who can afford
to test their systems on many different hardware configurations, Linux developers typically
have access only to their personal computers. Luckily, because so many Linux developers
exist, most of the standard hardware found in the PC world is supported.
The System's CPU
A basic system requires an IBM-compatible PC with an Intel 80386 or later CPU in any of
the various CPU types and Intel's various Pentium processors. Other CPU clones, such as the
80686 clone chips made by Cyrix and Advanced Micro Devices (AMD), are also compatible
with Linux.
The Linux kernel has also been ported to other processors. Among those now supported are
the DEC Alpha, the PowerPC (Macintoshes), Sun Sparcs, and even embedded systems
processors such as those used in Corel's Network PC running the Debian distribution.
The System's Bus
The type of bus used to communicate with the peripherals is also important. Linux works
with only the ISA, EISA, and PCI buses. The MicroChannel Architecture (MCA) bus used on
IBM's PS/2 isn't supported, although a port is in process. Some newer systems use a faster
bus, called the local bus, for disk access and video displays, for example. Linux does support the
VESA Local Bus but might not support a non-VESA Local Bus architecture.
34 Chapter 2 Linux Installation Overview
Memory Needs
Linux requires surprisingly little RAM to run, especially when compared to comparable
operating systems such as OS/2 and Windows NT. Linux requires at least 16MB of RAM,
although 64MB is highly recommended. If you have less than 32MB of RAM, you need to use
what's called a swap file. The basic rule of thumb is that the more memory your system
contains, the faster your system runs.
The next memory consideration for Linux is the use of the X Window System clone called
XFree86. XFree86 is a version of the X Window System that can be freely distributed and is
included with Linux for that reason. XFree86 is a GUI similar to Microsoft Windows.
? See ``Installing the XFree86 System,'' p. 508
Disk Drives and Space Requirements
Although you can run Linux from a floppy-drive-only system, running Linux from your
system's floppy drive isn't recommended.
You can boot Linux from a floppy drive. Booting a system refers to the process of
starting a computer system and loading the operating system into memory to start the
system. The term is derived from the phrase bootstrapping.
For a home-based system, you no longer need a floppy drive if you have a bootable CD-ROM
(check your BIOS to be sure).
For better system performance, you should install Linux on a hard drive. Linux supports all
IDE-enhanced drives. Linux also supports a wide range of SCSI hard-drive controllers. If
your controller is a true SCSI controllerÐthat is, not a proprietary version of SCSIÐLinux
can use your controller. Linux now supports SCSI controllers from Adaptec, Future Domain,
Seagate, UltraStor, the SCSI adapter on the ProAudio Spectrum 16 card, and Western
Digital. The following card types are supported:
Adaptec 152x/1542/1740/ Always IN2000
274x/284x/294X
Buslogic Pro Audio Spectrum 16
EATA-DMA (DPT, NEC, AT&T) Qlogic
Seagate ST-02 Trantor T128/T128F/T228
Future Domain TMC-8xx, 16xx UltraStor
Generic NCR5380 7000FASST
NCR 53c7, 8xx
If you have the proper drive controller, you must worry about disk space requirements. Linux
supports multiple hard drives and can be installed across drives. Unlike other operating
Understanding Linux's Hardware Requirements 35
systems, Linux doesn't need to be installed on the same hard drive; pieces can be installed on
different drives.
Unless you're installing Linux onto a brand-new hard drive, you need to repartition
and reformat the drive. This process destroys all information currently stored on that
portion of the drive. Thus, backing up your filesÐtwiceÐbefore installing Linux is
imperative. If space permits, you can split a single hard drive into multiple partitions
Part
and copy your files back to one of the partitions.
I
Ch
The amount of disk space required depends on the software you install and the amount of
2
data you expect that software will generate. Linux requires less disk space than most
implementations of UNIX systems. You can run a completely functional Linux system,
without X Window System support, in 600MB. For a complete installation of everything in
the distribution, more than 1.6 GBs (gigabytes) is recommended.
Swap Space
Finally, as mentioned earlier in the section ``Memory Needs,'' if you have limited RAM, you
need swap space. Whereas systems such as Microsoft Windows create a swap file that resides
on your hard drive as any other file, Linux allows the swap file to reside on a separate swap
partition. Most Linux installations use partitions rather than files. Because you can place
multiple partitions on the same physical hard drive, you can place the swap partition on the
same drive as Linux, but for better performance, you should place the swap partition on a
separate drive.
Linux allows up to eight swap partitions that can be no larger than 16MB. A rule of thumb is
to set the swap file size to twice the amount of physical RAM contained on your system.
Thus, if you have 8MB of physical RAM, your swap partition should be 16MB in size.
Tip #6 from
Jack Use the df command to see how much swap space, or amount of any disk space, you
have available. For example, to display the amount of space available in megabytes
use:
df -m
Monitor Requirements
For text-based terminals, Linux supports all standard Hercules, CGA, EGA, VGA, and
SuperVGA video cards and monitors. To take advantage of the color-coding directory
listings available with Linux, you need a color monitor. So for text-based operation, any
video/controller combination should work.
36 Chapter 2 Linux Installation Overview
The big problems occur when you run the X Window System distributed with Linux. To use
XFree86, you need a video adapter that uses one of the chipsets listed in Table 2.1. Chipsets
are a group of integrated circuits, or computer chips, used to take information from the
computer and convert the data into a form that can be displayed on a video monitor. To find
the chipset used by your video adapter, you can check the documentation included with your
card to determine whether it has any problems using XFree86.
Table 2.1 Video Chipsets Supported by Linux
Manufacturer Chipset(s)
Tseng ET3000, ET40000AX, ET4000/W32
Western Digital WD90C00, WD90C10, WD90C11, WD90C24, WD90C30,
WD90C31
Trident TVGA8800CS, TVGA8900B, TVGA8900C, TVGA8900CL,
TVGA9000, TVGA9000i, TVGA9100B, TVGA9200CX,
TVGA9320, TVGA9400CX, TVGA9420
ATI 28800-4, 28800-5, 28800-a
NCR 77C22, 77C22E, 77C22E+
Cirrus Logic CLGD5420, CLGD5422, CLGD5424, CLGD5426,
CLGD5428, CLGD6205, CLGD6215, CLGD6225,
CLGD6235
OAK OTI067, OTI077
S3 86C911, 86C924, 86C801, 86C805, 86C805i, 86C928
Compaq AVGA
Western Digital/ Paradise PVGA1
The release notes for the current version of XFree86 distributed with Linux should
contain a more recent list of supported and nonsupported chipsets.
Some problems encountered by the XFree86 developers are caused by adapter manufacturers
who don't provide the necessary information on programming the cards to display
information. Without this information, developers can't support the X Window System on
those adapters. Also, some manufacturers provide this information but require either royalty
payments or nondisclosure agreements for others to use the information. These types of
restrictions make it impossible to support these adapters on a freely distributed system like
the XFree86 system for Linux.
Understanding Linux's Hardware Requirements 37
CD-ROMs
To install the Linux system included on the accompanying CD-ROMs, you must have a
CD-ROM drive supported by Linux. Because most CD-ROMs use a SCSI interface
controller, any SCSI controller listed earlier in the section ``Disk Drives and Space
Requirements'' should also work with a CD-ROM attached to the controller. Linux also now
supports many of the new EIDE and ATAPI CD-ROMs available on the market.
Many of the CD-ROMs included with multimedia packages may or may not work with
Part
Linux, depending on whether the controller is a true SCSI adapter or a proprietary adapter. I
Proprietary adapters for the most part don't work. However, Linux does specifically support
Ch
the Creative Labs SoundBlaster line of CD-ROMs and provides a specific installation
configuration for those CD-ROMs. Other CD-ROMs known to work with Linux include the
2
following:
NEC CDR-74 Okano
Sony CDU-541 Wearnes CD with interface card
Sony CDU-31a or 33a
Plextor DM-3024 Most IDE/ATAPI CD-ROMs
Aztech Mitsumi CD-ROMs
Orchid SoundBlaster, Panasonic Kotobuki, Matsushita,
TEAC-55a, or Lasermate
Network Access
You can connect a Linux system to the world in several ways, but the two most popular (and
supported) methods are via network controller cards and modems. Network controller cards
include Token Ring, FDDI, TAXI, and Ethernet cards. Most common business networks use
an Ethernet controller card. Table 2.2 lists several of the Ethernet adapters supported.
Table 2.2 Ethernet Controller Cards Supported by Linux
Manufacturer Interface Card
3Com 3c503, 3c503/16, 3c509
Novell NE1000, NE2000
Western Digital WD8003, WD8013
Hewlett-Packard HP27245, HP27247, HP27250
At home, you'll more than likely connect to the outside world via a modem and a
communications protocol such as SLIP or PPP. Linux supports almost every type of modem
on the market, internal and external. If you can access the modem from MS-DOS, you'll have
no problem accessing the modem from Linux.
? See ``Understanding the Requirements for SLIP and PPP,'' p. 664
38 Chapter 2 Linux Installation Overview
Miscellaneous Hardware
The following sections list miscellaneous hardware supported by Linux, such as mouse
devices, tape drives, and printers. Although such hardware makes Linux easier to use and
more robust, it isn't required.
Mouse Devices
Using text-based Linux doesn't require a mouse. Unlike many UNIX implementations,
however, Linux does allow you to cut text from any area of the screen and paste it to the
command line by using a mouse. If you intend to use the X Window System clone, XFree86,
you must use a mouse.
Linux supports most serial mouse devices, including the following:
& Logitech
& MM series
& Mouseman
& Microsoft
& Mouse Systems
Linux also supports the Microsoft, Logitech, ATIXL, and PS/2 bus mouse devices. In fact,
any pointing devices, such as trackballs and touch screens, that emulate the previously listed
mouse devices should work with Linux.
Tape Drives
Tape drives provide a great deal of storage space for backing up your computer system. Linux
supports several SCSI-based tape systems, as shown in Table 2.3. Linux also supports the
popular Colorado Memory Systems tape drives (120 and 250 versions), which are plugged
into a system's floppy-disk controller. The versions that plug into the printer port aren't now
supported. Most drives supporting QIC-02 should also work with Linux.
Table 2.3 Tape-Backup Drives Supported by Linux
Manufacturer Model
Exabyte All SCSI models
Sanko CP150SE
Tandberg 3600
Wangtek 5525ES, 5150ES, 5099EN
Printers
Linux supports the complete range of parallel printers. Configuring Linux to support serial
printers is tedious and error-prone. Serial printer support isn't well documented or supported
Starting the Installation Process 39
by the basic Linux installation programs. If you have a serial printer, you might have problems
using it under Linux. If you have a parallel printer, your biggest problem is most likely the stair-step
effect, illustrated here:
This is line one.
This is line two.
This is line three.
How UNIX and Linux treat carriage returns and linefeeds produces the stair-step effect.
Under most UNIX systems, the command to move the paper down one line (linefeed) and
Part
then position the print head at the beginning of the line (carriage return) is represented by I
one control character. Under systems such as MS-DOS and Windows, however, each
Ch
command is represented by a different control character. When you print a UNIX file under
a printer configured for MS-DOS systems, you see the stair-step effect because the file
2
contains only the linefeed control character and not the carriage-return control character.
? See ``Knowing What You Need to Configure Printers,'' p. 392
Compiling Necessary Information
Before starting the installation, you need the following information about your system:
& The type of video card, chipset, and monitor used
& The serial port used by your mouse
& The serial port used by your modem
& The network information for your computer, if it's connected to a network (items
such as its IP address, gateway, and domain name)
& The type of hard drive and CD-ROM drive in your system and their controller types
& The name you intend to call your system
& A password for the system administrator's root account
If you're connecting to the Internet, you can get most of this information from either your
network administrator or your Internet service provider (ISP).
Starting the Installation Process
After compiling the information, you need to decide on the media you intend to use to install
Linux. Traditionally, you would use a program called rawrite to create boot and root floppies.
Today, if your system supports bootable CD-ROMs, you can boot from the CD. Caldera
allows you to install from the CD while you're running Windows (but you need the
unpartitioned space to do so). You can also install via a network connection using File
Transfer Protocol (FTP), Hypertext Transfer Protocol (HTTP), and Network File System
(NFS). The subsequent chapters guide you through each process.
40 Chapter 2 Linux Installation Overview
You can also run part of the Linux file system from the CD-ROM without installing all
the software. You can choose to do so during installation.
Understanding the Various Installation Methods
If you're reading this book, you will probably install Red Hat Linux from the accompanying
CD-ROM. However, you can use one of the following four methods to install Red Hat: from
CD-ROM, via NFS, via FTP, or from a hard drive.
To install Linux directly from a CD-ROM, you need access to DOS. From the DOS prompt,
execute the command
[cdrom-drive]:\dosutils\autoboot
where [ cdrom-drive] is the drive letter for your system's CD-ROM.
This method erases your hard drive. Be sure to back up any files you fear losing.
If you have another partition available, you can install Linux to coexist with your system
without erasing what's already there. To do so, you need the CD-ROM, an empty partition,
and a boot disk. You'll learn later in this chapter how to create the boot disk, as well as how to
repartition your hard drive.
NFS (Network File System) provides a way to install Red Hat across a network. First, you
must mount the CD-ROM drive on a machine supporting the ISO-9660 file system with
RockRidge extensions and then export the file system via NFS. You need to know the path to
the exported file system and the IP number or, if DNS is configured, the name of the system.
FTP (File Transfer Protocol) is a method for transferring files across the Internet. (Chapter
32, ``Accessing the Network with telnet, ftp, and the r- Commands,'' explains FTP in more
detail.) To install via FTP requires a boot disk and the supplemental disk described later in
this chapter.
Installing Red Hat from a hard drive requires the same boot and supplemental disks used for
an FTP installation. First, you create a directory named RedHat. Then you copy the
corresponding directory from the CD-ROM, and all the subdirectories there, to the RedHat
directory. You can use the following DOS commands to do so:
cd \RedHat
xcopy /s e:\RedHat
The cd command assumes that you're already on the installation hard drive; the xcopy
command assumes that your CD-ROM drive is drive E.
Understanding the Various Installation Methods 41
No matter which method you use, you need at least the boot disk to proceed with installation.
But first, you should gather the information described earlier in this chapter.
You can install Caldera's OpenLinux directly from the CD-ROM, even if you're running
Windows. The OpenLinux CD provides a multimedia-based setup program. You simply
double-click the Setup icon, shown in Figure 2.1, in the winsetup folder on the CD-ROM.
Figure 2.1
Part
You can use
Caldera's I
impressive
multimedia
Ch
installation 2
program from
Windows.
Debian installs much like Red Hat: You either can boot from the CD-ROM or use
installation diskettes you create.
If you intend to use other operating systems on the same computer (such as Windows 95,
Windows 98, Windows NT, or OS/2), you need to create the necessary partitions for these
operating systems. Typically, you need to use the operating system's partitioning software
because Linux can't handle these other partition types.
ON THE WEB
A product named System Commander, from V Communications, lets you install and switch between 32
different operating systems. You can find more information about this product at the following site:
http://www.v-com.com/
Next, you should check for any last-minute changes to your favorite distribution. The reasons
for checking are many, but the two major reasons are that Linux is constantly updated, and
42 Chapter 2 Linux Installation Overview
this chapter is being written at least a month before the CD-ROM is cut. In the interim, new
material or bug fixes may have been released.
ON THE WEB
You can also check for updated material on the Web at these sites:
http://www.redhat.com/errata
http://www.calderasystems.com/doc/openlinux/errata.html
http://www.debian.org/releases/slink/#errata
Table 2.4 lists the currently available update packages that fix known problems in the Red
Hat 6.0 distribution.
? See ``Installing Packages with RPM,'' p. 169
Table 2.4 Red Hat Errata Listing
Date Released Package
29-Jul-1999 squid
29-Jul-1999 gnome
29-Jul-1999 samba
23-Jul-1999 enlightenment
23-Jul-1999 gnumeric
09-Jul-1999 rdist
07-Jul-1999 rpm
07-Jul-1999 netkit-base, traceroute
24-Jun-1999 net-tools
24-Jun-1999 talk
21-Jun-1999 KDE
21-Jun-1999 mod_php3
17-Jun-1999 XFree86
16-Jun-1999 dev, rxvt, screen
10-Jun-1999 wu-ftpd
10-Jun-1999 utempter
03-Jun-1999 kernel update
29-May-1999 New Boot Images
27-May-1999 Sparc kernel and egcs
25-May-1999 Netscape
22-May-1999 INN
14-May-1999 Applications
Understanding the Various Installation Methods 43
Table 2.4 Red Hat Errata Listing
Date Released Package
13-May-1999 apmd
12-May-1999 pump
11-May-1999 xscreensaver
Part
/pub/OpenLinux/updates/README
For information on Caldera, see the file at
I
ftp://ftp.caldera.com/pub/OpenLinux/updates/.
Ch
These directories contain various updated boot disk images and update packages in RPM
2
format for the releases of Caldera OpenLinux 1.3:
kdelibs-1.1-2.i386.rpm
kdelibs-devel-1.1-2.i386.rpm
kdelibs-1.1-2.src.rpm
dosemu-0.98.5-1.i386.rpm
dosemu-0.98.5-1.src.rpm
opendos-hdimage-7.02-4.i386.rpm
opendos-hdimage-7.02-4.src.rpm
If you're not installing directly from the CD-ROM, you next need to repartition your current
hard drive to make room for Linux. This step may cause problems because repartitioning a
hard drive destroys any data contained on the affected partitions. After making room for
Linux, you need to boot the Linux system and create its new partitions and file systems.
Typically, Linux systems need a primary partition to store the files on and a swap file
partition, especially if you have a machine used in text mode with 8MB or less of memory.
A file system is basically a section of your hard drive specially formatted to hold
certain types of files. UNIX and Linux use file systems to represent entire sections of
the directory tree. This use is in contrast to MS-DOS, which places subdirectories in
the directory tree on the same logical drive. UNIX systems use the directory tree
format because placing subdirectories on different drives is safer. If one drive
malfunctions, only the information on that drive needs to be replaced or fixed.
? See ``Understanding File Systems,'' p. 440
After creating the file systems, you then can install the Linux operating system, its support
files, and various application packages distributed with the system. To install Linux, you must
first boot a stripped-down version of the operating system. You do so by creating a boot disk
and a supplemental disk set containing the stripped-down operating system.
44 Chapter 2 Linux Installation Overview
Creating the Boot and Supplemental Disks
You create the boot and supplemental disks by using the rawrite program. You can find this
program on the accompanying CD-ROM in the /dosutils subdirectory. For this step, you
need two formatted floppies: one labeled boot and the other labeled supp. Place the boot disk
in drive A and enter the following:
E:\dosutils>rawrite
Enter disk image source file name: e:\images\boot.img
Enter target diskette drive: A:
Please insert a formatted diskette into drive A: and press -ENTER-
If you want to abort the process, simply press Ctrl+c to stop. If rawrite fails, try a new
formatted disk. If the problem persists, you should have your hardware checked for possible
problems.
Partitioning Your Hard Drive
After you back up your system and make the necessary boot and supplemental disks, you must
prepare your system's hard drive for Linux.
This process is the most dangerous because maximum data loss is assured. If you
haven't backed up your system, do so now. Although you can use an experimental
program called FIPS and commercial programs such as Partition Magic to do
nondestructive repartitioning, a full backup is recommended, just in case problems
occur.
Red Hat provides a program called Disk Druid to partition your hard drive. For Caldera and
Debian, you use a program called fdisk (Debian also provides a program called cfdisk). Just
like its DOS counterpart, fdisk allows you to create partitions on your hard drive.
You should plan your disk partitions now and also plan on how much disk space you want to
allocate to each partition because this portion of the installation provides the greatest source
of problems, especially if you run out of disk space. The following partitions and sizes are a
good starting point:
swap 64MB
/ 128+MB
boot 16MB
usr 600+MB (if you install everything!)
var 256MB
home All available space
Maneuvering Through Linux 45
Installing the System
After you create the appropriate partitions, each distribution requires you to select which
components to install. If you program for a living, you should install the programming
components. Linux provides compilers and interpreters for a variety of languages, such as C,
C++, Java, Perl, and FORTRAN. If you like games, you should install the X Window System
and games components. After you select the components, the install program proceeds with
copying the necessary files from the source media (typically the CD-ROM or the network) to Part
your system. This process can take some timeÐeven when you're installing from the
I
CD-ROMÐso plan to wait awhile.
Ch
2
Tip #7 from
Jack During installation, you can monitor what the system does via virtual terminals.
Maneuvering Through Linux
After installing Linux and rebooting, you're faced with a system prompt based on the name
you gave your system during installation. Under Red Hat, the prompt looks similar to the
following:
Red Hat Linux release 6.0 (Hedwig)
Kernel 2.2.5-15 on an I686
web login:
The prompt might indicate a different version of Linux, however, depending on the kernel
version, machine name, and distribution.
You must now supply a username and a password. A username identifies you to the operating
system because Linux can support many different users, both at different times and
concurrently. An account also provides each user with a default directory, called the home
directory. Many accounts are also set up to restrict users to certain directories on the system
and to prevent them from using certain commands, primarily to protect the files of one user
from the prying eyes of another.
Entering Commands
You enter commands in Linux much as you do in DOS and other command-line-oriented
operating systems. Linux, like UNIX, is case sensitive; if Linux doesn't know a command,
make sure that you've spelled it correctly and that you've entered it in the proper case. Most
commands are executed after you press Enter.
Recalling Command History
Linux also provides a history function to recall previous commands. This history is kept
across sessions, too. You can press the : key to recall previous commands and then press
46 Chapter 2 Linux Installation Overview
Return to activate that command. To get a complete listing of all the prior commands you've
entered, use the history command, as shown here:
[tackett@web~]$ history
1 clear
2 adduser
3 history
When you have the preceding history list, you can repeat the command by using the : key
and cycling through the commands until the proper one appears on the command line.
Alternatively, you can press ! (the bang character) and enter the number of the command you
want to re-execute. For example, if you want to repeat the adduser command in the preceding
list, you enter the following:
[tackett@web~]$ !2
The number of entries in the history list is user-defined in the user account's .profile
configuration file. See Chapter 16, ``Understanding Linux Shells,'' for more information on
the .profile configuration file.
Tip #8 from
Jack Linux provides many different command shells, some of which don't provide the
history functions. To switch to a different shell, simply invoke the command for that
shell, for example
sh invokes the default shellÐtypically bash for Linux installations.
csh invokes the c shell and tcsh invokes the t shell.
ksh, or pdksh, invokes the korn shell.
Making Selections
If you have a mouse with your system, and you have installed the selection program, you can
also use your mouse to copy text from other areas of your screen to the command line. To
select the text, simply move the mouse cursor (which appears as soon as you click the left
mouse button) by holding down the left mouse button as you drag the cursor across the
desired text, and then press the right mouse button to copy the text to the command line.
This technique is useful if you need to enter a long filename on the command line.
Completing Commands
Linux also offers another nice feature when you're entering commands. You can start to type
a filename and then press Tab. Linux searches the directory for a file beginning with the same
letters you've typed and completes the filename it finds. If Linux can't find a unique filename,
it beeps and completes the filename to the last common character. For example, if you want
to copy a file called todo_monday to todo_today, you type cp to at the prompt and then press
Tab. Linux beeps and fills out the command line like this:
[tackett@web~]$ cp todo_
Managing Users 47
If you now type an m and press Tab, Linux places the entire todo_monday filename on the
command line.
Managing Users
On many systems, the person responsible for maintaining the user accounts is referred to
asthe system administrator. The system administrator sets up user accounts and performs other
Part
duties. For more information on the various aspects of system administration, check out the
chapters in Part II, ``System Administration.'' On your Linux system, you're the system
I
administrator, so it's your responsibility to set up accounts for yourself, family, and friends. Ch
To add an account for yourself, you must create that account as the system administrator. 2
System administrators are also sometimes referred to as superusers because they have so much
control over the system. To begin your trek through Linux, you must first log in as the
superuser via the root account.
? See ``Using the adduser Command,'' p.251
Logging In and Out
To log in as root, you enter root at the login prompt. Linux then asks for a password.
By using a password, you prevent unauthorized users from logging in to any account. Linux
wants to make sure that the username is, in fact, the correct user. You shouldn't share your
passwords with just anyone. Linux protects the password you type by not echoingÐthat is, not
displayingÐthe letters onscreen, so make sure that you enter the correct password.
If you enter an invalid username or password, Linux gives the following error message and
starts the process over:
web login: jack
Password: password
Login incorrect
web login:
Once logged in you can enter Linux commands. You enter most commands in the same way
you enter them in DOS: You type the command with any needed parameters and press
Return.
Tip #9 from
Jack If you want to switch to another user account you can do so in a variety of ways. You
can enter the command login, which logs you out of the current account and prompts
you for a new username and password. You can use the su command to switch
users, as in
su tackett
48 Chapter 2 Linux Installation Overview
You are prompted for the password (unless you are currently the superuser) and
allowed to execute commands as that user. When done enter the exit command and
Linux returns you to the previous user account.
To log out, you enter logout. This command returns you to the login prompt. If this
command doesn't work, try the exit command.
Using Basic Commands
You need to know some basic commands to get around the system. The following sections
provide some of the commands you need to use your Linux system. Many of the ``commands''
presented in the following sections are actually utility programs that Linux uses to extend its
command set. These programs are found in the /bin, /sbin, and /usr/bin directories.
Getting Help for Commands with man
To get online help for each of the various Linux commands, you can type man. Linux then
displays, a screen at a time, any information it has on the command. If you aren't sure of what
command to use, you can try the -k parameter and enter a simple keyword that represents the
topic of interest. man then searches through its help files (called man, or manual, pages) for a
topic that contains the keyword. Linux also provides an alias for this command, called
apropos.
If you enter the command man ls, Linux provides help on the ls command, including all its
parameters. The command man -k cls provides a listing of commands that have the word cls
in the help file; the command apropos cls is the same as man -k cls.
Using Directory-Manipulation Commands
Linux provides many commands to work with directories. Like other operating systems you
may have used, Linux allows you to create, delete, and move directories, as well as display
information about the directory.
Changing the Current Working Directory with cd
Linux, like DOS and other operating systems, stores files in a tree structure called a directory.
You can specify a file via a path from the root directory, specified with the / character, to the
file itself. Thus, the configuration file for Emacs for the user jack can be exactly specified as
follows:
/home/jack/.emacs
If you're familiar with the DOS limits of eight characters for a filename and three characters
for an extension, you'll be pleasantly surprised to learn that Linux has no such limit on
filenames.
? See ``Understanding Filenames and Pathnames,'' p. 408
Using Basic Commands 49
Linux also uses the concept of a home directory, which is specified when an account is added
to the system. A user's home directory is usually specified with the ~ (tilde) character. You can
use the tilde in place of the directory name, where the user wants to copy a file from the
current directory \usr\home\jack to his or her home directory:
cp .emacs ~
To move around the Linux directory structure, you use the change directory command, cd. If
you enter cd without any parameters, Linux immediately returns you to your home directory. Part
To move from one directory to another directory, you use the cd command much as you do I
in DOSÐthat is, cd new-directory. Linux also uses the . (single dot) to represent the current
Ch
directory and the .. (double dot) to represent the parent directory. In fact, it's DOS that
emulates UNIX, not UNIX/Linux emulating DOS. 2
Be careful how you specify the directory separator. DOS uses as its directory
separator the \ (backslash) character, which Linux uses as the character for continuing
a command on another line. To separate directory names in Linux, you must use the /
(forward slash) character.
Also, although DOS doesn't mind if you fail to use spaces when specifying the. and
.. parameters, Linux does. Linux doesn't understand cd.., but it understands cd ...
Linux needs the space separating the command and the parameter.
Displaying Information About Files and Directories with ls
ls stands for list and is used by Linux to display a list of files. This command is the
counterpart to the DOS DIR command. (Linux also accepts the dir command to list files in a
directory.) Under Linux, the ls command displays all the main files in a directory in color. By
default, blue indicates directories, and green indicates executable programs. You can change
the default colors by modifying the file /etc/DIR_COLORS.
? See ``Listing Files,'' p. 420
ls takes many parameters to specify not only how to display a file but also what files to
display. The most common parameter is -la, which tells ls to display information in a long
format for every file in a directory.
The command ls -la lists all information about every file in the current directory. The
command ls .emacs lists the file .emacs, whereas ls -l .emacs lists all information about the
file .emacs.
The command options -ltarÐthat is, ls -ltarÐlist the same information as the preceding
ls command, except that the file listings are displayed in order from oldest to most recent.
50 Chapter 2 Linux Installation Overview
Creating New Directories with mkdir
Because Linux's file system is based on directories, Linux provides the mkdir command so
that users can create new ones. Unlike DOS, which has an alias for the mkdir command called
MD, Linux requires that the full mkdir command be spelled out. You must specify a name for
each new directory, as shown in the following example:
mkdir backup
Tip #10 from
Jack Linux does provide a way, via the command shell, to make aliases for command
names; thus, if you simply can't live without the DOS MD command and hate typing
mkdir, you can alias MD to the mkdir command.
To create a new command called mk, use the following command:
alias mk=mkdir
? See ``Aliasing Commands,'' p. 347
Deleting Directories with rmdir
The rmdir command deletes Linux directories. The command takes the name of the
directory to delete. This directory must be empty; otherwise, Linux can't remove it.
For example, if the /backup directory has two directories within it, the command rmdir /
backup fails. The command rmdir/backup/jack/* removes all files in the /backup/jack
directory, and then rmdir /backup/jack removes the now-empty /backup/jack directory.
You can't delete a directory that contains files by using the rmdir command. Instead,
you can use the -r flag to the rm command. For example,
rm -r *
deletes everything from the current directory and every directory below the
current directory. Be very careful using this command because the moment you
delete a directory, you can't recover the directory or the files that were located in the
directory. Make backups.
Using File-Manipulation Commands
Because Linux treats directories and files similarly, it provides similar commands for
manipulation.
Using Basic Commands 51
Copying Files with cp
The cp command is similar to the DOS COPY command. You use this command to copy one
or more files from one directory to another directory. The syntax of cp is as follows:
cp from-filenameto-filename
You must supply both the from-filename and to-filename parameters for the files to be
.
copied. If you want to preserve the filename, you can use the dot ( ) as a placeholder for the
Part
to-filename parameter. This use is in contrast to DOS, where you can leave off the
to-filename. I
Ch
The command cp fred1 fred1.old copies the file fred1 to a backup file named fred1.old,
whereas the command cp ~fred1.old/backup/jack copies the file fred1.old from the home 2
directory to the /backup/jack directory. (The ~ character represents the user's home
directory.)
Moving Files with mv
The mv command, which is similar to the DOS MOVE command, allows you to move files from
one directory to another directory. Moving a file has the same effect as copying the file to a
new directory and then deleting the file in the old directory. mv doesn't make a copy of the
file.
The syntax of the mv command is identical to the cp command:
mv from-filenameto-filename
The command mv fred1 fred1.old copies the file fred1 to a backup file named fred1.old and
deletes the old fred1 file, whereas the command mv ~fred1.old /backup/jack moves the
fred1.old file from the home directory to the /backup/jack directory.
Deleting Files with rm
To delete files under Linux, you use the rm command. The rm command is dangerous because
as soon as a file is deleted, you can never recover it. For safety reasons, you should use the
following form of the rm command:
rm -i filename
The -i parameter tells the command to query, or inquire, you to see whether that's the file
you really want to remove. For example, the command rmfred1 removes the file named
fred1, whereas the command rm-ifred1 deletes the fred1 file after asking whether you really
want to remove this file.
As soon as you delete a file under Linux, that file is gone. You can't undelete a file or
directory under Linux like you can with DOS. If you delete a file, your only hope is a
backup copy.
52 Chapter 2 Linux Installation Overview
Displaying File Contents with more
The more command displays a screen of a text file. You can look through a text file without
invoking an editor, printing the file, or trying to pause the terminal as it displays the file. To
display the contents of your Emacs configuration file, for example, you can type the
following:
more .emacs
Tip #11 from
Jack If you try to pass a binary data file to more, you could have some unpleasant effects;
for example, your terminal can lock up. If your terminal does lock up, try pressing
Ctrl+q or Ctrl+s.
Use thestrings command if you want to see any text a binary file contains.
strings displays to the screen any printable ASCII characters contained in the binary
file.
A disadvantage of using more is that you can't back up to see a screen of information after it
passes. The command described in the following section overcomes that problem, though.
Using lessÐA Better more
less displays information one screen at a time on your terminal. The program's name is a
play on words for the program it's meant to replaceÐ more. Like more, less can display a
screen of information in a text file, but unlike more, less allows you to page back and forth
within the file. You can use the following command, for example, to browse through the
README file located in the info directory:
less /info/readme
Clearing the Screen with clear
Sometimes after filling your terminal screen with information, you want a blank screen while
you sit and contemplate your next action. Under DOS, you can use the cls command, but
under Linux, you must use the clear command.
Dealing with DOS Files Under Linux
During installation, you are given the chance to make any DOS partitions you have available
visible to Linux. These partitions are then placed in a directory you specify during
configurationÐfor example, /dosc.
If you want to copy these files to a floppy, using the cp command might cause problems
because UNIX and Linux treat text files a little differently than DOS, especially when dealing
with carriage returns and linefeeds. To help you overcome this problem, developers created a
group of programs to help deal with MS-DOS files under a UNIX environment. These
Dealing with DOS Files Under Linux 53
programs are the m- commands, which include such commands as mcopy and mdir. mcopy
works just like the DOS COPY command, and mdir provides a directory listing. As you may
notice, they resemble their DOS counterparts, except that they begin with the letter m, hence
the name `` m- commands.'' The m- commands are part of the mtools package, which is a
collection of public-domain programs that allows UNIX to interact with DOS files much
more easily.
These commands also make copying files to floppy disks much easier because you can use the
Part
DOS designation, like A, rather than the Linux designation /dev/fd0. For more information
I
on the m- commands, enter the following:
Ch
man mtools
2
Table 2.5 provides a brief listing of the various m- commands.
Table 2.5 The m- Commands
Command Description
mattrib Displays the file attributes for the specified file(s)
mcd Changes the directory to the specified path
mcopy Copies the files specified to the new path
mdel Deletes the specified files
mdir Provides a directory listing
mformat Formats a floppy
mlabel Labels the DOS file system
mmd Makes a directory
mrd Removes a directory (must be empty, just as in DOS)
mren Renames an existing DOS file
mtype Displays the text contents of a DOS file
Although you can view a DOS file with Linux and even do some editing on text files in
DOS partitions that Linux can see, you can't execute DOS or Windows programs
under Linux. However, projects are under way across the Internet to supply such
emulation for Linux. Although the prospects look very good for such emulators in the
future, at this time DOS and Windows emulation isn't fully available.
54 Chapter 2 Linux Installation Overview
Shutting Down Linux
When you're finished using a DOS machine, you can typically just turn off the power and
walk away. You could also do the same under Windows, although you're likely to cause file
damage. Under Linux, simply turning off the power presents even more chances for
damaging your system, both to hardware and file systems. You must shut down Linux in an
orderly fashion, or you might corrupt the operating system to the point where it can't boot
the next time you try.
Linux keeps a lot of information about itself and files in memory, in areas called buffers,
before writing the information to disk. This process helps improve system performance and
control access to the hardwareÐsomething a multitasking operating systems needs to
maintain so that one user doesn't try to use a hardware device that another user is using. If
you turn off the power, this information is lost, and you can corrupt your file system.
? See ``Shutting Down Linux,'' p. 243
Because Linux is a multiuser and multitasking operating system, it must make sure that every
user stops processing gracefully and saves any work in progress before shutting down the
system to prevent data loss and file damage. This shutdown process also gives each user
logged in to the system time to log out. To shut down Linux in an orderly fashion, you must
use the shutdown command. The shutdown command syntax is as follows:
shutdown [-r] time-to-shutdown [message]
Tip #12 from
Jack For a quick restart you can use the reboot command. Also most Linux distributions
for Intel PCs also trap the Ctrl+Alt+Delete three-finger salute and perform a reboot.
-h flag to shutdown or use the
If you want to halt the system, you can specify the
halt command.
reboot and halt can be dangerous since they do not perform a full shutdown
sequence as does shutdown. When using reboot or halt you may want to use the
sync command to flush all data cached in your system to the drives, just to be safe.
For example:
sync; reboot
or
sync; halt
the semicolons allow you to specify multiple commands on the same command line.
The optional -r flag indicates that Linux should immediately reboot after it shuts down. This
option is useful if you want to quit Linux and boot to another operating system.
time-to-shutdown indicates when the system should shut down. The time is specified on a
24-hour clock, so you can tell the machine to shut down at 11 p.m. by entering the following:
shutdown 23:00
Troubleshooting 55
The message parameter is a message sent to each user logged in to the system. This message
is displayed on their terminals. You can use this message to tell users why you're shutting
down the system. For example, if you need to do weekly backups, you can use the following
message to make sure that everyone logs out of the system:
shutdown -r 23:00 Shutting down at 11:00pm for maintenance
Remember, don't simply turn off the computer or press the reset button to exit Linux.
Part
I
Ch
On some systems, Linux traps the Ctrl+Alt+Del reboot keystroke and executes an
orderly shutdown as though the user had typed the shutdown command. However, 2
on some systems, Linux can't detect this keystroke combination and reboots
immediately.
If you do accidentally turn off your system and damage the file structure, you can use
the fsck command to try to repair the file system.
? See ``Using the fsck Command,'' p. 449
Troubleshooting
After your machine reboots, the LILO prompt should appear. Make sure that you can boot to
your old operating system if you left it on the hard drive. If that system was DOS, press Shift
and then type the short word you used to identify the DOS partition when you installed
LILO. If you enter an invalid word, press Tab to get a list of valid operating system types. If
you're having problems at that point, place your DOS boot disk in the boot drive and reboot.
You should be able to boot from your boot disk. When your system is up and running under
DOS, try the Linux boot disk you created during installationÐnot the ones you created to
originally install the entire system. If that boot disk doesn't work, you may have to reinstall
Linux. Potential problems to check initially are the kernels and your hardware. Before
starting over, make sure that you have the appropriate hardware. If you made notes during the
installation process, check which kernel you installed against what hardware you have. Make
sure your hardware is supported by Linux.
Below are some answers to common problems listed on Red Hat's Web site. These
troubleshooting tips are used under the provisions of GNU's GPL.
Q: Can I use a hard drive that has more than 1023 cylinders?
A:The infamous 1023 cylinder question. Yes, but not to boot Linux. You can install Linux on
partitions above the 1023 cylinder, but to boot Linux, the root directory and specifically the
/boot directory must be installed on the first hard drive below 1024.
56 Chapter 2 Linux Installation Overview
Q: How do I add arguments for LILO at the prompt?
A:Some hardware requires that extra parameters be fed to the kernel before the kernel will
recognize the hardware. You can accommodate this by editing the /etc/lilo.conf file to
provide the necessary parameters, or you can provide them manually during boot up. See the
LILO HOW-TO for more examples of LILO parameters.
Q: Why does LILO hang on LI?
A:This is a symptom of the 1023 cylinder problem addressed previously. If you have installed
the boot system above 1023, LILO will not be able to boot the system. You can try to boot
from a floppy using the rescue disk you made during installation, or you can repartition your
hard drive and reinstall Linux.
Q: The installation will not find the SCSI card.
A:To remedy this, you need to add a boot-time argument such as the following:
LILO: linux qlogicfas=0x230,11,5
This option can be made permanent so you don't have to re-enter it. See the LILO
configuration option append in the lilo.conf man page.
Q: How do I uninstall LILO?
A:If you want touninstall LILO and reinstall the original boot record, try using this
command
lilo -u /dev/hda
which represents the boot record of the first IDE drive. Parameters may vary for your
machine, for example, if your first hard drive is a SCSI drive, you would use /dev/sda.
Q: Can I use LILO and Win98 on one installation?
A:Yes, install Windows 98 first and then install Linux. During the installation, tell Linux to
place LILO in the MBR. You can also use a commercial program such as System
Commander.
Q: How do I mount a CD-ROM?
A:Installing Red Hat 6.0 should place the proper entries in your /etc/fstab file, as shown in
the following listing:
#
# /etc/fstab
#
# You should be using fstool (control-panel) to edit this!
#
# <device> <mountpoint> <filesystemtype> <options> <dump> <fsckorder>
/dev/sda1 / ext2 defaults 1 1
/dev/sda5 /home ext2 defaults 1 2
/dev/cdrom /mnt/cdrom iso9660 noauto,ro 0 0
/dev/fd0 /mnt/floppy ext2 noauto 0 0
/dev/sda6 /var ext2 defaults 1 2
/dev/sda2 none ignore 0 0 0
Troubleshooting 57
none /proc proc defaults
/dev/sda7 none swap sw
Note the use of noauto for the cdrom entry. Without this setting, Linux will try to automount
the CD-ROM when it boots, which isn't really a problem unless there's no CD in the drive.
If there is not an entry in your fstab file, you can either edit /etc/fstab or use the X
Window Control Panel tool to add the appropriate mount information. Also, make sure the
mount point /mnt/cdrom does indeed exist. If the entry is correct, you can cd to the mount Part
point and issue the following commands:
I
cd /mnt
mount cdrom Ch
Q: I have Red Hat 5.0 and have upgraded to the ld.so RPM package listed in the errata, but my
2
libc5 applications still create a seg fault. What is wrong?
A:The problem with crashing libc5 applications can be caused by several things.
Before or after the upgrade, another version of libc might have been installed that was not
obsoleted by the upgrade process, or the libc5 libraries might have been placed in a location
that causes conflict.
To find out if this is the case, run this command:
rpm -qa | grep libc
It should produce the following output:
glibc-devel-2.0.5c-12
libc-5.3.12-24
glibc-debug-2.0.5c-12
rpm-2.4.10-1glibc
rpm-devel-2.4.10-1glibc
glibc-profile-2.0.5c-12
glibc-2.0.5c-12
If you see items like libc-debug-5.3.12-18 or libc-5.4.44-2, you will need to remove these
packages (with the command rpm -e libc-debug) and run ldconfig Dv.
Your /etc/ld.so.conf file has been changed from an optimal setting. For optimal loading,
set your /etc/ld.so.conf file in the following order:
/usr/i486-linuxaout/lib
/usr/i486-linux-libc5/lib
/usr/openwin/lib
/usr/X11R6/lib
Q: Some of my older applications get the incorrect time.
A:Some libc5 apps want /usr/lib/zone info. You can either recompile them for libc6, or
you can provide a symlink with the following command so things will work.
ln -s ../share/zoneinfo /usr/lib/zoneinfo
58 Chapter 2 Linux Installation Overview
Q: I have all the latest updates installed, but my programs still get the incorrect time.
A:If you have installed all the latest updates and your programs still get the incorrect time, try
checking the settings in /etc/sysconfig/clock. They probably look something like this:
UTC=true
ARC=false
This means that Linux will assume your BIOS clock is set to the UTC or GMT time zone.
More than likely, the clock is set to your local time zone, and you need to change the UTC
line to the following:
UTC=false
Q: When the system boots up, I see a message that says I have unknown PCI hardware. What does this
mean?
A:The error ``unknown PCI device'' can occur for several reasons. The first and most
harmless one is that PCI isn't responding to Linux's queries in a way it understands, but
Linux is able to keep going. The more common occurrence is that the system hangs on,
querying PCI bus cards, and cannot get any further.
Because this is a hardware problem in the kernel, there is not much that Red Hat can do
except point you to the maintainer of that section of the kernel. That person might be able to
let you know what is going on and might want to look at what hardware you do have in your
system so she can better handle it in the future. The maintainer can be reached at:
[email protected]
Include the following information
/proc/pci
which is your exact hardware description. Try to find out which device is unknown. It may be
your main board chipset, your PCI-CPU bridge, or your PCI-ISA bridge. If you can't find
the actual information in your hardware booklet, try to read the references of the chip on the
board.
Q: Linux isn't detecting my NE2000 compatible network card.
A:It has been found that some NE2000s that worked with earlier kernels do not work with
the later 2.0.x kernels. For some, the following workaround will enable them to work.
You can try to get the card to work by entering the following settings:
insmod 8390
insmod ne io=0XXXX irq=Y
Replace XXXXand Y with your IO address and IRQ. Most common values for the IO
address are 0x300 and 0x310. The IRQ can be anything.
Troubleshooting 59
After this, use ifconfig or netcfg to configure the card. Sometimes, even though the card is
recognized, it fails to transfer TCP/IP packets. This is being looked into.
If the previous settings work, add them to /etc/conf.modules. It should look something like
this:
alias eth0 8390
alias eth0 ne
options eth0 io=0xXXX irq=Y Part
Q: I have installed Linux, and it seems to initially start booting. However, when it gets down to I
something called sendmail, the machine seems to hang. What is happening, and what should I do?
Ch
A:If after the install the machine seems to hang when it reaches certain processes like
sendmail, apache, or SMB, there is probably a network problem. The most common cause is
2
that Linux cannot look up the name of the machine you have called the box (if you set up
networking to have a machine name). The machine is currently paused waiting for the
network timeout of DNS lookups and will eventually bring up the login prompt. When you
get the prompt, log in as root and check the usual culprits for a problem.
If you are directly on a network with a DNS server, make sure that the /etc/resolv.conf file
has the correct values for your machine's DNS server. Check with your systems administrator
for the correct values.
If you are using Linux on a network without a DNS server (or if this box is going to be the
DNS server), you will need to edit the /etc/hosts file to have the hostname and IP address so
that the lookups will occur correctly. The format of the /etc/hosts file is like the following
example
127.0.0.1 localhost localhost.localdomain
192.168.200.1 mymachine mymachine.mynetwork.net
where the example machine is called mymachine.
CHAPTER
3
Installing Red Hat Linux
In this chapter by Jack Tackett, Jr.
Starting the Installation Process 62
Installing from Floppies or CD-ROM 63
Understanding the Various Installation Methods 64
Installing the Linux System 67
Configuring Your System 82
Configuring Your Network 84
Troubleshooting 88
Going Back to the Beginning 89
Case Study: Installing Red Hat Linux on DEC Alphas 89
62 Chapter 3 Installing Red Hat Linux
Starting the Installation Process
This chapter gives you the information you need to install the Red Hat distribution of Linux.
Although this book leads the way, you might find the need to use the resources, such as the
various How-Tos, provided on the Red Hat CD-ROM. However, Red Hat is one of the
easiest distributions to install, so take heart!
First, you need to decide how to install the product. Table 3.1 describes the various methods
available for installing Red Hat 6.0. No matter what medium you select, you need either one
or two blank floppy disks, depending on your selected installation method. If you have a PC
capable of booting from a CD-ROM, and you do not need PC card (PCMCIA) support, then
you need just the Red Hat CD-ROM and no floppy disks.
Table 3.1 Installation Methods
Method Description Pros/Cons
Floppy/CD-ROM You can use this traditional A floppy boot disk is used, and
method to install most Linux then installation continues
distributions. from the CD-ROM. This
installation method is the
most likely to succeed on the
first try.
CD-ROM You can boot and install This method is fast, but you
directly from CD-ROM by must have a PC that can boot
using autoboot. from a CD-ROM. Most older
systems do not have this cap-
ability.
Hard Drive You can copy CD-ROM This method is fast, but you
contents to an unused hard need the free space for the
drive or partition and install copy and free space for the
from the copy. installation.
HTTP You can install from a remote You must have a supported
location viathe Web. network card and a network
connection. Transfers may be
slow, depending on your
connection speed. In fact, you
should attempt an HTTP
install only if you have a cable
modem or xDSL connection.
FTP This method is the same as This method poses the same
for HTTP, but it uses a problems as an http
remote FTP server. installationÐslow connec-
tions and possible hardware
incompatibilities.
Installing from Floppies or CD-ROM 63
Table 3.1 Installation Methods
Method Description Pros/Cons
NFS You can install from a Net- This method is typically used
work File System server. in local area networks (LANs)
to install many Linux systems
from a single site. Network
problems and bandwidth may
cause problems.
If you're doing a full installation and not upgrading a previous version of Red Hat Linux, you
need to decide on what class of installation after you decide on the method. Red Hat provides
for three types of installations: workstation, server, and custom.
Part
Installing Red Hat Linux over another distribution, such as Caldera, Debian, Slackware,
I
or SuSE, may destroy important configuration files, data, and program files. Red Hat
Ch
does not save information from those distributions as it does when upgrading from
prior Red Hat distributions. If possible, you should put your home directories on a
3
separate drive so that upgrades can go smoothly from one system to another. You
should also save your /etc /var directory. If you have installed
directory and local
software, such as programs from /usr/local, then you should save that
information, too.
& WorkstationÐSelect workstation if you intend to use Linux for your day-to-day
tasks. Be careful, though, because selecting Workstation erases all prior Linux
partitions along with all the data selected. You need approximately 600MB for a
typical workstation installation.
& ServerÐSelect server if you want to have a complete Linux Internet server and a
capable workstation system, too. A server installation also destroys data and requires
more disk space. You need over 1.7GB of disk space for a complete server installation.
& CustomÐThis type of installation is typical of previous versions of Red Hat. With
this selection, you can pick and choose your partition sizes and installation packages.
Installing from Floppies or CD-ROM
To start the installation process, you need one or two (depending on the installation method)
3 1/2-inch 1.44MB formatted floppy disks. You will use these disks to create a boot disk for
the Linux installation. You should also have an extra disk available to create a rescue disk.
Next, you should make sure that you have enough hard disk space to install Linux. Everything
on the CD-ROM, if installed, requires about 1.7GB of disk space, but you can get by with
less, especially if you don't install the X Window System. To decide on the amount of space,
64 Chapter 3 Installing Red Hat Linux
you should decide how much space you want for user accountsÐthat is, the space you want to
provide to your users. On a single-user system, 500MB is adequate.
Next, you need to decide how much swap space your machine needs. You can figure on at
least 64MB. Finally, you can figure about 256MB for your root directory. It is the main
directory from which all other directories under Linux are accessed.
? See ``Linux Standard Directories,'' p. 418
Tip #13 from
Jack You can also run part of the Linux file system from the CD-ROM without installing all
the software, but at a great decrease in system performance. You can choose to do so
during installation.
If you decide to install and configure the X Window System (highly recommended), you
should also write down what type of chipset your video card uses. If you have a serial mouse
and modem, write down the serial port that each is using. You need this information later
during the configuration process.
? See ``Installing the XFree86 System,'' p. 508
Understanding the Various Installation Methods
If you're reading this book, you will probably install Red Hat Linux from the accompanying
CD-ROM. However, you can use one of the following five methods to install Red Hat: from
CD-ROM, via Network File System (NFS), via File Transfer Protocol (FTP), via a Session
Message Block (SMB) image on a shared drive, or from a hard drive.
To install Linux directly from a CD-ROM, you need access to DOS. From the DOS prompt,
execute the command
[cdrom-drive]:\dosutils\autoboot
where [ cdrom-drive] is the drive letter for your system's CD-ROM.
This method erases your hard drive. Be sure to back up any files you fear losing.
If you have another partition available, you can install Linux to coexist with your system
without erasing what's already there. To do so, you need the CD-ROM, an empty partition,
and a boot disk. You'll learn later in this chapter how to create the boot disk, as well as how to
repartition your hard drive.
Understanding the Various Installation Methods 65
For those of you who have systems that can boot from a CD-ROM (check your BIOS
settings), you can boot and install from the Red Hat CD-ROM.
NFS (Network File System) provides a way to install Red Hat across a network. First, you
must mount the CD-ROM drive on a machine supporting the ISO-9660 file system with
RockRidge extensions and then export the file system via NFS. You need to know the path to
the exported file system and the IP number or, if Domain Name Service (DNS) is configured,
the name of the system.
FTP (File Transfer Protocol) is a method for transferring files across the Internet. (Chapter
32, ``Accessing the Network with telnet, ftp, and the r- Commands,'' explains FTP in more
detail.)
Installing Red Hat from a hard drive requires the same boot disk used for an FTP installation.
First, you create a directory named RedHat. Then you copy the corresponding directory from Part
RedHat directory. You can use the
the CD-ROM, and all the subdirectories there, to the
I
following DOS commands to do so:
Ch
cd \RedHat
xcopy /s e:\RedHat 3
The cd command assumes that you're already on the installation hard drive; the xcopy
command assumes that your CD-ROM drive is drive E.
No matter which method you use, you'll need to gather some information.
Compiling Necessary Information
Before starting the installation, you need the following information about your system:
& The type of video card, chipset, and monitor used
& The serial port used by your mouse
& The serial port used by your modem
& The network information for your computer, if it's connected to a network (items
such as its IP address, gateway, and domain name)
& The type of hard drive and CD-ROM drive in your system and their controller types
& The directory structure you want to use on your system, such as placing /home on a
separate hard drive and /var on a separate partition from your swap file
& The name you intend to call your system (the hostname)
If you're connecting to the Internet, you can get most of this information from either your
network administrator or your Internet service provider (ISP).
If you intend to use other operating systems on the same computer (such as Windows 95/98,
Windows NT, or OS/2), you need to create the necessary partitions for these operating
systems. Typically, you need to use the operating system's partitioning software because
Linux can't handle these other partition types.
66 Chapter 3 Installing Red Hat Linux
ON THE WEB
A product named System Commander, from V Communications, lets you install and switch between 32
different operating systems. You can find more information about this product at the following site:
http://www.v-com.com/
Next, you should check for any last-minute changes to the Red Hat distribution. The reasons
for checking are many, but the two major reasons are that Linux is constantly updated, and
this chapter is being written at least a month before the CD-ROM is cut. In the interim, new
material or bug fixes may have been released.
ON THE WEB
You can also check for updated material on the Web at the following site:
http://www.redhat.com/errata
If you're not installing directly from the CD-ROM, you next need to repartition your current
hard drive to make room for Linux. This step may cause problems because repartitioning a
hard drive destroys any data contained on the affected partitions. After making room for
Linux, you need to boot the Linux system and create its new partitions and file systems.
Typically, Linux systems need a primary partition to store the files on and a swap file
partition, especially if you have a machine with 8MB or less of memory.
A file system is basically a section of your hard drive specially formatted to hold
certain types of files. UNIX and Linux use file systems to represent entire sections of
the directory tree. This use is in contrast to MS-DOS, which places subdirectories in
the directory tree on the same logical drive. UNIX systems use the directory tree
format because placing subdirectories on different drives is safer. If one drive
malfunctions, only the information on that drive needs to be replaced or fixed.
? See ``Understanding File Systems,'' p. 440
After creating the file systems, you then can install the Linux operating system, its support
files, and various application packages distributed with the system. To install Linux, you must
first boot a stripped-down version of the operating system. You do so by creating a boot disk
containing the stripped-down operating system.
Creating the Boot, Supplemental, and Rescue Disks
You create the boot and supplemental disks by using the rawrite program. You can find this
program on the accompanying CD-ROM in the /dosutils subdirectory. For this step, you
need two formatted floppies: one labeled boot and the other labeled rescue. Place the boot disk
in drive A and enter the following:
Installing the Linux System 67
E:\dosutils>rawrite
Enter disk image source file name: e:\images\boot.img
Enter target diskette drive: A:
Please insert a formatted diskette into drive A: and press -ENTER-
If you want to abort the process, simply press Ctrl+c to stop. If rawrite fails, try a new
formatted disk. If the problem persists, you should have your hardware checked for possible
problems.
Next, you should create a rescue disk. Unlike previous releases of Red Hat, with 6.0, you
cannot use the installation disk to boot the system in case of problems. To provide for better
problem resolution, Red Hat has introduced a rescue disk image ( rescue.img). To create the
rescue disk, you can use rescue.img as the source file in the preceding rawrite commands.
Part
Installing the Linux System I
To start the Linux installation, you place the boot disk you created into your disk drive and Ch
reset your computer. After your system does its hardware and BIOS checks, you should see
3
the following boot messages on your system:
Welcome to Red Hat Linux
To install or upgrade a system running Red Hat 2.0 or later, press the
<ENTER> key.
To enable expert mode, type expert <ENTER>. Press <F3> for more information
about expert mode.
This disk can no longer be used as a rescue disk. Press <F4> for more
information on the new rescue disk.
Use the Function Keys listed below for more information
[F1-Main] [F2-General] [F3-Expert] [F4-Rescue] [F5-Kickstart] [F6-Kernel]
boot:
Table 3.2 provides information on the various function keys and their uses. Typically, you
only need to press Enter to continue installation.
Table 3.2 Installation Function Keys
Function Key Description
F1 Displays the main screen shown earlier in this section. Pressing F1 always
returns you to this screen.
F2 Provides general information on the installation process.
F3 Explains expert mode. Basically, the Linux installation process autoprobes
the hardware trying to determine what is installed. This probing can hang
a system. If your system hangs, you must enter expert mode and specify
your system's hardware during installation.
F4 Provides instructions on creating and using the rescue disk.
F5 Allows you to perform an unattended installation using a configuration file
in Red Hat 6.0. This screen provides information on performing such an
installation.
68 Chapter 3 Installing Red Hat Linux
Table 3.2 Installation Function Keys
Function Key Description
F6 Tells you how to pass some extra parameters to the kernel during boot-up
if Linux can't boot properly.
Tip #14 from
Jack For more information on passing boot parameters, see the ``Linux BootPrompt
How-To'' on the CD-ROM at /doc/HOWTO/BootPrompt-HOWTO. You can use either
WordPad or a word processor such as Microsoft Word to read this file; however, the
file is too large to use Notepad.
The system displays the following prompt and then begins initializing the system:
Loading initrd.img....
loading vmlinuz....
After booting, your system displays the Welcome screen (see Figure 3.1).
Figure 3.1
The Red Hat
Linux Welcome
screen greets
you at the be-
ginning of your
adventure.
Press Enter to continue. The next screen asks you which language to use during installation.
Moving around in the dialog boxes is easy, and the installation program provides
reminders at the bottom of most screens. To move from element to element (field),
you can press Tab or Alt+Tab. You use the spacebar if you need to select an item
from a list or check a check box. To choose a button (typically OK or Cancel), you
press Enter. To scroll through a list of selections, you use the arrow keys.
Installing the Linux System 69
The next dialog box asks you to select the type of keyboard used by your system (see Figure
3.2).
Figure 3.2
Linux uses the
selections you
make about
your keyboard
during installa-
tion and when
you boot your
system in the
future.
Part
Tip #15 from I
Jack You can use /usr/sbin/kbdconfig to change your keyboard selection in the
Ch
future.
The next screen asks whether you need PCMCIA (PC Card) support for your system (see
Figure 3.3). Select the appropriate answer by using the Tab key, and press Enter.
Figure 3.3
Red Hat Linux
provides op-
tional support
for PCMCIA
cards.
Next, you need to select the installation method. After you select your installation method,
press Enter. The installation program prompts you to insert the Red Hat CD-ROM into the
CD-ROM drive. When you're done, simply press Enter to continue.
The rest of this chapter assumes that you're installing from the local CD-ROM drive. If
you select another method of installation, see the appropriate help topics or Red Hat's
Web site at http://www.redhat.com.
70 Chapter 3 Installing Red Hat Linux
Next, the installation program attempts to autoprobe the system's CD-ROM type. If the
program determines the CD-ROM correctly, installation continues; if not, you must select
the type from the following selections:
SCSI Use this selection for true SCSI devices
Other CDROM Use this selection for non-IDE or soundcard-
compatible CD-ROMs
If you select Other, the installation program displays the selection dialog box shown in Figure
3.4. Select the appropriate drive type and press Enter to continue installation.
Figure 3.4
The Red Hat
Linux installa-
tion program
needs to know
what type of
CD-ROM is
used in your
system.
The Other CDROM category includes such drives as those sold by Creative Labs
(SoundBlaster) and other multimedia kit-based CD-ROMs, as well as the following:
Aztech CD Sanyo
Goldstar R420 Sony CDU-31A
Mitsumi Sony CDU-5xx
Optics Storage 8000 SoundBlaster/Panasonic
Phillips CM206/CM260
Depending on your selection, the installation program may ask for some parameters, such as
IRQs or DMA addresses. Or the program may try to determine these values automatically by
probing your hardware. It's best to let the program autoprobe first before providing
parameters.
Tip #16 from
Jack Any time the installation program probes the system's hardware, the system may
hang. If that happens, you must reboot and redo the installation. Be sure to try to
collect the needed information, such as IRQs and DMA addresses, before attempting
to reinstall.
Installing the Linux System 71
After detecting your CD-ROM type, the system starts its installation from the CD-ROM
drive. First, it asks you whether you're installing a new system or upgrading an existing Red
Hat system. Red Hat 6.0 easily upgrades over versions 2.0 or greater, but no Linux
distribution easily upgrades over a different distribution version. So if you have a prior
distribution version, such as Slackware, you would be wise to just perform a new installation
and blow away your prior systemÐafter backing up important data files, of course. If you're
upgrading from a previous version of Red Hat, the installation program saves any current
configuration files with the .rpmsave extension.
? See ``Updating Packages with RPM,'' p. 171
All actions performed by the installation program are also saved in the file /tmp/upgrade.log.
Tip #17 from Part
Jack To see what the installation program is doing, you can press Alt+F3 to change to the
I
virtual terminal that displays every action taken.
Ch
3
Next, the installation program probes for any SCSI devices. The program may ask you to
select a SCSI controller and display the Configuration dialog box, where you tell the system
whether you have any SCSI adapters in your system. Choose the appropriate button and
continue.
If you have a SCSI adapter, the program displays the Load Module dialog box from which
you can select from the following SCSI drivers:
Adaptec 152x Iomega PPA3 (Parallel port Zip)
Adaptec 1542 NCR 5380
Adaptec 1740 NCR 53c406a
Adaptec 2740, 2840, 2940 NCR 53C810/53C820 PCI
AdvanSys Adapters Pro Audio Spectrum/Studio 16
Always IN200 Qlogic FAS
Buslogic Adapters Qlogic ISP
DTC 3180/3280 Seagate ST01/02
EATA DMA Adapters Trantor T128/T128F/T228
EATA PIO Adapters UltraStor 14F/34F
Future Domain TMC-885, UltraStor 14F/24F/34F
TMC-950
Future Domain TMC-16x0 Western Digital wd7000
72 Chapter 3 Installing Red Hat Linux
ON THE WEB
For current information on hardware that Red Hat 6.0 does and doesn't support, see Red Hat's Web site
at the following address:
htp://www.redhat.com/corp/support/hardware/intel/60/rh6.0-hcl-i.ld.html
Next, you must partition your disksÐor at least select the partitions you've already created.
The installation program displays the Disk Setup dialog box shown in Figure 3.5. You can use
the command-line fdisk program or the GUI-based Disk Druid program. If you want to use
fdisk, move to the fdisk button and press Enter. When you press this button, you drop into
the fdisk program to partition the selected hard drive. If you want to use the Disk Druid
program, skip forward to the section ``Using Disk Druid.''
Figure 3.5
You have to
prepare your
hard drives for
installation.
You can use
the GUI pro-
gram Disk
Druid or the
command-line
fdisk program.
Using the Linux fdisk Program
At the fdisk prompt, type m for a list of commands. Table 3.3 lists the available commands.
You use the fdisk program native to Linux for these actions. Be careful because this
program is different from the fdisk programs included with other operating systems
such as MS-DOS, Windows 95/98, and OS/2. You cannot use these programs
interchangeably! For example, you cannot use Linux's fdisk to rearrange a partition
for a DOS partition. Although you can use any fdisk to create partitions, you must use
the appropriate operating system's version of fdisk to perform such actions as setting
file types.
Table 3.3 The Linux fdisk Commands
Command Description
a Toggles a bootable flag
c Toggles the DOS compatibility flag
d Deletes a partition
l Lists known partition types
Installing the Linux System 73
Table 3.3 The Linux fdisk Commands
Command Description
m Displays this menu
n Adds a new partition
p Displays the partition table
q Quits without saving changes
t Changes a partition's system ID
u Changes display/entry units
v Verifies the partition table
w Writes the table to disk and exits
x Provides extra functionality for experts only
Part
I
To begin the partitioning, select the p command (press p and then Enter) to display the
Ch
current partition table, which should reflect the drive you partitioned earlier with the DOS
FDISK program. Listing 3.1 shows a possible listing from the p command. 3
Listing 3.1 Example of a Current Partition Table
Disk/dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
dev/hda1 1 1 41 5219 1 Novell?
Your screen might appear different than what's shown in Listing 3.1 because the
values are different for each drive type and the partitions already defined on that
drive.
Listing 3.1 indicates the various partitions already defined that it can detect, the starting and
ending locations of the partition, and how big it is in blocks. The listing also indicates the
partition type. Table 3.4 shows all the different types of partitions you can define by using the
Linux fdisk program. The primary partition types used here are 83-Linux Native and
82-Linux Swap. You can get a similar listing by using the l command.
Table 3.4 Partition Codes and Types in Linux fdisk
Hex Code Partition Type
0 Empty
1 DOS 12-bit FAT
2 XENIX root
74 Chapter 3 Installing Red Hat Linux
Table 3.4 Partition Codes and Types in Linux fdisk
Hex Code Partition Type
3 XENIX usr
4 DOS 16-bit < 32MB
5 Extended
6 DOS 16-bit >= 32MB
7 OS/2 HPFS
8 AIX
9 AIX bootable
a OS/2 Boot Manager
40 Venix 80286
51 Novell?
52 Microport
63 GNU HURD
64 Novell
75 PC/IX
80 Old MINIX
81 MINIX/Linux
82 Linux Swap
83 Linux Native
93 Amoeba
94 Amoeba BBT
a5 BSD/386
b7 BSDI fs
b8 BSDI swap
c7 Syrinx
db CP/M
e1 DOS access
e3 DOS R/O
f2 DOS secondary
ff BBT
In Listing 3.1, Linux prints a note about the different physical and logical endings at the
bottom of the screen. The difference exists because on the system used to write this chapter, a
prior partition containing the DOS D drive was left intact, whereas the C drive was
repartitioned to a smaller C drive to make room for Linux. Thus, space exists between the C
drive and the D drive. The necessary partitions required by Linux will be created here.
Installing the Linux System 75
The begin, start, and end numbers from Listing 3.1 are very important, so you should write
them down. You'll need them in a later step to specify the necessary sizes of the partitions
you'll add.
Tip #18 from
Jack While the numbers are very important, you can specify the disk space using +xxxM
bytes rather than cylinder numbers. Cylinder numbers are much more accurate
though.
Adding the Necessary Partition
Part
Because you've repartitioned the drive for DOS, you shouldn't have to delete any partitions
for Linux. You should only have to add partitions. To add a partition, you issue the n I
command, which displays the following:
Ch
Command Action 3
e extended
p primary(1-4)
Now, you can press p and then Enter. When fdisk asks for the partition number, enter your
selection and press Enter. If you indicate a partition number already in use, fdisk reports this
fact and asks you to delete the partition before trying to add it to the partition table. For this
example, enter 3 to add a third primary partition that's referred to as /dev/hda3.
Next, fdisk asks for the location of the first cylinder. This is usually the first available cylinder;
in fact, fdisk displays a default range for your selection, as shown in this example:
First cylinder (42-1024) :
Notice that the first partition ends at cylinder 41 and that the next partition begins at cylinder
1024. Thus, the range supplied by fdisk here allows you to start the next partition anywhere
in the range of 42-1024. It's a very good idea not to place partitions just anywhere throughout
the disk; instead, choose the next available location, which in this case is cylinder 42. Enter 42
and press Enter.
Linux can have trouble booting from partitions defined to start at cylinders above
1024. If you can create a Linux partition only in this range, you might have to boot
Linux from a floppy. If the /boot directory is located within the first 1024 cylinders,
the system boots fine. Otherwise you'll learn how to create a boot floppy (different
from the boot floppy used for installation) later in this chapter. The only downside is
that booting Linux from a floppy takes a little longer than booting it from the hard
drive.
76 Chapter 3 Installing Red Hat Linux
Now fdisk wants you to specify how much space to allocate for this partition. You can express
this size in number of cylinders or by the number of bytes ( +size), kilobytes (+sizeK), or
megabytes ( +sizeM). Because you should already know the approximate size you need for the
swap file, you can define this partition first and then leave the rest of the disk space for the
Linux program partitions. Thus, for this example, your machine has 8MB of RAM, so you
need to specify a 16MB partition size by replying as follows:
Last cylinder or +size or +sizeM or +sizeK (42-1023): +16M
You should then use the p command to look at the new partition table you've defined. In this
example, the new partition table looks like this:
Disk /dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
/dev/hda1 * 1 1 41 5219 1 DOS 12-bit FAT
/dev/hda2 1024 1024 4040 384667+ 51 Novell?
Partition 2 has different physical/logical endings:
phys=(967, 14, 17) Logical=(4039, 14.17)
/dev/hda3 42 42 170 16447+ 83 Linux native
By default, fdisk makes the new partition a Linux Native type. To change this type to a swap
partition, you need to use the t command. To do so, enter t, and then enter the partition
number you want to change; in this example, enter 3. fdisk then requests that you enter the
hexadecimal value of the desired partition type from Table 3.4 (if you don't have the table
handy, you can type l to get the list of codes). Because you want a swap partition in this case,
enter 82 at the prompt.
As you can see, fdisk reports the new partition type, but you can also use the p command to
double-check that partition 3 is now a Linux swap partition.
Now you can add your Linux partitions. For this example, add only one partition, but if you
want to have multiple partitions for various reasons, you can also do so at this time. To add a
partition, enter n, specify p for another primary partition, and then specify the number for
4
this partition ( ). To keep from fragmenting different partitions across the drive, start the last
partition where the other left off, at cylinder 171. Because you want to use the rest of the
space for the Linux system, you can specify the last cylinder instead of an exact byte count.
Thus, enter 1023, as shown here:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 4
First cylinder (171-1024): 171
Last cylinder or +size or +sizeM or +sizeK (171-1023): 1023
Now use the p command to verify the new partitions. If you need to make any changes, you
can do so now.
Installing the Linux System 77
When you're satisfied with the layout of your partitions, you can use the w command to write
the partition table information to the hard disk. None of your changes are permanent until
you use the w command; thus, if you feel you've made some changes in error, you can use the
q command to exit without altering the partition table. If you use the w command, Linux tells
you the partition table has been altered and then resynchronizes the disks to match the new
partition table. If your Linux system hangs at this point, you can reboot with the installation
boot and root disks until you're back at the # prompt.
Don't use the Linux fdisk program to create or modify partitions for other operating
systems. Doing so could leave the hard drive in a useless state for both operating
systems.
Part
I
Creating the Swap Partition Ch
Most current distributions of Linux provide automatic creation and activation of the swap file
3
during installation, so you don't have to worry about creating the swap file. However, if
you're using a distribution that does not create the swap file, you might need to create and
activate the swap file before continuing with the installation.
If you get an ``out-of-memory'' type error during the installation procedures that
follow, you should increase the size of your swap file. If you already have the
maximum of 16MB, you need to create and activate another swap partition by
following these instructions. Remember that the Red Hat installation program allows
only one swap partition.
Next, you need to activate the swap system by using the swapon command, as follows:
# swapon /dev/hda3
Again, if you're using the accompanying Red Hat CD-ROM, you shouldn't have to worry
about activating the swap system as long as you create the partition for one. During
installation, the installation program detects the swap partition and automatically starts the
system for installation.
After you create your partitions on the various hard drives and return to the Partitioning
Disks dialog box, choose the Done button to continue with the installation.
Next, the system asks you to select the active swap space, which should be the partition you
created and marked as type Linux Swap (82) in the preceding section. Select this partition and
choose OK. The program then initializes the swap space.
78 Chapter 3 Installing Red Hat Linux
After creating the swap space, the program displays the Select Root Partition dialog box. The
root partition is your main file system for Linux where all the boot files are located. Select the
device (hard drive) for your root partition from the list box and press Enter. Now you can
mount the other partitions, if any, from the Partition Disk dialog box. From here, you also
can mount any DOS or OS/2 file systems so that you can access them from Linux. Select the
partition to edit from the list box and press Enter. From the Edit Mount Point dialog box,
you can specify a mount pointÐthat is, a directoryÐto which you want this partition
mounted.
? See ``Mounting and Unmounting File Systems,'' p. 444
After you select the root and mount points for your various partitions, the program formats
those you selected. You select the partitions to format from the Format Partition dialog box.
Using Disk Druid
Selecting the Disk Druid button on the Disk Setup dialog box (refer to Figure 3.5) displays
the Disk Druid main screen shown in Figure 3.6.
Figure 3.6
Disk Druid
makes parti-
tioning disks
and creating
mount points a
snap.
Using Disk Druid, you can create partitions, set mount points on specified devices, set sizes
of partitions, and specify file system types. Disk Druid provides information on these
attributes. Table 3.5 lists the various fields and buttons on the Disk Druid main screen and
describes what task each field or button performs.
? See ``Understanding File Systems,'' p. 439
Table 3.5 Disk Druid Fields
Field/Button Description
Mount Point Mount point is another term for a directory. This field lists the location
in which the specified directory and all subdirectories will be placed.
Device This field indicates the physical hard drive and partition to which this
mount point belongs.
Requested Size In this field, you can specify either a default size for a partition or an
initial size that can be allowed to grow.
Actual Size This field indicates the actual size allocated for the partition.
Installing the Linux System 79
Table 3.5 Disk Druid Fields
Field/Button Description
Type This field indicates the file system type of the partition.
Drive Summaries This portion of Disk Druid's main screen provides information on the
specified device (hard drive/patron) including such information as the
amount of space available. Remember, a hard drive device can have
several partitions.
Add This button adds a new partition.
Edit This button edits the selected mount point.
Delete This button deletes the selected mount point.
Ok This button commits the specified changes to your system and
continues with the installation program.
Part
Back This button returns to the previous dialog box in the installation
program and aborts all changes you have specified. I
Ch
To add a new partition, you can click the Add button or press the F2 function key to display
3
the Edit New Partition dialog box shown in Figure 3.7.
Figure 3.7
You don't need
to remember
all those differ-
ent file system
types. The Edit
New Partition
dialog box pre-
sents them all
in a list box.
Here, you can enter the mount point for this new partition in the Mount Point field.
/
Examples are the root partition ( ) or the var partition ( /var). Next, you can specify the size
of the partition in megabytes and indicate whether you want the partition to grow in size as
necessary when you add and delete other partitions. Next, you must select from the Type list
box what type of file system you want on the partition. Finally, you can select which physical
hard drive to place the partition on by selecting the appropriate hard drives from the list of
Allowable Drives.
If you specify a size that's too big for the space available on the indicated device, Disk
Druid tells you and prompts you to reduce the amount of requested space. Disk Druid
also warns you of other potential problems and provides you with possible solutions.
80 Chapter 3 Installing Red Hat Linux
Installing the Software Components
Congratulations! Your system is now prepped for Linux, but you're only halfway finished.
Now you must select the various software components to install and then configure them.
The installation program displays the Components to Install dialog box, from which you can
select the various packages. Table 3.6 describes each package.
Table 3.6 The Installation Components
Component Description
Printer Support Allows you to print from your Linux system.
X Window System Provides the GUI for all UNIXÐhence, LinuxÐ-
workstations; the X Window System is a powerful
GUI like Windows 95/98 and OS/2.
Mail/WWW/News tools Provides programs to use email, surf the World
Wide Web, and read and post Usenet news.
DOS/Windows Connectivity Allows you to access DOS files, run DOS programs,
and run some Windows programs (with limited
success).
File Managers Provides tools such as Midnight Commander to
manipulate your file systems.
BRU Backup Util Provides a single-user version of the popular tape
backup program.
BRU X11 Front End Provides a GUI interface under the X Window
System to BRU.
Real Media Client Allows you to access Real Media programs on the
Internet.
Real Media Server Allows your Linux server to provide Real Media
content to the Internet.
Graphics Manipulation Provides programs to work with graphic images
such as xv and the popular The GIMP.
X Games Provides popular strategic and arcade type games
that run under the X Window System.
Console Games Provides games that run on a text console.
X Multimedia Support Provides multimedia support for the X Window
System.
Console Multimedia Support Provides multimedia support for text consoles.
Print Server Allows your Linux box to act as a print server on
your network.
Networked Workstation Provides networking applications and Simple Net-
work Management Protocol (SNMP) support.
Dialup Workstation Allows you access to the Internet via dial-up
linesÐthat is, via a modem.
Installing the Linux System 81
Table 3.6 The Installation Components
Component Description
News Server Allows your system to act as a news server (if you
can get a news feed), thus providing news to your
users.
NFS Server Allows your system to export and attach to other file
systems on your network.
SMB (Samba) Connectivity Provides SMB services, both client and server.
IPX/NetWare Connectivity Provides access to Novell NetWare networks.
Anonymous FTP/Gopher Server Allows you to set up your system so that others can
access it via anonymous FTP.
Web Server Includes the most-used Web server software today,
Apache.
Part
DNS Name Server Provides the software needed to run your own
I
Domain Name Server on your Linux system.
Ch
PostGres (SQL) Server Allows you to run the PostGres SQL database
system.
3
Network Management Workstation Provides utilities and tools to help troubleshoot and
monitor your network, including SNMP services.
TeX Document Formatting Provides a series of programs used to add formatting
codes to documents.
Emacs Installs the ubiquitous editor for Linux (you can do
anything with Emacs, or so the Emacs gurus say).
Emacs with X Window System Provides an X Window System front end to the
powerful Emacs editor.
C Development Provides the GNU gcc compiler and tools.
Development Libraries Provides various libraries needed by the various
development tools, such as gcc and g++.
C++ Development Installs the GNU C++ compiler, gcc.
X Development Provides the tools, libraries, and miscellaneous
items (such as fonts) needed to develop X Window
System applications.
Extra Documentation Provides the Linux documentation project contain-
ing the important How-Tos, along with other
helpful information.
Everything Installs everything on the CD-ROM. You need
about 350MB available, not counting free space for
your data files.
You can select individual packages by checking the appropriate check box in the
dialog box, or you can install everything by selecting that list option. To select a
82 Chapter 3 Installing Red Hat Linux
package to install, simply move to the desired component and press the spacebar.
After you select all your components, tab to the OK button and press Enter.
Tip #19 from
Jack You can use the RPM program described in Chapter 7, ``Upgrading and Installing
Software,'' to install any package in the future.
The next dialog box after the installation informs you that you can see the files installed by
viewing the file /tmp/install.log. Press Enter to continue with the installation.
Now comes the hard partÐwaiting. Transferring and decompressing upward of 1600MB of
programs can take awhile. Setup firsts installs a file system on your indicated partitions and
then starts installing software. In the Install Status dialog box, the system informs you of its
progress as it installs the various files you selected. Installation time varies depending on what
you're installing and how fast your machine can process the information. Relax and order a
pizza!
Configuring Your System
After installing the software, the installation program begins to configure your system. It first
configures your mouse by displaying the Configure Mouse dialog box. Here, you can simply
select the mouse type that best describes your mouse from the list box. Remember, many
mouse devices can emulate the Microsoft serial mouse if they have to. The Emulate 3 Buttons
check box is there because many PC mouse devices have only two buttons, and the X
Window System usually uses three buttons to maneuver and make program selections. If you
check this box, the system makes clicking both mouse buttons at the same time the same as
pressing the middle button on a three-button mouse. After you make your selection,
choose OK.
Tip #20 from
Jack You can use the command /usr/sbin/mouseconfig to reconfigure your mouse at
any time in the future.
Next, you must specify the serial port that your mouse connects to. After you make the
selection from the list box, tab to the OK button and press Enter. The program then prompts
you to select the type of video card in your system.
Configuring Your System 83
Try to select the correct video card because of all things software-based, the only
subsystem that software can easily destroy is your video card and monitor. If you
make the wrong decision, you might fry your monitor! Although this outcome is
highly unlikely, there's still the slight possibility, so choose wisely, young Linux
Walker.
The system now tries to install the proper XFree86 server for your hardware.
? See ``Installing the XFree86 System,'' p. 508
Next, you must select your monitor. Again, you should be as specific as possible. After you
select your monitor, the program asks for the amount of video memory your card contains.
Make the appropriate selection and choose OK to continue. Part
Remember all those warnings about frying your monitor? Well, now you really have a chance
I
to toast it, so be careful. The next screen prompts you to select the clockchips located on your Ch
video card. These chips are used to drive the video signals through your card and into your
3
monitor. If they're way out of synchronization, the signals canÐyou guessed itÐfry your
monitor (few actually explode, most just fizzle and smoke). Please be careful! If you have no
clue as to what clockchips your card is using, take the default selection, No Clockchip
Setting, and choose OK.
After you select your clockchips (or lack thereof), the system can autoprobe and try to
configure the X Window System. The autoprobe might hang your system, but as long as
nothing is seriously wrong (for example, you selected outrageous clock speeds for your card),
you can simply reboot and continue with the installation. You do have the option to skip the
autoprobe and continue with the installation.
I have installed Red Hat many times, and Red Hat 6.0 is the first distribution to
properly configure my X Window System. You may have worse luck than me,
especially when dealing with laptops, but don't worry if your X installation fails. You
can configure X after installation by using the various configuration programs
available.
? See ``Configuring XFree86,'' p. 514
If the autoprobe succeeds, the system displays an information screen on selecting the
resolutions you want to use with your system. You can select more than one, as long as your
video card and monitor can handle the resolutions. Finally, the program tells you how to start
and stop your X Window System.
84 Chapter 3 Installing Red Hat Linux
Configuring Your Network
After configuring the X Window System, the installation program continues with your
network. If your machine is or will be connected to the Internet and you installed the
networking components, you should choose Yes and continue.
First, the system asks which Ethernet driver to use in the Load Module dialog box. Select the
appropriate driver for your Ethernet card and choose OK.
Again, the installation program might try to autoprobe your hardware to determine certain
values for the card. This probing can hang the system and force you to reboot. If this
happens, hang in there. First, make sure that you selected the correct driver. Then see
whether you need to pass any special parameters to the device, such as IRQ or DMA address
settings. You can do so by selecting the Specify Parameter option instead of the autoprobe
option.
Ethernet is the most popular network interface for Linux today. Other technologies,
such as Token Ring, ISDN, and ATM, have some support.
If the system can detect your network card, it leads you through setting up your TCP/IP
network.
Configuring the TCP/IP Network
The installation software uses the Configure TCP/IP dialog box to gather your system's
TCP/IP information. Your network administrator or Internet service provider can provide
the following information: your machine's IP number, netmask, network address, and
broadcast address.
Next, the system must configure your network. It gathers information from the Configure
Network dialog box. You must specify your network's domain name and your system's host
name. The domain name is typically the last two parts of an Internet address. For example, if
the name is www.netwharf.com, then netwharf.com is the domain name and www is the host
name.
Next, your network administrator must give you the values for your system's default gateway
and the primary name server. Your network may also have a secondary name server, too, so
enter the value in the appropriate place.
Be careful what you name your host because this name will appear on your default
prompt line, in mail messages, and in log reports. Do you really want your boss to
receive mail from uradork.netwharf.com?
Configuring Your Network 85
Configuring the Clock
Next, you have to specify how your system will keep time and in what time zone it exists. You
do so in the Configure Timezones dialog box. Check whether you want to use local time or
Greenwich mean time (GMT), and then pick your time zone from the list box. After you
make your selections, choose OK.
Selecting the Services to Start on Reboot
For the next step, you need to specify the services (programs and daemons) that your system
will start automatically when it boots. You select from the list of services displayed in the
Services dialog box, shown in Figure 3.8. Table 3.7 provides a list of the available services and
a description of their uses. Those services marked with an asterisk (*) have been selected by
Red Hat to start on reboot by default.
Part
Figure 3.8
I
Linux gives you
control of Ch
which pro-
grams to start
3
at boot time,
which is similar
to the startup
folder under
Microsoft
Windows.
? See ``Understanding the Boot Process,'' p. 234
Table 3.7 Startup Services
Service Description
amd Runs the automount daemon
apmd* Monitors battery status and can shut down the system on low battery
conditions
atd* Runs at commands at their scheduled times
autofs Automatically mounts file systems when you use them
bootparamd Allows Sun servers to boot from a Linux box using bootp
crond* Runs the cron daemon
dhcpd* Provides Dynamic Host Configuration Protocol (DHCP) services
gated Runs the gate daemon to provide routing services for Boarder Gateway
Protocol (BGP) and other routing protocols
gpm* Runs the program that provides mouse support to Linux
86 Chapter 3 Installing Red Hat Linux
Table 3.7 Startup Services
Service Description
httpd* Runs the Apache Web server
Inet* Starts the internet super daemon (inetd) that provides all the sevices
specified in /etc/inet.conf
inmd Starts the Usenet news server innd
kerneld* Starts the kerneld daemon, which loads and unloads kernel modules as they
are needed
keytable* Loads the appropriate keyboard map
lpd* Provides printing services to Linux
mars-new Loads the MArs NetWare file and print server daemon
mcserv Provides midcommander remote file services
named* Provides Domain Name Service (DNS) services
network* Provides control of all the network interfaces
nfs* Provides Network File System (NFS) server services
nfsfs* Mounts and unmounts all NFS mount points specified in /etc/exports
pcmcia Provides access to PCMCIA (PC Cards) services
pnserver Starts the Real Media services
portmap* Provides Remote Procedure Call (RPC) support for other protocols like
Network File System (NFS)
postgresql Runs the postgres database and provides Structured Query Language (SQL)
services
random* Saves and restores a random value to help generate better random numbers
(which are used for various security systems)
routed Provides for automatic router table updates using the Routing Information
Protocol (RIP)
rusersd Provides services that allow users to find one another over the network
rwalld Enables users to use the rwall command to write messages on remote
terminals
rwhod Provides remote users with a list of all of the users logged into a machine by
running the rwho daemon
sendmail* Runs the sendmail daemon to provide email
smb* Provides Session Message Block (SMB, or Samba) client/server services
snmpd* Provides Simple Network Management Protocol (SNMP) support to Linux
sound* Provides access to sound cards
squid* Runs the squid proxy Web server
syslog* Provides logging capability to your Linux system
xntpd Starts the NTPv3 daemon
ypbind Binds YP/NIS clients to a yellow pages server
Configuring Your Network 87
Table 3.7 Startup Services
Service Description
yppasswd Allows users to change their passwords on systems running YP/NIS
ypserv Provides NIS server functions
Tip #21 from
Jack Although you can manually change the services that run on startup by editing the
appropriate rc.d files (see Chapter 10, ``Booting and Shutting Down''), you can use
the /usr/sbin/ntsysv command to go back to the Services dialog box and
reconfigure the services with the GUI.
Part
I
Selecting Your Root Password Ch
Now you must select your root password. This password is the ultimate key into your system,
3
so you should take some care in choosing it. The superuser, or root, on a Linux/UNIX
system can do great thingsÐand can also wreak awesome damage. You should pick a secure
password and be careful whom you give it to. In the Root Password dialog box shown in
Figure 3.9, you enter the password twice to confirm what you've typed.
Figure 3.9
You must
choose a root
password wi-
sely ``young
Linux Walker.''
Although you can fix users' accounts when they forget their passwords, if you forget the root
password, you might be forced to reinstall the system. However, booting from a floppy and
editing the password file may allow a recovery.
? See ``Dealing with Password Security,'' p. 277
Installing LILO
Next, you are asked to create a boot disk, as shown in the Bootdisk dialog box in Figure 3.10.
We highly recommend you create a disk, just in case you cannot boot your computer in the
future. A boot disk is your first tool for repair, followed by the rescue disk you created.
88 Chapter 3 Installing Red Hat Linux
Figure 3.10
A boot disk can
help you re-
cover from a
variety of sys-
tem crashes or
corrupted con-
figuration files.
After creating the boot disk, you are asked to install LILO. LILO stands for LInux LOader.
LILO, which is a program executed at system startup, lets you choose which operating system
is used to boot the computer. You can use LILO to boot several different operating systems,
such as Linux and MS-DOS. Press Tab to get a list of operating systems LILO can boot.
With LILO, you can specify a default operating system to boot and a default time limit before
it boots that system. For example, if you have MS-DOS and Linux on your computer, you can
configure LILO to boot either one. You could then tell LILO to boot MS-DOS if no one
intervenes after 30 seconds. Before that 30 seconds is up, however, a user can specify another
operating system to boot instead of the default. You can press the Ctrl, Alt, or Shift key to
stop the timed process.
You specify all this information while configuring LILO. You can later directly edit the
lilo.conf file located in the /etc directory. If you don't want to boot Linux automatically,
you can select the Skip button to continue. Otherwise, select a hard drive to install LILO and
press Enter to edit the entry.
Congratulations! After you load LILO, your system is up and runningÐand, let's hope,
problem free.
Troubleshooting
After your machine reboots, the LILO prompt should appear. Make sure that you can boot to
your old operating system if you left it on the hard drive. If that system was DOS, you can
press Shift and then type the short word you used to identify the DOS partition when you
installed LILO. If you enter an invalid word, you can press Tab to get a list of valid operating
system types. If you're having problems at this point, place your DOS boot disk in the boot
drive and reboot.
? See ``Troubleshooting,'' p. 55
You should be able to boot from your rescue disk. When your system is up and running under
DOS, try the Linux rescue disk you created during installation. If that rescue disk doesn't
work, you might have to reinstall Linux. Potential problems to check initially are the kernels
and your hardware. Before starting over, make sure that you have the appropriate hardware. If
Case Study: Installing Red Hat Linux on DEC Alphas 89
you made notes during the installation process, check which kernel you installed against what
hardware you have.
Going Back to the Beginning
After you complete the setup and configuration of your system, you should reboot the system
so that all your setup and configurations can take effect.
Rebooting Linux is more involved than rebooting DOS; you can't just turn off the power and
turn the system back on. If you do so in Linux, you can damage the file structures and
systems. Linux tries to repair itself on boot-up. So, remember, don't turn off the power while
you're running Linux.
? See ``Performing Backups and Restoring Files,'' p. 263
Part
To exit Linux, you use the following command:
I
shutdown [-r] time
Ch
-r time
The optional indicates that the system should reboot after shutting down, and
3
indicates the time that the system should shut down. You can use now in place of time to
indicate immediate shutdown. Linux also recognizes the warm boot keys used by DOS to
reboot the machine, Ctrl+Alt+Delete, which Linux interprets as the command
shutdown -r now
When you shut down this way, make sure that you've removed the root disk from the drive,
and reboot your new Linux machine.
Case Study: Installing Red Hat Linux on DEC
Alphas
Unlike other distributions, Red Hat also provides Linux for Digital Equipment Corporation
(DEC) Alphas. This distribution isn't included on the accompanying Red Hat CD-ROM, but
it's available from Red Hat. For more information on getting this distribution, see Red Hat's
Web site at http://www.redhat.com/products/rhl-alpha.html. When you have the correct
distribution, you can use the following instructions to install Red Hat Linux on an Alpha.
Red Hat also has a distribution for Sun's line of Sun Sparc processors. See Red Hat's
Web site for more information.
Before installing the distribution on an Alpha, you should read the installation instructions in
the first part of this chapter because many of the steps are the same. You also need access to a
computer capable of reading and writing MS-DOS disks because you must create an
installation floppy.
90 Chapter 3 Installing Red Hat Linux
Using Supported Alpha Hardware
Red Hat supports various Alpha hardware from both Digital Equipment Corporation (DEC)
and other vendors. The following hardware is supported:
& AlphaPC64 (Cabriolet, Aspen Telluride)
& AxpPCI133 (Noname)
& EB64+ (Aspen Alpine)
& EB66 (NekoTek Mach 1)
& EB66+
& Jensen (DEC PC 150, 2000 model 300, Cullean)
& Universal Desktop Box (UDB, a.k.a. Multia)
& AlphaStation 200, 250, 255, 400 (Avanti machines)
& EB164 (Aspen Avalanche, Timerline, Summit)
& Kinetic's Platform 2000 machines
& PC164 (Durango)
& Alcor AlphaStations 500, 600 (Maverick, Brett)
& Alpha-XL
& Alpha-XLT (XL 300, XL 366)
& Mikasa AlphaServer 1000Ðthe 1000A is NOT supported
All these systems have SCSI systems supported by Red Hat Linux. The video systems should
work, too, although S3 support for the Jensen systems isn't included by default. To run the X
Window System with a Jensen system, you need to download the X server from ftp://
ftp.azstarnet.com/pub/linux/axp/jensen. Finally, all Ethernet solutions for these systems
are supported, and the kernels for these machines also support Token-Ring adapters.
The hardware list changes frequently, so you should check the up-to-date list on Red Hat's
Web site at http://www.redhat.com.
Creating the Boot and Root Disks
You need to create a boot and root floppy for an Alpha installation. The boot disk contains a
program image allowing you to boot the system. The root floppy provides an image of the
Linux kernel for the system to use during installation. Just as for Intel machines, you use the
rawrite program to create these disk images.
The boot disk image depends on the type of Alpha used. These images are located in the
/images directory with a README file that provides more information on each image described
in Table 3.8.
Case Study: Installing Red Hat Linux on DEC Alphas 91
Table 3.8 Boot Images Available for DEC Alphas
Image Description
cab.img AlphaPC64, Cabriolet
noname.img AxpPCI33, Noname, Universal Desktop Box (Multias)
eb64p.img EB64+, Aspen Timberlines
eb66.img EB66
eb66p.img EB66+
jensen.img Jensens
avanti.img AlphaStation 200, 250, and 400
xl.img Alpha XL
xlt.img Alpha XLT
Part
eb164.img EB164-based machines
I
p2000.img Platform 2000 Ch
alcor.img Alcor-based machines 3
mikasa.img Mikasa-based machines
To create a boot image for a Universal Desktop Box, you use the command
E:\dosutils\rawrite -f E:\images\noname.img -d a: -n
where E: represents the drive letter of your CD-ROM. After creating the boot disk, you must
create the root disk, which contains the RAM disk image of the Linux kernel. You can create
the root disk by using the following command:
E:\dosutils\rawrite -f E:\images\ramdisk.img -d a: -n
Installing the Main Red Hat Distribution
After you have your boot media ready, you can install Linux. The installation procedure is
very much like the one outlined earlier in the section ``Installing the Linux System.'' The
installation program guides you through the process, prompting you to make selections from
a list of possible choices.
To begin, you place your boot disk into the floppy drive and restart your system. At the
prompt, you enter the following command:
boot fd0:vmlinux.gz root=/dev/fd0 load_ramdisk=1
You might see several SCSI messages flash by onscreen. Don't worry about them unless you
see a message such as scsi0 : 1, which indicates that you have a SCSI termination problem
that needs to be fixed before continuing with the installation. If all goes well, you should see
the message VFS: Insert Root floppy to be loaded into ramdisk. Insert the root disk you
created and press Enter to continue the installation process.
CHAPTER
4
Installing Caldera OpenLinux
In this chapter by Jack Tackett, Jr.
What You Need to Install OpenLinux 94
Installation Methods 94
Making the Preparations 95
Preparing the Installation Floppies 96
Installing Linux 100
Installing the Linux Software System 111
Configuring Your System 112
Installing LILO 114
Going Back to the Beginning 115
Troubleshooting 115
94 Chapter 4 Installing Caldera OpenLinux
What You Need to Install OpenLinux
This chapter gives you the information you need to install the Caldera OpenLinux
distribution. Like Red Hat and Debian, OpenLinux is a complete distribution of a multiuser,
multitasking operating system based on the Linux 2.2.5 kernel. The CD-ROM
accompanying this book contains a noncommercial version of Caldera's OpenLinux
distribution. The version on the CD-ROM is a subset of Caldera's commercial OpenLinux
Base product and as such does not contain programs such as Partition Magic or technical
support.
ON THE WEB
For more information on Caldera's product line, visit the company's Web site at the following URL:
http://www.calderasystems.com
Your system must have the following components to successfully install OpenLinux:
& An 80386 or higher Intel-based PC (Caldera doesn't support other processors at this
time).
& A 3 1/2-inch floppy drive.
& At least 16MB of RAM (32MB or more recommended).
& About 250MB of disk space, but a minimal system without the X Window System
requires only about 100MB. A full installation with everything installed requires
almost 1300MB.
& A mouse and video card supported by XFree86.
To briefly summarize, you need to partition your hard drive, create the boot floppy, and then
install and configure the system.
You're about to make major changes to your system, so be careful. If you intend to
install OpenLinux on your current Windows machine, you should make a backup
copy of your system first.
Installation Methods
To get started with your installation, you first need a distribution of Caldera OpenLinux,
which is supplied on the accompanying CD-ROM. To start the installation process, you need
to decide on an installation method. OpenLinux allows you to boot and install from the
CD-ROM if you have a drive capable of booting from CD-ROM. Caldera also enables you to
create the installation disks from Windows 98 or Windows NT. Finally, you can install the
Making the Preparations 95
system onto a Windows machine if you have enough disk space to create a proper installation
partition.
Next, you must decide how you intend to boot Linux. You have two choices:
Booting a system refers to starting up the system's operating system and is a
reference to the phrase ``pulling yourself up by your bootstraps.''
& You can boot Linux from a floppy disk, in which case you need an extra formatted
diskÐfor a total of three disks.
& You can use a program called LILO (the Linux Loader). LILO allows you to specify
which operating system to boot. Such programs as OS/2, Windows 98, and Windows
NT provide similar functionality.
Next, you should make sure that you have enough disk space to install Linux. Most people
can get by with 1000MB devoted to Linux; however, you can get by with less space if you
don't plan to use such applications as TeX and the X Window System.
Part
Having paper and pen nearby is a good idea, so you can take notes just in case something goes
wrong. Besides, you'll need to jot down some numbers along the way. For configuring
I
XFree86, the X Window System program distributed with Linux, you should write down Ch
what type of chipset your video card uses. If you have a serial mouse and modem, you should
4
write down the serial port that each uses. You'll need this information later during the
configuration process. The information gathered here will be used later on to help configure
your new system. The time that you spend listing all the information now will reduce any
aggravation later.
Making the Preparations
If you have a brand new system, or if you have an existing system but don't care what happens
to the data already stored on the computer, you can skip most of the following sections and
go directly to ``Creating the Install and Modules Disks.'' If, however, you're already using a
system and you simply want to add Linux, you must do some planning because Linux is
another operating system, not just a collection of programs.
First, you must decide whether you want to erase the current contents of your system or
whether you want to consolidate space and install OpenLinux in the free space generated.
In general, when you install LinuxÐa new operating systemÐyou must do the following if
you cannot boot from the CD-ROM:
& Create the Linux boot disksÐYou must create two floppies because you need to
bootstrap Linux onto the new system.
96 Chapter 4 Installing Caldera OpenLinux
& Repartition the hard drive to make room for LinuxÐRepartitioning a hard drive
may cause problems because it destroys any data stored on the affected partitions.
& Boot LinuxÐAfter making room for Linux, you need to boot the Linux system to
gain access to the tools required to create its new partitions and file systems.
& Create the Linux partitionsÐTypically, Linux systems need a primary partition to
store the files on and a swap file partition, especially if you have a machine with 32MB
or less of memory.
& Create the file systemsÐA file system is basically a section of your hard drive specially
formatted to hold files. UNIX and Linux use file systems to represent entire sections
of the directory tree. This use is in contrast to MS-DOS, which places subdirectories
in the directory tree on the same logical drive. UNIX systems use the file system
structure because placing subdirectories on different drives is safer. If one drive
malfunctions, only the information on that drive must be replaced or fixed.
& Install the Linux system and software applicationsÐAfter creating the file systems,
you install the Linux operating system, its support files, and various application
packages distributed with the system, such as the games and networking support
packages.
Preparing the Installation Floppies
You must make an installation disk for your PC. To install Linux, you need to repartition
your hard drive to make room for the new operating system. Unfortunately, you can't simply
copy the files over to an MS-DOS, OS/2, or Windows file system.
If your system can boot a CD-ROM, or if you have free partitions on your current hard drive,
or if you don't mind destroying your current file system, you can install directly from the
CD-ROM and not bother with making the installation disk. Making the modules disk is
recommended, however, because you might need one of the drivers found on that disk.
Tip #22 from
Jack If your system uses a Buslogic SCSI controller, you definitely need the modules disk.
To launch an OpenLinux installation from the CD-ROM, change the directory to
d:/col/launch (using the series of commands d:, cd col, and then cd launch)
and proceed according to the instructions in that directory's README.us.txt file.
Remember to create the modules disk discussed in the next section.
Preparing the Installation Floppies 97
If you use floppies, you need to decide on an installation program. Caldera provides two:
Lizard and LISA. The Lizard installation, which is a graphical user interface (GUI) program,
provides a user-friendly installation process, if it works. Unfortunately, if you need to specify
unique parameters to the installation programÐfor example, to use hardware drivers
available only on the modules diskÐthen the Lizard installation program is not a viable
option; you should use LISA instead. LISA, which stands for the Linux Installation and System
Administration program, provides a text-based menu system to first install OpenLinux and
then help administer the system.
Creating the Installation and Modules Disks
For the next step, you need to create the installation disks for Linux. Two floppies are needed
with OpenLinux: the installation and the modules floppies. You create these floppies in one
of two ways:
& From the CD-ROM installation program
& With an MS-DOS program called rawrite that's provided with most Linux
distributions
Creating the Installation Disks from the CD-ROM Part
To start the Windows Setup program (which is a very nice multimedia program) from the I
CD, simply double-click the setup icon in the winsetup folder. The main setup screen then Ch
appears, as shown in Figure 4.1.
4
Figure 4.1
Caldera offers
a variety of
multimedia
presentations
as well as a
variety of in-
stallation op-
tions.
From the main menu screen, you can do the following:
& Install Products
& View Online Documentation
& View Our Web site ( www.calderasystems.com)
& Multimedia Presentations
& Register Now (not supported on enclosed CD)
98 Chapter 4 Installing Caldera OpenLinux
& Contact Us
& Support and Training
To install OpenLinux, select the first menu option, Install Products. This selection displays
the installation screen shown in Figure 4.2.
Figure 4.2
Caldera allows
you to create
installation
disks from
Windows and
to install under
Windows.
The first selection, Launch Linux Install, must run in DOS mode so it closes all your running
programs and starts the installation process. You must prepare your hard drive prior to
selecting this option (See ``Preparing the Hard Disks'' later in this chapter.)
The next selection, Create Floppy Install Diskettes, provides you the choice of creating
floppies to use either the text-based installation program LISA or the GUI Lizard installation
program. LISA provides a more robust installation than Lizard if you need to specify specific
hardware during installation. After you create the necessary disks on your PC, you can install
the system at any time.
Tip #23 from
Jack The commercial version of OpenLinux provides several utilities to allow you to
repartition and install Linux on a Windows PC. If you want to install the program on a
PC running Windows, you should choose the Partition & Launch Linux Install
selection. This option installs a version of Partition Magic on your PC. You then can
repartition part of your hard drive to create enough space to install Linux (between
250MB and 1300MB).
Preparing the Installation Floppies 99
Repartitioning your drive is a dangerous undertaking. Make sure you have backups of
your system on zip disks, tape, or CDRs. An even better solution is to allocate an
entire drive to Linux.
To create the installation disks from the Setup program, select Create Floppy Install
Diskettes, which displays the screen shown in Figure 4.3.
Figure 4.3
OpenLinux
provides gra-
phical tools to
create the Li-
zard and LISA
installation
disks directly
from the
CD-ROM.
Part
I
You should make sure you have two blank floppies available to create the disks. Ch
4
Creating the Installation Disks from DOS
To create the installation disks from DOS, you use rawrite, which writes the contents of a file
directly to a floppy without regard to the format. You use rawrite to transfer the images to the
appropriate floppies.
The examples in this chapter assume that your CD-ROM is drive D:. If it is not, you
need to substitute the appropriate drive letter for your CD-ROM.
To create the installation disk, you can issue the following command:
D:/col/launch/floppies/rawrite3.com
This command starts the rawrite program, and you can follow the prompts. When you're
asked for the file, indicate the appropriate installation file. For 1.44MB floppies, use the
install.144 file (as inD:/col/launch/floppies/install.144). For 2.88MB installation
floppies, use the install.288 file.
Next, following the same instructions, you can create the modules disk. This disk can be
placed on either a 1.44MB floppy or a 2.88MB floppy. Specify the filename D:/col/launch/
floppy/modules.144.
100 Chapter 4 Installing Caldera OpenLinux
Installing Linux
To use the disks you just created, you simply place the installation disk into the drive and
reboot your system. If you are installing from a DOS system, you should read the instructions
found in the file d:/col/launch/dos/README.us. No matter which way you boot the system,
the installation program displays a splash screen and welcomes you to Caldera's OpenLinux
product. You are then asked to boot the system.
If you have any parameters to pass to the kernel before it boots, you can enter them at the
boot: prompt. (For information on these parameters, see the ``BootPrompt How-To'' in
/doc/HOWTO.) If you don't need to pass on any parameters, press Return to continue the
installation.
The system now goes through a series of probes trying to determine what type of equipment
your system uses. Then the Linux Installation and System Administration (LISA) program
begins. During the installation, you can maneuver around the various dialog boxes by using
the cursor keys to select choices from list boxes. The Tab key moves you from one item to
another in a dialog box, such as from list boxes to buttons. At any time, you can press the
Escape key to cancel a selection. To enter your selections, press the Return key.
Tip #24 from
Jack During installation, you can use the key sequence Alt+F6 to see the progress of the
installation.
On the first screen, you can select the language to be used during installation. You can choose
from English, German, French, Italian, Spanish, or Portuguese. After you make a selection,
LISA asks you to select a keyboard to use. The Linux system uses configuration information
based on your selection to bind certain keys to certain language characters.
Using a Previous Configuration
Next, LISA asks whether you want to use a previously saved configuration. The Caldera
distribution allows you to create several installation configurations and save them. If you want
to reinstall at a later date, you can reuse this predefined configuration so that you don't have
to go through the entire configuration process again. Because this is your first time installing
the OpenLinux system, you can simply answer No.
Configuring LISA
Next, you must configure LISA. Typically, the defaults shown on the Change LISA Setup
dialog box are acceptable. If you need to make changes, you can select the following options
and make the appropriate change:
Installing Linux 101
& Disable Plug-and-Play CardsÐThis option turns off the BIOS setting that interacts
with Plug-and-Play cards. During autoprobing and configuration, these cards can
cause problems with Linux.
& Automatic Network Configuration with bootpÐThis option allows another
computer to issue network configuration to the current machine via the bootp
protocol. At least initially, you should configure your own network options instead of
trying to use bootp.
& Automatic Network Configuration with NetprobeÐThis option provides for
network setup using Netprobe. Netprobe is a Caldera product similar in function to a
computer running bootp, which allows remote network configuration.
& Use Selection and ContinueÐThis option tells the program to accept your
selections and continue with the installation.
Probing for Hardware
Next, the system autodetects all the hardware it can. If LISA does not locate all your
hardware, you have to use the modules disk you created earlier to load the appropriate
hardware drivers. First, the system probes for IDE and ATAPI equipment. You can review
the hardware list displayed by the Hardware Found (IDE/ATAPI) dialog box to see whether Part
all your hardware has been detected. When you click the Continue button, LISA asks I
whether all hardware has been detected. If not, select No and continue with the hardware
Ch
probes.
4
If the probe still does not find all your hardware, LISA displays the Kernel Module Manager
dialog box. This dialog box gives you the following options:
& Continue with Installation
& Analyze Kernel Modules
& Load Kernel Modules
& Remove Modules
During installation, your typical selections are to load and analyze kernel modules. How
much hardware you have to add support for determines how many times you cycle through
the various dialog boxes associated with the Kernel Module Manager.
Analyzing Kernel Modules
You can select the Analyze Kernel Modules function if you want to see what hardware was
detected. You can also use this function to see what modules you or LISA have added through
the course of installation. You can also review any messages the system has generated during
the boot process. Table 4.1 describes each function in the Analyze Kernel dialog box.
102 Chapter 4 Installing Caldera OpenLinux
Table 4.1 The Analyze Kernel Functions
Function Description
Return to Previous Screen Returns you to the Kernel Modules dialog box.
Show Hardware That Has Already
Been Found Displays all the hardware found so far in your
system.
Show Loaded Kernel Modules Displays all kernel modules currently installed on
your system.
Verbose System Analysis Provides more detailed information messages dur-
ing installation.
Display Boot Process Messages Lists all information generated during the boot
process (similar to the information displayed by the
Alt+F6 key sequence). This information includes
the results of probing your system and installing
various kernel modules.
Store Information on a DOS Floppy Allows you to create a copy of all the information
available from the Analyze Kernel Module func-
tions.
Loading and Removing Kernel Modules
The Load Kernel Modules function allows you to load various device drivers into Linux to
support your hardware. These drivers are on the modules disk you created earlier. Table 4.2
lists the various subfunctions available in the Load Kernel Modules dialog box.
Table 4.2 The Load Kernel Modules Functions
Function Description
Return to Previous Menu Returns you to the Kernel Modules dialog box
Load Driver for CD-ROM Allows you to select a driver for your CD-ROM from the
modules disk
Load Driver for SCSI Adapter Allows you to load a driver for your system's SCSI
adapter
Load Driver for Network Card Allows you to select a driver for your Ethernet card
To load a driver not currently available in the installation program, you must remove the
installation disk from the floppy drive and then insert the modules disk. For example, several
SCSI controllers are available from the installation disk, including Adaptec 2940s. However,
the default installation disk does not contain drivers for Buslogic adapters. Therefore, to
support a Buslogic controller, you need the modules disk. You are led through a series of
dialog boxes to select the desired driver and also to provide any additional configuration
Installing Linux 103
information for the device. LISA offers context-sensitive help throughout the process; you can
simply press F1 at any time to get help.
You should work your way down the function list, first installing support for your CD-ROM,
then for any SCSI devices, and then for your network card. If you find you have installed the
wrong driver, or if LISA's autoprobe installed the wrong driver, you can select the Remove
Kernel Module function and indicate which module is to be removed. Next you need to
prepare a place on your hard drive to install Linux.
Preparing the Hard Disks
MS-DOS refers to most partitions and hard drives with a letter such as C or D. Linux refers
to them in a very different manner. Linux refers to everythingÐdevices, files, and so onÐin
the same manner.
Linux and MS-DOS communicate with hardware via a series of programs called device drivers.
Whereas MS-DOS device drivers usually have an .SYS extension and can reside anywhere on
the system, Linux stores all such device drivers in the /dev directory.
What Microsoft and Windows folks call device drivers are usually called modules in
Part
Linux.
I
Ch
The drivers Linux uses in installation were specified using the Kernel Modules dialog box. 4
The important point to remember, though, is that because the hard drive, floppy drives, and
CD-ROM drives are hardware, Linux uses device drivers in the /dev directory to access the
drives. Linux also references these drives by their subdirectory names instead of by letters.
Table 4.3 displays a typical Linux device directory.
Table 4.3 Linux Devices
Device Filename
Floppy drive A /dev/fd0
Floppy drive B /dev/fd1
First hard drive /dev/hda
First primary partition on hard drive A /dev/hda1
Second primary partition on hard drive A /dev/hda2
First logical partition on hard drive A /dev/hda4
Second hard drive /dev/hdb
First primary partition on hard drive B /dev/hdb1
First SCSI hard drive /dev/sda
104 Chapter 4 Installing Caldera OpenLinux
Notice that the entire hard drive is referred to as /hdletter. The primary partitions are given
the next set of four numbers, followed by the logical partitions. Thus, logical partitions
always start at /dev/hda4. SCSI hard drives and CD-ROMs follow the same convention,
except that the hd is replaced by sd.
Repartitioning Your DOS Drive
This section assumes that you need to repartition a DOS drive. First, you execute FDISK by
typing fdisk at the DOS prompt. The FDISK Options screen appears (see Figure 4.4).
Figure 4.4
From the FDISK
Options
screen, you can
look at current
partitions, cre-
ate new parti-
tions, and
delete old par-
titions.
The screen shown in Figure 4.4 might look different depending on which version of
MS-DOS you're using. Pick menu option 4, Display partition information. The Display
Partition Information screen appears (see Figure 4.5). Write down the information in this
screen. You need the current partition table information if you decide to abort the Linux
installation and put your system back the way it was before you started.
Figure 4.5
You can look at
current parti-
tion informa-
tion by using
the Display
Partition Infor-
mation screen
in MS-DOS 6.x.
An Alternative to Repartitioning Your Hard Drive
You may not need to repartition your hard drive, although it's thought that repartitioning offers the best
introduction to Linux. You can use FIPS to non-destructively repartition your hard drive.
FIPS stands for First non-destructive Interactive Partition Splitting . A program developed by Arno Schaefer as a
result of the Linux project, FIPS is used to move around DOS partitions to make room for Linux partitions.
Installing Linux 105
fips.doc located on the accompanying
You can find the complete instructions for using FIPS in the document
Red Hat CD-ROM in the /utils/fips directory. This program can help only if you have enough free space left
on your drive to install Linux; otherwise, you need to delete unneeded files or use the process described earlier
to repartition your hard drive.
Deleting Partitions
Unfortunately, FDISK doesn't allow you to simply resize a partition; you must first delete the
partition and then add it back with the desired size. From the FDISK Options screen, choose
menu option 3, Delete Partition or Logical DOS Drive, which deletes the necessary
partitions. The Delete DOS Partition or Logical DOS Drive screen appears (see Figure 4.6).
Figure 4.6
Use the Delete
DOS Partition
screen to de-
lete a specific
partition or lo-
gical drive.
Part
I
Ch
Pick the appropriate menu option for the type of partition you're deleting, such as a primary
4
DOS partition. For example, option 1, Delete Primary DOS Partition, allows you to delete
primary DOS partitions.
Choose option 1 to display the Delete Primary DOS Partition screen (see Figure 4.7). The
screen asks for a volume name of the partition and then a confirmation to see whether you
really want to delete the partition. Because all information on the partition will be destroyed,
FDISK wants to make absolutely sure that you want to delete the primary DOS partition.
Figure 4.7
MS-DOS warns
you when you
try to delete a
primary DOS
partition.
106 Chapter 4 Installing Caldera OpenLinux
Adding Partitions
After you delete all the necessary partitions, you must add the appropriate partitions for your
DOS system by selecting the Create a DOS Partition menu item on the FDISK Options
screen. Figure 4.8 shows the Create a DOS Partition or Logical DOS Drive screen.
Figure 4.8
Most operating
systems re-
quire a primary
active partition
to boot prop-
erly.
You can't add the Linux or OS/2 partitions with the DOS FDISK program. Partitioning
the hard drive for Linux is covered later in the section ``Using the Linux fdisk
Command.''
Providing all the space available for the partition and making the partition the active partition
are the FDISK defaults, as shown in Figure 4.9.
Figure 4.9
You can use all
the disk space
for one parti-
tion or spread
out the free
space across
several parti-
tions.
Active indicates that the partition is bootable. To boot DOS, you must specify the primary
partition as active. Choose N (no) for this first selection so that you can specify the exact
amount of disk space to provide to your DOS partition. Answering no to the question in
Figure 4.9 displays the Specify Disk Space for the Partition screen. Specify the desired space
for your DOS partition either in megabytes or in percentage of space available and press
Return.
Installing Linux 107
Next, you must set this partition as active. From the FDISK Options screen, choose menu
option 2, Set Active Partition, and simply follow the instructions on the set active menu
screen.
Formatting the Partition
After you repartition your hard drive, you need to prepare the new partition for DOS and
restore the appropriate files back to the DOS partition. Reboot your computer with the boot
disk you made earlier. Then format the appropriate drive and transfer the system files by
using the following DOS command:
format c: /s
When the partition is formatted, you can restore your backup to the new drive. Remember, if
you reduced the size of the partition, not all the files will fit on the new drive. It might be
necessary to place the files that don't fit on the new drive onto other DOS drives or partitions.
Using the Linux fdisk Program
At the fdisk prompt, you can enter m for a list of commands. Table 4.4 shows a list of
available commands.
Part
Table 4.4 The Linux fdisk Commands I
Command Description
Ch
4
a Toggles a bootable flag
c Toggles the DOS compatibility flag
d Deletes a partition
l Lists known partition types
m Displays this table
n Adds a new partition
p Prints the partition table
q Quits without saving changes
t Changes a partition's system ID
u Changes display/entry units
v Verifies the partition table
w Writes the table to disk and exits
x Provides extra functionality (experts only)
To begin the process of partitioning, you select the p command to display the current
partition table, which should reflect the drive you partitioned earlier with the DOS FDISK
program. Listing 4.1 shows a possible display from the p command.
108 Chapter 4 Installing Caldera OpenLinux
Listing 4.1 An Example of a Current Partition Table
Disk /dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
dev/hda2 1024 1024 4040 384667+ 51 Novell?
Partition 2 has different physical/logical endings:
phys=(967, 14, 17) Logical=(4096, 14.17)
You might see different information than what's shown here because the values are
different for each type of drive and the partitions are already defined on that drive.
The display in Listing 4.1 indicates the various partitions already defined, the starting and
ending locations of the partitions, and how big each partition is in blocks. The display also
indicates the partition type. Table 4.5 lists all the different partition types you can define by
using the Linux fdisk command. The primary partitions used here are 83-Linux Native and
82-Linux Swap. You can get a similar listing by using the l command.
Table 4.5 The Known Linux Partition Types
Reference Number Type
0 Empty
1 DOS 12-bit FAT
2 XENIX root
3 XENIX usr
4 DOS 16-bit < 32MB
5 Extended
6 DOS 16-bit >=32MB
7 OS/2 HPFS
8 AIX
9 AIX bootable
a OS/2 Boot Manager
40 Venix 80286
51 Novell?
52 Microport
63 GNU HURD
64 Novell
75 PC/IX
80 Old MINIX
Installing Linux 109
Table 4.5 The Known Linux Partition Types
Reference Number Type
81 MINIX/Linux
82 Linux Swap
83 Linux Native
93 Amoeba
94 Amoeba BBT
a5 BSD/386
b7 BSDI fs
b8 BSDI swap
c7 Syrinx
db CP/M
e1 DOS access
e3 DOS R/O
f2 DOS secondary
ff BBT Part
I
Notice the note about the different physical and logical endings at the end of Listing 4.1. The
Ch
difference exists because, on the system used to write this chapter, a prior partition containing
4
the DOS D drive was left intact, whereas the C drive was repartitioned to a smaller C drive to
make room for Linux. As a result, space exists between the C drive and D drive. The
necessary partitions required by Linux will be created here.
The begin, start, and end numbers from the display are very important. You should write
them down because you'll need them in a later step to specify the necessary sizes of the
partitions you'll add.
Adding the Necessary Partitions
Because you've repartitioned the drive for MS-DOS, you shouldn't have to delete any
partitions for Linux. You should only have to add partitions. A standard set of partitions
should include the following:
& A / (root) partition for the whole system
& A swap partition for the swap file
& A /usr partition for software
& A /home partition for user directories
& A /var partition for log files
To add a partition, issue the n command, which displays the following:
Command Action
110 Chapter 4 Installing Caldera OpenLinux
e extended
p primary(1-4)
Now, you can press the p key and then Return. When fdisk asks for the partition number,
enter your selection and press Return. If you indicate a partition number that's already in use,
fdisk reports this fact and asks you to delete the partition before trying to add it to the
partition table. For this example, enter 3 to add a third primary partition that's referred to as
/dev/hda3.
Next, fdisk asks for the location of the first cylinder. This is usually the first available cylinder.
In fact, fdisk displays a default range for your selection, such as the following:
First cylinder (42-1024) :
From the preceding example, you can infer that the first partition ends at cylinder 41, and the
next partition begins at cylinder 1024. Thus, the range supplied by fdisk allows you to start
the next partition anywhere in the range of 42 to 1024. It's a very good idea not to place
partitions just anywhere throughout the disk, so choose the next available locationÐin this
case, cylinder 42.
Linux can have trouble booting from partitions defined to start at cylinders above
1024. If you create a Linux partition starting above 1024, you might have to boot
Linux from a floppy. You'll learn how to create a boot floppy (which is different than
the boot floppy used for installation) later in this chapter. The only downside is that
booting Linux from a floppy takes a little longer than booting from the hard drive.
Now fdisk wants you to specify how much space to allocate for this partition. You can express
this size in number of cylinders or by the number of bytes ( +size), kilobytes (+sizeK), or
megabytes ( +sizeM). Because you should already know the approximate size you need for the
swap file, you can define this partition first and then leave the rest of the disk space for the
Linux program partitions. For this example, because your machine has 8MB of RAM, you
need to specify a 16MB partition size by replying like this:
Last cylinder or +size or +sizeM or +sizeK (42-1023): +16M
You should then use the p command to look at the new partition table you've defined. In this
example, the new partition table looks like this:
Disk /dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
/dev/hda1 * 1 1 41 5219 1 DOS 12-bit FAT
/dev/hda2 1024 1024 4040 384667+ 51 Novell?
Partition 2 has different physical/logical endings:
phys=(967, 14, 17) Logical=(4039, 14.17)
/dev/hda3 42 42 170 16447+ 83 Linux native
Installing the Linux Software System 111
By default, fdisk makes the new partition a Linux native type. To change this type to a swap
partition, you need to use the t command. To do so, enter t and then enter the partition
number you want to change; in this example, enter 3. fdisk then requests that you enter the
hexadecimal value of the desired partition type (refer to Table 4.5). If you don't have Table
4.5 handy, you can enter l to get the list of partition type codes. Because you want a swap
partition in this case, enter 82 at the prompt.
As you can see, fdisk reports the new partition type, but you can also use the p command to
verify that partition 3 is now a Linux swap partition.
Now you can add your Linux partitions. For this example, you will add only one partition.
But if you want multiple partitions for various reasons, you can add them at this time. To add
a partition, enter n, specifyp for another primary partition, and then specify the number for
this partition, which is 4. To keep from fragmenting different partitions across the drive, start
the last partition where the first left off, at cylinder 171. Because you want to use the rest of
the space for the Linux system, you can specify the last cylinder instead of an exact byte count.
Thus, enter 1023, as in the following example:
Command (m for help): n
Command action
e extended Part
p primary partition (1-4)
p I
Partition number (1-4): 4
First cylinder (171-1024): 171 Ch
Last cylinder or +size or +sizeM or +sizeK (171-1023): 1023
4
Now use the p command to verify the new partitions. If you need to make any changes, you
can do so now.
When you're satisfied with the layout of your partitions, you can use the w command to write
the partition table information to the hard disk. None of your changes are permanent until
you use the w command; thus, if you feel you've made some changes in error, you can use the
q command to exit without altering the partition table. When you issue the w command,
Linux tells you that the partition table has been altered and then resynchronizes the disks to
match the new partition table. If your Linux system hangs at this point, you can reboot with
the installation disk.
Creating the Swap Partition
After you partition your system as you see fit, LISA asks you to set up a swap partition. From
the Configure Swap Space dialog box, select the partition you created for swap and select the
Continue button. LISA then configures, formats, and activates the swap area.
Installing the Linux Software System
Now that the system is partitioned for Linux, you can install the various software packages
that form the OpenLinux system. LISA displays the Installation Source Selection dialog box.
112 Chapter 4 Installing Caldera OpenLinux
From here, you can select the installation media: CD-ROM, hard disk, or Network File System
(NFS).
Installing from the CD-ROM is a breeze with the enclosed CD package. However, you can
copy the contents of the CD-ROM to a hard drive partition and access it from there. Also, if
you have a working network connection, you can connect to another computer and install
from that machine. You might want to perform such an installation if OpenLinux cannot
recognize your CD-ROM.
After you make your selection, follow the instructions to begin software installation.
(Remember that you can access help by pressing F1.) For example, if you choose to install
from the CD-ROM, you are asked to select the appropriate hardware device. LISA typically
already has this device selected.
/
You then need to select the root partition ( ) to which all the software will be copied. After
you select from the list of partitions, LISA formats and prepares the root partition for
installation. During this time, you see a screen of numbers and hear lots of disk activity, but
don't worry, this activity is normal.
After preparing the root partition, LISA asks whether you want to place other directories
under the root onto their own partition. Be careful here: The default answer is No, but you
do want to place the /usr, /home, and /var directories onto the partitions you created earlier.
So answer Yes and repeat the process for each directory.
After creating the mount points for your file systems, LISA asks you what packages you want
to install. A basic system requires about 100MB; a full installation requires around 1300MB.
You can select from the following list:
Minimal System without the X Window System (94MB)
Minimal System with the X Window System (152MB)
Small Standard System (192MB)
Standard System (754MB)
All Packages (1300MB)
LISA generates a package listing after you select the installation size. LISA then begins the
automated package installation. You can relax, sit back, and watch the installation.
Configuring Your System
After package installation, you need to configure your system. (Here you'll need the system
and network information you gathered earlier in this chapter.) If you are connected to a
network, you need the appropriate information from your service provider or network
coordinator. You also need to select a root password. This password is important, so choose
wisely and don't forget it.
Configuring Your System 113
Tip #25 from
Jack For complete details on using LISA see http://www.calderasystems.com/doc/
openlinux/english/install.html.
To begin, LISA asks for a host name. This is the name others on the network will call your
machine. You also need to provide your domain name, which is typically something like
company.com. The host name and domain name create the fully qualified domain name for
your computer (such as opus.netwharf.com).
? See ``IP Addresses,'' p. 604
You need to provide the IP address of your computer, along with its netmask and the default
gateway. After supplying this information, you need to specify your network's domain name
server, or DNS machine.
After setting up your network, you need to configure your clock and time zone information.
It is recommended that you use local time, even though most Internet servers use Greenwich
mean time (GMT), because most PCs have their BIOS clocks set to local time, not GMT.
Setting a PC to GMT may cause problems, especially if you use other operating systems on
Part
the machine. At this point, select local or GMT time, and then specify your time zone, such
as EST.
I
Ch
Next, you must specify what type of mouse you are using. Most ATX systems use a PS/2 style
mouse. If you are using a serial mouse, make sure you remember which serial port your
4
mouse uses. The next step is to select the printer you intend to use with your system, if
necessary.
? See ``Understanding Key Terms Used in This Chapter,'' p. 164
Now you must select a root password. This password, which is for the superuser's account,
allows anyone to do anything he or she wants to your system. Do not give this password to
just anyone! And do not forget the root password. If you do forget, more than likely you will
have to reinstall the system.
See Chapter 13, ``Improving System Security,'' for password tips and rescue options.
After setting your root password, you need to create your first user account. LISA uses a
default value of col (for Caldera OpenLinux), but you can specify any name you wantÐeven
your own. This first user account allows you to use the system as a regular user instead of as
the superuser (root). Typically, you should not use the root account for day-to-day user tasks
because of the potential for creating problems. After you specify the new account name, you
114 Chapter 4 Installing Caldera OpenLinux
can simply accept the default values for the other items requested. These fields, such as group, are
explained in Chapter 11, ``Managing User Accounts.''
? See ``Adding a User'' p. 250
Next, you indicate how you want your system to boot.
Installing LILO
As mentioned earlier, LILO stands for the Linux Loader. LILO is a program executed at
system startup; it lets you choose which operating system will be used to boot the computer.
You can use LILO to boot several different operating systems, such as Linux and MS-DOS.
With LILO, you also can specify a default operating system to boot and a default time limit
the system should wait before it boots that system. For example, if you have MS-DOS and
Linux on your computer, you can configure LILO to boot either one. You can tell LILO to
boot MS-DOS if no one intervenes within 30 seconds. Before that 30 seconds is up, however,
a user can specify another operating system to boot instead of the default. You can press the
Ctrl, Alt, or Shift key to stop the timed process. You can press Tab to get a list of operating
systems LILO can boot.
You specify all this information while configuring LILO. Although you can directly edit the
lilo.conf file located in the /etc directory, the LILO installation screen provides a better
interface for editing the file.
After you configure your system, Setup lets you install LILO with the option to configure
LILO.
Uninstalling LILO
If you're running LILO version 0.14 or newer, you can uninstall LILO by using the following
command:
opus:~# lilo -u
If you have a previous version, you must remove or disable LILO from its primary partition
first. You can use the Linux fdisk or MS-DOS FDISK program to make another partition
active.
If you placed LILO within the MBR (master boot record), you must replace it with another
MBR from another operating system. With MS-DOS 5.0 or above, the command c:\>fdisk
/mbr restores the MS-DOS MBR.
When LILO is removed from the active partition or the MBR, you're free to remove the files
from /etc/lilo.
? See ``Removing Files or Directories,'' p. 426
Troubleshooting 115
Going Back to the Beginning
After you complete the setup and configuration of your system, the Setup program returns
you to the main menu. From there, you can choose the Exit option to leave Setup. If you want
to change options, you can do so here. (In case you don't change the options during
installation, Chapter 7, ``Upgrading and Installing Software,'' provides information on
updating and installing software after your initial installation.) For now, choose Exit to leave
the Setup program.
Choosing Exit returns you to the system prompt, indicated by the # sign. You're now in
Linux and can issue simple commands, such as ls for a directory listing of files. At this time,
though, you should reboot the system so that all your setup and configurations settings can
take effect.
Rebooting Linux is more involved than rebooting DOS. You can't turn off the power and
turn the system back on. If you do so in Linux, you can damage the file structures and
systems. Linux tries to repair itself on bootup. You should not turn off the power while
running Linux. To exit Linux, you use the following command:
shutdown [-r] time
Part
The optional -r flag indicates that the system should reboot after shutting down. time
indicates the time that the system should shut down; you can use now in place of time to I
indicate immediate shutdown. Linux also recognizes the warm-boot keys used by DOS Ch
(Ctrl+Alt+Delete) to reboot the machine, which Linux interprets as the command
4
shutdown -r now
? See ``Shutting Down Linux,'' p. 243
When you shut down this way, make sure you've removed all the floppy disks from the drive
and reboot your new Linux machine.
Troubleshooting
After you reboot your machine, the LILO prompt should appear. If not, the following list of
LILO messages may help deduce the problems:
LILO Prompt Description
Nothing Indicates the system can not find the LILO program.
L Indicates a possible media error.
LI LILO cannot find the kernel image.
LIL Indicates a media error.
LIL- Indicates a corrupted descriptor table.
LILO Success! LILO susccessfully loaded and you can now
boot your choice of operating systems.
116 Chapter 4 Installing Caldera OpenLinux
If there are errors, check the hardware and your configuration in /etc/lilo.conf. After
booting Linux you need to make sure that you can boot to your old operating system if you
left it on the hard drive. If that system was DOS, press the Shift key, and then type the short
word you used to identify the DOS partition when you installed LILO. If you enter an invalid
word, you can press Tab to get a list of valid operating system types. If you're having
problems at this point, you can place your DOS boot disk in the boot drive and reboot.
You should be able to boot from your boot disk. When your system is up and running under
DOS, try the Linux boot disk you created during installationÐnot the ones you created to
originally install the entire system. If that boot disk doesn't work, you may have to reinstall
Linux. Potential problems to check initially are the kernels and your hardware. Before
starting over, make sure that you have the appropriate hardware. If you made notes during the
installation process, check which kernel you installed against what hardware you have. Make
sure your hardware is supported by Linux.
Below are some answers to common problems.
Q: Can I use a hard drive that has more than 1023 cylinders?
A: The infamous 1023 cylinder question. Yes, but not to boot Linux. You can install Linux on
partitions above the 1023 cylinder, but to boot Linux, the root directory and specifically the
/boot directory must be installed on the first hard drive below 1024.
Q: How do I add arguments for LILO at the prompt?
A: Some hardware requires that extra parameters be fed to the kernel before the kernel will
recognize the hardware. You can accommodate this by editing the /etc/lilo.conf file to
provide the necessary parameters, or you can provide them manually during boot up. See the
LILO HOWTO for more examples of LILO parameters.
Q: Why does LILO hang on LI?
A: This is a symptom of the 1023 cylinder problem addressed previously. If you have installed
the boot system above 1023, LILO will not be able to boot the system. You can try to boot
from a floppy using the rescue disk you made during installation, or you can repartition your
hard drive and reinstall Linux.
Q: The installation will not find the SCSI card.
A: To remedy this, you need to add a boot-time argument such as the following:
LILO: linux qlogicfas=0x230,11,5
This option can be made permanent so you don't have to re-enter it. See the LILO
configuration option append in the lilo.conf man page.
Q: How do I uninstall LILO?
A: If you want to uninstall LILO and reinstall the original boot record, try using this
command
lilo -u /dev/hda
Troubleshooting 117
which represents the boot record of the first IDE drive. Parameters may vary for your
machine; for example, if your first hard drive is a SCSI drive, you would use /dev/sda.
Q: Can I use LILO and Win98 on one installation?
A: Yes, install Windows 98 first and then install Linux. During the installation, tell Linux to
place LILO in the MBR. You can also use a commercial program such as System
Commander. The commercial version of Caldera OpenLinux provides utilities to both
repartition your system and also manage booting multiple operating systems.
Q: How do I mount a CD-ROM?
A: Installing OpenLinux 2.3 places the proper entries in your /etc/fstab file, as follows:
#
# /etc/fstab
#
# You should be using fstool (control-panel) to edit this!
#
# <device> <mountpoint> <filesystemtype> <options> <dump> <fsckor-
der>
/dev/sda1 / ext2 defaults 1 1
/dev/sda5 /home ext2 defaults 1 2
/dev/cdrom /mnt/cdrom iso9660 noauto,ro 0 0
/dev/fd0 /mnt/floppy ext2 noauto 0 0 Part
/dev/sda6 /var ext2 defaults 1 2 I
/dev/da2 none ignore 0 0 0
none /proc proc defaults Ch
/dev/sda7 none swap sw
4
Note the use of noauto for the cdrom entry. Without this setting, Linux will try to automount
the CD-ROM when it boots, which isn't really a problem unless there's no CD in the drive.
If there is not an entry in your fstab file, you can either edit /etc/fstab or use the X
Window Control Panel tool to add the appropriate mount information. Also, make sure the
mount point /mnt/cdrom does indeed exist. If the entry is correct, you can cd to the mount
point and issue the following commands:
cd /mnt
mount cdrom
Q: When the system boots up, I see a message that says I have unknown PCI hardware. What does this
mean?
A: The error unknown PCI device can occur for several reasons. The first and most harmless
one is that PCI isn't responding to Linux's queries in a way it understands, but Linux is able
to keep going. The more common occurrence is that the system hangs on, querying PCI bus
cards, and cannot get any further.
Q: I have installed Linux, and it seems to initially start booting. However, when it gets down to
something called sendmail, the machine seems to hang. What is happening, and what should I do?
A: If, after the install, the machine seems to hang when it reaches certain processes like
sendmail, Apache, NFS, or SMB, there is probably a network problem. The most common
118 Chapter 4 Installing Caldera OpenLinux
cause is that Linux cannot look up the name of the machine you have called the box (if you set up
networking to have a machine name). The machine is currently paused waiting for the network
timeout of DNS lookups and will eventually bring up the login prompt. When you get the prompt,
log in as root and check the usual culprits for a problem.
If you are directly on a network with a DNS server, make sure that the /etc/resolv.conf file
has the correct values for your machine's DNS server. Check with your systems administrator
for the correct values.
If you are using Linux on a network without a DNS server (or if this box is going to be the
DNS server), you will need to edit the /etc/hosts file to have the hostname and IP address so
that the lookups will occur correctly. The format of the /etc/hosts file is like the following
example:
127.0.0.1 localhost localhost.localdomain
192.168.200.1 mymachine mymachine.mynetwork.net
The example machine is called mymachine.
CHAPTER
5
Installing Debian Linux
In this chapter by Jack Tackett, Jr.
About Debian 120
What You Need to Install Debian 120
Preparing to Install Debian 121
Partitioning Your Hard Drive 124
Installing Debian 131
Creating the Rescue and Driver Floppies 132
Installing the System Files 134
Installing the Base System 141
Configuring the Base System 142
120 Chapter 5 Installing Debian Linux
About Debian
This chapter gives you the information you need to install the Debian distribution of Linux.
Though this book leads the way, you might find the need to use the resources provided on the
CD-ROM. First, you must determine whether you have the appropriate hardware. After you
have determined that you have the necessary hardware, you must prepare for the installation.
Following the steps provided in this chapter should ensure a smooth installation.
Debian GNU/Linux, the distribution's official title, is the Linux distribution supported by
the GNU community and not by a company, such as Red Hat or Caldera. The community
subscribes to a set of high-quality standards, as indicated on its Web site at http://
www.debian.org/doc/debian-policy/.
The distribution is maintained by a group of volunteers started by Ian Murdock. Many
Debian proponents feel the only pure GNU/Open Source Linux solution is Debian.
What You Need to Install Debian
When you're ready to install Debian, having paper and pen nearby is a good idea so that you
can take notes just in case something does go wrong. Besides, you'll need to jot down some
numbers and information before starting and along the way. To install Debian, you need the
following:
& Your host name (be careful, you don't really want your boss to see email from
studmuffin, do you?)
& Motherboard
& Video card
& Monitor
& Disk system (EIDE or SCSI)
& Mouse
If you plan to access the Internet or a local network, you need to know the following network
information (most of which you can get from your local area network team at work or from
your Internet service provider):
& The system on your network that you should use as a Domain Name Service (DNS)
server
& Your domain name
& Your computer's IP address
& The IP address of your network
& The netmask to use with your network
& The broadcast address to use on your network
Preparing to Install Debian 121
& The IP address of the default gateway system you should route to, if your network has
a gateway
& Your network interface card, which is typically Ethernet but may be Token Ring
The general installation procedure is as follows:
1. Determine the hardware requirements, including disk space.
2. Determine the installation method, whether from floppies or from CD-ROM.
3. Install the system.
4. Configure the system.
5. Decide how to boot the system.
Debian supports the hardware supported by the Linux kernel, as indicated in the ``Linux
Hardware How-To.''
Tip #26 from
Jack You can find the ``Linux Hardware How-To'' at the following address:
http://metalab.unc.edu/LDP/HOWTO/Hardware-HOWTO.html
You can boot the installation program from either a floppy or the CD-ROM. You need to
create the rescue floppy to use a floppy drive, or you need to set your computer to boot from
CD-ROM. No matter what media you use to boot, the system then runs the dbootstrap
program to install and configure a base Debian system.
Part
You also must decide how to boot Linux. You have two choices. You can boot Linux from a
I
floppy disk, in which case you need an extra formatted disk. The other choice is to use a
program called LILO, the Linux Loader. LILO is a program that allows you to specify which Ch
operating system to boot. OS/2 and Windows NT provide similar functionality. 5
Next, you should make sure you have enough disk space to install Linux. A minimal system
requires at least 4MB of RAM and 35MB of free disk space, but this type of system is
minimalÐusable at best as a terminal. For a system running the X Window System and some
usable software, you can figure on 300MB of disk space. For a robust system, you need
around 800MB, and for a full installation, you need around 2000MB (2GB).
For configuring XFree86, you need to write what type of chipset your video card uses. If you
have a serial mouse and modem, you should write down the serial port that each is using.
You'll need this information later during the configuration process.
Preparing to Install Debian
If you have a brand-new system or a system on which you don't care what happens to the data
already stored on the computer, you can skip most of the following sections and go directly to
the section ``Creating the Rescue and Driver Floppies.'' If, however, you're already using a
122 Chapter 5 Installing Debian Linux
system and simply want to add Linux, you must do some planning. The main reason is that Linux
is another operating system, not just a collection of programs.
If you intend to install Debian on a system already in use, you should back up your essential
files to a zip disk or CD-R system. Installing Debian can be a destructive process on your
system. After making the necessary backups, you need to inventory your hardware as
described previously. Debian makes no additional hardware restrictions other than those
imposed by the Linux kernel. If you have a standard PC, you should have no problems.
If you have any Windows-specific hardware, such as a WinModem or printer, you will
have a problem when installing Debian. These devices rely on Windows to do much
of their processing, something Linux cannot do.
Tip #27 from
Jack For information on installing Debian or other Linux distributions on a laptop, see the
following Web site:
http://www.cs.utexas.edu/users/kharker/linux-laptop/
Debian supports four architectures: Intel x86-based architectures; Motorola 680x0 machines
such as Atari, Amiga, and Macintoshes; DEC Alpha machines; and Sun SPARC machines.
They are referred to as i386, m68k, Alpha, and Sparc, respectively.
Debian does not run on 286 or below Intel processors but should run on nearly all other Intel
x86 processors and nearly all clones.
Tip #28 from
and
Jack If you have problems installing Debian from your floppy drive you have a Cyrix
CPU, you need to disable the system's cache memory from the BIOS setup menu.
Remember to re-enable the cache after installation. The symptoms seem to be
problems with the floppy drive after a switch from 16-bit to 32-bit mode and more
investigations continue to see if Debian can supply a workaround to this issue.
Debian supports most industry standard motherboards, including some multiprocessor
boards, and all system buses.
Preparing to Install Debian 123
Tip #29 from
Jack To use multiple processors, called Symmetric Multiprocessing, or SMP, you need a
specialized kernel to take advantage of those processors. Otherwise, your system uses
only the first processor on the motherboard.
You need a VGA-compatible display interface for the console terminal. Older standards such
CGA, MDA, or HGA should also work, assuming you do not require X11 support.
Unlike some other distributions, Debian does not utilize a GUI environment during
the installation process. This way, you are ensured a successful installation across a
wide hardware base.
Debian's X Window System support, like the other distributions, is determined by the
underlying support found in the XFree86's X Window System, which includes support for
most of the newer AGP video cards.
? See ``Installing the XFree86 System,'' p. 508
Debian supports most storage devices such as floppies, IDE/EIDE/ATAPI, and SCSI
devices. Debian does not support older style drives such as MFM or RLL hard disks,
however. Support for other storage devices, such as parallel port zip drives, can be added later
with a modified kernel, as described in Chapter 14, ``Configuring the Linux Kernel.'' Debian
Part
does not support the following drive controllers:
I
& EATA-DMA protocol-compliant SCSI host adapters such as the SmartCache III/IV,
SmartRAID controller families, and the DPT PM2011B and PM2012B controllers
Ch
& The 53c7 NCR family of SCSI controllers (but 53c8 and 5380 controllers are 5
supported)
As shown in the ``Linux Hardware How-To,'' Linux supports a myriad of hardware
peripherals. Some items Debian does not support include sound cards, by default, but you can
add sound support later. Debian also does not support the following network interface cards
(NICs):
& AX.25 cards and protocols
& 3Com EtherLink Plus (3c505) and EtherLink16 (3c507)
& NI5210 cards
& Generic NE2100 cards
& NI6510 and NI16510 EtherBlaster cards
& SEEQ 8005 cards
& Schneider & Koch G16 cards
124 Chapter 5 Installing Debian Linux
& Ansel Communications EISA 3200
& Zenith Z-Note built-in network card
& Microchannel (MCA) network cards
Tip #30 from
Jack The following Web site contains unofficial images to support Microchannel NICs:
ftp://ns.gold-link.com/pub/LinuxMCA/
Now that you know your hardware is supported by Debian, you need to decide how you
intend to use the systemÐas a standalone workstation, a server, or a development machine.
Table 5.1, based on the Debian documentation, provides an overview of each type of system
and its requirements. During installation, you select which profile to install.
Table 5.1 System Profiles
Profile Description
Server_std A small server profile, useful for a stripped-down server that does not have a
lot of niceties for shell users. It basically has an FTP server, a Web server,
DNS, NIS, and POP. It takes up around 80MB. Of course, this is just the
size of the software; any data you serve up would be additional.
Dialup A standard desktop box, including the X Window System, graphics
applications, sound, editors, and so on. The size of the package is around
500MB.
Work_std A more stripped-down user machine, without the X Window System or X
applications, possibly suitable for a laptop or mobile computer. The size is
around 140MB. (Note that I have a pretty simple laptop setup including X11
in even less spaceÐaround 100MB.)
Devel_comp A desktop setup with all the development packages, such as Perl, C, C++, and
so on. The size is around 475MB. Assuming you are adding X11 and some
additional packages for other uses, you should plan around 800MB for this
type of machine.
Partitioning Your Hard Drive
Next, you need to partition your hard drive for installation. This process is the most
dangerous because maximum data loss is assured. If you have not backed up your system, do
so now. Although you can use an experimental program called FIPS or commercial programs
such as Partition Magic that do nondestructive repartitioning, a full backup is recommended,
just in case problems occur.
Partitioning Your Hard Drive 125
Why Use Partitions?
In the early days of PCs, hard drives were few and far between. Most computers used floppies
to hold the operating system and programs and their data. IBM, with the introduction of the
IBM PC XT, introduced a 10MB hard drive. Early operating systems such as DOS could
access only a limited amount of space on hard drives. Then hard drive manufacturers kept
expanding the space on their hard drives quicker than the operating system's capability to
access the additional space. The operating system got around this problem by letting the user
split the hard drive into sections, called partitions. These partitions can hold program files,
other operating systems, or data.
Typical MS-DOS systems have one partition, which is referred to as drive C. If you split the
drive into partitions, these partitions are typically referred to in alphabetical order as D, E,
and so on. MS-DOS also allows you to install multiple hard drives, so the next drive in this
chain might be referred to as F. UNIX and Linux do not use drive letters to refer to
partitions; instead, they use directory names to refer to partitions. Also, as indicated earlier,
Linux users can place different directories on different partitions and even on different drives.
You can also place different operating systems on different partitions.
Debian needs at least one partition for itself. You can have a single partition containing the
entire operating system, applications, and your personal files, but most people feel that the
swap partition is also a necessity, although it's not strictly true. ``Swap'' is scratch space for an
operating system, which allows the system to use cheap disk storage as ``virtual memory.'' By
putting the swap area on its own partition, Linux can make much more efficient use of it (it is
possible to force Linux to use a regular file as swap, but doing so is not recommended).
In general, I recommend a minimum of three partitions: one for the operating system itself, Part
one for all the home directories, and one for swap. As mentioned previously, for efficiency
I
reasons you need swap on its own partition. For safety reasons, you should place your home
directory on a separate partition, or better, on a separate drive so that problems on one
Ch
partition do not destroy data on another partition. For ease of upgrades, you need to split 5
system files across partitions. Thus, you can upgrade your operating system while protecting
user files in their home directories.
Naming Partitions
Linux refers to disks and partitions in a different way than Windows and MS-DOS. Table 5.2
illustrates the naming conventions used by Linux.
? See ``Understanding File Systems,'' p. 440
Table 5.2 Disk and Partition Naming Schemes for Linux
Device Name
First floppy drive /dev/fd0
Second floppy drive /dev/fd1
126 Chapter 5 Installing Debian Linux
Table 5.2 Disk and Partition Naming Schemes for Linux
Device Name
First SCSI disk (SCSI ID address-wise) /dev/sda
First partition on SCSI disk /dev/sda1
Second partition on first SCSI disk /dev/sda2
Second SCSI disk /dev/sdb
First partition on SCSI disk /dev/sdb1
Second partition on second SCSI disk /dev/sda2
First SCSI CD-ROM /dev/scd0 (sometimes /dev/CD-ROM)
Master IDE/primary controller /dev/hda
Slave disk IDE/primary /dev/hdb
Basically, you distinguish partitions on a drive by appending the next numeric value to the
device name.
Explaining Partitions
Partitions are specified in a section of the hard drive referred to as the boot record in what is
called a partition table. This table is used by the various operating systems to determine what
operating system to boot and where their files can be found physically on the hard drive. The
boot record is used to boot, or start up, the machine's operating system. LILO and other boot
managers use this section of the hard drive, typically found on the first sectors of the drive, to
control which operating system to start.
The partition table holds information about the locations and sizes of the various partitions
on the hard drive. The three kinds of partitions are primary, extended, and logical. DOS and
some other operating systems must boot from primary partitions. Hard drives can contain
only four primary partitions. An extended partition does not contain data itself; instead, it
allows the user to define other, logical partitions on the drive. Thus, to get around the four-
limit primary partition number, you can define an extended partition and then define other
logical partitions within the extended partition. Some operating systems such as MS-DOS
and versions of OS/2 before version 2.0 require that they be installed in a primary partition,
but they can access logical drives in extended partitions. This information is important to
remember if you are going to have both a DOS system and a Linux system reside on the same
drive. DOS must go in a primary partition.
Using FDISK
Partitions are created, destroyed, and managed by a program usually called FDISK. Each
operating system has its own version of FDISK, so you must be sure to use the correct one. If
you are currently using DOS or are planning to use DOS, you must first repartition the DOS
drive using DOS's FDISK. You later use the Linux version of cfdisk to create the Linux
Partitioning Your Hard Drive 127
partitions. If you are using OS/2, you also need to use the OS/2 version of FDISK to prepare the
OS/2 partitions.
Partition Requirements
When you're ready to start partitioning your hard drive, you first should plan what partitions
you need. DOS requires a primary partition. Linux and OS/2 can reside in other partitions. If
you are using the OS/2 boot manager, which also works well with Linux, you must prepare
for its use also. You must also be aware if you are shrinking a current DOS partition to make
room for Linux that not all of your files can be restored to the new, smaller DOS partition.
Note that you can access DOS partitions from Linux, moving, saving, and editing DOS files
under Linux. However, you cannot execute DOS programs under Linux.
Two experimental components of Linux allow you to emulate DOS under Linux and
also install Linux under DOS. Both systems are still in the implementation stage and
are, thus, more suited for Linux hackers. These topics are covered in Chapter 2, ``Linux
Installation Overview.'' You can also find plenty of information on these topics in the
Linux world.
Next, you should jot down the number of partitions you need and how much disk space to
provide for each.
DOS Requirements
If you want to boot DOS, it must go in a primary partition. A bootable version of DOS does
Part
not require much spaceÐjust enough for the system files, COMMAND.COM, CONFIG.SYS, and any I
driver files needed to start your system. For instance, I provide a 5MB DOS partition on my
Ch
first drive to boot DOS. When DOS is loaded and running, you can access any of the other
extended and logical drives on the system. Unfortunately, although Linux can access DOS
5
files in a DOS partition, DOS cannot access Linux files in a Linux partition.
OS/2 Requirements
OS/2 versions 2.0 and later do not need a primary partition. The OS/2 system can install and
boot from an extended partition. Thus, you can install DOS on a primary partition and create
an extended partition area for OS/2 and Linux. The space required for OS/2 is version- and
feature-dependent, so you should consult your OS/2 documentation for space requirements.
You should also subtract 1MB from available space if you intend to use the OS/2 boot
manager.
Linux Requirements
As explained earlier, Linux stores files on file systems, and these file systems can reside on
different partitions, basically as safety precautions. Linux requires one partition for each file
system. The next consideration is for a swap partition. Linux, like most operating systems
128 Chapter 5 Installing Debian Linux
that use disk space for memory (called a virtual memory configuration), needs a swap file or a swap
partition to simulate physical memory using disk space. Linux typically uses a swap partition of at
least 64MB.
Also, Linux limits the partitions per drive to 15 partitions for SCSI disks (3 usable primary
partitions, 12 logical partitions) and 63 partitions on an IDE drive (3 usable primary
partitions, 60 logical partitions).
Repartitioning
This section assumes you need to repartition a DOS drive. To do so, execute FDISK by
typing fdisk at the DOS prompt. You then see the FDISK Options screen (see Figure 5.1).
Figure 5.1
FDISK provides
many options
used to create,
delete, and
modify parti-
tions.
The screen shown in Figure 5.1 might appear different depending on which version of
MS-DOS you are using. Pick menu option 4, Display Partition Information. The Display
Partition Information screen appears (see Figure 5.2).
Figure 5.2
The Display
Partition Infor-
mation screen
in MS-DOS 6.0.
You should write down this information. You need the current partition table information if
you decide to abort the Linux installation and put your system back the way it was before you
started.
You might not need to repartition your hard drive, although it is thought that
repartitioning offers the best introduction to Linux. You can use FIPS to
nondestructively repartition your hard drive.
Partitioning Your Hard Drive 129
FIPS stands for the First nondestructive Interactive Partition Splitting program. It is a
program developed by Arno Schaefer as a result of the Linux project and is used to
move around DOS partitions to make room for Linux partitions.
You can find the complete instructions for using FIPS in the document fips.doc
located on the enclosed CD-ROM in the /tools/fips20 directory. This program can
help only if you have enough free space left on your drive to install Linux; otherwise,
you either need to delete unneeded files or use the process described earlier to
repartition your hard drive.
Deleting Partitions
Unfortunately, FDISK does not allow you to simply resize a partition; you must first delete
the partition and then add it back with the desired size. From the FDISK Options screen,
choose menu option 3, Delete Partition or Logical DOS Drive, which deletes the necessary
partitions. The Delete DOS Partition or Logical DOS Drive screen then appears (see
Figure 5.3).
Figure 5.3
The Delete
DOS Partition
or Logical DOS
Drive screen.
Part
I
Ch
Here, you should pick the appropriate menu option for the type of partition you are deleting,
5
such as a primary DOS partition. For example, option 1, Delete Primary DOS Partition,
allows you to delete primary DOS partitions. Choosing option 1 displays the Delete Primary
DOS Partition screen (see Figure 5.4).
Figure 5.4
The Delete Pri-
mary DOS Par-
tition screen.
130 Chapter 5 Installing Debian Linux
This screen asks for a volume name of the partition and then a confirmation to see whether
you really want to delete the partition. Because all information on the partition will be
destroyed, FDISK wants to be absolutely sure that you want to delete the primary DOS
partition.
Adding Partitions
After you have deleted all the necessary partitions, you must then add the appropriate
partitions for your DOS system. You cannot add the Linux or OS/2 partitions with the DOS
FDISK program. Partitioning the hard drive for Debian Linux is covered earlier in this
chapter. Figure 5.5 shows the Create DOS Partition or Logical DOS Drive screen.
Figure 5.5
The Create
DOS Partition
or Logical DOS
Drive screen.
Providing all the space available for the partition and making the partition the active partition
are the FDISK defaults (see Figure 5.6).
Figure 5.6
The Create
Primary DOS
Partition
screen.
Active indicates that the partition is bootable. To boot DOS, you must specify the primary
partition as active. Say No to this first selection so that you can specify the exact amount of
disk space to provide to your DOS partition. Answering No to the question in Figure 5.6
displays the Specify Disk Space for the Partition screen. There, you can specify the desired
space for your DOS partition either in megabytes or in percentage of space available and
press Enter.
Next, you must set this partition active. From the FDISK Options screen, choose menu
option 2, Set Active Partition, and simply follow the instructions on the Set Active menu
screen.
Installing Debian 131
Installing Debian
You can boot the installation system from floppies, bootable CD-ROM, or a non-Linux boot
loader. You can install the entire system from floppies, but this approach is not
recommended; it takes more than seven floppies and a lot of time to create and use them. The
typical installation procedure is to create a boot floppy and then install from the CD-ROM.
After you have booted the system, you can install from CD-ROM, a Network File System
(NFS) server, FTP, HTTP, or from another hard drive. The easiest way to boot and install, if
it works, is to boot directly from the CD-ROM.
Installing from CD-ROM
If you have a bootable CD-ROM drive, you need to make sure it is selected in your BIOS
settings as a boot device. Then you can place the CD-ROM that accompanies this book into
the drive and reboot your computer.
If the system boots properly, you can continue installing system files, as described later in this
chapter; otherwise, you need to create a bootable system for installation. If your hardware
does not support bootable CD-ROMs, you should boot into DOS and execute the boot.bat
file located in the \boot directory on your CD-ROM. Then you can skip down to ``Installing
the System Files.'' However you decide to boot, you can install the base Debian system from
the CD-ROM. To do so, simply boot using one of the other installation techniques; when it
is time to install the base system, and when you install the complete system, just point your
installation system at the CD-ROM drive.
Installing from DOS Part
Even though you can run DOS programs from Windows 98 and NT, you must be in DOS I
mode, not Windows, to perform this installation. To install from DOS, you need to copy the
Ch
following files into a directory on the DOS drive:
5
& resc1440.bin
& drv1440.bin
& base2_1.tgz
& root.bin
& linux
& install.bat
& loadlin.exe
These files are located in the install folder on the CD-ROM.
Next, you must boot into pure DOS mode (that is, you must not have any drivers loaded). To
enter DOS mode from Windows, go to the start, shutdown menu option and select Restart
the Computer in MS-DOS Mode, as shown in Figure 5.7.
132 Chapter 5 Installing Debian Linux
Figure 5.7
Installing
Debian from
DOS requires
you to be in
MS-DOS mode,
not in a DOS
compatibility
window while
running
Windows.
At the DOS prompt, execute the install.bat program you copied. You can now continue
installing the system as detailed in the section ``Installing the System Files."
Creating the Rescue and Driver Floppies
The Debian distribution of Linux included on the CD-ROM accompanying this book
contains the disk images of two floppies needed with Debian: the rescue and driver floppies.
Disk images are files containing the complete contents of a floppy disk in raw form. Disk
images, such as resc1440.bin, cannot simply be copied to floppy drives. A special program is
used to write the image files to floppy disk in raw mode. This program is required because
these images are raw representations of the disk; it is required to do a sector copy of the data
from the file onto the floppy. You create these floppies with an MS-DOS program called
rawrite2, which is provided with most Linux distributions. You use rawrite2 to transfer the
images to the appropriate floppies.
You use the rescue disk to start the Linux system for installation. It contains bare-bones
device drivers and a basic version of the operating system. This rescue disk is specific to the
hardware and type of floppy drive your system uses to boot (usually the A drive under
MS-DOS). To create this disk, you issue the following command from the DOS prompt:
rawrite2 -f file -d drive
file is one of the floppy disk images described next. drive indicates which floppy drive to
use, either a: or b:. You need to create the rescue and root disks at the minimum to install
Debian.
The Debian documentation describes the following image files:
The files resc1440.bin, resc1440tecra.bin, and resc1200.bin are the Rescue Floppy disk
images. The Rescue Floppy is used for initial setup and for emergencies, such as when your
system doesn't boot for some reason. Therefore, it is recommended you write the disk image
to the floppy even if you are not using floppies for installation. The tecra images are alternate
kernels for people who have problems with the standard disks.
Creating the Rescue and Driver Floppies 133
The files drv1440.bin, drv1440tecra.bin, and drv1200.bin are the Drivers Floppy disk
images. They contain the kernel modules, or drivers, for all kinds of hardware not necessary
for initial booting. You are prompted to choose the drivers you need during the installation
process. If you used a special Rescue Floppy image, you need to use the corresponding
Drivers Floppy image.
The following set of files make up the base system:
base2_1.tgz
base14-1.bin, base14-2.bin, base14-3.bin, base14-4.bin, base14-5.bin,
base14-6.bin, base14-7.bin
base12-1.bin, base12-2.bin, base12-3.bin, base12-4.bin, base12-5.bin,
base12-6.bin, base12-7.bin
These files contain the base system that will be installed on your Linux partition during the
installation process. This set of files is the bare minimum necessary for you to be able to
install the rest of the packages. The base2_1.tgz file is for installation from nonfloppy
mediaÐthat is, CD-ROM, hard disk, or NFS.
The file root.bin is the Root image. This file contains an image of a temporary file system
that gets loaded into memory when you boot. It is used for installations from hard disk and
from CD-ROM.
The program rawrite2.exe is a DOS utility to write a floppy disk image to a floppy. You
should not copy images to the floppy; instead, you should use this utility to copy them sector
by sector.
Part
The program loadlin.exe is the Linux boot loader. You need this boot loader if you are
installing from a DOS partition or from a CD-ROM. I
The file install.bat is a DOS batch file for starting Debian installation from DOS. This
Ch
batch file is used in installations from hard disk or CD-ROM. 5
The file linux is very important; it is the kernel program, the core of the operating system.
This file is the Linux kernel image to be used for hard disk and CD-ROM installations.
The files install.txt and install.html are in the Installation Manual in plain ASCII and
HTML format, respectively.
The files fdisk.txt and cfdisk.txt are the instructions for using your available partitioning
programs.
The file basecont.txt contains a list of the contents of the base system.
The file md5sum.txt lists the MD5 checksums for the binary files. If you have the md5sum
program, you can ensure that your files are not corrupt by running the following:
md5sum -v -c md5sum.txt
134 Chapter 5 Installing Debian Linux
You need to create only the rescue disk to install from the CD-ROM. To create it, use the
following command:
rawrite2 -f resc1440.bin -d a:
Now you just place the resc1440 disk into the drive and reboot your computer.
Installing the System Files
After you reboot, the system displays startup messages, the welcome to Debian GNU/Linux
2.1, and then the following prompt:
boot:
You can press the F4 or F5 keys for more information or press Enter to continue the
installation.
Tip #31 from
Jack Keep the rescue floppy because, as its name implies, you can use this floppy in the
future should you have problems booting the system. At the boot: prompt, you can
press F3 to get more information on the rescue disk.
Pressing Return starts the dbootstrap program, which is responsible for the initial system
installation and configuration. The primary goal is to install and configure the core of the
system so that you can continue with the full installation. dbootstrap is a text-based menu-
driven program that leads you through the installation and configuration process. You use the
up- and down-arrow keys to move between menu selections and then Return to accept a
selection. If you make a mistake, you can return to a previous step via the menu selection to
correct the error.
Tip #32 from
Jack Debian displays any error messages on the third virtual terminal. You reach this
display by pressing Alt+F3 and return to the installation screen by pressing Alt+F1.
When the system has finished booting, you should see the Select Color or Monochrome
Display dialog box. If your monitor can display only black and white, press Return to
continue with the installation. Otherwise, you can use the arrow keys to move the cursor to
the Color menu item and then press Return. The display should change from black and white
to color. Then you can press Return again to continue with the installation.
Installing the System Files 135
Next, you configure the keyboard by selecting the appropriate entry from the Configure the
Keyboard dialog box. After installation, you can select from a wider range of keyboards by
running the kbdconfig program.
After selecting a keyboard, you need to partition your hard drives. If you prepped the drives
previously, you can continue; otherwise, now you need to partition and format the drives
using fdisk.
Tip #33 from
Jack You can also use the program cfdisk to partition your drives. You can find more
information on cfdisk in the tools folder on the Debian CD-ROM.
Using the Linux fdisk Program
When you're ready to partition your drives, at the fdisk prompt, type m for a list of
commands. Table 5.3 lists the available commands.
You use the fdisk program native to Linux for these actions. You should be careful
because this program is different from the fdisk programs included with other
operating systems such as MS-DOS, Windows 95/98, and OS/2. You cannot use these
programs interchangeably! For example, you cannot use Linux's fdisk to rearrange a
partition for a DOS partition. Although you can use any fdisk to create partitions, you
must use the appropriate operating system's version of fdisk to perform such actions Part
as setting file types.
I
Ch
5
Table 5.3 The Linux fdisk Commands
Command Description
a Toggles a bootable flag
c Toggles the DOS compatibility flag
d Deletes a partition
l Lists known partition types
m Displays this menu
n Adds a new partition
p Displays the partition table
q Quits without saving changes
t Changes a partition's system ID
u Changes display/entry units
136 Chapter 5 Installing Debian Linux
Table 5.3 The Linux fdisk Commands
Command Description
v Verifies the partition table
w Writes the table to disk and exits
x Provides extra functionality for experts only
To begin the partitioning, select the p command (press p and then Return) to display the
current partition table, which should reflect the drive you partitioned earlier with the DOS
FDISK program. Listing 5.1 shows a possible listing from the p command.
Listing 5.1 Example of a Current Partition Table
Disk /dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
dev/hda1 1024 1024 4040 384667+ 51 Novell?
Partition 2 has different physical/logical endings:
phys=(967, 14, 17) Logical=(4096, 14.17)
Your screen may appear different than what's shown in Listing 5.1 because the values
are different for each drive type and the partitions already defined on that drive.
Listing 5.1 indicates the various partitions already defined that it can detect, the starting and
ending locations of the partition, and how big each partition is in blocks. The listing also
indicates the partition type. Table 5.4 shows all the different types of partitions you can
define by using the Linux fdisk program. The primary partition types used here are 83-Linux
Native and 82-Linux Swap. You can get a similar listing by using the l command.
Table 5.4 The Known Linux Partition Types
Reference Number Type
0 Empty
1 DOS 12-bit FAT
2 XENIX root
3 XENIX usr
4 DOS 16-bit < 32MB
5 Extended
6 DOS 16-bit >= 32MB
Installing the System Files 137
Table 5.4 The Known Linux Partition Types
Reference Number Type
7 OS/2 HPFS
8 AIX
9 AIX bootable
a OS/2 Boot Manager
40 Venix 80286
51 Novell?
52 Microport
63 GNU HURD
64 Novell
75 PC/IX
80 Old MINIX
81 MINIX/Linux
82 Linux Swap
83 Linux Native
93 Amoeba
94 Amoeba BBT
a5 BSD/386
b7 BSDI fs
Part
b8 BSDI swap
c7 Syrinx
I
db CP/M Ch
e1 DOS access 5
e3 DOS R/O
f2 DOS secondary
ff BBT
In Listing 5.1, Linux prints a note about the different physical and logical endings at the
bottom of the screen. The difference exists because, on the system used to write this chapter,
a prior partition containing the DOS D drive was left intact, whereas the C drive was
repartitioned to a smaller C drive to make room for Linux. Thus, space exists between the C
drive and the D drive. The necessary partitions required by Linux will be created here.
The begin, start, and end numbers from Listing 5.1 are very important. You should write
them down because you'll need them in a later step to specify the necessary sizes of the
partitions you'll add.
138 Chapter 5 Installing Debian Linux
Adding the Necessary Partitions
Because you've repartitioned the drive for DOS, you shouldn't have to delete any partitions
for Linux. You should only have to add partitions. To add a partition, you issue the n
command, which displays the following:
Command Action
e extended
p primary(1-4)
Now, you can press p and then Return. When fdisk asks for the partition number, enter your
selection and press Return. If you indicate a partition number already in use, fdisk reports this
fact and asks you to delete the partition before trying to add it to the partition table. For this
example, enter 3 to add a third primary partition that's referred to as /dev/hda3.
Next, fdisk asks for the location of the first cylinder. This is usually the first available cylinder.
In fact, fdisk displays a default range for your selection, such as the following:
First cylinder (42-1024) :
Notice that the first partition ends at cylinder 41 and that the next partition begins at cylinder
1024. Thus, the range supplied by fdisk here allows you to start the next partition anywhere
in the range of 42 to 1024. It's a very good idea not to place partitions just anywhere
throughout the disk; instead, choose the next available location, which in this case is cylinder
42. Enter 42 and press Return.
Linux can have trouble booting from partitions defined to start at cylinders above
1024. If you can create a Linux partition only in this range, you might have to boot
Linux from a floppy. You'll learn how to create the Rescue floppy. The only downside
is that booting Linux from a floppy takes a little longer than booting from the hard
drive.
Now fdisk wants you to specify how much space to allocate for this partition. You can express
this size in number of cylinders or by the number of bytes ( +size), kilobytes (+sizeK), or
megabytes ( +sizeM). Because you should already know the approximate size you need for the
swap file, you can define this partition first and then leave the rest of the disk space for the
Linux program partitions. For this example, because your machine has 8MB of RAM, you
need to specify a 16MB partition size by replying as follows:
Last cylinder or +size or +sizeM or +sizeK (42-1023): +16M
You should then use the p command to look at the new partition table you've defined. In this
example, the new partition table looks like this:
Disk /dev/hda: 15 heads, 17 sectors, 1024 cylinders
Units = cylinders of 255 * 512 bytes
Device Boot Begin Start End Blocks Id System
/dev/hda1 * 1 1 41 5219 1 DOS 12-bit FAT
/dev/hda2 1024 1024 4040 384667+ 51 Novell?
Installing the System Files 139
Partition 2 has different physical/logical endings:
phys=(967, 14, 17) Logical=(4039, 14.17)
/dev/hda3 42 42 170 16447+ 83 Linux native
By default, fdisk made the new partition a Linux native type. To change this type to a swap
partition, you need to use the t command. To do so, enter t and then enter the partition
number you want to change; in this example, enter 3. fdisk then requests that you enter the
hexadecimal value of the desired partition type from Table 5.4. (If you don't have the table
handy, you can type l to get the list of partition type codes.) Because you want a swap
partition in this case, enter 82 at the prompt.
As you can see, fdisk reports the new partition type, but you can also use the p command to
double-check that partition 3 is now a Linux swap partition.
Now you can add your Linux partitions. For this example, you will add only one partition.
But if you want to have multiple partitions for various reasons, you can add them at this time.
To add a partition, enter n, specify p for another primary partition, and then specify the
number for this partition, which is 4. To keep from fragmenting different partitions across
the drive, start the last partition where the first left off, at cylinder 171. Because you want to
use the rest of the space for the Linux system, you can specify the last cylinder instead of an
exact byte count. Thus, enter 1023, as shown here:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 4
First cylinder (171-1024): 171 Part
Last cylinder or +size or +sizeM or +sizeK (171-1023): 1023
I
Now use the p command to verify the new partitions. If you need to make any changes, you
can do so now. Ch
When you're satisfied with the layout of your partitions, you can use the w command to write
5
the partition table information to the hard disk. None of your changes are permanent until
you use the w command; thus, if you feel you've made some changes in error, you can use the
q command to exit without altering the partition table. When you issue the w command,
Linux tells you the partition table has been altered and then resynchronizes the disks to match
the new partition table. If your Linux system hangs at this point, you can reboot with the
installation boot and root disks until you're back at the # prompt.
Don't use the Linux fdisk program to create or modify partitions for other operating
systems. Doing so could leave the hard drive in a useless state for both operating
systems.
140 Chapter 5 Installing Debian Linux
Configuring a Swap Partition
After configuring your drives, the installation asks whether you want to initialize and activate
a swap partition. This disk area is used as slow memory when the system runs out of RAM. A
swap partition is strongly recommended if you have less than 32MB of RAM. Select the next
menu item to continue the installation. If you do not want a swap partition, you can select the
menu option Do Without a Swap Partition; otherwise, you should select the partition to
activate as a swap partition and press Return.
After selecting the partition, you can scan the entire swap area for bad disk blocks and remove
them from the pool of available blocks, thus averting a future problem. Select Yes from the
dbootstrap menu to initialize the swap area.
Tip #34 from
Jack Most of today's disk drives do not suffer from bad blocks, so you can generally save
the time and skip initializing the disk space. However, if you have older drives or well-
used drives, initialization is a small price to pay for peace of mind.
Initializing Linux Partitions
Next, dbootstrap displays the Initialize a Linux Partition dialog box. You can now initialize
various Linux partitions for your system. Although you can have just one partition for
everything, I recommend at least two: one for the operating system and one for users' home
directories. Select Next to initialize and mount the / disk partition, called the root.
? See ``Linux Standard Directories,'' p. 418
You can use the arrow keys to continue to initialize and mount the various partitions you
created, such as /usr, /var, and /opt. After initializing all the partitions, you can install the
kernel.
Configuring the System
Install Operating System Kernel and Modules is the next menu selection. dbootstrap
presents you with a selection of drives from which you can install Debian, typically your
CD-ROM. You can also select to install from another hard drive or via the network.
Although you can specify several devices or locations from which to install Debian, for the
typical installation from CD-ROM, you can simply press Return for the default selections.
After installing the kernel, you need to configure the various devices on your system, such as
Ethernet cards. Using the Configure Device Driver Modules menu selection, you can install
on your system the drivers needed for various devices to be made available on startup.
Installing the Base System 141
Tip #35 from
Jack You can reconfigure any module or device driver after installation by using the
modconf program.
Configuring the Network
After installing the various device drivers needed by your system, you need to configure the
network, even if your system is not connected to a network. If you are not connected to a
network, you just have to select a host name for your system and then answer No to the Is
Your System Connected to a Network selection.
You need the information listed in the previous section What You Need to Install Debian to
configure your system for connection to a network. However, if you will be connecting to the
network via PPP, you should not configure the network at this time. To configure, you can
simply answer the various questions asked by dbootstrap from the information you have. The
primary network connection should be eth0 if you are using Ethernet.
After configuring the base network, you can then proceed to install the base system.
Installing the Base System
When you're ready to install the base system, you are prompted to specify the path to the
base2_1.tgz file. You can simply press Return and let dbootstrap find the file, or you can
type the path at the prompt. At this point, you have a minimal Debian system, but now you
Part
must perform some configuration before the system will run.
I
dbootstrap next asks you to select your time zone. You can specify your time zone in many
ways, but the suggested method is to go to the Directories: pane of the dialog box and select
Ch
your country (or continent). That selection changes the listed time zones so that you can 5
select your geographic location in the Timezones: dialog box.
Next, dbootstrap asks if your system clock is to be set to Greenwich Mean Time (GMT) or
local time. Select GMT if you intend to run only Debian on your computer; otherwise, select
local time.
Next, you need to decide whether to boot Linux from the master boot record (MBR) or from
some other boot manager or floppy. dbootstrap asks you to make a selection. If you select to
boot from the MBR, you cannot boot directly into any other operation system. If you do not
select to boot from the MBR, you can later use either the fdisk command or the activate
command to set the bootable partition.
Tip #36 from
Jack If you want to uninstall Debian booting from the MBR so that you can boot into DOS/
Windows, you can use the command fdisk /mbr to restore the DOS boot block.
142 Chapter 5 Installing Debian Linux
dbootstrap next asks you to create a boot floppy, which is a very good idea. If your
installation fails, or at some future time you have problems booting into Debian, the boot
floppy can usually boot Debian so that you can troubleshoot the problems.
Finally, you need to select the Reboot the System menu item to reboot your system and
continue with the rest of the installation.
Configuring the Base System
After rebooting, you are asked to set the password for the superuser account, called the root
account. This username (root) is the most powerful account on any UNIX system, allowing
the person logged in as root full access to everything while being bound to none of the
security precautions other users must follow. The root account is meant to be used only for
system administration, so you must not use the account for day-to-day activities.
Next, Debian asks you to create a normal user account to use for your day-to-day tasks. You
can use almost any name, but your password should not be a word easily guessed by others.
The same goes for the root password. Try to select a password of at least six characters, and
use a number to help make the password harder to break.
? See ``Dealing with Password Security,'' p. 277
After you create the normal user account, Debian then requests you to enable shadow
password support. Shadow passwords provide even more security to your system, not only for
the root account but for all users. Without shadowed passwords, your password file
/etc/passwd)
( can be read by anyone on your system, and even though passwords are
encrypted, crackers can easily break passwords using a variety of methods.
Although unencrypting a password from the encrypted one is practically impossible,
some programs can encrypt millions of words or character patterns and then
compare them to password files. A UNIX system checks a password this way: You
enter the password, the system encrypts the password, and then it compares the
password just encrypted to the encrypted password stored in the password file. If they
match, the user is allowed access to the system.
Enabling shadow password support allows the system to store the encrypted passwords in a
file that can be accessed only by the superuser. Thus, crackers cannot gain access to even the
encrypted passwords normally stored in /etc/passwd. I highly recommend that you enable
shadow password support.
Configuring the Base System 143
Tip #37 from
Jack If you decide not to use shadow passwords in the beginning but decide later that you
want to use them, you can use the shadowconfig command to reconfigure your
system.
After you create the root and a user account, the system then asks you whether you want to
use one of the preconfigured profiles, as described in Table 5.1, to install. Selecting the
various packages to install can be time-consuming, considering over 2,000 packages are
available. The canned selections provide for an easy choice. However, you can use the dselect
program to individually select from the various packages available. You can run the dselect
command at any point to install packages. After you have made your selections, either a
profile or packages, the system then starts installing your selections.
Congratulations! After installing all the selected packages, your Debian system is ready to go.
Part
I
Ch
5
CHAPTER
6
Adding Sound Cards and Other
Multimedia Hardware
In this chapter by Jeff Tranter
Sound Cards 146
Audio CDs 155
Joysticks 157
Other Multimedia Devices 158
Multimedia Applications 158
Information Resources 159
Troubleshooting 160
146 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Sound Cards
Most PCs today are sold with sound cards. Compared with just a few years ago, current sound
cards provide very high quality digital audio at an extremely affordable price. In contrast to
the traditional use of UNIX systems for running multiuser text-based applications, Linux
systems today make use of multimedia, including sound. Applications for multimedia are
limitless, ranging from a single user running a graphical desktop environment that uses sound
for feedback all the way to using Linux as the basis for a professional digital audio recording
studio.
A Little History
Actually, several different sound card drivers are available under Linux. A little history is in
order to help you understand where they came from.
In the early days of Linux (prior to version 1.0), Hannu Savolainen of Finland developed a
kernel driver for the Creative Labs SoundBlaster card. He and others extended the driver over
time to support many other types of popular sound cards. While working on a part-time
basis, Hannu could see that with more effort the sound drivers could be ported quite easily to
other UNIX-compatible operating systems and extended to support more sound cards. To
this end, the U.S. company 4Front Technologies hired Hannu to work full time on turning
the sound card drivers into a commercial product. This product has come to be known as the
Open Sound System (OSS). OSS now runs on a many different operating systems and is sold
as a commercial product. A freely available sound driver derived from OSS, now usually
called OSS/Lite, continues to be included with the Linux kernel.
Some time after Hannu developed the sound drivers, a group of people headed by Jaroslev
Kysela were using Gravis UltraSound sound cards, one of the first affordable sound cards that
supported wavetable synthesis technology. They were unhappy with the kernel sound card
driver and started the Gravis UltraSound Project to develop their own. Eventually, the
project expanded in scope to include support for more sound cards and became the Advanced
Linux Sound Architecture (ALSA).
To make things somewhat more confusing, the OSS/Lite sound drivers in the 2.0 kernel were
modified by Alan Cox under the sponsorship of Red Hat Software so that they were
configured as separate kernel loadable modules, making it easier to configure sound when
installing a Red Hat Linux system. Red Hat shipped these modified drivers with several
versions of Red Hat Linux. With the release of Linux 2.2, these changes were included as part
of the standard kernel distribution.
One other class of kernel sound driver is available, one that may become more common as
hardware manufacturers realize the importance of supporting Linux. Some hardware vendors
may choose to develop their own drivers and release them separately. For example, Creative
Labs has developed Linux drivers for their SoundBlaster Live! card. Often these drivers are
made available only as precompiled binaries. This approach allows hardware vendors to keep
information about their card designs proprietary, presumably to make it harder for
Sound Cards 147
competitors to copy. The disadvantage is that, without source code, users cannot enhance the
driver or fix bugs themselves, and the modules may work only with specific versions of the
Linux kernel. Some people feel that this restriction goes against the Open Source spirit of
Linux. Whether it is truly advantageous to the hardware vendor is a subject of debate in the Linux
community.
Sound Drivers
In effect, kernel sound drivers come in four flavors, all of which are partially compatible but
have subtle differences and their own advantages and disadvantages. Although this
information is somewhat debatable, it is summarized in Table 6.1.
Table 6.1 Sound Driver Comparison
Driver Advantages Disadvantages
OSS/Lite Freely available Not very actively developed
Included with Linux kernel anymore
Supports most sound cards No direct support for
plug-and-play
OSS Supports many sound cards Commercial product
Easy installation (costs $)
Multiplatform No source code provided
Supports plug-and-play
High compatibility with
OSS/Lite
Has additional features
ALSA Freely available Currently supports only a
Mostly compatible with few sound cards
OSS/Lite Not fully compatible with
Has additional features OSS/Lite
Is actively developed
Supports plug-and-play
Vendor Supplied May support sound cards May be binary only
that are not supported by any May not be fully compatible
Part
other driver with OSS/Lite
I
Looking ahead, the ALSA team plans to submit their code for inclusion into the standard Ch
kernel source sometime during the 2.3 kernel series, making it the ``official'' sound driver in 6
2.4. It is unclear if OSS/Lite will continue to be included as an alternative or will disappear
entirely.
Having said all that, although much of the information in this section is applicable to all
sound drivers, I will focus on the OSS/Lite drivers since they are shipped with most current
Linux distributions. Depending on your needs, you might want to explore ALSA or OSS, in
which case you should consult the documentation that comes with those products.
The information in this chapter is applicable to Linux on the Intel x86 platform, and to a
lesser extent on systems that use ISA or PCI bus-based sound cards. On other Linux
148 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
platforms where sound hardware is built in, the process is different, so you should consult
documentation specific to that platform.
Sound Card Technology
The typical PC sound card consists of several different functional blocks.
A digital-to-analog (D/A) converter, sometimes inaccurately called a Digital Signal Processor
(DSP), converts between the digital form that sound files are stored in a computer to the
analog form required for playback to a loudspeaker.
In digital form, sound is represented as a sequence of discrete numbers, known as samples,
representing the value of the sound pressure at specific points in time. The number of bits
used to store each sample is known as the sample size. The greater the sample size, the more
accurately the sound can be represented. The most common sound sample sizes are 8 and 16
bits (hence, the common classification of sound cards as 8- or 16-bit cards). The rate at which
the sound samples are measured is known as the sample rate. The greater the sampling rate,
the more accurately the sound card can be reproduced. A sampling rate of 8,000 samples per
second with an 8-bit sample size provides telephone-quality sound. CD audio uses 44,100
16-bit samples per second. The trade-off with using large sample sizes and sample rates is that
the amount of data storage required increases.
The complement to the D/A converter is the analog-to-digital (A/D) converter used for
recording sound. It converts analog signals such as that from a microphone to digital form
that can be stored in a computer. These two devices, the A/D and D/A converters, allow
sound to be recorded, stored on a computer, and later played back. The sound can be speech,
music, or anything else that can be captured by a microphone.
The first generation of sound cards used a simple and inexpensive technique called FM
synthesis for generating sound. This type of sound generation requires very little effort on the
part of the processor to produce sound but is generally limited to music and sound effects.
The disadvantage of FM synthesis is that it is artificially created and therefore generally does
a poor job of reproducing the sound of real musical instruments (and cannot handle human
speech at all). Most sound cards today continue to provide an FM synthesis chip for backward
compatibility.
Many modern sound cards use a technology called wavetable synthesis that is essentially a
hybrid of the D/A converter and FM synthesis techniques. The sounds of real musical
instruments, converted to digital format, can be stored in memory on the sound card itself
and played back. Circuitry on the sound card allows this to be done mostly in hardware. This
technology produces a much more realistic sound when producing computer music.
Tip #38 from
Jeff The timidity program performs wavetable synthesis in software using any sound
card with a D/A converter. This provides sound quality comparable to that of
wavetable cards at the expense of some processing power.
Sound Cards 149
An essential component of all sound cards is a mixer. It is the circuit that controls volume
levels of signals passing through the sound card, controls effects such as bass and treble, and
selects the input sources when recording and playing back sounds. You could think of it as a
set of volume and tone controls that can be adjusted in software.
Many sound cards include a MIDI bus interface. MIDI, an acronym for musical instrument
digital interface, is a standardized protocol for connecting together and communicating with
electronic musical instruments. Common MIDI devices include sound synthesizers, piano
keyboards, and even computer-controlled lighting systems. MIDI is most commonly used by
professional and serious amateur musicians.
Because sound cards are often used for games, many also include one or more joystick ports.
Collecting Hardware Information
Configuring sound cards under Linux can be a little tricky and unpredictable. It might be
completely straightforward (particularly if you have one of the more common sound cards) or
next to impossible (for example, the sound hardware built in to some laptop computers can be
unique). Fortunately, you can postpone sound card support until after your basic Linux
system is up and running. In addition, Linux distributions are continually being improved to
make configuration simpler.
The first step in configuring your sound card is to collect as much information about your
sound hardware as possible. If you've been reading this book in sequence, then you probably
did that in Chapter 2, ``Linux Installation Overview.'' If not, then it's time to collect that
information now.
Although you can often successfully configure sound support with incomplete information, as
much of the following information as you can get will be useful:
& The sound card manufacturer and model name
& Settings for IRQ number, DMA channels, and I/O ports
& The type of sound chip used (particularly for ``no-name'' cards)
Part
& The type of bus used (for example, ISA, ISA plug-and-play, or PCI)
I
If necessary, you can open the PC and physically examine the card to determine these settings Ch
(for example, look at jumper switch settings).
6
It is also helpful if you can verify that the card was working under another operating system
before Linux was installed. Using the same settings under Linux is not strictly necessary but is
recommended.
Tip #39 from
Jeff Under Microsoft Windows 95 or 98 you can use the System Properties applet to
display the resources used by sound cards and other devices.
150 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Tip #40 from
Jeff A note on compatibility: Virtually all sound card manufacturers claim their products
are ``SoundBlaster compatible.'' Essentially, this claim means only that these products
provide software drivers for MS-DOS and Windows that are compatible with the
SoundBlaster card drivers. Under Linux, this software cannot be used, so the cards
must be compatible at the hardware level with a card that is supported under Linux.
Although few people have this luxury, the ideal situation is that you select the model
of sound card being purchased to ensure that it is compatible with Linux. Fortunately,
most popular sound cards do work under Linux, and some manufacturers are now
indicating Linux compatibility in their product marketing literature.
Configuration Methods
You can choose from three main methods of configuring sound support under Linux. They
are covered here in order of preference, starting with the simpler methods that may not work
in all cases and then the more complex methods that you might need to fall back on.
Using an Automated Configuration Program
Some Linux distributions include a utility to configure sound support in an automated or
semi-automated manner. It can be a standalone tool or part of a larger system administration
utility. When this method works, using such a utility is the easiest way to configure sound and
appears to be the way that Linux distributions are headed in the futureÐeasy-to-use, often
graphical, system administration tools as an alternative to (but not replacement for) the
traditional configuration file-based system administration. Note that you generally need to be
logged in as the system administrator (user root) to perform these functions.
Under Red Hat Linux, using the sndconfig program is the recommended way to configure
sound. This utility first probes for plug-and-play sound cards and displays any cards found. If
no plug-and-play cards are available, it presents you with a list to select from. You then select
the settings for I/O port, IRQ number, and DMA channels. For plug-and-play cards, the
program can optionally pick the settings automatically. The program then saves the card type
in a file (/etc/sysconfig/soundcard) and for plug-and-play cards creates the appropriate
configuration file (/etc/isapnp.conf) for the ISA plug-and-play utilities to configure the
card on system startup. The file /etc/conf.modules is configured to load the appropriate
kernel modules for the selected type of sound card. As a final check, it even plays a sound file
of Linus Torvalds speaking to verify that the sound card is working.
Under Caldera OpenLinux, the COAS system administration tool provides a similar
capability to configure sound cards.
Other Linux distributions may include sound configuration tools. If not, or if the automated
tools fail for your system, you should use the manual method described in the next section.
If you are using the OSS or ALSA sound drivers, you should use the configuration tools that
come with that software.
Sound Cards 151
Tip #41 from
Jeff You can obtain utilities such as sndconfig separately and install them on other Linux
distributions even if they were not provided by your vendor. Make sure you install
any other packages required by sndconfig.
Loading Kernel Modules
The second method is to manually perform the same steps that were described in the
preceding section.
The first step is optional and applies only to ISA plug-and-play sound cards. Unless you're
using a sound driver that directly supports plug-and-play devices (that is, OSS or ALSA), you
need to use the isapnp tools to configure the sound card settings. Follow these steps:
1. Save a copy of the existing /etc/isapnp.conf file, if any.
2. Use the pnpdump tool to generate a new/etc/isapnp.conf file. This is typically done
with the command pnpdump >/etc/isapnp.conf.
3. Edit the file (typically by uncommenting the appropriate lines) to select the desired
device settings.
4. Run isapnp to configure the ISA plug-and-play cards in your system.
You can find more details on the ISA plug-and-play tools in the system manual pages for
ispnp, pnpdump, and isapnp.conf.
The appropriate kernel sound driver modules are loaded by the kerneld daemon (or in newer
kernels, the kmod thread in the kernel). The configuration file /etc/conf.modules needs to be
configured to indicate the drivers to be loaded. Again, you can find the detailed
documentation in the manual pages for kerneld. A typical portion of the configuration file for
a SoundBlaster 16 card using the OSS/Lite drivers is shown in Listing 6.1.
Part
Tip #42 from
I
Jeff An advantage of using sound drivers built as kernel loadable modules is that
kerneld (or kmod) will automatically unload them if they are not used, freeing up
Ch
memory for other uses.
6
Listing 6.1 Typical /etc/conf.modules Entries for the OSS/Lite Sound Driver
# 2.2.0 OSS/Lite sound drivers
alias sound sb
alias synth0 midi
alias midi opl3
options opl3 io=0x388
options sb io=0x220 irq=10 dma=3 dma16=7 mpu_io=0x330
152 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
For the ALSA driver, the module names are different. An example is shown in Listing 6.2.
You should consult the ALSA documentation to determine the correct options for your
sound card.
Listing 6.2 Typical /etc/conf.modules Entries for the ALSA Sound Driver
# 0.4.0 ALSA sound drivers with OSS emulation
alias char-major-116 snd
alias char-major-14 soundcore
alias snd-card-0 snd-sb16
alias snd-slot-0 snd-card-0
alias sound-service-0-0 snd-mixer-oss
alias sound-service-0-1 snd-seq-oss
alias sound-service-0-3 snd-pcm1-oss
alias sound-service-0-4 snd-pcm1-oss
options snd snd_cards_limit=1
# options for SB16
options snd-sb16 snd_port=0x220 snd_mpu_port=0x330 snd_irq=10 \
snd_dma8=3 snd_dma16=7 snd_dma8_size=64 snd_dma16_size=128 \
snd_mic_agc=1 snd_isapnp=0
# after loading, initialize mixer
post-install snd-sb16 alsactl restore
Under OSS, you generally use the soundon script to load and configure the drivers.
When the sound drivers are configured properly, they should be loaded automatically when
an application opens a sound device. You can proceed now to the section on testing.
The presumption here is that the appropriate kernel sound modules were provided for you
when the system was installed. Typically, they are located in the directory /lib/modules/
x.y.z/misc, where x.y.z is the kernel version. If that is not the case, you need to create them
as part of building a new kernel. This task is beyond the scope of this chapter and is covered
in Chapter 14, ``Configuring the Linux Kernel.''
If you are new to Linux and are not yet comfortable with system administration tasks such as
editing configuration files and rebooting the system, I recommend that you gain a basic
familiarity with the system, such as is covered in Part II of this book, ``System
Administration,'' before attempting to set up kernel sound support.
Building Sound Support into the Kernel
The third and final method of setting up sound support is to compile the appropriate sound
drivers into the kernel. With this method, you also have to configure plug-and-play cards, but
because the drivers are linked into the kernel itself, you don't need to worry about the loading
of modules.
Again, the process of configuring, building, and installing a new kernel is beyond the scope of
this chapter and is the subject of Chapter 14.
Sound Cards 153
Testing Your Sound Card
To verify that your sound card configuration is correct, answer these three questions:
& Can you load the sound drivers into the kernel?
& Did the drivers find the sound card?
& Do the sound devices work?
If configured properly, the kernel drivers should automatically be loaded if any sound devices
are accessed. An easy way to access them is to run the command cat /dev/sndstat, which
displays the contents of the sound status device. If all goes well, you should see a result that
shows information about the loaded drivers. A typical output is shown in Listing 6.3.
Listing 6.3 Example of /dev/sndstat Output for the OSS/Lite Sound Driver
Sound Driver:3.5.4-960630 (Sat Jan 4 23:56:57 EST 1997 root,
Linux fizzbin 2.0.27 #48 Thu Dec 5 18:24:45 EST 1996 i586)
Kernel: Linux fizzbin 2.0.27 #48 Thu Dec 5 18:24:45 EST 1996 i586
Config options: 0
Installed drivers:
Type 1: OPL-2/OPL-3 FM
Type 2: Sound Blaster
Type 7: SB MPU-401
Card config:
Sound Blaster at 0x220 irq 5 drq 1,5
SB MPU-401 at 0x330 irq 5 drq 0
OPL-2/OPL-3 FM at 0x388 drq 0
Audio devices:
0: Sound Blaster 16 (4.13)
Synth devices:
0: Yamaha OPL-3
Part
Midi devices: I
0: Sound Blaster 16
Ch
Timers:
0: System clock
6
Mixers:
0: Sound Blaster
Keep in mind that this output is for OSS/Lite. For OSS, sample output is shown in
Listing 6.4.
154 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Listing 6.4 Example of /dev/sndstat Output for the OSS Sound Driver
OSS/Linux 3.9.2p 4Front Technologies 1996-1999
License serial number: N12345678
Options: AWE APCI MIX
This copy of OSS is licensed to Jeff Tranter
Build: 2.2.10-UP
Card config:
Software mixing (audio)
Generic PnP support
SoundBlaster PnP at 0x220 irq 10 drq 3,7
SB MPU-401 at 0x330 irq 10
Audio devices:
0: Creative SB16 PnP (4.13) (DUPLEX)
1: SB secondary device (DUPLEX)
2: SoftOSS v1.2 CH #0
3: SoftOSS v1.2 CH #1
4: SoftOSS v1.2 CH #2
5: SoftOSS v1.2 CH #3
6: SoftOSS v1.2 CH #4
7: SoftOSS v1.2 CH #5
8: SoftOSS v1.2 CH #6
9: SoftOSS v1.2 CH #7
Synth devices:
0: SoftOSS v1.2
Midi devices:
0: Sound Blaster 16
Timers:
0: System clock
1: SoftOSS
Mixers:
0: Sound Blaster
1: SoftOSS
If reading/dev/sndstat produces an error such as cat: /dev/sndstat: Operation not
supported by device, then your modules are not being loaded. Proceed to the
``Troubleshooting'' section at the end of this chapter.
The output of /dev/sndstat should indicate that the sound drivers were correctly detected at
the correct addresses. If devices were not found, then check out the ``Troubleshooting''
section.
Note that if you are using the ALSA drivers, you don't have /dev/sndstat; the equivalent is
the /proc/asound directory. Looking at the files in this directory should give you an
indication of whether the drivers are loaded.
Audio CDs 155
If the drivers are loaded, it's time to test the card. A good first test is to run a mixer program
and verify that you can set the gain settings without any errors. Common mixer programs
include xmixer, aumix, and kmix. Setting all levels to their maximums is a good idea for now;
this way, you can make sure that if the card is working, you'll be able to hear it.
To test the D/A converter, you can use a sound player program such as play or kmedia to play
a sound file. You likely have some sound files on your system; look for files with the extension
.wav or .au. (Using the command locate *.wav *.au is a good way to find some.) You should
hear the sound played through the speakers. Under Red Hat Linux, for example, the famous
sound file of Linus Torvalds speaking can be found in /usr/share/sndconfig/sample.au.
To test the A/D converter, you need a microphone and some software for recording sound to
a sound file. You might have to dig around on your machine or download some software to
do your testing. Look for programs called rec or vrec.
To test the FM synthesizer on a sound card, you need a MIDI player application such as
playmidi, kmid, and kmidi and some MIDI files (the file extension is usually .mid). Try
playing a MIDI file with one of these applications. Because you're checking FM synthesis,
you probably won't be very impressed by the sound quality.
Testing the MIDI bus interface on a sound card is outside the scope of this chapter. If you are
fortunate enough to have MIDI devices, you are definitely in the ranks of the advanced users.
Audio CDs
If your sound card is working now, you've probably made it over the biggest hurdle. This
section describes how to get your CD-ROM drive to work with the sound card to play music
CDs.
Audio CD players and CD-ROM drives use the same basic technology. Most CD-ROM
drives can play audio CDs. Listening to some music while programming is a great way to
relax.
Part
To use your CD-ROM drive, you need software to control it. Many CD player programs
exist, ranging from simple command-line tools to complex graphical applications. Some of I
the more common programs include cdplay, cdp, kscd, and xplaycd. Ch
Although most CD-ROM drives include a front panel headphone jack and volume control, 6
most users connect the drive to their sound card and play it through speakers.
Tip #43 from
Jeff When you're playing audio CDs, the drive always runs at single speed. Your drive may
make a noticeably different sound when playing audio. If you want to use a CD-ROM
drive only to play music, then don't waste your money on an expensive high-speed
drive. In fact, if you have an old drive lying around and a spare IDE device port in
your machine, you might want to consider installing it as a second drive. That way,
you can listen to music even when you're using the CD-ROM for data.
156 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Installation
The first prerequisite for installation is to make sure that the CD-ROM drive is working
properly for accessing data CDs. If you installed Linux from CD, then you should be in
business; otherwise, go back and review the appropriate sections in Chapters 2 through 5.
Now you can insert a music CD, plug headphones into the front panel headphone jack, and
run a CD player application. You should be able to listen to music without any involvement
from your sound card. In fact, if you don't have a sound card, you can plug amplified speakers
into the headphone jack and listen to CDs. If you don't get any sound from the front panel
jack, look at the suggestions in the ``Troubleshooting'' section.
You may already hear sound from the speakers connected to your sound card. If not, fire up a
sound mixer program and set the CD and master gain levels to suitable levels for listening.
One note on a point of possible confusion: If you have accessed a CD-ROM drive under
Linux, you may know that you need to mount the drive. You don't mount audio CDs; doing
so only produces error messages.
Doing More with CD Audio
Although audio CDs do not contain information on the artist or track names, they do have a
unique signature made up of the number of tracks and lengths of each track. Many CD player
programs can match this signature against a list of data for known discs and display the track
information. You can enter the information yourself and save it so that the application will
display it when the disc is played in the future. You can access databases of CD-ROM
information on the Internet that some CD player applications can directly download when
playing a CD, saving you the trouble of entering it.
Programs such as cdparanoia can ``rip'' digital audio from CDs and save it to a sound file.
Because the data is stored entirely in digital format, no degradation of the quality occurs. At
the high sampling rates and sample sizes used for CD audio, this format requires a lot of
storage space. Using an MP3 audio encoder, you can compress this format down by a factor
of approximately 10. You then can play the resulting file with an MP3 player program.
Tip #44 from
Jeff CDs can have both data tracks for storing software as well as audio tracks containing
sound. Some games use audio tracks to play the music that accompanies the game.
Even if the game does not run under Linux, you can listen to the audio tracks using a
CD player program.
Joysticks 157
Joysticks
Joysticks are most commonly used for playing games, and several Linux games can make use
of them. They have other applications too, being a suitable input device whenever a
continuously variable analog input is required. Other creative applications include using a
joystick as a way to shut down or reboot a Linux system that lacks a keyboard.
In Linux 2.0 and earlier kernels, the joystick driver was a separate package that had to be
obtained and compiled separately. With Linux 2.2 and later, the joystick drivers are included
in the standard kernel distribution. Many improvements were made in the drivers in 2.2,
including support for many more types of joysticks and those found on the non-Intel x86
versions of Linux.
At the time of this writing, at least 32 different models of joysticks are supported by 11
different kernel drivers. They include devices that connect through a joystick connector,
parallel port, serial port, or (on some platforms) dedicated joystick interfaces built in to the
computer. Joysticks using USB interfaces are not yet supported.
Loading the Kernel Joystick Modules
Like most kernel drivers, the joystick drivers can be compiled into the kernel or built as
kernel loadable modules. Using modules is preferred because this approach provides more
flexibility and allows the drivers to be unloaded when not needed, freeing up more memory.
The presumption here is that the appropriate kernel joystick modules were provided for you
when your Linux system was installed. Typically, they are located in the directory /lib/
modules/x.y.z/misc, where x.y.z is the kernel version. If that is not the case, you need to
create them as part of building a new kernel. This task is beyond the scope of this chapter and
is covered in Chapter 14.
If the joystick drivers are built and installed, you should be able to manually load them. For a
traditional PC analog joystick, you can load the driver by using the command modprobe
joystick-analog. This command should result in a message either sent to the console or to
Part
the kernel message log (displayed using the command dmesg) such as this:
I
js: Version 1.2.13 using 166 MHz RDTSC timer.
js0: Analog 3-axis 2-button joystick at 0x201 Ch
With a properly configured /etc/conf.modules file, the kernel automatically loads the 6
drivers when the joystick devices are accessed. The following is a typical entry for an analog
joystick:
# joystick - 2 button, 3 axes
alias char-major-15 joy-analog
options joy-analog js_an=0x201,0x3f
A program called jstestis included with the joystick driver package; it verifies the operation
of the driver with your joystick.
158 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Joystick support is still being actively developed. For the most up-to-date documentation,
you should consult the documentation on joysticks included with the kernel in /usr/src/
linux/Documentation. You also might want to download the full package from the joystick
driver author's Web site to pick up the latest changes and the test programs.
Other Multimedia Devices
As Linux matures, support for more unusual and interesting hardware devices increases. This
support can sometimes be challenging if the hardware vendors do not provide much
information on programming the devices. Some of the general categories of multimedia
devices include the following:
& Full-motion video cards than can capture video for storage and later playback
& TV and radio tuner cards for turning your PC into a software-controlled radio or
television
& Digital cameras for capturing images that can be displayed and manipulated
& Scanners for digitizing paper documents
Linux currently supports many CD-R, CD-RW, DVD-ROM, and DVD-RAM drives. DVD
video support is still in the early stages at the time of this writing. USB support is coming but
is very incomplete, mostly limited to the most common devices such as keyboards and mice.
This area is very dynamic, with support for new devices appearing almost daily. To find out
the current status of these devices, you can check Internet Web sites related to Linux. Several
sources are listed later in this chapter. You might need to run the latest development version
of the kernel or even install special kernel patches.
If you have a hardware device and are uncertain whether it is supported under Linux, you can
do a little research to dig up some information. Specialized hardware is most likely to work if
it uses a standard interface such as a serial, SCSI, or parallel port. Specialized interface cards
and protocols can be a problem.
If no driver exists, writing a kernel driver is not necessarily the daunting task that it may
appear. If you have a good understanding of how to communicate with the hardware device
and a knowledge of C programming, writing a kernel driver can be within the ability of a
Linux hobbyist.
Multimedia Applications
When your sound card or other multimedia hardware is up and running, many applications
are available for you to make use of that hardware. Only a few of the major categories are
listed here. Some of the resources listed in the next section can point you to specific
programs. Popular multimedia applications include the following:
Information Resources 159
& Sound file playback and recording tools
& CD-player programs
& MP3 players
& MIDI file players, sequencers, and editors
& Sound mixers
& Video players (for file formats such as MPEG, QuickTime, and AVI)
& Graphics editing and manipulation tools
& Image rendering tools
& Toolkits for developing multimedia applications
& Games of all types
Trying Linux multimedia applications can be an interesting experience. The quality can
range from a hack full of bugs written by a beginning programmer to a professional
application that is as good as or better than commercial applications costing hundreds of
dollars.
Information Resources
You can find a plethora of information about Linux and multimedia, much of it available on
the Internet.
Tip #45 from
Jeff The Linux Documentation Project (LDP) is an initiative to create freely available
documentation for Linux. Out of this project has come the well-known Linux How-To
documents and several book-length manuals. Often you can find LDP manuals in
Linux systems in the /usr/doc directory or on the Internet at http://
www.linuxdoc.org or http://metalab.unc.edu/LDP .
Part
I
Most Linux commands have manual pages (also called man pages) that you can access by using
man command. GNU tools are documented using the info format that can be viewed using
the
Ch
GNU Emacs or the info program. In addition, graphical environments such as KDE and 6
GNOME typically provide graphical browsers for searching and accessing online
documentation, including info and man pages.
The Linux kernel comes with some documentation. Although usually quite terse, these
README files are often the only documentation for recent development in the kernel. They
are conventionally located in the directory /usr/src/linux/Documentation if you have the
kernel source code installed.
The Web is the primary information resource on the Internet. A few useful sites (subject to
change) related to the topics in this chapter are listed in Table 6.2.
160 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Table 6.2 Web Sites for Multimedia on Linux
URL Description
http://www.alsa-project.org The ALSA Project
http://www.opensound.com Open Sound System
http://www.linuxdoc.org The Linux Documentation Project
http://www.mostang.com/sane SANE (Scanner Access Now Easy)
http://www.freshmeat.net Announcements of Linux Applications
http://sound.condorow.net Linux MIDI and Sound Applications
For answers to specific questions, you can check out the many Linux-related Usenet
newsgroups and mailing lists.
Finally, you might be able to find a local Linux user group in your area. Joining this user
group can be a great way to get answers and find out what is happening locally. If you can't
find a user group, why not consider starting one?
Troubleshooting
Symptom: Kernel modules don't load
You might have an incorrect /etc/conf.modules file. Edit the file and make sure that the
correct drivers are specified. You can manually load the modules using the insmod or
modprobe commands to experiment with the settings.
The kerneld daemon may not be running (or kmod support is not compiled into the kernel).
Check if the kerneld process is running using the ps command. It should be started by one of
the system startup scripts.
Also, the kernel sound modules may not have been installed in the correct directory, usually
/lib/modules/x.y.z/misc. If you built the kernel, make sure that you ran the commands
make modules and make modules_install to build and install the modules.
Symptom: Sound card not detected
You may have the wrong sound card drivers loaded. Try loading a different driver (for
example, many cards listed as SoundBlaster compatible are really equivalent to the Windows
Sound System).
The wrong I/O address may have been specified for the sound driver. Check if the settings
used match the card jumpers or plug-and-play settings.
The plug-and-play settings may not have been configured correctly. Check the
/etc/isapnp.conf file and verify that the selections there are correct.
Troubleshooting 161
Symptom: IRQ/DMA timeout
If this occurs when playing a sound file, you most likely selected the wrong IRQ number or
DMA channel. Check that you used the setting that match the card jumpers or plug-and-play
setting.
The plug-and-play settings may not have been configured correctly. Check the
/etc/isapnp.conf file and verify that the selections there are correct.
Symptom: Device Conflicts
You have an IRQ, DMA, or I/O setting conflict between two devices. Note that conflicts can
occur, even for plug-and-play devices. You need to ensure that resources do not conflict
between cards. Find out what settings are possible for jumper settings and Plug and Play
cards. If this becomes complex, I suggest drawing a table. Find a combination of settings that
avoid conflicts and configure your cards accordingly.
Symptom: ``kernel-module version'' mismatch error
The sound modules may have been built with a different kernel from the one being run; you
need to recompile using the current kernel version. You may have built a kernel but neglected
to build the modules, install them, and update the dependencies file. Make sure you have run
the commands make modules, make modules_install, and depmod -a.
Symptom: Sound works only for user root
You need to change the permissions on the device files to allow non-root users to read/write.
For example, the command chmod a+rw /dev/dsp would grant all users access to the DSP
device.
Symptom: No sound playback but no error messages
If this occurs, the most likely reason is that you are not setting the mixer gain levels. Use a
mixer program to set the levels to a suitable value.
Symptom: Cannot record
This can be caused by not setting the mixer gain levels or not setting the mixer recording
Part
source. Use a mixer program to correct it. Another possibility is a microphone that is bad or
plugged into the wrong input jack. Use a known good microphone and check where it is
I
plugged in. Ch
Symptom: ``device busy'' error
6
This can be due to a device conflict (described earlier). Another possibility is that another
application has opened the audio device. Use the ps and fuser commands to check for a
process which is using sound.
Symptom: No sound after a reboot
Make sure that the system startup scripts initialize plug-and-play devices. Verify that the
sound drivers get automatically loaded on system startup. Set the mixer volume levels to an
appropriate value using a mixer program.
162 Chapter 6 Adding Sound Cards and Other Multimedia Hardware
Note: OSS uses the soundon script to load the drivers. This can be automated by adding a call
to soundon in a system startup script such as rc.local.
Symptom: CD doesn't play through headphones
This indicates a problem that is not related to a sound card. First, make sure the drive works
with data CDs. Use a known good audio CD. If it still fails, try another CD player program.
Symptom: Plays through headphones but no sound from sound card
This indicates a problem with the sound card. Make sure that there is an audio cable
connected from the CD-ROM drive to the sound card. Check that the mixer volume and CD
gain levels are correctly set using a mixer program. Make sure that the speakers are turned on
and have power (if applicable), volume controls are turned up, and they are connected to the
correct sound card jack.
It is possible that you have a sound card with a nonstandard mixer circuit that is not
supported under Linux. A new kernel sound driver or alternative drivers such as OSS or
ALSA may support the card. You may also be able to work around the problem by initializing
it under another operating system and soft-booting into Linux.
Symptom: Sound skips
This can be caused by a dirty or scratched CD; replace with a known good one. The
CD-ROM drive may also need cleaning or is simply defective.
Symptom: Only user root can play CDs
You need to set file permissions on the device file for the CD-ROM drive to allow users to
read and write.
CHAPTER
7
Upgrading and Installing
Software
In this chapter by Jack Tackett, Jr.
Understanding Key Terms Used in This Chapter 164
Understanding the Politics of Upgrading 165
Installing Software 166
Using the Red Hat Package Manager 167
Using the Debian Package Management System 174
Installing Non-Linux Software 175
Case Study: Upgrading Your Kernel 178
164 Chapter 7 Upgrading and Installing Software
Understanding Key Terms Used in This Chapter
The base Linux system initially contains only a core set of utilities and data files. The system
administrator installs additional commands, user application programs, and various data files
as required. Applications get updated frequently. System software changes as new features are
added and bugs are fixed. The system administrator is responsible for adding, configuring,
maintaining, and deleting software from the Linux system.
The word installing means copying the associated program files onto the system's hard disk
and configuring the application (assigning resources) for proper operation on a specific system.
The configuration of a program instructs it as to where parts of the application are to be
installed and how it is to function within the system environment in general.
Both the Red Hat and Caldera distributions of Linux ease the pain of installing and
upgrading software by including the Red Hat package management system accessed via the
rpm command. However, you'll also find yourself installing software that isn't in rpm format.
Many of the software packages available on the Internet are in compressed tar format.
On large systems, an administrator usually installs applications because most users don't have
access to the tape or floppy drives. Administrative permission is also often needed to install
components of the applications into system directories. Components may include shared
libraries, utilities, and devices that need to go into directories that normal users can't access.
Table 7.1 lists some terms and definitions that you should become familiar with.
Table 7.1 Terms Related to Application Installation
Term Definition
Superuser The highest privileged user on the system. Also called the root user.
System administrator The person in charge of keeping the Linux system optimized and
properly running. The system administrator has superuser
privileges and can install new software onto the system.
Dependencies The reliance/requirement of various software components to each
other. A given program may require a specific library or other
daemon in order to work properly.
Installing applications The initial installation or update of a program for a UNIX system.
The process usually requires superuser privileges and access to the
computer's tape or floppy disk drive.
Configuring The act of setting up an application to work with your particular
system. Configuring can include setting up the application for
many users to use, putting it in accessible directories, or sharing it
with the network.
Understanding the Politics of Upgrading 165
Understanding the Politics of Upgrading
What software should you upgrade? How often should you upgrade? The answers to these
questions are largely determined by the purpose of your systemÐpersonal or businessÐand
the demands of your users. Software versions are changing all the time. Various parts of the
Linux system are constantly being updated. You wouldn't have time to use your system if you
tried to keep up with each and every upgrade that comes out.
Typically, you shouldn't have to reinstall the entire Linux system when you upgrade your
system software. Usually, only a tiny portion of the system software changes with a new
release. You might have to upgrade your kernel or upgrade your system libraries, but you
probably won't have to do a full reinstallation. However, when you upgrade software
packages, you quite often have to completely install a new version, especially if you're several
versions behind when you upgrade.
Tip #46 from
Jack The Red Hat package manager used by Red Hat and Caldera provides a mechanism to
ftp:
install packages across the net via
rpm -u ftp://ftp.netwharf.com/pub/rpms/somepkg-1999.rpm
The Red Hat distribution also allows you to install over the network from an
automated script called a kickstart installation. See http://www.redhat.com/
mirrors/LDP/HOWTO/KickStart-HOWTO.html for more information.
Making a current backup of your system before upgrading software is a good idea.
That way, if something goes wrong, you can always get back to your original system.
? See ``Considering Backup Tips,'' p. 261
In general, you should upgrade your system if a new version of either system or application
software fixes a serious problem or adds functionality that you need. It's up to you to
determine what constitutes a serious problem. If a new release of a software package fixes
something that has caused problems on your system, poses a serious security hole, or fixes a
bug that could damage your system, installing it is probably worth your time.
Part
I
Don't try to keep up with every release of every piece of software; upgrading for the
Ch
sake of upgrading takes too much time and effort. With a little research, you can keep
your system working in good condition and update only the parts that need 7
upgrading as you go along.
166 Chapter 7 Upgrading and Installing Software
Installing Software
Installing a major program onto a Linux system is more complicated than installing a similar
program on a single-task operating system, such as MS-DOS or Apple Macintosh System 7.6.
The multiuser nature of Linux means that every application on the system sometimes receives
simultaneous calls for access.
To further complicate installation, most application programsÐwith the exception of very
simple onesÐrequire configuration to your specific system before they can be used. It's up to
the system administrator installing the software to identify items specific to the system's
configuration when prompted during an application's configuration process.
For example, one user may have only an older character-based terminal, whereas another has
a fancy new X Windows terminal. The superuser must make sure that the application
responds correctly to the older terminal, sending only ASCII charactersÐthat is, letters and
numbersÐand that the X Windows terminal receives full advantage of the application's
colors and graphics. The system administrator manages the system and has the responsibility
of keeping it optimized (all programs up to current versions, proper user accounts assigned,
and so forth).
As already stated, loading a program onto a Linux system is more complicated than doing so
on single-user operating systems. The system administrator who's installing an application
might have to create new directories to house the files associated with a particular program.
Some software packages call for the configuring or reconfiguring of system devices. Although
the end user worries only about learning the new program's features and operating
commands, the superuser must make sure that system resources are properly allocated,
configured, and maintained for the program (while, of course, not messing up any already
installed applications).
Installing software by using menus or commands is outwardly a relatively simple task; to the
system itself, however, the task is complex. Applications for single-user operating systems,
such as DOS programs, usually run only one copy of themselves at a time and have no
competing programs. In even a simple Linux installation with only one user logged in, many
processes can be running at the same time. Multiply this activity by several users all running
programsÐincluding some users who use the same applicationÐand the complexity increases
dramatically.
The Linux operating system excels at juggling a multitude of processes, programs, users, and
peripherals simultaneously. To live in a Linux environment, an application must be properly
loaded. An ill-behaving application, or one improperly installed, can cause a system crash
(when a process or program goes wild and locks the CPU, causing it to lose control of all the
currently running programs). The system shuts down, all users are kicked off, and their
programs are interrupted. There's often much wailing and gnashing of teeth from frustrated
persons in the midst of some complicated task.
As the one loading a new application, the system administrator or superuser is responsible for
making sure that the application is compatible with the system and testing the application
Using the Red Hat Package Manager 167
after it's installed. Understanding the loading of software onto a Linux system first requires a basic
knowledge of the responsibilities and privileges of the system administrator.
Understanding the System Administrator's Job
If you use Linux on a small system, you're probably your own system administrator. You
install and run your applications. It's your responsibility to keep a current backup of files,
maintain a proper amount of free space on the hard disk, make sure that the system runs
optimally through memory management and other means, and do everything else required in
the administration of an efficient and productive system. If you're a user in a larger system
environment, a specific person probably handles system administration. The following list
briefly summarizes what the system administrator does:
& Starts and stops the system, as needed.
& Makes sure that enough free disk space is available and that file systems are free of
error.
& Tunes the system so that the maximum number of users have access to the system's
hardware and software resources and so that the system operates as fast and as
efficiently as possible.
& Protects the system from unauthorized entry and destructive actions.
& Sets up connections to other computer systems.
& Sets up or closes user accounts on the system.
& Works with software and hardware vendors and with those with training or other
support contracts for the system.
& Installs, mounts, and troubleshoots terminals, printers, disk drives, and other pieces of
system and peripheral hardware.
& Installs and maintains programs, including new application programs, operating
system updates, and software-maintenance corrections.
& And nothing else. Too often system administrators log in as root and do everything
from there, but doing so can cause a myriad of problems on your system. Use system
administration tasks, and use your user account for day-to-day tasks!
Using the Red Hat Package Manager
Both the Red Hat and the Caldera OpenLinux distributions make use of Red Hat Package
Part
Manager (RPM) packages (or RPMs) for managing software installation.
I
Ch
The Debian distribution also uses packages but in a different package system format.
Also, typically (although not required of either system), RPMs have an .rpm 7
extension, whereas Debian packages have a .deb extension.
168 Chapter 7 Upgrading and Installing Software
A package contains a complete, fully tested, and configured program. The package is typically
built from a source code package so that developers and users know what they are getting. To
manage these packages, Red Hat Software developed the Red Hat Package Manager and
released it to the world.
The current version, RPM 3.0, is backward-compatible with previous versions of RPM;
however, parts of the program were totally rewritten to improve the product. For more
information, see the white paper on Red Hat's Web site at http://www.redhat.com/
knowledgebase/rpm3.0/.
RPM has six modes of operation, five of which can be used from either the command line or
the X Windows-based tool called Glint. The various modes are installing, uninstalling,
updating, querying, verifying, and building. You can build an RPM package only in text mode
from the command line.
For more information on building packages with RPM, see the book Maximum RPM
(Sams Publishing) or Red Hat software's Maximum RPM.
You use RPM from the command line in the format
rpm [options] package_name
where optionsis one of many different flags used by RPM to manipulate packages, and
package_name indicates the software package to be used. The package name usually looks like
this: quota-1.55-4.i386.rpm. The package name follows this format:
Name quota
Version 1.55
Release 4
Computer architecture i386
Extension .rpm (typically)
However, the package file can be any name because the information about the package itself
is contained inside the file.
Locating Packages
You can find most packages provided with your distribution on the CD-ROM under the
directory /RedHat/RPMS. To mount the CD-ROM and list the various packages available, use
the following commands:
cd /mnt
mount CD-ROM
cd CD-ROM/RedHat/RPMS
ls | more
Using the Red Hat Package Manager 169
Most of these packages were installed during your installation of Linux. However, if you
decided not to install certain packages, you can install them now from this collection of
packages.
RPM also allows you to install packages located on other computers by using FTP, as you
will see in the next section.
Installing Packages with RPM
To install a package from the command line, you use the -i option like this:
rpm -i quota-1.66-6.i386.rpm
This command installs the quota package on your system. The -i option instructs the rpm
command to install the package quota-1.66-6.i386.rpm onto the local system. To run the
installation, RPM goes through a series of steps:
& Checks dependencyÐEach package may depend on other software already being
installed.
& Checks conflictsÐRPM checks to see whether a component is already installed or
that the component is not older than the one currently installed.
& Processes configuration filesÐRPM attempts to provide a proper configuration file,
and if it finds a previous configuration file, it saves that file for future reference.
& Installs filesÐRPM unpacks the various components from the package and places
them in the proper directories.
& Performs post-installation processingÐAfter installing the various components,
RPM performs any necessary tasks to properly configure the system.
& Updates the databaseÐRPM keeps track of all its action via a database.
The command provides no feedback during this installation, but you can use the -v (verbose)
option to get more information. Table 7.2 provides a list of other options you can use during
installation.
Table 7.2 Installation Options
Option Description
-vv Provides very verbose information.
-h #
Prints hash marks ( ) periodically during installation. These marks allow Part
you to see that RPM is actually doing something and is not just hung.
I
-percent Prints the percent completed during installation instead of #.
Ch
-test Does not install the package, but performs a dry run to test installation
and reports any errors. 7
-replacefiles Replaces files from other packages.
-force Tells RPM to ignore certain conflict errors and install the package
anyway.
170 Chapter 7 Upgrading and Installing Software
To install a package located on another machine, you can use an FTP-type URL to designate
the package:
rpm -i ft p://ftp.netwharf.com/pub/RPMS/quota-1.66-6.i386.rpm
Tip #47 from
Jack You can use the vh flags, eg
rpm -ivh ftp://ftp.netwharf.com/pub/RPMS/quota-1.66-6.i386.rpm
to provide verbose output and hash marks as the package is installed. The -ivh
argument is preferred; the output and hash marks are helpful to eliminate confusion
if the package does not install fully or the core dumps.
This command assumes that the remote machine accepts anonymous FTP.
? See ``Using FTP with a Web Browser,'' p. 706
If you need to specify a username and password to install the file, you can use the following
command:
rpm -i ftp://[email protected]/pub/RPMS/quota-1.66-6.i386.rpm
Password for [email protected]: <enter your password here>
You can enter your username and password in the command at the same time, like
this:
rpm -ivh ftp://mark:passw [email protected]/pub/RPMS/
quota-1.66-6.i386.rpm
However, this method for entering commands is not secure because someone could
look over your shoulder or (more likely) recall the command from your history file.
Uninstalling Packages with RPM
One of the benefits of using RPM is the ease of installing new programs. If you've heard
about a new program on the Internet, you can install the package and test the new program.
What happens, then, if you decide the software is not for you and you want to get rid of it?
Fortunately, RPM makes uninstalling a package just as easy as installing one. To uninstall a
package, you use the -e option:
rpm -e quota-1.66-6.i386.rpm
Using the Red Hat Package Manager 171
Tip #48 from
Jack You can indicate the RPM package by name only such as rpm -e quota without the
version indicators and it will work. The only exception is if you have more than one
version of the package installed.
When erasing a package from your system, RPM goes through the following sequence of
actions:
& Checks dependenciesÐRPM checks its database to see whether any other packages
depend on this database. If so, RPM does not delete the package unless explicitly told
to do so.
& Executes scriptsÐRPM executes a pre-uninstall script.
& Checks configuration filesÐRPM saves a copy of any modified configuration files.
& Deletes filesÐRPM deletes every file associated with the specified package.
& Executes scriptsÐRPM executes a post-uninstall script.
& Updates the databaseÐRPM removes all references to the package from its database.
As with the -i option, you can use the -v and -vv options to get verbose information from the
erase command. You can also use the -test option to see what problems might occur if you
were to really remove the package. Finally, you can use the -nodeps option to tell RPM to
ignore dependencies and go ahead and remove the package.
Tip #49 from
Jack You can also use the -nodeps option to force an installation.
Be careful using the -nodeps option. If you remove a package on which another
program depends, that program might not work correctly in the future.
Updating Packages with RPM Part
After you install a package, you eventually will need to install upgrades either for bug fixes or
I
for new features. RPM makes the typically horrendous task of upgrading a program effortless Ch
-U
with the (note the uppercase) option. Say someone has added several new features to the
7
quota program and released a new package called quota-2.01-1.i386.rpm. To upgrade to the
new version, you would use the following command:
rpm -Uvh quota-2.01-1.i386.rpm
172 Chapter 7 Upgrading and Installing Software
While upgrading, RPM installs the specified package and then erases all the older versions of
the packages (if any exist). RPM also spends a great deal of time processing any configuration
files associated with the package. Thus, while RPM is upgrading a package, you might see a
message like the following, indicating that a configuration file is being saved to a new file:
Saving syslog.conf to syslog.conf.rpmsave
This message indicates that RPM has created a new configuration file that may be compatible
with your system. After upgrading, you should compare the two configuration files and make
any necessary modifications to the new file.
Querying Packages with RPM
To see what packages are installed on your system, you can use the following command:
rpm -qa
This command lists every package currently installed on your system. To get information on
a specific package, just use the -q option. Table 7.3 provides the various options you can use
with the rpm -q command to query RPM packages.
Table 7.3 RPM Query Options
Option Description
-q name Provides the package name, version, and release number
-qa Lists all packages currently installed
-qf file Queries the package associated with file
-qp package Queries package
-qi package Provides the name, description, release, size, build date, installation date, and
other miscellaneous information about package
-ql package Lists the files associated with package
The various -q options do not work well when specifying symbolically linked files. For
the best results, use cd to change to the appropriate directory where the real file is
located before using the -q options.
If you find a new package, for example, and want to know more information about it, you can
use the following query command:
rpm -qip quota-1.66-6.i386.rpm
Using the Red Hat Package Manager 173
The command displays output similar to the following:
# rpm -qip quota-1.66-6.i386.rpm
Name : quota Relocations: (not relocateable)
Version : 1.66 Vendor: Red Hat Software
Release : 6 Build Date: Tue Apr 13 11:05:47 1999
Install date: (not installed) Build Host: porky.devel.redhat.com
Group : System Environment/Base Source RPM: quota-1.66-6.src.rpm
Size : 79332 License: BSD
Packager : Red Hat Software <http://developer.redhat.com/bugzilla>
Summary : System administration tools for monitoring users' disk usage.
Description :
The quota package contains system administration tools for
monitoring and limiting users' and or groups' disk usage, per
filesystem.
Verifying Packages with RPM
The final RPM mode verifies a package. You might need to check the consistency of a file on
your system at some point. Say you suspect that a file has been corrupted accidentally by an
errant program or a user. You need to compare the current files against the originals you
installed. RPM allows you to do so with the -V option (note the uppercase). Verifying a
package compares the size, MD5 checksum, file permissions, file type, and file owner and
group settings. To verify that a particular package's files have not been modified since they
were installed, you can use rpm -V packagename. For example, to verify the quota package, you
enter the following:
rpm -V quota
If nothing has changed, RPM does not display any output. If something has changed, RPM
displays a string of eight characters indicating what has changed and the name of the file that
has changed. You then need to inspect the various files in the package and determine whether
you need to reinstall the damaged package. Table 7.4 lists the possible output codes.
Table 7.4 Verification Failure Codes
Code Meaning
c The file is a configuration file.
5 The file failed the MD5 checksum test.
S The file size has changed since installation.
L A problem exists with the symbolic links.
Part
T The file modification time does not match the original.
I
D This code indicates a device attribute.
Ch
U The user setting is different.
7
G The group setting is different.
M The mode differs, either in permission or file type.
174 Chapter 7 Upgrading and Installing Software
Tip #50 from
Jack The following command verifies all installed packages on your system:
rpm -Va
Using the Debian Package Management System
A Debian package, like an RPM package, contains all the files necessary to install a program
.deb extension, and you install the package
or suite of programs. A Debian package ends in a
using thedpkg tool. The naming convention used for Debian packages follows this format:
packagename-version-debian-revision-number.deb, where version is the program version
number and debian-revision-number refers to the version of Debian (currently 2.1) supported
by the package. dpkg is analogous to RPM, in that you can install, remove, and upgrade
programs. The command dpkg ±helpprovides an extensive listing of instructions as does the
man page.
To install a package, you use the --install flag and specify the name of the Debian archive
package file ( .deb extension). For example, to install the quota program you use the following
command:
dpkg ±install quota-1.6-2.1.deb
dpkg checks dependencies to make sure all required components are installed and installs
them if need be. Then like RPM, it runs a pre install and then post install script.
To completely remove a package, use the --purge flag command as follows:
dpkg --purge quota
Tip #51 from
Jack You do not need to use the entire package name as long as dpkg can distinguish
between similar named packages.
If you need to keep the program's configuration files for future reference or use, use the
--remove flag to remove only the executable components of the package.
If you need to find which package deals with a given program, use the --search option, for
example:
dpkg --search quota
Debian maintains its package database in /var/lib/dpkg/info/.
Installing Non-Linux Software 175
Installing Non-Linux Software
Unfortunately, most of the software programs you'll find aren't in RPM or Debian package
formats. Typically, these pieces are downloaded via anonymous FTP from some archive site.
The process of installing software can range from extremely simple to almost impossible. The
level of difficulty depends on how well the software authors wrote their installation scripts
and how good their installation documentation is.
Deciphering Software Package Formats
The software packages that you get via anonymous FTP are virtually all in the form of
compressed tar files. These files can be created in a couple of different ways. Typically, a
directory tree contains source files, libraries, documentation, executables, and other necessary
files that are bundled into a tar file by using the tar program. This tar file is then usually
compressed to save space.
The software package probably has an extension at the end of the filename that tells you what
format it's in. If the file ends in .gz, it was compressed with the GNU gzip program. It is the
most common file-compression format for Linux software packages. If the archive name ends
with a .Z, it was compressed with the compress program. For example, the software package
foo.tar.gz is a tar archive that has been compressed with gzip.
Sometimes, a tar file that has been compressed with gzip is written with the .tgz
extension instead of .tar.gz.
? See ``Using tar,'' p. 265
Installing the Software
After you figure out the package format, you need to figure out where you want to place the
source files so that you can build the software package. Some software packages are fairly
large, so placing them on a file system that has a good bit of free space is a good idea. Some
people create a separate file system for sources and mount it under a directory, such as
/usr/local/src or /src. Wherever you decide to build your software packages, make sure
that you have enough disk space so that the software can be compiled successfully.
Part
Now you can move the software package to the source tree that you've set up and then
I
decompress it and expand the archive. If a file is compressed with gzip, you can decompress it
by using the gzip -d command. For example, the command Ch
gzip -d foo.tar.gz 7
expands the compressed file foo.tar.gz and replaces it with the tar archive named foo.tar.
See Table 7.5 for gzip command-line flags.
176 Chapter 7 Upgrading and Installing Software
Table 7.5 Flags for the gzip Command
Flag Flag Name Description
-a ascii ASCII text; converts end-of-line
characters by using local conventions
-c stdout Writes on standard output; keeps
original files unchanged
-d decompress Decompresses
-f force Forces overwrite of output file and
compresses links
-h help Gives a help listing
-l list Lists compressed file contents
-L license Displays software license
-n no-name Doesn't save or restore the original
name and time stamp
-N name Saves or restores the original name and
time stamp
-q quiet Suppresses all warnings
-S suffix suffix .suf Uses suffix .suf on compressed files
-t test Tests compressed file integrity
-v verbose Changes to verbose mode
-V version Displays the version number
-1 fast Compresses faster
-9 best Compresses better (that is, the file is
smaller)
file Specifies file(s) to (de)compress; if none
given, uses standard input
For files that have been compressed with the compress command, you can use the uncompress
command to expand them. For example, the command
uncompress foo.tar.Z
expands the compressed file foo.tar.Z and replaces it with the tar archive named foo.tar.
After you expand the compressed file, you need to expand the tar file into a directory tree.
You should put the source for each separate package in its own directory in your source tree.
Before un- tar-ing the file, you should look at its tar listing to see whether it was created with
a directory as the first entry. You can use the following command to see whether the first
entry in the tar file is a directory:
tar -tvf tarfile-name | more
Installing Non-Linux Software 177
If so, the tar file creates the directory when it's expanded. If no directory entry is listed at the
top level of the tar file, all the files at the top level are extracted into the current directory. In
this case, you need to make a directory and move< the tar file into it before you expand it.
You should always check for a top-level directory before expanding a tar file. You
might have quite a mess if the tar file expands and places a few hundred files in the
current directory instead of in a subdirectory.
When you have the tar file where you want to expand it, you can use this command to expand
the source tree in the tar file:
tar -xvf tarfile-name
The next step depends on how the software package that you're installing was written.
Typically, you change directory to the top-level directory of the software sources and look for
a file named something like README.1ST. The top-level source directory should contain a few
documentation files that explain the installation process.
On most versions of Linux, you can decompress a tar file on-the-fly as you extract it.
Simply add the z flag to the tar command, as intar -zxvf foo.tar.gz.
The typical installation process involves editing the file named Makefile to edit the
destination directories where the software places its compiled binaries. You then usually run
make followed by make install.
The make process probably varies with each package that you install. For some packages,
some sort of configuration shell script may ask you questions and then compile the software
for you. Make sure that you read the documentation files that come with the package.
Reviewing File Permissions
Permissions for a software package are usually set automatically during installation. The
installation script that comes with your application usually installs each file with the proper
ownership and permissions. Only when something goes wrong and a user who should be able
to access the program can't do so are you required to find the directory the application was
Part
copied to and check the permissions. I
Typically, the executable file that you run to start the application is installed with permissions Ch
that let any user run the file; however, only the superuser can delete or overwrite it. The 7
application usually is installed in a directory with read and execute permissions, but no write
permissions.
? See ``File Permissions,'' p. 414
178 Chapter 7 Upgrading and Installing Software
Solving Problems
A well-written and well-supported application installs onto your system with minimal
requests for information from you. It sets permissions properly so that all you have to do is
test the program and inform your usersÐoften through emailÐthat the application is now
available. But things can and do go wrong in the installation of programs and their
subsequent operation (or nonoperation). If, for whatever reason, the program doesn't
complete the loading process or fails to operate correctly after installation, it's your
responsibility to determine why and to fix the problem.
If a program doesn't install completely, your troubleshooting efforts often require no more
than reading the documentation and README files supplied with the application and looking
for a list of exceptions or problems and their solutions. However, no one expects you to
possess expertise and familiarity with the scores of software packages available for Linux.
Occasionally, you might require outside help.
If you can't solve the problem by using the information that came with the package, you
should try looking on Usenet news to see whether you can find any discussion of the package
in question. A question posted in the appropriate Linux group on Usenet can solve a lot of
problems. If you can't find help on the Net, you can try to contact the application developer,
usually via email. Remember, Linux is free, and so are most of the software packages available
for Linux. Don't expect shrink-wrapped manuals and 24-hour technical support lines. But if
you weren't the adventurous type, you wouldn't be using Linux, right?
Removing Applications
If an application is superseded by a better package or is no longer used by any user on the
system, removing it is a good idea. Disk space is always precious; you certainly don't want old,
unused programs to hog space required by new applications.
Like installation, removal of a program on a Linux system is more complicated than for
single-user operating systems. Sometimes just erasing the application's files and removing its
directory aren't enough. Drivers and other software connections must be disconnected to
avoid future problems. By taking notes and capturing the installation messages to a log file,
you can usually figure out what was changed when the software was installed. You can then
deduce what files to remove and which files to change to successfully delete a package.
Case Study: Upgrading Your Kernel
Along with upgrades for other software, new versions of the kernel sources are released
regularly. These versions may fix bugs or add new functionality. Alternatively, you might
decide to upgrade your kernel because you need to reconfigure it or add new device drivers.
In any case, the process is fairly straightforward. You should make sure that you have a
backup of your system software and a Linux boot floppy before you start so that you can
recover if you should damage your system. For a complete description of how to rebuild the
Linux kernel, see Chapter 14, ``Configuring the Linux Kernel.''
Case Study: Upgrading Your Kernel 179
The process for upgrading your kernel is detailed in the ``Kernel How-To'' document, which
is regularly posted to the Linux newsgroups on the Internet and is available on the various
Linux FTP sites, including metlab.unc.edu. Be sure to get a copy of this How-To and read it
thoroughly before you start your kernel upgrade.
The first step in the basic process of upgrading your kernel is getting the new kernel sources,
which are available via anonymous FTP from the various Linux archive sites. When you have
the sources, you need to preserve your current kernel sources. To do so, move your
/usr/src/linux directory to another name, such as /usr/src/linux.old. Unpack the kernel
sources in the/usr/src directory, which creates the linux subdirectory in the process. At that
point, change to the linux directory and look at the documentation and the README files.
Things may change as new kernels are released, so be sure to read the documentation.
From here, the process may vary a bit. Typically, you enter make config, which runs a
configuration script and asks you questions about your system. If the configuration phase
completes successfully, you then enter something similar to make dep. This command checks
for all the file dependencies to make sure that the new kernel finds all the files it needs to
compile. The basic building commands are as follows:
cd /usr/src/linux
make mrporper
make menuconfig
make dep
make zImage
make modules
make modules install
After the dependency check is complete, you typically enter make mrproper to delete any old
object files that are left lying around in the kernel source directory. If everything goes okay up
to this point, you can enter make bzImage or zImange to compile the new kernel. After it
compiles, you can install it with the LILO boot manager, and off you go.
Again, be sure to read the ``Kernel How-To'' before trying to perform this operation. The
How-To goes into detail on setting up your kernel and will probably save you hours of
frustration. Also, it might keep you from trashing your current Linux system in the process.
Part
I
Ch
7
PART
II
System Administration
8 Understanding System Administration 183
9 Using the vi Editor 203
10 Booting and Shutting Down 233
11 Managing User Accounts 249
12 Backing Up Data 259
13 Improving System Security 275
14 Configuring the Linux Kernal 295
15 Linux on PowerPC Platforms 307
CHAPTER
8
Understanding System
Administration
In this chapter by Jack Tackett, Jr.
Understanding the Importance of Proper Administration 184
Understanding Multiuser Concepts 185
Understanding Centralized-Processing Systems 186
Understanding Distributed-Processing Systems 188
Understanding the Client/Server Model 192
Performing Administration in a Networked Environment 192
Defining the Role of the Network Administrator 192
184 Chapter 8 Understanding System Administration
Understanding the Importance of Proper
Administration
A Linux system should have at least one person designated as the system administrator to
manage the system and oversee its performance. The system administrator is responsible for
seeing that the system is functioning properly. He or she knows who to call if things can't be
fixed locally and knows how to provide software and hardware facilities to current and new
users.
A Linux system requires initial configuration and continuous attention to ensure that the
system remains effective, trustworthy, and efficient for all users. The system administrator is
the person responsible for attending to the Linux system's needs. As such, this person is
responsible for many different tasks.
All UNIX systems are different in one way or another, and each is unique in the way it must
be administered. Linux is no exception. Your administrative duties vary, based on such
variables as the number of users you manage, the kinds of peripherals (printers, tape drives,
and so on) attached to the computer, networking connections, and the level of security you
require.
A system administrator, alone or with a support staff, must provide a secure, efficient, and
reliable environment for system users. The administrator has the power and responsibility to
establish and maintain a system that provides effective and dependable service. In a multiuser
environment, a number of competing purposes and priorities exist. The administrator
exercises the power and responsibility necessary to provide a well-functioning system.
Delegation of administrative responsibilities varies from system to system. On large systems,
administrative tasks can be divided among several people. Conversely, some small systems
don't even require a full-time administrator; such systems simply designate a certain user to
act as system administrator. If you work in a networked environment, your system may be
administered over the network by a network administrator.
Each Linux system has a single user who can perform virtually any operation on the
computer. This user is called the superuser and has a special login name called root. The home
directory for the root user, when logged in to the system, is typically / (the root directory of
the file system) or a specific home directory, such as /home/root.
The system administrator logs in as the superuser to perform tasks that require special access
privileges. For normal system work, the system administrator logs in as an ordinary user. The
superuser's login nameÐrootÐis used only for limited special purposes. The number of
users who can log in as root should be kept to a minimum (two or three at most). When any
person logs in to the system as root, that person is a superuser and has absolute power on the
system. With this privilege, the superuser can change the attributes of any file, stop the
system, start the system, back up the system's data, and perform many other tasks.
The administrator must be aware of many of the technical aspects of the computer system.
Also, the administrator must be aware of the users' needs, as well as the system's primary
185
Understanding Multiuser Concepts
purpose. Any computer system is a finite resource, and therefore, policies regarding its use
Part
must be established and enforced. Thus, the administrator must play a policy-enforcing role
as well as a technical role. That policy-enforcing role, combined with the power to perform II
virtually any possible action, requires a responsible, skillful, and diplomatic person in the role
Ch
of administrator.
8
The precise job description of the system administrator often depends on the local
organization. As system administrator, you might find yourself involved in a wide variety of
activities, from setting policy to installing software to moving furniture. However, all system
administrators have to perform or manage a number of tasks, such as the following:
& Manage usersÐAdd users, delete users, and modify users' capabilities and privileges.
& Configure devicesÐMake available and share such devices as printers, terminals,
modems, and tape drives.
& Make backupsÐSchedule, make, and store backups for possible restoration in case
the system's files are lost or damaged.
& Shut down the systemÐShut down the system in an orderly manner to avoid
inconsistencies in the file system.
& Train usersÐProvide or obtain training for users so that they can use the system
effectively and efficiently.
& Secure the systemÐKeep users from interfering with one another through accidental
or deliberate actions.
& Log system changesÐKeep a log book to record any significant activity concerning
the system.
& Advise usersÐAct as the ``local expert'' to aid the system's ordinary users.
Understanding Multiuser Concepts
A multiuser system employs two main concepts: multitasking and multiuser services. Linux
has the apparent capability to execute multiple tasks concurrentlyÐthough transparently to
the user. For example, you can read your email while compiling a program.
Each taskÐwhether it's a simple command entered on the command line or a complex
applicationÐstarts one or more processes. Everything running on a Linux system is
associated with a process. And because Linux can run many processes simultaneously, Linux
is a multitasking operating system.
You can connect to a computer running UNIX (referred to as a server) in many ways. You can
use a terminal or a computer; you can be located physically near the server and connected
with a cable, or you can be on the other side of the planet connected with high-speed data
lines or ordinary phone lines. Whether you're using a terminal or computer and how you're
connected to the server determine whether the computer's resources are considered to be
distributed or centralized.
186 Chapter 8 Understanding System Administration
A single-user computer operating system, such as DOS, is designed to be used by one person
at a time. All the processing is done on one computer that has sole access to resources, such as
printers, storage area, and processing.
Multiuser systems use the centralized-processing and distributed-processing models
described here to accommodate many users simultaneously:
& In a centralized-processing environment, many users (large systems can have hundreds of
users) access the resources of one computer; storage, printer, memory, and processing
tasks are all performed by that one computer.
& In the distributed-processing environment, processing can occur on the user's own
workstation, and the central processor is used to distribute applications and data.
Printers and storage can be connected to the user's workstation or to the main server.
Understanding Centralized-Processing Systems
As technology during the 1950s and 1960s advanced, operating systems began to allow
multiple users to share resources from separate terminals. By using a batch-processing
sequence, two users could execute two sets of instructions while sharing a processor, storage,
and output.
With the advent of a switched telephone network, computers began to use telephone
resources to extend computer resources geographically. In this model, each processor used
communications-processing resources to connect with remote terminals. This process
created a need for computers and terminals to communicate in a better way. The result was
the development of front-end processing for communications tasks and the centralized-
processing model.
Until personal computers became inexpensive, powerful, and ubiquitous, most UNIX
systems used the centralized-processing model. With centralized processing, mainframe
computers handled all the processing. Users connected to the mainframe and shared its
resources. This model is used less and less today, although it's still appropriate for computing
sites where users are separated geographically.
For example, your bank may have one main processing center, yet all the bank's branches can
access the data center regardless of their locations. On each user's desk is a terminal,
including a keyboard, a monitor, and a direct connection to the mainframe so that the
terminal can access the centralized resources: processing, printing, and storage (see Figure
8.1). The centralized-processing model is usually made up of many elements, such as the
server, front-end processors, terminals, modems, and multiport adapters.
When a user requests data, the request is processed by the computer in the bank's main office.
Results of the processing are then sent back to the terminal in the branch office. All data is
processed and stored by the mainframe computer.
187
Understanding Centralized-Processing Systems
Anything
Figure 8.1 typed at the
Part
Big Iron refers terminal is
to the processed by II
centralized-
the mainframe in
the main office Ch
processing
model of a
Branch A Branch B
8
computing en-
vironment. Terminals
Printer
Data
Storage
Data link
(dedicated phone lines)
Main frame
Computer
Front end
Branch C Processor Branch D
fcp
Elements of the Centralized-Processing Model
To make the centralized-processing model work, you need many elements, including the
server, front-end processors, terminals, modems, and multiport adapters.
A server can be defined as any computer set up to share its resources (processing power,
storage, printers, and so on). For example, you can use an IBM-compatible PC as a server as
long as it has enough hard disk space and RAM.
A front-end processor connects the communication channels and the server. It handles the
details of communication so that the server is free to process its data.
Two popular types of terminals are used today: dumb terminals and smart terminals.
Traditionally, UNIX is used with dumb terminals, which have keyboards and monitors but
nothing else. The most important point to realize about dumb terminals is that they have no
local processing power. The communications port on the terminal is connectedÐdirectly or
through a modemÐto the server. When you type at a dumb terminal, each keystroke is
transmitted to the server, where it's processed. Smart terminals can complete minimal
processing at the local site. Cash registers and other point-of-sale devices are examples of
smart terminals, as are the familiar automated teller machines (ATMs). The local device
stores the transaction request and then transmits the entire request instead of transmitting
each keystroke as a dumb terminal does.
To connect your terminal to a telephone line, you use a modem. Modems translate the digital
signals of terminals and computers into analog signals required by telephone lines. Modems
188 Chapter 8 Understanding System Administration
are always used in pairs. The first one connects your terminal to the telephone line; the second
connects the server to the telephone line. To make the connection, you dial out on the terminal.
When the modem on the other end (the one connected to the server) answers, your terminal can
communicate with the server.
To expand the number of available ports that users can connect to, you can install a multiport
adapter. Typically, a PC has only two serial ports: COM1 and COM2. If you want to use a PC
as a server for more than two users, however, you need more ports. The multiport adapter, in
this case, consists of a card that you install inside the computer, a small box with eight or
more connectors (locations to plug in eight or more serial cables), and a cable that connects
the box and the card. Software is supplied with the adapter to permit the added connectors to
function as additional serial ports.
Understanding Distributed-Processing Systems
In distributed processing, the terminal is replaced by a workstation, which is itself a computer
usually running DOS or UNIX. Programs can be located and run from the server or from
your workstation. Similarly, files can be located on either system. If you process a file on your
workstation, you store it on the server so that others can access it. You can print on local
printers connected to your workstation or on printers connected to the server.
Because workstations are in common use, your bank probably uses a distributed-processing
system instead of the centralized system described in the preceding section. Figure 8.2 shows
the same bank with a distributed-processing system.
Figure 8.2
Processing, printing, and
storage take place in the
Web-centric or
branch.
client-server ty- Files are also updated in
pifies the main office
distributed-
Branch A Branch B
processing
Local Personal
model.
Network Computers
Wiring (Workstations)
Main Office
Printer
Printer
Data
Storage Data link
(dedicated phone lines)
Main frame
Computer
Branch C Branch D
189
Understanding Distributed-Processing Systems
Elements of the Distributed-Processing Model
Part
Distributed processing uses file servers, workstations, network interface cards, hubs,
II
repeaters, bridges, routers, and gateways. The purpose of the file server is to distribute files
Ch
and segments of programs to workstations, print from a central location, and control flow on
the connection between workstations. More than 90 percent of processing occurs at the 8
workstation level, leaving 5 to 10 percent of the load at the file server for administrative tasks.
In addition to using a personal computer as a file server, you can use it as a Linux workstation.
A workstation is a computer used by one person for their own activites. A server is a
computer used to provide services for many people. Linux can function as either. As a
workstation Linux allows a single person to perform daily tasks such as reading email
and writing reports. As a server, Linux can function as a Web server providing Web
pages to many different users, or as a file server, allowing many people to store files
on the Linux box rather than their workstation.
Linux was designed to run in a minimal hardware configuration. In fact, you can run Linux
with a 386SX microprocessor and 4MB of RAM! Because most current systems are more
powerful than Linux's minimum requirements, you should have no problem with computing
power. The amount of disk space required depends on the software you install and the
amount of data you expect that software will generate. Linux requires less disk space than
most implementations of UNIX systems. You can run a completely functional Linux system,
without X Windows support, in 600MB. For a complete installation of everything in the
distribution, 1.6+ GBs (gigabyte) is recommended.
Generally, resources should be applied to the workstation level, where most of the processing
occurs. The amount of additional resources depends on the types of tasks you plan to do. For
example, word processors take minimal resources (hard drive, RAM, quality of monitor)
compared with graphics-intensive tasks such as those you might perform in multimedia and
computer-aided design (CAD) programs. For applications involving CAD, you need very
large hard disks (4 gigabytes or more), a lot of RAM (64MB, or even 128+MB), and high-
resolution monitors and video cards (1,280 61,024 or higher). You might even want a tape
drive for backing up your system and a CD-ROM drive for loading large applications.
A network interface card (NIC) attaches to a slot on the motherboard and is the physical link
between the computer and the cabling for the network. Network interface cards are generally
available for coaxial or twisted-pair cabling.
The hub serves as a connecting point for network cables, such as 10BaseT Ethernet, and can
be passive or active. A passive hub usually has four connectors. An active hub usually has at
least eight ports and amplifies or relays the signal.
Repeaters amplify or regenerate the signal over the network so that you can extend the normal
distance limitations of network cabling.
190 Chapter 8 Understanding System Administration
Use a bridge when you need to connect two different network types. For example, you use a
bridge to connect ISDN to Ethernet, Ethernet to Token Ring, ATM to FDDI, etc.
Routers are used in large, complex networks that provide many paths for network signals to
travel to the same destination. The router determines which is the most effective route and
sends the signal along that route.
Use a gateway when you need to connect dissimilar network types, which use different
protocols. The gateway performs the necessary protocol conversions so that the two networks
can communicate. For example, an SNA network connected to a TCP/IP network would
require a gateway.
Topologies
The term topology refers to how workstations and file servers are connected in a network. The
names of various topologies are derived from the pattern the cables make after you connect
the various terminals, workstations, and file servers. The most common topologies are star,
bus, and ring. When more than one topology is used in a network, it's referred to as a hybrid
network.
Star Topology
With the star topology, all workstations are connected to a central file server or hub (see
Figure 8.3). You can have passive or active hubs in this scheme.
Figure 8.3
Information
radiates out
from a central
server in a star Workstation Workstation
topology.
Server
Workstation Workstation
Workstation
A passive hub is simply a connecting point for the workstations. An active hub, called a switch,
modifies the signals from workstations and offers amplification of the signal, among other
functions. AT&T's StarLan is an example of a network using star topology.
191
Understanding Distributed-Processing Systems
Bus Topology
Part
In a bus topology, all workstations and file servers share a common pathway (see Figure 8.4).
II
They are, in fact, connected directly. The bus topology is the foundation for Ethernet and
token bus. Ch
8
Figure 8.4
A bus topology
Server
distributes in-
formation to
various work- Workstation
stations along
a common
pathway.
Workstation Workstation Workstation
Ring Topology
A ring topology looks like a wagon wheel without the hub (see Figure 8.5). The server is
connected to the workstations in bus fashion, except that the last items along the network are
connected to make a closed loop. Ring topologies use a repeater, which IBM refers to as a
Multistation Access Unit (MAU). The IBM Token-Ring Network is an example of a ring
topology.
Figure 8.5
In a ring topol-
ogy, data flows
around a net-
work from the Workstation Workstation
server to the
workstations in
bus fashion,
usually nego-
Server
tiating access
with some
Workstation
form of token.
Workstation Workstation
Hybrid Topology
In the 1970s and 1980s, firms with decentralized purchasing departments experienced the
growth of different topologies on their networks. For example, the accounting department in
192 Chapter 8 Understanding System Administration
a given company used a bus network; the purchasing department installed Token-Ring machines;
manufacturing used an Ethernet bus; and administration relied on mainframe technology. This
combination of networks planted the seeds for enterprise computing and hybrid wide area
networks. The integration of these networks linked dissimilar topologies, such as rings, stars, and
buses.
Understanding the Client/Server Model
The result of the development of distributed processing is the client/server model. Today,
Linux can be used in this model as the client, the server, or both.
To understand a client/server setup, assume that several Linux workstations (the clients) are
connected in a bus topology to a server (a high-end PC with lots of disk space, also running
Linux). The server has directories for each client where important files can be stored and
backed up with the server's nightly backup. The server also has directories from which clients
can share files. Connected to the server is a fast laser printer that everyone can access and a
tape drive suitable for backing up the large hard disks. Also, several of the clients have their
own slower, less-expensive laser printers connected locally.
The server in this example is a PC running LinuxÐjust like the clients' PCs, although
the server is more powerful. There's no reason why the server can't act like a client at
times and share resources from the clients. In other words, any Linux system can
function as both a client and a server.
Performing Administration in a Networked
Environment
A UNIX network usually takes the form of many computers, large and small, tied together
over directly connected wires or common telephone lines. Administering the network is
usually the task of a person or persons located at one of the sites in the network.
Most people can learn Linux and administer a network. In a production environment, finding
a qualified person right away would be nice; however, such people are somewhat rareÐand
usually well paid. In fact, the growth of the Linux market has exacerbated this fact. With
practice and patience, even people with limited backgrounds in computers can learn to
administer a corporate UNIX/Linux computer.
Defining the Role of the Network Administrator
Any time you have more than a few UNIX/Linux systems connected in a network, you should
probably have a dedicated network administrator. Some expertise is needed to decide how
systems are connected (local area networks or modems), the level of security needed, and how
193
Defining the Role of the Network Administrator
shared peripherals (printers, tape backups, and so on) are distributed. On a day-to-day basis, the
Part
administrator maintains lists of system names, network addresses, and user access and generally
makes sure that the network is running properly.
II
Corporations with networks of hundreds of computers can afford to have several Ch
administrators with extensive training in selected topics. Having these administrators can be a
8
necessity if you have complex printing needs, for example. Administering printers and
printing can require extensive knowledge of specific printers and how to interface that
equipment to Linux.
Understanding Hardware and Software Issues
If, as system administrator, you're required to choose the networking software and hardware
for the computers under your control, you should consider several factors. As with most
things in life, you balance what you need with what you can afford.
If your systems are close together in the same building, a local area network (LAN) is a low-
cost, high-speed means of networking your computers. You can put an Ethernet board in
each Linux system and use TCP/IP as the networking protocol software. TCP/IP is a
standard component of Linux distributions.
To connect over greater distances, you can use modems for lower-speed transmissions such as
Point-to-Point Protocol (PPP) or Serial Line Internet Protocol (SLIP) to provide
asynchronous TCP/IP connections. You also can use Unix to Unix CoPy (UUCP) software
for email, news, and file transfers (although UUCP has limitations and is all but supplanted
by TCP/IP in today's Internet). For higher speeds over long distances, you can use ISDN,
cable modems, xDSL or get leased lines from the telephone company.
Don't buy just any old networking hardware. Although many off-the-shelf networking
hardware products come with the drivers needed to make them work with DOS, the same
isn't true with Linux. As a result, Linux systems have many standard networking drivers built
in. Table 8.1 shows some of the Ethernet cards now supported by Linux. Check the
``Ethernet How-To'' for updates to this list.
Tip #52 from
Jack You can find the HOW-TOs on your system in the /usr/doc/HOWTO directory. They
are also available on the RedHat CD-ROM in the /RedHat/doc/HOWTO directory. If
you mount the CD-ROM with the following command
mount /mnt/cdrom
then you can cd to the directory and read the various How-Tos with your favorite
editor.
On the Web you can find the How-Tos at http://www.linux.org/help/
howto.html.
? See ``The Linux How-To Index,'' p. 843
194 Chapter 8 Understanding System Administration
Table 8.1 Some Ethernet Cards Now Supported Under Linux
Manufacturer Cards
3Com 3c503, 3c503/16, 3c509, 3c579, 3c590/3c595
SMC (Western Digital) WD8003, WD8013, SMC Elite, SMC Elite Plus, SMC Elite 16
ULTRA
Novell Ethernet NE1000, NE2000, NE1500, NE2100
D-Link DE-600, DE-650, DE-100, DE-200, DE-220-T
Hewlett-Packard 27245A, 27247B, 27252A, 27247A, J2405A
Digital DE200, DE210, DE202, DE100, DEPCA (rev. E), 21040
AMD Pcnet-32 cards
Allied Telesis AT1500, AT1700
Intel Ethernet express
PureData PDUC8028, PDI8023
Applications that aren't integrated with networking products can be used in a network
environment. For example, you can install an application on a Linux system and have many
users from other computers use the application by running the remote execution commands
built into UNIX. Alternatively, you can share an application by remotely mounting the file
system that contains the application and then running it from the local system.
Performing Common Networking Administrative Tasks
Administration of a network takes on several dimensions. Most networks don't just occur;
typically, they evolve. In the ideal situation, the administrators are involved with the purchase
of the computers and software so that they know what's expected of them as administrators
and what the users are getting.
Setting Up the System
Network software should be installed and ready to connect onsite. If you're using Ethernet
for your Linux network segment, it's a good idea to have the continuity tests completed. If
you're using telephone lines, have them tested as well. Wiring and terminals for users also
should be tested and ready. Installation should be plug-and-play, but it never is. Connection
?
problems always occur.
See ``Understanding the TCP/IP Configuration Files,'' p. 620
The advantage of buying a computer for a situation in which the operating system isn't yet
installed is that you can set up file systems to accommodate your specific needs. You must
know what software is going on the computer, the number of users who will be using the
system, and the intensity of their usage.
195
Defining the Role of the Network Administrator
Tip #53 from
Part
Jack You've invested time and money in setting up the network to this point. You should
immediately back up the configuration files you've set up. Most of these files are
II
located in the /etc subdirectory. The easiest way to back up these files is to tar the
/etc directory to another file system that resides on another disk drive, such as
Ch
/home. For example: 8
tar -cvf /home/etc-backup.tar /etc
When the system is fully functional, the application software should be installed. Software on
a Linux computer is often more complex than on a single-user system, so be prepared to
spend some time installing, tuning, and making the software fully operational. This task can
take from a couple of hours to several daysÐor longer.
You're now ready to start adding users to the system, although you're still not onsite. Add
login IDs for a few key users, and put in a common startup password, such as temp01. This
step provides some initial security and gives you a chance to get key people onto the system
and operational right away when you install the system.
After installation, the computer should be attached to the network. Make sure that you can
communicate from any point in the network to any other point. Test communications by
moving large and small files from one computer to another. Electronic mail should be
directed to and from other nodes in the network. All computers must ``know'' this new
computer in the network. This means that you need to add it to your host name database
that's used by any other computers on your network. If you use the Domain Name Service
(DNS) locally, you must add the host name to the DNS name database. If you aren't using
DNS, add the name to the /etc/hosts files on your other systems.
If you are using the Network Information Service (NIS), then make sure to configure your
NIS configuration files correctly. NIS is used to manage a number of computers from a
central NIS server. Originally called yellow pages, NIS today is being supplanted by light
?
weight directory services, known as ldap.
?
See ``Database Files and Resource Records,'' p. 800
/etc/hosts
?
See ``The File,'' p. 620
See ``What Is Using NIS,'' p. 476
Handling Peripherals
Printing can present a major issue to an administrator. Monitoring and maintaining printers
is a significant task and can take a lot of an administrator's time. Understanding the spooling
of print jobs, interface tools, and equipment peculiarities requires time and patience. You can
use tools such as printtool and magicfilterconfig to install and configure many different
?
types of printers.
See ``Printing,'' p. 391
196 Chapter 8 Understanding System Administration
Using modems is the cheapest way to link a network that spans long distances. Modems and
PPP or UUCP are tools that can make it practical for a small staff to administer many
computers. As with printers, however, modems have some problems that require time to get
?
them running right. Choose one or two brand names and really learn their idiosyncrasies.
See ``Using dip in Command Mode,'' p. 665
Monitoring the System
When the installation is complete, you can set up UNIX tools to monitor this new system. As
the administrator, you should start getting a feel for how the system is performing.
Monitoring running systems in a network is an ongoing process, but the administrative load
should stabilize after a while if you aren't constantly adding peripherals or software.
Occasionally, something fails, or tweaking might be necessary. A good administrator learns to
determine whether the problem is related to hardware or software.
A competent system administrator monitors at least the following items: disk space, memory,
network processes, and CPU load. Running out of disk space is an inconvenience to users and
potentially catastrophic to a running system. Use the df command to track disk space, as
shown in the following output:
/dev/hda1 1536971 299483 1158060 21% /
/dev/hdb2 2554589 345188 2077311 14% /home
/dev/sbpcd 499012 499012 0 100% /mnt/cdrom
If space exceeds a given threshold, such as 80 percent, then the system administrator needs to
do something to prevent problems. If a file system exceeds this value, the administrator
should move to correct the situation. A system administrator can delete unneeded files,
compress or move important files that are not needed everyday, or add a new hard drive to the
system.
Tip #54 from
Jack Most of the information provided by various system administration utilities, such as
top, make use of information kept in the /proc file system. For example, to see how
much memory your system consumes, you can cat the file /proc/meminfo.
The top command provides information on a variety of system resource topics, such as
processor load and memory usage. The following illustrates sample output from the top
command:
6:49pm up 21 days, 18:14, 1 user, load average: 0.00, 0.00, 0.00
56 processes: 55 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 3.1% user, 4.0% system, 0.0% nice, 93.0% idle
197
Defining the Role of the Network Administrator
Mem: 30796K av, 30076K used, 720K free, 39332K shrd, 1172K buff
Swap: 0K av, 0K used, 0K free 14016K cached Part
II
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND
15302 root 12 0 732 732 564 R 0 6.4 2.3 0:01 top Ch
15288 root 0 0 1224 1224 596 S 0 0.5 3.9 0:03 sshd
342 root 0 0 228 228 188 S 0 0.1 0.7 59:14 update 8
1 root 0 0 344 344 284 S 0 0.0 1.1 0:03 init
2 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 kflushd
3 root -12 -12 0 0 0 SW< 0 0.0 0.0 0:00 kswapd
4 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 md_thread
5 root 0 0 0 0 0 SW 0 0.0 0.0 0:00 md_thread
343 root 0 0 296 296 248 S 0 0.0 0.9 0:00 mingetty
6847 root 0 0 872 872 676 S 0 0.0 2.8 0:06 bash
5486 root 0 0 296 296 248 S 0 0.0 0.9 0:00 mingetty
37 root 0 0 332 332 280 S 0 0.0 1.0 0:00 kerneld
66 root 0 0 500 500 404 S 0 0.0 1.6 8:35 sshd
202 root 0 0 428 428 352 S 0 0.0 1.3 0:34 syslogd
211 root 0 0 536 536 324 S 0 0.0 1.7 0:01 klogd
222 daemon 0 0 400 400 324 S 0 0.0 1.2 0:00 atd
233 root 0 0 468 468 388 S 0 0.0 1.5 0:00 crond
The top command provides a wealth of information, such as how many users are logged in,
what load the system is currently under (0.00 in this example), how much memory is being
consumed and how much is swapped out to the swap file, and so on. The output also provides
the top several processes currently running and how much of the resources they are
consuming. If a single process is consuming too many resources, then the administrator may
have to kill that process.
The top command automatically refreshes, which can be a great help when you're
troubleshooting a resource problem, but refreshing is also a resource hog itself. Be
careful when using the top command on a heavily loaded system.
Table 8.2 provides a list of other useful utilities for the system administrator. For more
information, see their respective man page.
Table 8.2 Useful SysAdmin Utilities
Utility Example Description
ps ps aux Lists all the current running processes on
your system.
df df -m Provides a listing of disk space used and
available.
ifconfig ifconfig -a Displays information about the various net-
work connections and the ip addresses and
hardware mac addresses associated with each
interface.
198 Chapter 8 Understanding System Administration
Table 8.2 Useful SysAdmin Utilities
Utility Example Description
arp arp Provides a listing of the current arp table.
whois whois mcp.com Queries Internic registration authority for
information on who owns the requested
domain name.
nslookup nslookup www.mcp.com Provides the ip address of the requested
domain name. If an ip address is entered
instead, then the name associated with that ip
is provided.
vmstat vmstat 5 12 Provides information on memory usage, CPU
usage, interupts, and other items.
Coping with Software Upgrades
Some software packages are constantly being updated. Although upgrading is a concern with
commercial UNIX, it's a special issue under Linux because much of the software is publicly
available over the Internet and is continually modified. The good news might be that a bug is
fixed. The bad news might be that each system in the network has to be updated. You can
expect a new challenge with each update.
The best advice isn't to immediately put all new versions on your systems, but to test the
upgrade or patch on one non-critical system. When you're sure that the new version is okay,
you can upgrade the other systems. A good administrator learns how to install these patches
or new versions without going to the other sites in the network. This solution sounds
impossible at first, but you'll find that many UNIX tools facilitate the installation of patches
and upgrades. For the distributions included in this book, check out dpkg and apt-get for
?
Debian and autorpm and autoupdate for Red Hat and Caldera systems.
See ``Understanding the System Administrator's Job,'' p. 167
Training the Administrator
Training in most organizations is very hit or miss. Perhaps the person has some computer
background in some computer topic, but little is done to formally train that person to
administer the system. Fortunately, many organizations, such as Red Hat and Global
Knowledge provide indepth and detailed training for Linux Administrators. Administration
requires attention and a solid knowledge of the following topics:
& Linux/UNIX design and usageÐThe administrator has to have a thorough
understanding of such issues as redirection, pipes, background processing, and so on.
199
Defining the Role of the Network Administrator
& The vi editorÐThe vi editor is on virtually every credible UNIX computer, including
Part
Linux. Many people criticize it, and many people substitute other editors for their own
use, but it's advisable to have an administrator learn and become proficient in the use II
of vi because it's the ``common denominator'' among UNIX editors.
Ch
& Shell script programmingÐMany of the key programs used to administer UNIX are
8
written in shell script language and might require modification for your specific needs.
Many of the tools outlined in this chapter require knowledge of how to put together
and use a shell program. Almost every user has a favorite shell.
The Bourne Again shell, or bash, is a Bourne shell clone that's the default shell under
Linux. Also, the Z and T shells are available in the distribution. You should, however,
stay with the common Bourne shell until you master this shell language. Also, virtually
all the shell programs written by the Linux creators are written in the Bourne shell.
You should also investigate the Perl system administration language. It provides a very
robust set of tools for system administration in one programming environment.
& CommunicationsÐCommunications training has improved since previous editions
of this book and training is continuing to improve. To set up computer networking
effectively, you must have knowledge of TCP/IP and the related protocols. Similarly,
you should understand PPP if you're going to set up an asynchronous Internet
connection. Ideally, these protocols should be taught in a laboratory environment with
the many options available. You can attend classes or at least buy manuals on the
subject, but you should accept the fact that you'll be spending a great deal of time
experimenting.
& UNIX conventionsÐUNIX conventions aren't taught or even mentioned in many
UNIX classes, and you'll probably have to pick them up by observing as you go
through training. For example, you'll learn that binary executable programs are
generally stored in the bin directories, such as /usr/bin, /bin, and /usr/local/bin.
You can put your own executable programs in /usr/local/bin. Likewise, the lib
directories, such as /usr/lib, are used for library files, and you can put your own
libraries in a directory such as /usr/local/lib. Understanding and following standard
Linux/UNIX conventions such as these can save time when you need to find and fix
problems.
Several reputable companies, perhaps including the company that you bought your computer
from, offer training on all these topics. However, this training is probably not specific to
Linux. A few vendors sell various distributions of the Linux operating system and offer classes
on selected topics. You should also look for user groups in your area and check the
comp.os.linux newsgroup hierarchy on Usenet news on the Internet. Also check out
[email protected] for an active mailing list for Linux administrators.
Training is best done in small pieces. You should take a course and then come back and use
what you learned right away on your network. Linux has an elaborate set of tools that will
probably never be completely mastered, but you have to know where to find information in
manuals.
200 Chapter 8 Understanding System Administration
Troubleshooting the Network
Tracking down problems and fixing them is one of, if not the most important command tasks
a system administrator faces. Many times the problem is acute, but it causes a mystery and a
solution not readily apparent.
No access to a network can cause many problems, such as a system to hang at boot, hang on
traversing an nfs mounted directory, or being unable to surf the Web or telnet to other
machines. And no one likes a slow network connection, but how do you know where the
bottle neck isÐyour system or somewhere out on the Internet? The basic tools to use for
investigating are ping, traceroute, and netstat.
Ping sends an ICMP echo request out through the network to a targeted machine and is the
beginning point for any network investigation. Ping provides information on whether
another computer is accessible on the net and how long it takes to get a packet of data to the
computer and back. For example to see if the host www.netwharf.com is available use:
[root@ns /root]# ping www.netwharf.com
PING www.netwharf.com (209.42.203.228): 56 data bytes
64 bytes from 209.42.203.228: icmp_seq=0 ttl=255 time=1.8 ms
64 bytes from 209.42.203.228: icmp_seq=1 ttl=255 time=1.0 ms
64 bytes from 209.42.203.228: icmp_seq=2 ttl=255 time=1.0 ms
--- www.netwharf.com ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 1.0/1.2/1.8 ms
The command provides a wealth of information. First the program indicates the ip address
associated with the host.
Tip #55 from
Jack If ping cannot translate the host name into an IP, you may have dns problems. Check
the/etc/resolv.conf for proper settings for your name servers. If all appears fine,
nslookup to see if your name server can translate names:
you can use the command
nslookup www.netwharf.com. Finally, make sure you have spelled the host name
correctly.
Next, ping provides a sequence number to let you know if the connection is dropping data
packets. Finally, ping lets you know how long a packet took to traverse the network. Both
items provide a glimpse into the health of the network and potential problems. But just
knowing a network is sending and receiving packets may not be enough informationÐwhere
is the problem on a slow network?
Use traceroute to find out where the bottleneck occurs on the net. The program sends out
data packets and then reports how the packet snakes through the Internet. As a packet heads
out to the net, it is routed through various routers (a hop), the more routers, the more
potential for problems. To see how many hops a packet takes, use:
201
Troubleshooting the Network
[root@ns /root]# / usr/sbin/traceroutewww.mcp.com
traceroute to www.mcp.com (198.70.146.70), 30 hops max, 40 byte packets Part
1 gateway (209.42.203.225) 3.764 ms 3.150 ms 3.052 ms
2 max4000-1.rtp.intrex.net (209.42.192.1) 34.528 ms 32.074 ms 31.190 ms II
3 209.42.192.30 (209.42.192.30) 33.264 ms 33.012 ms 31.607 ms
4 rtp-rtr1-e1.intrex.net (209.42.255.17) 33.393 ms 44.959 ms 32.609 ms Ch
5 sl-gw4-atl-4-0-2xT1.sprintlink.net (144.228.86.53) 61.957 ms 47.754 ms 8
A51.439 ms
6 sl-bb10-atl-1-2.sprintlink.net (144.232.12.29) 80.908 ms 82.393 ms 48.498 ms
7 sl-bb10-fw-4-0.sprintlink.net (144.232.8.98) 66.180 ms 66.824 ms 69.790 ms
8 144.232.11.10 (144.232.11.10) 91.006 ms 64.449 ms 70.304 ms
9 144.232.9.253 (144.232.9.253) 76.865 ms 74.102 ms 78.176 ms
10 sl-bb11-chi-4-0.sprintlink.net (144.232.9.118) 92.362 ms 101.582 ms 93.251 ms
11 sl-bb11-chi-8-0.sprintlink.net (144.232.10.5) 85.014 ms 85.514 ms 85.111 ms
12 sl-gw14-chi-8-0-0.sprintlink.net (144.232.0.194) 87.983 ms 90.575 ms
A92.611 ms
13 sl-napnet-2-0-0-T3.sprintlink.net (144.228.159.18) 90.760 ms 75.226 ms
A77.879 ms
14 chi-f0.iquest.net (206.54.225.250) 92.193 ms 114.950 ms 81.379 ms
15 204.180.50.9 (204.180.50.9) 85.988 ms 108.888 ms 86.364 ms
16 iq-ind-core1.iquest.net (206.53.249.1) 99.583 ms 91.685 ms 111.068 ms
17 * iq-ss2.iquest.net (206.246.190.161) 96.229 ms 82.107 ms
traceroute again provides a name to IP translation and then for each hopÐthat is, for each
router the package passes through on the journeyÐdisplays an information line. Each line
consists of the hop number, the name and IP number of the router, and the times recorded
for reaching that particular router. An asterisk indicates the time for the packet to reach its
destination exceeded the amount of time allotted. If a site is not responding, traceroute keeps
repeating a line of three asterisks for each hop attempted. The final router shown provides an
idea of where the problem might be located, since the next hop is unavailable. Of course you
need to be able to get outside your current network to reach another destination.
The netstat program provides many options to report information on the TCP/IP stack
running on your system. Data packets are routed based on entries in your systems routing
tables. To see the current settings, use:
[root@ns /root]# netstat -rv
Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt Iface
ns2.netwharf.co * 255.255.255.255 UH 1500 0 0 eth0:0
www.nightskyweb * 255.255.255.255 UH 1500 0 0 eth0:1
209.42.203.224 * 255.255.255.224 U 1500 0 0 eth0
127.0.0.0 * 255.0.0.0 U 3584 0 0 lo
default gateway.netwhar 0.0.0.0 UG 1500 0 0 eth0
The important item to check for access to the Internet is the default gateway item listed in the
Destination column. This is the route through which your computer sends a packet if it does
not know where to send the data. You should ping the computer shown in the Gateway
column to make sure it is available. If you suspect you have the wrong gateway configure,
check with your service provider or your network administrator.
CHAPTER
9
Using the vi Editor
In this chapter by Steve Burnett
Why vi? 204
Using vi 207
vi Command Summary 226
Setting the vi Environment 228
204 Chapter 9 Using the vi Editor
Why vi?
In earlier chapters, you learned how helpful it can be to have sequences of commands or shell
scripts stored in a file. You probably have to create data, email, lists, memos, notes, reports,
and so on; you use some type of text editor to perform these tasks. You may have several
editors or word processors available on your Linux system to help you with these tasks. To
put commands or shell programs in a file, however, you need an editor that can save your
work in a text fileÐa file in ASCII format. Linux comes with a standard text editor called vi,
which you can use for all but the most complex writing and editing projects.
vi is very useful to system administrators because it is available on every UNIX platform.
Thus, after you learn how to use vi, you can use it on any system running UNIX. vi is also
useful because it takes up very few resources when executing, which means you can use vi
when other programs might not run because of hardware or other system problems.
Tip #56 from
Steve This chapter doesn't discuss ed or ex very much because you'll find vi easier to use
and available on every UNIX machine, including Linux.
The vi and ex editors that ship with the Red Hat distribution are actually other names for an
editor called vim (for vi improved). The names vi and ex are symbolically linked to vim, so
when you type vi, you're actually running vim. Read /usr/share/vim/vim_diff.txt for a
summary of the differences between vim and vi.
Your Linux system also has other text editors: a graphical editor for use under the XFree86
system and two standard nongraphical text editors called ed and ex. They're both line-
oriented editors; that is, using them, you work with only one line at a time. Another editor,
called Emacs, is also supplied with most Linux distributions. vi and Emacs are full-screen
editors; when you use them, you see a screen's worth of information, so you can make
changes or additions in context.
To understand vi (pronounced vee-eye, not vie), you need to understand some of vi's history
within the UNIX world. Moreover, although today's systems (including Linux) have much
more user-friendly and robust editors, you should learn how to use vi because every UNIX
(and, thus, Linux) system has a copy of vi available. Sometimes vi is the only editor available,
so you should know some of its basic operations.
UNIX was developed in an environment in which users' terminals were teletypes or some
other slow, hard copy terminals; video display monitors generally weren't used. A natural
editor for that environment was a line-oriented editorÐone that the users could see and work
on one line of text at a time. As mentioned previously, two line-oriented editors are on UNIX
systems today: ed and ex.
205
Why vi?
In its early days, UNIX was made available to universities essentially free of charge. Students
and faculty at several universities made many contributions to the UNIX working
environment. Several notable improvements came out of the University of California at
Berkeley, including a full-screen editorÐone that lets the users work with a screen of
information at once rather than a single line of text. As you've probably guessed by now, that
full-screen editor is called vi, which stands for visual. The time was right for the transition to
screen-oriented work. Users were working with video terminals rather than hard copy
Part
devices.
II
Ch
Tip #57 from
man vi 9
Steve You don't have to become an expert to use vi. Simply type at the command
prompt for help. You also can ask for help by pressing Esc and then entering :help.
What Is vi?
Because vi is part of the standard UNIX environment, millions of UNIX users have learned
and used it (to one degree or another). You will find that vi starts quickly and can be used for
simple and complex tasks. As you would expect, you can use it to enter, modify, or delete text;
search or replace text; and copy, cut, and paste blocks of text. You also can customize it to
match your needs. You can move the cursor to any position onscreen and move through the
file you're editing. You use the same methods with any text file, regardless of its contents.
The vi editor isn't a word processor or desktop publishing system. It has no menus and
virtually no help facilities. This chapter doesn't cover all of vi's features; that requires more
space than is available. (In fact, entire books are written just on vi.) Instead, you'll learn the
commands to do the most necessary editing tasks. If you want to know about the more
advanced features of vi, consult the man pages supplied with Linux.
The original version of vi doesn't have a help facility. However, newer versions of vi,
such as vim for the Red Hat distribution, provide some online help.
Word processing systems usually offer screen and hard copy formatting and printing, such as
representing text as bold, italic, or underlined, but vi doesn't. Other Linux commands can
perform some of these functions; for example, lp can print and nroff can format text. Some
text processing programs, such as TeX (pronounced tek) and LaTeX, can process embedded
commands into text, such as bold and underlined attributes.
206 Chapter 9 Using the vi Editor
The vi editor operates in two modes:
& In command mode, your keystrokes are interpreted as commands to vi. Some of the
commands allow you to save a file; exit vi; move the cursor to different positions in a
file; and modify, rearrange, delete, substitute, and search for text.
& In input or text-entry mode, your keystrokes are accepted as the text of the file you're
editing. When vi is in input or text-entry mode, the editor acts as a typewriter.
In an editing session, you can freely switch between modes. You do, however, have to
remember the mode you're using and know to change modes. Some people may find
switching modes uncomfortable at first. Later in this chapter, you'll learn about the showmode
option, which tells you vi's current mode. With a little practice, you'll find vi extremely
convenient for editing Linux ASCII files, especially configuration files and shell scripts.
Understanding the Editing Process
You edit text by creating new text or by modifying existing text. When you create new text,
you place the text in a file with an ordinary Linux filename. When you modify existing text,
you use the existing filename to call a copy of the file into the editing session. In either case,
as you use the editor, the text is held in the system's memory in a storage area called a buffer.
Using a buffer prevents you from directly changing the contents of a file until you decide to
save the buffer. This is to your benefit if you decide you want to forget the changes you've
made and start over.
As you change and add to the text, these edits affect the text in the buffer, not in the file stored
on disk. When you're satisfied with your edits, you can issue a command to save the text. This
command writes the changes to the file on the disk. Only then are the changes made
permanent. You can save changes to disk as often as you like. (It's usually a good idea to save
any file you're editing frequently in case of lockups or power loss.) You don't even have to exit
the editor when you save changes. This chapter shows you several ways to exit the editor;
some of those ways write the buffer to the text file on the disk.
The vi editor is said to be interactive because it interacts with you during the editing session.
The editor communicates with you by displaying status messages, error messages, or
sometimes nothing onscreen (in typical Linux fashion). The last line onscreen, called the
status line, holds the messages from Linux. You see the changes you make in the text onscreen.
You can use the editor to modify, rearrange, delete, substitute, and search for text. You
conduct these editing operations while using the editor in command mode. In several
instances, a command is a single letter that corresponds to the first letter of an action's name.
For example, i corresponds to the insert action, and r is used when replacing a character.
Most commands operate on a single line or range of lines of text. The lines are numbered
from 1 (the top line) to the last line in the buffer. When you add or delete lines, the line
numbers adjust automatically. A line's number is its address in the buffer. An address range is
simply two addresses or line numbers separated by a comma. If you want to specify the range
consisting of the third through the eighth line of the buffer, you use 3,8.
207
Using vi
The position of the cursor always indicates your current location in the editing buffer. Some
of the commands you issue in command mode affect the character at the cursor position.
Unless you move the cursor, changes take place at that position. Naturally, vi has several
commands for moving the cursor through the edit buffer.
You know now that vi is a full-screen editor. You can give vi commands to move the cursor to
different positions in a file, and you see the changes you make as you make them. Therefore,
vi has to be able to move to and modify the text on your terminal as well as on a host of other Part
terminal types. It knows what terminal you're using and what its video capabilities are by
II
checking the shell variable TERM. Linux uses the TERM variable to determine your terminal's
Ch
capabilities, such as underlining, reverse video, the screen-clearing method, function-key
assignment, and color capability. 9
Using vi
To start vi, simply type its name at the shell prompt (command line). If you know the name of
the file you want to create or edit, you can issue the vi command with the filename as an
argument. For example, to create the file myfile with vi, enter vi myfile.
~
When vi becomes active, the terminal screen clears, and a tilde character ( ) appears on the
left side of every screen line, except for the first. The ~ is the empty-buffer line flag. The
following is a shortened version of what you should see on your screen (only five lines are
listed to save space):
_
~
~
~
~
The cursor appears at the leftmost position of the first line (represented here as an underscore
character). You'll probably see 20 to 22 of the tilde characters at the left of the screen. If that's
not the case, check the value of TERM (as described in the troubleshooting section at the end of
the chapter) and perhaps talk with your system administrator.
When you see this display, you've successfully started vi; vi is in command mode, waiting for
your first command.
Unlike most word processors, vi starts in command mode. Before you start entering
text, you must switch to input mode by using the a or i keys, both of which are
described in the next section.
208 Chapter 9 Using the vi Editor
Looking at vi's Two Modes
As mentioned earlier, the vi editor operates in two modes: command mode and input mode.
In command mode, vi interprets your keystrokes as commands; vi has many commands. You
can use commands to save a file; exit vi; move the cursor to various positions in a file; and
modify, rearrange, delete, substitute, and search for text. You can even pass a command to the
shell. If you enter a character as a command, but that character isn't a command, vi beeps.
Don't worry; the beep is an audible indication for you to check what you're doing and correct
any errors.
You can enter text in input mode (also called text-entry mode) by appending characters after
the cursor or inserting them before the cursor. At the beginning of the line, which method
you use doesn't make much difference. To go from command mode to input mode, press one
of the following keys:
a To append text after the cursor
I To insert text in front of the cursor
Use input mode only for entering text. Most word processors start in input mode, but vi
doesn't. When you use a word processing program, you can type away, entering text; to issue
a command, you have to use function keys or keys different than those you use when typing
normal text. vi doesn't work that way: You must go into input mode by pressing a or i before
you start entering text and then explicitly press Esc to return to command mode.
Creating Your First vi File
The best way to learn about vi is to use it. This section gives a step-by-step example of how to
create a file by using vi. In each step, you see an action to perform and then the necessary
keystrokes. Don't be concerned with complete accuracy here. The example takes you through
the motions and concepts of using vi to create a file, moving between command and input
modes, and saving your results. If you run into difficulties, you can quit and start over by
pressing Esc; then you can enter :q!. When you do, you'll lose all your changes since the last
time you saved, but that's the point: You can avoid saving your recent work if it's all mistakes.
1. Start vi by entering vi. You see the screen full of flush-left tildes.
2. Go into input mode to place characters on the first line. To do so, press the a key;
don't press Return. Now you can append characters to the first line. You shouldn't see
the character a onscreen.
3. Add lines of text to the buffer by typing the following:
Things to do today.
a. Practice vi.
b. Sort sales data and print the results.
You can use the Backspace key to correct mistakes on the line you're typing. Don't
worry about being precise here; this example is for practice. You'll learn other ways to
make changes in some of the later sections of this chapter.
209
Using vi
4. Go from input mode to command mode by pressing Esc. You hear a beep from your
system if you press Esc when you're already in command mode.
5. Save your buffer in a file called vipract.1 by entering :w vipract.1. The characters :w
vipract.1 appear on the bottom line of the screen (the status line). The characters
shouldn't appear in the text. The :w command writes the buffer to the specified file.
This command saves or writes the buffer to the file vipract.1.
6. Look at your action confirmed on the status line. You should see the following on the Part
status line:
II
"vipract.1'' [New File] 3 lines, 78 characters
Ch
This statement confirms that the file vipract.1 has been created, is a new file, and
9
contains 3 lines and 78 characters. Your display might be different if you didn't type
the information exactly as specified.
7. Exit vi by entering :q.
When you type :q, you're still in command mode, so you see these characters on the status
line. When you press Return, however, vi terminates, and you are returned to the login shell
prompt.
You use these steps, or variations of them, for all your editing tasks. Make sure that you can
work through them before continuing.
Points to Remember About vi
The following are some points to remember about vi:
. vi starts in command mode.
. To move from command mode to input mode, press a (to append text) or i (to insert text).
. You add text when you're in input mode.
. You give commands to vi only when you're in command mode.
. You give commands to vi to save a file and can quit only when you're in command mode.
. To move from input mode to command mode, press Esc.
Starting vi by Using an Existing File
To edit or look at a file that already exists in your current directory, type vi followed by the
filename. Try this procedure with the file you created in the preceding section by entering the
following:
vi vipract.1
You should see the following display (the number of lines shown here are fewer than you will
actually see onscreen):
Things to do today.
a. Practice vi.
b. Sort sales data and print the results.
~
~
210 Chapter 9 Using the vi Editor
~
"vipract.1'' 3 lines, 78 characters
As before, tilde characters appear on the far left of empty lines in the buffer. Look at the
status line: It contains the name of the file you're editing and the number of lines and
characters.
TROUBLESHOOTING
I type a filename that I know exists, but vi acts as though I'm creating a new file.
No one is a perfect typist; you may have typed the name of a file that doesn't exist in your current directory.
Suppose that you type vivipract.1, but you don't have a file named vipract.1 in your current directory.
You still start vi, but vi acts as though you were creating a new file.
I try to edit a file, but vi displays a message about read permission denied and I see the shell prompt again.
You've tried to edit a file you aren't permitted to read. Also, you can't edit a directory; that is, if you type vi
directory_name, where directory_name is the name of a directory, vi informs you that you opened a
directory and doesn't let you edit it. If you try to use vi with a file that's an executable program in binary, as
opposed to ASCII, you see a screen full of strange (control) characters. It isn't something you can read and edit.
vi expects files to be stored as lines.
I open a file in vi, but I see a message that the line is too long.
You're trying to use vi on a data file that's just one long string of bytes. You can modify this file, but doing so will
probably corrupt the data file.
I open a file in vi, but I see some very strange characters onscreen.
You may be using vi with a file produced by a word processor.
In all these cases, exit vi to return to your login shell prompt by pressing Esc to go to command mode and then
typing :q!. Using :q! ensures that you quit vi and make no changes to the existing file.
Exiting vi
You can exit or quit vi in several ways. Table 9.1 lists the commands you can use to exit.
Remember that you must be in command mode to quit vi. To change to command
mode, press the Esc key. (If you're already in command mode when you press Esc,
you hear a harmless beep from the terminal.)
Table 9.1 Ways to Quit or Exit vi
Command Action
:q Exits after making no changes to the buffer, or exits after the buffer is
modified and saved to a file
:q! Exits and abandons all changes to the buffer since it was last saved to a file
:wq, :x, or ZZ Writes the buffer to the working file and then exits
211
Using vi
As you can see in Table 9.1, several keystrokes accomplish the same end. To practice, you can
use vi to edit the file vipract.1 created earlier in this chapter. To edit the file, enter vi
vipract.1. You should see a display similar to this:
Things to do today.
a. Practice vi.
b. Sort sales data and print the results.
~
~ Part
~
"vipract.1'' 3 lines, 78 characters II
The cursor is indicated by an underscore character. When you first open the file, it's under Ch
the first character of the file (the T in Things). Because you haven't made any changes to the
9
file since you opened it, you can exit by entering :q. When you do, you see the shell prompt.
You can also type :wq to exit the file; if you do so, you see the following message before the
shell prompt appears:
"vipract.1'' 3 lines, 78 characters
This message appears because vi first writes the buffer to the file vipract.1 and then exits.
You can start vi again with the same file (by typing vi vipract.1). You then should see a
display similar to this:
Things to do today.
a. Practice vi.
b. Sort sales data and print the results.
~
~
~
"vipract.1'' 3 lines, 78 characters
Although vi starts in command mode, just to be sure, press Esc. Now press the spacebar
enough times so that the cursor moves under the period following today in the first line. To
replace that character with an exclamation mark, press the r key (for replace) and type !. The
first line now looks like this:
Things to do today!
Because you've changed the buffer, vi doesn't let you exit unless you save the changes or
explicitly give a command to quit without saving the changes. If you try to exit vi by typing
:q, vi displays the following message to remind you that you haven't written the file to disk
since you changed it:
No write since last change (:quit! overrides)
To abandon the changes you've made to the file, quit by typing :q!. To save the changes, quit
by typing :wq or any other equivalent form ( ZZ or :x).
vi doesn't keep backup copies of files. After you enter :wq, the original file is modified
and can't be restored to its original state. You must make your own backup copies of
vi files.
212 Chapter 9 Using the vi Editor
? See Chapter 12, ``Backing Up Data'', p. 259
Use the :q! command sparingly. When you enter :q!, all the changes you've made
to the file are lost.
Rather than issue a :q! command, you could save the file to a different filename. That subject
is covered later in this chapter in the section ``Saving as a New File."
Undoing a Command
In vi, you can ``undo'' your recent actions or changes to the buffer as long as you haven't
saved those changes to the disk file. You do so in command mode. Suppose that you've
inadvertently deleted a line of text, changed something you shouldn't have, or added some
text incorrectly. To undo your changes, press Esc to change to command mode, and then
press u. Pressing these keys returns things to the way they were before the buffer was changed
with the undo command sequence.
The following is an example of using the undo command. To follow along, start vi again with
the file vipract.1 (by entering vi vipract.1). You should see a display similar to the
following:
Things to do today!
a. Practice vi.
b. Sort sales data and print the results.
~
~
~
"vipract.1'' 3 lines, 78 characters
To add the phrase for 60 minutes between vi and the period on the second line, move to the
second line by pressing Return. The cursor now appears under the first character of the
second line. Now move the cursor to the period after the word vi by pressing the spacebar
until the cursor moves to that location. Insert the phrase by pressing the i key to give the
input command and then typing for 60 minutes. Press Esc to return to command mode. Your
screen now looks like this:
Things to do today!
a. Practice vi for 60 minutes.
b. Sort sales data and print the results.
~
~
~
Is 60 minutes a good idea? Maybe not. To undo the change to the second line, make sure that
you're in command mode (press Esc), and then press u. The second line of the file now looks
like this:
a. Practice vi.
213
Using vi
Then again, maybe practicing for 60 minutes was a good idea. Press the u key again (you're
already in command mode), and you see the phrase for 60 minutes reappear. Will you or won't
you practice for that long? You decide. Use the undo command to undo the change (and
undo the undo) as many times as you want. Even if you decide to leave the buffer in its
original form, vi assumes that the buffer has changed, and you must exit by using :q!
(abandon changes) or :wq (save the changes).
If you decide to save the file with the changes, save it to another file by entering :w vipract.2. Part
II
Tip #58 from Ch
Steve You can use the Backspace key to correct mistakes you make while typing a single
9
line. Unfortunately, as you backspace, you erase all the characters you go back over.
The left-arrow key doesn't erase characters. The arrow keys are covered later in this
chapter.
Writing Files and Saving the Buffer
You've seen how to write the buffer to a file and quit vi. Sometimes, however, you might want
to save the buffer to a file without quitting vi. You should save your file regularly during an
editing session. If the system goes down because of a crash or a power failure, you may lose
your work if you haven't saved it recently. To save the buffer, you issue the :w (write)
command from command mode.
Before you issue the write command, first press Esc to change to command mode if
you aren't already there. If you're already in command mode, you hear a harmless
beep.
You'll notice some variations to the steps you follow to save a file. The form of the write
command you use depends on the case, of which there are four distinct ones. The following
sections describe these cases, and Table 9.2 lists the variations of the write command.
Table 9.2 Commands to Save or Write a File
Command Action
:w Writes the buffer to the file vi is editing
:w filename Writes the buffer to the named file
:w! filename Forces vi to overwrite an existing file
214 Chapter 9 Using the vi Editor
Saving a New File
If you started vi without specifying a filename, you must provide a filename if you want to
save the file to disk. The write command you issue in this case has the following format:
:w filename
This command writes the buffer to the file filename. If the command is successful, you see
the name of the file and the number of lines and characters in the file. If you specify the name
of an existing file, an appropriate message appears on the status line:
File exists - use ``w! filename'' to overwrite.
This condition is described later in the section ``Overwriting an Existing File."
Saving to the Current File
You might want to save the buffer to the file you're now editing. For example, if you started vi
with an existing file, made some changes to the file, and want to save the changes to the
original file, you can simply enter :w, a form of the write command.
Tip #59 from
Steve Save the changes you're making to a file regularly. Use the :w command
frequentlyÐat least every 15 minutesÐduring an edit session. You never know when
the system might go down.
The :w command saves the buffer to the file you're now working with (your working file).
The status line tells you the name of the file and the number of lines and characters written to
the file.
Saving As a New File
You might want to save the buffer to a new file, giving it a different filename from the one you
originally started with. For example, if you started vi with the file vipract.1, made some
changes to the file, and want to save the changes to a new file without losing the original
vipract.1 file, you can save the file as a new file. Type this form of the write command to
save the file with a new filename:
:w filename2
This form of the write command is essentially the same as the original form described earlier
in the section ``Saving a New File.'' The buffer is written to the file named filename2. If the
command is successful, you see the name of the file and the number of lines and characters in
the file. If you specify the name of an existing file, an appropriate message appears on the
status line:
File exists - use ! to overwrite.
215
Using vi
The following section explains this scenario.
Overwriting an Existing File
If you try to save the buffer to an existing file different from the one you started with, you
must explicitly tell vi that you want to overwrite or replace the existing file. If you specify an
existing filename when you try to save the buffer, vi displays the following message:
File exists - use ! to overwrite. Part
If you really want to save the buffer over the existing file, use this form of the write command: II
:w! existing_file Ch
In this syntax, existing_file is the name of the file you want to replace. Be careful; after you 9
overwrite a file, you can't restore it to its original form.
Positioning the Cursor
When you edit text, you need to position the cursor where you want to insert additional text,
delete text, correct mistakes, change words, or append text to the end of existing text. The
commands you enter in command mode to select the spot you want are called cursor-
positioning commands.
The Arrow Keys
You can use the arrow keys on many, but not all, systems to position the cursor. You can
easily see whether the arrow keys work: Just start vi with an existing file and see what effects
the arrow keys have. You should also be able to use the Page Up and Page Down keys on the
Linux keyboard, providing you have the correct terminal type indicated in your TERMCAP
environment variable.
To create a new file called vipract.3 that contains a list of the files and directories in the
directory usr, enter the following command:
ls /usr ]] vipract.3
You can use this file to experiment with cursor-positioning commands.
After you create the file, start vi with the vipract.3 file (by entering vi vipract.3). Now try
using the arrow keys and the Page Up and Page Down keys to move around the editing
buffer.
You might find that, although the cursor-positioning keys appear to work, they introduce
strange characters into the file. To check whether the keys are entering characters instead of
just moving the cursor, press Esc to make sure that you're in command mode, and then enter
:q. If vi allows you to quit and doesn't complain that the file was modified, everything is fine.
Tip #60 from
Steve In vi, you can clear the screen of spurious or unusual characters by pressing Ctrl+l.
216 Chapter 9 Using the vi Editor
Other Cursor-Movement Keys
You can position the cursor in vi in other ways without using the arrow keys. You should
become familiar with these methods in case you can't or don't want to use the arrow keys.
This section also shows you some ways to position the cursor more efficiently than using the
arrow keys.
When vi was developed, many terminals didn't have arrow keys. Other keys were and still are
used to position the cursor. vi uses the h, j, k, and l keys to position the cursor because they're
in a convenient position for touch-typists. Getting comfortable with these keys takes a little
practice, but some experienced vi users still prefer these keys over the arrow keys.
The following are some other keys that move the cursor:
& Press the spacebar or the l key to move the cursor one position to the right.
& Press Return or + (the plus sign) to move to the beginning of the next line. (Note that
using the j key to go down one line preserves your position in the line.)
& Press - (the minus sign) to move to the beginning of the previous line. (Note that using
the k key to go up one line preserves your position in the line.)
& Press h to move one character to the left.
& Press 0 (zero) to move to the beginning of a line.
& Press $ (the dollar sign) to move to the end of a line.
Some vi commands allow you to position the cursor relative to words on a line. A wordis
defined as a sequence of characters separated from other characters by spaces or usual
punctuation symbols such as these:
. ? , -
Those commands include the following:
Keystroke Action
w Moves forward one word
b Moves to the beginning of the current word
e Moves to the end of the current word
The following example demonstrates some of these actions. To follow along, start vi and
open the vipract.1 file by entering vi vipract.1. Now use any of the cursor-positioning
commands just described to move the cursor (indicated by an underscore) to the t in the word
data on the third line of the file. The third line looks like this:
b. Sort sales data and print the results.
To move to the beginning of the next word, press w; the cursor is positioned under the a of
the word and. To move to the end of that word, press e; the cursor is positioned under the d in
and. To move to the beginning of that word, press b; the cursor is then positioned under the a
in and again.
217
Using vi
You can move forward several words to the beginning of another word by pressing a number
key before pressing w. For example, to move the cursor from its current position (under the a
of the word and) to the beginning of the word three words forward (under the r of the word
results), press 3+w. Likewise, you can move backward four words by pressing 4+b; you can
move forward to the end of the second word by pressing 2+e.
You can also use this whole-number technique with the h, j, k, l, +, and - keys. For example,
press 1+5+j to position the cursor down 15 lines. If 15 lines aren't left in the buffer, you hear a Part
beep, and the cursor stays where it is.
II
Ch
Big-Movement Keys
You can quickly position the cursor at the top, middle, or bottom of the screen. In each case,
9
the cursor appears at the beginning of the line. The following commands allow you to
position the cursor onscreen:
& Press Shift+h to move to the first line of the screen. This line is sometimes called the
home position.
& Press Shift+m to move to the line in the middle of the lines now displayed.
& Press Shift+l to move to the last line onscreen.
If you want to move through a file one screen at a time (which is more efficient than pressing
Return or the j key 23 times), use commands that scroll through a file. Pressing Ctrl+f moves
you forward one screen. Pressing Ctrl+b moves you backward one screen.
To move quickly to the last line of the file or buffer, press Shift+g. To move to the first line of
the file, press 1+Shift+g. In fact, to move to a specific line in the buffer, type the line number
before you press Shift+g. For example, to move to line 35 of the file (if the file has a line 35),
press 3+5+Shift+g.
Take a little time to practice positioning the cursor by using the commands described
in the preceding few sections. Remember that you must be in command mode for the
cursor-positioning commands to work. Press Esc before you issue a cursor-positioning
command.
Adding Text
To add text to the editing buffer, you must go from command mode to input mode. Any
usual text characters you type are then added to the buffer. If you press Return while you're in
input mode, vi ``opens,'' or adds, a line to the buffer. Before you start adding text, first
position the cursor at the location you want to add text. Press a to go to input mode and
append text after the cursor position. Press i to go to input mode and insert text in front of the
cursor position. When you're done adding text, press Esc to return to command mode.
218 Chapter 9 Using the vi Editor
The following are two examples of typing in input mode. The position of the cursor is
represented by an underscore character. For each case, before and after views are shown:
& Example showing the use of the i key (the insert command) to add text.
Before:
This report is important.
Press the i key to insert text in front of the word important, type very, press the
spacebar, and press Esc.
After:
This report is very_important.
Note that the cursor is positioned under the last character you added (in this case, the
space).
& Example showing the use of the a key (the append command) to add text.
Before:
This report i s important.
Press the a key to append text after the word is, press the spacebar, type very, and press
Esc.
After:
This report is ver y important.
Note again that the cursor is positioned under the last character you added (in this
case, the y in very).
When you want to append text to the end of a line, you can position the cursor at the end of
the line and press the a key. You can also position the cursor anywhere in the line and press
Shift+a to position the cursor at the end of the line, put you in input mode, and allow you to
append textÐall with one command. Likewise, you can move to the beginning of the current
line and insert text at the beginning of a line by pressing Shift+i.
To add a line of text below or above the current line, you press the o key or Shift+o,
respectively. Each keystroke ``opens'' a line in the buffer and allows you to add text. In the
following two examples, you add a line to some existing text:
& Example showing the use of the o key to insert lines below the current line.
Before:
All jobs complete
please call
if you have any questions.
The cursor is on the second line. Press the o key to add a line or lines below that line.
Now type the following lines:
John Baucom
555-2222
Press the Esc key.
219
Using vi
After:
All jobs complete
please call
John Baucom
555-2222
if you have any questions.
& Example showing the use of Shift+o to insert lines above the current line.
Part
Before:
All jobs complete II
please call
if you have any questions. Ch
The cursor is on the third line. Press Shift+o to add a line or lines above that line.
9
Now type the following lines:
John Baucom
555-2222
Press the Esc key.
After:
All jobs complete
please call
John Baucom
555-2222
if you have any questions.
In both cases, when you press Esc, the cursor is positioned under the last character you typed
(the last 2 in the phone number). Although you added only two lines, you could have added
more lines by pressing Return at the end of each line. Naturally, you could have added only
one line by not pressing Return at all.
Table 9.3 summarizes the commands for adding text. Press Esc to make sure that you're in
command mode before using these commands.
Table 9.3 Commands for Adding Text
Keystroke Action
a Appends text after the cursor position
Shift+a Puts you in input mode and appends text to the end of the current line
i Inserts text in front of the cursor position
Shift+i Puts you in input mode and inserts text at the beginning of the current line
o Opens a line below the current line to add text
Shift+o Opens a line above the current line to add text
220 Chapter 9 Using the vi Editor
Deleting Text
Making corrections or changes to a file may involve deleting text. You must be in command
mode to delete characters. If you're in input mode when you type the delete-character
commands, the letters of the commands appear as characters in the buffer file. If that should
happen, press Esc to go to command mode, and press the u key to undo the mistake.
With vi, you can delete a character, a word, a number of consecutive words, all the text to the
end of a line, or an entire line. Because vi is a visual editor, the characters, words, or lines are
removed from the screen as you delete them. Table 9.4 describes the delete commands.
Table 9.4 Commands for Deleting Text
Keystroke Action
x Deletes the character at the cursor position
d+w Deletes from the cursor position in the current word to the beginning of the
next word
d+$ Deletes from the cursor position to the end of the line
Shift+d Same as d+$: deletes the remainder of the current line
d+d Deletes the entire current line, regardless of cursor position in the line
All these commands take effect from the current cursor position. You move the cursor to the
character, word, or line you want to change and then issue the desired delete command.
Practice using them to see their effect. You'll find they're helpful in making corrections to
files.
You can apply these commands to several objectsÐcharacters, words, or linesÐby typing a
whole number before the command. (This whole-number technique was introduced earlier in
this chapter in the section on positioning the cursor.) Some examples are as follows:
& Press 4+x to delete four characters.
& Press 3+d+w to delete three words.
& Press 8+d+d to delete eight lines.
Tip #61 from
Steve To have vi display line numbers, press Esc to make sure that you're in command
mode, and then enter :se number. To turn off the line numbers, enter :se
nonumber.
You can also specify a range of lines to delete. To do so, press the colon (Shift+;), type the two
line numbers you want to delete (inclusive) separated by a comma, press the d key, and press
Return. For example, to delete lines 12 through 36 (inclusive), type :12,36d and press Return.
221
Using vi
When you delete two or more lines, the status line states how many lines were deleted.
Remember that you can press the u key to undo the deletion.
Searching
Finding a word, phrase, or number in a file can be difficult if you have to read through each
line yourself. Like most editors and word processors, vi has a command that allows you to
search for a string of characters. You can search forward or backward from your current
Part
position in the buffer. You also can continue searching. vi starts searching from the beginning
II
of the buffer file when it reaches the end, and vice versa. Table 9.5 summarizes the commands
for searching. In each case, vi searches for the string you specify in the direction you specify Ch
and then positions the cursor at the beginning of the string. 9
Table 9.5 The Search Commands
Command Action
/string Searches forward through the buffer for string
?string Searches backward through the buffer for string
n Searches again in the current direction
Shift+n Searches again in the opposite direction
When you type the search command, it appears on the status line. To search forward for the
string sales > 100K in a file, for example, first you make sure you're in command mode, and
then you enter the following:
/sales > 100K
The typed command appears on the status line. If the string is in the buffer, vi positions the
cursor under the first s in the word sales. If the string isn't in the buffer, vi displays the message
Pattern not found on the status line. To search for another occurrence of the string, press the
n key; vi positions the cursor under the next occurrence of the string or, if there's no ``next
occurrence,'' the cursor doesn't move.
TROUBLESHOOTING
I typed a string I know exists in the file, but vi can't find it.
The most common cause for this error is that you typed the string incorrectly. vi (and computers in general)
doesn't do a good job of thinking; vi has a terrible time figuring out what you really mean when you type
something. If you're looking for the string vegi-burger, but you type vigi-burger, vi can't find what you
want (unless you happened to misspell vegi-burger in the buffer, and it matches the search string). Check the
search string carefully before you press Return.
I searched for a phrase that incorporates a punctuation mark, and vi returned some odd results.
Searching in vi may not give you the results you want if you're looking for characters that are ``special'' to vi. For
example, if you want to find a word you know is located at the end of a sentence (for example, the string end.),
you must ``escape'' the period; to vi, the period means ``any character,'' not ``end of sentence.'' If you enter
/end. and press Return, vi would locate such things as the word ending, the word end followed by a space,
and the word end followed by a period. To find only end followed by a period, you need to enter /end\..
222 Chapter 9 Using the vi Editor
Searches in vi are also case sensitive. If you're looking for the word Tiger in your buffer, you must enter
/Tiger, not /tiger.
Changing and Replacing Text
Another often-faced editing task is changing text or replacing one text string with another
(the two operations aren't very different). The change commands in vi allow you to change a
word or the remainder of a line. In effect, you replace one word or the remainder of a line
with another. You use the replace commands to replace or change a single character or
sequence of characters. Table 9.6 summarizes the change and replace commands. After you
enter the command, simply type the new material as appropriate.
Table 9.6 The Change and Replace Commands
Keystroke Action
r Replaces a single character
Shift+r Replaces a sequence of characters
c+w Changes the current word, from the cursor position to the end of the word
c+e Changes the current word, from the cursor position to the end of the word
(same as c+w)
c+b Changes the current word, from the beginning of the word to the character
before the cursor position
c+$ Changes a line, from the cursor position to the end of the line
Shift+c Changes a line, from the cursor position to the end of the line (same as c+$)
c+c Changes the entire line
The changes take place relative to the position of the cursor. You must be in command mode
before you can use these commands. Position the cursor at the location in the buffer file you
want to correct and press Esc before using these commands. Because vi is visual, the changes
are made to the buffer as you execute the commands.
Each of these commands puts you into input mode. Except for when you use r to replace a
single character, you must press Esc to finish making changes and return to command mode.
Tip #62 from
Steve To change several words, use a whole number (representing the number of words to
change) before pressing c+w. The specified number of words are deleted, and you're
put back into insert mode.
The following are three examples of how to use the change and replace commands, with
before and after scenarios for each:
223
Using vi
& Example showing the use of c+e to change to the end of the word.
Before:
The report demonstra its thw,strengths of are apporach.
The cursor is located at the point in the incorrectly spelled word where corrections are
to begin. To change the spelling, press c+e, type tes, and press Esc.
After:
Part
The report demonstrate s thw,strengths of are apporach.
II
& Example showing the use of Shift+r to replace a sequence of characters.
Ch
Before:
The report demonstrates th w,strengths of are apporach. 9
The cursor is located at the point in the incorrectly spelled word where you want to
start replacing characters. To change thw, to the and a space, press Shift+r, type e,
press the spacebar, and press Esc.
After:
The report demonstrates the_strengths of are apporach.
& Example showing the use of c+w to change text, beginning with the current word and
continuing for two words.
Before:
The report demonstrates the strengths of are apporach.
The cursor is positioned under the letter of the word where you want to begin making
changes. To fix the last two words on the line, press 2+c+w, type our approach, and
press Esc.
After:
The report demonstrates the strengths of our approac h.
Remember to press Esc after you make changes to the lines and return to command mode.
Copying, Cutting, and Pasting
When you delete or cut characters, words, lines, or a portion of a line, the deleted object is
saved in what's called the general-purpose buffer. The name isn't too important; what's
important is that you can put or paste the contents of that buffer anywhere in the text you're
editing. You do so by using the p or Shift+p command. The p command pastes the object to
the right of or after the cursor position; the Shift+p command pastes the object to the left of
or before the cursor.
The following are some examples of cutting and pasting text with before and after scenarios
for each:
& Example showing the use of p to paste the contents of the general-purpose buffer after
the cursor.
224 Chapter 9 Using the vi Editor
Before:
Carefully carry these out instructions.
Delete the characters out and a space by pressing d+w. Now move the cursor to the
space after the y in carry and press the p key.
After:
Carefully carry out_these instructions.
& Example showing the use of Shift+p to paste the contents of the general-purpose
buffer in front of the cursor.
Before:
Carefully carry these out instructions.
Delete the characters these and a space by pressing d+w. Now move the cursor to the
first i in instructions and press Shift+p.
After:
Carefully carry ou t these instructions.
Tip #63 from
Steve To change the order of two characters, position the cursor under the first character
and press x+p. Try it to change the word tow to the word two, for example.
The preceding examples show you how to paste after deleting text. But you don't have to
delete before you can paste. You can use an operation called yank, which is the same as the
copy operation in some word processors. The forms of the yank command are similar to the
forms of the delete command. The idea is that you yank, or copy, a portion of text and then
paste it somewhere else by pressing the p key or Shift+p. The following list names some of the
yank commands (notice that most of the yank commands use the lowercase letter y):
Keystroke Action
y+w Yanks from the cursor position in the current word to
the beginning of the next word
y+$ Yanks from the cursor position to the end of the line
Shift+y Same as y+$: yanks the remainder of current line
y+y Yanks the entire current line
You can apply all these commands to several objectsÐcharacters, words, or linesÐby typing a
whole number before the command.
To copy a sequence of four lines to another portion of the text, follow these steps:
1. Position the cursor at the beginning of the first of the four lines.
225
Using vi
2. Press 4+y+y to yank from the cursor to the end of the line four times. The buffer (what
you see onscreen) is unchanged.
3. Position the cursor elsewhere in the text.
4. Press p to paste the yanked lines below the line holding the cursor.
You can also search and replace words throughout the file or within a specified range of lines.
The format of the command is
Part
:[range]s/oldstring/newstring/g
II
where
Ch
range Indicates the range on which to operate; for example,
9
you can use the percent symbol (%) to operate on the
entire file, or you can use specific line numbers (such as
1,4) to operate on specific lines (in this case, lines 1
through 4)
s Indicates this is a search-and-replace operation
oldstring Specifies the string to search for in the file and replace
with newstring
newstring Specifies the string to insert; newstring replaces
oldstring
For example, to replace the incorrectly spelled word recieved with the correct spelling
throughout an entire file, you can use the following command:
:%s/recieved/received/g
Repeating Commands
Not only does vi keep the text just deleted or yanked for future use, it also stores the last
command you used for future use. You can repeat the last command that changed the buffer
by pressing the . (period) key.
Suppose that you've completed a report but think it would be a good idea to put two lines
containing this text at key points in the report:
*************** Please comment ******
*************** On this section ******
To do so, follow these steps:
1. Position the cursor in the buffer file where you want to place these lines the first time.
2. Insert the lines by pressing the o key to open a line and typing the two lines of asterisks
and text.
3. Press Esc to make sure that you're in command mode.
4. As often as necessary, position the cursor to another section of the report and press the
. key to insert these same two lines again and again.
226 Chapter 9 Using the vi Editor
vi Command Summary
You now have a basic understanding of using vi for text processing. Table 9.7 provides a
summary of the keystrokes and commands you can use in vi.
Table 9.7 vi Command Summary
Keystroke/Command Description
i Inserts text before the cursor
I Enters text at the start of the line
a Inserts text after the cursor
A Enters text at the end of the line
o Opens a new line below the cursor
O Opens a new line above the cursor
d+w Deletes a word
d+d Deletes an entire line
D Deletes to the end of the line
x Deletes the character under the cursor
c+w Changes a word
c+c Changes a line
C Changes to the end of the line
R Replaces the character under the cursor
J Joins lines together
e Moves to the end of a word
w Moves to the next word
$ Moves to the end of the line
l Moves one space right
k Moves one line up
j Moves one line down
h Moves one space left
f+x Moves the cursor to the first occurrence of x
F+x Moves the cursor to the last occurrence of x
; Repeats the last f/F command
number+| Moves the cursor to the specified column number
H Moves the cursor to the top line onscreen (not the top line
of the file)
L Moves the cursor to the bottom line onscreen
M Moves the cursor to the middle line onscreen
227
vi Command Summary
Table 9.7 vi Command Summary
Keystroke/Command Description
G Moves the cursor to the bottom line of the file
number+G Moves the cursor to the specified line number (same as
Esc+: number)
^ Moves to the beginning of the line
Part
m+x Marks the current position with the letter x
II
Ctrl+d Scrolls forward one half of the screen
Ch
Ctrl+u Scrolls backward one half of the screen
9
Ctrl+f Scrolls forward one screen
Ctrl+b Scrolls backward one screen
Ctrl+l Redraws the screen
Ctrl+g Shows the filename, current line, and column number
z+z Redraws the screen with the current line in the middle of
the screen
y+y Yanks the entire line into the buffer
p Puts the contents of the buffer below the cursor
P Puts the contents of the buffer above the cursor
x"[number]"y+y Yanks the indicated number of lines into the buffer named x
( x can be any single character A 7Z)
x p Places the contents of buffer x after the cursor
:w [file] Writes the contents to disk as file
:q Quits vi
:q! Quits the file without saving changes
:wq Saves changes and quits vi
:r file Reads the specified file into editor
:e file Edits file
:!command Executes the specified shell command
:number Moves to the specified line number
:f Prints the current line and filename (same as Ctrl+g)
/string Searches forward for string
?string Searches backward for string
:x,ys/oldstring/newstring Replaces oldstring with newstring from line x to line y
(entering y = $ replaces to the end of the file)
Esc+u Undoes the last command
n Finds the next occurrence of the string
. Repeats the last command
~ Changes the character to the opposite case
228 Chapter 9 Using the vi Editor
Table 9.7 vi Command Summary
Keystroke/Command Description
Esc Switches to command mode
Setting the vi Environment
The vi editor has several options you may or may not choose to use. Some of these options
can be set on a system-wide basis by the system administrator. You can customize your
environment with a number of options that are in effect whenever you start vi. Table 9.8
summarizes all the environment options you can set for vi. When setting environment
options (as described in the next section), you can use the abbreviation shown in the first
column of the table or the full name used in the second column.
Table 9.8 Environment Options for vi
Abbreviated Option Full Name and Function of Option
ai autoindent indents each line to the same level as the line above
(useful for writing programs). The default is autoindent off.
ap autoprint prints the current line to the screen when the line is
changed. The default is autoprint on.
eb errorbells causes the computer to beep when you introduce a
command error. The default is errorbells off.
nu number displays line numbers when you're editing a file. The
default is number off.
redraw redraw keeps the screen up-to-date as changes occur. The default is
redraw on.
report report sets the size of an editing change that results in a message
on the status line. For example, report=3 triggers a message when
you delete three lines but not when fewer than three lines are
deleted. The default is report=5.
sm showmatch shows a matching open parenthesis when the closing
parenthesis is entered. This option is useful mainly for program-
mers writing program code. The default is showmatch off.
smd showmode displays INPUT, REPLACE, or CHANGE on the right side of the
status line when the associated command is given. The default is
showmode off.
warn warn displays a warning message when an attempt is made to exit vi
if the buffer has been changed and not saved to the disk file. The
default is warn on.
229
Setting the vi Environment
Table 9.8 Environment Options for vi
Abbreviated Option Full Name and Function of Option
wm=n wrapmargin defines the right margin. In the syntax of this
command, n is a whole number. If n is greater than 0, the command
forces a carriage return so that no word is n or fewer characters
from the right margin. For example, wm=5 tells vi to wrap the line
when a character occurs within five characters of the end of the
Part
line. Turn this option off by specifying wm=0, which is the default.
ws word search (called wrapscan on some systems) wraps from the II
<eof> (end-of-file) character to the < bof > (beginning-of-file)
character during a search. Default is word search on.
Ch
Using set to See and Set Options
To see the options now set for your system, you can enter :set while in command mode in vi.
The options currently set for this session of vi are displayed on the status line. The options
displayed with the set command vary depending on the default options and on your
particular implementation of vi. The following is an example of what you might see when you
issue the set command:
autoprint errorbells redraw report=1 showmatch showmode term=vt100 wrap margin=5
Issuing theset command with no arguments results in a display of only the user-set
set command as se. To set a number of options on
options. You can abbreviate the
the same line, use the se command and separate the options with a space, as in the
following example:
:se ap eb redraw report=1 sm smd warn wm=5 ws
Notice that the first character is the colon character, which indicates to vi that a
command is to be entered.
To see the list of all possible options and their settings, enter :set all. The options and their
settings listed in Table 9.7 are displayed.
Setting the showmode Option
One of the most used options in vi is showmode. To learn about the showmode option, start vi
again with the vipract.1 file (by entering vi vipract.1).
When vi executes, you see the text from your first vi session onscreen. In your first session,
you may have noticed that you could not determine whether you were in input mode when
you entered the text for this file. You can tell vi to inform you when you're in input mode by
230 Chapter 9 Using the vi Editor
using the showmode option. The showmode option identifies the mode you're in on the status
line.
When you set the showmode option, vi displays whatever type of input mode it's in: regular
INPUT MODE, APPEND MODE, REPLACE 1 CHAR mode, and so on. To set showmode in vi, press Esc to
make sure that you're in command mode, and then enter :set showmode. Now go to input
mode (press the i key). You should see the message INPUT MODE on the status line. Press Esc to
return to command mode. You might want to see what happens when you give the commands
to replace or change text.
Setting Toggle Options
Any option that doesn't take a number argument is like a toggle switch: You can turn it on or
off. For example, as you learned in the preceding section, you set the showmode option by
entering this command:
:se showmode
To turn off the showmode option, you simply add no in front of the option like this:
:se noshowmode
Changing Options for Every vi Session
Setting an option during a vi session sets that option for the current session only. You can
customize your vi sessions by putting the set commands in a file named .exrc in your home
directory. To see whether such a file exists, type the following commands:
cd
vi .exrc
The first command takes you to your home directory. The second starts vi by using the .exrc
file. If the file exists, it appears on the vi screen. If the file doesn't exist, vi lets you know it's a
new file.
The set commands in the .exrc file start with the word set but no colon. For example, the
following line sets the options number and showmode:
set number showmode
The .exrc file is read when you start vi. If you create it while you're in vi, you must
restart vi to put the settings into effect.
The options you set and the values you give to some options depend on your preferences and
the type of editing you'll be doing. Experiment with some options or talk with more
experienced users.
231
Troubleshooting
Troubleshooting
My vi editor doesn't appear to be working correctly with my terminal or screen; I see ``strange''
characters.
The TERM variable may not be set correctly. Another symptom of an improper terminal setup
$TERM expression gives the value of
is that blocks of characters overwrite legible text. The
your current terminal setting. To check the value of TERM, enter echo$TERM. If you work at a
Part
terminal that isÐor emulatesÐa vt100, this command displays the following result (type the
command on the terminal, not while in the vi editor):
II
vt100 Ch
9
If the proper terminal type isn't echoed back, set the value of TERM by entering the following
command, if you're using the bash shell:
TERM=vt100
export TERM
If you're using the C shell, enter the following (the spaces around the = sign are important):
setenv TERM = vt100
export TERM
?
Your specific terminal type may be different from vt100; set TERM appropriately.
See Chapter 16, ``Understanding Linux Shells,'' p. 317
I start vi but don't get the expected responses.
Check to see whether your terminal is properly set up. Your terminal type isn't the same as
the name of your terminal; your terminal type must match one of the terminal types
contained in the directory /usr/lib/terminfo.
CHAPTER
10
Booting and Shutting Down
In this chapter by Jack Tackett, Jr.
Understanding the Boot Process 234
Booting Linux from a Floppy 240
Booting from a Boot Manager 241
Understanding LILO, the Linux Loader 242
Shutting Down Linux 243
Troubleshooting Startup and Shutdown 245
234 Chapter 10 Booting and Shutting Down
Understanding the Boot Process
Two of the most common tasks that you encounter when administering a Linux system are
booting the system and shutting it down. As you might have guessed, booting and shutting
down Linux are operations that require special consideration.
To use Linux, you must boot the operating system. Although this process sounds pretty
straightforward, you need to consider that most people run at least one additional operating
system on their PCs other than Linux. This means that you must have some way to specify
which operating system you want to boot when you start the system. You can do so in two
basic ways: You can boot Linux from a floppy, or you can boot from your hard drive by using
a boot manager.
Red Hat and most modern distributions of Linux use the SysV init boot process instead of
the older BSD style init. init is the first program the kernel executes at startup and, hence,
is given the process ID (or PID) of 1. It becomes the parent process for all other processes
running in the Linux system.
The PID of a process is a number the operating system uses to identify that process.
Many Linux commands use this PID number as a qualifying parameter.
Tip #64 from
Jack The ps aux command displays the PID of all running processes.
Linux follows these steps to boot:
1. The kernel runs the init program, which is located in the /sbin directory.
2. init runs the shell script /etc/rc.d/rc.sysinit.
3. rc.sysinit sets various system variables and performs other startup initializations.
4. init runs all the scripts specified for the default run level.
?
5. init runs the script /etc/rc.d/rc.local.
See ``Understanding Multitasking,'' p. 366
This program starts various processes and writes information to the console and to the system
log file /var/log/messages about the status of each process that's started.
Tip #65 from
Jack The /var/log/message log file is an excellent aid for debugging startup problems.
The kernel stores all error messages here, so you don't have to worry about writing
235
Understanding the Boot Process
down the messages as they scroll by during startup. The log files/var/log/boot
and /var/log/dmesg also contain log entries associated with startup and shutdown.
To scroll through these files, use the command less, as in
less /var/log/dmesq
init starts all the processes required by the operating system to perform its duties, such as
allowing network operations, use of the mouse, and basic functions like I/O to the terminal.
The SysV init program knows which processes to start by reading config files located in
/etc/rd.d. These files are further segregated according to run levels, specified by directories.
A run level specifies what types of services are available, from single-user mode (run level 1) to
full multiuser, multitasking, all-processes-running mode (run level 3). Table 10.1 outlines the
various run levels available in Linux.
Part
Table 10.1 Linux Run Levels II
Ch
Run Level Description
10
0 Halt
1 Single-user mode
2 Multiuser, no NFS
3 Full multiuser mode
4 Unused
5 X11
6 Reboot
You can change run levels by using the telinit command, which has the following form:
/sbin/telinit [-t sec ] [0123456sSQqabcUu]
-t sec Indicates the number of seconds to delay before
executing the command
0 to 6 Indicate the run level to switch to from the current level
sS Indicates to switch to single-user mode
abc Indicates to switch to level a, b, or c specified in
/etc/inittab
qQ Causes init to re-examine the /etc/inittab file
wW Re-executes init while preserving current system states
The init program uses the following directory structure:
init.d
rc0.d
rc1.d
rc2.d
rc3.d
rc4.d
236 Chapter 10 Booting and Shutting Down
rc5.d
rc6.d
The various numbers in the directory names correspond to the run levels in Table 10.1. Each
directory contains various shell scripts that start or stop the necessary services required in
?
each run level. These scripts also initialize the file system and lock files to a known state.
See ``Working with Shell Scripts,'' p. 348
Each directory contains various shell scripts. Each script's filename begins with either an S or
a K (for Start or Kill) and a two-digit number. The numbers are used to order the sequence
but have no other meaning.
Each script usually accepts either a start or stop command-line argument, although it can
accept other parameters. init supplies either a start or stop to the script, depending on
whether rc has been called to change run levels. You can also execute the scripts by hand if
you need to reconfigure a service; for example, you can use sendmail with the following
command (you must be logged in as root to execute the initscripts):
/etc/rc.d/init.d/sendmail stop
/etc/rc.d/init.d/sendmail stop
/etc/rc.d/init.d/sendmail start
You should notice two things about this command. First, the command is repeated twice with
the stop parameter. Repeating the command ensures that the system has time to stop the
process. Then the start command is called. Next, notice that the script is executed from the
init.d directory, not from the directory for the run level. Also, the script does not have a
letter (S or K) or a number. If you list the files in any run level directory, you will note that
they are actually linked to files in the init.d directory, as you can see in Listing 10.1.
Following the listing, Table 10.2 outlines a few crucial startup scripts in this directory.
Listing 10.1 Directory Listing of a Typical rc3.d Directory
lrwxrwxrwx 1 root root 16 Jan 25 21:56 K08autofs -> ../init.d/autofs
lrwxrwxrwx 1 root root 18 Dec 14 12:17 K10pnserver -> ../init.d/pnserver
lrwxrwxrwx 1 root root 17 Dec 14 12:17 K20rusersd -> ../init.d/rusersd
lrwxrwxrwx 1 root root 15 Dec 14 12:17 K20rwhod -> ../init.d/rwhod
lrwxrwxrwx 1 root root 15 Dec 14 12:17 S15nfsfs -> ../init.d/nfsfs
lrwxrwxrwx 1 root root 16 Dec 14 12:17 S20random -> ../init.d/random
lrwxrwxrwx 1 root root 16 Dec 14 12:17 S30syslog -> ../init.d/syslog
lrwxrwxrwx 1 root root 13 Dec 14 12:17 S40atd -> ../init.d/atd
lrwxrwxrwx 1 root root 15 Dec 14 12:17 S40crond -> ../init.d/crond
lrwxrwxrwx 1 root root 14 Dec 14 12:17 S50inet -> ../init.d/inet
lrwxrwxrwx 1 root root 15 Dec 14 12:17 S55named -> ../init.d/named
lrwxrwxrwx 1 root root 13 Dec 14 12:17 S60lpd -> ../init.d/lpd
lrwxrwxrwx 1 root root 13 Jan 31 20:17 S72amd -> ../init.d/amd
lrwxrwxrwx 1 root root 18 Dec 14 12:17 S75keytable -> ../init.d/keytable
lrwxrwxrwx 1 root root 18 Dec 14 12:17 S80sendmail -> ../init.d/sendmail
237
Understanding the Boot Process
Table 10.2 rc.3 init Scripts
Script Name Daemon Description
S15nfsfs nfs Handles Network File System (NFS)
S30syslog syslog Allows logging of system messages in
/var/log/messages
S40atd atd Allows the user to perform a task at an
indicated time
S40crond cron Acts as the batch scheduler for Linux
S50inet inetd Acts as the super server (PID 1)
S55named Name server Provides DNS name services
S60lpd lpd Acts as the line printer daemon to allow
printing Part
II
init loops through the files in the specified run level directory and passes either the start or Ch
?
stop parameter as indicated by the first character of the filename.
10
See ``Links,'' p. 412
The rc.d directory also contains three files called rc, rc.local, and rc.sysinit. The rc shell
script is responsible for restarting the system in a different run level. That script takes one
parameter, which is a number corresponding to the new run level. The rc.local file is
executed after all the other scripts are executed during startup. You can place any local
initialization instructions in this file. The rc.local file (the contents of which are shown in
Listing 10.2) provides an example of starting a local process, called secure shell, which allows
secure remote access to the system.
Listing 10.2 A Sample rc.local Shell Script
#!/bin/sh
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.
if [ -f /etc/redhat-release ]; then
R=$(cat /etc/redhat-release)
else
R="release 3.0.3''
if
arch=$(uname -m)
a="a"
case ``_$arch'' in
_a*) a="an'';;
_i*) a="an'';;
esac
# This will overwrite /etc/issue at every boot. So, make any changes you
# want to make to /etc/issue here or you will lose them when you reboot.
echo ``'' > /etc/issue
238 Chapter 10 Booting and Shutting Down
Listing 10.2 continued
echo ``Red Hat Linux $R'' ]]]] /etc/issue
echo ``Kernel $(uname -r) on $a $(uname -m)'' ]]]] /etc/issue
cp -f /etc/issue /etc/issue.net
echo >> /etc/issue
## Start sshd
## Added By Lance Brown 1/29/1998
/usr/local/sbin/sshd
The rc.sysinit file is the first file init runs at startup. This script performs various
functions, such as setting systemwide variables (like the hostname), checking the file system
and starting repairs, turning on user quotas, and mounting the /proc file system. The script
in Listing 10.2 also starts a local process called sshd, which is the secure shell daemon that
provides secure Telnet and remote commands to Linux.
ssh is not part of most Linux distributions because of export restrictions on munitions
(the United States government has classified encryption utilities in the same category
as nuclear weapons). You can install the utility yourself, though. Just check out
http://www.cs.hut.fi/ssh/ for more information.
The default run level is decided in /etc/inittab with the following command:
id:3:initdefault:
This command tells the system to start in run level 3 (full multiuser and multitasking). Listing
10.3 shows a sample /etc/inittab file.
Listing 10.3 A Sample /etc/inittab File
# inittab This file describes how the INIT process should set up
# the system in a certain run-level.
#
# Author: Miquel van Smoorenburg, [[[email protected]]]
# Modified for RHS Linux by Marc Ewing and Donnie Barnes
#
# Default run level. The run levels used by RHS are:
# 0 - halt (Do NOT set initdefault to this)
# 1 - Single user mode
# 2 - Multiuser, without NFS (The same as 3, if you do not have
Anetworking)
# 3 - Full multiuser mode
# 4 - unused
# 5 - X11
# 6 - reboot (Do NOT set initdefault to this)
#
id:3:initdefault:
# System initialization.
239
Understanding the Boot Process
Listing 10.3 continued
si::sysinit:/etc/rc.d/rc.sysinit
l0:0:wait:/etc/rc.d/rc 0
l1:1:wait:/etc/rc.d/rc 1
l2:2:wait:/etc/rc.d/rc 2
l3:3:wait:/etc/rc.d/rc 3
l4:4:wait:/etc/rc.d/rc 4
l5:5:wait:/etc/rc.d/rc 5
l6:6:wait:/etc/rc.d/rc 6
# Things to run in every run level.
ud::once:/sbin/update
# Trap CTRL-ALT-DELETE
ca::ctrlaltdel:/sbin/shutdown -t3 -r now Part
# When our UPS tells us power has failed, assume we have a few minutes II
# of power left. Schedule a shutdown for 2 minutes from now. Ch
# This does, of course, assume you have power installed and your
# UPS connected and working correctly. 10
pf::powerfail:/sbin/shutdown -f -h +2 ``Power Failure; System Shutting Down''
# If power was restored before the shutdown kicked in, cancel it.
pr:12345:powerokwait:/sbin/shutdown -c ``Power Restored; Shutdown Cancelled''
# Run gettys in standard run levels
1:12345:respawn:/sbin/mingetty tty1
2:2345:respawn:/sbin/mingetty tty2
3:2345:respawn:/sbin/mingetty tty3
4:2345:respawn:/sbin/mingetty tty4
5:2345:respawn:/sbin/mingetty tty5
6:2345:respawn:/sbin/mingetty tty6
# Run xdm in run level 5
x:5:respawn:/usr/bin/X11/xdm _nodaemon
Do not specify either run level 0 or run level 6 as the default run level because either one will
render your system unusable. If, for some reason, your inittab file does become corrupted,
you can boot to single-user mode and fix the problem. To do so, at the LILO boot prompt,
enter the parameter LINUX single, like this:
LILO boot: LINUX single
The default listed in LILO is linux (lower case). It would be good to note that if the
kernel image name is changed when recompiling a kernel, the command becomes
ineffective. So, keeping the linux image name is a good thing.
? See ``Configuring LILO,'' p. 242
240 Chapter 10 Booting and Shutting Down
LILO is the Linux loader and is discussed later in this chapter.
Tip #66 from
Jack You can pass any run level to the LILO prompt by specifying the run level as a
number. For example, to enter run level 2, you enter the following command at the
LILO prompt:
LILO boot: linux 2
? See ``Understanding LILO, the Linux Loader,'' p. 242
Booting Linux from a Floppy
Many people use a boot floppy to start Linux. This boot floppy contains a copy of the Linux
kernel that points to the root Linux file system on the appropriate hard drive partition. The
Red Hat and Caldera Linux installation programs give you the opportunity to create a
bootable floppy during the installation process.
You should make a bootable Linux floppy disk during the installation, even if you
intend to install a boot manager on your hard drive. If your hard disk should crash,
the bootable floppy might be the only way to boot your system. Also, if you try to use
a ``generic'' boot disk made up after a crash from another Linux computer, it probably
won't work.
You can also use the installation disks in an emergency. At the boot prompt, pass the option
rescue to the kernel. After asking a few questions, the system prompts you to insert the
?
supplementary disk to finish the boot process.
See ``Creating the Boot, Supplemental, and Rescue Disks,'' p. 66
After booting, the system provides you a minimal command shell called ash, as well as several
other utilities. Table 10.3 lists the utilities provided, which should be enough to repair your
system.
Table 10.3 Rescue Utilities
Utility Description
cat Displays the contents of a file
chmod Changes the access permissions of files
cpio Copies files from archives
e2fsck Checks a Linux second extended file system
241
Booting from a Boot Manager
Table 10.3 Rescue Utilities
Utility Description
fdisk Partitions a table manipulator for Linux
gzip/gunzip Compresses or expands files
insmod Installs a loadable kernel module
ls Lists files
mkdir Creates a directory
mke2fs Creates a Linux second extended file system
mount Mounts a file system
rm Removes a file
rmmod Unloads a loadable module
Part
? See ``Using Basic Commands,'' p. 48
II
Ch
10
Tip #67 from
Jack For Intel systems with the root partition on an IDE hard drive, you can also use the
boot disk to boot Linux. At the boot prompt, enter the following command:
Linux single root=/dev/hda1 initrd=
Be sure to specify the appropriate device for your root file system if it is on an IDE
device other than /dev/hda1. This command mounts the root partition, dumps you
into single-user mode immediately, and then skips the rest of the boot process on the
boot disk. Unfortunately, this procedure does not work if your root partition is on a
SCSI device.
Booting from a Boot Manager
Linux comes with a boot manager known as LILO, which stands for Linux Loader. This
program modifies the master boot sector of your boot hard disk and allows you to choose
which operating system you want to boot when you turn on your computer.
Boot Managers
Using a boot manager has its advantages and disadvantages. With a boot manager, you don't need a floppy
disk to boot your system. Also, you can choose to boot different operating systems from a menu at boot time or
have the system default to a given operating system.
As for the disadvantages, a boot manager adds another level of complexity to the boot process. It must be
modified or possibly reinstalled if you add, delete, or upgrade a version of any of the operating systems on your
disk. It modifies the master boot record of your hard disk, so if something goes wrong, you might not be able to
boot with anything other than a floppy disk until you reformat your hard drive. Also, the boot manager that you
choose might not be compatible with some operating systems.
You should consider your own computing needs carefully before deciding whether to use a floppy or boot
manager for booting Linux.
242 Chapter 10 Booting and Shutting Down
You also can set up LILO so that it can be started from the OS/2 boot manager.
Understanding LILO, the Linux Loader
LILO is a boot manager that comes bundled as part of almost every Linux distribution,
including the Red Hat, Debian, and Caldera Linux distributions. It can be installed in the
master boot record, on a formatted floppy disk, or on the boot partition's super block for
booting OS/2.
When LILO is installed, you can use the master boot record to select from a set of different
operating systems at boot time. Depending on its configuration, LILO counts to a timeout
value and then boots a default operating system.
The easiest way to install LILO is to do so via the Red Hat or Caldera Linux installation
program, which takes you through a menu-driven system that automates much of the
installation process.
Installing LILO from the Red Hat or Caldera installation program is highly
recommended. Installing a boot manager is an inherently dangerous process; you can
easily corrupt data on your hard disk if the installation isn't done correctly.
Configuring LILO
LILO reads a configuration file, /etc/lilo.conf, and uses it to figure out what operating
systems are installed on your system and where their boot information is located. The
/etc/lilo.conf file starts with some information that tells LILO how to operate in general.
It then contains several sections that list the boot information specific to each operating
system that LILO can boot. LILO is configured to boot one section for each operating
system on your Linux system.
Two sections from a LILO configuration file follow:
# Section for the Linux partition
image=/vmlinuz
label=Linux
root=/dev/hda1
# Section for MS-DOS
other=/dev/hda3
table=/dev/hda
label=msdos
The first section gives the boot information for Linux. The image line tells LILO where the
Linux kernel is located. The label line that appears in both sections gives the name of the
243
Shutting Down Linux
operating system that appears in the LILO boot menu. The root line specifies the location of the
Linux root file system.
In the MS-DOS section, the other line indicates that the partition for an additional operating
system is located on the disk partition hda3. The table line tells LILO where to find the
partition table for /dev/hda3.
Using LILO
When you install LILO, you typically should set a default timeout value and a default
operating system to boot. Doing so allows you to have a certain amount of time to select
another operating system at boot time. In the event that you don't select an operating system,
LILO boots the one that you've set as the default at the end of the timeout count.
When you boot your computer with LILO installed, you get a prompt that reads LILO:. At
Part
this point, you have several options. You can wait and have Linux boot your default operating
system, or you can press Ctrl, Alt, or Shift to have LILO boot the default operating system
II
immediately. You can also type the name of one of the operating systems to have LILO boot Ch
the one you specify. Finally, you can press the Tab key to have LILO display a list of the
10
different available operating systems.
Tip #68 from
Jack If you are having problems booting your system, you can pass the flag -b emergency
to LILO as follows so that the system will boot directly into a single-user shell without
running any other startup scripts:
LILO boot: linux -b emergency
You are provided a minimal set of commands, and only the / and /proc file systems
are loaded. Also, you are required to provide the root password to perform any work
on the system.
Shutting Down Linux
With a Linux system, you have to be careful when you shut down the system. You can't
simply turn off the power. Linux maintains file system I/O information in memory buffers. If
you just power down a Linux system, file system corruption can result.
You should never turn off a Linux system without shutting down properly. The file
systems need to synchronize properly when the system is shutting down. You can
cause severe damage to the Linux file system if you just power off the system.
244 Chapter 10 Booting and Shutting Down
The best way to shut down a Linux system is to use the shutdown command. The syntax of
the command is as follows:
/sbin/shutdown [flags] time [warning-message]
[warning-message] is a message sent to all users who are currently logged on, and time is the
time that the shutdown is to occur. The time argument can take a couple of formats:
& It can be specified as an absolute time in the format hh:mm, where hh is the hour (in one
or two digits) and mm is the minute of the hour. The mm value must be specified with
two digits.
& The time value can also be given in the format +m, where m is the number of minutes to
wait before the shutdown. You can substitute the word now for +0.
Table 10.4 lists the flags that you can use with the shutdown command.
Table 10.4 Command-Line Flags for the shutdown Command
Flag Message to Linux
-tsec Wait the specified number of seconds between sending the warning and the kill
signal to all processes. This delay gives processes time to finish any shutdown
processing that they have to do.
-k Don't really shut down the system; just send the warning message to all users.
-r Reboot after shutdown.
-h Halt after shutdown.
-n Don't synchronize disks before rebooting or halting. Use this flag with caution; it can
cause your data to become corrupted.
-f Do a ``fast'' reboot. This type of reboot creates the /etc/fastboot file. The rc boot
script should check for this file and should not do a fsck if it's found.
-c Cancel an already running shutdown. When you use this option, you cannot
specify the time argument.
Tip #69 from
Jack To reboot your system quickly, simply use the command
shutdown -r now
The shutdown command prevents any users from logging on, notifies all users on the system
that the system will be shut down, waits until the time that you specify, and then sends a
SIGTERM signal to all processes so that they can exit cleanly. shutdown then calls halt or
reboot, depending on your command-line choice in the shutdown command.
245
Troubleshooting Startup and Shutdown
You can halt or reboot the system by entering halt or reboot directly. However, if
you use either of these commands, no warning is given to the users, and the system
goes down immediately. You should use these commands only if you're the only user
on the system. To see who is logged on to the system, either press w or use the
command who.
Troubleshooting Startup and Shutdown
Why does LILO hang on LI? Can I use a hard drive that has more than 1023 cylinders?
The infamous 1023 cylinder question. Yes, you can use such a hard drive but not to boot
Linux. You can install Linux on partitions above the 1023 cylinder, but to boot Linux, the Part
root directory and specifically the /boot directory must be installed on the first hard drive II
below 1024. See the following site for more information:
Ch
http://metalab.unc.edu/LDP/HOWTO/mini/Large-Disk.html
10
How do I add arguments for LILO at the prompt?
Some hardware requires that extra parameters be fed to the kernel before the kernel
will recognize the hardware. You can accommodate this requirement by editing the
/etc/lilo.conf file to provide the necessary parameters, or you can provide them
manually during bootup. See the ``LILO How-To'' for more examples of LILO parameters.
The installation cannot find the SCSI card.
To remedy this problem, you need to add a boot-time argument such as the following:
LILO: linux qlogicfas=0x230,11,5
You can make this option permanent so that you don't have to re-enter it. See the LILO
configuration option append in the lilo.conf man page.
How do I uninstall LILO?
If you want to uninstall LILO and reinstall the original boot record, try using this command:
lilo -u /dev/hda
It represents the boot record of the first IDE drive. Parameters may vary for your machine;
for example, if your first hard drive is a SCSI drive, you use /dev/sda. You can also use the
fdisk/mbr command from DOS/Windows.
When my system boots up, I see a message that says I have unknown PCI hardware. What does this
mean?
The unknown PCI device error can occur for several reasons. The first and most harmless one
is that PCI isn't responding to Linux's queries in a way it understands, but Linux is able to
keep going. The more common occurrence is that the system hangs on, querying PCI bus
cards, and cannot get any further.
246 Chapter 10 Booting and Shutting Down
Because this is a hardware problem in the kernel, Red Hat cannot do much except point you
to the maintainer of that section of the kernel. That person might be able to let you know
what is going on and might want to look at what hardware you do have in your system so that
he or she can better handle it in the future. You can reach the maintainer at the following
address:
[email protected]
Include the following information for your exact hardware description:
/proc/pci
Try to find out which device is unknown. It may be your main board chipset, your PCI-CPU
bridge, or your PCI-ISA bridge. If you can't find the actual information in your hardware
booklet, try to read the references of the chip on the board.
While I'm booting, the machine seems to hang when Linux gets to sendmail (or some other network
program). What is happening, and what should I do?
If, after the installation, the machine seems to hang when it reaches certain processes such as
sendmail, apache, or SMB, you probably have a network problem. The most common cause is
that Linux cannot look up the name of the machine you have called the box (if you set up
networking to have a machine name). The machine is currently paused waiting for the
network timeout of DNS lookups and will eventually bring up the login prompt. When you
get the prompt, log in as root, and check the usual culprits for a problem.
If you are working directly on a network with a DNS server, make sure that the
/etc/resolv.conf file has the correct values for your machine's DNS server. Use the
following command to check:
more /etc/resolve.conf
After that, see whether you can ping the name servers using the ping command, like the
following:
ping ns.you-dns-srv.com
If you are using Linux on a network without a DNS server (or if this box is going to be the
DNS server), you need to edit the /etc/hosts file to include the hostname and IP address so
that the lookups occur correctly. The format of the /etc/hosts file is
127.0.0.1 localhost localhost.localdomain
192.168.200.1 mymachine mymachine.mynetwork.net
where the sample machine is called mymachine.
How many times has my machine been rebooted?
You can use the following command to see how many times your machine has been rebooted:
last reboot
247
Troubleshooting Startup and Shutdown
You can also use the commands w and uptime w
to see how long your box has been up. The
command shows the users' login times and the uptime command shows how long the system
has been up and running.
Part
II
Ch
10
CHAPTER
11
Managing User Accounts
In this chapter by Steve Burnett
Working with Users 250
Working with Groups 253
Managing Home Directories 254
Web-Based Administration 255
Project: Using Userconf 255
250 Chapter 11 Managing User Accounts
Working with Users
As the system's administrator, you're in charge of managing users. Managing involves adding
users so that they can log in to the system, setting user privileges, creating and assigning
home directories for users, assigning users to groups, and deleting users when necessary. In
this chapter, you will learn about the various tools and techniques that enable you to perform
user account management.
Every user should have a unique login name. Login names make it possible to identify each
user and avoid the problem of one user deleting another's files.
Each user also must have a password. About the only exception to having a password is a case
in which only one user is on a system, and the system has absolutely no connection by modem
or network to any other computer. Even then, a password for every account is strongly
?
encouraged.
See ``Dealing with Password Security,'' p277
When a person has no real reason to have access to your system, you must make sure that
individual can't log in. You should remove that person's login name, along with any files that
your remaining users no longer need.
Adding a User
When you add a user to your system, the result is an entry for the user in the password file,
/etc/passwd. That entry has the following form:
login_name:encrypted_password:user_ID:group_ID:user_information:
Alogin_directory:login_shell
In this syntax, fields are separated by colons. Table 11.1 lists the fields.
Table 11.1 Fields in an /etc/passwd File Entry
Field Description
login_name The name used to log in.
encrypted_password The password required to authenticate the user; the password is the
primary line of defense against security violations.
user_ID A unique number the operating system uses to identify the user.
group_ID A unique number or name used to identify the primary group for
this user. If a user is a member of multiple groups, he or she can
switch group membership to another group if permitted by the
administrator.
user_information A description of the user, such as the user's name or title.
login_directory The user's home directory (where the user ends up after logging in).
login_shell The shell used by a user when logging in (for example, /bin/bash if
using the bash shell).
251
Working with Users
If you are using the Shadow Security Suite, your /etc/passwd file might look like this instead:
root:x:0:0:root:/root:/bin/bash
bin:x:1:1:bin:/bin:
daemon:x:2:2:daemon:/sbin:
adm:x:3:4:adm:/var/adm:
lp:x:4:7:lp:/var/spool/lpd:
Notice that the second field (between the first and second colons) is x on every line. Shadow
removes the actual passwords from the /etc/passwd file and hides them in a separate, less
?
accessible file.
See ``Shadow Passwords: What Good Are They?'' p. 289
The adduser command enables you to add a user to your Linux system. You invoke the
command with the name of the user that you want to add. The following section provides
greater detail on this subject.
Using the adduser Command
When you add a user, you simply use the adduser command and provide the name of the user
you want to add, as shown in Listing 11.1. Part
II
Listing 11.1 An Example of an adduser Session
Ch
# ./adduser jschmoe
# 11
The adduser command copies the files whose names begin with . (dot) from the /etc/skel
directory into the user's home directory. The /etc/skel directory should contain template
files you want every user to have. These files typically include ``personal'' configuration files,
such as .profile, .cshrc, and .login for shell configuration; .mailrc for email setup; .emacs
for your users using Emacs as an editor; and so on.
The adduser command is a Bourne shell script located in the /usr/sbin directory. You can
customize this script if you need to perform some additional actions when you create a user
account. A common modification is to have adduser prompt for the user's full name rather
than hard-code a default username into the password file. If you don't change the script so
that it asks for the user's name, you have to change it by hand by using the chfn command, as
shown here:
# chfn jschmoe
Changing finger information for jschmoe.
Name [RHS Linux User]: Joseph A. Schmoe
Office []:
Office Phone []:
Home Phone []:
Finger information changed.
#
252 Chapter 11 Managing User Accounts
Tip #70 from
Steve The adduser command doesn't set the password for the account. You have to do
that by using the passwd command.
Setting User Passwords
You set a user's password by using the passwd command. You should set a password for each
user added to the system; each user can then change his or her password when logging in.
The following steps outline the basic procedure for using passwd:
1. Type the command and login name (for example, passwd jschmoe), and press Return.
2. At the New password: prompt, enter the password (you don't see the password
onscreen).
3. You're prompted to type the password again. Enter the password again:
New password (again): newpassword
The password is encrypted and put into the /etc/passwd file.
It's important that you take the time to make sure your password follows these rules:
& Passwords should be at least six (preferably eight) characters long.
& Passwords should contain both upper- and lowercase letters as well as punctuation
?
symbols and numerals.
See ``Dealing with Password Security,'' p. 277
When you're adding a number of users, you might be tempted to enter short, easy passwords.
Don't fall for this trap. Good passwords are your first line of defense against intruders. Be
sure to tell your users why you've assigned a particular type of password. Further, it's a good
idea to change passwords regularly, but remember to educate system users about the choice of
good passwords.
After a user is assigned a password, the file entry looks something like this:
jschmoe:Zoie.89&^0gW*:123:21:Joseph A. Schmoe:/users/jschmoe:/bin/bash
The second field is the passwordÐnot as it was typed, but in encrypted form.
Users occasionally forget their passwords. You cannot tell users their own passwords.
You can delete a forgotten password, however, by editing the /etc/passwd file and
deleting the second field in the user's file entry. You can then set the user's new
password by using the passwd command. You should establish a procedure for
dealing with such a situation and let your users know about it.
253
Working with Groups
Removing a User
There are several different degrees of user removal. Removing a user from the system doesn't
have to be a final, irrevocable act. The following are some possibilities:
& Remove only the capability to log inÐThis type of removal is useful if the user is
away for a while and needs to be reinstated some time in the future. The user's
directory, files, and group information are kept intact. You can edit the password file
(/etc/passwd) and put an * in the second field of the user's entry like this:
jschmoe:*:123:21:Joseph A. Schmoe:/users/jschmoe:/bin/bash
If you are running the Shadow Suite, editing the /etc/passwd file doesn't matter; you
need to edit /etc/shadow instead.
& Remove the user from the password file, but keep the user's files on the
systemÐThis type of removal is useful if the files are used by others or if a new
person will be taking over the duties of the old user. You can delete the user's entry
from the password file or files by using an editor or the userdel login_name command.
You can then change the ownership and location of the deleted user's files by using the
chown and mv commands.
& Remove the user from the password file, and remove all files the user
Part
ownsÐThis type of removal is the ultimate and complete form of deleting a user. You
II
must delete the user's entry from the password file and delete the user's files from the
system. You can do so by using thefind command, as follows: Ch
find user's-home-directory -exec rm {} \; 11
Then you can remove the directory by using rmdir user's-home-directory and
remove the appropriate entry from the password file or files.
If you use other configuration files at your site, such as email alias files, you also have
to remove the user from those files.
Working with Groups
Each user is a member of a group. You can give different types of groups different capabilities
or privileges. For example, it's reasonable to give a group of users who use the system to
analyze the company's sales data access to a different set of files than a user group whose main
function is researching new products.
The password file contains information for a single user. Information about groups is kept in
the /etc/group file. The following is a sample entry:
sales::21:tuser, jschmoe, staplr
254 Chapter 11 Managing User Accounts
In this example, the group name is sales, the group ID (GID) number is 21, and the
members are tuser, jschmoe, and staplr. Files and directories have permissions associated
with them for the owner, group, and others. A user can be a member of more than one group,
and you can change group memberships.
Adding a Group
You create a new group by editing the /etc/group file directly and entering the new group
information.
Tip #71 from
Steve Each group in the /etc/group file has a unique group ID number associated with it.
Linux pays attention to the number that you assign, not to the name. Hence, if you
assign two groups with the same number, they are treated as though they're the
same group.
Deleting a Group
You delete a group by editing the /etc/group file and removing the entry for the specific
group that you want to delete. Also, you should reassign all files that have that associated GID
to a different group. An easy way to reassign them is to use the find command, as in this
example:
find / -gid group-id find users-home-directory -exec chgrp newgroup {} \;
Managing Home Directories
You should give some thought to grouping your home directories logically if you plan to have
many users on your system. In general, you should try to place all the home directories on a
given machine under one single top-level directory. That way, you can group them according
?
to whatever arrangement makes sense for your needs.
See ``Mounting and Unmounting File Systems,'' p. 444
/home be the top-level directory for user directories. Under
For example, you can specify that
/home, you can group users by department. The sales users would have accounts under
/home/sales, development under /home/develop, and so on. Your user home directories
would then fall under these directories or under another set if additional grouping is needed.
Because user directories can use a lot of disk space, you could consider placing logical groups
of users on different physical file systems. As you need additional space, you can simply create
an additional category for home directories and mount it on a file system as a mount point
under /home.
255
Project: Using Userconf
Web-Based Administration
The Red Hat distribution of Linux includes Jacques Gelinas's system administration tool
called Linuxconf. Linuxconf enables you to manage many system administration tasks,
including working with users and groups. In addition to the familiar character-line and X
Window System access, Linuxconf supports administration of the Linux system over the
World Wide Web. If the option Linuxconf HTML Access Control is selected in Linuxconf,
you can enter the URL http://<hostname>:98/ to display the top Web page of the
Linuxconf tool.
To do anything to the system, you need to go to one of the subpages. You then are
prompted for the root password, so have it ready.
Project: Using Userconf
Although the command line is powerful, especially because you can create shell scripts to
Part
automate tasks at the command line, graphics-based tools are sometimes nice to have as well.
II
One excellent tool that provides graphic access to user configuration management is
Userconf. In this exercise, you're going to do the following: Ch
& Add a new imaginary user named Susan Jones. (If you already have a Susan Jones 11
account on your Linux system, make up someone else, of course.)
& Change her group assignment.
& Delete her account.
Adding a User with Userconf
To begin adding a user, assume you're using a default installation of Red Hat Linux. Follow
these steps:
1. In an xterm session, use su to change to root, and then enter the following to start
userconf, as shown in Figure 11.1:
userconf &
2. To add a new user, click the User Accounts button. The resulting window displays the
currently defined users (see Figure 11.2).
3. Click the Add button to display the User Account Creation window shown in Figure
11.3.
4. Enter the following list of values in the fields:
. Login name: sjones
. Full name: Susan Jones
256 Chapter 11 Managing User Accounts
Figure 11.1
The Userconf
window pro-
vides graphic
control of user
accounts and
group mem-
berships.
Figure 11.2
The Users Ac-
counts window
displays cur-
rent users and
their group
memberships.
Figure 11.3
On the User
Account Crea-
tion window,
you can define
user account
parameters.
5. For the group value, click the down-arrow to the right of the group field and choose
Users.
6. Click the Accept button to accept the defined values. The Changing Password window
then appears (see Figure 11.4).
257
Project: Using Userconf
Figure 11.4
The Changing
Password win-
dow appears
when a new
user is created
in userconf.
7. In this window, enter a password. Remember that userconf checks the password for
security, so choose a good password. The Changing Password window appears a
second time for you to retype the password. Then the Users Accounts window
reappears with the new user account (see Figure 11.5).
Figure 11.5
The Users Ac-
counts window
is updated with
a newly cre-
ated account.
Part
II
Ch
11
8. Click the Quit button to close the window, and you're done.
Adding a User with Userconf
For the next task, assume that Susan is now a member of the system administrators' team for
your network. She needs to have root group access, so you must change her group assignment
from users to root. To do so, follow these steps:
1. On the Userconf window, click the User Accounts button to display the Users
Accounts window (refer to Figure 11.5).
2. Click the line for Susan Jones's account to display the User Information window
shown in Figure 11.6.
3. Click the down-arrow to the right of the Group field and select root from the drop-
down list box. The group name root appears in the field.
4. Click the Accept button, and you're done.
Deleting a User with Userconf
For the third task of this project, assume that Susan is leaving the network and no longer
needs a user account on the network (especially with root privileges). To delete her user
account, follow these steps:
258 Chapter 11 Managing User Accounts
Figure 11.6
On the User
Information
window, you
can change in-
formation for a
selected user.
1. On the Userconf window, click the User Accounts button to display the Users
Accounts window (refer to Figure 11.5).
2. Click the line for Susan Jones's account to display the User Information window (refer
to Figure 11.6).
3. Click the Del button. A dialog appears warning you that you're about to erase files and
directories (see Figure 11.7).
Figure 11.7
The Deleting
account
prompt gives
you options for
disposing of a
user's files.
4. Accept the default option of archiving the account's files, and you're done.
CHAPTER
12
Backing Up Data
In this chapter by Jack Tackett, Jr.
Considering Backup Issues 260
Considering Backup Tips 261
Planning a Backup Schedule 262
Performing Backups and Restoring Files 263
Using taper 269
Using dump 270
260 Chapter 12 Backing Up Data
Considering Backup Issues
Various kinds of problems can result in loss of data: files get accidentally removed, hardware
fails, and important information stored in files is no longer available. Users should feel
confident that, in such cases, they can access a timely backup of the ``lost'' files.
Your company's futureÐand your future with your companyÐmay depend on making those
backup files available. At such times, you and others will be thankful that you've taken the
time and effort to copy files to some sort of storage media according to a regular, rigorous,
and well-documented schedule. Backing up files isn't very glamorous, but no administrator
can ignore the process.
The following are several issues to consider when backing up a system:
& Full or incremental backupsÐA full backup (sometimes called a complete backup)
copies every file. Is it necessary to do that every day? A full backup usually requires a
good deal of time and enough media to hold all the files on the system. An incremental
backup copies the files that have changed since the last full backup.
& File systems to back upÐNaturally, active file systems must be backed up regularly.
Others can be backed up less frequently. Make sure that you have current copies of all
the file systems.
& Types of backup mediaÐDepending on the devices on your system, you might be
able to use nine-track tape, 1/4-inch cartridge tape, 4mm or 8mm DAT tapes, or
floppy disks. Each has advantages over the other in terms of sheer bulk, storage
capacity, and cost for devices and media. Choose the backup medium to fit your
budget, remembering that the least-expensive medium may be the most time-
consuming.
& Effect of backups on usersÐPerforming a backup operation increases the load on a
system. Will that be an unreasonable burden on users? Also, files that are changed
during the backup process might not be backed up, which can merely be an
inconvenience or a very important consideration if you're backing up an active
database. Should you perform backups when the system is quiet?
& Commands to use for backupsÐSome relatively simple, time-honored commands
are available for creating backups, such as tar and cpio. Are they sufficient?
Sometimes not, so we will explore more options in the section ``Using taper.''
& Documentation of the backed-up filesÐYou must label all backed-up material so
that you can use it to recover files when necessary. Some procedures and commands
allow you to prepare a table of contents or list of the material that has been backed up.
From an administrator's point of view, the file system should be backed up according to some
automated process with as little operator intervention as possible. It should also be done
when the system is relatively quiet so that the backup is as complete as possible. This
consideration must be balanced with convenience and costs. Should an operator or
administrator have to stay until midnight on Friday to perform a full backup? Is it worth
261
Considering Backup Tips
$2,000 for a DAT tape drive so that the entire system can be backed up automatically at 3 a.m.
with no operator intervention?
Consider the alternatives, determine the true costs, and make a decision or recommend a
course of action. Restoring well-managed backup information is generally a lot cheaper and
always easier than re-creating it or doing without it.
Considering Backup Tips
The purpose of performing backups is being able to restore individual files or complete file
systems as rapidly and easily as possible. Whatever you do about backups should be focused
on that central purpose.
Set up a backup plan. Include the files to be backed up, how often they'll be backed up, and
how the files are to be restored. Let all users know the backup schedule and how they can
request restoration of files. Be sure to stick with the plan.
Be sure to verify your backups. This step could include reading a table of contents from the
backup medium after it's stored or restoring an arbitrarily chosen file from the medium.
Remember that the backup mediumÐdisk or tapeÐ possibly can have flaws.
Make backups so files can be restored anywhere on the file system or on another computer
system. Use backup or archive utilities that create archives that can be used on other Linux or
UNIX computer systems.
Be sure to label all mediaÐtapes, disks, whateverÐused in a backup. If you have to use
multiple tapes or disks, make sure that they're numbered sequentially and dated. You must be
Part
able to find the file or files you need.
II
Plan for a disaster. Make copies of the files on your system so that the entire system can be
Ch
restored in a reasonable amount of time. Store copies of backup tapes or disks offsite. The
preceding sentence is very important! You should store at least one copy of your backup 12
material offsite, away from your computers. If a disaster (such as a fire) wipes out your system,
it will more than likely also destroy your nearby backups. Many businesses rent a safe deposit
box to store their tapes and disks in. You should store a complete hardware list in the same
offsite location so that you can reorder identical parts should a disaster occur.
Plan to re-evaluate your backup procedures periodically to make sure that they're meeting
your needs.
Several tools are available to help automate your backup procedure. Check out the Linux
archives on sunsite.unc.edu for more information. Also, Linux supports the ftape
extensions. ftape lets you perform backups to QIC-80 magnetic tape units that run off a
floppy controller on your system. For detailed information, refer to the ``ftape How-To.''
262 Chapter 12 Backing Up Data
Tip #72 from
Jack Linux can take advantage of several commercial backup programs too such as BRU
and systems from Legatto. See the Commercial-HOW-TO for more information. For
information on BRU see:
http://www.estinc.com/
For information on Legatto systems, see:
http://wwwm.legato.com/
Planning a Backup Schedule
You must come up with a backup schedule that meets your needs and makes it possible to
restore recent copies of files. After you decide on a schedule, stick to it.
The ideal situation is to be able to restore any file at any time. Taken to an extreme, that's not
possible, but you should be able to restore files on a daily basis. To do so, you use a
combination of complete and incremental backups.
Incremental backups can be at different levelsÐincremental to the last complete backup or
incremental to the last incremental backup. You might think of backups as occurring at
different levels:
Level 0:Full backup
Level 1: Incremental to the last complete backup
Level 2: Incremental to the last level 1 backup
The following are some sample backup schedules:
& Full backup one day, incremental other days.
Day 1 Level 0, complete backup
Day 2 Level 1, incremental backup
Day 3 Level 1, incremental backup
Day 4 Level 1, incremental backup
Day 5 Level 1, incremental backup
If you create and save an index of each backup, you should need only one day's backup
to restore an individual file and only two days' backups (that of day 1 and another day)
to completely restore the system. For example, if your system failed on day 4, before
the Day 4 Level 1 incremental, then to fully recover you need to install the Day 1
Level 0 complete backup and the Day 3 level 1 incremental to fully recover.
& Full backup once a month, weekly incremental, and daily incremental. (This example
is built around Tuesday, but you can choose any day of the week.)
First Tuesday Level 0, complete backup
Any other Tuesday Level 1, incremental backup
Any other day Level 2, incremental backup
263
Performing Backups and Restoring Files
To restore an individual file under this schedule, you might need the complete backup
if the file wasn't changed during the month, the level 1 backup if the file was changed
the previous week but not this week, or the level 2 backup if the file was changed this
week. This schedule is more complex than the previous example, but backups take less
time per day.
You also might want to consider keeping backup files for an extended period, in case you need
to restore an older version. A common schedule is to keep one weekly copy of a full backup
for four weeks. For periods of longer than four weeks, you might consider keeping a biweekly
backup for about three months.
Tip #73 from
Jack Don't back up unnecessary files such as /tmp and /proc. /tmp is temporary space
and files there are not needed by definition. Also /proc is a 'make-believe' file
system used by Linux to keep information on various system functions, like type of
cpu and memory in use. The information on these file systems is transitory and
worthless if restoredÐso why back them up to begin with? Excluding these directories,
and directories like them, from your backup file lists speeds up backup and restore
times.
Performing Backups and Restoring Files
Several different utilities are available for backing up and restoring files in a Linux system.
Part
Some are simple and straightforward; others are more complex. The simple methods have
II
their limitations, however. You should choose the one that meets your needs.
Ch
Because backing up and restoring files are very important, a number of available software
systems are dedicated to these tasks. The following sections present four of them:
12
& tar is a tape archive utility available on every Linux or UNIX system. This easy-to-use
Linux version can use several tapes or disks.
& dump is an archive utility that can perform various levels of backups from a full backup
to various levels of incrementals.
& taper is an archive utility with a text-based interface.
& cpio is a general-purpose utility for copying files available on every UNIX system. cpio
is easy to use and more robust than tar, and it can use several tapes or disks.
After selecting a program, you must select a medium. Hard drives (files) are the easiest to use
but present a high risk. Tapes provide better risk management but are slower to access and
slower to archive data. Table 12.1 provides an overview of the various archive media.
264 Chapter 12 Backing Up Data
Table 12.1 Backup Media Pros and Cons
Medium Access Speed Reliability Comments
Hard Drives Random Fast Medium Crashing hard
drives and cor-
rupt file sys-
tems are the
primary reason
for making
backups. How-
ever, hard
drives can be
utilized for
backups on
RAID and non-
RAID systems,
but the disad-
vantages are
1) they are
nonportable
2) they cost
more to replace
than their tape
media counter-
parts
Floppies Random Slow Good Floppies are in-
secure and can-
not hold large
amounts of
data.
Tape Sequen- Fast High Tape is the sta-
tial ple of the in-
dustry.
Removable Media Random Fast Good Zip, Jazz, and
other remova-
ble media are
now supported
under Linux
and provide ex-
cellent archive
media, though
they are more
expensive than
tape.
CD-R Random Fast High CD-Rs can be
used only once,
and not all
drives are cur-
rently sup-
ported.
265
Performing Backups and Restoring Files
Table 12.1 Backup Media Pros and Cons
Medium Access Speed Reliability Comments
CD_RW Random Fast High CD-RWs are
expensive and
can be hard to
configure under
Linux.
Using tar
The UNIX tar utility was originally designed to create a tape archive (to copy files or
directories to tape and then to extract or restore files from the archive). You can use it to copy
to any device. It has the following advantages:
& It's simple to use.
& It's reliable and stable.
& Archives can be read on virtually any Linux or UNIX system.
It also has a few disadvantages:
& For some versions of tar, the archive must reside on one disk or tape, which means that
if a portion of the medium failsÐfrom a bad sector on a disk or bad block on a tape,
for exampleÐthe entire backup may be lost.
& tar can't back up special files, such as device files.
& On its own, tar can perform only complete backups. If you want to create incremental Part
?
backups, you have to do a little shell programming.
II
See ``Working with Shell Scripts,'' p. 348 Ch
Table 12.2 lists some options that are commonly used with tar. You can use many other
12
command parameters with tar; refer to the man page for a complete list.
Table 12.2 Common Options for tar
Option Description
c Creates an archive
x Extracts or restores files from the archive that's on the default device or on the
device specified by the f option
f name Creates the archive or reads the archive from name, where name is a filename or a
device specified in /dev, such as /dev/rmt0
Z Compresses or decompresses the tar archive
z Compresses or decompresses the tar archive with gzip
M Creates a multivolume tar backup
266 Chapter 12 Backing Up Data
Table 12.2 Common Options for tar
Option Description
t Creates an index of all files stored in an archive and lists on stdout
v Uses verbose mode
Consider some examples of the use of tar in backing up and restoring files. The following
command copies the directory /home to the floppy drive /dev/fd0:
tar -cf /dev/fd0 /home
In this case, the f option specifies that the archive is created on the floppy drive device
/dev/fd0.
The following command also archives the directory /home:
tar -cvfzM /dev/fd0 /home | tee homeindex
The v option indicates verbose mode, the z option indicates that the archive should be
compressed to save space, and the M option tells tar to create a multivolume backup. When
one floppy disk is full, tar prompts you for another. A list of the copied files is directed to
homeindex. It's a good idea to look at that file to see what was copied.
The find command is useful for locating files that have been modified within a certain time
period so that they can be scheduled for incremental backups. The following example uses the
command find to create a list of all files that have been modified in the last day:
find /home -mtime -1 -type f -print > bkuplst tar cvfzM /dev/fd0
Á cat bkuplstÁ | tee homeindex
To use the list as input to the tar command, place the command cat bkuplst in back quotes
(backward single quotation marks, also known as grave accentsÐ cat bkuplstÁ). They tell the
shell to execute the command as a subshell and place the output from the command on the
command line in the location of the original back-quoted command.
The following command restores the /home/dave/notes.txt file from the device /dev/fd0
(note that you have to give the complete filename to restore it):
tar -xv /usr2/dave/notes.txt
Tip #74 from
Jack You can automate any of these commands by putting them in root's crontab file. For
example, you could put the following entry in the root's crontab file to perform a
backup of /home every day at 1:30 a.m.:
30 01 * * * tar cvfz /def/fd0 /home > homeindex
If you need to do more complicated backups, you can create shell scripts to control
your backups. These shell scripts can also be run via cron.
267
Performing Backups and Restoring Files
? See ``Scheduling Commands with cron and crontab,'' p. 373
You also can use the tar command to create archive files in the Linux file system rather than
write to a backup device. This way, you can archive a group of files along with their directory
structure in one file. To do so, simply give a filename as the argument to the f option instead
of a device name. The following is an example of archiving a directory and its subdirectories
with the tar command:
tar cvf /home/backup.tar /home/dave
This command creates the file /home/backup.tar, which contains a backup of the /home/dave
directory and all files and subdirectories below /home/dave.
The tar command by itself doesn't perform any file compression. To compress the resulting
tar z option with the tar command or use a compression program,
file, either specify the
such as gzip, on the final tar file.
Tip #75 from
Jack If you do compress the tar file, then you can use the zcat command as follows to
get a listing of files stored in the archive:
Zcat backup.tar.Z | tar -tf - | more
When you use the tar command to make archive files, it's usually a good idea to try to make
the top-level entry in the tar file a directory. This way, when you extract the tar file, all the
files in it are placed under a central directory in your current working directory. Otherwise,
you could end up with hundreds of files in your directory if you extract a tar file in the wrong Part
place.
II
Suppose that below your current directory is a directory named data, which contains several Ch
hundred files. You can create a tar file of this directory in two basic ways. You can change
12
directories to the data directory and create the tar file from there, as in this example:
$ pwd
/home/dave
$ cd data
$ pwd
/home/dave/data
$ tar cvf ../data.tar *
This example creates a tar file in /home/dave that contains just the contents of data without
containing an entry for the directory. When you extract this tar file, you don't create a
directory to put the files in; you just get several hundred files in your current directory.
Another way to create the tar file is to start from data's parent directory and specify the
directory name as the thing to archive. The command sequence is as follows:
$ pwd
/home/dave
$ tar cvf data.tar data
268 Chapter 12 Backing Up Data
This example also creates an archive of the data directory, but it puts the directory entry as
the first thing in the archive. This way, when the tar file is extracted, the first thing that's
created is the directory data, and all the files in data are placed in the data subdirectory.
If you want to create a tar file of all the files in the directory, specifying a different
location for the tar file (other than the current directory) is a good idea. That way, if
you try to archive all the files in the current directory, tar doesn't get confused and try
to add its tar file recursively to the tar that it's creating.
Using cpio
cpio is a general-purpose command utility for copying file archives. You can use it to create
backups by using the -o option or to restore files by using the -i option. It takes its input
from standard input and sends its output to standard output.
The advantages of cpio include the following:
& It can back up any set of files.
& It can back up special files.
& It stores information more efficiently than tar.
& It skips bad sectors or bad blocks when restoring data.
& Its backups can be restored on almost any Linux or UNIX system.
Some people find cpio's syntax to be a bit more confusing than tar's syntax. Also, to perform
?
incremental backups, you have to do some shell programming.
See ``Working with Shell Scripts,'' p. 348
Table 12.3 lists the commonly used options for cpio. See cpio's man page for a complete
description of the options you can use with this command.
Table 12.3 Common Options for cpio
Option Description
-o Copy out. Creates an archive on standard out.
-B Blocks input or output at 5,120 bytes per record; useful for efficient storage on
magnetic tape.
-i Copy in. Extracts files from standard input. This option is typically used when the
standard input is the result of a copy out action of another cpio command.
-t Creates a table of contents of the input.
269
Using taper
The following list provides some examples of using the cpio command to back up and restore
files:
& The following command copies the files in the directory /home to the device /dev/fd0:
ls /home | cpio -o > /dev/fd0
& The following command extracts the files on the device /dev/fd0 and creates an index
in thebkup.indx file:
cpio -it < /dev/fd0 > bkup.indx
& The following example uses the find command to create a list of all files in /home that
have been modified in the last day:
find /home -mtime 1 -type f -print | cpio -oB > /dev/fd0
The output of that command is piped to cpio, which creates an archive on /dev/fd0,
where the data is stored at 5,120 bytes per record.
& The following command restores the file /home/dave/notes.txt from the device
/dev/fd0:
echo ``/home/dave/notes.txt'' | cpio -i < /dev/fd0
You must give the complete filename to restore a file with cpio.
Tip #76 from
Part
Jack You can automate any of these commands by putting them in root's crontab file. For II
example, you could put the following entry in the root's cron file to perform a daily
Ch
backup of /home at 1:30 a.m.:
30 01 * * * ls /home | cpio -o > /dev/fd0 12
If you need to do more complicated backups, you can create shell scripts to control
your backups. You also can run these shell scripts via cron.
Using taper
taper is an open source program created by Yusuf Nagree. The program comes with a text-
based GUI for backing up and restoring files. Figure 12.1 shows the main menu screen for
taper.
taper is a high-end program that makes use of other low-level programs for accessing the tape
device. Thus, you must have a device driver or other program, such as ftape, available to
transfer information to the tape.
270 Chapter 12 Backing Up Data
taper is not installed by default, so you need to install the program. An RPM package is
available for Red Hat and Caldera ( taper-6.9-2.i386.rpm); you can install it by using the
rpm -i command. You can download the code from http://www.omen.net.au/~yusuf and
then build the program for Debian. To install a Debian package use dpgk -i <package-name>.
Figure 12.1
Using taper's
interface
makes backing
up and restor-
ing much ea-
sier.
taper can fail to work if the correct libraries or ncurses program is not installed on
your system. Also, you must have a driver for your archive device so that taper can
work properly. Look at /usr/doc/Taper-6.9/TAPER.txt and FAQ.txt for more
information.
To run the program, use the following command line to display the GUI shown in
Figure 12.1:
taper -T < device>
Here, device can be any of the following:
z Floppy tape drive using zftape
f Floppy tape drive using ftape
r -b /dev/<device> A removable device, such as a Zip disk or Jazz drive
s A SCSI tape drive
If you want to use a file instead of a device, use the following form of the command:
taper -f < filename>
To get a full help listing, use the following command:
taper -?
Using dump
If you use only the ext2fs or minix file systems, you can use the dump program to back up
your system and the restore program to retrieve backup files. dump works like tar, but it
271
Using dump
allows easier management of backups, as detailed in the section ``Planning a Backup Schedule''
earlier in this chapter. The command line for dump is as follows:
/sbin/dump [0123456789BbhfusTdWn [ argument ...]] filesystem | directory
The numbers indicate the backup level, with 0 indicating a full backup and 9 (the default
value) indicating the highest incremental backup. You can specify a file or a tape device just as
you can with tar. Thus, to do a full backup to a file of the /usr directory, you would use the
following command:
/sbin/dump 0uf /backup_usr.dmp /usr
The f flag indicates that dump should create a dump file named /backup_usr.dmp in the
root directory containing an archive of /usr. The u flag instructs dump to update the file
/etc/dumpdates. Table 12.4 provides information on each flag value.
Table 12.4 Common Flags for Creating Archives
Flag Description
0 to 9 Indicates the backup levelÐ 0=full, 9=highest incremental. A level number above 0,
incremental backup, tells dump to copy all files new or modified since the last
dump of the lower level.
f Writes the backup to file; file may be a special device file like /dev/tape, /dev/hda1
(a disk drive), an ordinary file, or ``-'' (the standard output).
u Updates the file /etc/dumpdates after a successful dump.
W Tells the operator what file systems need to be dumped.
w Prints only those file systems that need to be dumped.
Part
II
Ch
12
dump is not installed by default by any of the distributions included on the CDs. To
install dump for Red Hat or Caldera, use the following rpm command:
rpm -ivh dump-0.3-14.i386.rpm
To install it for Debian use, enter the following command:
dpkg -i <filename>.i386.deb
dump uses the file /etc/dumpdates to store information about file systems. The following
listing indicates that a level 0 dump occurred on the /hda1 file system on June 20, 1999 at
8:31 p.m.:
[root@ns RPMS]# more /etc/dumpdates
/dev/hda 1 0 Sun Jun 20 20:31:09 1999
272 Chapter 12 Backing Up Data
You use the restorecommand to restore files from the dump. You can restore entire file
systems by using dump/restore. restore allows an interactive mode (-i), so you can select
exactly what files to restore from the dump archive. To interactively restore the dump file
created previously, you would use the following command:
/sbin/restore -if /backup_usr.dmp
In this case, restore would interactively prompt you for information. If necessary, you can
request help by using the help command, as shown in the following output:
/sbin/restore -if /backup_usr.dmp
restore > help
Available commands are:
ls [arg] - list directory
cd arg - change directory
pwd - print current directory
add [arg] - add CËarg' to list of files to be extracted
delete [arg] - delete CËarg' from list of files to be extracted
extract - extract requested files
setmodes - set modes of requested directories
quit - immediately exit program
what - list dump header information
verbose - toggle verbose flag (useful with ÁÁ#180;ls'')
help or `?' - print this list
If noÂarg' is supplied, the current directory is used
restore >
You need to follow this procedure to restore files interactively:
1. Add files to be extracted by using the add command.
2. Provide the tape (or filename) to the program.
3. Extract the files you have added.
See the man pages for detailed information on both commands.
Case Study: Copying Files
Backing up files to an external media is one procedure to insure data safety, but sometime you
might need to ensure that backup systems used for disaster recovery have the needed files so
you can recover from a disaster in a timely manner. Transporting tapes or removable drives
takes time. To ensure the proper distribution of essential files, you can use ftp to copy the
files, but this is cumbersome and error prone. You can use the program rdist and cron to
?
automate this process, or if the files needed are for your Web site then you can use wget.
See ``Scheduling Commands with cron and crontab,'' p. 373
rdist uses a configuration file, shown in Listing 12.1, to specify which files and directory trees
to copy and where to copy them. You can specify multiple machines so that you can distribute
an entire directory tree to various servers anywhere on the net.
273
Case Study: Copying Files
Tip #77 from
Jack If you are distributing files across an unsecure link, you should investigate secure
copy programs such as scp from datafellows. See http://www.datafellows.com
for more information.
Listing 12.1 Sample rdist Configuration File
#
# rdist configuration file to sync authdb files on other web servers
#
#
# list of hosts to replicate files to
#
HOSTS = (root@machine1 root@machine2)
#
# list of files and/or directories to replicate
#
FILES = (/opt/netscape/nse-home/authdb )
#
# and now the mapping(s)
#
# copy files that are new and remove any extraneous files
#
(${FILES}) -> (${HOSTS})
install /opt/netscape/nse-home/authdb;
notify root@mmail;
Any line beginning with a # is a comment and ignored by rdist. The HOSTS directive specifies Part
which machines to copy the files to and also which user account to use. The FILES directive
II
specifies which files to copy. You can specify an entire directory tree or individual files by
seperating each entry with spaces. The line (${FILES}) -> (${HOSTS}) tells rdist to copy the Ch
items specified in FILES to the machines specified in HOSTS and then to perform the following 12
actions on those files. The install keyword tells rdist where on the remote machines to place
the files. Next rdist sends an email notification to the specified email account. Notice that
each command terminates with a semicolon.
After you create the config file, you invoke rdist by using the following syntax:
rdist -f configfile
The final configuration to make is to create a .rhosts file on the remote machines in the
home directory of the specified users (root in the above sample listing) and place the a plus
sign (+) and appropriate username in the file, such as the following:
+root
Using wget
wget provides an even simpler mechanism to copy Web-based files from remote machines.
Many popular Web sites do not run on a single system; instead, the same content is served
274 Chapter 12 Backing Up Data
from various Web servers around the country. The sites use techniques such as DNS round robin,
special software like resonate, or special hardware such as a Cisco Local Director to point inbound
Web requests to the proper server.
The following is the general syntax of wget:
wget options url-list
A simple request to download a Web page is as follows:
wget http://www.usinglinux.com
This downloads the index.html file from the usinglinux.com Web site into the current
directory. Table 12.5 provides a useful subset of options available to wget. For more details,
see the man page.
Table 12.5 Usefule wget options
Option Description
-o filename Specifies wget to copy the files from the
indicated Web site to filename.
-http-user username,-http-passwd password Allows wget to access Web sites that are
password protected.
-a filename Any output messages generated by wget
goes to the indicated filename rather than
to the screen.
-h Displays help information.
-v Provides verbose output messages.
-q Quiet mode in which output messages are
suppressed.
-i filename URLs are read from the indicated filename
rather than from the command line.
-r Recursive option that starts at the indi-
cated URL and then downloads all files
from that point. A great way to copy an
entire Web site.
A final warningÐwget is not a tool to be used to steal Web content! Nearly all Web sites and
the files on them, including image files, are copyrighted!
CHAPTER
13
Improving System Security
In this chapter by Steve Burnett
Handling Physical Security 276
Dealing with Password Security 277
Developing Login Security 278
Handling File Security 281
Avoiding Social Engineering Threats 282
Recording Use of the su Command 283
Developing a Secure System 284
PAM: The Pluggable Authentication Modules Architecture 287
Shadow Passwords: What Good Are They? 289
Project: Establishing Security Procedures 292
276 Chapter 13 Improving System Security
Handling Physical Security
With all the mass media hype about viruses, computer break-ins, and diabolical computer
crackers with their modems and network connections, too little attention is paid to the
physical security of computer systems. Computer equipment is fairly sensitive to various
environmental conditions.
Over the past several years, the mass media has changed the meaning of the word
hacker from ``a computer enthusiast'' to ``someone who breaks into computers.'' In
the computer community, the commonly accepted term for someone who breaks into
computers is cracker. This is the term used throughout this chapter.
Fire and smoke can obviously mean a quick end for your computing equipment. If you have
any sort of business computer installation, you should consider installing smoke detectors,
automatic fire extinguishers, and a fire alarm system.
In addition to fire and smoke, dust can wreak havoc with computer equipment. Dust is
abrasive and can shorten the life of magnetic media and tape and optical drives. Dust can
collect in ventilation systems and block the airflow, letting computers overheat. Also, dust can
be electrically conductive and can cause circuit boards to short out and fail.
Electricity poses a special threat to computer equipment. Computers are very sensitive to
surges in electrical current. All computer equipment, including modems connected to
telephone lines, should be connected to surge suppression equipment to reduce the chances
of damage. Many areas suffer from ``dirty power'' that fluctuates in current and voltage.
Although surge suppressors can help protect against spikes in the electrical current,
they're virtually worthless against any kind of lightning strike. If lightning hits an
incoming line to your house or business, simple surge suppressors are unlikely to
save your equipment. In the case of a severe thunderstorm, you would be wise to
unplug your surge suppressor and wait out the storm.
Computers are also common targets for theft. Many computer components are small and
expensive. As a result, they're easily stolen and sold. You should evaluate how secure your
computers are and try to protect them against theft as you would any valuable possession.
Another aspect of physical computer security is preventing access by unauthorized persons. If
someone can walk into your computer room, sit down at a console, and start working
unchallenged, you have a problem. By controlling access to your computers, you make it
more difficult for someone to steal or damage data or equipment. Establish access policies for
your computing facilities and educate your users as to these policies.
277
Dealing with Password Security
The following are some tips you can follow to improve the physical security at your
installation:
& Don't leave a system, tape drives, disk drives, terminals, or workstations unattended
for a prolonged period of time. It's a good idea to have some restrictions regarding
access to the rooms that house your primary system and associated tape and disk
drives. A lock on the door goes a long way in providing security. An unauthorized
person can remove backup mediaÐdisks or tapesÐfrom an unlocked area.
& Don't leave the system console or other terminal device logged in as root and
unattended. If users know the system, they can easily give themselves root privileges,
modify important software, or remove information from the system.
& Educate system users about physical security risks. Encourage them to report any
unauthorized activity they may witness. Feel free to courteously challenge someone
you don't recognize who is using the system.
& If possible, don't keep sensitive information on systems that have modem or network
connections.
& Keep backups in a secure area and limit access to that area.
Dealing with Password Security
The first line of defense against unauthorized access to a system is password protection. This
type of protection is also often the weakest link in the chain. This section describes some
steps you can take to keep passwords secure.
The reality is that users want simple, easy-to-remember passwords. They don't want to
change their passwords. They like to write down their passwords so that they can reference
them. Unfortunately for you, the system administrator, these approaches are all bad from a
computer security standpoint. Password security requires almost constant attention.
The root password is special. Anyone who knows it can access anything on your system and
Part
perhaps other systems that your computer is connected to through a network. Change the
root password often, choose it wisely, and keep it secure. It's best committed to memory. In II
most organizations, it's a good idea for two people to know the passwordÐbut no more than
Ch
that!
13
A password should be at least six characters long; however, only the first eight characters in
any password are recognized. This means that your password is truncated to eight characters
if you enter one that's longer than eight characters.
Tip #78 from
Steve The MD5 module allows passwords to be as long as 256 characters.
278 Chapter 13 Improving System Security
Writing a program that can attempt to guess a password is not too difficult. If the password-
guessing program tries to guess a random password, it will take longer to be successful if the
password itself is longer.
Computers are very good at doing the same thing over and over, such as encrypting every
word in a dictionary and comparing it to your password to try to break into your system. You
should never choose a password that's a dictionary wordÐin any language. Also, try not to
choose a password that's easily associated with you. Your name, address, spouse's name,
child's name, pet names, phone number, driver's license number, and so on are all obvious
targets for a cracker.
So how do you pick a good password if all the easy ones are also easy to guess? One technique
is to pick two random short words and connect them with a punctuation character. This
technique makes an almost random sequence of characters as far as a password guesser is
concerned, but the password still is fairly easy for you to remember. The following are a few
examples of passwords that use this technique:
joe&day
car!pan
modem!at
Another method for picking passwords is to take a phrase that you'll remember and use the
first letter from each word for the password. This technique results in a random sequence of
characters, but one that you can easily recall. For example, the phrase ``Ladies and
Gentlemen, Elvis has left the building'' translates into the password L&GEhltb.
The crucial point is that the password should be remembered. It shouldn't be written down
anywhere. If your users feel they must write down their passwords at all, give them a tip to
disguise their passwords in some type of list or sentence. For instance, if your password is
modem!at, a note on a small piece of paper saying ``Don't forget to pick up modem! At
computer shop for repairs'' looks like an ordinary reminder in case another person sees the
paper, but the password is disguised.
Developing Login Security
Each account on your Linux system is a door into your computer. All someone needs is the
right keyÐthe password. If you've instituted good password-management practices, you
already have a head start toward developing a more secure system. One aspect of computer
security that goes hand in hand with password security is login or account security.
Login or account security involves looking for accounts on your system that may be potential
security problems and dealing with them. Login security can pose several different kinds of
problems.
279
Developing Login Security
Accounts Without Passwords
Many computer crackers succeed in breaking into a computer by simply finding an account
that doesn't have a password. You should check your password file regularly for such accounts
and disable them. The password is stored in the second field of the password file under Linux.
You can check for a blank password field with several tools, such as grep, awk, or perl. You
can disable logins to an account by editing the password file and changing the password field
to an * (asterisk) character. Changing this field prevents anyone from logging in with that
?
login ID.
See ``Setting User Passwords,'' p. 252
Unused Accounts
If a login name won't be used anymore, you should delete the account so that it can't be
compromised. At the very least, you should edit the password file and set the password to the
* character, which prevents anyone from logging in to the account. If you choose to delete the
account, you should use the find command to locate all files owned by the account and then
?
change their ownership or delete them.
See ``Removing a User,'' p. 253
If you use other configuration files, such as system mail alias lists, you have to remove
the account from those files as well.
Default Accounts
Linux comes with several standard login IDs, which are required for the operating system to
work correctly. For example, the root account may have no password when Linux is first
installed. You should check the password file after you've finished your installations to make
sure that all your default accounts have good passwords or that they have been disabled by Part
setting the password field to a * character. If you are using the Shadow Password Suite, all the II
password fields in /etc/password are set to * by default.
Ch
Some software packages automatically create accounts on your system during their
13
installation processes. Remember to disable them or set their passwords accordingly.
Guest Accounts
It's not uncommon for computer centers to provide some type of guest access accounts for
visitors so that they can use the local computers temporarily. These accounts usually don't
have passwords, or they have passwords that are the same as the login IDs. For example, the
login guest might not have a password or might have a password of guest. As you might guess,
these accounts are security disasters waiting to happen.
280 Chapter 13 Improving System Security
Because these accounts and passwords are probably widely known, an intruder could use one
to gain initial access to your system. When a cracker breaks into your system, the intruder can
then try to get root access from the inside or use your system as a starting point from which to
attack other computers over a network. Tracing an attack back to an open public account
makes it much harder to find the true source of the attack.
Using guest or open accounts really isn't a good idea on any system. If you really must use
one of these accounts, keep it disabled until it's needed. Randomly generate a password for
the account when it needs to be used and, when you can, disable it immediately. Also,
remember not to send the password via email.
Command Accounts
Computers commonly have several command accountsÐlogin IDs that run given commands
and then exit. For example, finger is an account that has no password. When a user logs in as
finger, the finger program is run, showing who is on the system, and then the session
terminates. Other such accounts may be sync and date, which typically don't have passwords.
Even though they don't run a shell, and they do run only one command, they can still pose a
security risk.
If you allow command accounts on your system, you should ensure that none of these
commands accept command-line input. Also, these commands shouldn't have any type of
shell escape that can allow a user to get to an interactive shell.
A second reason for not using these types of accounts is that they can give away information
about your system that can be useful to an intruder. Using programs such as finger or who as
command accounts can allow intruders to get the login IDs of users on your system.
Remember that the login ID/password combination protects your accounts. If an intruder
gets the login ID of a user, that person now has half the information needed to log in to that
account.
Group Accounts
A group account is an account for which more than one person knows the password and logs in
under the same ID. Using group accounts is almost always a bad idea. If you have an account
shared by several people that is broken into and is being used as a base to attack other
computers, finding the person who gave out the password is difficult. If you have an account
that's shared by 5 people, it may, in fact, be shared by 25 people. You have no way of
?
knowing.
See ``Working with Groups,'' p. 253
Linux allows you to provide file access based on group membership. This way, a group of
people who need access to a set of files can share them without needing to share an account.
Rather than create group accounts, you should make wise use of groups under Linux. Stay
with the ``one login ID, one person'' philosophy.
281
Handling File Security
Handling File Security
The file system under Linux is a tree structure that's built from files and directories. Linux
stores several types of information about each file in its file system, including the following:
& The filename
& The file type
& The file size
& The file's physical location on disk
& Various access and modification times
& The owner and group ID of the file
& The access permissions associated with the file
If a user can modify some of the file information on certain files, security breaches can occur.
As a result, the file system plays a very important role in system security.
Permissions
Linux file permissions control which users can access which files and commands. These
permission bits control access rights for the owner, the associated group members, and other
users. By using the ls -l command, you can generate a file list that shows the permissions
field. The leftmost field shown by ls -l specifies the file permissions. For example, this field
may look like -rw-r--r--. The first - in the field shows the file type. For regular files, this
field is always -.
The next nine characters represent the file access permissions for the owner, group, and
world, respectively. Each category takes up three characters in the permissions field,
consisting of the characters r (for read permission), w (for write permission), and x (for
execute permission). Any or all of these characters may be present.
If one of the permissions has been granted, the corresponding character is present. If
permission isn't granted, a - appears instead. For example, if a file has a permission field that Part
looks like -rw-r--r--, -), the
it indicates that the file is a regular file (the first character is
II
owner has permissions rw- (which means read and write but no execute), and the other group
members and the world at large have permissions r-- (which means read permission but no Ch
?
write or execute access). File permissions are changed via the chmod command.
13
See ``File Permissions,'' p. 414
You can specify the permissions to thechmod command as octal values instead of the
rwx symbolic values. To do so, simply treat the three characters in a permission field
as bits in an octal number; if the character is present, count it as a 1. So, the
permissions -rw-r--r-- are represented numerically as 644.
282 Chapter 13 Improving System Security
SUID and SGID Programs
Two additional permission bits are associated with a file: the SUID and SGID bits. SUID
stands for Set User ID, and SGID is Set Group ID. Programs with these permissions behave as
though they were owned by different UIDs when they're run. When an SUID program is
run, its effective UID is set the same as the user who owns the program on the file system,
regardless of who is actually running the program. SGID is similar except that it changes the
group ID instead.
Although the SUID/SGID feature can be useful, it can present a big security hole. SUID
programs are generally used when a program needs special permissions, such as root
permission, to run.
Programmers usually go to great lengths to ensure that their SUID programs are secure.
Most security holes in SUID programs occur when the program executes a command line,
activates a shell, or runs a file that users can change to contain their own commands.
Although some SUID programs are necessary, you should try to keep them to a minimum.
You should also regularly scan your file systems to check for new SUID programs by using
the find command (refer to the man page for the exact syntax).
Avoiding Social Engineering Threats
With all the different security features available on a Linux system, the biggest security hole is
typically your users. After all, your users already have valid accounts.
But what do your users have to do with social engineering? What is social engineering,
anyway? Social engineering is about convincing people to do what you want, either by playing
on their assumptions or behavior, or by outright misrepresentation and lying. People, in
general, want to be helpful. And, if given the opportunity, they usually try to help out as much
as possible. Crackers with good social engineering skills play on this characteristic.
Assume that you have a computer user named Mr. Jones. He's just your average userÐnot a
guru at all. One day, Mr. Jones gets a call at the office that goes something like this:
Mr. Jones: Hello?
Caller: Hello, Mr. Jones. This is Fred Smith in tech support.
Due to some disk space constraints, we're going to be
moving some user home directories to another disk at
5:30 this evening. Your account will be part of this move
and will be temporarily unavailable.
Mr. Jones: Uh, okay. I'll be home by then, anyway.
Caller: Good. Be sure to log out before you go. I just need to
check a couple of things. What was your login ID
againÐjones?
Recording Use of the su Command
283
Mr. Jones: Yes, it's jones. None of my files will get lost during the
move, will they?
Caller: No, sir. But I'll check your account just to make sure.
What was the password on that account so I can get in
to check your files?
Mr. Jones: My password is tuesday.
Caller: Okay, Mr. Jones. Thanks for your help. I'll be sure to
check your account and verify that all the files are there.
Mr. Jones: Thank you. Bye.
So what just happened here? Someone called one of your users on the phone and managed to
get both a valid username and password in the course of the conversation. And you guessed
itÐif Mr. Jones calls tech support tomorrow, he'll probably find that no Fred Smith is
working there!
How do you prevent situations like this from happening? Educate your users. Your users
should never give out passwords over the phone to a caller. The preceding conversation
played on Mr. Jones's assumption that ``Oh, this caller is in tech support, so I trust him, and
he can look up my password if he needs it.'' If the caller has a legitimate reason to have your
password, he or she should not get it from you. You should never leave passwords on email or
voice mail either. Crackers use social engineering by convincing users to give them what they
want; these intruders don't even have to try to break into your system.
Recording Use of the su Command
Linux verifies your identity by your login ID/password combination. As you log in, your
process is tagged with an ID that identifies you to the system. It's this UID that's checked for
file and directory access.
Linux offers the capability to switch to another UID while you're working. When users use
the su command, they can become root or another user. They must know the password of the Part
user that they're changing to. For example, for a user to change user ID to that of user ernie,
II
the command is
Ch
su ernie
13
The user is then prompted for the password associated with the login ID ernie.
To change to root, the command is as follows:
su root
The user is then prompted for the root password.
Typically, all attempts at using su are automatically logged in a system log file, such as /var/
adm/syslog. Examine this file periodically to check on this sort of activity.
284 Chapter 13 Improving System Security
Developing a Secure System
Along with power comes responsibility. If not handled carefully, Linux's power to share
information, process resources, and handle peripherals can leave your system open to abuse.
Your job is to set up system security so that only the right users and systems can connect to
yours, and that they can use only the parts of your computer you want to share.
Security Threats
You can monitor your system for security threats. To determine who is using your system and
the type of work they're doing, you can use the ps command.
Be wary of jobs that seem to be running a very long time or users who seem to be using more
resources than normal. They can be an indication that a login has been compromised and an
unauthorized user is running a program to guess passwords.
Controlling the Root
The root login is reserved for your administrators. The person who logs in as the root has the
power to erase any file, restrict use by any person on the network, and quite literally cause
havoc among users. That's the downside of the picture. Linux was designed to give the people
having root access the tools to do their jobs better than in other environments.
Many proprietary operating systems have blockages established by the creators to avoid
accidental damage to files and other operating factors on the system. The creators of UNIX
and Linux took a different attitude toward the administrator. You'll find tools that permit you
to connect almost any computer device. You'll find software that monitors the performance
of the computer. You can create an endless array of software and adapt it to just about any
business environment.
Also, you can force your users to do only specified things on the computer, or you can give
them limited rights until they grow in their knowledge. The root user, the administrator, has
the power to do these things.
Because access to the root is so important, some companies restrict use to a select
few. Even if your Linux system isn't being used for business, you should still keep the
root password limited.
Even if the Linux system is your own, used by no one else, you should create at least
one other account, log in as that account, and switch to root using su or log in as root
only when you need to do something the normal user account isn't allowed to do,
and then log off or exit the root account as soon as you're done.
285
Developing a Secure System
Controlling Modems and Crackers
Allowing access from a common modem, similar to those that people have at home, can
permit someone to ``crack'' the system and destroy important data. As a result, many
companies insist that their computers have elaborate security mechanisms, which can make
these computers almost impossible to work with. Some companies put a dial-back option on
the computers so that you must dial the computer and then wait for a return call before you
can interact with the system.
Most of the time, a traditional UNIX/Linux approach is recommended. Make sure that all
your user logins have passwords. Restrict the systems that can connect to your system. Keep
permissions closed on sensitive files. Be careful of set UID bit programs (those that give the
user who runs the program the permissions to run as another user). Most break-ins occur
because someone left the door open.
Ultimately, security is a problem with people rather than systems. You can't allow
passwords to be etched in the wall near a terminal or have DOS computers with root
passwords embedded in communication programs.
Preventing Idle Terminals
Users should log out or use some kind of terminal lock program when they leave at the end of
the day. Most UNIX systems have such a program that shuts down terminals left on beyond a
prescribed length of time.
Enforcing Security
Security in defense firms is clearly understood. Companies that have highly sensitive products
in the design cycle understand the need for security. But employees who work for a small
distributor of plumbing parts, for example, may have a hard time understanding what
Part
everybody is so concerned about. Security in this example isn't an issue until you can't figure
out who removed a file that included a key proposal. II
Ch
You should give employees a quick lesson about the sensitivity of data on your computer. A
business has a significant investment in the data on the computer. Loss of data can be a 13
distraction, or it can mean chaos. Employees who are unwilling to participate in securing
a computer should understand that this can be cause for dismissal.
For an administrator, the task becomes apparent. If you're the chief security officer for the
network, how can you be sure that files and directories are adequately secured? Fortunately,
many tools are available to help you, such as umask, cron, and Linux itself.
Permissions seem to be a significant source of worry for most administrators. New
administrators typically tighten up permissions and then field calls from people saying they
can't gain access to a file they need or can't execute a program on the system. After a while,
286 Chapter 13 Improving System Security
these administrators loosen up the permissions so that anybody can do anything. The balancing act
of securing the computer while permitting the proper people the tools to do their jobs is
sometimes frustrating.
Handling Security Breaches
Security on a computer can require a little detective work. For example, look at the following:
# who -u
root tty02 Jan 7 08:35 old Ofc #2
martha ttym1d Jan 7 13:20 . Payroll #1
ted ttyp0 Jan 7 08:36 8:25 Warehouse
margo ttyp2 Jan 7 07:05 9:45 CEO Ofc
root ttyp4 Jan 7 08:36 . Modem #1
# date
Tue Jan 7 19:18:21 CST 1997
Suppose you know that Martha left the office at 5:00 p.m. Has someone found her password,
or did she leave the terminal on when she left? You can see that she logged in at 13:20 today.
It's now 19:18, and somebody is active on the system using her login. Do you dispatch
security?
But what do you do if someone does break into your system? First, you should try to
determine whether you really do have an intruder. Many times, what you notice may be just
the result of human error. If you do have an intruder, you have several options. You need to
decide whether any damage was done and determine the extent of the damage. Do you
prosecute those responsible if you can catch them? If so, you should start trying to gather and
protect evidence.
You must decide how to go about securing your system and restoring any damage from your
backups. Probably the most important task of all is to document what you do. Start a log
immediately. Sign and date any printouts showing evidence of intrusion; they may be useful
as evidence. Your log may be invaluable in helping you figure out what you've done when you
have to change or restore files.
Two other preventive measures that you should take are to make printouts of your basic
system configuration files, such as /etc/fstab, and establish a site security policy. You must
make sure that your users are aware of your site policy and that they're reminded frequently.
Another area of concern occurs when an employee leaves the company. When an employee
leaves, for whatever reason, personnel should contact the computer staff to retire the login.
With all the different security considerations, how much security is enough? Can you have
too much? You might be surprised to learn that, yes, you can have too much security. In
general, if the cost of recovering from a security breach is less than the cost of security, you
should reduce the security level for your systems. Note that these cost factors include much
more than monetary costs. Among other things, you should take into consideration the
content of your files, the amount of time and money required take to replace them, any lost
productivity time that an attack would produce, and the effect that publicity of a computer
security problem will have on your organization.
287
PAM: The Pluggable Authentication Modules Architecture
Performing Backups
Few issues that the typical Linux administrator deals with are as important as backing up or
archiving a system. An administrator can be fired or a company can literally fail because of the
loss of valuable data. If you use a Linux system for personal use, what's your time worth to
you if you've, for example, personally typed your entire cookbook's worth of favorite recipes
over the last two months and the system crashes, wiping out all your files? The disk or disks
on a computer are electromechanical devices, and they will fail at some time. If you're ever
worried about a hostile attack on your system, backups can at least let you rebuild the system.
The data on the system is typically much more valuable to the users than the system itself.
Most new hard disks are rated at around 150,000 hours mean time between failuresÐmore
than five years. But the mean-time statistic can be deceptive. Your disk could fail at the 50,000
hour mark, or it might last for more than 10 years (highly unlikely). You're gambling if you
back up your systems only occasionally, and you take an even greater chance if you aren't
?
checking your backup tapes regularly.
See ``Planning a Backup Schedule,'' p. 262
PAM: The Pluggable Authentication Modules
Architecture
Users need to be able to perform the tasks they want, even if their desired goal is winning that
game of Solitaire. To do so, users affect the system and its contents in varying degrees. In
general, users should be able to run applications and create, change, and delete files that do
not affect the system's continued performance or that do not change items belonging to
another user that that user has not decided to share. One way of assigning authority over a
system is based on login name and password combination: When you log in, the system asks
you for a name and password. Based on the proof that you are who you say you are, the
system allows you to do essentially anything you want to your own area of the system and
restricts you if you try to affect a part of the system you're not supposed to.
Part
Other methods exist for verifying a user's identity besides the name/password combination.
II
The Pluggable Authentication Modules (PAM) architecture allows you to change
Ch
authentication policy without having to change the applications themselves. This section
presents the structure and relationship of the PAM architecture. 13
The four types of PAM modules are as follows:
& Auth performs the authentication activity.
& Account defines whether the authentication is allowed. For example, consider a user
who's only supposed to be on the system during the daytime and not work evenings or
weekends. An account module would detect the user if he or she attempted to perform
an action in the middle of the night.
& Password sets passwords.
288 Chapter 13 Improving System Security
& Session provides services to the user after the account module allows the authentication
module to verify the user's identity.
Modules can be stacked in sequence to allow multiple methods of access or to restrict access by
requiring success of multiple methods.
Understanding PAM Configuration Files
The configuration files for PAM are located in the directory /etc/pam.d/.
In older Linux systems, the file /etc/pam.conf provided configuration definitions.
/etc/pam.conf is still supported for backward compatibility, but its use is
discouraged.
The best way to understand the syntax is to examine a configuration file. The PAM file for
passwd appears as follows. If you installed PAM as part of your Linux installation, this is the
default file /etc/pam.d/passwd:
#%PAM-1.0
auth required /lib/security/pam_pwdb.so shadow nullok
account required /lib/security/pam_pwdb.so
password required /lib/security/pam_cracklib.so retry=3
password required /lib/security/pam_pwdb.so use_authtok nullok
The first line is a comment, indicated by the octothorp ( # sign) at the beginning of the line. The
second line causes the user to be prompted to enter a password and for that password to be
checked. The third line does the same if shadow passwords aren't being used (you'll learn
more details on shadowing later). The fourth line calls a password-cracking application to see
whether the new password is a good one, and the fifth line specifies which module should be
used to change the password.
Required, Requisite, and Optional: Module Order and Necessity
In the preceding section, you can see that all four of the called modules are marked as
``required.'' Labeling a module as required means that the module is called regardless of the
success or failure of earlier modules. As a security guideline, all of them are called, so the
reply from a failure at any point looks the same. If you hide the location of the failure, a
malicious attacker's task is made harder.
If every module is required, the order of the modules is unimportant. However, PAM allows
for these other control flags to be used instead of required:
& OptionalÐOptional is entirely secondary to all other modules; the success or failure
of an optional module does not affect the success of the authentication if another
module appears in the PAM configuration file. If an optional module is the only one
defined for authentication, its success or failure determines the success or failure of the
authentication itself.
289
Shadow Passwords: What Good Are They?
& SufficientÐA sufficient module acts like an optional module, except that it overrides
any or all optional modules. A required or requisite module's response supersedes a
sufficient module, however.
& RequisiteÐIf a requisite module fails, control is directly returned to the application.
If you want a PAM stack to stop at a particular module, you can edit the configuration
file and change the control flag from required to requisite.
If you want more information, Red Hat Software provides documentation for PAM on its
Web site at http://www.redhat.com/linux-info/pam/.
Shadow Passwords: What Good Are They?
On a Linux system without the Shadow Suite installed, user information (including
passwords) is stored in the /etc/passwd file. The password is stored in an encoded format;
although the password looks like gibberish to a human, it is simply encoded with the UNIX
crypt command, with the text set to [null] and the password used as the key.
It is difficult but possible to take a given encoded password and re-create the original
password. However, because people may get lazy sometimes, on any system with more than a
few users, some of the passwords are likely to be common words or simple variations. It's
quite possible, and within the means of many, to encrypt a dictionary list and compare it to
the password list in /etc/passwd. Other attacks are possible and used often, but this brute-
force approach is simple and easy to do. In addition to passwords, the /etc/passwd file also
contains information such as user IDs and group IDs that are read by many system programs,
so the /etc/passwd file must remain world readable.
Shadow passwording moves the passwords to another file, usually /etc/shadow, which is set
to be readable only by root. Moving the passwords to the /etc/shadow file prevents an attacker
from having access to the encoded passwords with which to perform a dictionary attack.
The Shadow Suite is included with most of the standard distributions of Linux.
However, in some cases such as the following, installing the Shadow Suite is not a good idea: Part
& The system does not contain user accounts. II
& The system is running on a LAN and uses Network Information Services (NIS) to get Ch
or supply usernames and passwords to other machines on the network.
13
& The system is used by terminal servers to verify users via Network File System (NFS),
NIS, or some other method.
& The system runs other software that validates users, and no shadow version is available,
and you don't have the source code.
The /etc/password and /etc/shadow Files
A nonshadowed /etc/passwd file has the following format:
username:passwd:UID:GID:full_name:directory:shell
290 Chapter 13 Improving System Security
Now consider this example:
username:Npje044eh3mx8e:507:200:Full Name:/home/username:/bin/csh
A shadowed /etc/passwd file would instead contain the following:
username:x:507:100:Full Name:/home/username:/bin/csh
The x in the second field in this case is now a placeholder for the real passwords stored in the
shadow file /etc/shadow. The /etc/shadow file has the following format:
username:passwd:last:may:must:warn:expire:disable:reserved
Table 13.1 outlines the fields in the /etc/shadow file.
Table 13.1 Fields in an /etc/shadow File Entry
Field Description
username The name used to log in
password The encoded password
last Days since January 1, 1970, that the password was last changed
may Days before the password may be changed
must Days after which the password must be changed
warn Days before the password is to expire that the user is warned
expire Days after the password expires that the account is disabled
disable Days since January 1, 1970, that the account is disabled
reserved A reserved field
Adding, Changing, and Deleting Users with Shadowed Passwords
The Shadow Suite adds the following command-line-oriented commands for adding,
modifying, and deleting users: useradd, usermod, and userdel.
useradd
You use the useradd command to add users to the system. You also invoke this command to
change the default settings.
The first thing that you should do is examine the default settings and make changes specific to
your system by using the following command:
useradd -D
usermod
You use the usermod utility to modify the information on a user; this utility is very similar to
the useradd program. For example,
291
Shadow Passwords: What Good Are They?
usermod -g 200 username
changes the specified user from their initial group to group id 200.
userdel
Using userdel, you can delete the user's account with this command:
userdel -r username
The -r deletes all files in the user's home directory to be removed, along with the home
directory itself. A less drastic way to eliminate a user from the system is to use the passwd
command to lock the user's account.
passwd
In addition to setting and changing passwords, the root user can use the passwd command to
perform the following tasks:
& Lock and unlock accounts (with the -l and -u options)
& Set the maximum number of days that a password remains valid ( -x)
& Set the minimum number of days between password changes ( -n)
& Set the number of days of warning that a password is about to expire ( -w)
& Set the number of days after the password expires before the account is locked ( -i)
pwck
Using the program pwck, you can check on the consistency of the /etc/passwd and
/etc/shadow files. This program checks each username and verifies that each entry
has the following:
& Correct number of fields
& Unique username
& Valid user and group identifier
&
Part
Valid primary group
II
& Valid home directory
& Valid login shell
Ch
13
Finally, pwck also warns of any account that has no password.
Running pwck after installing the Shadow Suite is a good idea. Running this program
periodicallyÐperhaps weekly or monthlyÐis also a good idea. If you use the -r
option, you can use cron to run the program on a regular basis and have the report
mailed to you.
292 Chapter 13 Improving System Security
grpck
grpck is the consistency-checking program for the /etc/group and /etc/gshadow files. It
checks for the correct number of fields, unique group names, and a valid list of members and
administrators.
Again, the -r option generates an automated report, so you can use cron to trigger this check
automatically.
Enabling Dial-Up Passwords
If you want to limit who can dial in and connect, you can use dial-up passwords to control
who accesses the systems remotely. To enable the use of dial-up passwords, you must
examine the file /etc/login.defs and see that DIALUPS_CHECK_ENAB is set to Yes.
Two files contain the dial-up information:
& /etc/dialups contains the ttys (one per line, with the leading /dev/ removed). If a tty
is listed, dial-up checks are performed.
& /etc/d_passwd contains the fully qualified pathname of a shell, followed by an
optional password.
If a user logs in to a line that is listed in /etc/dialups, and that user's shell is listed in the file
/etc/d_passwd, he or she is allowed access only after entering the correct dial-up password.
The command dpasswd assigns passwords to the shells in the /etc/d_passwd file.
Project: Establishing Security Procedures
Let's look at some procedures to better secure your Linux system. You cannot make your
system perfectly secure, but you can make it more difficult for the lazy, bored, or less-
committed to hurt you. Securing a system is like buying security for a house or office:
Deadbolt locks are fairly cheap but harder to break than standard door latches. After
deadbolts, people add alarm systems, automatic calls to law enforcement, barrier doors,
watchdogs, and human guards, depending on how concerned.
First, the following are two good rules for system security:
& Know what you are allowing. Leaving a default password on a default account is
exactly like forgetting to lock the side door of your house or office: Anyone who
bothers to look can walk in. At the least, make these unwanted visitors bring a crowbar
or lockpick, metaphorically speaking.
& Everything not explicitly permitted is denied. ``Do as thou wilt is the whole of the
law'' is a bad guideline for system security.
We'll start from the beginning. Before you order anything, plan (you can use paper for this if
you wantÐyes, paper). What do you want to do from this system? Will you be the only user?
If not, who else do you want to give access to? What services do you want to supply? Do you
293
Project: Establishing Security Procedures
want to let people log in from a remote computer? After you've decided what you want to do with
the system, you can order the hardware and Linux distribution you want.
When you're installing, most distributions of Linux prompt you to create a root account and
password, and some also prompt you to create a second account for general use. If Linux
doesn't prompt you to create a second account during installation, do so after you finish the
installation.
Remember to pick good passwords.
The following are some tips about installation:
& Install only what you want. Installing a Linux system requires a little work for a
decently secure setup. Many Linux systems install a whole raft of services and
applications by default. Remember the list you made earlier of the services you want to
offer on this system? Have it with you during the installation. If you don't want a
service installed, such as a Telnet server daemon, don't install it at all. You can always
go back and install it later if you change your mind.
& If you install a service, turn it off until you want it. Some distributions of Linux
automatically enable a server if you install it; others allow you to install a service
without enabling it. This point is important if it's new to you. You should set up your
individual services knowing that you're doing so, changing default passwords as you go
along, and so on.
& After installation (did you create a second user account for general use yet?),
turn off every service you don't want. Remember the first rule: If you don't
specifically want it, you don't need it. Obviously, any remote access is a big area of
security risk. You should pay careful attention to all services. Start by opening the file
/etc/inetd.conf in a text editor such as vi and disable everything that's enabled. Turn
off Web server access, turn off FTP server access, turn off POP or IMAP access, and Part
turn off everything by commenting out the lines from the file. Typically, putting * as
II
the first character of the line disables the service. Some GUI-based tools such as
Ch
linuxconf let you turn services on and off through windows-based applications.
13
Tip #79 from
Steve Turning off services is especially relevant for services such as telnet, finger, rsh,
and rlogin. finger is a ``look-up'' program that gives information about the user,
and if you need it, you'll know why you need it. The telnet and r- commands can
be replaced for general use by the ssh (secure shell) client and server discussed in
Chapter 32, ``Accessing the Network with telnet, ftp, and the r- Commands.''
294 Chapter 13 Improving System Security
You'll learn how to choose decent passwords elsewhere in this book. You might consider
running a program such as crack against your password file, only if you have administrative
privileges on this system. If you are using a company system, just having a crack program or
similar design in your system or directory could be grounds for immediate dismissal from the
company, or grounds for cancellation of your account from your Internet service provider.
Do not ``help out'' the official system administrators unless you get agreement from them in
advance. However, if you are the (or a) system administrator, you should test the system's
security before someone else decides to.
Security should not be a ``one-time'' issue; it should be an ongoing concern. Change
passwords as often as you feel appropriate. Check to see whether any services that you didn't
turn on are suddenly running (automatic monitoring can help with this check). Resist the
urge to write the root password (or any password) on a sticky note and put the note on your
monitor or mousepad, and so on.
Other aids to security are available. For example, you can find more than one file
system for Linux that automatically encrypts the entire file system (Matt Blaze's
Cryptographic file system is a well-known example). IPSec encrypts the IP traffic over
your network. ssh (discussed elsewhere in the book) can provide a more secure
alternative to Telnet. Tripwire can detect if someone's changed a selected file, such as
the password file, and warn you.
CHAPTER
14
Configuring the Linux Kernel
In this chapter by Jack Tackett, Jr.
What Is the Kernel? 296
Preparing to Build a New Kernel 296
Configuring a New Kernel 297
Compiling the New Kernel 301
Project: Building a Modularized Kernel 302
296 Chapter 14 Configuring the Linux Kernel
What Is the Kernel?
The kernel is the core of the Linux operating system and provides the basic system services to
the rest of Linux. Remember, Linux isn't a commercial product, so you might find some
problems after a new distribution is released. Or someone may discover a serious security
hole in the kernel. This happens all the time with both commercial and ``free'' operating
systems. The difference is that with Linux, because the source code is available, you can patch
any problems immediately after they are discovered. You don't have to wait for your
commercial vendor to release a new service pack to fix a hole in your system.
In addition, a new feature in the current releases of the Linux kernel enables you to load
specific device and program support into the kernel without precompiling the support into a
large kernel. As a result, Linux can load into memory only those parts of the kernel it needs.
Modules also provide a way for you to modify the kernel to solve a problem or to add a new
feature without recompiling the entire system.
Preparing to Build a New Kernel
Sometimes a problem has only one solutionÐa new kernel. The kernel is the core operating
system for Linux. Although not for the faint of heart, downloading a new kernel from the Net
and building the kernel are sometimes necessary. If you have some programming experience
and know your way around the C programming language, you should be able to build and
install a new kernel.
You might have to install a new kernel for the following reasons:
& A patch is released to run new hardware.
& You want to remove features you don't use from the kernel to lower the memory
requirements for your system.
The starting point is to determine what kernel version you're now running. You can find out
the kernel version by using the following command:
uname -- release
or
uname -r
The response indicates the kernel version currently running on your system. The command
displays the version number in the following format:
MajorVersionNumber.MinorVersionNumber.PatchLevel
Linus Torvalds is the official release point for new kernels, although anyone can modify
Linux (because of the General Public License, or GPL). Because Linus is the official release
point, the Linux development and user community have a common baseline from which to
work and communicate.
297
Configuring a New Kernel
Be sure to read the ``Kernel How-To'' ( /usr/doc/Kernel-HOWTO) for up-to-date
information before actually trying to build and configure a new kernel. If you mess
up, you could render your system useless. You should also make sure to keep an
older, working copy of a kernel around just in case of problems. You can then boot
that kernel instead of the worthless kernel.
Configuring a New Kernel
To build a new kernel, you first need to configure the source code files. The source files
should be located in the /usr/src/linux directory. You also must have the C compiler
package loaded. If you didn't install that package during installation, you can use RPM to do
so now with the following commands:
rpm -i kernel-source-2.2.5-15.i386.rpm
rpm -i egcs-2.91.66.i386.rpm
?
You also might have to install the kernel headers and various compiler libraries.
See ``Installing Packages with RPM,'' p. 169
First, you must get the new kernel sources or patches. You can usually find the new sources
on the Internet; check metlab.unc.edu for the latest and greatest kernels and also
ftp.kernel.org. (If you're modifying your current kernel, this step is, of course,
unnecessary.) The source files are usually in a tar file and need to be unarchived.
Tip #80 from
Jack Making a backup copy of your current kernel with the following commands is a very
good idea:
cd /usr/src
mv linux linux.old
These commands copy the entire Linux source directory to another directory called
linux.sav.
Next, you should use the patch command to apply any patch files. After preparing the source
files, you can configure and build your new system. Depending on your personal preferences
Part
and hardware available, you can choose from three methods for configuring the kernel: a text-
II
based program, a text-based menu program, and if you have installed X Windows System, an
X-based program. Ch
14
298 Chapter 14 Configuring the Linux Kernel
The AC kernel patches will require the linux directory have a sym-link with
a directory called linux.ac linux - linux.ac, apply
or even better rename
the patches and rename the directory linux. Patch commands are gzip -dc
<filename>.tar.gz | patch -p0 or bunzip2 cd <filename>.tar.bz2 | patch
-p0. Standard patches do not require re-naming the linux directory.
Tip #81 from
Jack To use kernel modules, you must answer Yes to kerneld support and module version
CONFIG_MODVERSIONS)
( support during your kernel configuration.
The current command sequence to build the kernel is as follows:
cd /usr/src/linux
make mrporper
invoke one of three commands, make config (CLI), make menuconfig (text GUI), or make
xconfig (X Window System GUI), depending on which user interface you intend to use.
make dep
make zImage or make bzImage
make modules
make modules install
depmod -a
The Interactive Text-Based Program
If you are using the text-based interactive program, you start by entering the following
command from the /usr/src directory:
cd /usr/src/linux
# make config
The make command asks you various questions about the drivers you want to install or
configure. By pressing Enter, you accept the default value for each question; otherwise, you
must supply the answer. Some of the questions are listed in Table 14.1. You might have to
answer other questions depending on the version of the kernel you're installing or the patches
you've applied. This list of options is supported by all the configuration utilities described in
this chapter.
299
Configuring a New Kernel
Table 14.1 Kernel Configuration Options
Configuration Option Description
Code Maturity Level For use with experimental components in this kernel
Processor type and features Used to specify processor type, math emulation,
MTRR support, SMP support
Loadable Module Support Needed if you intend to use modules instead of a
monolithic kernel
General Setup Asks a series of questions about general components,
such as math coprocessor support and PCI BIOS
supports
Block Devices Asks questions about the type of IDE hard drives and
other block I/O devices
Networking Options Asks several questions about how to support various
network support features, such as firewalls and IP
Masquerading
SCSI Support Enables support for SCSI controllers
SCSI Low-Level Support Enables low-level support for SCSI controllers and for
reporting on various SCSI statistics
Network Device Support Enables support for various network controllers and
processes
Ethernet (10 or 100Mbit) Enables support for 10baseT and 100baseT ethernet
connections
ISDN Subsystem Enables kernel support for Integrated Services Digital
Network (ISDN)
CD-ROM drivers (not for SCSI Provides support for proprietary CD-ROM drivers
or IDE/ATAPI drivers)
Character Devices Provides support for various character and similar
devices, such as system watchdogs
Mice Provides support for various types of pointing devices
ftape Provides support for floppy-based tape drives and
other removable devices
Filesystems Lets you configure support for various file systems,
including foreign language DOS code pages
Network Filesystems (NFS) Provides support for NFS
Console drivers Provides support for various consoles (display devices) Part
Sound Provides configuration support for various sound II
cards
Ch
Additional low level sound drivers Provides support for low level, machine and sound
card specific drivers 14
Kernel Hacking Provides for profiling support in the kernel
300 Chapter 14 Configuring the Linux Kernel
Using the Menu-Based Program
If you are using the text-based interactive program, you start by entering the following
command from the /usr/src/linux directory:
# make menuconfig
Linux then displays the main screen, as shown in Figure 14.1.
Figure 14.1
Using a graphi-
cal menu sys-
tem can help
speed config-
uration of a
new kernel.
The advantage to using the graphical system is that you must configure only those parts of
the kernel that need to be modified. The interactive text-based system leads you through the
entire configuration process.
Using the X Windows System-Based Program
If you are using the text-based interactive program, you start by entering the following
command from the /usr/src/linux directory:
# make xconfig
Linux then displays the main screen (see Figure 14.2).
Figure 14.2
X Windows
System pro-
vides a less
cluttered sys-
tem for config-
uring a new
kernel.
301
Compiling the New Kernel
The X-based configuration tool allows you to configure only those kernel components you
want to change, just as the graphical text-based tool does. When you click a button, you are
presented with another dialog box you use to configure various components. For example,
Figure 14.3 displays the Loadable Module Support dialog box. From this dialog box, you can
configure the entire kernel for module support.
Figure 14.3
You must spe-
cify module
support during
configuration
to enable such
support in your
new kernel.
To select an item, simply click the appropriate radio button (the diamonds). If you need help
on a specific topic, such as Enable Loaded Module Support, you can click the Help button
along the right side of the dialog box. The resulting dialog box provides helpful information
(see Figure 14.4).
Figure 14.4
Helpful infor-
mation about
the component
is just a mouse
click away.
You must save your configuration after answering the appropriate questions. Simply click the
Save and Exit button to save your new kernel configuration and exit the configuration system.
Compiling the New Kernel
After you answer the various questions to configure your new kernel, you must compile it.
You can use the following commands to build the new kernel:
make dep
make zImage
The build process can take anywhere from a few minutes to many hours, depending on your
Part
hardware. So relax and order another pizza!
II
When the compilation is complete, you need to set up your system to use the new kernel on
Ch
boot. The new kernel is /usr/src/linux/arch/i386/boot/zImage, and you need to copy this
image into the boot directory. But before that, you should create a copy of your current
14
kernel image, just in case something goes wrong. To save the old kernel, use the following
command:
mv /boot/vmlinuz.old /boot/vmlinuz.old
302 Chapter 14 Configuring the Linux Kernel
Then you can copy over the new kernel with this command:
?
cp /usr/src/linux/arch/i386/boot/zImage /boot/vmlinuz
See ``Installing LILO,'' p. 87
To change the default kernel that Linux boots into, you edit the /etc/lilo.conf file and add
another entry for a new kernel. The example in Listing 14.1 shows the addition of the older
kernel to the list of operating systems the machine can boot. To change the kernel, you must
rename /boot/vmlinuz to /boot/vmlinuz.old with the preceding commands and then change
its label to old in lilo.conf, as shown in Listing 14.1.
Listing 14.1 A Sample /etc/lilo.conf File
boot=/dev/hda
map=/boot/map
install=/boot/boot.b
prompt
timeout=50
image=/boot/vmlinuz
label=linux
initrd=/boot/initrd
root=/dev/hda1
read-only
image=/boot/vmlinuz.old
label=old
root=/dev/hda1
read-only
After you make the changes to /etc/lilo.conf, run the following command:
/sbin/lilo -v
The updated lilo is then written to the boot device. From then on when you reboot,
the machine will boot into the new kernel (linux) as default instead of the older kernel, with
a 50-second delay to give you time to choose the old kernel at the boot prompt if you want
to boot that kernel.
Project: Building a Modularized Kernel
With the introduction of modularization in the Linux 2.0.x kernel, some significant changes
have been made in the process of building customized kernels. In the past, you were required
to compile support into your kernel if you wanted to access a particular hardware or file
system component. For some hardware configurations, the size of the kernel could quickly
reach a critical level, so to require ready support for items that were used only occasionally
was an inefficient use of system resources. With the capabilities of the 2.0.x kernel, if certain
hardware components or file systems are used infrequently, driver modules for them can be
loaded on demand. To see the current modules in use, use the following command:
lsmod
303
Project: Building a Modularized Kernel
The output, shown here, lets you know what modules are loaded and how they loaded, as well
as how many pages of memory they are using:
Module Pages Used by
isofs 5 1 (autoclean)
ne2k-pci 1 1 (autoclean)
8390 2 [ne2k-pci] 0 (autoclean)
BusLogic 20 4
Only Red Hat Linux/Intel and Red Hat Linux/SPARC support modular kernels; Red Hat
Linux/Alpha users must build a monolithic kernel as described in the earlier section
``Preparing to Build a New Kernel.'' These instructions provide you with the knowledge
required to take advantage of the power and flexibility available through kernel
modularization.
You need to have the kernel-headers and kernel-source packages already installed.
Also, you must issue all commands from the /usr/src/linux directory.
To make the modules, go to /usr/src/linux and run the following command:
make modules
Then use this command to install the modules and run the following command:
make modules_install
Working with Kernel Modules
Now that you have compiled and installed the modules, you are ready to extend your kernel
with loadable modules. Table 14.2 shows the basic commands that are available.
Table 14.2 Module Commands Available in Linux
Command Description
lsmod Lists the modules currently loaded in the kernel
insmod Inserts a specified module into the kernel
rmmod Removes the specified module from the kernel
Part
depmod Creates a dependency file for use by modprobe
modprobe Loads modules from a list generated by depmod II
Ch
If you are running X Window System, you can take advantage of the kerneld daemon from 14
the Control Panel (shown in Figure 14.5) to work with modules from a GUI instead of from a
command line. Clicking this button brings up the Kernel Configurator dialog box shown in
Figure 14.6.
304 Chapter 14 Configuring the Linux Kernel
Figure 14.5
The Control
Panel provides
access to many
administrative
functions, in-
cluding work-
ing with
kerneld.
To list the currently loaded modules, use the lsmod command. To add a module you have
compiled to the kernel, you can use the following command:
insmod module-name
Or you can click the Add button on the kerneld dialog box and specify the module (see Figure
14.7).
Figure 14.6
Working with
kernel mod-
ules is easy
with the X in-
terface to ker-
neld.
Figure 14.7
Adding mod-
ules is a breeze
using X Win-
dow System.
To delete a module from the kernel, you can use the following command:
rmmod module-name
Alternatively, you can select the module from the list displayed in Figure 14.6 and click the
Remove button.
Restarting kerneld
The changes you make with the Kernel Daemon Configuration tool are made in the /etc/
conf.modules file, which kerneld reads whenever it is started. Listing 14.2 provides a sample
listing.
305
Project: Building a Modularized Kernel
Listing 14.2 A Sample /etc/conf.modules File
alias scsi_hostadapter BusLogic
alias eth0 ne2k-pci
To restart kerneld, you can use the tool shown in Figure 14.6 and click the Restart Kerneld
button. You can also restart the daemon via the command line, as shown here:
/etc/rc.d/init.d/kerneld stop
/etc/rc.d/init.d/kerneld start
Restarting kerneld does not cause any modules that are currently in use to be reloaded, but
kerneld will use the configuration when it loads modules in the future.
Part
II
Ch
14
CHAPTER
15
Linux on PowerPC Platforms
In this chapter by Steve Burnett
Linux for the PowerPC 308
MkLinux 308
Yellow Dog Linux 309
LinuxPPC 310
SheepShaver: Use Linux and Have Your Macintosh Too 312
Project: Adding Linux to a Macintosh Environment with Netatalk 312
308 Chapter 15 Linux on PowerPC Platforms
Linux for the PowerPC
Although Linux is predominantly developed and used on the Intel series of processors, the
operating system has been ported to other hardware platforms. For example, the Red Hat
distribution Linux was at one point shipped for the Intel processor, Motorola's PowerPC
RISC processor, Sun's Sparc, and the Alpha processor. Other efforts working with non-Intel
processors include a minimal but functioning port of Linux to the PDA (personal digital
assistant) PalmPilot from 3Com. Although not much appears to be active with Linux on
Sparc and Alpha, an active development effort focuses on work supporting Linux on the
PowerPC family of processors from Motorola. The two primary vendors of PowerPC-based
hardware are Apple Computer, which switched from the Motorola 68k family of CPUs for its
Macintosh computers in the mid-1990s, and IBM, which uses PowerPCs as the CPU in its
series of RS/6000 workstations and servers.
PowerPC support was recently integrated into the main Linux kernel tree in version 2.2.x.
Work is in progress to port to 64-bit versions of the PowerPC architecture, and also to
include the hard-realtime kernel extension. Although binaries of Linux for Intel applications
don't run on PowerPC versions of Linux, recompiling a given application is usually all that's
required; a primary goal of the development effort was to make software porting relatively
painless. More than one of the Linux for PPC distributions has standardized on the glibc C
libraries. For example, Applix's port of its Applixware Office Suite to LinuxPPC took about a
month and was reported to be a ``flawless'' transition.
Informational sources for Linux on PowerPC include a Usenet newsgroup,
comp.os.linux.powerpc, which includes MkLinux, Linux-pmac, and LinuxPPC.
You can find the Web site for the Linux/PPC project at http://www.linuxppc.org/,
with a kernel development group at http://www.ppc.kernel.org/. You also can
find a general informational Web site devoted to Linux on the PowerPC platform at
http://linux.macnews.de/index.shtml.
The following sections of this chapter present some of the distributions of Linux for the
PowerPC. Other distributions include a German language-localized version, a Red Hat-
derived distribution named TurboLinux, a version for non-PowerPC-based Macintoshes
called Linux/m68k, and Debian's distribution.
MkLinux
MkLinux is Apple Computer's Mach 3-based version of the Linux operating system. Versions
of MkLinux are available for the Intel, PA-RISC, and PowerPC architectures. Apple's
Developer Release 3 disc contains the version of MkLinux that runs on PowerPC-based
Apple Power Macintosh and related systems. The following lists the supported and
unsupported systems for the current version.
309
Yellow Dog Linux
& Runs MkLinuxÐApple Power Macintosh 6100, 7100, 8100, 9100, 7200, 7500, 7600,
Part
8500, 9500, 7300, 8600, 9600, 4400, 5400, 5500, 6400, and 6500; 20th Anniversary
Mac; G3 Desktop; G3 Minitower; Powerbooks 5300, 1400, 2400, and 3400; and G3. II
Similar clones are likely to function but may not have been explicitly tested.
Ch
& Does not run MkLinuxÐ68k Macintoshes; PowerPC Performas 52xx, 53xx, 62xx,
15
and 63xx; Powerbook 2300; iMac/A; iMac/B; colored iMacs; iBook; and Blue and
White G3.
As of DR3, MkLinux binary executables are compatible with the other PowerPC Linux
systems (for example, LinuxPPC). In addition, MkLinux file systems are byte-order
compatible with other Linux file systems (for example, Intel-based). MkLinux DR3 added
support for dynamic shared libraries to reduce disk and RAM storage and improve program-
loading speed.
For more information on MkLinux, you can find Apple's Web site for it at
http://www.mklinux.apple.com. You also can find a community-based Web site
for the project at http://www.mklinux.org/.
Yellow Dog Linux
Yellow Dog Linux is made available in two flavors:
& Yellow Dog Linux Champion Server
& Yellow Dog Linux Gone Home
The Server edition ships with a suite of applications intended for business or technical use,
whereas the Gone Home edition ships with user-based tools for use as a home desktop.
Commercial support is available for the Yellow Dog distribution. Table 15.1 indicates which
platforms support Yellow Dog Linux.
Table 15.1 Yellow Dog Linux-Capable Hardware
Company Platforms That Support Yellow Dog
Apple Computer Blue & White G3s; iMac; G3 desktop; All-in-Ones; and Powerbooks,
9600, 9500, 8600, 8500, 7600, 7500, 7300, 7200, 6500, 6400, 5500,
5400, 4400
Apple Computer Twentieth Anniversary Macintosh, PowerBook 2400 and 3400, Apple
Network Server (ANS) 500 and 700
Apus 2000, 3000
Be BeBox
IBM RS6000 (830, 850, 40P, Nobis, INDI)
Motorola StarMax (and all StarMax clones from APS, PowerTools, and
MacTell), PowerStack, FirePower, Series E, and PowerStack II
Power Computing PowerBase, PowerWave, PowerCenter, PowerCenter Pro,
PowerTower, PowerTower Pro
310 Chapter 15 Linux on PowerPC Platforms
Table 15.1 Yellow Dog Linux-Capable Hardware
Company Platforms That Support Yellow Dog
Umax C500, C600, J700, and S900; Apus 2000 and 3000
Only the Apple models in the first row of the table are supported; the Yellow Dog
documentation states that the rest of them work but are unsupported.
The Black Lab Linux distribution is built on Yellow Dog Linux Server and is the Beowulf
clustering capability added to LinuxPPC. The reason for this is the newer Apple Macintoshes
such as the blue-and-white minitowers, the iMac and iBook, do not require a video card or
CD-ROM. Macintosh hardware built on the New World ROM design automatically
attempts a ``net boot,'' searching for a remote machine to boot from, loading the operating
system into RAM without use of internal storage devices.
For more information on Yellow Dog and Black Lab Linux, check out the Yellow Dog Linux
http://www.yellowdoglinux.com
Web site at and the Black Lab Linux Web site at http://
www.blacklablinux.com/.
LinuxPPC
LinuxPPC is a port of the Linux operating system to the PowerPC processor-based
platforms, such as the Apple Macintosh and IBM RS/6000 platforms. LinuxPPC has been
developed under the GNU General Public License (GPL), and as such, the source code is
freely available. Unlike MKLinux, it uses the conventional Linux kernel instead of the Mach
microkernel. LinuxPPC is compatible with programs compiled for MkLinux, except for the
few that make Mach microkernel system calls.
LinuxPPC is offered as a commercial release by LinuxPPC, Inc. Similar to MkLinux,
LinuxPPC is based on the Red Hat Linux distribution and ships with a selection of
applications and utilities, including the KDE GUI, Netscape Communicator, the GIMP,
Apache, sendmail, sshd, and a suite of C and C++ software-development tools. LinuxPPC
currently supports glibc 2.1 and the Linux 2.2 kernel. In addition to the freely distributed
material that comes with the operating system, Applix has ported the Applixware Office Suite
to LinuxPPC, as well as Stalker Software's Communigate Pro mail server.
Being a PowerPC-native operating system, LinuxPPC can coexist peacefully on the same
system as the original operating system, such as AIX, Mac OS, or BeOS, in the same way
Linux for Intel can share a system with Microsoft Windows or other operating systems. A few
features of LinuxPPC that are notable from a Linux for Intel processor perspective include
the following:
& Support for PCI-based 601, 603/603e, 604/604e/604r, and G3 machines. However,
NuBus-based Macintosh systems such as the Apple Power Macintosh 6100, 7100, and
8100 family are not supported.
311
LinuxPPC
& Serial and USB device support for many printers and modems.
&
Part
X Windows and several window managers such as AfterStep, WindowMaker,
II
Enlightenment, and desktop environments such as KDE and GNOME.
& Java support (JDK 1.1.7, 1.2).
Ch
Being a full port of Linux, LinuxPPC supports all the expected hardware: SCSI and IDE 15
drives, Ethernet and other networking gear, serial port devices such as printers and modems,
and so on.
For Apple Macintosh users, the BootX application is a LILO-like chooser that allows you to
select Mac OS or Linux at boot-up. You also can choose to quit Mac OS and boot LinuxPPC
from the Mac OS Finder. BootX consists of the application itself and a Mac OS system
extension. An included MkLinux plug-in supports the use of BootX for booting into
MkLinux instead of LinuxPPC. BootX allows you to use Mac OS initialized video, and
installation of LinuxPPC does not require a floppy drive. In addition to an X-based installer,
you can also use a Red Hat-style installer by adding the word redhat to the kernel arguments
line in the BootX window that comes up when starting Install LinuxPPC on the CD.
Tip #82 from
Steve The BootX application may be found at the author's Web site at
http://calvaweb.calvacom.fr/bh40/.
LinuxPPC runs on PCI-based Power Macintosh computers and compatibles, as well as
PowerPC BeBox, PReP, and CHRP machines. Machines that can run LinuxPPC are shown
in Table 15.2.
Table 15.2 LinuxPPC Supported Hardware
Company Platforms That Support LinuxPPC
Apple Computer Blue & White G3s; iMac; G3 desktop and Powerbooks, 9600, 9500,
8600, 8500, 8200, 7600, 7500, 7300, 7200, 6500, 6400, 6360, 5500,
5400, 4400; Powerbook 2400 and 3400; 20th Anniversary Macintosh
Be BeBox
IBM RS6000 (PowerPC-based), 830, 850, 40P, Nobis, INDI
Motorola StarMax (and all StarMax clones from APS, PowerTools, and
MacTell), PowerStack, Series E, PowerStack II
Power Computing PowerBase, PowerWave, PowerCenter, PowerCenter Pro,
PowerTower, PowerTower Pro
Umax C500, C600, J700, and S900; Apus 2000 and 3000
312 Chapter 15 Linux on PowerPC Platforms
LinuxPPC maintains an actively supported mailing list for those interested in its distribution
linuxppc-
of Linux. To subscribe to the LinuxPPC-user mailing list, you can send email to
[email protected] and add the word subscribe in the body of the message,
or you can send email to [email protected] if you want
announcements only.
For more information and other lists, you can check out the informational Web page at
http://lists.linuxppc.org/.
SheepShaver: Use Linux and Have Your
Macintosh Too
SheepShaver is originally a Mac OS runtime environment for the BeOS developed by
Christian Bauer and Marc Hellwig; it allows users of PowerPC-based BeOS and Linux
systems to run Mac OS applications at native speed inside the BeOS multitasking
environment. This means that both Linux and Mac OS applications can run at the same time
and exchange data between them. Because SheepShaver is not an emulator running in full-
screen mode, you won't notice any difference in speed compared to running Mac OS without
SheepShaver. You need a copy of Mac OS to use SheepShaver.
SheepShaver currently supports the operation of Mac OS 7.5.2, 7.5.3, 7.5.5, 7.6, 7.6.1, 8.0,
and 8.1, and with any Macintosh application that doesn't access Mac hardware directly. In
addition, you can also run an emulator such as VirtualPC under SheepShaver or WINE for
Linux.
SheepShaver for LinuxPPC should run on any BeOS/Linux-ready PowerPC system,
including such rarities as the BeBox and PowerPC Amigas. It does not run on Intel machines
because Mac OS and PowerPC Mac applications run natively under SheepShaver accessing
the PowerPC directly. SheepShaver/Linux requires at least glibc 2.0, GTK 1.2, and a 2.2.x
kernel. For more information, see Sheepsaver's Web site at http://www.sheepshaver.com/.
Project: Adding Linux to a Macintosh
Environment with Netatalk
Consider a prepress graphics shop that is an all-Macintosh operation. The office
administrator decides he wants to experiment with Linux and, in the by now venerable
tradition, finds a system he can set up as a Linux-based file and print server. However, all his
employees use Macintoshes, which communicate among themselves with the network
protocol, AppleTalk, over which Mac OS provides file sharing and network printing. How
does the office administrator make the Linux server a benefit to his users? With the use of
Netatalk on the Linux server.
313
Project: Adding Linux to a Macintosh Environment with Netatalk
Through the AppleTalk protocol, a machine running Mac OS can use file and print services
Part
from a Linux box. The reverse is also possible: A Linux machine can print to remote
AppleTalk print queues and even mount AppleTalk disk shares. II
Ch
The platform for the Linux server is irrelevant in this case. The Linux server can be 15
PowerPC-based, Intel-based, or based on another processor.
The Mac OS file system (HFS) is included in the Linux kernel from release 2.2.x and higher
(HFS+ is not yet supported, however). Netatalk is a kernel-level implementation of the
AppleTalk Protocol Suite; it was originally developed for BSD UNIX-derived systems. It
includes support for routing AppleTalk, serving UNIX and AFS file systems over AFP
(AppleShare), serving UNIX printers, and accessing AppleTalk printers over the Printer
Access Protocol (PAP), as well as including a small suite of utilities. The basic Netatalk is
available from the University of Michigan at http://www.umich.edu/~rsug/netatalk/.
Adrian Sun has developed a set of patches and modifications including bug fixes and
performance enhancements to the basic Netatalk suite; you can find this set under the name
netatalk+asun at ftp://ftp.u.washington.edu/public/asun/ or from Red Hat Software's
site at ftp://contrib.redhat.com/.
An AppleTalk network address is described by a network number and a node number.
Network numbers can be in the range of 1 to 65279 (65280 through 65534 are reserved for
networks with no AppleTalk routers present). Each network can have up to 253 defined
nodes. AppleTalk uses zones for organizational and access control, as well as making
AppleTalk networking more friendly. Zones conceal the numerical network address from
clients in the same way the domain name service masks the IP number of a server
(192.168.0.45) with a name such as www.afakecompany.com.
AppleTalk support has been included in Linux since release 1.3.x, specifically for the DDP
protocol. DDP is the AppleTalk datagram protocol over which the other protocols are
implemented, just as FTP is implemented over TCP/IP. DDP lives in the kernel, but the
other protocols are supported by the Netatalk package itself. Adding AppleTalk support as a
module (typically named appletalk.o) is more flexible for the user than compiling AppleTalk
support into the kernel. As a module, AppleTalk can be reset without disturbing other
networking activity by removing and reinserting the module. To disable the AppleTalk
module, you can enter the following at a terminal prompt:
# rmmod appletalk
To restart the AppleTalk module, enter the following
# insmod appletalk
To find out whether your Linux distribution's kernel has AppleTalk support as a module, see
whether the following exists:
/lib/modules//misc/appletalk.o
314 Chapter 15 Linux on PowerPC Platforms
Tip #83 from
Steve You can check your kernel version by using the uname -r command.
To check whether AppleTalk support has been compiled directly into the kernel, see whether
the file /proc/net/appletalk/ exists, or check the boot messages (using dmesg) for a line like
the following:
'NET4: AppleTalk 0.18 for Linux NET4.0'
If the sources used for the current kernel are available, you can also check /usr/src/linux/
.config to see whether CONFIG_ATALK is set to y.
Netatalk's default configuration files are nicely self-documented, and in many cases suffice
without much, if any, tweaking. One default setting you might want to change is the number
of simultaneous users, which defaults to five.
PART
III
Working with Linux
16 Understanding Linux Shells 317
17 Managing Multiple Processes 365
18 Printing 391
19 Understanding the File and Directory System 407
20 Managing File Systems 439
21 Managing NFS 463
22 Managing NIS and LDAP 475
23 Using Samba 485
CHAPTER
16
Understanding Linux Shells
In this chapter by Steve Burnett
Logging In 318
Understanding Shells 319
Understanding Shell Command Parsing 332
Doing Background Processing 343
Understanding Command Feedback 346
Editing and Aliasing Shell Commands 346
Working with Shell Script 348
Customizing Linux Shells 361
Troubleshooting 364
318 Chapter 16 Understanding Linux Shells
Logging In
As a new user and novice system administrator on your recently installed Linux system,
you've chosen a login ID and password. Because Linux is a multiuser operating system, it
must be able to distinguish between users and classes of users. Linux uses your login ID to
establish a session in your name and determine the privileges you have. Linux uses your
password to verify who you are.
Because any user can log in to any terminal in theory (there is an exception), the UNIX
operating system begins by displaying a login prompt on every terminal. Because it's unlikely
you'll have multiple terminals connected to your initial Linux system (although connecting
multiple terminals is certainly possible), you'll have the alternate, or virtual, terminals
available to you.
To switch to the various virtual terminals, you can press the Alt key and any of the first six
function keys. For example, to log in to virtual terminal one as root, you can press Alt+F1,
which displays the following prompt:
Red Hat Linux release 5.1 (Manhattan)
Kernel 2.0.36 on an i686
login:
The prompt line in the code line declares this sample session to be running under the
2.0.36 version of the Linux kernel. As newer kernels are released, this number is
incremented, so you might see a different version on the accompanying CD-ROMs.
The stable released kernels are given even numbers for the middle number, and the
odd numbers indicate the latest (and beta) releases.
At this point, you can enter your user ID (root) and password.
When you log in to any terminal, you own the session on that terminal until you log out.
When you log out, Linux displays the login prompt for the next user. Between logging in and
logging out, Linux makes sure that all the programs you run and any files you might create
are owned by you. Conversely, Linux doesn't allow you to read or alter a file owned by
another user unless that user or the system administrator has given you permission to do so.
Your login ID and password allow Linux to maintain the security of your files and those of
others.
As the system administrator for your Linux system, you assign every user a user ID,
temporary password, group ID, home directory, and shell. This information is kept in a file
named /etc/passwd, which is owned and controlled by the system administrator, also known
as root or the superuser. After you successfully log in, you can change your password, which is
then encrypted in a form that no one else can read. If you forget your password, you (the
system administrator) have to log in as the root user to create a new password. You can
319
Understanding Shells
change your own password by using the passwd command (although you have to type in the old
password).
For more information on basic system administration duties, such as adding users
and fixing forgotten passwords, see the chapters in Part II, ``System Administration,''
particularly Chapter 11, ``Managing User Accounts.''
Part
III
Ch
Understanding Shells 16
After you log in, Linux places you in your home directory and runs a program called a shell. A
shell is really nothing more than a program designed to accept commands from you and
execute them. Many kinds of programs can be used as shells, but several standard shells are
available with almost all versions of Linux.
Although graphical interfaces have been added to the UNIX system in recent years,
you run most of the utilities for using and administering Linux (and other UNIX-like
systems) by typing commands. In Linux, the command-line interpreter is called the
shell. Linux shells are equivalent to the COMMAND.COM program used by MS-DOS:
they accept and execute commands, run batch files, and execute programs.
Looking at Different Shells
Red Hat Linux provides the following shells: sh, bash (Bourne Again Shell), tcsh, csh, pdksh
(Public Domain Korn Shell), zsh, ash, and mc. You should try each shell and pick the one you
like. This chapter concentrates on the sh and bash shells because most Linux distributions
install bash as the default shell. Also, sh is available on most UNIX systems, and you'll find
many shell scripts written with sh commands.
Because the shell serves as the primary interface between the operating system and the user,
many users identify the shell with Linux. They expect the shell to be programmable, but the
shell isn't part of the kernel of the operating system. With enough background in systems
programming and knowledge of the Linux operating system, you can write a program that
can become a shell.
Although many different shells have been created, several shells are prevalent: the
Bourne, C, T, and Korn shells. The Bourne shell is the oldest, and the others have
some features not in the Bourne shell. In fact, Linux uses a variation of the Bourne
shell, the bash shell, as its default shell. (To the novice user, the Bourne and Korn
shells look identical; indeed, the Korn shell was developed from the Bourne shell.)
320 Chapter 16 Understanding Linux Shells
The Red Hat distribution provides a version of the Korn shell called pdksh, which
stands for Public Domain Korn Shell.
The C shell was developed at the University of California at Berkeley as a shell more suitable
for programmers than the Bourne shell. The T shell is a derivative of the C shell. The Korn
shell has all the features of the C shell but uses the syntax of the Bourne shell. If all this
information sounds confusing at the moment, don't worry. You can do a lot of work without
knowing or worrying about the shell you're using.
$
In their simplest forms, the Bourne and Korn shells use the dollar sign ( ) as the standard
%
prompt; the C shell uses the percent sign ( ) as the prompt. Fortunately (or not, depending on
your disposition), you can change these prompts so that you can or cannot see either the
dollar sign or the percent sign when you first log in.
The Bourne shell, known as sh, is the original UNIX shell. It was written by Steve Bourne
with some help and ideas from John Mashey, both of AT&T Bell Laboratories; this shell is
available on all Linux systems. The executable program for this shell is in the file /bin/sh.
Because the Bourne shell is available on all Linux systems, and it has all the properties
described in the preceding sections as well as powerful programming capabilities, it has
become widely used.
Tip #84 from
Steve Many of the shell script examples in this chapter are written so that they can be used
with the Bourne shell. Shell scripts are sequences of shell commands, normally
written with an ASCII editor such as vi. You can think of shell scripts as similar to DOS
batch files.
? See ``What Is vi?'' p. 205
The C shell, known as csh, was developed by Bill Joy at the University of California at
Berkeley. The students and faculty at Berkeley have had a great deal of influence on UNIX
and hence Linux. Two results of that influence are the C shell and the vi text editor. The
Bourne shell has superior shell programming capabilities, but the C shell was developed to
reflect the fact that computing is becoming more interactive. The executable program for the
C shell is in the file /bin/csh.
The syntax of the C shell closely resembles the C programming language. This is one reason
that shell scripts written for the C shell often can't run under the Bourne or Korn shell
(executables compiled under the C shell often behave properly, though). But the C shell has
some desirable features not available in the Bourne shell: command editing, history, and
aliasing.
321
Understanding Shells
The default Linux shell is the bash shell. bash is located in /bin/bash and provides several
enhanced features detailed in the next few paragraphs, such as command editing, command
history, and command completion.
Tip #85 from
Steve Man pages are not available for the various shells, but the shells each have their own
internal help. To find help, you can open a terminal window and type help. For the Part
bash shell, for example, you see a list of commands defined internal to the bash shell.
III
Each command also has its own help included in the bash shell.
Ch
16
All Linux systems have the bash shell. You also might have installed several other shells
during installationÐfor example, the C shell or the T shell. To determine which shell you're
using, you can enter the following:
echo $SHELL
The echo command prints whatever follows the word echo to the terminal screen. SHELL is a
variable, maintained by the shell, that holds the name of your current shell; $SHELL is the
value of that variable.
To see whether the C shell is available, enter this command:
Csh
%
If you see the percent sign ( ) as the prompt, the C shell is available and running (enter exit
to return to your previous shell). If you're logged in as root, the prompt for the C shell is #. If
you get an error message, the C shell isn't available to you.
The shell you use as a login shell is specified in the password file. Each login ID is
represented by a record or line in the password file; the last field in the record specifies your
login shell. To change your login shell, you must change that field. The act of changing to
another shell is relatively easy. Before you change shells, however, you need to decide
whether learning a new syntax and operating method are worth the change. See the man
pages for detailed information on your shell's syntax.
Tip #86 from
Steve Never directly edit the password file ( /etc/passwd) in Linux. Because of the security
features added to these releases, the password file should be manipulated only with
the appropriate commands. This caution is especially important if you are using the
Shadow Suite of utilities.
Several other shells are available; some are proprietary, and others are available on the
Internet or through other sources. To determine which shell you want to use, simply read the
322 Chapter 16 Understanding Linux Shells
man pages for the various shells and give each a try. Because shells are programs, you can run them
just like any other application.
Configuring Your Login Environment
Before you see the shell prompt, Linux sets up your default environment. The Linux
environment contains settings and data that control your session while you're logged in. Of
course, as with all things in Linux, you're completely free to change any of these settings to
suit your needs.
Your session environment is divided into two components:
& The first component, called the terminal environment, controls your terminal (more
properly, the behavior of the computer's port to which you connect the cable from
your terminal).
Because Linux runs on a PC, the ``terminal'' is actually your monitor and keyboard.
You may or may not have other terminals connected to your Linux system. Of course,
you do have six virtual terminals from which you can log in.
& The second component, called the shell environment, controls various aspects of the
shell and any programs you run.
You should first know about your terminal environment.
Setting the Terminal Environment
Your login session actually consists of two separate programs that run side by side to give you
the appearance of having the machine to yourself. Although the shell is the program that
receives your instructions and executes them, before the shell ever sees your commands,
everything you type must first pass through the relatively transparent program called the
device driver.
The device drivercontrols your terminal. It receives the characters you type and determines
what to do with themÐif anythingÐbefore passing them on to the shell for interpretation.
Likewise, every character generated by the shell must pass through the device driver before
being delivered to the terminal. This section is first concerned with how to control the
behavior of your device driver.
Linux is unique in that every device connected to the system looks, to a program, just like
every other device, and all devices look like files. It's the task of the different device drivers in
your system to accomplish this transformation. A hard disk in the system behaves very
differently from your terminal, yet it's the job of their respective device drivers to make them
look identical to a program.
323
Understanding Shells
For example, a disk has blocks, sectors, and cylinders, all of which must be properly addressed
when reading and writing data. Your terminal, on the other hand, accepts a continuous
stream of characters, but those characters must be delivered to the terminal in an ordered and
relatively slow manner. The device driver orders this data and sends it to you at 1200, 2400,
9600, or higher bits per second (bps) and inserts stop, start, and parity bits in the data stream.
Because your terminal is always connected to the system, the device driver allows you to
define special characters, called control characters, that serve as end-of-file and end-of-line Part
markers for your shell. The device driver also allows you to define control characters that
III
send signals to a running process (such as the interrupt signal, which can, in most cases, stop a
Ch
running process and return you to the shell). Figure 16.1 shows one way that the Linux
kernel, shell, and device driver behave. 16
Figure 16.1 Linux Data Shell Data Device Data Terminal
Kernel Driver
You can un-
derstand how
Linux interacts
with the user
through the
command
shell.
You can set dozens of parameters for your terminal, but most of them are handled
automatically. However, you should know about a few parameters and modes.
The device driver has two modes of operation, called cooked and raw. In raw mode, all the
characters you type pass directly to the shell or to a program run by the shell. Programs such
as editors and spreadsheets require raw mode and set it up automatically. When such
programs end, they normally reset your terminal to cooked modeÐbut not always. When
your terminal is in raw mode, it doesn't respond to control keys such as the interrupt key.
When your terminal is in cooked mode, every key you type is interpreted by the device driver.
Normal keys are stored in a buffer until the end-of-line key is pressed. In most cases, the
end-of-line key is the Enter or Return key (however, this key can be changed). When the
device driver receives the end-of-line character, it interprets the entire line before passing
the interpreted or parsed line on to the shell or application program. Table 16.1 lists the most
important control keys.
324 Chapter 16 Understanding Linux Shells
Table 16.1 Control Keys
Key Name Description
Interrupt Interrupts the execution of a running program. When you give Linux a
command and press the end-of-line key, a program typically runs until
normal completion. If you press the interrupt key, you send a signal to the
running program, telling it to stop. Some programs ignore this signal; if your
terminal is in raw mode, the interrupt key passes directly to the program and
may not have the desired effect. The UNIX convention is to use the Del key
as the interrupt key, but Linux changes this key to Ctrl+c for the convenience
of people familiar with MS-DOS and other systems that use this key
combination.
Erase Deletes the last character in the buffer. This key is defined as the Backspace
key. The erase key works just like the Backspace key on a typewriter. On
some terminals and systems, there's confusion between the Delete and
Backspace keys.
Kill Deletes everything in the buffer before it passes to the shell or application
program. This key is normally defined as the @ character. Unlike when you
press the interrupt key, you don't see a new shell prompt when you press the
kill key; the device driver simply waits for you to type more text.
End-of-line Tells the device driver that you've finished entering text and want the text
interpreted and passed on to the shell or application program. Linux uses the
Enter or Return key.
End-of-file Tells the shell to exit and display the login prompt. The end-of-file character
is the Ctrl+d character. Linux treats all devices as though they were files;
because your terminal is a source of virtually unlimited characters, Linux uses
the end-of-file key as a way for you to signal that you're done with your login
session.
The command used to set and display these control-key parameters is stty, which stands for
set teletype. In the old days, a teletype terminal was the only terminal available; a lot of UNIX
terminology is left over from this era. For example, your terminal is defined as a tty device
with a name such as tty14. To display all your present settings, enter stty -a on the command
line. If you use this command, you see something like this:
speed 38400 baud; rows 25; columns 80; line = 0;
intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;
eol2 = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W;
lnext = ^V; flush = ^O; min = 1; time = 0;
-parenb -parodd cs8 hupcl -cstopb cread -clocal -crtscts
-ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon ixoff
-iuclc -ixany -imaxbel
opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
isig icanon iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
echoctl echoke
Notice that on this system, the interrupt key ( intr) is defined as Ctrl+c (shown as ^C), and the
kill key is Ctrl+u. Although you can set all the settings listed here, as a matter of practicality,
325
Understanding Shells
users usually reset only the interrupt and kill keys. For example, if you want to change the kill key
from ^U to ^C, enter the following:
stty kill '^C'
If your terminal is behaving strangely, reset it to a ``most reasonable'' setting by giving
the command stty sane.
Part
III
Ch
Tip #87 from 16
Steve If you want a certain setting to take effect every time you log in, place the command
in your .profile file (located in your home directory) if you're running the bash,
Bourne, or Korn shell. For the C shell, place the command in your .login file.
Setting the Shell Environment
Part of the process of logging inÐthat is, of creating a Linux sessionÐis the creation of your
environment. All Linux processes (as running programs are called) have their own environment
separate and distinct from the program itself. It could be said that a program runs from within
an environment. The Linux environment, called the shell environment, consists of a number of
variables and their values. These variables and values allow a running program, such as a shell,
to determine what the environment looks like.
Environment refers the shell that you use, your home directory, and what type of terminal
you're using, for example. Many of these variables are defined during the login process and
either can't or shouldn't be changed. You can add or change as many variables as you like as
long as a variable hasn't been marked read-only.
Variables are set in the environment in the form VARIABLE=value. You can set the meaning of
VARIABLE to anything you like. However, many variables have predefined meanings to many
standard Linux programs. For example, the TERM variable is defined as being the name of your
terminal type, as specified in one of the standard Linux terminal databases. Digital
Equipment Corporation for years made a popular terminal named the VT-100. The
characteristics of this terminal have been copied by many other manufacturers and often
emulated in software for personal computers. The name of such a terminal type is vt100; it's
represented in the environment as TERM=vt100.
Many other predefined variables exist in your environment. If you use the C shell, you can list
these variables by using the printenv command; with the Bourne or Korn shell, you use the
set command. Table 16.2 lists the most common environment variables and their uses. The
Variable column shows what you type at the command line.
326 Chapter 16 Understanding Linux Shells
Some environment and system variables can be changed, and some can't.
Table 16.2 Common Bourne Shell Environment Variables
Variable Description
HOME=/home/login HOME sets your home directory, which is the location from which you
start. Replace login with your login ID. For example, if your login ID is
jack, HOME is defined as /home/jack.
LOGNAME=login LOGNAME is automatically set the same as your login ID.
PATH=path The path option represents the list of directories that the shell looks
through for commands. For example, you can set the path like this:
PATH=/usr:/bin:/usr/local/bin.
PS1=prompt PS1 is the primary shell prompt that defines what your prompt looks like.
If you don't set it to anything specific, your prompt is the dollar sign ( $).
If you prefer, you can set it to something more creative. For example,
PS1="Enter Command >'' displays Enter Command > as your command-line
prompt.
PWD=directory PWD is automatically set for you. It defines where you are in the file
system. For example, if you checked PWD (by entering echo $PWD at the
command line) and Linux displays /usr/bin, you're in the /usr/bin
directory. The pwd command also displays the current directory.
SHELL=shell SHELL identifies the location of the program that serves as your shell.
For example, you can set SHELL in your .profile or .login file as
SHELL=/bin/ksh to make the Korn shell your login shell.
TERM=termtype This variable sets the name of your terminal type, as specified by the
terminal database. For example, you can set TERM in your .profile or
.login file as TERM=vt100.
Tip #88 from
Steve If you want an environment variable defined every time you log in, you can place the
definition in your .profile file (located in your home directory) if you're running
the bash or Bourne shell. For the C shell, place the definition in your .login file.
Perhaps the single most important variable in your environment is the PATH variable.
MS-DOS users should be familiar with the PATH variable. It performs the same
function under both DOS and Linux.
327
Understanding Shells
The PATH variable contains a colon-delimited string that points to all the directories
containing the programs you use. The order in which these directories are listed determines
which directories are searched first. The list order is important on systems that support
several different forms of the same command. Your system may also have locally created
commands you want to access. For example, your PATH variable might contain the following
values:
/usr/ucb:/bin:/usr/bin:/usr/local/bin Part
This statement tells your shell to explore the /usr/ucb directory first. If the shell finds the III
command in the first directory it searches, it stops searching and executes that command.
Ch
The /bin and /usr/bin directories contain all the standard Linux commands. The
/usr/local/bin directory often contains the local commands you and other users of 16
your system added. This task of adding local commands is usually the responsibility
of the system administrator.
If you are acting as the system administrator, or if you want access to the more system-
oriented commands, you will probably want to add /usr/sbin or /usr/local/sbin or both to
shorten the effort of typing /usr/sbin/traceroute.
If you intend to create your own commands, you can modify the PATH variable to include
directories that contain your own commands. How you do so depends on which shell you use.
For example, if you use the Bourne or Korn shells, you can add a directory to your PATH
variable by typing the following at the command prompt:
$ PATH=$PATH: newpath
Tip #89 from
Steve When you place a $ in front of the name of a variable, its current value is substituted.
In this command, the $PATH variable represents whatever the current path is; the
colon and the newpath parameters add to the current path.
The following sections describe several other ways of manipulating variables in your
environment. For now, it's sufficient to say that the shell environment contains variables and
functions and that these objects can be manipulated by both shells and application programs.
Application programs can access and modify the environment, but they generally manipulate
variables within the program. Shells, on the other hand, can only manipulate variables in the
environment.
Using Special Shell Variables
The shell keeps track of a number of special variables. You can see what they are by using the
env command, which lists the variables available to you within your working environment.
Following is an abbreviated list of what you might see when you enter env:
HOME=/usr/wrev
328 Chapter 16 Understanding Linux Shells
SHELL=/bin/sh
MAIL=/usr/mail/wrev
LOGNAME=wrev
PATH=/bin:/usr/bin:.
TZ=PST8PDT
PS1=$
TERM=vt100
You can use any of these special variables in the same way you use any other shell variable.
Table 16.3 defines the special variables.
Table 16.3 Special Environment Variables
Variable Name Meaning
HOME Full path name of your home directory
SHELL Name of your current shell
MAIL Full pathname of your mailbox
LOGNAME Your login name
PATH Directories the shell searches for commands
TZ Time zone for the date command
SECONDS Number of seconds since invoking the shell
PS1 System prompt
TERM The type of terminal you're using
The HOME Variable The HOME variable always specifies your home directory. When you log
in, you're in your home directory. Occasionally, you use the cd command to move to other
directories. To change to the directory/usr/local/games, for example, you enter
cd /usr/local/games. To get back to your home directory, all you have to do is enter cd.
You can use the HOME variable when you're writing shell scripts that specify files in your home
directory. Rather than write a command such as grep $number /usr/wrev/sales/data.01,
it's better to enter the command as grep $number $HOME/sales/data.01 for these reasons:
& The command line is easier to read.
& If your home directory is moved, the command still works.
& $HOME always represents the home directory of whoever is using the command. If you
enter the command by using $HOME, others can use the command as well.
The PATH Variable The PATH variable lists the directories in which the shell searches for
commands. The shell searches those directories in the order they're listed. If you enter
PATH=/bin:/usr/bin/:., whenever the shell interprets a command, it first looks in the
directory /bin. If it can't find the command there, the shell looks in the directory /usr/bin.
Finally, the shell searches the . directory (remember that the dot represents your current
directory). If you enter cal to print this month's calendar, the shell first looks in /bin.
Because the command isn't there, the shell then looks in /usr/bin and finds it.
329
Understanding Shells
Tip #90 from
Steve If you have a personalized command named cal, the shell never finds it; the shell
executes the cal command in /usr/bin first whenever you give the command. You
should give your commands names that aren't the same as system commands.
You might want to put all your shell scripts in one directory and change the PATH variable to
Part
include that directory. This arrangement allows you to execute your shell scripts from
III
whatever directory you happen to be in. To do so, follow these steps:
Ch
1. Create a directory to hold the scripts. Use the mkdir $HOME/bin command to create the
bin subdirectory in your home directory. 16
2. Move each shell script to that subdirectory. For example, to move a shell script named
stamp to your bin subdirectory, use the mv stamp $HOME/bin command.
3. Add the script subdirectory to your PATH variable by using the PATH=$PATH:$HOME/bin
command. Do so in your .profile file so that the change takes effect every time you
log in to your system.
You need to create the new bin directory and modify the PATH variable only once. Under
Linux, the directory called /usr/local/bin is created to hold ``local'' commands and scripts
that aren't part of the standard Linux package but that you've added locally and have made
available to all users. In this case, you should expect that /usr/local/bin is also part of PATH.
The MAIL Variable The MAIL variable contains the name of the file that holds your email.
Whenever mail comes into the system for you, it's put into the file specified by the MAIL
variable. If you have a program that notifies you when new mail has arrived, it checks the file
associated with the MAIL variable.
The PS1 Variable The PS1 variable holds the string of characters you see as your primary
prompt. The prompt is the string of characters the shell displays whenever it's ready to
receive a command. You'll see how you can change this variableÐand any of the othersÐin
the section ``Customizing Linux Shells'' near the end of this chapter.
The TERM Variable The TERM variable is used to identify your terminal type. Programs that
operate in full-screen mode, such as the vi text editor, need this information.
The TZ Variable The TZ variable holds a string that identifies your time zone. The date
program and some other programs require this information.
Your computer system keeps track of time according to Greenwich Mean Time (GMT). If
theTZ variable is set to PST8PDT, the time and date are determined as Pacific Standard Time
PST), eight hours west of GMT, with support for Pacific Daylight Savings Time (PDT). Your
(
computer system automatically changes between daylight savings time and standard time.
330 Chapter 16 Understanding Linux Shells
The LOGNAME Variable The LOGNAME variable holds your login name, the name or string of
characters that the system associates with you. Among the tasks the LOGNAME variable is used
for is to identify you as the owner of your files, as the originator of any processes or programs
you may be running, and as the author of mail or messages sent by the write command.
The following example is an extension of safrm, a shell script created for the safe removal of
files. TheLOGNAME variable is used to remove all the files you own from the directory /tmp. To
remove those files, the shell script uses the find command. The find command has a number
of options; the shell script uses this find command line:
find /tmp -user $LOGNAME -exec rm {} \;
The first parameter, /tmp, is the directory to search. The option -user indicates that you
want to search for all files that belong to a specified user. Before the command is executed,
the shell replaces $LOGNAME with the current user's login name. The option -exec indicates
that the following command is to be applied to every file found by the find program. In this
case, the rm program is used to remove the found files. The braces ( {}) represent the position
of each filename passed to the rm command. The last two characters, \;, are required by the
find command (an example of using the backslash to pass a character on to a program
without being interpreted by the shell). You can add this command line to the shell script in
Listing 16.1 to obtain a program that removes files safely and also cleans up anything a user
has in the /tmp directory that's more than 10 days old.
Listing 16.1 The safrm Shell Script
# Name: safrm
# Purpose: copy files to directory /tmp, remove them
# from the current directory, clean up /tmp,
# and finally send mail to user
# first copy all parameters to /tmp
cp $* /tmp
# remove the files
rm $*
# create a file to hold the mail message
# The file's name is set to msg
# followed by process ID number of this process
# For example, msg1208
msgfile=/tmp/msg$$
# construct mail message
date > $msgfile
echo "These files were deleted from /tmp" >>$msgfile
# get list of files to be deleted from tmp
# -mtime +10 gets all files that haven't been
# modified in 10 or more days, -print displays the names.
find /tmp -user $LOGNAME -mtime +10 -print >> $msgfile
# remove the appropriate files from /tmp
find /tmp -user $LOGNAME -mtime +10 -exec rm {} \;
# mail off the message
mail $LOGNAME < $msgfile
# clean up
rm $msgfile
331
Understanding Shells
Understanding Processes
A running program in Linux is called a process. Because Linux is a multitasking system, many
processes can run at the same time. To distinguish between processes, Linux assigns each new
process a unique ID called a process ID (PID).
The process ID is simply a number that uniquely identifies each running process. To see what
process IDs are now associated with your process, you can use the ps command. To look at
Part
most of the process IDs now running on your system, you can issue the command with the
flags -guax, and you see something like the following: III
USER PID %CPU %MEM SIZE RSS TTY STAT START TIME COMMAND Ch
jack 53 3.2 7.0 352 468 p 1 S 02:01 0:01 -bash
jack 65 0.0 3.5 80 240 p 1 R 02:01 0:00 ps -guax 16
root 1 0.8 3.1 44 208 con S 02:00 0:00 init
root 6 0.0 1.8 24 124 con S 02:00 0:00 bdflush (daemon)
root 7 0.0 1.9 24 128 con S 02:01 0:00 update (bdflush)
root 40 1.0 3.5 65 240 con S 02:01 0:00 /usr/sbin/syslogd
root 42 0.2 2.9 36 200 con S 02:01 0:00 /usr/sbin/klogd
root 44 0.5 3.2 68 216 con S 02:01 0:00 /usr/sbin/inetd
root 46 0.2 3.0 64 204 con S 02:01 0:00 /usr/sbin/lpd
root 52 0.1 2.0 32 140 con S 02:01 0:00 selection -t ms
root 58 0.2 2.4 37 164 p 6 S 02:01 0:00 /sbin/agetty 38400 tt
The process ID is identified by the column titled PID. Also note the boldfaced line, which
indicates the first process started by the systemÐ init. The init process is also described
later in this chapter.
When Linux is told to run a program (that is, to create a process), it does so by making an
exact copy of the program making the request. In the simplest case, you request that a
program be run by telling your shell; the shell makes a fork request to the Linux kernel.
Fork, init, and the exec Process
A fork is the process of cloning an existing process. Linux creates all new processes through
the mechanism of forking. When a process is forked, an almost exact duplicate of an existing
process (including its environment and any open files) is created; what keeps the duplicate
from being exactly the same as its parent application is a flag that tells the forked process
which is the parent and which is the child.
Because all processes are created in this fashion, all processes have a parent process and a
parent's process ID. Every process running on a Linux system can trace its lineage back to
init, the mother of all processes. init itself, process ID 1, is the only process run directly by
the Linux kernel that you, as a user, have any contact with. Every process you create during a
session has your login shell as an ancestor, and your login shell has init as its parent.
After a process successfully forks, the child process calls the exec routine to transform itself
into the process you requested. The only thing that changes after an exec function is the
identity of the running process; the environment of the new process is an exact copy of the
environment of its parent.
332 Chapter 16 Understanding Linux Shells
Standard Input and Output
Every new process is created with three open ``files.'' Because Linux treats files and devices
exactly the same, an open ``file'' can be a real file on a disk or a device such as your terminal.
The three open files are defined as standard input ( stdin), standard output (stdout), and
standard error output ( stderr). All Linux commands, as well as application programs, accept
input from the standard input and place any output on the standard output. Any diagnostic
messages are automatically placed on the standard error output.
When you first log in, the standard input, output, and error files are attached to your
terminal; any programs you run (processes you create) inherit your terminal as the three open
files.
Understanding Shell Command Parsing
Parsingis the act of splitting the command line, or what you type, into its component parts for
processing. In Linux, parsing constitutes a lot more than simply splitting the command line.
The command string is first split into its component parts: the filenames expand if you use
any wildcards, shell variables expand, I/O redirection sets up, any command groupings or
subshells set up, and command substitution is performed. Only then can the command line,
as you type it, be executed.
If terms such as wildcards and I/O redirection are new to you, you can find explanations of
them, in the order they're performed, later in this chapter. You must first start, however, with
the basic command syntax.
Using Commands, Flags, and Parameters
To execute a Linux command, you merely type the name of the file. The command to list
files is ls; you can find a file by that name in the /bin directory. If /bin is listed in your PATH
variable (and it should be), your shell finds and executes /bin/ls.
Some Linux commands aren't independent files. These commands are built into the shells
themselves. For example, the cd (change directory) command is built into most shells and
executed directly by the shell without looking up a file. You can read the man pages for the
shell you're using to determine what commands are executed internally or externally. Some
shells have a command file that contains commands executed directly by the shell.
Flags
If a command is to execute properly, you must present it to your shell in the proper fashion.
The command name itself must be the first item on the line; it's followed by any flags and
parameters. Flags (sometimes called options) are single letters preceded by a hyphen ( ) that -
modify the behavior of a command. For example, the list command, ls, simply lists the names
of the files in the current directory in alphabetical order. By adding various flags, you can list
the contents of a directory in many different ways. You can list files and all their attributes by
using the ``long'' flag, -l. This command takes the following form:
333
Understanding Shell Command Parsing
ls -l
Here, -l is the flag. When you want to use more than one flag, you can simply string the flags
together, as in ls -lF. The -F flag displays an asterisk (*) if the file is executable, an at sign (@)
if the file is a symbolic link, and a slash ( /) if the file is a subdirectory. The man page for every
command usually lists all the modifying flags and their meanings before describing any
parameters. Flags can also be listed separately; the shell parses them before passing them on
to the program. For example, you can write the ls -lF command as ls -l -F.
Part
III
Ch
Linux provides a popular feature: color highlighting. When you issue the ls 16
command, Linux can display files in different colors depending on the file's type. This
feature allows you to quickly identify files that are executable, directories, or files that
are linked to other files located in other directories. Also, if you redirect the output
from ls to a file, this file contains the control codes used to indicate color. The control
codes' information may cause problems with other programs, such as less, when
used with this file. For Red Hat Linux, you must provide the -color flag to ls to get the
effect:
ls _-color
One type of flag signals that the next parameter has some special meaning. For example, the
-t flag in the sort command is used to indicate that the next character is a field separator. If
you want to sort the /etc/passwd file, whose fields are separated by a colon (:), you can enter
the following:
sort -t: /etc/passwd
In the case of the sort command, the -t flag is needed only if the file uses a field separator
other than the default. The default field separator is defined in the IFS (Inter Field Separator)
environment variable. The shell uses the IFS variable to parse the command line so that the
shell knows to use the standard field separator unless the -t flag indicates otherwise.
Parameters
Flags must be presented to the command before any other parameters. Parameters are strings
separated by any of the characters defined in the IFS environment variable. The default string
in IFS is a space, a tab, and a newline character. You can place any number of field-separator
characters between parameters; when the shell parses the command line, it reduces these
characters to one character before proceeding. For example, if a command is followed by
three spaces, a tab character, and then the first parameter, the shell automatically reduces the
three spaces and a tab to one tab character. Thus, the line
334 Chapter 16 Understanding Linux Shells
command<spacebar><spacebar><spacebar><Tab>parameter
becomes
command<Tab>parameter
Parameters are usually filenames or strings that tell the command to perform some function.
If a parameter contains an embedded space, you must place the string in quotation marks to
prevent the shell from expanding it. The following command line contains two parameters;
the shell attempts to find the word New in a file named York:
grep New York
If the intent is to find the string "New York'' in the standard input, you must enter the
command as follows:
grep ``New York''
In this case, the string "New York'' is passed to the grep command as one parameter.
Performing Filename Matching
Most modern operating systems (including all versions of Linux and DOS) support the use of
wildcards for file and string searches. Table 16.4 summarizes the filename completion
characters, otherwise known as wildcards.
Table 16.4 Filename Completion Characters
Character Meaning
* Represents any collection of characters except a period when it's the first
character in a filename. For example, the command cat sales* > allsales
combines all files whose names begin with sales into a file named allsales.
? Represents a single character. For example, the command lp sales.9? prints a
collection of files with names in the form of sales.yy, where yy represents a year
in the nineties (such as sales.90, sales.91, and so on).
[] Represents a single character in a range. For example, the command rm
sales.9[0-3] removes the collection of files with the names sales.90,
sales.91, sales.92, and sales.93.
Tip #91 from
Steve If you place a filename wildcard or expression inside quotation marks, the filename
expansion is suppressed during command-line parsing. For example, if you type ls *,
ls ''*'', you
you get all files in the current directory. On the other hand, if you type
probably will get the file not found error message because you're instructing ls to
search for a file named *.
335
Understanding Shell Command Parsing
The * Wildcard
*
The asterisk ( ) is the most universal wildcard used. It simply means any and all characters.
For example, the string a* means all filenames beginning with a. You can use as many
asterisks in a single expression as you need to define a set of files. For example, the expression
*xx*.gif means any filename with the .gif extension that has xx anywhere in the rest of the
name. Matches include the filenames abxx.gif, xxyyzz.gif, and xx.gif.
*
You can use the asterisk character ( ) to represent any sequence of characters. For example, to
Part
print all files in your current directory with names that end with .txt, you enter the III
following:
Ch
lp *.txt 16
Pay attention when using the asterisk wildcard. If you enter the following command, you
print all files whose names end with txt:
lp *txt
The file named reportxt is included with the files printed with the second command but not
with the first. If you enter the following command, the shell passes the name of every file in
your directory, as well as the single file named txt, to the command lp (the file named txt in
your directory is passed twice to lp):
lp * txt
In the preceding example, the lp command first prints the files represented by the *; that is, it
lp command then moves to the second item in the list of files it is to print
prints all files. The
* and txt as a delimiterÐin effect, as a
(Linux interprets the space character between the
comma in an English command). The lp command processes txt as the name of the next file
it is to print.
Tip #92 from
Steve You can use the * symbol anywhere in a string of characters. For example, if you
want to use the ls command to list the names of all files in your current directory
whose names contain the characters rep, you can enter this command:
ls *rep*
Linux lists files with names such as frep.data, report, and janrep. The one exception is that
files with names starting with a period aren't listed. To list files with names starting with a
period (often called hidden files), you must specify the leading period. For example, if you have
a file named .reportrc and want to see it listed, you can enter the following variation of the
preceding command:
ls .*rep*
336 Chapter 16 Understanding Linux Shells
Be careful of using the asterisk wildcard when you're deleting or removing files. The
command rm * removes all files in your directory. An all-too-common mistake is to
accidentally delete all files when you mean to delete a collection of files with a
common suffix or prefix. If, instead of rm *txt (which would remove all files with
names ending in txt), you enter rm * txt, Linux first deletes all files and then
attempts to delete a file named txt. But at that point, no files are left.
To be safe, you should use the -i option with rm if you use the asterisk for filename
completion. The rm -i *txt command prompts you for confirmation before each file
is deleted.
The ? Wildcard
?
You use the question mark ( ) wildcard to represent a single character. Suppose that you have
the filesreport1, reportb, report10, reportb3, report.dft, and report.fin in your current
directory. You know that the lp rep* command prints all the files, but to print just the first
two (report1 and reportb), you can enter the following command:
lp report?
To list the names of all files whose names are three characters long and end with the character
x, enter the following:
ls ??x
This command lists a file with the name tax but not trax.
Because the question mark represents a single occurrence of any character, the string ???
represents all files consisting of just three letters. You can generate a list of files with three-
letter extensions by using the string *.???. For example, if you're searching a directory
containing graphic images as well as other data, the following command lists all files with
extensions such as .tif, .jpg, and .gif, as well as any other files with three-letter extensions:
ls *.???
Tip #93 from
Steve Remember that Linux isn't MS-DOS; filenames aren't limited to eight characters with a
three-character extension. Also, remember that filenames are case sensitive under
Linux.
The [] Expression
Sometimes you must be more selective than either of the more general-purpose wildcards
allow. Suppose that you want to select the files job1, job2, and job3, but not jobx. You can't
337
Understanding Shell Command Parsing
select the right files by using the ? wildcard because it represents one occurrence of any
character. You can, however, use job[123] as the file descriptor.
You can also represent a single character by enclosing a range of characters within a pair of
square brackets. To list the names of all files that begin with an uppercase letter, for example,
you can enter the following:
ls [A-Z]*
Part
Suppose that you have files named sales.90, sales.91, sales.92, and sales.93 and want to
copy the first three to a subdirectory named oldstuff. Assuming that the subdirectory III
oldstuff exists, you could enter the following command: Ch
cp sales.9[0-2] oldstuff 16
Like the question mark, items inside square brackets ( []) represent exactly one character. You
can describe a discrete series of permissible values, such as [123], which permits only the
characters 1, 2, or 3; you can also describe a range of characters, as in [A-Z], which represent
any character between uppercase A and uppercase Z, inclusive.
You can also specify a set of ranges, which incorporates more than one range. For example, if
you want to specify only alphabetic characters, you can use [A-Z,a-z]. The ASCII character
set contains special characters between ASCII Z and ASCII a; if you specify [A-z], you
include those special characters in your request.
Connecting Processes with Pipes
Frequently, you need to use the output of one program or command as the input of another.
Rather than enter each command separately and save results in intermediate files, you can
connect a sequence of commands by using a pipe ( ). |
For example, to sort a file named allsales and then print it, you can enter the following:
sort allsales | lp
The name pipe is appropriate. The output of the program on the left of the pipe (the vertical
bar) is sent through the pipe and used as the input of the program on the right. You can
connect several processes with pipes. For example, to print a sorted list of the data in all files
with names that begin with sales, you can enter the following command:
cat sales* | sort | lp
Redirecting Input and Output
Many programs expect input from the terminal or keyboard; many programs send their
output to the terminal screen. Linux associates keyboard input with a file named stdin; it
associates terminal output with a file named stdout. You can redirect input and output so that
rather than come from or go to the terminal, it comes from a file or is sent to a file.
You use the < (less than) symbol to redirect input into a command or program so that it
comes from a file instead of the terminal. Suppose that you want to send a file named info by
email to someone whose address is sarah. Rather than retype the contents of the file to the
338 Chapter 16 Understanding Linux Shells
mail command, you can give this command to use the info file as the input ( stdin) to the
mail command:
mail sarah < info
You use the > (greater than) symbol to redirect the output of a program to a file. Instead of
going to the terminal screen, the output is put into a file. The command date displays the
current time and date on the terminal screen. If you want to store the current time and date in
a file named now, for example, you can enter this command:
date > now
If the filename on the right side of the > already exists, it is overwritten. Be careful not
to destroy useful information this way.
If you want to append, or concatenate, information to an existing file, you use the two-
character >> symbol. To append the current date to a file named report, for example, you can
enter the following command:
date >> report
For a slightly more lengthy example, suppose that the file named sales consists of sales data.
The first field of each line contains a customer ID code. The first command line shown here
puts the output of thedate command into a file named sales_report. The second command
sales file as input to the sort command and appends the output to the
line uses the
sales_report file. The last line sends the sales_report file to users sarah and brad by email:
date > sales_report
sort < sales >> sales_report
mail sarah brad < sales_report
Be careful not to redirect the same file as both input and output to a command. Most
likely, you'll destroy the contents of the file.
Table 16.5 summarizes the redirection symbols used in Linux.
Table 16.5 Linux's Redirection Symbols
Symbol Meaning Example
< Take input from a file mail sarah < report
> Send output to a file date > now
>> Append to a file date >> report
339
Understanding Shell Command Parsing
Substituting Shell Variables
You learned about shell variable expansion earlier in this chapter when you set your PATH
variable to PATH=$PATH:newpath. The shell replaced $PATH with the current values of the PATH
variable. Shells are really interpreted languages, almost like BASIC; the shell variable is the
primary object manipulated. Because shell variables are frequently manipulated, each shell
provides methods of testing and defining the shell variables. Part
Shell variables are stored as strings. When two variables are placed together, their respective
III
strings are concatenated. For example, if you have two variables, X=hello and Y=world, the Ch
expression $X$Y results in the string helloworld. If you give the following command, the shell 16
parses the two parameters, and the values of X and Y (the two strings hello and world) are
substituted before being passed to the echo command:
echo $X $Y
The echo command then prints hello world.
If you place a dozen tab characters between $X and $Y, the output results are still the
same.
If the substitution can be ambiguous, the shell picks the most obvious substitutionÐoften
with unpredictable results. For example, if you type echo $XY, the shell substitutes helloY. If
you also have a variable XY, its value is substituted instead. To get around these ambiguities,
you can use a simple shell mechanism to define exactly what you mean. If you type ${X}Y, the
shell substitutes the value of X before appending the character Y to the string.
The Bourne and Korn shells have a rich collection of shell-variable expansion techniques that
perform various tests on the variable before making the substitution. See the man pages for sh
and ksh for more details.
Substituting Command Results
After the shell performs its substitution of variables, it scans the line again for commands to
be run before the command line is finally ready. Command substitution means that Linux
substitutes the results of a command for a positional parameter. This is specified in the
following way:
command-1parameter ' command-2'
'
Be careful in the use of apostrophes, or single quotes ( ), and back quotes ( Á, also known as
grave accents). Table 16.6 lists what each mark does.
340 Chapter 16 Understanding Linux Shells
Table 16.6 Quotation Marks and Apostrophes
Symbol Meaning
" Quotation marks disable filename generation and suppress parameter expansion;
however, shell-variable and command substitution still take place.
' The apostrophe disables all parsing. Whatever is enclosed within the apostrophes
is passed on as a single parameter.
The back quoteor grave accent, implies command substitution. Whatever is
Á
enclosed within back quotes is executed as though the command were performed
on a line by itself. Any output placed on the standard output then replaces the
command. The command line is then parsed again for parameters.
Consider the following command line:
echo Today\'s date and time are 'date'
It produces this output:
Today's date and time are Mon May 17 14:35:09 EST 1999
To make the echo command behave properly, you must precede the 's in Today's in the
\
preceding command with a backslash ( ), also called the escape character ( Today\'s). Virtually
every nonalphanumeric character on your keyboard has some special meaning to the shell. To
use any of the special characters in a string and to prevent the shell from interpreting the
character, you must ``escape'' the character; that is, you must precede it with the backslash. If
you want to pass the backslash character itself, you can use \\. To pass a dollar sign to a
command, you use \$.
Regular Expressions
A regular expression is a series of standard characters and special operators. Regular
expressions are useful for searching for a string of characters in a file. Regular expressions are
often used with the grep family of tools: grep, egrep, and fgrep, but are also used with other
UNIX commands.
The simplest kind of regular expression is a string. A string is a set of characters, such as and.
The syntax for a grep command is as follows:
grep stringfilename
For example, to search for the word hand in a specific file named michael, you enter this
command:
grep hand michael.txt
This command might return the result
on the other hand, michael has been working hard this past
if that were the only line of the text file that contained the word hand. grep returns every line
of a text file that has a match to the string.
341
Understanding Shell Command Parsing
Regular expressions use special characters. The special characters used with regular
. *
expressions are the period ( ), asterisk ( ), square brackets ( []), slash (/), caret (^), and dollar
$
sign ( ). Table 16.7 summarizes these special characters and their behavior in regular
expressions.
Table 16.7 Special Characters in Regular Expressions
Character Description Part
III
. The period matches a single character, unless the single character is a line
return. For example, b.d matches bad and bod. Ch
* The asterisk matches zero or more of the preceding regular expression. So the
16
pattern 4* matches no 4s, one 4, two 4s, and so on.
[] The brackets are used to group a set of multiples for matches. Remember that
unlike DOS, UNIX is case sensitive. So to search for all instances of the name
Michael, you could use [Mm]ichael to search for both michael and Michael, but
not MICHAEL. If you want to search for an actual ] character, either you can use
[]], or you can use the backslash as an escape character to treat the right
bracket as a text character, like so: /]. A dash inside brackets acts as a range, so
[a-j] is the same as [abcdefghij].
/ The slash is used to escape the special behavior of these special characters and
treat them as text to be searched for in a string. So * matches everything, but /*
matches only a line with the * character in it. You use // to search for a slash
character, of course.
^ If ^ is at the beginning of the string, it matches a line only if the string is at the
beginning of the line. So if you have a text file of telephone numbers sorted by
area code, the regular expression ^704 matches all telephone numbers with the
area code 704, but not a telephone number such as 407-555-7043.
$ If a $ is the last character in a regular expression, it matches the expression to a
line in the file if the expression is at the end of the file.
You can define how many of a given character to match by using the curly braces ( {}). For
example, the command
g\{3,4}
matches any line in the text file that contains either ggg or gggg.
If you maintain a large file of older mail, the command
grep 'whatever' ~/mail/*
searches for the string whatever in the mail directory. This capability is useful if you recall
that your recent acquaintance Dave Quigman included his telephone number in his sigfile,
but you can't remember what folder you saved his message into. The command
grep 'Quigman'' ~/mail/*
matches all instances of the name.
342 Chapter 16 Understanding Linux Shells
However, a more efficient way might be to match the telephone number itself. Say his
telephone number is in the area code 408. So the command
grep '408.[0-9]\{3\}.[0-9]\{4\}' ~/mail/*
finds all telephone numbers that start with 408. Notice the periods separating the 408, the
[0_9]\{3\}, and [0_9]\{4\}. The period matches any single character, which allows you to
match the telephone number 408-555-1212 or 408.555.1212 because some people use
periods to separate the telephone numbers.
Understanding Command Groups, Subshells, and Other Commands
You terminate a simple command by using a carriage return. If you want to place more than
one command on the command line before pressing Return, you can delimit individual
;
commands by using a semicolon ( ), thus forming a group of commands. When the shell
parses the command line, it treats the semicolon as an end-of-line character. If you type the
following string, the shell executes each command sequentially as though you had typed each
on a line by itself:
command-1;command-2;command-3
For example, you can enter clear;ls to clear your screen and display a directory listing.
Command Groups
If you want to redirect input or output to all the commands as a group, you can do so by
making the command line a command group. A command group is defined as any number of
commands enclosed in braces ( {}). For example, the following command string directs the
output of both commands to the file named output-file:
{command-1;command-2} > output-file
You also can use any form of redirection. The output of a command group can be piped, as in
the following example:
{command-1;command-2} | command-3
In this case, the output of command-1 is fed into the pipe, the output of command-2 is then fed
into the same pipe, and command-3 sees just one stream of data.
Commands executed in a command group run in the current shell. That means that
they may modify the environment or change directories.
Subshells
When you run a series of commands as a command group, those commands run in the
current shell. If one of the commands modifies the environment or changes the directory, the
343
Doing Background Processing
changes are in effect when the command group finishes running. To avoid this problem, you can
run a command group in a subshell.
A subshell is a clone of the present shell, but because child processes can't modify the
environment of their parent process, all commands run in a subshell have no effect on the
environment when the command group finishes. To run a command group in a subshell, you
can replace the braces with parentheses. The command-group example in the preceding
section then becomes the following:
Part
(command-1;command-2) | command-3
III
Only command-3 runs in the current shell, but the output of the subshell is piped into the
Ch
standard input of command-3.
16
Doing Background Processing
Because Linux is a multitasking operating system, you can run commands in the background
in several ways. The simplest form of background processing allows you to run a command
concurrently with a command in the foreground. Other methods place commands deeper and
deeper in the background.
Arranging for Processes to Run in the Background
The shell allows you to start one process and, before the first one completes, start another.
When you do so, you put the first process in the background. You put a process in the
&
background by using the ampersand ( ) character as the last character on the line containing
the command you want to run in the background. Consider the following command:
sort sales > sales.sorted &
If you enter this command, you see a number onscreen. This number is the process ID (PID)
number for the process you put in the background. The PID is the operating system's way of
identifying that process.
Normally, when you run a command, the shell suspends operation until the command is
complete. If you append the ampersand to the end of a command string, the command string
runs concurrently with the shell. If you place the ampersand aftera command string, the shell
resumes operation as soon as the background command is launched. Unless you use I/O
redirection with the background command, the background command and the present shell
expect input from and produce output to your terminal. Unless your background command
takes care of I/O itself, the proper syntax for background processing is as follows:
command-string [input-file] output-file &
For example, to copy a collection of files whose names end with the characters .txt to a
subdirectory named oldstuff and, without waiting for that process to finish, print a sorted
list of the data in all files with names that begin with sales, you can use the following two
commands:
344 Chapter 16 Understanding Linux Shells
cp *.txt oldstuff &
cat sales* | sort | lp
Tip #94 from
Steve You can put jobs in the background when you don't want to wait for one program to
finish before starting another. You can also put jobs in the background when you
have a collection of tasks in which at least one can run on its own. Just start that one
and put it in the background.
You can also use the virtual terminals offered by Linux to execute a command and
then log in to another terminal; then switch between the virtual terminals with
Ctrl+Alt+Fx.
Because the background process is a child of your shell, it's automatically killed when you log
out. All child processes are killed when their parent dies.
Using the nohup Command
To place a command deeper in the background than the & operator allows, you can use the
nohup command which stands for no hang up). The nohup command takes as its argument a
command string. However, you must use nohup with the & operator if you want the command
to actually be placed in the background. If a command is run with nohup in the foreground,
the command is immune to being killed when you disconnect your terminal or hang up a
modem (its original purpose). The syntax for the nohup command is as follows:
nohup command-string [input-file] output-file &
Using the cron Daemon
If you run a command with the nohup command, the command executes immediately. If you
want to run the command at a later time or on a ``time-available'' basis, you must invoke the
services of the cron daemon.
The cron daemon is a command run in the background by LinuxÐor, more specifically, by
init, the master program. cron provides scheduling services to all Linux processes. You can
ask cron to run a program at a specific time, periodically, at a particular time every day, or
?
whenever the load on cron permits.
See ``Scheduling Commands with cron and crontab,'' p. 373
The at Command
The at command expects a time or date as a parameter and takes any number of command
strings from its standard input. When the at command detects an end-of-file marker, it
creates a Bourne shell script for execution at the time you specified.
345
Doing Background Processing
Tip #95 from
Steve The at command is flexible about the types of dates and times it accepts. For
example, if you enter the command at now + 1 day, the next commands, taken from
the standard input, are executed tomorrow at this time.
One way to use the at command is from within a shell script. A shell script is nothing more
Part
than a file containing all the commands necessary to perform a series of commands. The
III
name of the file then becomes your own addition to the Linux command language. One way
of using the at command is shown here: Ch
at now + 1 day 16
command-1
command-2
When placed in a shell script, these lines let you conveniently run one or more commands the
next day. To run any number of different commands, you can simply enter new commands
after the at command line. You can run any number of commands from this script.
The batch Command
The batch command is the logical equivalent of at now. If you attempt to use the at now
command, you see an error message that says something along the lines of now has passed.
The batch command works exactly as at now works if it were logically possible, with one
minor exception: The cron daemon maintains a separate queue for commands generated by
at, batch, and cron. Suppose that you entered the following commands into the file named
backup:
tar -cvf tackettbkup /usr/home/tackett
Then you can tell the system to back up the directory /usr/home/tackett by using this
command:
?
batch backup
See ``Creating Your First vi File,'' p. 208
The crontab Command
One of the best uses of thecron daemon is in automating the maintenance of a system. With
cron, you, as the system administrator, can set up automatic backups of your system every day
at 4 a.m., Monday through Saturday. You install, delete, and list commands you want run in
this fashion by using the crontab command.
To run commands periodically, you must create a file in the crontab format. The crontab file
consists of six fields separated by spaces or tabs. The first five fields are integers specifying
minute (00-59), hour (00-23), day of the month (01-31), month of the year (01-12), and day of
the week (0-6, with 0 referring to Sunday). The sixth field is a command string. Each numeric
field can contain an inclusive range of numbers (such as 1-5 to indicate Monday through
346 Chapter 16 Understanding Linux Shells
Friday) or discrete sets of numbers (such as 0,20,40 to indicate that an instruction should be run
every 20 minutes). A field can also contain an asterisk to indicate all legal values.
The following example runs the calendar command every 20 minutes, starting at midnight
Monday and ending at 11:40 p.m. Friday:
0,20,40 * * * 1-5 calendar -
Tip #96 from
Steve cronfile,
If you name the file you can install it in the cron system by issuing the
command crontab cronfile.
The cron daemon has a time granularity of one minuteÐmeaning, the shortest time duration
you can work with is one minute. You, as system administrator, can place limits on the
number of commands allowed to be run at any one time. Just because you ask cron to run an
at, batch, or crontab file doesn't mean that it runs at precisely the time you've indicated.
Understanding Command Feedback
Linux provides instant feedback for commands that abort for one reason or another. In most
cases, errors are limited to misspellings of the command name or badly formed filenames. If
you attempt to run a nonexistent command, Linux replies with the following message:
command: command not found
If you try to use a nonexistent filename, Linux responds with this message:
command: file: No such file or directory
If the error is caused by something other than a command-line error, the command itself
usually reports what happenedÐalthough not always in an easily decipherable form.
If you try to run a command with nohup and haven't redirected the standard error, Linux
automatically places any error messages in a file named nohup.out in the directory from
which the command was run.
Because commands run by cron have less urgency, any errorsÐindeed, any output placed on
the standard output and not redirectedÐare sent to you through email.
Editing and Aliasing Shell Commands
Different shells include features that provide shortcuts for running commands. Command
editing lets you modify commands that have already been typed in. By using Linux's command
history feature, you can recall commands you've previously entered. Aliasing lets you create
commands that represent other commands. Command completion lets you fill in the rest of a
filename after you type part of it.
347
Editing and Aliasing Shell Commands
Editing Commands
Command editing means that after you type a commandÐand before you press ReturnÐyou
can edit or change parts of the command without having to retype most of it. To edit a
command, press Esc to get into editing mode, and then use any of the line-movement
commands from the vi editor to modify the command. You can press Backspace to return to
the portion of the command you want to change, and use other vi commands, such as x to
delete a character, r to replace a character, and so on. Part
III
Viewing Command History
Ch
The command history feature allows you to look back at previously entered commands and
recall them. This feature saves you the time and trouble of retyping commands. When you 16
combine this feature with command editing, you can easily correct mistakes in complicated
commands and deal effectively with some repetitive tasks.
In both shells, the history command displays the list of past commands the shell has saved.
The commands are numbered. To execute command 10, for example, you enter ! 10. The
bash shell also takes advantage of your PC's arrow keys; you can recall previous commands by
pressing the _ key.
Aliasing Commands
Command aliasing allows you to define a name for a command. Consider this example: The
man command displays Linux documentation, or man pages. To make the word help an alias,
or alternative, for man, you can enter the following:
alias help=man
Now you can enter help cp or man cp to display Linux man pages about the cp command.
You also can use aliases with commands that have options or arguments. For example, if you
want to list the names of all the files in the current directory sorted in descending order by the
time they were last modified (so that the most recent files are at the bottom of the list), you
can use this command:
ls -art
The ls command is the command to list files. The -a -r option
option specifies all files; the
arranges the files in reverse, descending order; and the-t option sorts by time last modified.
That's a lot to remember. You can assign the alias timedir to this complex command by using
the following command:
alias timedir="ls -art''
The quotation marks ( "") are necessary because the shell expects the alias for timedir to be
terminated by a space or Return. Now, if you enter timedir, you get the directory listing you
want.
348 Chapter 16 Understanding Linux Shells
Setting an alias from the command line keeps that alias in effect only for the current
session. To have the alias active whenever you log in, you can include the alias
definition in the .profile file if you use the Bourne shell; keep it in the .login file
if you use the C shell.
Completing Commands
Command completion allows you to type the beginning of a filename and then press the Tab
key to expand the filename. This technique can save time and spelling mistakes when you're
entering a command. If two files share a common prefix, Linux expands the command to the
last common character, stops expanding the filename, and then beeps. You need to provide
the unique filename.
Adding Text with Cut and Paste
Red Hat Linux offers a program that you can start at boot time; this program allows you to
use the mouse to select text from anywhere onscreen and then paste the text onto the
command line for the shell to interpret. To get a mouse cursor, you simply press one of the
mouse buttons. You then select the desired text from anywhere onscreen by first clicking on
the beginning of the text and, while holding down the button, dragging the cursor to the
desired end point of the text. After you select the text, you right-click to copy the text to the
command line.
Working with Shell Scripts
The shell accepts commands, interprets them, and arranges for the operating system to
execute the commands in the manner you specify. In the preceding sections, you saw how the
shell interprets special characters to complete filenames, redirects input and output, connects
processes through pipes, and puts jobs or processes in the background.
You can type commands at the terminal, or they can come from a file. A shell script is a
collection of one or more shell commands in a file. To execute the commands, you type the
name of the file. The advantages to this approach include the following:
& You don't have to retype a sequence of commands.
& You determine the steps to accomplish a goal once.
& You simplify operations for yourself and others.
By using variables and keywords, you can write programs that the shell can interpret. This
capability is useful because it allows you to create general shell scripts you or others can use in
various situations.
349
Working with Shell Scripts
Suppose that after you log in, you regularly like to see who's logged in to your system, run a
program named calendar that displays your appointments for today and tomorrow, and print
the current date and time to the screen. To do all that, you enter the following commands:
who
calendar
date
If you put these three commands into a file named whatsup and make that file executable, you Part
have a shell script that you can execute just like any other command. The file whatsup must be
III
a text file. You can use the vi or Emacs text editor to put the commands in the whatsup file.
To make the file executable, you enter this command: Ch
chmod +x whatsup 16
The chmod command modifies or sets the permissions for a file. The +x option makes the file
executable; that is, it makes the file work just like a standard Linux command. Putting
commands into the file and making the file executable are both one-time operations. From
that point on, you can enter whatsup to execute your shell script. You can use the shell script
just like any other command. For example, to print the results of the whatsup command, you
can enter the following:
whatsup | lp
To put the results of the whatsup command into a file named info for future reference, you
can enter the command below:
whatsup > info
As a review, follow these steps to create a shell script that you can use whenever you want:
1. Use a text editor, such as vi or Emacs, to put the shell commands into a text or ASCII
file. In the preceding example, the commands were put in the file named whatsup.
2. Make sure you have execute permission on the file. To do so, use chmod +x filename
(for example, chmod +x whatsup).
3. Test the command by typing the name of the command and pressing Return.
After using this process a few times, you'll see how easily you can create useful scripts. Of
course, the hardest part is figuring out which shell commands to use and how to use the
shell's programming capabilities to express the steps you need to carry out.
You can test a shell script and see all the steps it goes through by entering this command:
sh -x script-name
In this syntax, script-name is the name of the file that holds the script you're considering.
The sh -x command displays all the steps the script goes through and is useful when you're
trying to debug a script.
350 Chapter 16 Understanding Linux Shells
Writing Programs with the Shell
To write programs that use the shells, you must know about variables and control structures.
Don't let either term scare you. A variable is an object that, at any one time, has one of
possibly many different values assigned to it. Control structures specify the way you can
control the flow of execution of a script. The two basic types of control structures are
decision structures (such as if...then...else structures or case structures), and iterative
structures or loops (such as a for or while loop). With a decision structure, you choose a
course of action from one or more alternatives, usually depending on the value of a variable or
the outcome of a command. With an iterative structure, you repeat a sequence of commands.
The earlier section ``Setting the Shell Environment'' describes shell variables; the later
section ``Programming with Control Structures'' provides more information on control
structures.
Using echo
You can use the echo command to display informative messages about what's happening in a
shell script. The echo command displays its argumentsÐthat is, whatever follows the word
echoÐonscreen. Putting a string of characters in quotation marks ensures that all the
characters are displayed. You also can redirect the results of echo to a file.
The command
echo ``Please stand by ...''
displays the following line on the terminal screen:
Please stand by ...
The following command puts Please stand by ... in the file named messg:
echo ``Please stand by ...'' > messg
Tip #97 from
Steve Using the echo command can make your users feel as though something is
happening when they enter a commandÐa particularly good idea if the command
doesn't give any output for several seconds or longer.
The echo command is also useful when you want to trace a shell script. Using the echo
command at key points tells you what's happening in a script. The following is the file
whatsup with an echo command or two added:
echo '' Let's see who is on the system.''
who
echo '' Any appointments? ''
calendar
date
echo '' All done''
351
Working with Shell Scripts
When you run the whatsup file, you see the following:
$ whatsup
Let's see who is on the system.
sarah tty01 Dec 20 08:51
brad tty03 Dec 20 08:12
ernie tty07 Dec 20 08:45
Any appointments?
12/20 Sales meeting at 1:45
12/21 party after work! Part
Mon Dec 20 09:02 EST 1993
All done III
$ Ch
16
Using Comments
After you write a shell script and don't use it for a while, you might forget what the shell
script does or how it accomplishes its task. To avoid this situation, you should put comments
in your shell scripts to explain the purpose of the task and how the task is achieved. A comment
is a note to yourself or whoever is reading the script. The shell ignores comments; they're
important to and for human beings.
The pound sign ( #) signals the beginning of a comment to the shell. Every character from the
pound sign to the end of the line is part of that comment. For example, you might comment
the shell script whatsup like this:
# Name: whatsup
# Written: 1/19/97, Patty Stygian
# Purpose: Display who's logged in, appointments, date
echo "Let's see who is on the system."
who # See who is logged in
echo " Any appointments? "
calendar # Check appointments
date # Display date
echo " All done"
If you run the shell script again, you see the same results as before. The comments don't
change the behavior of the shell script in any way.
Using Variables in Shell Programs
To use variables, you must know how to give a variable a value and how to access the value
stored in a variable. Using the value of a variable is straightforward, but you can give a value
to a variable in these four ways:
& Using direct assignment
& Using the read command
& Using command-line parameters
& Substituting the output of a command
352 Chapter 16 Understanding Linux Shells
Using Direct Assignments The most direct way to give a value to a variable is to write an
expression such as this:
[email protected]
This expression gives the variable myemail the value [email protected]. Don't include spaces
=
on either side of the equals sign ( ). The direct-assignment method of assigning a value to a
variable takes the following form:
variable-name=variable-value
If variable-value contains blanks, you enclose the value in quotation marks. To assign an
office address of Room 21, Suite C to the variable myoffice, for example, you use the
following command:
myoffice="Room 21, Suite C''
$
The shell retrieves the value of the variable whenever it sees a dollar sign ( ) followed by the
name of a variable. You can see that when the following two statements are executed:
echo '' My e-mail address is $myemail''
echo '' My office is $myoffice''
Suppose that you frequently copy files to a directory named /corporate/info/public/sales.
To copy a file named current to that directory, you enter this command:
cp current /corporate/info/public/sales
To make using this command easier, you can assign the long directory name to the variable
corpsales by using the following expression:
corpsales=/corporate/info/public/sales
Now, to copy the current file to that directory, you enter the following:
cp current $corpsales
The shell replaces $corpsales with the value of the variable corpsales and then issues the
copy command.
Using the read Command The read command takes the next line of input and assigns it
to a variable. The following shell script extends the preceding corpsales example to ask the
user to specify the name of the file to be copied:
# Name: copycorp
# Purpose: copy specified file to
# /corporate/info/public/sales
corpsales=/corporate/infor/public/sales
echo "Enter name of file to copy" # prompt user
read filename # get file name
cp $filename $corpsales # do the copy
The read command pauses the script and waits for input from the keyboard. When Return is
pressed, the script continues. If Ctrl+d (sometimes represented as ^D) is pressed while the
read command is waiting for input, the script is terminated.
353
Working with Shell Scripts
Using Command-Line Parameters When the shell interprets a command, it attaches
variable names to each item on the command line. The items on the command line are the
sequences of characters separated by blanks or tab characters. (You use quotation marks to
signal that a collection of characters separated by spaces represents one item.) The variables
attached to the items in the command line are $0, $1, $2, and so on through $9. These 10
variables correspond to the positions of the items on the line. The command name is $0, the
first argument or parameter for the command is $1, and so on. To demonstrate this concept,
named shovars:
Part
consider the following sample shell script
III
# Name: shovars
# Purpose: demonstrate command-line variables Ch
echo $0
echo $2 $4! 16
echo $3
Now suppose that you enter this command:
shovars -s hello ``look at me'' bart
The output of the shell script is this:
shovars
hello bart!
look at me
In this output, the first line is the command's name (variable $0), the second line is made up of
the second and fourth arguments (variables $2 and $4), and the last line is the third argument
(variable $3).
Following is a more serious example. This shell script deletes a file but first copies it to the
directory /tmp so that you can retrieve it if necessary:
# Name: safrm
# Purpose: copy file to directory /tmp and then remove it
# from the current directory
# first copy $1 to /tmp
cp $1 /tmp
# now remove the file
rm $1
If you enter safrm abc def, only the file abc is removed from the current directory because the
safrm shell script deletes only variable $1. You can, however, represent all the parameters on
the command line by using $*. You can make safrm more general by replacing each
occurrence of $1 with $*. If you then enter safrm abc def xx guio, all four files ( abc, def, xx,
and guio) are removed from the current directory.
Substituting the Output of a Command You can assign to a variable the result of an
executed command. To store the name of the current working directory in a variable named
cwd, for example, you enter the following:
cwd=Ápwd Á
Notice that pwd, the print working directory command, is set in back quotes instead of single
quotation marks.
354 Chapter 16 Understanding Linux Shells
The following shell script changes the name of a file by appending the current month, day,
and year to the filename:
# Name: stamp
# Purpose: rename file: append today's date to its name
# set td to current date in form of mmddyy
td='+%m%d%y'
# rename file
mv $1 $1.$td
In this example, the variable td is set to the current date. In the final line, this information is
appended to variable $1. If today is February 24, 1997, and you use this script on a file called
myfile, the file is renamed (moved) to myfile.022497.
Using Special Characters in Shell Programs
You've seen how the shell gives special treatment to certain characters, such as >, *, ?, $, and
others. What do you do if you don't want those characters to get special treatment? This
section provides a few answers.
You can use quote to make the shell ignore special characters. You enclose the character
string with a pair of single quotes, as in this example:
grep '^Mary Tuttle' customers
The result of this grep Mary Tuttle in the file
command is that the lines that begin with
customers ^ grep to search from the beginning of the line. If
are displayed. The caret ( ) tells
you don't enclose the text Mary Tuttle in single quotes, it might be interpreted literally (or as
a pipe symbol on some systems). Also, the space between Mary and Tuttle isn't interpreted by
the shell when it occurs within the single quotes.
You can also use quotation marks to make the shell ignore most special characters, with the
exception of the dollar sign and back quote. In the following example, the asterisks, spaces,
and the greater-than sign are treated as regular characters because the string is surrounded by
quotation marks:
echo '' ** Please enter your response -->''
In this next example, however, $LOGNAME evaluates correctly, but $5 doesn't have any value:
echo '' >>>Thanks for the $5, $LOGNAME''
\
You use the backslash ( ) to make the shell ignore a single character. For example, to make
the shell ignore the dollar sign in front of the 5, you issue this command:
echo '' >>>Thanks for the \$5, $LOGNAME''
The result is what you expect:
>>>Thanks for the $5, wrev
355
Working with Shell Scripts
Programming with Control Structures
The two primary control structures in shell programming are decision structures and iterative
structures. In decision structures, such as if...then...else and case, you can have the shell
script decide which commands to execute based on the value of an expression (such as a
variable, the properties associated with a file, the number of parameters in a script, or the
result of executing a command). In iterative structures, such as for and while loops, you can
execute a sequence of commands over a collection of files or while some condition holds. Part
The following sections use examples that aren't too complicated yet demonstrate the III
essentials of programming with some control.
Ch
16
Using case
The case structure is a decision structure that lets you select one of several courses of action,
based on the value of a variable. Listing 16.2 shows a short menu program.
Listing 16.2 Implementing Menu Shell Script with case
# Name: ShrtMenu
# Purpose: Allow user to print a file, delete a file,
# or quit the program
# Display menu
echo "Please choose either P, D, or Q to "
echo " [P]rint a file"
echo " [D]elete a file"
echo " [Q]uit"
# Get response from user
read response
# Use case to match response to action
case $response in
P|p) echo "Name of file to print?"
read filename
lp $filename;;
D|d) echo "Name of file to delete?"
read filename
rm $filename;;
*) echo "leaving now";;
esac
The syntax of the case statement is as follows:
case word in
pattern) statement(s);;
pattern) statement(s);;
...
esac
The word parameter is matched against each pattern parameter, starting with the pattern at
the top of the list. The statements that execute if word matches a pattern are terminated by
two semicolons ( ;;). The end of the case statement is marked by the word esac (that's case
spelled backward).
356 Chapter 16 Understanding Linux Shells
In Listing 16.2, the pipe character is used to give a choice for a match. For example, P|p
means that either an uppercase or lowercase letter P is considered a match.
The pattern * is used to represent all other patterns not explicitly stated. If users press any key
besides P, p, D, or d, they exit from the menu.
Listing 16.3 uses a case statement that makes a selection based on the number of parameters
the shell represents as $#.
Listing 16.3 Command-Line Parsing with case
# Name: recent
# Purpose: list the most recent files in a directory
# If user types recent <Return> then the names of
# the 10 most recently modified files are displayed
# If the user types recent n <Return> then the names of
# the n most recently modified files are displayed
# Otherwise, user is notified of incorrect usage
#
# Case based on number of parameters
case $# in
0) ls -lt | head ;;
# ls -lt lists names of file in order of
# most recently modified
# head displays the first 10 lines of a file
1) case $1 in
[0-9]*) ls -lt | head -$1 ;;
*)echo "Usage: recent number-of-files";;
esac;;
*) echo "Usage: recent number-of-files";;
esac
Finding the Exit Status
When a shell command executes, it's either successful or not. If you use the command grep
``American Terms'' customers to see whether the string American Terms is in the file
customers and the file exits, the following is true:
& You have read permission to the file
& American Terms is in the file
& The shell command has executed successfully
If any of those conditions isn't true, the shell command executes unsuccessfully.
The shell always reports back about the status of the termination of a command, program, or
shell script. The value reported back is called the exit status of a command and is represented
by the variable #?. If you enter the following commands, you see the value of $?:
grep ``American Terms'' customers
echo $?
357
Working with Shell Scripts
If $? has a value of 0, this command was successful; otherwise, the command was
unsuccessful.
The following is an example in which the exit status of the command who|grep $1 is used in
the case statement:
Part
# Name: just.checking
# Purpose: Determine if person is logged in III
# Usage: just.checking login_name
# Ch
case 'who | grep $1 > /dev/nullCË in
16
0) echo "$1 is logged in.";;
*) echo "$1 is not here. Try again later.";;
esac
echo "Have a great day!"
If you enter just.checking rflame and rflame is logged in, you see the following:
rflame is logged in.
Have a great day!
If rflame isn't logged in, you see this response instead:
rflame is not here. Try again later.
Have a great day!
Using if Structures
The if...then...else...fi structure is a decision structure that allows you to select one of
two courses of action based on the result of a command. The else portion of the structure is
optional. One or more commands go in place of the ellipsis (...). Provided that the exit status
of the last command following the if is zero (that is, the command executed successfully), the
commands following the then and preceding the else (if one is included) are executed.
Otherwise, the commands following the else are executed.
In other words, one or more commands are executed. If the last command is successful, the
commands in the then portion of the statement are performed, and then the commands
following the fi (the end of the structure) are executed. If the last commands aren't
successful, the commands after the else are performed.
This familiar example behaves exactly the same as when it was written using the case
statement:
# Name: just.checking
# Purpose: Determine if person is logged in
# Usage: just.checking login_name
#
if
who | grep $1 > /dev/null
then
echo "$1 is logged in."
else
358 Chapter 16 Understanding Linux Shells
echo "$1 is not here. Try again later."
fi
echo '' Have a great day!"
Using the test Command
Many of the shell scripts used in this chapter expect users to behave nicely. The scripts have
no check to see whether users have permission to copy or move files or whether what the
users were dealing with was an ordinary file rather than a directory. The test command can
deal with these issues as well as some others. For example, test -f abc is successful if abc
exists and is a regular file.
You can reverse the meaning of a test by using an exclamation point in front of the option.
For example, to test that you don't have read permission for file abc, you use test ! -r abc.
Table 16.8 lists several options for the test command.
Table 16.8 Options for Using the test Command with Files
Option Meaning
-f Successful if the file exists and is a regular file
-d Successful if the file is a directory
-r Successful if the file exists and is readable
-s Successful if the file exists and isn't empty
-w Successful if the file exists and can be written to
-x Successful if the file exists and is executable
Listing 16.4 is an example of the use of the test command.
Listing 16.4 A Sample Script That Uses the test Command
# Name: safcopy
# Purpose: Copy file1 to file2
# Check to see we have read permission on file1
# If file2 exists then
# if file2 is a file we can write to
# then warn user, and get permission to proceed
# else exit
# else
# copy file
#
# Check for proper number of arguments
case $# in
2) if test ! -r $1 # cannot read first file;;
then;;
exit (1) # exit with non-zero exit status;;
fi;;
if test -f $2 # does second file exist?;;
then;;
if test -w $2 # can we write to it?;;
359
Working with Shell Scripts
Listing 16.4 continued
then;;
echo " $2 exists, copy over it ? (Y/N)";;
read resp # get permission from user;;
case $resp in;;
Y|y) cp $1 $2;; # go ahead;;
*) exit(1);; # good bye!;;
esac;;
else;; Part
exit (1) # Second file exists but can't write;;
fi III
else # Second file doesn't exist; go ahead and copy!;
cp $1 $2;; Ch
fi;; 16
*) echo "Usage: safcopy source destination";;
exit (1);;
esac
You can also use the test command to test numbers. To determine whether a value in the
variable hour is greater than 12, for example, you use test $hour -gt 12. Table 16.9 lists some
options you can use with test when you're comparing numbers.
Table 16.9 Options for Using test When Comparing Numbers
Option Meaning
-eq Equal
-ne Not equal
-ge Greater than or equal
-gt Greater than
-le Less than or equal
-lt Less than
Listing 16.5 shows these options used to display a timely greeting.
Listing 16.5 Displaying a Greeting with the test Command
# Name: greeting
# Purpose: Display Good Morning if hour is less than 12
# Good Afternoon if hour less than 5PM
# Good Evening if hour is greater than 4PM
# Get hour
hour='date +%H'
# Check for time of day
if test $hour -lt 12
then
echo "Good Morning, $LOGNAME"
else
if test $hour -lt 17
then
360 Chapter 16 Understanding Linux Shells
Listing 16.5 continued
echo "Good Afternoon, $LOGNAME"
else
echo "Good Evening, $LOGNAME"
fi
fi
Using Iterative Structures
Iterative control structures allow you to write shell scripts that contain loops. The two basic
types of loops are for and while loops.
With for loops, you specify a collection of files or values to use with some commands. To
copy all the files whose names end with the characters .txt to the directory textdir, for
example, you can use the following for loop:
for i in *.txt
do
cp $i textdir/$i
done
The shell interprets the statement for i in *.txt and allows the variable i to take on the
name of any file in the current directory whose name ends with .txt. You can then use the
variable $i with any statements between the do and the done keywords.
The script in Listing 16.6 prints a collection of files, each with its own banner page. It also
sends mail to the user concerning the status of the print requests. The characters $* represent
all the parameters given to the shell command.
Listing 16.6 Processing Files with the for Command
# Name: Prntel
# Purpose: Print one or more files
# each with own title page
# Notify user which files were sent to the printer
# and which were not.
# Do this for all parameters to the command
for i in $*
do
if lp -t $i -dlasers $i > /dev/null
then
echo $i >> printed
else
echo $i >> notprinted
fi
done
# end of loop
if test -s printed
then
echo "These files were sent to the printer " > mes
cat printed >> mes
mail $LOGNAME < mes
rm mes printed
361
Customizing Linux Shells
Listing 16.6 continued
fi
if test -s notprinted
then
echo "These files were not sent to the printer " >mes
cat notprinted >> mes
mail $LOGNAME < mes
rm mes notprinted
fi Part
III
A while loop looks at the exit status of a command in the same way the if statement looks at
the status. The script in Listing 16.7 notifies users when they've received new mail. The script Ch
makes the assumption that if a mailbox changes, a user has new mail. The script uses the
16
command diff to compare two files and then reports on the differences. If the files are the
same, the exit status is zero (the command is successful).
Listing 16.7 Repeating Commands with while
# Name: checkmail
# Purpose: Notify user if their mail box has changed
# Suggestion: Run this in the background
# get a size of mail box for comparison
cp $MAIL omail # Get set for first time through
# MAIL is a "special" variable indicating the user's mailbox
# while omail and $MAIL are the same, keep looping
while diff omail $MAIL > /dev/null
do
cp $MAIL omail
sleep 30 # sleep, pause for 30 seconds
done
# There must be a change in the files
echo ``New mail!!'' | write $LOGNAME
You can see that some of the commands and concepts used with if...then...else
statements can be transferred to while loops. The difference, of course, is that with while
loops, you're dealing with an iterative, repetitive process.
Customizing Linux Shells
The shell starts when you log in. Tables 16.2 and 16.3 show you that special variables are
given values by the shell to help define your shell environment. The shell sets some of these
variables. You can change these settings and give other variables values by editing the file
.profile if you're using the Bourne or bash shell. If you're using the C shell, you set the
variables by editing the file .login. You can also use command aliasing to define aliases for
commands.
Whenever you issue a command, a new shell starts; it inherits many of the characteristicsÐor
much of the environmentÐof the existing shell. Note these two things about the new shell:
362 Chapter 16 Understanding Linux Shells
& The new shell runs in your current directory. The pwd command returns the same
value within a shell as it gives before the shell was started.
& The new shell receives many of its variables from the existing shell. You can make sure
that variables set in the existing shell are exported to the new shell in different ways.
Exporting Variables to the New Shell
When you create shell variables or give values to existing variables, they exist in the running
shell. A variable set in the login shell is available to all command-line arguments. A variable
set within a shell has that value only within that shell. The value disappears or is reset when
you exit that shell.
For example, enter these two commands from the command line:
today=Thursday
echo $today
Suppose that the echo command displays Thursday. Now suppose that you write and execute
the following shell script named whatday:
# Name: whatday
# display the current value of the variable today
echo ``Today is $today.''
# set the value of today
today=Friday
# display the current value of the variable today
echo ``Today is $today.''
Now enter the following four commands from the command line:
chmod +x whatday
today=Thursday
whatday
echo $today
The following lines then appear onscreen:
Today is .
Today is Friday.
Thursday
The value of the variable today Thursday. When you execute the shell
in the login shell is
script whatday, you see that initially the variable today isn't defined (as shown by the display
Today is .). Then the today variable has the value Friday in the shell. When the whatday
script terminates, you return to the login shell and today has its original value, Thursday.
To give the variable today the same value that it has in the login shell when the shell script
whatday starts, you can use the command export. This command ``exports,'' or passes on, the
variables from one shell to subsequent shells:
export today
Now any shell started from the login shell inherits the value of the variable today. You can
add the export command to the preceding sequence of commands as follows:
363
Customizing Linux Shells
today=Thursday
export today
whatday
echo $today
You then see the following output:
Today is Thursday.
Today is Friday.
Thursday Part
Notice that the value the variable receives in the shell started by the whatday script isn't III
carried back to the login shell. Exportation or inheritance of variable values goes in only one
Ch
directionÐfrom a running shell down to the new shell, never back up. That's why when you
change your current directory inside one shell, you're back to the place you started when that 16
shell terminates.
You can export any variable from one shell down to another shell by using the following
syntax:
export variable-name
In this syntax, variable-name is the name of the variable you want to export. To change your
terminal type from its current setting to a vt100, for example, you can enter the following
commands to make the new value of TERM available to all subsequent shells or programs:
TERM=vt100
export TERM
When you change or set bash shell variables in the .profile file, be sure to export them. For
example, if you want the PATH variable to be PATH=/bin:/usr/bin:/usr/local/bin:., you can
set it in the .profile file and follow it with this export command:
export PATH
To change the shell prompt, you must set a value for PS1 in the file .profile. To change it
from $ to Ready $, for example, you can use a text editor to put these lines in the file named
.profile:
PS1="Ready $''
export PS1
Tip #98 from
Steve Changes you make to .profile or .login don't take effect until you log out and log
in again.
Defining Command Aliases
Command aliases are useful for defining commands you use regularly but for which you don't
want to bother remembering the details. Command aliases are also useful for enhancing your
364 Chapter 16 Understanding Linux Shells
working environment with a set of useful tools. This command assigns the alias recent to a
command that lists the 10 most recently modified files in the current directory:
alias recent="ls -lat|head''
To avoid typing your command aliases each time you log in, you can put them in the .login
file if you're using the C shell or the .profile file if you're using bash or a similar shell. The
command aliases are now available to you when you're in your shell.
Troubleshooting
The shell is the primary interface between you and the Linux operating system. Although a
shell can be almost any executable program, several standard shells are supplied with Linux or
are freely available in source code (written in C) or already compiled for your machine. All
Linux shells can be viewed as highly sophisticated, special-purpose programming languages
containing all the usual constructs found in a programming language. The special purpose of
Linux shell languages is to tie together the many small commands and utilities found in the
Linux environment.
I set an environment variable, but nothing changed.
Did you remember to export the environment variable after you set it?
I changed my .profile configuration file , but nothing happened.
Changes you make to the .profile or .login configuration files don't take effect until you
log out and log back in.
I am trying to use a shell script, and it's not doing what it's supposed to.
The shell script may have been written for a different shell and may need to be rewritten. For
a simple fix, you might be able to add lines to the beginning and end of the shell script to set
the shell to the shell the script was originally written for and then set it back to the shell you
commonly use. Possibly, the script was written in a nonshell scripting language such as Perl
or python. In this case, you might need to set the path to the Perl, python, or other script
engine or define it in the script to be somewhere Perl isn't located on your system.
CHAPTER
17
Managing Multiple Processes
In this chapter by Jack Tackett, Jr.
Understanding Multitasking 366
Initiating Multiple Processes 368
Using the Scheduling Commands 370
Reporting On and Monitoring the Multitasking Environment 376
Controlling Multiple Processes 383
366 Chapter 17 Managing Multiple Processes
Understanding Multitasking
Linux is a multiuser and multitasking operating system. Multiuser means that several people
can use the computer system simultaneously (unlike a single-user operating system, such as
MS-DOS). Multitasking means that Linux, like Windows NT, can work on several tasks
concurrently; it can begin work on one task and take up another before the first task is
finished.
Taking care of several user requests and multitasking are the jobs of the operating system.
Most systems have only one CPU and one collection of chips that make up main memory, or
RAM. A system may have more than one disk or tape drive for secondary memory and several
input/output devices. All these resources must be managed and shared among several users.
The operating system creates the illusion that each user has a dedicated computer system.
As mentioned earlier, it's Linux's job to create the illusion that when you make a request, you
have the system's undivided attention. In reality, hundreds of requests may be handled
between the time you press Enter and the time the system responds to your command.
Imagine having to keep track of dozens of tasks simultaneously. You have to share the
processing power, storage capabilities, and input and output devices among several users or
several processes belonging to a single user. Linux monitors a listÐalso known as a queueÐof
tasks waiting to be done. These tasks can include user jobs, operating system tasks, mail, and
background jobs such as printing. Linux schedules slices of system time for each task. By
human standards, each time slice is extremely shortÐa fraction of a second. In computer
time, a time slice is adequate for a program to process hundreds or thousands of instructions.
The length of the time slice for each task may depend on the relative priority of each task.
Linux works on one task from the queue for a while, puts the task aside to begin work on
another task, and so on. It then returns to the first task and works on that task again. Linux
continues these cycles until it finishes a task and takes the task out of the queue, or until the
task is terminated. In this arrangement, sometimes called time-sharing, the resources of the
system are shared among all the tasks. Naturally, time-sharing must be done in a reliable and
efficient manner. The UNIX term for a task is process. Table 17.1 shows several types of
processes.
Table 17.1 Types of Processes
Process Type Description
interactive Initiated by a shell and running in the foreground or background
batch Typically a series of processes scheduled for execution at a specified point in
time
daemon Typically initiated at boot time to perform operating system functions on
demand, such as LPD, NFS, and DNS
367
Understanding Multitasking
You've already seen that you can put or run a program in the background. While the program
runs in the background, you can continue entering commands and working with other
material. This is a feature of multitasking: Linux uses the time-sharing method to balance
your immediate commands and the ones running in the background. This chapter shows
?
other ways to schedule processes so that they can run without your attention ( batch processes).
See ``Doing Background Processing,'' p. 343
The Linux operating system has the primary responsibility of handling the details of working
with several users and several processes. As a user, you have the power to specify which
programs you want to run. Some Linux commands let you specify when you want a process to
start. You also can monitor your processes as well as see what other processes are running. In
some cases, you can change their relative priority. And you can always terminate your
processes if the need arises. If you're the system administrator, you have all these capabilities,
Part
plus the responsibility and power to initiate, monitor, and manage processes that belong to
the operating system or any user. III
Ch
Table 17.2 lists the commands that make it possible to control the multiuser and multitasking
capabilities of Linux. 17
Table 17.2 Multiuser and Multitasking Commands
Command Action
at Executes commands at a given time
batch Executes commands when the system load allows
cron Executes scheduled commands
crontab Maintains crontab files for individual users
kill Stops processes
nice Adjusts the priority of a process before it starts
nohup Allows a process to continue after you log out
ps Displays process information
renice Adjusts the priority of a running process
w Shows you the users who are logged in and what they're doing
who Displays the system's logged-in users
368 Chapter 17 Managing Multiple Processes
For more information on the commands in Table 17.2, you can consult the following
man page:
man command
You also can use the -help option:
command -help
Initiating Multiple Processes
You can start running a program by entering its name. You can also start programs from files
that contain shell commands. Running programs can interact with many different parts of the
system. A program can read from or write to files; manage its information in RAM; or send
information to printers, modems, or other devices. The operating system also attaches
information to a process so that the system can keep track of and manage it.
A process is a running program but is different from a program. In one sense, a process is
more than a program because a program is only a set of instructions; a process is dynamic
because it uses the resources of a running system. On the other hand, a single Linux program
can start several processes.
Linux identifies and keeps track of processes by assigning a process ID number (PID) to each
process.
Starting Multiple Processes
You've already seen that your login shell is always running. Whenever you enter a command,
you start at least one new process while the login shell continues to run. If you enter the
following command, for example, the file named report.txt is sent to the lp program:
?
lp report.txt
See ``Understanding Shells,'' p. 319
When the lp program completes its task, the shell prompt reappears. However, before the
shell prompt reappeared, the login shell and the lp command were running; you initiated
multiple processes in that case. The shell waited until the lp command finished before
putting the shell prompt back onscreen.
Starting a Background Process
You can run a process as a background job by giving the command to start a process and
&
placing an ampersand ( ) after the command. For example, if you enter the command , the
shell responds immediately with a numberÐthe PID for that process. The shell prompt
reappears without waiting for the process to complete. The following is a sample of what you
would see:
369
Initiating Multiple Processes
$ lp report.txt &
3146
$
In this example, 3146 is the PID of the process started by the lp command.
Regardless of whether you run the lp command in the background, the process associated
with lp is started from the current shell. The lp process is a child process of the current shell.
This example points to a common relationship between processesÐthat of parent and child.
Your current shell is the parent process, and the running lp process is a child process.
Usually, a parent process waits for one or more of its child processes to complete before it
continues. If you want the parent to continue without waiting for the child to finish, attach
&
the ampersand ( ) to the command that spawns, or initiates, the child process. You can
continue with other work or commands while the child runs.
Part
If you're working from a character terminal or a remote login, your current shell is III
usually your login shell. However, if you're using a virtual terminal or a terminal
Ch
window from a GUI, a separate shell is associated with each session.
17
Using Pipes to Start Multiple Processes
Another way to start multiple processes is to use one or more pipes on a command line. To
print a long listing of the 10 most recently modified files in your current directory, enter this
command:
ls -lt | head | lp
This command starts three processes simultaneously, and they're all children of the current
|
shell. A pipe works this way: Commands on either side of the vertical bar ( ) begin at the same
time. Neither is the parent of the other; they're both children of the process that was running
when they were created. In this sense, you can think of commands on either side of the pipe
symbol as sibling processes.
Some programs are written so that they themselves spawn several processes. One example is
the ispell command, which lists the words in a document that Linux can't find in a system
dictionary. The ispell command spawns some child processes. Suppose you enter the
following:
ispell /usr/doc/HOWTO/Commercial-HOWTO ]] final.errs &
You'll see the following results displayed:
[1]1286
$
Here, [1] indicates the number of background processes you have started and 1286 is the PID
of the background process, in this case ispell; the $ prompt indicates that the shell is ready
to handle another command from you. Even though ispell may spawn some children and
370 Chapter 17 Managing Multiple Processes
wait for them to complete, you don't have to wait. In this example, the current shell is the parent of
ispell, and ispell's children can be thought of as grandchildren of the login shell. Although
a parent can wait for its children, a grandparent doesn't.
All these examples show how it's possible for users to start multiple processes. You can wait
until child processes are finished before continuing or not. If you continue without waiting
for child processes to complete, you make the children background processes. The following
sections look at some Linux commands you can use to schedule processes to run at specified
times or at a lower relative priority.
Using the Scheduling Commands
The Linux environment provides many ways to handle command execution. Linux lets you
create lists of commands and specify when they're to be run. The at command, for example,
takes a list of commands typed at the keyboard or from a file and runs them at the time
specified by the command. The batch command is similar to the at command, except that
batch runs commands when the system finds time for them rather than allows users to specify
a particular time. The cron command allows for commands to be run periodically, and the
crontab command allows users to edit the files used by cron.
All scheduling commands are useful for running tasks at times when the system isn't too busy.
They're also good for executing scripts to external servicesÐsuch as database queriesÐat
times when it's least expensive to do so.
Running Commands at Specified Times with at
To schedule one or more commands for a specified time, use the at command. With this
command, you can specify a time, a date, or both. The command expects two or more
arguments. At a minimum, you specify the time you want the command(s) executed and the
command(s) you want to execute.
The following example performs its job at 1:23 a.m. If you're working in the wee hours of the
morning before 1:23 a.m. (that is, between midnight and 1:23 a.m.), the command is done
today, at 1:23 a.m. Otherwise, it's done at 1:23 a.m. the following day. The job prints all files
in the directory /usr/sales/reports and sends a user named boss some mail announcing that
the print job was done at 1:23 a.m. Type the following commands on the terminal, pressing
Return at the end of each line. After you enter each line, press Ctrl+d to finish the command.
at 1:23
lp /usr/sales/reports/*
echo ``Files printed, Boss!'' | mail -s"Job done'' boss
cron jobs, discussed later in this chapter, are the most commonly used mechanisms
for running automated system administration jobs under Linux. However, you must
be the root user to create and edit cron job entries. The at command allows anyone
to run tasks even if he or she does not have root privileges.
371
Using the Scheduling Commands
? See ``Setting the Terminal Environment,'' p. 322
Commands to be scheduled by at are entered as a list of commands on the line following the
at command.
After you terminate the at command, you see a display similar to the following:
job 756603300.a at Tues July 27 01:23:00 1999
This response indicates that the job will execute at 1:23 as specified. The job number,
756603300.a, identifies the job. If you decide you want to cancel the job, do so by using the
job number associated with it, like so:
at -d 756603300.a
If you have several commands you want to schedule by using at, it's best to put them in a file.
If the filename is getdone, for example, and you want to schedule the commands for 10 a.m., Part
type one of the following:
III
at 10:00 [[ getdone
at 10:00 -f getdone Ch
<
Remember that the less-than symbol ( ) indicates the use of the contents of the getdone file as 17
input to the at command. By using the -f option, you can specify the command file without
using redirection.
You can also specify a date for an at job. For example, to schedule a job at 5 p.m. on July 27,
enter these commands:
at 17:00 July 27
lp /usr/sales/reports/*
echo ``Files printed, Boss!'' | mail -s"Job done'' boss
The jobs you schedule with at are put into a queue that the operating system checks
periodically. You don't have to be logged in for the job to be executed. The at command
always runs in the background, freeing resources but still accomplishing the job. Any output
produced by the commands in your at job is automatically mailed to you.
To see which jobs you scheduled with at, enter at -l. Working with the preceding examples,
you see the following results:
job 756603300.a at Tue July 27 01:23:00 1999
job 756604200.a at Tue July 27 17:00:00 1999
Only your at jobs are listed.
To remove a scheduled at job, enter at -d followed by the job number. To remove the
second job just listed, for example, enter this command:
at -d 756604200.a
Table 17.3 summarizes the different ways to use the at command.
372 Chapter 17 Managing Multiple Processes
Table 17.3 Summary of at Commands
Format Action
at hh:mm Schedules the job at the hour ( hh) and minute ( mm) specified,
using a 24-hour clock
at hh:mmmonthdayyear Schedules the job at the hour ( hh), minute ( mm), month, day,
and year specified
at -l Lists scheduled jobs; an alias for the atq command
at now + count time-units Schedules the job right now plus the count number of time-
units; time units can be minutes, hours, days, or weeks
at -d job_id Cancels the job with the job number matching job_id; an alias
for the atrm command
As the root user, you can use any of these commands; for other users, the files /etc/at.allow
and /etc/at.deny determine the permission to use the commands. If /etc/at.allow exists,
only the usernames listed in the file are allowed to use the at command. If the /etc/at.allow
file doesn't exist, the system checks /etc/at.deny, and every username not mentioned in
/etc/at.deny is allowed to use at (in other words, any user listed in /etc/at.deny isn't
allowed to use at). If neither file exists, only the superuser (root) can use at. If /etc/at.deny
is empty, every user can use at.
Running Long Tasks with batch
Linux offers more than one command for scheduling tasks. The preceding section describes
the at command, which gives you the power to dictate when a task will run. However, it's
always possible that the system can be loaded down with more jobs scheduled at one time
than it can handle comfortably. The batch command lets the operating system decide an
appropriate time to run a process. When you schedule a job with batch, Linux starts and
works on the process whenever the system load isn't too great. Jobs run under batch execute
in the background, just as those run with at. In fact, batch is an alias for at -b in Red Hat
Linux.
Tip #99 from
Jack It's useful to put commands you want to run with at or batch in a file so that you
don't have to retype the commands each time you want to run the jobs. To use
batch to schedule the commands in the file getdone, enter the following command:
batch < getdone.
The format forbatch commands is to enter the list of commands on the lines following the
batch command; you terminate the list of commands by pressing Ctrl+d. You can put the list
of commands in a file and then redirect the input of the file to batch. To sort a collection of
373
Using the Scheduling Commands
files, print the results, and notify the user named boss that the job is done, enter the following
commands:
batch
sort /usr/sales/reports/* | lp
echo ``Files printed, Boss!'' | mailx -s"Job done'' boss
The system returns the following response:
job 7789001234.b at Fri Feb 21 11:43:09 1997
The date and time listed are the date and time you pressed Ctrl+d to complete the batch
command. When the job is complete, check your mail; anything that the commands normally
display is mailed to you.
Scheduling Commands with cron and crontab
Part
Both at and batch schedule commands on a one-time basis. To schedule commands or
processes on a regular basis, you use the cron program. You specify the times and dates you III
want to run a command in crontab files. Times can be specified in terms of minutes, hours,
Ch
days of the month, months of the year, or days of the week.
17
The cron program is started only once, when the system is booted. Individual users shouldn't
have permission to run cron directly. Also, as the system administrator, you shouldn't start
cron by typing the name of the command; you should list cron in a shell script as one of the
commands to run during a system boot-up sequence.
When started, cron (short for chronograph) checks queues for at jobs to run and also checks to
see whether users or the root have scheduled jobs by using crontab files. If cron doesn't have
anything to do, it ``goes to sleep'' and becomes inactive; it ``wakes up'' every minute, however,
to see whether it needs to run commands. You can see how important and useful this facility
is; also, cron uses very few system resources.
Tip #100 from
Jack While cron uses few system resources itself, the programs cron spawns can
overload a system. Thus, try to schedule resource intensive jobs for off-peak hours.
Also try to schedule similar jobs, such as those that use the find command, at the
same off-peak time to lesson the overall impact.
You can use crontab to install a list of commands that will be executed on a regular schedule.
The commands are scheduled to run at a specified time (such as once a month, once an hour,
once a day, and so on). The list of commands to be performed on the specified schedule must
be included in the crontab file, which is installed with the crontab command. After you
install the crontab file, cron reads and executes the listed commands at the specified times.
Also, with the crontab command, you can view the list of commands included in the file and
cancel the list if you want.
374 Chapter 17 Managing Multiple Processes
Before you install your crontab file with the crontab command, create the file containing the
list of commands you want to schedule by using the crontab-e option.
Each user has only one crontab file, created when the crontab command is issued. This file is
placed in a directory that's read by the cron command.
Linux stores the user's crontab file in the /usr/spool/cron/ directory in a file named for the
user's login name. If your username is mcn, and you use a text editor to create a file called
mycron and install it by typing crontab mycron, the file /usr/spool/cron/mcn is created. (In
this example, the mcn file is created, or overwritten, with the contents of mycron, which might
contain entries that launch one or more commands.)
For users to use the crontab command, they must be listed in the /etc/cron.d/
cron.allow file. If you add a user to the system from the command line (by using
the useradd command), he or she isn't added automatically to the /etc/cron.d/
cron.allow file. As the root user, you must add the new user to the cron.allow
file with a text editor.
Although you can initially create your crontab file with a text editor, after you create
your crontab file, you can modify it by using only the crontab-e command. Don't
try to replace or modify the file that cron examines (that is, the /usr/spool/cron/
crontabs/user file) by any means other than by using the crontab command.
Each line in the crontab file contains a time pattern and a command. The command is
executed at the specified time pattern. The time pattern is divided into five fields separated by
spaces or tabs. Any output that usually appearsÐthat is, information that isn't redirected to
stdout or stderrÐis mailed to the user.
The following is the syntax for the commands you enter in a file to be used by crontab:
minute hour day-of-month month-of-year day-of-week command
The first five fields are time option fields. You must specify all five of these fields. You can,
*
however, use an asterisk ( ) in a field if you want to ignore that particular field.
crontab field means ``any valid value'' instead of ``ignore
Technically, an asterisk in a
the value"Ðthat is, match anything. The crontab entry 02 00 01 * * date, for
example, says to run the date command at two minutes after midnight (zero hour)
on the first day of the month. Because the month and day of the week fields are both
asterisks, this entry runs on the first day of every month and any day of the week that
the first of the month happens to land on.
375
Using the Scheduling Commands
Table 17.4 lists the time-field options available with crontab.
Table 17.4 Time-Field Options for the crontab Command
Field Range
minute 00 through 59
hour 00 through 23 (midnight is 00)
day-of-month 01 through 31
month-of-year 01 through 12
day-of-week 01 through 07 (Monday is 01, Sunday is 07)
You can have as many entries as you want in a crontab file and can designate them to run at
Part
any time. This means that you can run as many commands as you want in a single crontab
file.
III
To sort a file named /usr/wwr/sales/weekly and mail the output to a user named twool at
Ch
7:30 a.m. each Monday, use the following entry in a file: 17
30 07 * * 01 sort /usr/wwr/sales/weekly |mail -s"Weekly Sales'' twool
This command specifies the minute as 30, the hour as 07, any day of the month with the
asterisk, any month of the year with another asterisk, and the day of the week as 01 (which
represents Monday).
Notice the pipe between the sort and mail commands in the preceding example. The
command field can contain pipes, semicolons, arrows, or anything else you can enter on a
shell command line. At the specified date and time, cron runs the entire command field with a
standard shell (bash).
To specify a sequence of values for one of the first four fields, use commas to separate the
values. Suppose you have a program, chkquotes, that accesses a service that provides stock
quotes and puts the quotes in a file. To get those quotes at 9 a.m., 11 a.m., 2 p.m., and 4 p.m.
on Monday, Tuesday, and Thursday of every weekÐand definitely on the 10th of March and
SeptemberÐuse the following entry:
* 09,11,14,16 10 03,09 01,02,04 chkquotes
You can put the command lines into a file by using vi or some other editor that allows you to
save files as text files. Assume that you put your commands in a file named cronjobs. To use
crontab to put the file where cron can find it, enter this command:
crontab cronjobs
Each time you use crontab this way it overwrites any crontab file you may have already
launched.
The crontab command has three options:
376 Chapter 17 Managing Multiple Processes
& The -eoption edits the contents of the current crontab file. (The -e option opens your
file by using the ed editor or whatever editor is assigned to the EDITOR variable in your
?
shell.)
See ``Setting the Shell Environment,'' p. 325
& The -roption removes the current crontab file from the crontabs directory.
& The -loption lists the contents of the current crontab file.
In all these cases, crontab works with the crontab file that has your login name. If your login
name is mcn, your crontab file is /usr/spool/cron/crontabs/mcn. The crontab command
names your file automatically.
The system administrator and users share responsibility for making sure that the system is
used appropriately. When you schedule a process, be aware of the impact it may have on the
total system. Linux allows you, as the system administrator, to grant access to the at, batch,
and cron commands to all users, specific users, or no users (or to deny access to individual
users).
Reporting On and Monitoring the Multitasking
Environment
Youknow that Linux is a multiuser, multitasking operating system. Because so many people
can do so many things with the system at the same time, users find it useful to determine
who's using the system and what processes are running, as well as to monitor processes.
Knowing that others can keep track of the commands you enter is important. Most users can't
access your files without your permission, but they can see the names of commands you enter.
Also, you (as the system administrator) or someone else who has the root password can peruse
all the files on the system.
Although you don't have to be paranoid about privacy on a Linux system, you should know
that the system can be monitored by anyone who wants to take the time to do it. The
information you can gain about what's going on in the system is more useful than just
satisfying curiosity: By seeing what jobs are running, you can appropriately schedule your
tasks. You can also see whether a process of yours is still active and whether it's behaving
properly.
Finding Out Who's on the System with who
The purpose of the who command is to find out who's logged in to the system. The who
command lists the login names, terminal lines, and login times of users currently logged in.
The who command is useful in many situations. If you want to communicate with someone on
the computer by using the write command, for example, you can find out whether that
person is on the system by using who. You can also use who to see when certain users are
logged in to the computer to keep track of their time spent on the system.
377
Reporting On and Monitoring the Multitasking Environment
Using who to List Users Logged In to the System
To see everyone who's currently logged in to the system, you can enter who. You then see a
display similar to the following:
$ who
root console Dec 13 08:00
ernie tty02 Dec 13 10:37
bkraft tty03 Dec 13 11:02
jdurum tty05 Dec 13 09:21
ernie ttys7 Dec 11 18:49
$
This listing shows that root, ernie, bkraft, and jdurum are now logged in. It shows that root
logged in at 8 a.m., bkraft at 11:02, and jdurum at 9:21. You can also see that ernie is logged
in to two terminals and that one login occurred at 6:49 p.m. (18:49) two days earlier (which
may be some reason for concern, or it might just be ernie's usual work habits). Part
III
Using Headers in User Listings
Ch
Several options are available with who, but this chapter describes how to use only two to
17
monitor processes on the system:
-u Lists only users who are currently logged in
-H Displays headers above each column
With these two options, you can get more information about the users now logged in. The
headers displayed with the -H option are NAME, LINE, TIME, IDLE, PID, and COMMENTS. Table
17.5 explains the terms appearing in the heading.
Table 17.5 Output Format for the who Command
Field Description
NAME Lists the user's login name.
LINE Lists the line or terminal being used.
TIME Lists the time the user logged in.
IDLE Lists the hours and minutes since the last activity on that line. A period is
displayed if activity occurred within the last minute of system time. If more than
24 hours has elapsed since the line was used, the word old is displayed.
PID Lists the process ID number of the user's login shell.
COMMENTS Lists the contents of the comment field if comments have been included in /
etc/inittab or if there are network connections.
You probably won't see the COMMENT field filled in very often in any recent Linux
systems. In the old days, processes that let you log in to UNIX (getty or uugetty)
were started directly from entries in the /etc/inittab file and usually listened for
login requests from a particular terminal. The COMMENT field might identify the
378 Chapter 17 Managing Multiple Processes
location of that terminal and could tell you which users were logged in and at what
terminal they were sitting. Today, processes that listen for login requests are typically
handled by the Service Access Facility and are no longer listed in /etc/inittab.
The following example uses the -u and -H options and shows the response Linux returns:
$ who -uH
NAME LINE TIME IDLE PID COMMENT
root console Dec 13 08:00 . 10340
ernie tty02 Dec 13 10:37 . 11929 Tech-89.2
bkraft tty03 Dec 13 11:02 0:04 4761 Sales-23.4
jdurum tty05 Dec 13 09:21 1:07 10426
ernie ttys7 Dec 11 18:49 old 10770 oreo.coolt.com
$
You can infer from this listing that the last session associated with ernie is from a network site
named oreo.coolt.com and that no activity has occurred in that session in more than 24
hours (which might signal a problem). The session for root and the first one for ernie have
both been accessed within the last minute. The last activity on the session for bkraft was four
minutes ago; it has been one hour and seven minutes since any activity was reported on the
session for jdurum.
Also note that this listing includes the PID (process ID number) for the login shell of each
user's session. The next section shows how you can use the PID to further monitor the
system.
Using the finger Command to Learn More About Who Is on the System
A command that complements the who command is finger. To see more information about a
specific user, you can enter finger username (or finger username@domain if the user is on
another computer). For example, to see more information on a user named tackett, you enter
this command:
finger tackett
You then see the following output:
Login: tackett Name: Jack Tackett Jr
Directory: /home/tackett Shell: /bin/tcsh
Office: 2440 SW Cary Parkway 114 Office Phone: 919 555 1212
Home Phone: 919 555 1212
Never logged in.
Mail last read Fri Jul 3 17:42 1998 (EDT)
Plan:
----------------------
Jack Tackett, Jr.
In the immortal words of Socrates:
I drank WHAT?
----------------------
379
Reporting On and Monitoring the Multitasking Environment
This output shows the login and real name associated with the specified account. You can also
see which shell this user prefers to use, his address, when he last read his email, and when he
was last logged in. If he is currently logged on, the finger command tells you how long he
has been logged on and which program he is currently using. The finger command also
displays any information this user may have placed in his .plan file in his home directory.
As you can see, the finger command displays a lot of information about a user; this
information could be used by crackers to hack the system. For this reason, many system
administrators disable the finger command so that others cannot see this information.
If you allow the use of the finger command on your system, or if your system
administrator allows it on a system you are using, you can use the chfn command to
change the information displayed by finger. See the related man page (use the man
chfn command) for more information. Part
III
Ch
17
Reporting On the Status of Processes with ps
The ps (process status) command reports on the status of processes. You can use it to
determine which processes are running, whether a process has completed, whether a process
is hung (having some difficulty), how long a process has run, the resources a process is using,
the relative priority of a process, and the PID needed before you can kill a process. All this
information is useful to a user and very useful to a system administrator. Without any
options, ps lists the PID of each process associated with your current shell. You also can see a
detailed listing of all the processes running on a system.
Monitoring Processes with ps
A common use of the ps command is to monitor background jobs and other processes on the
system. Because background processes don't communicate with your screen and keyboard in
most cases, you use ps to track their progress.
The ps listing displays four default headings as indicators of the information in the fields
below each heading: PID, TTY, TIME, and COMMAND. Table 17.6 explains these headings.
Table 17.6 Headings in the Output of ps
Field Explanation
PID The process identification number
TTY The terminal on which the process originated
TIME The cumulative execution time for the process, in minutes and seconds
COMMAND The name of the command being executed
380 Chapter 17 Managing Multiple Processes
Suppose that you want to sort a file named sales.dat, save a copy of the sorted file in a file
named sales.srt, an mail the sorted file to the user sarah. If you also want to put this job in
the background, you can enter the following command:
sort sales.dat | tee sales.srt | mailx -s"Sorted Sales Data'' sarah &
To monitor this process, enter ps to see a display such as this one:
PID TTY TIME COMMAND
16490 tty02 0:15 sort
16489 tty02 0:00 mailx
16492 tty02 0:00 ps
16478 tty02 0:00 bash
16491 tty02 0:06 tee
16480 tty02 96:45 cruncher
You see the accumulated time and PID for each process started with the command. You also
see information for your login shell (bash) and for ps itself. Notice that all the commands in
the pipe are running at once, just as you would expect (this is the way the piping process
works). The last entry is for a command that has been running for more than an hour and a
half. If that's a problem, you might want to terminate the process by using the kill command
(described later in this chapter). If you enter ps and see only the following listing, the
previous job you put into the background is complete:
PID TTY TIME COMMAND
16492 tty02 0:00 ps
16478 tty02 0:00 bash
16480 tty02 99:45 cruncher
Use ps occasionally to check the status of a command. If, however, you use ps every
second while waiting to see whether the background job is complete, putting the job
in the background doesn't make much sense in the first place.
Obtaining More Information About Processes with ps
Sometimes you need to know more about your processes than what the default ps listing
provides. To generate additional information, you can invoke some of the flags listed in Table
17.7.
Table 17.7 Commonly Used Flags for the ps Command
Flag Description
a Shows processes of other users also.
c Displays the command name from the task_struct environment.
e Shows the environment after the command line and and.
f Shows the ``forest'' family tree format (processes and subprocesses).
381
Reporting On and Monitoring the Multitasking Environment
Table 17.7 Commonly Used Flags for the ps Command
Flag Description
h Specifies no header.
j Indicates the jobs format.
l Indicates the long format.
m Displays memory info.
n Specifies numeric output for USER and WCHAN. WCHAN is the name of the kernel
function where the process is sleeping, with the sys_ stripped from the function
name. If /etc/psdatabase doesn't exist, the number is hexadecimal instead.
r Specifies running processes only.
s Specifies the signal format.
S Adds child CPU time and page faults.
Part
txx Processes associated with tty xx only, where xx is a place holder for the actual III
number of the terminal.
Ch
u Indicates the user format; gives username and start time.
17
v Specifies vm (virtual memory) format.
w Specifies wide output; doesn't truncate command lines to fit on one line.
x Shows processes without controlling terminal.
Tip #101 from
Jack Although many Linux commands require you to preface a flag with the - symbol, you
do not have to do so with the ps command. In fact, you might get a warning message
about using the dash ( -).
The ps command gives only an approximate picture of process status because things can and
do change while the ps command is running. The ps command gives a snapshot of the
process status at the instant ps executed. The snapshot includes the ps command itself.
The following examples show three commands. The first command is the login shell (bash).
The second command is sort, which is used to sort the file named inventory. The third
command is the ps command you're now running.
To find out what processes you're currently running, use the following command:
$ ps
PID TTY TIME COMMAND
65 tty01 0:07 -bash
71 tty01 0:14 sort inventory
231 tty01 0:09 ps
To obtain a full listing, use this command:
382 Chapter 17 Managing Multiple Processes
$ ps -uax
UID PID PPID C STIME TTY TIME COMD
amanda 65 1 0 11:40:11 tty01 0:06 -bash
amanda 71 65 61 11:42:01 tty01 0:14 sort inventory
amanda 231 65 80 11:46:02 tty01 0:00 ps -f
If you are familiar with UNIX systems such as Solaris, then these flags are different,
and you will have to remember which system you are using because each uses
different flags to provide the same information. For example, to get a full listing under
many System V versions of UNIX, you use the ps -ef command.
Notice a few things about this full listing. In addition to the PID, the PPID is listed. The
PPID is the process ID number of that process' parent process. In this example, the first
process listed, PID 65, is the parent of the following two. The entry in the fourth column (the
column headed C) gives the amount of CPU time a process has used recently. In selecting the
next process to work with, the operating system chooses a process with a low C value over one
with a higher value. The entry in the STIME column is the time at which the process started.
To monitor every process on the system and get a full listing, enter ps -uax.
Tip #102 from
Jack For more system resource information, use the top command. top provides an
ongoing look at processor activity in real time. It displays a listing of the most
CPU-intensive tasks on the system and can provide an interactive interface for
manipulating processes. It can sort the tasks by CPU usage, memory usage, and
runtime.
By piping the command through the grep $LOGNAME command, the processes belonging to
your login name are displayed while all others are filtered out. To see a full listing of all your
processes, enter the following:
ps uax | grep $LOGNAME
To list processes for two terminals (for example, tty1 and tty2), use the following command:
$ ps -t ``1 2''
PID TTY TIME COMMAND
32 tty01 0:05 bash
36 tty02 0:09 bash
235 tty02 0:16 vi calendar
In this example, the -t option is used to restrict the listing to the processes associated with
terminals tty01 and tty02. Terminal tty02 is running the shell command (PID 32) and using
vi to edit the calendar (PID 235). The cumulative time for each process is also listed. If you're
383
Controlling Multiple Processes
using shells from a graphical interface (the xterm command), you can use device names pts001,
pts002, and so on with the -t option to see the processes from those sessions.
Tip #103 from
Jack The /proc file system contains much of the information reported by programs such
as ps. Thus, you can read the information directly from proc without using other
programs. For example, to check on memory usage, you type the following
command:
cat /proc/meminfo
Sometimes a process is marked as defunct, which means that the process has terminated and
its parent process has been notified, but the parent hasn't acknowledged that the process is
Part
``dead.'' A process like that is called a zombie process. The parent might be busy with something
else, and the zombie will soon disappear. If you see a number of defunct processes or ones III
that linger for some time, this is a sign of some difficulty with the operating system.
Ch
17
Because a zombie process has no parent, you can't kill the zombie. The only way to
get rid of a zombie process is to reboot your machine.
Controlling Multiple Processes
Linux gives you the power to run several processes concurrently. It also allows a user or an
administrator to have control over running processes. This control is advantageous when you
need to do the following:
& Initiate a process that continues after its parent quits running (use the nohup
command)
& Schedule a process with a priority different than other processes (use the nice
command)
& Terminate or stop a process (use the kill command)
Using nohup with Background Processes
Normally, the children of a process terminate when the parent dies or terminates. This means
that when you start a background process, it terminates when you log out. To have a process
continue after you log out, you can use the nohup commandPut nohup at the beginning of a
command line as follows:
nohup sort sales.dat &
384 Chapter 17 Managing Multiple Processes
This sample command tells the sort command to ignore the fact that you log out of the
system; it should run until the process completes. In this way, you can initiate a process that
can run for days or even weeks. What's more, you don't have to be logged in as it runs.
Naturally, you want to make sure that the job you initiate behaves nicelyÐthat is, eventually
terminates and doesn't create an excessive amount of output.
When you use nohup, the command sends all the output and error messages of a command
that normally appear onscreen to a file named nohup.out. Consider the following example:
$ nohup sort sales.dat &
1252
Sending output to nohup.out
$
The sorted file and any error messages are placed in the file nohup.out. Now consider this
example:
$ nohup sort sales.dat ]] sales.srt &
1257
Sending output to nohup.out
$
Any error messages are placed in the nohup.out file, but the sorted sales.dat file is placed in
sales.srt.
When you use nohup with a pipe, you must use nohup with each command in the
pipe:
nohup sort sales.dat | nohup mailx -s"Sorted Sales Data'' boss &
Scheduling the Priority of Commands with nice
You can use the nice command to run a command at a specific scheduling priority. The nice
command gives you some control over the priority of one job over another. If you don't use
nice, processes run at a set priority. You can lower the priority of a process by using the nice
command so that other processes can be scheduled to use the CPU more frequently than the
nice job. The superuser (the person who can log in as the root user) can also raise the priority
of a process.
The commands nice -help and nice -version don't work in the GNU
implementation ofnice.
385
Controlling Multiple Processes
The general form of the nice command is as follows:
nice - numbercommand
The priority level is determined by the number argument (a higher number means a lower
priority). The default is set to 10, and number is an offset to the default. If the number
argument is present, the priority is incremented by that amount up to a limit of 20. If you
enter the following command, the sort process starts with a priority of 10:
sort sales.dat > sales.srt &
If you want to start another processÐsay, with the lp commandÐbut give preference to the
sort command, you can enter the following:
nice -5 lp mail_list &
To give the lp command the lowest possible priority, enter this:
Part
nice -10 lp mail_list &
III
Ch
number flag in the nice command is preceded by the flag specifier -, which you
17
The
shouldn't confuse with the negative number sign.
Only superusers can increase the priority of a process. To do so, they use a negative number
as the argument to nice. Remember, the lower the nice value, the higher the priority (up to a
maximum priority of 20). To give a job ``top priority,'' a superuser initiates the job as follows:
nice -10 job &
&
The ampersand ( ) is optional; if job is interactive, you don't use the ampersand to place the
process in the background.
Scheduling the Priority of Running Processes with renice
Using the renice command, available on some systems, you can modify the priority of a
running process. Berkeley UNIX systems have the renice command; it's also available in the
/usr/ucb directory in Linux System V systems for compatibility with Berkeley systems. With
renice, you can adjust priorities on commands as they execute. The format of renice is
similar to that of nice:
renice - numberPID
To change the priority on a running process, you must know its PID. To find the PID of all
your processes, enter this command:
ps -e | grep name
In this command, name represents the name of the running process. The grep command
filters out all processes that don't contain the name of the process you're looking for. If
several processes of that name are running, you have to determine the one you want by
386 Chapter 17 Managing Multiple Processes
looking at the time it started. If you want to affect all processes belonging to a certain group or a
certain user, you can specify the GID or UID of the running processes to the renice command.
The entry in the second column of the ps listing is the PID of the process. In the following
example, three processes are running for the current user (in addition to the shell). The
current user's name is pcoco:
$ ps -ef | grep $LOGNAME
pcoco 11805 11804 0 Dec 22 ttysb 0:01 sort sales.dat]]sales.srt
pcoco 19955 19938 4 16:13:02 ttyp0 0:00 grep pcoco
pcoco 19938 1 0 16:11:04 ttyp0 0:00 bash
pcoco 19940 19938 142 16:11:04 ttyp0 0:33 find . -name core -exec rm {};
$
To lower the priority on the process with PID 19940 (the find process), enter the following:
renice -5 19940
As you would expect, the following statements are true about renice:
& You can use renice only with processes you own.
& The superuser can use renice on any process.
& Only the superuser can increase the priority of a process.
Terminating Processes with kill
Sometimes you want or need to terminate a process. The following are some reasons for
stopping a process:
& It's using too much CPU time.
& It's running too long without producing the expected output.
& It's producing too much output to the screen or to a disk file.
& It appears to have locked a terminal or some other session.
& It's using the wrong files for input or output because of an operator or programming
error.
& It's no longer useful.
Most likely, you'll come across a number of other reasons to kill a process as well. If the
process to be stopped is a background process, use the kill command to get out of these
situations.
To stop a command that isn't in the background, press Ctrl+c. When a command is in the
background, however, pressing an interrupt key doesn't stop it. Because a background process
isn't under terminal control, keyboard input of any interrupt key is ignored. The only way
you can stop background commands is to use the kill command.
387
Controlling Multiple Processes
Normal Termination of Background Processes
The kill command sends signals to the program to demand that a process be terminated or
killed. To use kill, use either of the following forms:
kill PID(s)
kill -signal PID(s)
To kill a process whose PID is 123, enter kill 123. To kill several processes whose PIDs are
123, 342, and 73, enter kill 123 342 73.
By using the -signaloption, you can do more than simply kill a process. Other signals can
cause a running process to reread configuration files or stop a process without killing it. Valid
signals are listed by the command kill -l. An average user, however, will probably use kill
with no signal or, at most, with the -9 signal (the I-mean-it-so-don't-ignore-me signal, Part
described in the next section).
III
Ch
Use the correct PID with the kill command. Using the wrong PID can stop a process 17
you want to keep running. Remember that killing the wrong process or a system
process can have disastrous effects. Also, remember that if you're logged in as the
system administrator, you can kill any process.
Stopping a Process by Name
Most Linux distributions provide a command called killall that allows you to give a process name rather
than a PID. For example, if you need to kill a series of processes started by the Web server, you could use the
following command:
killall httpd
If you do not have access to this command, you can simulate its usage with the grep, awk, and xargs
commands as follows:
ps uax | grep httpd | grep -v grep | awk '{print $2}' | xargs kill -9
This command line performs the following actions:
. ps aux lists all active processes.
. grep httpd separates out only those lines that contain the name httpd.
. grep -v grep filters out any lines containing grep commands.
. awk '{print $2}' prints the second field in the resulting list, which is the PID for that process.
. xargs kill -9 pipes each resultant value to the kill -9 command via the xargscommand.
If you successfully kill the process, you get no notice from the shell; the shell prompt simply
reappears. You see an error message if you try to kill a process you don't have permission to
kill or if you try to kill a process that doesn't exist.
Suppose that your login name is chris and that you're now logged in to tty01. To see the
processes you have running, enter ps -f, and you'll see the following response:
388 Chapter 17 Managing Multiple Processes
UID PID PPID C STIME TTY TIME COMMAND
chris 65 1 0 11:40:11 tty01 0:06 -bash
chris 71 65 61 11:42:01 tty01 0:14 total_updt
chris 231 65 80 11:46:02 tty01 0:00 ps -f
chris 187 53 60 15:32:01 tty02 123:45 crunch stats
chris 53 1 0 15:31:34 tty02 1:06 -bash
Notice that the program total_updt is running at your current terminal. Another program,
crunch, is running on another terminal, and it has used an unusually large amount of CPU
time. To kill that process, you can enter kill 187, which may be sufficient. To kill the parent
of that process, enter kill 53.
You might want to kill a parent and child process if you log in as the system administrator and
see that someone has left his or her terminal unattended (if you've set up Linux with remote
terminals). You can kill a clock process that the user has running (the child process) and the
login shell (the parent process) so that the unattended terminal is no longer logged in.
Stopping the parent of a process sometimes terminates the child process as well. To be sure,
stop the parent and its children to halt all activity associated with a parent process. Using the
preceding example, enter kill 187 53 to terminate both processes.
Tip #104 from
Jack If your terminal locks up, log in to another virtual terminal by pressing Alt combined
with a function key (F1-F6), enter ps -ef | grep $LOGNAME, and then kill the login
shell for the locked terminal.
Unconditional Termination of Background Processes
Issuing the kill command sends a signal to a process. Linux programs can send or receive
more than 20 signals, each of which is represented by a number. For example, when you log
out, Linux sends the hang-up signal (signal number 1) to all the background processes started
from your login shell. This signal kills or stops those processes unless they were started with
nohup (as described earlier in this chapter).
Using nohup to start a background process lets the process ignore the signal that tries to stop
it. You might be using programs or shell scripts written to ignore some signals. If you don't
specify a signal when you use kill, signal 15 is sent to the process. The command kill 1234
sends signal 15 to the process whose PID is 1234. If that process is set to ignore signal 15,
however, the process doesn't terminate when you use this command. You can use kill in a
way that a process ``can't refuse,'' however.
The signal 9 is an unconditional kill signal; it always kills a process. To unconditionally kill a
process, use the following command:
kill -9 PID
389
Controlling Multiple Processes
Suppose you enter ps-f and see the following response:
UID PID PPID C STIME TTY TIME COMMAND
chris 65 1 0 11:40:11 tty01 0:06 -bash
chris 71 65 61 11:42:01 tty01 0:14 total_updt inventory
chris 231 65 80 11:46:02 tty01 0:00 ps -f
chris 187 53 60 15:32:01 tty02 123:45 crunch stats
chris 53 1 0 15:31:34 tty02 1:06 -bash
To kill process 187, normally you would enter kill 187. If you then enter ps -f again and see
kill command. To
that the process is still there, you know the process is set up to ignore the
kill it unconditionally, enter kill -9 187. When you enter ps -f again, you see that the
process is no longer around.
Part
A disadvantage to using this unconditional version of the kill command is that kill III
-9 doesn't allow a process to finish what it's doing before it terminates the process. If
you use kill -9 with a program that's updating a file, you could lose the updated Ch
material or the entire file.
17
You should use the powerful kill -9 command responsibly. In most cases, you don't
need the -9 option; the kill command, issued without arguments, stops most
processes.
Termination of All Background Processes
To kill all background jobs, you can enter kill 0. Commands that run in the background
sometimes initiate more than one process; tracking down all the PID numbers associated with
the process you want to kill can be tedious. Because kill 0 terminates all processes started by
the current shell, it's a faster and less tedious way to terminate processes. Enter the jobs
command to see what commands are running in the background for the current shell.
Using kill to Send Signals to Processes
You also can use kill to send nontermination signals to a running process. You can tell a
process to restart or to enter a different logging phase by sending a signal. Linux can send
many different signals, with Table 17.8 listing the more common signals used. To get a
complete list of signals, use the kill -l command.
Table 17.8 Signals Delivered to a Running Process by the kill Command
Number Signal Description
1 SIGHUP Reread configuration files
9 SIGKILL Terminate the process immediately
390 Chapter 17 Managing Multiple Processes
Table 17.8 Signals Delivered to a Running Process by the kill Command
Number Signal Description
10 SIGUSR1 Send the process a program-specific signal
15 SIGTERM Terminate in a normal manner
Looking for PIDs to send a signal to can be an interesting exercise, which is why many
processes record their PIDs in a file located in /var/run. Then, if you need to know their
PIDs, you can look at the appropriate file for that information. For example, to see syslogd's
PID, you can use the following command:
cat /var/run/syslogd.pid
You can use this ability to send a signal directly to the program, as shown here:
kill -SIGHUP 'cat /var/run/syslogd.pid'
This signal then tells the system logger to reread its configuration file and restart.
Troubleshooting
The commands I put in my crontab file don't work.
The cron command runs your crontabentries by using the Bourne Again shell (bash). Your
entries fail if you use shell features not supported by bash. For example, the Public Domain
Korn shell (pdksh) allows you to use either a tilde (~) to represent a home directory or the
alias command to designate aliases for certain commands.
When I try to use the at command, I'm told I don't have permission to use it.
You haven't added your login ID to the /etc/cron.d/at.allow file.
I tried to use the at now command to run a command immediately.
No matter how fast you type, at now always responds with the message ERROR: Too late. The
best alternative is to use the batch command to run the command for you. You can, however,
use at now +5 min to run the command in five minutes. After you press Return, type quickly to
enter your command before the five minutes expire.
CHAPTER
18
Printing
In this chapter by Jack Tackett, Jr.
Selecting a Printer to Work with Linux 392
Knowing What You Need to Configure Printers 392
Knowing How Printing Works Under Linux 393
Understanding the Important Programs for Printing 394
Understanding the Important Directories 396
Understanding the Important Files 397
Understanding the /etc/printcap File 397
Creating a Test printcap Entry 401
Putting It All Together 401
Configuring Red Hat Printers 402
Troubleshooting 405
392 Chapter 18 Printing
Selecting a Printer to Work with Linux
Although everyone thought the computer revolution would bring the paperless office, it
hasn't. More paper is used today than was used 20 years ago. When the UNIX operating
system was in its infancy, Bell Labs used it to produceÐand printÐtechnical documentation.
As a result, UNIX, and thus Linux, has a great many utilities designed around printing (or at
least formatting data to be printed). This chapter concentrates on the mechanics of actually
printing a file.
The printing systems common to BSD UNIX/Linux are called the LPR systems (line printer).
If you can access the printer from MS-DOS, you should be able to print ASCII characters to
the printer from Linux. The only downside is that you might not be able to access certain
features of your printer from Linux. One of the main reasons is that, under Linux, the system
first sends the file to be printed to another file. Linux sends the files to a temporary area
because printers are relatively slow peripherals, and the system doesn't want to slow down
your session just to print a file. This process is called spooling, and printers are thus called
spooled devices. When you print a file in Linux, the file doesn't go directly to a printer; instead,
it goes to a queue to wait its turn to be printed. If your file is the first in the queue, it prints
almost immediately.
Spool is an acronym for Simultaneous Peripheral Operation Off Line. The term was
coined in the early days of the big IBM mainframes, when smaller computers were
used to print reports offline from the mainframe. This technique allowed expensive
mainframes to continue their tasks without wasting time on such trivial matters as
printing.
Because Linux inherits a great deal of UNIX functionality, Linux supports many types of
printers. If you can access your printer from DOS (as mentioned earlier), you should be able
to access the printer from Linux.
Knowing What You Need to Configure Printers
This chapter assumes that you know how to edit a text file under Linux and that you have a
basic understanding of file ownership and permissions. It also assumes that you have your
Linux system set up and running correctly. In particular, if you're going to use remote
printing, your networking subsystems must be installed and operating correctly. Check out
the man pages on the commands chmod and chown for more information. Also review Chapter
9, ``Using the vi Editor,'' for information on using the vi editor because you need to edit
several files when configuring your printers.
393
Knowing How Printing Works Under Linux
Knowing How Printing Works Under Linux
The simplest way to print under Linux is to send the print data directly to the printer device.
The following command sends a directory listing to the first parallel printer (LPT1 in DOS
terms):
ls > /dev/lp
This method doesn't take advantage of Linux's multitasking capabilities because the time
taken for this command to finish is however long the printer takes to actually physically print
the data. With a slow printer or a printer that's deselected or disconnected, you could wait a
long time. A better method is to spool the dataÐthat is, to collect the print data into a file and
then start a background process to send the data to the printer.
Spooling files to be printed later is essentially how Linux works. For each printer, a spool area
is defined. Data for the printer is collected in the spool area, one file per print job. A
background process (called the printer daemon) constantly scans the spool areas for new files
to print. When one appears, the data is sent to the appropriate printer, or despooled. When
more than one file is waiting to be printed, the files are printed in the order they're
completedÐfirst in, first out. Thus, the spool area is effectively a queue, and the waiting jobs
are often referred to as being in the print queue, or queued. In the case of remote printing, the Part
data is first spooled locally as for any other print job, but the background process is told to
III
send the data to a particular printer on a particular remote machine.
Ch
The necessary information that the printer daemon needs to do its jobÐthe physical device to
18
use, the spool area to look in, the remote machine and printer for remote printing, and so
onÐis all stored in a file called /etc/printcap. The details of this file are discussed later in
the section ``Understanding the /etc/printcap File.'' The printcap file is basically a text file
containing the information used by Linux to control the printing device.
Tip #105 from
Jack To Linux, the printer is just another file. But because the printer is a physical piece of
hardware, it has an entry in the /dev directory. Linux likes to treat physical devices as
if they are part of the file system. To list the various devices use the command
ls -l /dev
The term printer is used to mean a printer as specified in /etc/printcap. The term physical
printer is used to mean the thing that actually puts characters on paper. /etc/printcap can
have multiple entries that all describe one physical printer but do so in different ways. If this
point isn't clear to you, read the section on /etc/printcap.
394 Chapter 18 Printing
Understanding the Important Programs for
Printing
Five programs comprise the UNIX print system. By default, they are in the locations shown
in Table 18.1, are owned by root, belong to the group daemon, and have the permissions listed
in the table.
Table 18.1 The Important Printing Programs
File Permissions File Locations
-rwsr-sr-x /usr/bin/lpr
-rwsr-sr-x /usr/bin/lpq
-rwsr-sr-x /usr/bin/lpc
-rwsr-sr-x /usr/bin/lprm
-rwxr-s-- /usr/sbin/lpd
The first four file permissions in Table 18.1 are used to submit, cancel, and inspect print jobs.
/usr/sbin/lpd is the printer daemon. A daemon is a program run by the operating system in
response to an event. A daemon typically sits in the background until requested to do its job.
Tip #106 from
Jack The locations, ownerships, and permissions in Table 18.1 have been simplified and
may be wrong for your system, so note the lpd files and permissions. Use the ls -l
command to list the file permissions. If you need to change the permissions use the
chmod command. If you need to change ownership use the chown command.
All these commands have man pages, which you should consult for more information. The
important point to remember and understand is that by default the commands lpr, lprm, lpc,
and lpq operate on a specific printer called lp0. If you define an environment variable called
PRINTER, the name defined is used instead. You can override lp and the PRINTER environment
variable by specifying the printer name to use on the command line as follows:
lpc -PMYPRINTER
The lpd Daemon
Linux handles all print jobs via the lpd daemon. If this process isn't running, no printing can
take place; print files remain in their spool directories until the lpd process is started. (More
information about spool directories appears later in the section ``Understanding the
?
Important Directories.'')
See ``Understanding Multitasking,'' p. 366
395
Understanding the Important Programs for Printing
If your system doesn't load lpd at startup, or if you must kill and then restart the lpd daemon
for some reason, the following command starts the printer daemon:
lpd [options]
The man page on lpd gives a list of options, but one important option when configuring your
Linux printers is -l, which creates a log file that logs each print request to the system. This
log file can be useful when you're debugging your printing system.
The lpr Command
The lpr command submits a job to the printer or queues a print job. What actually happens
is that the file you specify is copied to the spool directory. Each printer specified for your
Linux system must have its own spool directory. The size of this spool directory is specified in
the minfree file located in each directory. The minfree file specifies the number of disk
blocks to reserve for spooling files to the printer. This is done to keep the lpd daemon from
using up the entire hard drive when spooling a print request.
lpd finds the file, which then takes care of moving the data to the physical printer. If you
don't specify a file, lpr uses standard input.
The lpq Command
Part
The lpq command shows you the contents of the spool directory for a given printer. One
III
important piece of information displayed by lpq is the job ID, which identifies a particular Ch
job. You must specify this number if you want to cancel a pending job. 18
lpq also assigns a number to indicate a rank for each job in the queue (meaning, where the job
is in the queue). active means the file is actually printingÐor at least that lpd is trying to
print it.
The lprm Command
The lprm command removes a job from the queue; that is, it removes unprinted files from the
spool directory. You can specify a job ID (obtained by using thelpq command), or you can
specify - as the job ID to cancel all jobs belonging to you.
If you issue lpq - as root, all jobs for the printer are canceled. If you are root and want to
remove all the jobs belonging to a specific user, you can specify the user's name.
The lpc Command
Using the lpc command, you can check the status of printers and control some aspects of
their use. In particular, lpc lets you start and stop despooling on printers, enable or disable
printers, and rearrange the order of jobs in a print queue. The following commands disable
printing on myprinter, enable the spool queue on yourprinter, and move job number 37 to
the top of the queue:
396 Chapter 18 Printing
lpc down myprinter
lpc enable yourprinter
lpc topq 37
If you invoke lpc without any command arguments, lpc is interactive, prompting you for
actions to take. Some of the more important commands are shown in Table 18.2; read the
man page for complete instructions. Most lpc commands take the name of the printer, as
specified in /etc/printcap, as the parameter.
Table 18.2 Some Common lpc Commands
Command Parameter Description
stop printer Stops the printer, but print requests are still spooled.
start printer Allows the printer to start printing previously
spooled files and any new files spooled to this
printer.
exit, quit (None) Leaves lpc in interactive mode.
status printer Displays the current status of the printer. status
provides such information as whether the queue is
enabled, whether the printer is enabled, and the
number (if any) of jobs now in the queue waiting to
be printed.
Tip #107 from
Jack Keep in mind that some lpc commands are restricted to rootÐthat is, the superuser.
To switch to the root account from a user account, use the switch-user command su.
For security reasons you must have the root password to gain root access.
Understanding the Important Directories
Only one directory is important in printingÐthe spool area where data to be printed is
accumulated before /etc/lpd prints it. However, a system is typically set up with multiple
spool directories, one for each printer, to make printer management easier. For example, my
system is set up to use /usr/spool/lpd as the main spool area, with each separate printer
having a directory under that with the same name as the printer. Thus, a printer named
ps_nff has /usr/spool/lpd/ps_nff as its spool directory.
The spool directories should belong to the daemon group and should be user and group read/
writable and world readable. That is, after you create the directory, you should make sure that
it has the permissions -rwxrwxr-x (0775) with the chmod command. For the directory
myprinter, the appropriate command is as follows:
Understanding the /etc/printcap File
397
chmod ug=rwx,o=rx myprinter
?
chgrp daemon myprinter
See ``File Permissions,'' p. 414
The locations, ownerships, and permissions given here are a simplification and may
be incorrect for your system, so you should take notes on the lpd files and
permissions.
Understanding the Important Files
Apart from the programs discussed so far, each spool directory contains files that have the
permissions -rw-rw-r-:
& The /etc/printcap file contains the printer specifications for each named printer in
your system.
& The .seq file contains the job number counter for lpr to assign.
& The status file contains the message to be reported by lpc stat.
Part
III
& The lock file is used by lpd to prevent itself from trying to print two jobs to the same
Ch
printer at once.
& The errs file logs printer failures. 18
The errs file isn't required by Linux in order to print, but the file must exist for lpd to be
able to log printer failures. You can call the errs file whatever you like as long as you specify
the name in /etc/printcap. You usually create the errs file manually when you set up the
spool area. The section ``Putting It All Together,'' later in this chapter, has more information
on this topic.
Understanding the /etc/printcap File
The /etc/printcap file is a text file that you can edit with your favorite editor.
/etc/printcap should be owned by root and have the permissions -rw-r-r-.
The contents of /etc/printcap typically look very cryptic, but when you know how the file
works, the contents are much easier to understand. To compound the problem, some
distributions don't have a man page for printcap, and most printcap files are created either
by programs or by people with no thought for readability. For your own sanity, make the
layout of your printcap file as logical and readable as possible with lots of comments. And get
the man page from the lpd sources if you don't already have it.
One printcap entry describes one printer. Essentially, a printcap entry provides a logical
name for a physical device and then describes how data sent to that device should be handled.
398 Chapter 18 Printing
For example, a printcap entry defines what physical device is to be used, what spool directory
any data for that device should be stored in, what preprocessing should be performed on the
data, where errors on the physical device should be logged, and so forth. You can limit the
amount of data that can be sent in a single job, or you can limit access to a printer to certain
classes of users. The following shows how a printer is defined in the printcap file:
# Sample printcap entry with two aliases
myprinter|laserwriter:\
# lp is the device to print to - here the first parallel printer.
:lp=/dev/lp0: \
# sd means spool directory - where print data is collected
:sd=/usr/spool/lpd/myprinter:
It's okay to have multiple printcap entries defining several different ways to handle data
destined for the same physical printer. For example, a physical printer might support
PostScript and HP LaserJet data formats, depending on some setup sequence being sent to
the physical printer before each job. It makes sense to define two printers: one that
preprocesses the data by preappending the HP LaserJet sequence and one that preappends
the PostScript sequence. Programs that generate HP data send it to the HP printer, whereas
programs generating PostScript print to the PostScript printer.
If you don't designate a default printer via an environment variable or don't specify a
printer on the lpr command line, Linux routes the print job to the lp0 printer. Thus,
you should specify one of the printers in the printcap file as the lp0 printer.
Programs that change the data before it's sent to the physical printer are called filters. It's
possible for a filter to send no data at all to a physical printer; that is, the filter filters out
everything.
Understanding the Fields in /etc/printcap
The printcap file has too many fields to describe fully in this chapter, so only the most
important ones are described. All fields in /etc/printcap (except for the names of the printer)
are enclosed between colons and denoted by a two-letter code. The two-letter code is
followed by a value that depends on the type of field. The three types of fields are string,
Boolean, and numeric. Table 18.3 describes the most common and most important fields; the
following sections go into more detail.
Table 18.3 The /etc/printcap Fields
Field Type Description
lp String Specifies the device to print toÐfor example, /dev/lp0
sd String Specifies the name of the spool directory for this printer
lf String Specifies the file that errors on this printer are to be logged to
Understanding the /etc/printcap File
399
Table 18.3 The /etc/printcap Fields
Field Type Description
if String Specifies the input filter name
rm String Specifies the name of a remote printing host
rp String Specifies the name of a remote printer
sh Boolean Suppresses headers (banner pages)
sf Boolean Suppresses end-of-job form feeds
mx Numeric Specifies the maximum allowable print job size (in blocks)
The lp Field
If you specify /dev/null as the print device, all other processing is performed correctly, but
the final data goes to the bit bucketÐthat is, to nowhere. Printing to nowhere is rarely useful
except for test printer configurations or with weird printers. When you're setting up a remote
printer (that is, you've specified rm and rp fields), specify :lp=:.
Don't leave the lp field empty unless you're using a remote printer. The printer daemon
Part
complains if you don't specify a print device.
III
The lf Field
Ch
Whatever file you specify in the lf field should already exist, or logging doesn't occur.
18
The if Field
Input filters are programs that take print data on their standard input and generate output on
their standard output. A typical use of an input filter is to detect plain ASCII text and convert
it into PostScript; that is, raw text is its input and PostScript is its output.
When you specify an input filter in the if field, the printer daemon doesn't send the spooled
print data to the specified device. Instead, it runs the input filter with the spooled data as
standard input and the print device as standard output.
The rm and rp Fields
Sending your print data to a printer attached to another machine is as simple as specifying the
remote machine rm and the remote printer rp and making sure that the print device field lp is
empty.
Data is still spooled locally before it's transferred to the remote machine. Any input
filters you specify are also run.
400 Chapter 18 Printing
The sh and sf Fields
Unless you have many different people using your printer, you're most likely not interested in
banner pages, so you should specify sh.
Suppressing form feeds, by specifying sf, is most useful if your printer is typically used for
output from word processing packages. Most word processing packages create complete
pages of data, so if the printer daemon adds a form feed to the end of each job, you get a blank
page after each job. If the printer is usually used for program or directory listings, however,
having that form feed ensures that the final page is completely ejected, so each listing starts at
the top of a new page.
The mx Field
The mx field allows you to limit the size of the print data to be spooled. The number you
specify is in BUFSIZE blocks (1KB under Linux). If you specify zero, the limit is removed,
allowing print jobs to be limited only by available disk space.
The limit is on the size of the spooled data, not the amount of data sent to the
physical printer.
If a user tries to exceed this limit, the file is truncated. The user sees a message saying this:
(lpr: file-name: copy file is too large)
For non-PostScript printers, this limit is useful if you have users or programs that may
deliberately or accidentally create excessively large output. For PostScript printers, the limit
isn't useful at all because a very small amount of spooled PostScript data can generate a large
number of output pages.
Setting the PRINTER Environment Variable
You might want to add a line to your login scriptÐor even to the default user login
PRINTER environment variable. Under the bash shell, a suitable line is
scriptÐthat sets up a
export PRINTER=myprinter. This line prevents people from having to specify -Pmyprinter
every time they submit a print job.
To add more printers, you can just repeat this process with different printer names.
Remember that you can have multiple printcap entries, all using the same physical device.
This way, you can treat the same device differently, depending on what you it when you
submit a print job to it.
401
Putting It All Together
Creating a Test printcap Entry
The following shell script is a very simple input filter; it simply concatenates its input onto
the end of a file in /tmp after an appropriate banner. You can specify this filter in the printcap
entry and specify /dev/null as the print device. The print device is never actually used, but
you have to set it to something; otherwise, the printer daemon complains:
#!/bin/sh
# This file should be placed in the printer's spool directory and
# named input_filter. It should be owned by root, group daemon, and
# be world executable (-rwxr-xr-x).
echo ------------------------------------------------ >> /tmp/
date >> /tmp/
echo ------------------------------------------------ >> /tmp/
cat >> /tmp/
In the following printcap entry, notice the reasonably readable format and the use of the
\
continuation character ( ) on all but the last line:
myprinter|myprinter: \
:lp=/dev/null: \
:sd=/usr/spool/lpd/myprinter: \
:lf=/usr/spool/lpd/myprinter/errs: \
:if=/usr/spool/lpd/myprinter/input_filter: \ Part
:mx#0: \
:sh: \
III
:sf: Ch
18
Putting It All Together
To put all the preceding bits together, the following steps guide you through setting up a
single printer on /dev/lp0. You can then extend this concept to other printers (you have to be
root to perform all these steps, by the way).
1. Check the permissions and locations of lpr, lprm, lpc, lpq, and lpd. Earlier in this
chapter, Table 18.1 listed the correct settings and directories.
2. Create the spool directory for your printer (named myprinter for now). Make sure
that both the directory and printer are owned by root, belong to the daemon group,
and have write permissions for user and group and read-only permission for others
(-rwxrwxr-x). Use the following commands:
mkdir /usr/spool/lpd
mkdir /usr/spool/lpd/myprinter
chown root.daemon /usr/spool/lpd /usr/spool/lpd/myprinter
chmod ug=rwx,o=rx /usr/spool/lpd /usr/spool/lpd/myprinter
3. In the /usr/spool/lpd/myprinter directory, create the necessary files and give them
the correct permissions and owner. Use the following commands:
cd /usr/spool/lpd/myprinter
touch .seq errs status lock
chown root.daemon .seq errs status lock
chmod ug=rw,o=r .seq errs status lock
402 Chapter 18 Printing
4. Create the shell script input_filter in the /usr/spool/lpd/myprinter directory. Use
the input filter given earlier in the section ``Creating a Testprintcap Entry'' for your
filter. Make sure that the file is owned by root, belongs to the daemon group, and is
executable by anyone. Use the following commands:
cd /usr/spool/lpd/myprinter
chmod ug=rwx,o=rx input_filter
5. Create the /etc/printcap file if it doesn't already exist. Remove all entries in it and
add the test printcap entry given in the ``Creating a Test printcap Entry'' section.
Make sure that the file is owned by root and is read-only to everyone else. You can use
the chmod command to set the proper file permissions: -rw-r-r- (or 644 in octal).
6. Edit the rc.local file (you can use any ASCII editor, such as vi or Emacs). Add the
line /etc/lpd to the end to run the printer daemon each time the system boots. You
?
don't need to boot now, however; just run it by hand using the lpd command.
See ``Starting vi by Using an Existing File,'' p. 209
7. Do a test print by entering the following:
ls -l | lpr -Pmyprinter
8. Use the ls command to look in /tmp for a file named testlp.out. It should contain
your directory listing, which you can check by using the more, less, or cat commands.
See Chapter 19, ``Understanding the File and Directory system,'' for more
?
information on these commands.
See ``Viewing the Contents of a File,'' p. 428
9. Use an ASCII editor such as vi to make the following edits to /etc/printcap:
. In the first printer entry, change both occurrences of myprinter to testlp
only in the first line.
. In the second entry, change /dev/null to your real print deviceÐfor example,
/dev/lp0.
. In the second entry, remove the if line completely.
10. Copy the myprinter entry so that you have two identical entries in the file.
11. Either reboot the system or kill the printer daemon and restart it. You do so because
the printer daemon looks only at the /etc/printcap file when it first starts up.
12. Run a test print again using the command ls -l | lpr -Pmyprinter. This one should
come out on your physical printer.
Configuring Red Hat Printers
If you've installed XFree86 under Red Hat, you can use the printer configuration tool shown
in Figure 18.1 to add and delete printers as well as maintain the /etc/printcap and spooler
files and directories. You can find this tool in the Control Panel; Table 18.4 describes each of
the printtool's menu items.
403
Configuring Red Hat Printers
Figure 18.1
Managing prin-
ters is easy
with Red Hat's
graphical utili-
ties.
Table 18.4 The printtool Menus
Menu Name Submenu Description
printtool Reload Rescans the directory for
printcap files
About Displays information about
the printtool
Quit Exits the printtool
lpd Restart Restarts the lpd daemon after
Part
making changes
Tests Print ASCII Test Page Prints a test page in plain text to III
the selected printer
Ch
Print PostScript Prints a PostScript test page to
Test Page the selected printer
18
Print ASCII Directly Prints a test page directly to the
to Port device and not via the lpd system
Help General Provides general help on the
printtool
Troubleshooting Provides help on various pro-
blems with printing
To add a new printer, click the Add button. You must first specify whether it's a local,
remote, or SMB printer (see Figure 18.2). A local printer is connected to your parallel or serial
port; a remote printer is connected to your network. A LAN Manager Printer is a printer
attached to a different system via Session Message Block protocol (SMB, or Samba), typically a
Microsoft Windows system.
Figure 18.2
To add a prin-
ter, you must
select the type
of printer.
? See ``Using Samba,'' p. 485
404 Chapter 18 Printing
To edit an existing printer configuration, select the entry and click the Edit button. Both
actions bring up the dialog box shown in Figure 18.3. You must enter a value for each field in
the dialog box. Table 18.5 describes each field.
Figure 18.3
To print prop-
erly from Linux,
you must spe-
cify certain op-
tions, such as
printer name
and physical
port location.
Table 18.5 Field Items for Each Printer
Field Name Description
Names The name of the printer and its queue. You can specify multiple names
by using the | character to separate names.
Spool Directory The directory for spooling documents for this printer, such as /usr/
spool/lpd/myprinter.
File Limit The maximum document size (in kilobytes). A 0 (zero) value indicates
no limit.
Printer Device The physical connection for your printer, such as lp0.
Input Filter The full path and filename of your custom filter. If you need to
configure a printer, click the Select button.
Suppress Headers Check this box if you don't want a header page printed with each
document.
Remote Host This field in the Remote Host dialog box specifies the name of the
remote host to which the printer is connected.
Remote Queue This field in the Remote Host dialog box specifies the printer queue on
the remote machine. Enter the full path.
To configure a print filter, click the Select button, which displays the Configure Filter dialog
box shown in Figure 18.4. Table 18.6 describes the various fields in the Configure Filter
dialog box.
Table 18.6 Field Items for Each Filter
Field Name Description
Printer Type Indicates the type of printer for this filter
Driver Description Provides a description of the selected printer
Resolution Selects the desired resolution for this printer
405
Troubleshooting
Table 18.6 Field Items for Each Filter
Field Name Description
Paper Size Selects the desired paper size for this printer
Color Depth Selects the desired color operation for this printer
Printing Options Sends EOF forces the printer to eject the page
Fix stair-stepping text fixes the stair-step effect
Fast text printing enables a non-PostScript printer to attain faster
print speeds
Margins Specifies the desired margins
Extra GS Options Specifies extra ghostscript options for the selected printer
Figure 18.4
Configuring an
input filter for
your printing
system is made
simple with the
Configure Filter
Part
dialog box.
III
Ch
18
After you add or change a printer entry, you might find that you need to restart the lpd
daemon. To do so, simply select the lpd menu item on the RHS Linux Print System Manager
and click the Restart lpd item.
Troubleshooting
I get a message saying lpd: connect: No such file or directory.
The printer daemon /etc/lpd isn't running. You may have forgotten to add it to your
/etc/rc.local file. Or maybe you did add it, but you haven't booted since then. Add it and
reboot, or just run /etc/lpd. Remember that you have to be root to add the printer daemon.
I get a message saying Job queued, but cannot start daemon.
This message often appears right after the lpd: connect message; same answer as the
preceding problem.
406 Chapter 18 Printing
I get a message saying lpd: cannot create spooldir/.seq.
You haven't created the spool directory specified in the printcap entry, or you've misnamed
it. An alternative (though much less likely) answer is that you have too little disk space left.
I get a message saying lpr: Printer queue is disabled.
As root, use lpc enable printer_name to enable the printer. Note that as root, you can submit
jobs even to a disabled printer.
I submit a print job and don't get any error messages, but nothing comes out on the physical printer.
You might have this problem for many reasons:
& Make sure that the physical printer is switched on, selected, and physically connected
to the device specified in the /etc/printcap file.
& Use the lpq command to see whether the entry is in the queue. If it is, the device may
be busy, the printer may be down, or an error may have occurred on the printer. If you
do have errors, check the error log specified in the printcap entry for clues.
& You can use the lpc status command to check whether the printer is down, and you
can use lpc up printer_name or lpc restart printer_name to bring it back up if it is
(you need to be root to use these commands).
If, after checking, your print jobs still don't come out, make sure that any input filter you've
specified is present in the correct directory and has the correct permissions. If you're running
syslogd, you can look in your logs for messages from lpd. If you see log entries saying cannot
execv name of input filter, this is almost certainly the problem.
Another possibility is that you have a PostScript printer, and you're not sending PostScript to
it. Most PostScript printers ignore non-PostScript data. You might need to install an
appropriate text-to-PostScript input filter.
Last (and you'll feel silly if this is the cause), make sure that your input filter actually
generates output and that the output device isn't /dev/null.
My printer seems to have locked up. None of the techniques described here seem to solve the problem.
When all else fails in the case of a nonprinting printer, the next-to-last resort is to kill the lpd
daemon and restart it. If that approach doesn't work, the last resort is to reboot your Linux
system with the shutdown -r now command. Make sure that no one else is logged in and that
you've saved any files before using the now option; otherwise, specify a time and also give a
message to your other users before shutting down the system. You also can test the printer on
a DOS or Windows machine to make sure that the physical device itself is working.
CHAPTER
19
Understanding the File and
Directory System
In this chapter by Jack Tackett, Jr.
Understanding Filenames and Pathnames 408
Linux Standard Directories 418
Managing Files and Directories 420
Listing Files 421
Organizing Files 424
Copying Files 425
Moving and Renaming Files 426
Removing Files or Directories 426
Viewing the Contents of a File 428
Searching for Files 432
Changing File Time and Date Stamps 434
Compressing Files 435
Case Study: Undeleting Files 436
408 Chapter 19 Understanding the File and Directory System
Understanding Filenames and Pathnames
Every physical and logical entity in Linux is represented as a file in the Linux file system. The
physical entities include disks, printers, and terminals; logical entities include directories and,
of course, ordinary filesÐthe kind that store documents and programs.
The term Linux file system has two different and often conflicting meanings: the file system of
disks and mechanisms of the disks, and the logical file systemthat the user sees and
manipulates. This chapter is about the logical Linux file system that you see and manipulate.
If you're familiar with PC operating systems such as MS-DOS and OS/2, you'll find many of
the following topics familiar because the file structures of MS-DOS from version 2.0 onward
were modeled on those of UNIX, which is the file structure used by Linux.
In Linux, just as in other operating systems such as MS-DOS, you must distinguish between a
filename and a pathname. A filename consists of a simple series of contiguous letters, numbers,
and certain punctuation marks. Filenames can't contain spaces or any characters that
represent a field separator. For example, the filename johns.letter is valid, but johns letter isn't.
A filename shouldn't contain any characters that have special meaning to the shell. The
following are the ``forbidden'' special characters:
! @ # $ % ^ & * ( ) [ ] { } ' '' \ / | ; < > Á
Also, a filename can't contain the slash character (/) because this character is used to indicate
pathnames. (Pathnames are discussed later in this section.)
Tip #108 from
Jack You can use any of the ``forbidden'' characters if you place double quotation marks
around the filename like this:
"! johns.letter''
However, you'll have a hard time accessing such a file with most programs, and the
file isn't very portable to other UNIX systems.
Most early versions of UNIX, on which Linux is based, limited filenames to 14 characters;
however, Linux allows 256 characters in a filename. Some recent UNIX versions, such as the
Berkeley version (BSD), allow 64-character filenames, but only the first 14 are significant.
Because one of the goals of Linux is portability, in the interest of writing portable programs
and shell scripts, you might want to limit yourself to 14-character filenames.
A pathname can contain any number of characters. In Linux, files don't exist in a vacuum;
they exist in a directory. The highest directory in Linux is called the root and is symbolized by
/
the slash character ( ). If a file named fred exists in the root directory, its absolute pathname
is /fred. When you add a user to the system by using the adduser command, he or she is
assigned a home directory. By convention, this home directory is usually found under root in
a directory named, appropriately enough, home. Therefore, if a user named Fred is assigned a
409
Understanding Filenames and Pathnames
directory named /home/fred, all files that Fred creates are attached to the /home/fred
directory. An absolute pathname for one of Fred's files might be /home/fred/freds.file. An
absolute pathname specifies exactly where a file is stored in the file system.
Another kind of pathname is a relative pathname, which unambiguously points to a file's
location as relative to the current directory. If Fred is in his home directory, for example, the
filename freds.file is also a relative pathname, relative to his current directory. To find out
which directory is your current directory, you can use the command pwd (print working
directory). You can also check the contents of the $PWD environment variable by using the
command echo $PWD to see which directory is the current working directory.
You can define a file anywhere in the Linux file system with relative pathnames by using two
.
pseudonyms found in all directories. The single dot ( ) refers to the current directory, and the
double dot ( ..) refers to the parent directory. MS-DOS and OS/2 use this same convention.
If Fred is in /home/fred, /fred by using ../../fred. In this relative
he can point to
pathname, the second double dot points to /home (the parent directory of /home/fred), and
the first double dot points to the parent directory of /homeÐnamely, the root.
The pseudonym for the current directory, the single dot, comes in handy if you want to move
files. If Fred wants to move /fred to his current directory, he can do so with absolute
pathnames by using this command:
mv /fred fred
Alternatively, Fred can use the pseudonym for the current directory by using this command:
mv /fred .
Part
Most Linux commands operate on pathnames. In most cases, the pathname you use is the
name of a file in the current directory. The default pathname points to your current directory. III
If Fred is in his home directory ( /home/fred), all three of the following are equivalent
Ch
commands:
19
command freds.letter
command /home/fred/freds.letter
command ./freds.letter
Although filenames and pathnames are different, directories are files, too. When
you're naming directories, remember that you must follow the same naming
guidelines as for ordinary files.
Also note that unlike many PC-based operating systems, Linux doesn't have the
concept of disk drive letters, only directory paths. Linux deals with disk drive letters
only when working with MS-DOS file systems on floppies with the m- commands
(such as mcopy).
410 Chapter 19 Understanding the File and Directory System
? See ``Understanding File Systems,'' p. 440
File Types
Linux lumps everything into four basic types of files: ordinary files, directories, links, and
special files. You'll work with several kinds of ordinary files, links, and special files and a large
number of standard directories. The basic file types are described in the following sections.
You can use the command file to determine the type of a file. file can recognize a file type
as executable, text, data, and so on. Many UNIX commands are only shell scripts or are
interpreted programs similar to MS-DOS batch files, and file can report whether a UNIX
command is a binary executable program or simply a shell script. It's also useful for
determining whether the file is text-based and, therefore, whether it can be viewed or edited.
The syntax for the file command is as follows:
file [-vczL] [-f namefile] [-m magicfile] filelist
Table 19.1 explains the arguments for the file command.
Table 19.1 file Command Arguments
Argument Description
-c Prints the parsed form of the magic file ( /usr/lib/magic), which is a number
in the first part of a binary file that identifies the file type. This argument is
usually used with -m to debug a new magic file before installing it.
-z Looks inside a compressed file and tries to figure out the file type.
-L Causes symbolic links to be followed.
-f namefile Tells file that the list of files to identify is found in namefile, which is a text
file. This argument is useful when many files must be identified.
-m magicfile Specifies an alternative file of magic numbers to use for determining file
types. The default file is /usr/lib/magic.
filelist Lists space-delimited files whose type you want to know.
Ordinary Files
Ordinary files are what you spend most of your time manipulating. Ordinary files can contain
text, C language source code, shell scripts (programs interpreted by one of the Linux shells),
binary executable programs, and data of various types. As far as Linux is concerned, a file is a
file. The only difference that Linux knows is files marked as executable. Executable files can
be executed directlyÐprovided, of course, that the file contains something to execute and that
it's in your search path. Basically, the search path is a list of pathnames you've specified that
?
Linux searches to find an executable file.
See ``Understanding Shells,'' p. 319
Executable files are binary filesÐthat is, files that execute machine code and shell scripts. The
Linux file command discussed in the preceding section looks at the data in a file and makes a
411
Understanding Filenames and Pathnames
reasonable guess as to what's inside. If you type file *, for example, you might see something
similar to this:
INSTALL: symbolic link to /var/adm
ghostvw.txt: ascii text
linux: symbolic link to /usr/src/linux
mbox: mail text
mterm.txt: English text
seyon.txt: English text
xcalc.txt: English text
xclock.txt: English text
xeyes.txt: English text
xgrap.txt: English text
xlock.txt: English text
xspread.txt: English text
xtris.txt: empty
All the files named in the first column are ordinary files that contain different kinds of data.
All the files are located within the directory where the file command was executed.
Directory Files
Directories are files that contain the names of files and subdirectories, as well as pointers to
those files and subdirectories. Directory files are the only places that Linux stores names of
files. When you list the contents of a directory by using the ls command, all you're doing is
listing the contents of the directory file. You never touch the files themselves.
When you rename a file by using the mv command and that file is in the current directory, all
you're doing is changing the entry in the directory file. If you move a file from one directory
to another, all you're doing is moving the description of the file from one directory file to
Part
anotherÐprovided, of course, that the new directory is on the same physical disk or partition.
If not, Linux physically copies each byte of the program to the other disk.
III
Ch
Directories and Physical Disks
19
Every file in a Linux system is assigned a unique number called an inode. The inode is stored
in a table called the inode table, which is allocated when the disk is formatted. Every physical
disk or partition has its own inode table. An inode contains all the information about a file,
including the address of the data on the disk and the file type. File types include ordinary files,
directories, and special files.
Tip #109 from
Jack You can use the df -icommand to see the number of inodes available on various file
systems. To see the information in a more readable manner, you can add the -h flag.
For example, the command shown here produces this table:
$ df -ih
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda7 65k 6.9k 58k 11% /
/dev/sda1 5.9k 24 5.9k 0% /boot
412 Chapter 19 Understanding the File and Directory System
/dev/sda8 1.1M 412 1.1M 0% /home
/dev/sda9 257k 45k 212k 17% /usr
/dev/sda5 65k 240 64k 0% /var
The Linux file system assigns inode number 1 to the root directory. This assignment gives
Linux the address on disk of the root directory file. The root directory file contains a list of
file and directory names and their respective inode numbers. Linux can find any file in the
system by looking up a chain of directories, beginning with the root directory. The contents
of the root directory file might look like this:
1 .
1 ..
45 etc
230 dev
420 home
123 .profile
Notice that the files . (dot) and .. (double dot) are shown in the directory. Because this is the
root directory, . and its parent directory, .., are identical. The contents of the /home
directory file would be different and might look something like this:
420 .
1 ..
643 fred
.
Notice that the inode of the current directory ( ) matches the inode for /home found in the
root directory file, and the inode for the parent directory ( ..) is the same as that of the root
directory.
Linux navigates its file system by chaining up and down the directory file system. If you want
to move a file to a directory on another physical disk, Linux detects this information by
reading the inode table. In such a case, the file is physically moved to the new disk and
assigned a new inode on that disk before being deleted from its original location.
As with the mv command, when you delete a file by using the rm command, you never touch
the file itself. Instead, Linux marks that inode as free and returns it to the pool of available
?
inodes. The file's entry in the directory is erased.
See ``Moving and Renaming Files,'' p. 425
Links
Ordinary links aren't really files at all; they're actually directory entries that point to the same
inode. The inode table keeps track of how many links go to a file, and only when the last
directory reference is deleted is the inode finally released back to the free pool. Obviously,
ordinary links can't cross device boundaries because all the directory references point to the
same inode.
413
Understanding Filenames and Pathnames
To create a link, you use the ln command, which has the following form:
ln [options] sourcedestination
For example to create a link between a file named mainfile.txt and a file named
tempfile.txt, you enter the following command:
ln mainfile.txt tempfile.txt
Linux, as well as most modern versions of UNIX, has another kind of link called a symbolic
link. For such a link, the directory entry contains the inode of a file that is itself a reference to
another file somewhere else in the logical Linux file system. A symbolic link can point to
another file or directory on the same disk, to another disk, or to a file or directory on another
computer.
One major difference between an ordinary link and a symbolic link is that with ordinary links,
every link has equal standing (that is, the system treats every link as though it were the
original file), and the actual data isn't deleted until the last link to that file is deleted. With
symbolic links, when the original file is deleted, all symbolic links to that file are also deleted.
Symbolically linked files don't have the same standing as the original file.
To create a symbolic link, you use the -s option to the ln command. For example, to create a
symbolic link from a file called named in the/etc/rc.d/initd directory to the file S55named,
you use the following command:
ln -s /etc/rc.d/initd/named /etc/rc.d/rc3.d/S55named
Other than these subtle differences between links and files, links are treated and accessed
exactly as files are.
Part
You can tell a file is a link by using the ls -l command. If it is a link, the response shows the
local filename and then an indication of the linked file like this:
III
lrwxrwxrwx 1 root root 4 Oct 17 15:27 Info -]] info/
Ch
19
The file permission flags begin with l to indicate that the file is a linked file.
Special Files
All physical devices associated with a Linux system, including disks, terminals, and printers,
are represented in the file system. Most, if not all, devices are located in the /dev directory.
For example, if you're working on the system console, your associated device is named
/dev/console. If you're working on a standard terminal, your device name might be
/dev/tty01. Terminals, or serial lines, are calledtty devices (which stands for teletype, the
original UNIX terminal). To determine what the name of your tty device is, you can type the
command tty. The system responds with the name of the device to which you're connected.
Printers and terminals are called character-special devices. They can accept and produce a
stream of characters. Disks, on the other hand, store data in blocks addressed by cylinder and
sector. You can't access just one character on a disk; you must read and write entire blocks.
The same is usually true of magnetic tapes. This kind of device is called a block-special device.
414 Chapter 19 Understanding the File and Directory System
To make life even more complex, disks and other block-special devices must be able to act like
character-oriented devices, so every block-special device has a matching character-special device.
Linux makes the translation by reading data being sent to a character device and translating it for
the block device. This translation happens without your doing anything.
You might run into at least one other type of special device: a FIFO (first-in-first-out buffer),
also known as a named pipe. FIFOs look like ordinary files: If you write to them, they grow.
But if you read FIFOs, they shrink in size. FIFOs are used mainly in system processes to allow
many programs to send information to a single controlling process. For example, when you
print a file by using the lp command, lp sets up the printing process and signals the lpsched
daemon by sending a message to a FIFO. A daemon, sometimes called a demon, is a system
process that acts without a user requesting an action.
One device-special fileÐthe bit bucket, or /dev/nullÐis very useful. Anything you send to
/dev/null is ignored, which is useful when you don't want to see the output of a command.
For example, if you don't want any diagnostic reports printed on the standard error device,
you can pour them into the bit bucket by using the following command:ls -la",4> /dev/null
File Permissions
File permissions mean more in Linux than just what permissions you have on a file or
directory. Although permissions determine who can read, write, or execute a file, they also
determine the file type and how the file is executed.
You can display the permissions of a file by using the long form of the listing command, ls -l.
The -l flag tells the lscommand to use the long listing. If you type ls -l, you might see a
directory listing that looks like this:
Drwx--- 2 sglines doc 512 Jan 1 13:44 Mail
Drwx--- 5 sglines doc 1024 Jan 17 08:22 News
-rw--- 1 sglines doc 1268 Dec 7 15:01 biblio
drwx--- 2 sglines doc 512 Dec 15 21:28 bin
-rw--- 1 sglines doc 44787 Oct 20 06:59 books
-rw--- 1 sglines doc 23801 Dec 14 22:50 bots.msg
-rw-r-- 1 sglines doc 105990 Dec 27 21:24 duckie.gif
This listing shows virtually everything that can be known about a file from the directory entry
and the inode of the file. The first column shows the file permissions, the second column
shows the number of links to a file (or extra blocks in a directory), and the third column shows
who owns the file. (In Linux, ownership has three possibilities: the owner, the owner's group,
and everyone else. Ownership is detailed later in this chapter.) The fourth column shows the
group to which the file belongs. The fifth column shows the number of bytes in the file, the
sixth column shows the date and time of creation, and the seventh column shows the name of
the file itself.
The permissions field (the first column) can be broken into four distinct subfields:
- rwx rwx rwx
415
Understanding Filenames and Pathnames
-
The first subfield defines the file type. A normal file has a hyphen ( ) as a placeholder;
directories are marked with a d. Table 19.2 shows the permissible values for the file-type
subfield.
Table 19.2 Valid Entries for the File-Type Subfield
Character Meaning
- Ordinary file
b Block-special file
c Character-special file
d Directory
l Symbolic link
The next three subfields show the read, write, and execute permissions of the file. For
example, an rwx in the first of these subfields means that the file has read, write, and execute
permission for the owner. The next three characters show the same information for the group
ownership of the file. Finally, the third set of characters shows the permissions allowed for
everyone else.
These permission subfields can show more information; in fact, several attributes are packed
into these three fields. Unfortunately, what these attributes mean is determined by the version
of Linux you use and whether the file is executable.
Part
Normally, a running program is owned by whoever ran it. If the user ID bit is on, the
running program is owned by the owner of the file. This means that the running III
program has all the permissions of the owner of the file. If you're an ordinary user
Ch
and the running program is owned by the root user, that running program has
automatic permission to read and write any file in the system regardless of your 19
permissions. The same is true of the Set Group ID bit.
The sticky bit can also be set in these subfields. The sticky bit tells the system to save a copy of
a running program in memory after the program is complete. If the program is used often,
the sticky bit can save the system a little time the next time it runs the program because the
program doesn't have to be reloaded into memory from disk each time someone runs it.
You can change permissions on any file for which you have write permission by using the
chmod command. This command has two different syntaxes: absolute and relative. With
absolute permissions, you define exactly what the permissions on a file will be in octal, or base
8. An octal number can have a value from 0 to 7. UNIX was originally created on a series of
DEC minicomputers that used the octal numbering system, hence the current use of octal
numbers. The octal numbers are added together to arrive at a number that defines the
permissions. Table 19.3 lists the valid octal permissions.
416 Chapter 19 Understanding the File and Directory System
Table 19.3 Absolute Octal Permissions Used with the chmod Command
Octal Value Permissions Granted
0001 Execute permission for the owner
0002 Write permission for the owner
0004 Read permission for the owner
0010 Execute permission for the group
0020 Write permission for the group
0040 Read permission for the group
0100 Execute permission for all others
0200 Write permission for all others
0400 Read permission for all others
1000 Sticky bit on
2000 Group ID bit on if the file is executable; otherwise, mandatory file locking is
on
4000 User ID bit on if the file is executable
Group and user IDs refer to who has permission to use, read, or execute a file. These initial
file permissions are granted by the system administrator when a user's account is first created.
Only users of an indicated group can access files in a group, and only if a user has given group
members permission to those files.
To give a file read and write permissions for everyone, you must add the required permissions
together, as in the following example:
0002 Write permission for the owner
0004 Read permission for the owner
0020 Write permission for the group
0040 Read permission for the group
0200 Write permission for all others
0400 Read permission for all others
______________________________________________________________________
0666 Read and write permission for everyone
To give a file these permissions, you use the following command:
chmod 666 file
417
Understanding Filenames and Pathnames
Tip #110 from
Jack You can change the mode for a directory and all files in the directory by using the -R
flag. For example, to change the read/write mode for all files under the /home/
website/gifs, you use the following commands:
cd /home/website
chmod -R 666 gifs
Relative permissions use a slightly different format. With relative permissions, you must state
the following:
& Whom you're giving permissions to
& What operation you intend (add, subtract, or set permissions)
& What the permissions are
For example, if you type chmod a=rwx file, you give read, write, and execute permission to all
users. Table 19.4 summarizes the commands for relative permissions.
Table 19.4 Relative Permissions Used with the chmod Command
Value Description
Whom
a All users (the users, their group, and all others)
g Owner's group
o All others not in the file's group
Part
u Just the user
III
Operator
Ch
+ Adds the mode
- Removes the mode
19
= Sets the mode absolutely
Permission
x Sets execute
r Sets read
w Sets write
s Sets user ID bit
t Sets sticky bit
If a file has been marked as having the user ID bit on, the permissions displayed by the ls -l
command look like this:
-rws--- 1 sglines 3136 Jan 17 15:42 x
418 Chapter 19 Understanding the File and Directory System
If the group ID bit is added, the permissions look like this:
-rws-S- 1 sglines 3136 Jan 17 15:42 x
If you then turn on the sticky bit for the file, the permissions look like this:
-rws-S-rws-S-T 1 sglines 3136 Jan 17 15:42 x
Note the use of uppercase S and T to indicate the status of the user ID bit and the sticky bit,
respectively.
Linux Standard Directories
You're already familiar with the concept of directories. When you log in, the system places
you in your home directory. The PATH environment variable is set to point to other directories
that contain executable programs. These other directories are part of the standard Linux
directory structure.
These directories include the classic set of directories for UNIX and what can be called the
``emerging standard set of directories,'' which Linux basically follows. They are described in
the following sections.
Classic UNIX Directories
Before UNIX System V Release 4 (for example, UNIX System V Release 3.2 and earlier),
most versions of UNIX settled on a regular system of organizing the UNIX directories that
looked like this:
/
/etc
/lib
/tmp
/bin
/usr
/spool
/bin
/include
/tmp
/adm
/lib
The /etc directory contains most of the system-specific data required to boot, or bring the
system to life. It contains such files as passwd and inittab, which are necessary for the proper
operation of the system.
The /lib directory contains a library of functions needed by the C compiler. Even if you
don't have a C compiler on your system, this directory is important because it contains all the
shared libraries that application programs can call. A shared library is loaded into memory
only when the command calling it is run. This arrangement keeps executable programs small.
Otherwise, every running program contains duplicate code, requiring a lot more disk space to
store and a lot more memory to run.
419
Linux Standard Directories
The /tmp directory is used for temporary storage. Programs that use /tmp generally clean up
after themselves and delete any temporary files. If you use /tmp, you should be sure to delete
any files before logging out. Because the system automatically deletes the contents of this
directory periodically, don't keep anything you might need later in it.
The /bin directory keeps all the executable programs needed to boot the system and is
usually home for the most commonly used Linux commands. Note, however, that an
executable program doesn't have to be binary (which the name bin implies). Several smaller
programs in /bin are, in fact, shell scripts.
The /usr directory contains everything else. Your PATH variable contains the string
/bin:/usr/bin because the /usr/bin directory contains all the Linux commands that aren't
in the /bin directory. This arrangement has a historical precedence. In the early days of
Linux, hard disks weren't very big. Linux needs at least the /etc/tmp/ and /bin directories
to bootstrap (that is, start executing) itself. Because the disks of the early Linux era held only
those three directories, everything else was on a disk that could be mounted after Linux was
up and running. When Linux was still a relatively small operating system, placing additional
subdirectories in the /usr directory wasn't much of a burden. It allowed a moderately sized
Linux system to exist with just two disks: a root disk and a /usr disk.
The /usr/adm directory contains all the accounting and diagnostic information needed by the
system administrator. If both system accounting and diagnostic programs are turned off, this
directory is effectively empty.
The /include directory contains all the source code used by #include statements in C
programs. You'll have at least read permission for this directory because it contains all the
code fragments and structures that define your system. You shouldn't modify any of the files Part
in this directory because they were crafted (carefully, you can assume) by your system vendor.
III
The /usr/spool directory contains all the transient data used by the lp print system, the cron Ch
daemon, and the UUCP communications system. Files ``spooled'' to the printer are kept in
19
the /spool directory until they're printed. Any programs waiting to be run by cron, including
all the crontab files and pending at and batch jobs, are also stored here.
The /usr/lib directory contains everything else that's part of the standard Linux system. In
general, the /usr/lib directory represents the organized chaos hidden beneath the relatively
well-disciplined Linux system. This directory contains programs called by other programs
found in /bin and /usr/bin as well as configuration files for terminals and printers, the mail
system, cron, and the UUCP communications system.
The /usr directory contains all the subdirectories assigned to users. The general convention
is this: If your login ID is mary, your home directory is /usr/mary.
This directory arrangement made a lot of sense when disks were small and expensive, but with
the advent of very large disks at (relatively) inexpensive prices, Linux can be organized in
better ways, as evidenced by the new directory structure discussed in the next section.
420 Chapter 19 Understanding the File and Directory System
Linux Directories
One problem with the classical structure of UNIX is that backing up your data files is difficult
with a fragmented /usr directory. Three different levels of backup generally are required in a
system: the basic system itself, any changes to the tables that define the basic system for a
specific site, and user data.
The basic system should be backed up only once, with changes to the controlling tables
backed up when changes are made. User data changes all the time and should be backed up
frequently. The typical Linux directory structure is shown here, but your structure might be a
little different depending on what packages you have installed:
/
/etc
/passwd (user database)
/rc.d (system initialization scripts)
/sbin
/bin
/tmp
/var
/lib
/home
/ < your user name here > (user accounts)
/install
/usr
/bin
/proc
The /bin, /etc, and /tmp directories have the same function as they do in the classic
structure. System definition tables are moved into the /var directory so that whenever the
operation of the system changes, you can back up only that directory.
What's new is that all system programs are moved into the /sbin directory. All the standard
Linux programs are in /usr/bin, /bin. For compatibility, all the classic
which is linked to
directories are maintained with symbolic links. The /usr directory, which no longer contains
user data, has been reorganized to make sense from the chaos that once was the /usr/
libdirectory.
Managing Files and Directories
The vast majority of Linux commands manipulate files and directories. Indeed, Linux shell
scripts are particularly adept at manipulating files and directories. File manipulations that are
difficult in a conventional language (even in C) are made easy from within a shell, largely
because of the rich selection of file-manipulation commands available in Linux.
File-manipulation commands can be grouped roughly into two categories:
& Commands that manipulate files as objects
& Commands that manipulate the contents of files
421
Listing Files
Listing Files
The basic command to list files is ls. The way ls displays files depends on how you use the
command. If you use the ls command in a pipe, every file is displayed on a line by itself. This
is also the default for some versions of UNIX, such as SCO UNIX. Other versions of UNIX
list files in several columns. For most uses, the columnar format is more convenient; systems
that list files one per row often have an alternative command, usually lc, for lists in column
format.
The ls command's behavior is modified with the use of flags that take the form -abcd. In
general, versions of the ls command fall into two categories: versions of ls derived from
Linux System V and those derived from Berkeley. Because the Berkeley Linux systems are
slowly giving way to Linux System V, this chapter concentrates on the flags used by System
V. If you're in doubt about which version of ls you have, consult the manuals for your system
or try the command man ls.
Most man pages for commands in this chapter are no longer being maintained and
may be inaccurate or incomplete under Red Hat Linux as the system is moved to
more graphical-based systems such as HTML and TexInfo. However, for the time
being, this information is accurate for this release of Red Hat Linux 6.0.
Flags used with the ls command can be concatenated or listed separately. This means that
the following commands are effectively identical:
Part
ls -l -F
ls -lF III
Table 19.5 lists in alphabetical order several of the flags used with ls and their uses.
Ch
19
Table 19.5 Flags for the ls Command
Flag Description
-a Lists all entries. In the absence of this or the -A option, entries whose names
begin with a period ( .) aren't listed. Linux has a way of ``hiding'' files; all files
that begin with a period by default aren't listed because they're generally files
used to customize applications. For example, .profile is used to customize the
Bourne and Korn shells; .mailrc is used to customize your systemwide email
configuration file. Because almost every major command you use has a startup
file, your home directory looks cluttered if the ls command lists all those
startup files by default. If you want to see them, you can use the -a flag.
-A Acts the same as -a, except that . and .. aren't listed. Recall from the section
``Understanding Filenames and Pathnames'' earlier in this chapter that . is a
pseudonym for the current directory and .. is a pseudonym for the parent
directory. Because these filenames begin with a period, the -a flag lists them. If
you don't want to see these pseudonyms, use the -A flag instead.
422 Chapter 19 Understanding the File and Directory System
Table 19.5 Flags for the ls Command
Flag Description
-b Forces printing of nongraphic characters to be in octal \ddd notation. -b is
more useful than the -q flag because it allows you to figure out what the
characters are.
-c Uses the time of the last edit (or last mode change) for sorting or printing.
Linux maintains three time and date stamps on every file: the file creation date,
the date of last access, and the date of last modification. Normally, files are
listed in ASCII order (alphabetical order, except that capitals are sorted before
lowercase letters).
-C Forces multicolumn output with entries sorted down the columns. This is the
default format of ls when output is to a terminal.
-d Lists only the name of the argument if it is a directory (not its contents); this
filename argument is often used with the -l flag to get the status of a directory.
Normally, the contents of a directory are listed if a directory name is explicitly
listed or implied with the use of a wildcard. Thus, the simple command ls lists
just the directory names themselves, but ls * lists files, directories, and the
contents of any directories encountered in the current directory.
-F /
Marks directories with a trailing slash ( ), executable files with a trailing asterisk
* @
( ), symbolic links with a trailing at sign ( ), FIFOs with a trailing bar ( ), and |
sockets with a trailing equals sign ( ). =
-i Prints each file's inode number ( inodes are described in the section ``Directories
and Physical Disks'' earlier in this chapter) in the first column of the report. If
you list linked files, notice that both files have the same inode number.
-l Lists directory entries in long format, giving mode, number of links, owner, size
in bytes, and time of last modification for each file. If the file is a special file, the
size field instead contains the major and minor device numbers. If the time of
last modification is greater than six months ago, the month, date, and year are
shown; otherwise, only the date and time are shown. If the file is a symbolic
link, the pathname of the linked-to file is printed, preceded by the characters ->.
You can combine -l with other options, such as -n, to show user and group ID
numbers instead of names.
-n Lists the user and group ID numbers, instead of names, associated with each file
and directory. Usually, only the names are listed. If you're setting up
networking products, such as TCP/IP, it's useful to know ID numbers when
you're setting up permissions across several systems.
-q Displays nongraphic characters in filenames as the character ?. For ls, this is
the default action when output is to a terminal. If a file has accidentally been
created with nonprintable characters, the -q flag displays the file.
-r Reverses the sort order to show files in reverse alphabetical or oldest-file-first
order, as appropriate.
-s Gives the size of each file, including any indirect blocks used to map the file, in
kilobytes. If the environment variable POSIX_CORRECT is defined, the block size is
512 bytes.
-t Sorts by time modified (latest first) rather than by name. If you want to see the
oldest file first, use the -rt combination.
423
Listing Files
Table 19.5 Flags for the ls Command
Flag Description
-u Uses time of last access, instead of last modification, for sorting (with the -t
option) or printing (with the -l option).
-x Forces multicolumn output with entries sorted across rather than down the
page.
If you installed the Slackware distribution of Linux, you'll find ls also provides color output
for each file type. The color definitions are defined in the configuration file DIR_COLORS in the
/etc directory. The default configuration highlights executable files in green, directories in
blue, and symbolic links in cyan. To customize the colors, you must copy the DIR_COLORS file
to your home directory and change its name to .dir colors. Table 19.6 provides the color
definitions available; see the man pages and the DIR_COLORS file for more information.
ls ±color to get the color effect. To set
For the Red Hat distribution, you must type
ls mini-How-To'' at http://
this behavior as the default, see the ``Colour
metalab.unc.edu/LDP/HOWTO/mini/Colour-ls.html.
Table 19.6 DIR_COLORS Values to Create Color Highlighting
Value Description
Part
0 Restores default color
III
1 For brighter colors
Ch
4 For underlined text
19
5 For flashing text
30 For black foreground
31 For red foreground
32 For green foreground
33 For yellow (or brown) foreground
34 For blue foreground
35 For purple foreground
36 For cyan foreground
37 For white (or gray) foreground
40 For black background
41 For red background
42 For green background
43 For yellow (or brown) background
424 Chapter 19 Understanding the File and Directory System
Table 19.6 DIR_COLORS Values to Create Color Highlighting
Value Description
44 For blue background
45 For purple background
46 For cyan background
47 For white (or gray) background
To find more options than the ones shown here, you can consult the man pages for ls.
Organizing Files
Linux doesn't have any fixed rules for organizing files. Files don't have extensions (such as
.exe for executables) as they do in MS-DOS. You can (and perhaps should) make up your
own system of naming files, but the classic system of organizing files in Linux is to use
subdirectories.
More and more, however, Linux applications that have come from the DOS world are
bringing their conventions to Linux. Although they may not require it, vendors encourage
you to use certain extensions to name files that you use with their applications.
If you're going to write your own commands, a useful way to organize your directories is to
mimic Linux's use of the /bin, /lib, and /etc directories. You can create your own structure
of subdirectories with these names, perhaps under your /home directory, and follow the Linux
tradition of placing executable commands in your /bin directory, subsidiary commands in
your /lib directory, and initialization files in your /etc directory. Of course, you aren't
required to organize your directories this way, but it's one way of organizing your files.
You create directories by using the mkdir command. Its syntax is simple:
mkdir directory-name
In this syntax, you replace directory-name with the name you want to assign to the new
directory. Of course, you need write permission in the directory before you can create a
subdirectory by using mkdir, but if you're making a subdirectory within your home directory,
you should have no problem.
Suppose you've written three programs called prog1, prog2, and prog3, all of which are found
in $HOME/bin. Remember that $HOME is your home directory. If you want your private
programs to run as though they were a standard part of the Linux command set, you must add
$HOME/bin to your PATH environment variable. You do so by using the following command in
the Bourne or Korn shell:
PATH=$PATH:$HOME/bin;export PATH
In the C shell, you use this command:
setenv PATH ``$PATH $HOME/bin''
425
Copying Files
Remember that $HOME is the placeholder for the complete path that refers to your
home directory. If your home directory is /home/ams, $HOME/bin is interpreted as
/home/ams/bin.
If your programs call subsidiary programs, you might want to create subdirectories within
your $HOME/lib directory. You can create a subdirectory for each program. The private
command pgm1 can then explicitly call, for example, $HOME/lib/pgm1/pgm1a.
Similarly, if your command prog1 requires a startup table, you can name that table
$HOME/etc/pgm1.rc, and your data can be in your $HOME/data/pgm1 directory.
Copying Files
The command for copying files is cp from to. You must have read permission for the file
you're copying from and write permission for the directory (and the file if you're overwriting
an existing file) you're copying to. Other than that, no restrictions are placed on your ability
to copy files.
You need to watch for the following as you copy files:
& If you copy a file and give it the name of a file that already exists and that you have
write permission for, you overwrite the original file.
& If you give the name of a directory as the destination of the cp command, the file is
copied into that directory with its original name. For example, if you type the
Part
command cp file directory, the file is copied into directory as directory/file.
III
& You can copy a list of files into a directory by using the command cp file1 file2
file3 ... directory. If the last item in the list isn't a directory, an error message
Ch
appears. Likewise, if any element in the list other than the last item is only a directory, 19
an error message appears.
& When you use wildcards with the cp command, you can copy more than you intend to,
so be careful.
Because many Linux users also have MS-DOS files on their systems and usually make
the DOS file system accessible from Linux, most of the Linux commands recognize
when a file is being copied to or from a DOS partition. Thus, Linux can handle the
necessary file translation when copying files. This translation is required because most
DOS files embed the carriage return/linefeed characters into an ASCII file to indicate a
line break. Most Linux and UNIX systems embed only a linefeed character, called
newline, in the file to indicate a line break.
426 Chapter 19 Understanding the File and Directory System
Moving and Renaming Files
In Linux, you move and rename files by using the same command: mv. The syntax and rules
are the same for mv as they are for the copy command, cp. That is, you can move as many files
as you want to a directory; however, the directory name must be last in the list, and you must
have write permission to that directory.
One thing you can do with mv that you can't do with cp is move or rename directories. When
you move or rename a file, only the entry in the directory file is changed. Unless the new
location is on another physical disk or partition, the file and the contents of the directory are
physically moved.
If you try to use rm (for remove) or cp without options on a directory, the command fails and
displays a message telling you that the item you're dealing with is a directory. To remove or
copy directories, you must use the -r flag (for recursive) with rm and cp. The mv command,
however, moves directories quite happily.
Removing Files or Directories
The command to remove a file is rm. To delete a file you don't own, you need read and write
permission. If you own the file, you're allowed to delete it, provided that you haven't closed
off your own permission to the file. For example, if you turn off write permission to a file by
typing chmod 000 file, you must open permission again by using the chmod command (by
typing chmod 644 file) before you can delete it.
If you accidentally type rm *, you delete all the files you have permission to delete in the
current directory; you don't delete the subdirectories. To delete subdirectories, you must use
the recursive option ( -r).
Some versions of rm stop and ask whether you really want to delete files that you own but
don't have at least write permission for. Other versions of rm prompt you for any files marked
for removal with wildcards. Indeed, you can write a macro or shell script that gives you a
second chance before actually deleting a file.
If your version of rm balks at removing files you own but don't have write permission for, you
can partially protect yourself from accidentally deleting everything in your directory by
following these steps:
1. Create a file named 0. In the ASCII string sequence, the number 0 is listed before any
files that begin with letters.
2. Remove all permissions from the file named 0 by typing the command chmod 000 0.
This command removes read, write, and execute permissions for everyone, including
yourself.
3. If you type the command rm *, the file named 0 is the first file that rm attempts to
remove.
427
Removing Files or Directories
If your version of rm balks at removing the 0 file when you type rm *, you have the chance to
think about what you just did. If you didn't intend to delete everything in your directory, you
can press Del or Ctrl+c to kill the rm process. To test this procedure, try removing just the file
named 0. Don't use rm *, because if your version of rm doesn't stop at the file 0, you'll erase all
the files in your directory.
A better way to protect yourself from accidentally deleting files is to use the -i flag with rm.
The -i flag stands for interactive. If you give the command rm -i filename, you're asked
whether you really want to delete the file. You must answer yes before the file is actually
deleted. If you type the command rm -i *, you must answer yes for every file in your directory.
Having to respond for every file should give you enough time to think about what you really
want to do.
Think before you delete files. In most versions of Linux, unlike Windows, DOS, or
Macintoshes, when you delete a file, it's gone and the only sure way to recover a lost
file is from a backup. You did make a backup, didn't you? Some possible ways to
recover parts of files are discussed later in this chapter, but you have no guarantees.
?
?
See ``Performing Backups and Restoring Files,'' p. 263
See ``Case Study: Undeleting Files,'' p. 436
If you use the rm -i command frequently, you can implement it in two ways: by writing a shell
script or by creating a shell function. If you write a shell script, remember that the shell Part
searches for commands in the directories listed in your PATH variable in the order in which
III
they're listed. If your $HOME/bin directory is listed last, a shell script named rm will never be
found. You can place your $HOME/bin directory first in the PATH variable's list or create a new
Ch
command, such as del. If you create a shell script called del, you must mark it as executable 19
with the chmod command before the shell can recognize it. When you create your del
command, you need to give it only one command: rm -i $*. If you then type the command del
?
*, the shell translates it into rm -i *.
See ``Editing and Aliasing Shell Commands,'' p. 346
Another way to accomplish the same task is to use an alias, which takes precedence over
commands that must be looked up. You can think of an alias as an internal shell command
(similar to the doskey commands introduced in MS-DOS version 5.0).
To add an alias if you're using the C shell, you must edit the file named .cshrc. You can use
any text editor, such as vi (see Chapter 9, ``Using the vi Editor''), to edit this file. For the C
shell, add the following lines to the top of your .cshrc file:
rm ()
{
428 Chapter 19 Understanding the File and Directory System
/bin/rm -i $*
}
To add an alias to the Korn shell, add the following line to your $HOME/.kshrc file:
alias rm 'rm -i $*'
rm command, you're told that it's a directory and
If you try to delete a directory by using the
can't be deleted. If you want to delete empty directories, you can use the rmdir command, as
with MS-DOS.
Linux offers another way to delete directories and their contents, but it's far more dangerous.
The rm -r command recursively deletes any directories and files it encounters. If you have a
directory named ./foo that contains files and subdirectories, the command rm -r foo deletes
the ./foo directory and its contents, including all subdirectories.
If you give the command rm -i -r, each directory that the rm command encounters triggers a
confirmation prompt. You must answer yes before the directory and its contents are deleted.
If you left any files in the directory you were attempting to delete, rm balks, just as it does if
you attempt to remove the nonempty directory by using the rm command with no options.
You don't have to issue each flag individually for a Linux command. If the flag doesn't
take an argument, you can combine the flags. Thus, you can issue rm -i -r as
rm -ir.
Viewing the Contents of a File
Almost every Linux command prints to the standard output device, typically your screen. If
the command takes its input from a file after manipulating the file in some way, the command
prints the file to your screen. The trick in choosing a Linux command depends on how you
want the file displayed. You can use three standard commands: cat, more, and less.
Linux, like all UNIX systems, opens four system files at startup: standard input,
standard output, standard error, and AUX. These files are actually physical devices:
Name Alias Device
Standard Input standard in (stdin) The keyboard
Standard output standard out (stdout) The screen
Standard error standard err (stderr) The screen
AUX auxiliary An auxiliary device
429
Viewing the Contents of a File
Using cat to View a File
For displaying short ASCII files, the simplest command is cat, which stands for concatenate.
The cat command takes a list of files (or a single file) and prints the contents unaltered on
standard output, one file after another. Its primary purpose is to concatenate files (as in cat
file1 file2 > file3), but it works just as well to send the contents of a short file to your
screen.
If you try to display large files by using cat, the file scrolls past your screen as fast as the
screen can handle the character stream. One way to stop the flow of data is to alternatively
press Ctrl+s and Ctrl+q to send start and stop messages to your screen, or you can use one of
the page-at-a-time commands, more or less.
Using more to View a File
Both more and less display a screen of data at a time. Although they both perform roughly
the same job, they do it differently. more and less determine how many lines your terminal
can display from the terminal database and from your TERM environment variable.
The more command is older than less, and it's derived from the Berkeley version of UNIX. It
has proved so useful that, like the vi editor, it has become a standard. This section covers just
the basics of the command.
The simplest form of the more command is more filename. When you use this command, you
see a screen of data from the file. If you want to go on to the next screen, you press the
spacebar. If you press Enter, only the next line is displayed. If you're looking through a series
of files (by using the command more file1 file2...) and want to stop to edit one, you can do
so by using the e or v command. Pressing e within more invokes whatever editor you've Part
defined in your EDIT shell environment variable on the current file. Pressing v uses whatever III
editor has been defined in the VISUAL variable. If you haven't defined these variables in your
Ch
environment, more defaults to the ed editor for the e command and to the vi editor for the v
19
?
command.
See ``Setting the Shell Environment,'' p. 325
The more command has only one real drawback: You can't go backward in a file and redisplay
a previous screen. However, you can go backward in a file by using less.
Using less to View a File
One disadvantage to using the less command is that you can't use an editor on a file being
displayed. However, less makes up for this deficiency by allowing you to move forward and
backward through a file.
The less command works almost the same way that more does. To page through a file, type
the command less filename. One screen of data is then displayed. To advance to the next
screen, press the spacebar as you do with the more command.
430 Chapter 19 Understanding the File and Directory System
To move backward in a file, press the b key. To go to a certain position expressed as a
percentage of the file, press p and specify the percentage at the : prompt.
Searching Through a File and Escaping to the Shell
Using the less and more commands, you can search for strings in the file being displayed.
The less command, however, allows you to search backward through the file as well. You
can use the search syntax less / string to search backward through the file. With the less
and more commands, if a string is found, a new page is displayed with the line containing the
matching string at the top of the screen. With less, pressing the n key repeats the previous
search.
The more and less commands also allow you to escape to the shell by using the ! command.
When you escape to the shell by using the ! command, you're actually in a subshell; you must
exit the subshell just as you do when you log out from a session. Depending on which shell
you're using, you can press Ctrl+d or type exit to return to the same screen in more or less
that you escaped from. If you press Ctrl+d and get a message to use logout instead of Ctrl+d,
you use the logout command.
Viewing Files in Other Forms
Other commands display the contents of files in different forms. For example, if you want to
look at the contents of a binary file, you can display it by using the od command, which stands
for octal dump. The od command displays a file in octal notation, or base 8. By using various
flags, od can display a file in decimal, ASCII, or hexadecimal (base 16) notation.
Octal, Decimal, and Hexadecimal Notation
Representing binary data is an intriguing problem. If the binary data represents ASCII, you have no problem
displaying it (ASCII is, after all, what you expect when you look at most files). If the file is a program, however,
the data most likely can't be represented as ASCII characters. In that case, you have to display it in some
numerical form.
The early minicomputers used 12-bit words. Today, of course, the computer world has settled on the 8-bit byte
as the standard unit of memory. Although you can represent data in the familiar decimal (base 10) system, the
question becomes what to displayÐa byte, a word, or 32 bits? Displaying a given number of bits compactly
requires that base 2 be raised to the required number of bits. With the old 12-bit systems, you could represent
3
all 12 bits with four numbers (represented by 2 , which was the octal or base 8 format). Because early UNIX
systems ran on these kinds of minicomputers, much of the UNIXÐand, thus, LinuxÐnotation is in octal. Any byte
can be represented by a three-digit octal code that looks like this (this example represents the decimal value of
8):
\010
Because the world has settled on an 8-bit byte, octal is no longer an efficient way to represent data.
4
Hexadecimal ( base 16 or 2 ) is a better way. An 8-bit byte can be represented by two hexadecimal digits; a
byte whose decimal value is 10 is represented as 0A in hexadecimal.
The od command lets you choose how to display binary data. The general form of the
command is one of the following:
431
Viewing the Contents of a File
od [option]... [file]...
od ±traditional [file] [[+] offset [[+] label]]
Table 19.7 summarizes the flags you can use with od.
Table 19.7 The od Command Flags
Short Flag Full Flag Description
-A --address-radix=radix Decide how file offsets are printed
-N --read-bytes=bytes Limit dump to bytes input bytes per file
-j --skip-bytes=bytes Skip bytes input bytes first on each file
-s --strings[=bytes] Output strings of at least bytes graphic
characters
-t --format=type Select output format or formats
-v --output-duplicates Don't use * to mark line suppression
-w --width[=bytes] Output bytes bytes per output line
--traditional Accept arguments in pre-POSIX form
--help Display this help and exit
--version Output version information and exit
radix in Table 19.7 stands for number system and is d for decimal, o x for
for octal,
hexadecimal, or n for none. bytes is hexadecimal with a prefix of 0x
0X; it's multiplied by
or
512 with a b suffix, by 1,024 with k, and by 1,048,576 with an m suffix. -s without a number
implies 3; -w without a number implies 32. By default, od uses -A o -t d2 -w 16. Part
The pre-POSIX format specifications in Table 19.8 can be intermixed with the commands in
III
Table 19.7, and their effects accumulate. Ch
19
Table 19.8 Pre-POSIX Format Specifications for od
Short Flag POSIX Equivalent Description
-a -t a Select named characters
-b -t oC Select octal bytes
-c -t c Select ASCII characters or backslash escapes
-d -t u2 Select unsigned decimal shorts
-f -t fF Select floats
-h -t x2 Select hexadecimal shorts
-I -t d2 Select decimal shorts
-l -t d4 Select decimal longs
-o -t o2 Select octal shorts
-x -t x2 Select hexadecimal shorts
432 Chapter 19 Understanding the File and Directory System
For older syntax (second-call format), offset means -j offset. label is the pseudo-address
at first byte printed, incremented when the dump is progressing. For offset and label, an 0x
or 0X prefix indicates hexadecimal. Suffixes may be . (dot) for octal and may be multiplied by
512. The type parameter is made up of one or more of the specifications listed in Table 19.9.
Table 19.9 Type Parameters
Parameter Description
a Named character
c ASCII character or backslash escape
d[size] Signed decimal, size bytes per integer
f[size] Floating point, size bytes per integer
o[size] Octal, size bytes per integer
u[size] Unsigned decimal, size bytes per integer
x[size] Hexadecimal, size bytes per integer
In Table 19.9, size is a number and also may be C for sizeof(char), S for sizeof(short),
I for sizeof(int), or L for sizeof(long). If type is f, size may also be F for sizeof(float),
D for sizeof(double), or L for sizeof(long double).
sizeof is a C language function that returns the number of bytes in the data
structure passed as the parameter. For example, you use the following function call to
determine the number of bytes in an integer on your system because the number of
bytes in an integer is system-dependent:
sizeof(int);
Searching for Files
If you can't find a file by looking with the ls command, you can use the find command. The
find command is an extremely powerful tool, which makes it one of the more difficult
commands to use. The find command has three parts, each of which can consist of multiple
subparts:
& Where to look
& What to look for
& What to do when you find it
433
Searching for Files
If you know the name of a file but don't know where in the Linux file structure it's located,
the simplest case of the find command works like this:
find / -name filename -print
Be careful when you're searching from the root directory. On large systems, searching
every directory can take a long time, beginning with the root directory and continuing
through every subdirectory and disk (and remotely mounted disk) before finding
what you're looking for.
Limiting your search to one or two directories, at most, might be more prudent. For example,
if you know that a file is probably in the /usr or /usr2 directory, you can use the following
command instead:
find /usr /usr2 -name filename -print
You can use many different options with find; Table 19.10 lists just a few. To see all the
available options, use the man find command.
Table 19.10 A Sample of the find Command Flags
Command Description
-name file The file variable can be the name of a file or a wild-carded
filename. If it's a wild-carded filename, every file that
Part
matches the wildcards is selected for processing.
-links n Any file that has n or more links to it is selected for III
processing. Replace n with the number you want to check.
Ch
-size n[c] Any file that occupies n or more 512-byte blocks is selected
for processing. A c appended to n means to select any file 19
that occupies n or more characters.
-atime n With this command, you can select any file that has been
accessed in the past n days. Note that the act of looking for
a file with find modifies the access date stamp.
-exec cmdfind /home/jack After you select a list of files, you can run a Linux command
-exec chown jack {} \; that uses the selected files as an argument. Two simple rules
are associated with -exec: the name of a selected file is
represented by {}, and the command must be terminated by
an escaped semicolon, which is represented by \;. Suppose
you've created a user directory while logged in as root; thus,
all the files are owned by root, but the files should be owned
by the user. You would issue the following command to
change the owner of all the files in /home/jack and all
subdirectories from root to jack:
-print This instruction, the most often used, simply prints the
name and location of any selected files.
434 Chapter 19 Understanding the File and Directory System
Using the find command, you can perform many logical tests on files as well. For example, if
you want to find a selection of filenames that can't be collectively represented with wildcards,
you can use the oroption ( -o) to obtain a list, as shown here:
find /home (-name file1 -o -name file2) -print
You can combine as much selection criteria as you want with the find command. Unless you
specify the -o option, find assumes you mean and. For example, the command find -size 100
-atime 2 means find a file that's at least 100 blocks in size and that was last accessed at least
two days ago. You can use parentheses, as in the preceding example, to prevent ambiguous
processing of your criteria, especially if you combine and/or select criteria.
Changing File Time and Date Stamps
Each Linux file maintains three time and date stamps: the date of the file's creation, the date
of the file's last modification, and the date of the last access. You can't change the file creation
date artificially except by deliberately copying and renaming a file. Whenever a program
reads or opens a file, the file's access date stamp is modified. As you learned in the preceding
section, using the find command also causes the access date to be modified.
If a file is modified in any wayÐthat is, if it's written to, even if the file is actually not
modifiedÐthe file modification and file access date stamps are updated. The date stamps on a
file are useful if you need to back up selectively only files that have been modified since a
given date. You can use the find command for this purpose.
If you want to modify the date stamps on a file without actually modifying the file, you can do
so by using the touch command. By default, touch updates the access and modification date
stamps on a file with the current system date. By default, if you attempt to touch a file that
doesn't exist, touch creates the file.
You can use touch to fool a command that checks for dates. For example, if your system runs
a backup command that backs up only files modified after a particular date, you can touch a
file that hasn't been changed recently to make sure that it's picked up.
You can use the following three tags with the touch command to modify its default behavior:
-a Updates only the file's access date and time stamp
-m Updates only the file's modification date and time
stamp
-c Prevents touch from creating a file if it doesn't already
exist
The default syntax for touch is the following:
touch -am filelist
435
Compressing Files
Compressing Files
If space is tight on a system, or you have large ASCII files that aren't used often, you can
reduce the size of the files by compressing them. The standard Linux utility for compressing
files is gzip. The gzipcommand can compress an ASCII file by as much as 80 percent. Most
UNIX systems also provide the command compress, which typically is used with tar to
compress groups of files for an archive. A file compressed with the compress command ends
with a.Z extensionÐfor example, archive1.tar.Z. Red Hat's distribution also provides the
zip and unzip programs for compressing and archiving lists of files.
Tip #111 from
Jack For those of you who like GUIs rather than a command line interface (CLI), check
out;programs;TkZip TkZip. You can find the program at:
http:// www.pcnet.com/~proteus/TkZip/TkZip.html
Tip #112 from
Jack Compressing is helpful before you mail or back up a file.
If a file is successfully compressed with the command gzip filename, the compressed file is
named filename.gz, and the original file is deleted. To restore the compressed file to its
original components, you can use the gunzip filename command. Part
III
Tip #113 from Ch
Jack You don't have to append the .gz to the filename when you decompress a file. The 19
.gz extension is assumed by the gunzip command.
If you want to keep the file in its compressed form but want to pipe the data to another
command, you can use the zcat command. The zcat command works just like the cat
command but requires a compressed file as input. zcat decompresses the file and then prints
?
it to the standard output device.
See ``Using Pipes to Start Multiple Processes'' p. 369
For example, if you've compressed a list of names and addresses stored in a file named
namelist, the compressed file is named namelist.gz. If you want to use the contents of the
compressed file as input to a program, you can use the zcat command to begin a pipeline, as
follows:
zcat namelist | program1 | program2 ...
436 Chapter 19 Understanding the File and Directory System
zcat suffers from the same limitations as does cat: It can't go backward within a file. Linux
offers a program called zless that works just like the less command, except that zless
operates on compressed files. The same commands that work with less also work with zless.
The compress command's legal status is in limbo; someone has claimed patent infringement.
The compression program of choice for Linux is the freely distributed compression utility
gzip. The gzip command has none of the potential legal problems of compress, and almost
all the compressed files installed by Linux were compressed with gzip. gzip should work with
most compressed files, even those compressed with the older compress program.
If you're familiar with PKWARE's PKZIP line of products, you can use the zip and unzip
programs provided with the Red Hat distribution. The zip command compresses several files
and stores them in an archive, just like PKZIP. The unzip command extracts files from an
archive. See the man pages zip/unzip for more information.
Case Study: Undeleting Files
What happens if you delete a file by mistake? Many Windows systems have an undelete
command, but alas most Linux distributions do not. Although you can recover some
undeleted files, it is best not to delete them in the first place.
Tip #114 from
Jack Your first line of defense is to back up your system on a regular basis. Chapter 12,
``Backing Up Data,'' provides information on how to back up your system. Then, if you
do delete a file, you can retrieve it from the backup.
To help prevent you from deleting files, many Linux distributions, such as Red Hat and
Caldera, alias the rm and rmdir commands to rm -i and rmdir -i to ask you for confirmation
before executing the command. You can also set the file permissions to important files to
modes below 440 so that you need to explicitly confirm their deletion.
You should not do everyday work as root because you can do a lot of damage to your
system as a superuser, especially to files of other users. For example, if you give the
command
rm -rf *
from the / directory, you wipe out your entire system!
DO NOT USE THE rm -rf COMMAND UNLESS YOU ARE ABSOLUTELY SURE WHAT
YOU ARE DOING!
437
Case Study: Undeleting Files
You can find detailed information on recovering text and binary files in the following
How-Tos:
& Ext2fs Undeletion mini How-To:
http://metalab.unc.edu/LDP/HOWTO/mini/Ext2fs-Undeletion.html
& Tips How-To:
http://metalab.unc.edu/LDP/HOWTO/Tips-HOWTO.html
Both techniques work best on a single-user systemÐthat is, a personal workstation rather
than a serverÐbecause Linux reuses disk space and can thus reuse the space holding the
deleted data at any time. This process is much more likely to happen on a server system. The
best course of action is to unmount the file system and cease any activity as soon as you realize
?
you've deleted a needed file.
See ``Unmounting File Systems,'' p. 448
The text-recovery process suggested by Michael Hamilton in the ``Tips How-To'' uses the
egrep and strings commands. This technique requires that you remember some of the text
in the file, hopefully several words' worth of text. You also need to know which partition
contained the file and have available disk space on another partition to save the recovery data.
You cannot save the recovery data onto the same partition as the deleted file because your
recovery actions might overwrite the data you're trying to undelete.
To try to undelete a file, you must figure out which partition has your file. You can use the
pwd and df commands to help identify the partition, as in the following:
[tackett@ns ~]$ cd
[tackett@ns ~]$ pwd Part
/home/tackett
Filesystem 1024-blocks Used Available Capacity Mounted on III
/dev/hda1 1536971 301132 1156411 21% / Ch
/dev/hdb2 2554589 346646 2075853 14% /home
19
The output indicates that /home/tackett exists on the partition /dev/hdb2. If you think
something might write to a directory on /dev/hdb2 before you can perform the undelete
function, then you might want to unmount the file systems on that partition. For example, to
unmount the /home directory, you can use the following command (note that no n appears in
this command name):
umount /home
You must be root to unmount the directory, and you need to be in another partition to do so;
otherwise, you get a device busy error message.
Next, you need to search this file system to find the text of the deleted file by using the egrep
command. You also need space on a different file system to place the results of your search.
From the preceding example, you have only two partitions and thus have to write your search
data to the / directory on /dev/hda1 by using this command:
egrep -150 'Four score and seven years ago' /dev/hdb2 > /tmp/recovery
438 Chapter 19 Understanding the File and Directory System
This command tells egrep to search for the phrase Four score and seven years ago on /dev/hdb2
and to save the text 150 lines before the phrase and 150 lines after the phrase to the file
/tmp/recovery.
Make sure to look for text and an appropriate number of lines before and after the
search phrase. If you are trying to recover a small shell script, maybe looking for 50
lines is sufficient. If you are trying to recover a 1,000-line program, then 1,500 lines
might be a more appropriate value.
Also, realize that searching an entire partition may take some time, especially if you
specify a large number of lines to recover before and after your search phrase.
After saving the file, you can then use the command strings to narrow the search:
strings /tmp/recovery | more
The strings command makes sure that you get only ASCII data to view; otherwise, you
might see garbage characters you can't recognize from the recovery file.
If you need to undelete a binary file, you can follow the steps outlined by Aaron Crane in his
``Ext2fs Undeletion mini How-To."
CHAPTER
20
Managing File Systems
In this chapter by Jack Tackett, Jr.
Understanding File Systems 440
Mounting and Unmounting File Systems 444
Maintaining File Systems 449
Using the fsck Command 449
Creating and Formatting File Systems 451
Project: Using Swap Files and Partitions 459
440 Chapter 20 Managing File Systems
Understanding File Systems
File systems form the basis for all data on a Linux system. Linux programs, libraries, system
files, and user files all reside on file systems. Proper management of file systems is critical
because all your data and programs exist on top of file systems.
Many of the steps outlined in this chapter are performed automatically when you install
Linux. However, you should learn to manage your file systems so that you can create,
manage, and maintain your Linux system. Understanding file system management is critical
to successful system administration. Your file system must work properly for your Linux
system to work at all.
Under Linux, the file space that's visible to users is based on a tree structure, with the root at
the top. The various directories and files in this space branch downward from the root. The
top directory, /, is known as the root directory. Figure 20.1 gives a graphical example of a tree
structure.
Figure 20.1
/
Picture the
Linux file
system as an
upside-down /bin /sbin /usr /etc /lib /var
tree, with the
root at the top
and the
branches and
/bin /etc /lib /spool /adm
leaves spread-
ing downward.
/mail /news
To users, this directory tree looks like a seamless entity; they just see directories and files. In
reality, many of the directories in the file tree are physically located on different partitions on
a disk, on different disks, or even on different computers. When one of these disk partitions is
attached to the file tree at a directory known as a mount point, the mount point and all
directories below it are referred to as a file system.
The Linux operating system is made up of several directories and many different files.
Depending on how you selected your installation, these directories may be different file
systems. Typically, most of the operating system resides on two file systems: the root file
system, known as /, and a file system mounted under /usr (pronounced user).
If you change directories to the root directory by using the cd / command and ask for a
directory listing, you see several directories. They make up the contents of the root file
system and provide the mount points for other file systems as well.
The/bin directory contains executable programs, known as binaries. (In fact, the directory
named/bin is short for binary.) These programs are essential system files. Many Linux
commands, such as ls, are actually programs found in this directory.
441
Understanding File Systems
The /sbin directory is also used to store system binary files. Most files in this directory are
used for system administration purposes.
The /etc directory is very important, containing many of the Linux system configuration
files. Essentially, these files give your Linux system its ``personality.'' The password file,
passwd, is found here, as is the list of file systems to mount at startup, fstab. Also, this
directory contains the startup scripts for Linux, the list of hosts with IP addresses that you
want permanently recorded, and many other types of configuration information.
The shared libraries that programs use when they run are stored in the /lib directory. By
using shared libraries, many programs can reuse the same code, and these libraries can be
stored in a common place, thus reducing the size of your programs at runtime.
The /dev directory contains special files known as device files, which are used to access all the
different types of hardware on your system. For example, the /dev/mouse file is for reading
input from the mouse. By organizing access to hardware devices in this way, Linux effectively
makes the interface to a hardware device look like any other piece of software. This means
that you, in many cases, can use the same syntax that you use with software to perform
operations on computer hardware devices. For example, to create a tape archive of your home
directory on a floppy drive, you can use the following command:
tar -cdf /dev/fd0 ~tackett
?
/dev/fd0 indicates that the tar command should use the floppy drive identified by fd0.
See ``Using tar,'' p. 265
Many of the devices in the /dev directory are in logical groups. Table 20.1 lists some of the
most commonly used devices in the /dev directory.
Table 20.1 Commonly Used Devices in the /dev Directory
Device File Description
/dev/console The system console, which is the computer monitor physically connected to
Part
your Linux system.
III
/dev/hd The device driver interface to IDE hard drives. The /dev/hda1 device refers
to the first partition on hard drive hda. The device /dev/hda refers to the Ch
entire hard disk hda.
20
/dev/sd The device driver interface for SCSI disks. The same conventions for SCSI
disks and partitions apply as they do to the IDE /dev/hd devices.
/dev/fd Device drivers that provide support for floppy drives. /dev/fd0 is the first
floppy drive, and /dev/fd1 is the second floppy drive.
/dev/st The device driver for SCSI tape drives.
/dev/tty Device drivers that provide different consoles for user input. The name
comes from the time when terminals known as teletypes were physically
hooked to a UNIX system. Under Linux, these files provide support for the
virtual consoles that you can access by pressing Alt+F1 through Alt+F6.
These virtual consoles provide separate simultaneous local login sessions.
442 Chapter 20 Managing File Systems
Table 20.1 Commonly Used Devices in the /dev Directory
Device File Description
/dev/pty Device drivers that provide support for pseudo-terminals, which are used for
remote login sessions such as login sessions using Telnet.
/dev/ttyS The serial interface ports on your computer. /dev/ttyS0 corresponds to
COM1 under MS-DOS. If you have a serial mouse, /dev/mouse is a
symbolic link to the appropriate ttys device that your mouse is connected to.
/dev/cua Special call-out devices used with modems.
/dev/null A very special deviceÐessentially a black hole. All data written to /dev/null
is lost forever. Writing to this device file can be very useful if you want to
run a command and throw away the standard output or the standard error.
Also, if /dev/null is used as an input file, a file of zero length is created, aka
a data sink.
The /proc directory is actually a virtual file system. It's used to read process information from
memory.
Tip #115 from
Jack You can query system information from the /proc file system. For example, to check
on the amount of memory your system is using, you can enter the following:
cat /proc/meminfo
This command provides the following information:
total: used: free: shared: buffers: cached:
Mem: 31535104 30867456 667648 30359552 520192 7835648
Swap: 0 0 0
MemTotal: 30796 kB
MemFree: 652 kB
MemShared: 29648 kB
Buffers: 508 kB
Cached: 7652 kB
SwapTotal: 0 kB
SwapFree: 0 kB
The /tmp directory is used to store temporary files that programs create when running. If you
have a program that creates many large temporary files, you might want to mount the /tmp
directory as a separate file system rather than just have it as a directory on the root file system.
If /tmp is left as a directory on the root file system and has many large files written to it, the
root file system can fill up.
443
Understanding File Systems
Tip #116 from
Jack Red Hat provides a program called tmpwatch that deletes files in specified directories
at specified times. For other distributions, you can create acron job to perform the
crontab, use the
same action. In root's following command:
find /tmp -atime 2 -exec rm {} \;
This command deletes all files not accessed in the last two days.
? See ``Searching for Files,'' p. 432
The /home directory is the base directory for user home directories. It's common to mount
this directory as a separate file system so that users can have plenty of room for their files. In
fact, if you have many users on your system, you might need to separate /home into several file
systems. To do so, you could create subdirectories such as /home/staff and /home/admin for
staff members and administrators, respectively. You can mount each of them as different file
systems and then create the users' home directories under them.
The /var directory holds files that tend to change in size over time. Typically, various system
log files are located below this directory. The /var/spool directory and its subdirectories are
used to hold data that's of a transitory nature, such as mail and news that's recently received
from or queued for transmission to another site.
Tip #117 from
Jack You can create other mount points under the / directory if you want. You might want
to create a mount point named /mnt/cdrom if you routinely mount CD-ROMs on
your system. In fact, Red Hat already creates such a mount point for you during
installation.
The /usr directory and its subdirectories are very important to the operation of your Linux
system. This directory contains several subdirectories with some of the most important
Part
programs on your system. Typically, subdirectories of /usr contain the large software
packages that you install. Table 20.2 describes some of the /usr subdirectories. The /usr III
directory is almost always mounted as a separate file system. Ch
20
Table 20.2 Important Subdirectories in the /usr File System
Subdirectory Description
/usr/bin This directory holds many of the executable programs found on your
Linux system.
/usr/etc This directory contains many miscellaneous system configuration files.
/usr/include Here and in the subdirectories of /usr/include, you find all the
include files for the C compiler. These header files define constants
and functions and are critical for C programming.
444 Chapter 20 Managing File Systems
Table 20.2 Important Subdirectories in the /usr File System
Subdirectory Description
/usr/g++-include This directory contains the include files for the C++ compiler.
/usr/lib This directory contains various libraries for programs to use during
linking.
/usr/man This directory contains the various manual pages for programs on
your Linux system. Below /usr/man are several directories that
correspond to the different sections of the man pages.
/usr/src This directory contains directories that hold the source code for
different programs on your system. If you get a package that you want
to install, /usr/src/packagename is a good place to put the source
before you install it.
/usr/local This directory is designed for local customizations to your system. In
general, much of your local software is installed in this directory's
subdirectories. The format of this directory varies on almost every
UNIX system you look at. One way to set it up is to have a
/usr/local/bin for binaries, a /usr/local/etc for configuration files,
a /usr/local/lib for libraries, and a /usr/local/src for source code.
You can mount the entire /usr/local directory tree as a separate file
system if you need a lot of room for it.
Mounting and Unmounting File Systems
By now, you should have a good feel for what a file system is. So how do you set up a
directory as a separate file system?
To mount a file system in the Linux directory tree, you must have a physical disk partition,
CD-ROM, or floppy disk that you want to mount. You also must make sure that the directory
to which you want to attach the file system, known as the mount point, actually exists.
Mounting a file system doesn't create the mount point directory. The mount point must exist
before you try to mount the file system. Suppose that you want to mount the CD-ROM in
drive /dev/sr0 under the mount point /mnt. A directory named /mnt must exist, or the mount
fails. After you mount the file system under that directory, all the files and subdirectories on
the file system appear under the /mnt directory. Otherwise, the /mnt directory is empty.
Tip #118 from
Jack You can use the command df. if you need to know which file system the current
directory is located on. The command's output shows the file system as well as the
free space available.
445
Mounting and Unmounting File Systems
Mounting File Systems Interactively
As you may have guessed by now, Linux uses the mount command to mount a file system. The
syntax of the mount command is as follows:
mount devicemountpoint
device is the physical device that you want to mount; mountpoint is the point in the file
system tree where you want it to appear.
Only superusers can use the mount command. This restriction helps ensure system
security. Several software packages that allow users to mount specific file systems,
especially floppy disks, are available.
mount accepts several command-line arguments in addition to the two mentioned previously
(see Table 20.3). If a needed command isn't given, mount attempts to figure it out from the
/etc/fstab file.
Table 20.3 Command-Line Arguments for the mount Command
Argument Description
-f Causes everything to be done except for the actual mount system call.
This argument ``fakes'' mounting the file system.
-v Provides additional information about what mount is trying to do
(Verbose mode).
-w Mounts the file system with read and write permissions.
-r Mounts the file system with read-only permission.
-n Mounts without writing an entry in the /etc/mtab file.
-t type Specifies the type of the file system being mounted. Valid types are
minux, ext, ext2, xiafs, msdos, hpfs, proc, nfs, umsdos, sysv, and Part
iso9660 (the default).
III
-a Causes mount to try to mount all file systems in /etc/fstab.
-o list_of_options Causes mount to apply the options specified to the file system being
Ch
mounted when followed by a comma-separated list of options. Many 20
options are available here; for a complete list, refer to the mount man
page.
Several forms of the mount command are very common. For example, the command
mount /dev/hdb3 /mnt mounts the hard disk partition /dev/hdb3 under the
directory /mnt. Similarly, mount -r -t iso9660/dev/sr0 /mnt mounts the SCSI
CD-ROM drive /dev/sr0, which is read-only and of the ISO 9660 file format, under
446 Chapter 20 Managing File Systems
the directory /mnt. And the command mount -vat nfs mounts all the NFS file
systems listed in the /etc/fstab file.
Tip #119 from
If a file system doesn't mount correctly, you can use the command mount -vf
Jack
device mountpoint to see what mount is doing. This command gives a verbose
listing and tells mount to do everything except mount the file system. This way, you
can fake out the mount command and get a lot of information about what it's trying
to do.
Mounting File Systems at Boot Time
Under most circumstances, the file systems that your Linux system uses don't change
frequently. For this reason, you can easily specify a list of file systems that Linux mounts
when it boots and that it unmounts when it shuts down. These file systems are listed in a
special configuration file named /etc/fstab, for file system table.
The /etc/fstab file lists the file systems to be mounted, one file system per line. The fields in
each line are separated by spaces or tabs. Table 20.4 lists the different fields in the /etc/fstab
file.
Table 20.4 Fields in the /etc/fstab File
Field Description
File System Specifier Specifies the block special device or the remote file system to be
mounted.
Mount Point Specifies the mount point for the file system. For special file systems
such as swap files, you can use the word none, which makes swap
files active but not visible within the file tree.
447
Mounting and Unmounting File Systems
Table 20.4 Fields in the /etc/fstab File
Field Description
Type Gives the file system type of the specified file system. The following
types of file systems are supported:
minixÐA local file system supporting filenames of 14 or 30
characters.
extÐA local file system with longer filenames and larger inodes.
(This file system has been replaced by the ext2 file system and
should no longer be used.)
ext2ÐA local file system with longer filenames, larger inodes, and
other features.
xiafsÐA local file system.
msdosÐA local file system for MS-DOS partitions.
hpfsÐA local file system for OS/2 High Performance File System
partitions.
iso9660ÐA local file system used for CD-ROM drives.
nfsÐA file system for mounting partitions from remote systems.
swapÐA disk partition or special file used for swapping.
umsdosÐA UMSDOS file system.
sysvÐA System V file system.
Mount Options Contains a comma-separated list of mount options for the file
system. At a minimum, it must contain the type of mount for the file
system. See the mount man page for more information on mount
options.
Dump Frequency Specifies how often the file system should be backed up by the dump
command. If this field isn't present, dump assumes that the file
system doesn't need to be backed up.
Pass Number Specifies in what order the file systems should be checked by the
fsck command when the system is booted. The root file system
should have a value of 1. All other file systems should have a value of
2. If a value isn't specified, the file system isn't checked for
consistency at boot time.
Part
Tip #120 from III
Jack I recommend you mount your file systems at boot time via the/etc/fstab file
instead of by using the mount command. Remember, only superusers can use mount.
Ch
20
The following is a sample fstab file:
# device directory type options
/dev/hda1 / ext2 defaults
/dev/hda2 /usr ext2 defaults
/dev/hda3 none swap sw
/dev/sda1 /dosc msdos defaults
/proc /proc proc none
448 Chapter 20 Managing File Systems
In this sample file, you can see several different file systems. First, notice that comments in
the file are prefixed by a # character. In thisfstab file, two normal Linux file systems are
mounted: the disk partitions /dev/hda1 and /dev/hda2. They are listed as being of type ext2
and are mounted under the root (/) and /usr directories, respectively.
The entry defaults, listed under the options field, indicates this file system should be
mounted by using a common set of default options. Specifically, the file system is mounted
read/write enabled, and it's to be interpreted as a block special device. All file I/O should be
done asynchronously. The execution of binaries is permitted, and the file system can be
mounted with the mount -a command. The Set UID (user ID) and Set GID (group ID) bits
on files are interpreted on this file system, and ordinary users aren't allowed to mount this file
?
system. As you can see, just typing defaults for the option is a lot easier.
See ``Creating a Swap Partition,'' p. 459
The partition /dev/hda3 is a swap partition that's used for kernel virtual-memory swap space.
Its mount point is specified asnone so that it doesn't appear in the file system tree. It still has
to be in the /etc/fstab file so that the system knows where it's physically located. Swap
partitions are also mounted with the option sw.
The /proc file system is a virtual file system that points to the process information space in
memory. As you can see, it doesn't have a corresponding physical partition to mount.
Tip #121 from
Jack For full information on all options available in the /etc/fstab file, refer to the man
page for fstab.
MS-DOS file systems can also be mounted automatically. The partition /dev/sda1 is the first
partition on the SCSI hard drive sda. It's mounted as an MS-DOS partition by specifying
msdos as the type and by giving /dosc as its mount point. You can place the mount point for
the MS-DOS file system anywhere; it is not required to be under the root directory.
Unmounting File Systems
Now that you know all sorts of stuff about mounting file systems, it's time to look at
how to unmount. You use the umount command to unmount file systems. You might want
to unmount a file system for several reasons: to check or repair a file system with fsck, to
unmount NFS-mounted file systems in case of network problems, or to unmount a file
system on a floppy drive.
This command is umount, not unmount. Make sure that you type it correctly.
Using the fsck Command
449
The three basic forms of the umount command are as follows:
umount device | mountpoint
umount -a
umount -t fstype
device is the name of the physical device to unmount; mountpoint is the mount point
directory name (you should specify only one or the other). The umount command has only
two command-line parameters: -a, which unmounts all file systems, and -t fstype, which
acts only on file systems of the type specified.
The umount command doesn't unmount a file system that's in use. For example, if
you have some file system mounted under /mnt and you try
cd /mnt
umount /mnt
you get an error telling you that the file system is busy. You have to change to a
different directory in another file system to unmount the file system mounted under
/mnt.
Maintaining File Systems
As the system administrator, you're responsible for maintaining the integrity of the file
systems. Typically, this means you have to check the file systems periodically for damaged or
corrupted files. Linux automatically checks file systems at boot time if they have a value
greater than 0 specified in the pass number field of the /etc/fstab file.
Part
The ext2 file system commonly used under Linux has a special flag known as a clean III
bit. If the file system has been synchronized and unmounted cleanly, the clean
Ch
bit is set on the file system. If the clean bit is set on a file system when Linux boots,
it's not checked for integrity.
20
Using the fsck Command
Checking your file systems occasionally for damaged or corrupt files is a good idea. Under
the Slackware distribution of Linux, you use the fsck (file system check) command to check
450 Chapter 20 Managing File Systems
your file systems. The fsck command is really a ``front end'' for a series of commands that are
designed to check specific file systems. The syntax for the fsck command is as follows:
fsck [-A] [-V] [-t fs-type] [-a] [-l] [-r] [-s] filesys
However, the most basic form of the command is this:
fsck filesys
Table 20.5 describes the command-line options for the fsck command.
Table 20.5 Command-Line Arguments for fsck
Argument Description
-A Goes through the /etc/fstab file and tries to check all file systems in one pass.
This option is typically used during the Linux boot sequence to check all
normally mounted file systems. If you use -A, you can't use the filesys
argument as well.
-V Prints additional information about what fsck is doing (Verbose mode).
-t fs-type Specifies the type of file system to be checked.
filesys Specifies which file system is to be checked. This argument can be a block
special device name, such as /dev/hda1, or a mount point, such as /usr.
-a Automatically repairs any problems found in the file system without asking any
questions. You should use this option with caution.
-l Lists all the filenames in the file system.
-r Asks for confirmations before repairing the file system.
-s Lists the superblock before checking the file system.
The fsck command is actually a front-end program that calls the command to check the file
system that matches the type you specify. To use this command, Linux needs to know the file
system type that it's checking. The easiest way to make sure that fsck calls the right
command is to specify a file system type with the -t option to fsck. If you don't use the
-t option, Linux tries to figure out the file system type by looking up the file system in
/etc/fstab and by using the file type specified there. If fsck can't find the file type
information in /etc/fstab, it assumes that you're using a Minix file system. For example,
the command
fsck -rV /dev/hda1
checks the file system on device /dev/hda1, provides verbose information, and asks for
confirmation before making changes.
The fsck command assumes that the file system you're checking is a Minix file
-t argument or by listing the
system if you don't tell it differentlyÐeither by using the
type in /etc/fstab. Because your Linux file systems are probably of type ext2 and
not Minix, you should be careful and make sure that fsck knows the correct type.
451
Creating and Formatting File Systems
Doing so is especially important if you're checking a file system that isn't listed in the
/etc/fstab file.
Unmounting a file system before checking it is a good idea. This way, you ensure that none of
the files on the file system are in use when they're being checked.
Remember, you can't unmount a file system if any of the files on it are busy. For
example, if a user is now in a directory on a file system that you try to unmount, you
get a message saying that the file system is busy.
Trying to check the root file system presents an additional problem. You can't directly
unmount the root file system because Linux must be able to access it in order to run. To
check the root file system, you should boot from a maintenance floppy disk that has a root file
system on it; then you can run fsck on your real root file system from the floppy by specifying
the special device name of your root file system. If fsck makes any changes to your file
system, it's important that you reboot your system immediately. Rebooting allows Linux to
reread important information about your file system and prevents your file system from
further corruption.
To prevent further corruption to your file system, you should be sure to reboot your
computer immediately after you run fsck if any changes were made to your file
system. You can use the shutdown -r command now or issue the reboot command
at the command line.
Part
Creating and Formatting File Systems III
Ch
When you add a new hard disk to your computer or want to change the partition information
on an old hard disk, you go through the steps of creating a file system from a raw disk. 20
Assuming that you've added a new hard disk to your system, you must set the disk partition
information and then create the actual file systems on the disk before Linux can use the disk.
To change disk partition information, you use the fdisk command. After you partition the
hard drive, you need to create the file systems by using the mkfs command.
Using fdisk to Create Disk Partitions
The fdisk command is used to create disk partitions and set the attributes that tell Linux
what type of file system is on a particular partition. If you installed Linux from scratch on an
452 Chapter 20 Managing File Systems
MS-DOS system, you had to run fdisk to change the disk partition information before you
could install Linux.
Using fdisk on a disk can destroy all data on the disk. Because fdisk completely
rewrites the file table on the disk, all your former files may be lost. Make sure that
you have a complete, current backup of your disks before using fdisk.
You should always run the fdisk command on an unmounted file system. fdisk is an
interactive, menu-driven program, not just a single command. To start fdisk, type this
command:
fdisk [drive]
drive is the physical disk drive that you want to work on. If you don't specify a disk, the disk
/dev/hda is assumed. For example, to run fdisk on the second IDE hard drive in your system,
enter the following at the superuser command prompt:
fdisk /dev/hdb
Because fdisk is a menu-driven program, several different commands are available when
you're using fdisk, as summarized in Table 20.6.
Table 20.6 Commands Available from the fdisk Menu
Command Description
a Toggles the bootable flag on a partition
c Toggles the DOS compatibility flag on a partition
d Deletes a partition
l Lists partition types known to fdisk
m Displays a menu listing of all available commands
n Adds a new partition
p Prints the partition table for the current disk
q Quits without saving any changes
t Changes the file system type for a partition
u Changes display/entry units
v Verifies the partition table
w Writes the table to disk and exits
453
Creating and Formatting File Systems
Table 20.6 Commands Available from the fdisk Menu
Command Description
x Lists additional functions for experts:
bÐMoves the beginning location of data in a partition
cÐChanges the number of cylinders
dÐPrints the raw data in the partition table
eÐLists extended partitions on disk
hÐChanges the number of heads on disk
rÐReturns to the main menu
sÐChanges the number of sectors on disk
fdisk can set the file system type of a disk partition to any of several different types. You
should use Linux fdisk only to create partitions used under Linux. For MS-DOS or OS/2
partitions, you should use the fdisk tool that's native to that operating environment and then
use Linux's fdisk to tag the partitions as Linux native or Linux swap.
Table 20.7 lists the partitions supported by Linux fdisk. Each partition type has an
associated hexadecimal code that identifies it. You must enter the appropriate code in fdisk
when you want to set a partition type.
Table 20.7 Partition Codes and Types in Linux fdisk
Hex Code Partition Type
0 Empty
1 DOS 12-bit FAT
2 XENIX root
3 XENIX usr
4 DOS 16-bit file system, less than 32MB
5 Extended
6 DOS 16-bit file system supporting more than 32MB Part
7 OS/2 High Performance File System (HPFS) III
8 AIX
Ch
9 AIX bootable
20
a OS/2 Boot Manager
40 Venix 80286
51 Novell?
52 Microport
63 GNU HURD
64 Novell NetWare
65 Novell NetWare
75 PC/IX
454 Chapter 20 Managing File Systems
Table 20.7 Partition Codes and Types in Linux fdisk
Hex Code Partition Type
80 Old MINIX
81 Linux/MINIX
82 Linux swap, used for swap files under Linux
83 Linux native, common Linux file system type
93 Amoeba
94 Amoeba BBT
a5 BSD/386
b7 BSDI file system
b8 BSDI swap file system
c7 Syrinx
db CP/M
e1 DOS access
e3 DOS R/O
f2 DOS secondary
ff BBT
The following sections show how to use fdisk. In these sections, you'll see an example of how
to use fdisk to set up the partitions on a hard disk for use by Linux. Assume that you want to
configure the first IDE drive in your system for Linux. Make sure that you have a backup of
your data because all data on your hard disk will be destroyed in the process. The name of the
first IDE hard disk is /dev/hda, which is the default device for Linux.
Running fdisk
You run fdisk by using this command:
# fdisk
fdisk responds with the following:
Using /dev/hda as default device!
Command (m for help):
This response tells you that fdisk is using disk /dev/hda as the device that you're working
with. Because this is the result you wanted, you're fine. You should always check to make sure
that you're really on the disk that you think you're on. Linux then displays the fdisk
command prompt.
Displaying the Current Partition Table
When you're working with fdisk, the first thing you should do is display the current
partition table. You do so by entering the p command:
455
Creating and Formatting File Systems
Command (m for help): p
Disk /dev/hda: 14 heads, 17 sectors, 1024 cylinders
Units = cylinders of 238 * 512 bytes
Device Boot Begin Start End Blocks Id System
Command (m for help):
This listing shows that the current disk, /dev/hda, has a geometry of 14 heads, 17 sectors, and
1,024 cylinders. The display units are in cylinders of 238 * 512 (121,856) bytes each. Because
the disk has 1,024 cylinders and each cylinder is 121,856 bytes, you can deduce that the disk
can hold 1,024 6121,856 = 124,780,544 bytes, or about 120MB. You can also see that
/dev/hda has no partitions.
Creating a New Partition
Now assume that you want to create a 100MB Linux file partition for user home directories
and a 20MB swap partition. Your next step is to use the n command to create a new partition:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-1023): 1
Last cylinder or +size or +sizeM or +sizeK (1-1023): + 100M
Using the n command to create a new partition displays another menu. You must choose
whether you want to create an extended partition or a primary partition. You typically should
create a primary partition unless you have more than four partitions on a disk. fdisk then asks
you for the partition number that you want to create. Because this is the first partition on the
disk, you answer 1. You're then prompted for the first cylinder for the partition. This
information determines where on the disk the data area starts. Again, because this will be the
first partition on the disk, you can start the partition at cylinder 1.
Part
The next line asks you how large you want the partition to be. You have several options as to
how to answer this question. fdisk accepts either a number, which it interprets as the size in III
cylinders, or the size in bytes, kilobytes, or megabytes. The size in bytes is specified as +bytes,
Ch
where bytes is the size of the partition. Similarly, +sizeK and +sizeM set the partition size to
size kilobytes or size megabytes, respectively. You know that you want a 100MB partition,
20
so the easiest answer to the prompt is +100M.
Rechecking the Partition Table
Now you should check the partition table again to see what fdisk has done:
Command (m for help): p
Disk /dev/hda: 14 heads, 17 sectors, 1024 cylinders
Units = cylinders of 238 * 512 bytes
456 Chapter 20 Managing File Systems
Device Boot Begin Start End Blocks Id System
/dev/hda1 1 1 861 102400 81 Linux/MINIX
Command (m for help):
The partition table shows that you have one partition, /dev/hda1, that goes from cylinder 1 to
cylinder 861 and uses 102,400 blocks. It's listed as being type 81, Linux/MINIX.
Creating the Swap Partition
Now you need to create the 20MB swap partition by using the remaining disk space. Creating
it is just like creating the first partition:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 2
First cylinder (862-1023): 862
Last cylinder or +size or +sizeM or +sizeK (862-1023): 1023
Tip #122 from
Jack It's usually better to go ahead and enter the size of the last partition in cylinders to
make sure that you use all the disk space.
Here, you specify partition number 2 for the second partition. When fdisk prompts for the
first cylinder, notice that it gives a range of 862 to 1023 because the first partition takes up
everything before cylinder 862. So you can enter 862 as the starting cylinder for the second
partition. In this case, you want to use all the remaining space on the disk for the swap
partition. You should have about 20MB left, but if you specify the size in megabytes, the
internal fdisk calculations could leave you with a couple of unused cylinders. So you can
enter 1023 for the last cylinder on the size prompt.
You might see an error similar to
Warning: Linux cannot currently use the last xxx sectors of this
partition.
where xxx is some number. You can ignore such an error. It's left over from the days
when Linux couldn't access file systems larger than 64MB.
457
Creating and Formatting File Systems
Making Sure the Sizes Are Correct
At this point, you've created both partitions that you wanted to create. You should take a look
at the partition table one more time to check that the sizes are correct:
Command (m for help): p
Disk /dev/hda: 14 heads, 17 sectors, 1024 cylinders
Units = cylinders of 238 * 512 bytes
Device Boot Begin Start End Blocks Id System
/dev/hda1 1 1 861 102400 81 Linux/MINIX
/dev/hda2 862 862 1023 19159 81 Linux/MINIX
Command (m for help):
As you can see,/dev/hda1 uses cylinder 1 through cylinder 861 with a size of 102,400 blocks,
which is approximately 100MB. Partition /dev/hda2 goes from cylinder 862 to cylinder 1023
with a size of 19,156 blocks, or almost 20MB.
Changing the Partition Type
Next, you need to change the partition type for each partition. To changethe partition type,
you use the t command at the fdisk command prompt. The most common choice for a
standard Linux file system partition is to set it to partition type 83, Linux native. Swap
partitions should be set to partition type 82, Linux swap, as shown here:
Command (m for help): t
Partition number (1-4): 1
Hex code (type L to list codes): 83
Command (m for help): t
Partition number (1-4): 2
Hex code (type L to list codes): 82
When you use the t command, you're prompted for the partition number that you want to
change. You're then prompted for the hex code for the file system ID that you want to set the
partition to. Typically, Linux file systems are set to type 83 for normal file systems and type
82 for swap partitions. You can type l at this point to see a list of file systems if you want.
Part
Finishing Up
III
Now that you've created the partitions and labeled them, you should take one last look at the
Ch
partition table before you exit just to make sure that everything is okay. You do so as follows:
Command (m for help): p 20
Disk /dev/hda: 14 heads, 17 sectors, 1024 cylinders
Units = cylinders of 238 * 512 bytes
Device Boot Begin Start End Blocks Id System
/dev/hda1 1 1 861 102400 83 Linux native
/dev/hda2 862 862 1023 19159 82 Linux swap
Command (m for help):
As you can see, the partitions are in the right place, they're the right size, and the file system
types are set correctly. Finally, you need to use the w command to write the partition table to
disk and exit:
458 Chapter 20 Managing File Systems
Command (m for help): w
#
None of the changes that you make during an fdisk session take effect until you write them
to disk by using the w command. You can always quit by using the q command and not save
any changes. This said, you should still always have a backup of any disk that you want to
modify with fdisk.
After you make changes to a disk with fdisk, you should reboot the system just to make sure
that Linux has the updated partition information in the kernel.
Using mkfs to Build a File System
After you create a file system partition with fdisk, you must build a file system on it before
you can use it for storing data. You do so by using the mkfs command. As an analogy, think of
building a parking lot. If you think of fdisk as physically building the parking lot, mkfs is the
part of the process that paints the lines so that the drivers know where to park.
Just like fsck is a ``front-end'' program for checking different types of file systems, mkfs
actually calls different programs to create the file system, depending on what file system type
you want to create. The following is the syntax of the mkfs command:
mkfs [-V] [-t fs-type] [fs-options] filesys [blocks]
filesys is the device of the file system that you want to build, such as /dev/hda1.
The mkfs command also accepts the name of a mount point, such as /home, as the
file system name. You should be extremely careful about using a mount point. If you
run mkfs on a mounted ``live'' file system, you might very well corrupt all the data on
that file system.
Table 20.8 lists the various command-line parameters that you can specify with mkfs.
Table 20.8 Command-Line Parameters for the mkfs Command
Option Description
-V Causes mkfs to produce verbose output, including all file system-specific
commands that are executed. Specifying this option more than once inhibits
execution of any file system-specific commands.
-t fs- type Specifies the type of file system to be built. If the file system type isn't specified,
mkfs tries to figure it out by searching for filesys in /etc/fstab and using the
corresponding entry. If the type can't be deduced, a Minix file system is created.
459
Project: Using Swap Files and Partitions
Table 20.8 Command-Line Parameters for the mkfs Command
Option Description
fs-options Specifies file system-specific options that are to be passed to the actual file
system-builder program. Although not guaranteed, the following options are
supported by most file system builders:
-C ÐChecks the device for bad blocks before building the file system
-l file-name ÐReads a list of the bad blocks on the disk from file-name
-vÐTells the actual file system builder program to produce verbose output
filesys Specifies the device on which the file system resides. This parameter is required.
blocks Specifies the number of blocks to be used for the file system.
Although -t fs-type is an optional argument, you should get in the habit of specifying the
file system type. Just like fsck, mkfs tries to figure out the type of the file system from the
/etc/fstab file. If it can't figure out the type, it creates a Minix file system by default. For
a normal Linux file system, you probably want an ext2 partition instead.
Project: Using Swap Files and Partitions
Swap space on your Linux system is used for virtual memory. A complete discussion of all the
issues involved with virtual memory is beyond the scope of this book. Any good general
computer operating system textbook describes the issue in detail.
Linux supports two types of swap space: swap partition and swap files. A swap partition is a
physical disk partition with its file system ID set to type 82, Linux swap, and it is dedicated for
use as a swap area. A swap file is a large file on a normal file system that's used for swap space.
You're better off using a swap partition instead of a swap file. All access to a swap file is
performed through the normal Linux file system. The disk blocks that make up the swap file
are probably not contiguous; therefore, performance isn't as good as it is with a swap
partition. I/O to swap partitions is performed directly to the device, and disk blocks on a swap
partition are always contiguous. Also, by keeping the swap space off a normal file system, you Part
reduce the risk of corrupting your regular file system if something bizarre happens to your III
swap file.
Ch
Creating a Swap Partition 20
To create a swap partition, you must first create a disk partition by using fdisk and tag it as
type 82, Linux swap. After you create the swap partition, you have two additional steps to
follow to make the swap partition active.
For the first step, you must prepare the partition in a manner similar to creating a file system.
Instead of mkfs, the command you use to prepare the partition is mkswap. The syntax of the
mkswap command is as follows:
mkswap [-c] devicesize_in_blocks
460 Chapter 20 Managing File Systems
device is the name of the swap partition, such as /dev/hda2, and size_in_blocks is the size of
the target file system in blocks. You can get the size in blocks by running fdisk and looking at
the partition table. In the example in the section ``Making Sure the Sizes Are Correct,'' the
size of /dev/hda2 was 19,159 blocks. Linux requires that swap partitions be between 9 and
65,537 blocks in size. The -c argument tells mkswap to check the file system for bad blocks
when creating the swap space, which is a good idea.
Following the example in ``Making Sure the Sizes Are Correct,'' the command for setting up
a swap partition on /dev/hda2 is as follows:
mkswap -c /dev/hda2 19159
After you run mkswap to prepare the partition, you must make it active so that the Linux
kernel can use it. The command to make the swap partition active for this second step is
swapon. The syntax for the swapon command is as follows:
swapon filesys
filesys is the file system that you want to make available as swap space. Linux makes a call to
swapon -a during boot, which mounts all available swap partitions listed in the /etc/fstab
file.
Remember to put an entry for any swap partitions or swap files that you create into
the /etc/fstab file so that Linux can automatically access them at boot time.
Creating a Swap File
Swap files can be useful if you need to expand your swap space and can't allocate disk space to
create a dedicated swap partition. Setting up a swap file is almost identical to creating a swap
partition. The main difference is that you have to create the file before you can run mkswap
and swapon.
To create a swap file, you use the dd command, which is used for copying large chunks of
data. For a full description of this command, see the man page for dd. The main pieces of
information that you have to know before creating the file are the name of the swap file you
want to create and its size in blocks. A block under Linux is 1,024 bytes. For example, to
create a 10MB swap file named /swap, enter the following:
# dd if=/dev/zero of=/swap bs=1024 count=10240
of=/swap specifies that the file to be created is named /swap, and count=10240 sets the size of
the output file to be 10,240 blocks, or 10MB. You then can use mkswap to prepare the file as a
swap space:
# mkswap /swap 10240
Remember that you have to tell mkswap how big the file is. Before you run swapon, you need
to make sure that the file is completely written to disk by using the /etc/sync command.
461
Project: Using Swap Files and Partitions
Now you're ready to make the swap file active. As you do with the swap partition, you use the
swapon command to make the file active, as shown in this example:
# swapon /swap
If you need to get rid of a swap file, you must make sure that it's not active. You use the
swapoff command to deactivate the swap file, as in the following:
# swapoff /swap
You can then safely delete the swap file.
Part
III
Ch
20
CHAPTER
21
Managing NFS
In this chapter by Jack Tackett, Jr.
Understanding the Network File System 464
Installing NFS 464
Exporting an NFS File System 468
Understanding the /etc/exports File 469
Mounting NFS File Systems 471
Troubleshooting 472
464 Chapter 21 Managing NFS
Understanding the Network File System
The Network File System (NFS) is a system that allows you to mount file systems from a
different computer over a TCP/IP network. Using NFS, you can share data among PC, Mac,
UNIX, and Linux systems. Under NFS, a file system on a remote computer is mounted
locally and looks just like a local file system to users. The illusion of being mounted locally
has numerous uses. For example, you can have one machine on your network with a lot of
disk space acting as a file server. This computer has all the home directories of all your users
on its local disks. If you mount these disks via NFS on all your other computers, your users
can access their home directories from any computer.
Sun developed NFS during the 1980s and released the protocol to the Unix community as
described in RFC1094 (RFC stands for Request For Comment). Since no single authority
created the Internet, most of its development grew from these RFCs. The Web site
www.faqs.org provides a plethora of information on RFCs and specifically on the one
describing NFS. See the following URL for more information:
http://www.faqs.org/rfcs/rfc1094.html
Like HTTP, NFS is a stateless protocol, thus each transaction is complete in and of itself.
NFS has three essential components:
& The computers with the file systems that you want to mount remotely via NFS must
be able to communicate with each other via a TCP/IP network.
& The computer with the file system that you're interested in as a local file system must
make that file system available to be mounted. This computer is known as the server,
and the process of making the file system available is known as exporting the file system.
& The computer that wants to mount the exported file system, known as the client, must
mount the file system as an NFS file system via the /etc/fstab file at boot time or
mount
?
interactively via the command.
See ``Mounting File Systems at Boot Time,'' p. 446
NFS under Linux is not as robust as NFS on other systems. For light to moderate usage, such
as serving up email for 500-1000 users from an NFS mounted file system, you should be OK.
But for larger operations you may find problems using the file system via NFS. Developers,
including Alan Cox and Olaf Kirch, are working to improve NFS performance under Linux.
Installing NFS
Most distributions, including Red Hat, Caldera, and Debian, automatically install the
necessary files needed to run NFS, but you should know the files and their functions, as listed
in Table 21.1, in order to help troubleshoot problems.
465
Installing NFS
Table 21.1 NFS Components
Component Description
rpc.portmaster Routes remote procedure calls (RPC) to the appropriate
daemonsÐ nfsd for NFS services.
rpc.statd Kernel statistics daemon
rpc.rquotad Remote quota server daemon
rpc.nfsd Provides access to the local file system in response to remote
procedures calls for NFS requests.
rpc.mountd The daemon mounts and unmounts file systems.
mount Local command that mounts file systems to local directories. One
uses this command to mount a remote file system via NFS.
umount Unmounts local file systems thus making them unavailable to
remote hosts.
/etc/exports This config file specifies which local file systems are available to
remote systems via NFS.
/var/lock/subsys/nfs Lock file used to prevent running multiple copies of NFS
programs.
Use the pmap_dump command to check the status of your system's RPC daemons. The
command displays all the registered RPC programs currently running on your system, like so:
[root@web /root]# /usr/sbin/pmap_dump
100000 2 tcp 111 rpcbind
100000 2 udp 111 rpcbind
100024 1 udp 1017 status
100024 1 tcp 1019 status
100011 1 udp 604 rquotad
100011 2 udp 604 rquotad
100005 1 udp 614 mountd
100005 1 tcp 616 mountd
100005 2 udp 619 mountd
100005 2 tcp 621 mountd
100005 3 udp 624 mountd
100005 3 tcp 626 mountd
100003 2 udp 2049 nfs
100021 1 udp 1026 nlockmgr
100021 3 udp 1026 nlockmgr
100021 1 tcp 1024 nlockmgr
100021 3 tcp 1024 nlockmgr Part
The NFS programs should startup automatically at boot time, but should you need to start, III
stop, query, or restart them you can use the script located in /etc/rc.d/init.d called nfs as
Ch
shown in Listing 21.1.
21
466 Chapter 21 Managing NFS
Listing 21.1 nfs init Script
#!/bin/sh
#
#nfs This shell script takes care of starting and stopping
# the NFS services.
#
# chkconfig: 345 60 20
# description: NFS is a popular protocol for file sharing across TCP/IP \
# networks. This service provides NFS server functionality, \
# which is configured via the /etc/exports file.
# probe: true
# Source function library.
. /etc/rc.d/init.d/functions
# Source networking configuration.
if [ ! -f /etc/sysconfig/network ]; then
exit 0
fi
. /etc/sysconfig/network
# Check that networking is up.
[ ${NETWORKING} = ``no'' ] && exit 0
[ -x /usr/sbin/rpc.nfsd ] || exit 0
[ -x /usr/sbin/rpc.mountd ] || exit 0
[ -x /usr/sbin/exportfs ] || exit 0
[ -f /etc/exports ] || exit 0
# Number of servers to be started uo by default
RPCNFSDCOUNT=8
# See how we were called.
case ``$1'' in
start)
# Start daemons.
action ``Starting NFS services: '' /usr/sbin/exportfs -r
echo -n ``Starting NFS statd: ''
daemon rpc.statd
echo
echo -n ``Starting NFS quotas: ''
daemon rpc.rquotad
echo
echo -n ``Starting NFS mountd: ''
daemon rpc.mountd
echo
echo -n ``Starting NFS daemon: ''
daemon rpc.nfsd $RPCNFSDCOUNT
echo
touch /var/lock/subsys/nfs
;;
stop)
# Stop daemons.
action ``Shutting down NFS services: '' /usr/sbin/exportfs -au
echo -n ``Shutting down NFS mountd: ''
killproc rpc.mountd
echo
echo -n ``Shutting down NFS daemon: ''
killproc nfsd
echo
echo -n ``Shutting down NFS quotas: ''
467
Installing NFS
Listing 21.1 continued
killproc rpc.rquotad
echo
echo -n ``Shutting down NFS statd: ''
killproc rpc.statd
echo
rm -f /var/lock/subsys/nfs
;;
status)
status rpc.statd
status rpc.mountd
status nfsd
status rpc.rquotad
;;
restart)
echo -n ``Restarting NFS services: ''
echo -n ``rpc.statd ''
killall -HUP rpc.statd
echo -n ``nfsd ''
killall -HUP nfsd
echo -n ``rpc.mountd ''
killall -HUP rpc.mountd
echo -n ``rpc.quotad ''
killall -HUP rpc.rquotad
touch /var/lock/subsys/nfs
echo ``done.''
;;
reload)
/usr/sbin/exportfs
touch /var/lock/subsys/nfs
;;
probe)
if [ ! -f /var/lock/subsys/nfs ] ; then
echo start; exit 0
fi
/sbin/pidof rpc.mountd >/dev/null 2>&1; MOUNTD="$?''
/sbin/pidof nfsd >/dev/null 2>&1; NFSD="$?''
if [ $MOUNTD = 1 -o $NFSD = 1 ] ; then
echo restart; exit 0
fi
if [ /etc/exports -nt /var/lock/subsys/nfs ] ; then
echo reload; exit 0
fi
;;
*)
echo ``Usage: nfs {start|stop|status|restart|reload}''
exit 1 Part
esac
exit 0 III
Ch
Table 21.2 provides a listing of all the commands you can pass to the nfs script.
21
468 Chapter 21 Managing NFS
Table 21.2 nfs Script Commands
Command Description
Start Starts all the daemons and creates the lock file.
Stop Stops the daemons and removes the lock file.
Status Indicates the current status of each NFS program, as shown here:
[root@web /root]# /etc/rc.d/init.d/nfs status
rpc.statd (pid 416) is running...
rpc.mountd (pid 438) is running...
nfsd (pid 460 459 458 457 456 455 454 453) is running...
rpc.rquotad (pid 427) is running...
restart Stops and then starts the programs.
Reload Reloads the available NFS file system from /etc/exports.
Exporting an NFS File System
For clients to mount an NFS file system, this file system must be made available by the server.
Before the file system can be made available, you must ensure that it's mounted on the server.
If the file system is always going to an NFS exported file system, you should make sure that
you have it listed in the /etc/fstab file on the server so that it automatically mounts when the
server boots.
When you have the file system mounted locally, you can make it available via NFS. This
process involves two steps. First, you must make sure that the NFS daemons rpc.mountdand
rpc.nfsdare running on your server. These daemons are usually started from the startup
/etc/rc.d/init.d/nfs script. Usually, all you need to do is make sure that the following lines
are in your script:
daemon rpc.mountd
daemon rpc.nfsd
rpc.mountd and rpc.nfsd daemons aren't managed
As RPC-based programs, the
inetd daemon but are started up at boot time, registering themselves with the
by the
portmap daemon. You must be sure to start them only after rpc.portmap is
running.
Second, you must enter the NFS file system in a configuration file named /etc/exports. This
file contains information about what file systems can be exported, what computers are
allowed to access them, and what type and level of access are permitted.
Understanding the /etc/exports File
469
Understanding the /etc/exports File
The /etc/exports file is used by the mountd and nfsd daemons to determine what file systems
are to be exported and what restrictions are placed on them. File systems are listed in
/etc/exports, one per line. The format of each line is the name of the mount point for a local
file system, followed by a list of computers that are allowed to mount this file system. A
comma-separated list of mount options in parentheses may follow each name in the list.
Table 21.3 lists the mount options available in the /etc/exports file.
Table 21.3 Mount Options Available in the /etc/exports File
Option Description
insecure Permits nonauthenticated access from this
machine.
secure Requires secure RPC authentication from this
machine.
root_squash Maps any requests from root, UID 0 on the
client, to the UID NOBODY_UID on the server.
no_root_squash Doesn't map any requests from UID 0 (default
behavior).
ro Mounts the file system as read-only (default
behavior).
rw Mounts the file system as read-write.
link_relative Converts absolute symbolic links (where the
link contents start with a slash) into relative
links by prefixing the link with the necessary
number of ../ characters to get from the
directory containing the link to the root on the
server.
link_absolute Leaves all symbolic links as they are (normal
behavior for Sun NFS servers). This is the
default behavior for Linux.
map_daemon Maps local and remote names and numeric IDs
by using an lname/uid map daemon on the client
where the NFS request originated. This option
is used to map between the client and server
UID spaces.
all-squash Maps all UIDs and GIDs to the anonymous Part
user. This option is useful for NFS-exported
public directories, such as those housing FTP
III
and news. Ch
no-all-squash Acts the opposite of the all-squash option.
21
This is the default option for Linux.
squash-uids Specifies a list of UIDs subject to anonymous
mappings. A valid list of IDs looks like this:
squash uids=0-15,20,25-50.
470 Chapter 21 Managing NFS
Table 21.3 Mount Options Available in the /etc/exports File
Option Description
squash-gids Specifies a list of GIDs subject to anonymous
mappings. A valid list of IDs looks like this:
squash gids=0-15,20,25-50.
anonuid Sets the UID for the anonymous account. This
option is useful for PC/NFS clients.
anongid Sets the GID for the anonymous account. This
option is useful for PC/NFS clients.
noaccess Excludes certain subdirectories from a client.
This option makes everything below the
directory inaccessible to the client.
The following is a sample /etc/exports file:
/home bill.tristar.com(rw) fred.tristar.com(rw)
george.tristar.com(rw)
/usr/local/bin/bin *.tristar.com(ro)
/projects develop.tristar.com(rw) bill.tristar.com(ro)
/pub (ro,insecure,root_squash)
In this example, the server exports four different file systems. /home is mounted with read/
write access on three different computers: bill, fred, and george. This information indicates
that the directory probably holds user home directories because of the directories' names.
The /usr/local/bin file system is exported as read-only with access allowed for every
?
computer in the tristar.com domain.
See ``File Permissions,'' p. 414
Tip #123 from
Jack The command ls -l displays the read/write permissions of files in the current
directory. To change permissions, use the chmod command.
The /projects file system is exported with read/write access for the computer
develop.tristar.com but with read-only access for bill.tristar.com.
The /pub file system doesn't have a list of hosts that are allowed access. This means that any
host is allowed to mount this file system. It has been exported as read-only with
nonauthenticated access allowed, and the server remaps any request from root on a remote
machine that accesses this file system.
471
Mounting NFS File Systems
Mounting NFS File Systems
Mounting an NFS file system is similar to mounting any other type of file system. You can
mount NFS file systems from the /etc/fstab file at boot time or interactively via the mount
command.
You must be sure to separate the host name and file/system/path portions of the
remote file system name with a colon, such as
mailserver:/var/spool/mail
when you use the mount command or when you make an entry in /etc/fstab. If
you don't separate the host name from the directory, your system won't mount the
remote directory correctly.
Mounting NFS File Systems via /etc/fstab
When you specify an NFS file system in the /etc/fstab file, you identify the file system with
the following format:
hostname:/file/system/path
hostname is the name of the server where the file system is located, and /file/system/path is
the file system on the server.
The file-system type is specified as nfs in the mount options field of the file system entry.
Table 21.4 lists the most commonly used mount options.
Table 21.4 Commonly Used Options for NFS Mounts
Option Description
rsize=n Specifies the datagram size in bytes used by the NFS clients on read requests.
The default value is 1,024 bytes.
wsize=n Specifies the datagram size in bytes used by the NFS clients on write requests.
The default value is 1,024 bytes.
timeo=n Sets the time, in tenths of a second, that the NFS client waits for a request to Part
complete. The default value is 0.7 seconds.
III
hard Mounts this file system by using a hard mount. This is the default behavior.
Ch
soft Mounts this file system by using a soft mount.
intr Allows signals to interrupt an NFS call. This option is useful for aborting an
21
operation when an NFS server doesn't respond.
472 Chapter 21 Managing NFS
Hard Mounts Versus Soft Mounts
Hard mounts and soft mounts determine how an NFS client behaves when an NFS server stops responding. NFS
file systems are hard-mounted by default. With either type of mount, if a server stops responding, the client
waits until the timeout value specified by the timeo option expires and then resends the request (this is known
minor timeout). If the requests to the server continue to time out and the total timeout reaches 60 seconds,
as a
amajor timeout occurs.
If a file system is hard-mounted, the client prints a message to the console and starts the mount requests all
over again by using a timeout value that's twice that of the previous cycle. This process has the potential to go
on forever. The client keeps trying to remount the NFS file system from the server until the file mount succeeds.
Soft mounts, on the other hand, just generate an I/O error to the calling process when a major timeout occurs.
Linux then continues on its merry way.
Typically, important software packages and utilities that are mounted via NFS should be mounted with hard
mounts. This is why hard mounts are the default. You don't want your system to start acting strange if the
Ethernet gets unplugged for a moment; you want Linux to wait and continue when the network is back up. On
the other hand, you might want to mount noncritical data, such as remote news spool partitions, as soft mounts
so that if the remote host goes down, it won't hang your current login session.
A typical NFS file system entry in the /etc/fstab file might look like this:
mailserver:/var/spool/mail /var/spool/mail nfs timeo=20,intr
This entry mounts the /var/spool/mail file system located on the host mailserver at the
local mount point /var/spool/mail. It specifies that the file system type is nfs. Also, it sets
the timeout value to 2 seconds (20 tenths of a second) and makes operations on this file
system interruptible.
Mounting NFS File Systems Interactively
NFS file systems can be mounted interactively, just like any other type of file system.
However, you should be aware that the NFS mount command isn't very pretty due to all the
options that you can specify on the command line.
By using the previous example, the interactive mount command that you use to mount the
/var/spool/mail file system becomes the following:
# mount -t nfs -o timeo=20,intr mailserver:/var/spool/mail /var/spool/mail
If you need to specify datagram sizes and timeouts, interactive mount commands can become
very complex. Placing these mount commands in your /etc/fstab file is highly recommended
so that they can be mounted automatically at boot time.
Troubleshooting
How do I determine what directories are currently nfs mounted?
You can use the mount command to display all disk information and look for nfs-mounted
file systems. You can also use the df -t nfs command to see nfs-mounted file systems and
information on disk space consumed by each system. The showmount -a command displays
information in hostname:directory pairs while showmount -d displays just the directories.
473
Troubleshooting
I cannot mount my directory.
There are several items to remember when mounting any file system. First, make sure the
mount point exists (for example, use the ls command to check the directory). Next, make
sure that you and no one else is currently logged into that directory. Make sure you have the
necessary permissions to mount a directory. Finally, make sure the remote host is accessible
across the network.
My system stops responding anytime I try to cd to an nfs-mounted directory.
Make sure there are no network problems. Ping the remote host to make sure the host is
available. If not, begin troubleshooting the network. The traceroute, netstat, and ipconfig
commands are all useful tools to use in checking your network connections. While there are
problems accessing the remote host, you should unmount the file system to prevent the
hangups. Use umount filesystem to unmount the offending system.
I'm having network problems accessing my nfs systems.
First, use the ping command to make sure the remote host is accessibleÐ ping host. If you can
not ping the host, then use the traceroute command to find out where in the network the
connection is down, as in the following example:
[root@ns /root]# /usr/sbin/traceroute www.yahoo.com
traceroute to www.yahoo.com (204.71.200.68), 30 hops max, 40 byte packets
1 gateway (209.42.203.225) 3.688 ms 3.242 ms 3.062 ms
2 max4000-1.rtp.intrex.net (209.42.192.1) 35.924 ms 36.155 ms 34.144 ms
3 209.42.192.30 (209.42.192.30) 34.459 ms 37.508 ms 35.809 ms
4 rtp-rtr1-e1.intrex.net (209.42.255.33) 37.974 ms 35.533 ms 35.959 ms
5 * * *
6 * * *
The *** indicates a break in the network, and the last location, #4, indicates the last host
reachable and where you should concentrate your troubleshooting. If this host is outside your
control, you need the help of your network administrator to proceed.
Part
III
Ch
21
CHAPTER
22
Managing NIS and LDAP
In this chapter by Steve Burnett
What Is NIS? 476
LDAP: What Is It, and Why Is It Better Than NIS? 481
Project: Installing an LDAP Server 482
476 Chapter 22 Managing NIS and LDAP
What Is NIS?
The Network Information Service (NIS) is a tool used to distribute a set of configuration files
common to a set of UNIX machines. NIS was originally named the Yellow Pages (YP) but
had to change because of copyright issues. You can still see the original name in the command
names, most of which begin with yp.
NIS is a distributed database system that aids the system administrator in sharing password
files, group files, host tables, and other files between networked systems. NIS can simplify the
management of a network because all the desired account and configuration information is
stored on a single computer, referred to as the NIS master server. NIS is included with many
flavors of UNIX.
A set of shared NIS database files is called a map; hosts that belong to the same NIS domain
share the same set of maps. NIS slave servers, which receive updated versions of the maps from
the NIS master server, are used to provide information when the NIS master server is down
and to spread the load of requests more evenly if necessary.
NIS Members
A system can fit into an NIS system in one of three ways:
& As a client
& As a slave server
& As a master server
A server hosts an NIS map from which clients get their information. A server can be a slave
server that gets its NIS map from a master server or a master server that is the actual resource
for the distributed files. A master server is the top of the tree for a given NIS domain. A slave
server has a complete copy of all NIS maps created on a master server; when the master
server's maps are updated, the new maps are pushed out to the slave server.
Finally, either a slave or master server can also be an NIS client; this is the common
configuration. Otherwise, you would have to maintain the given server's configuration
separate from the NIS map.
Clients can contact a master server directly or be configured to request from a slave server to
spread out the load. Clients are therefore configured to work on a pull model, rather than a
push model, characterizing the master-to-slave server relationship.
NIS Files, Maps, and Domains
NIS uses three major objects to distribute information. Data is taken from files in the Linux
or other UNIX system and is compiled into maps. These maps are distributed from master
server to clients and slave servers, and from slave servers to other clients, according to the
domain a system is a member of. This section explains in more detail which system
477
What Is NIS?
configuration files (and which information from those files) is commonly compiled into maps, and
Part
how domains are used to define the map distribution.
III
NIS Files and Maps
Ch
These files are identified, determined, and copied from master server to slave server to client
22
by the use of NIS maps. The maps for each domain are located in a separate /var/yp/
domainname directory on the NIS server. For example, the maps for a machine that belong to
the domain marketing are located in the directory /var/yp/marketing on the corresponding
NIS server.
Running ypbuild in the directory /var/yp of a master server calls the makedbm command to
compile the default NIS maps from the input files.
Never make the maps on a slave server; they will be overwritten as soon as the slave
server pulls the map from the master server again.
NIS maps are in a non-ASCII format, unlike the majority of the source files such as /etc/
hosts and /etc/passwd.
Table 22.1 lists the commonly created NIS maps, their source material in the directory
structure, and the contents of the map.
Table 22.1 A Summary of NIS Maps
Map Name NIS Filename Description
bootparams /etc/bootparams Contains pathnames of files that
clients need during booting, such
asroot, swap, and others.
/etc/bootparams is never consulted
after the map is created.
ethers.byaddr /etc/ethers Contains machine names and
Ethernet addresses. The Ethernet
address is the key in the map.
/etc/ethers is never consulted
after the map is created.
ethers.byname /etc/ethers Same as ethers.byaddr, except the
key is the machine name instead
of the Ethernet address. Again,
/etc/ethers is never consulted
after the map is created.
group.byid /etc/group Contains group security informa-
tion with the group ID as the key.
With NIS, the local /etc/group is
consulted first and then the map.
478 Chapter 22 Managing NIS and LDAP
Table 22.1 A Summary of NIS Maps
Map Name NIS Filename Description
group.byname /etc/group Contains group security informa-
tion, with the group name used as
the key. If NIS is active on a system,
the local /etc/group is consulted
first and then the map.
hosts.byaddr /etc/hosts Contains host names and IP ad-
dresses. The IP address is the key
used in the map. The source for
host information is determined by
the entries in the /etc/netconfig
file.
hosts.byname /etc/hosts Same as hosts.byaddr, except the
key used in this table is the host
name instead of the IP address.
mail.aliases /etc/aliases Contains aliases and mail addresses.
The alias is the key in the map. As
with host inquiries, the local file
/etc/aliases is consulted first and
then the NIS map.
mail.byaddr /etc/aliases Same as mail.aliases, except the
key used in the map is the mail
address instead of the alias.
netgroup.byhost /etc/netgroup Contains the group name, user-
name, and host name. The host
name is the key in the map. The
NIS map overrides the base file.
After the NIS map is created,
/etc/netgroup is never consulted.
netgroup.byuser /etc/netgroup Contains the same information as
netgroup.byhost, except the key
used is the username instead of the
host name.
netgroup /etc/netgroup Contains the same information as
netgroup.byhost, except the key
used is the group name instead of
the host name.
netid.byname0 /etc/passwd, /etc/ Contains the machine name and
hosts, /etc/group mail address (including domain
name). If a netid file is available, it
is consulted in addition to the data
available through the other files.
479
What Is NIS?
Table 22.1 A Summary of NIS Maps
Part
Map Name NIS Filename Description
III
netmasks.byaddr /etc/netmasks Contains network masks to be used
Ch
with IP subnetting. The network
mask is the key in the map. The 22
NIS map overrides the base file.
After the NIS map is created,
/etc/netmasks is never consulted.
networks.byaddr /etc/networks Contains names of networks known
to your system and their IP ad-
dresses. The network address is the
key in the map. After the NIS map
is created, /etc/networks is never
consulted.
networks.byname /etc/networks Contains the same information as
networks.byaddr, except the key
used is the network name instead of
the network address.
passwd.byname /etc/passwd Contains password information.
The user name is the key in the
map. The local /etc/passwd is
consulted first and then the map.
passwd.byuid /etc/passwd Contains the same information as
networks.byaddr, except the key
used is the user ID instead of the
username.
protocols.byname /etc/protocols Contains network protocols known
to your system. The protocol name
is the key used in the NIS map.
After the NIS map is created,
/etc/protocols is never consulted.
protocols.bynumber /etc/protocols Contains the same information as
protocols.byname, except the key
used is the protocol number instead
of the protocol name.
publickey.byname /etc/publickey Contains public and secret keys.
Use of the NIS publickey map
permits running secure RPC (Re-
mote Procedure Call) across the
network of machines. After the NIS
map is created, /etc/publickey is
never consulted.
rpc.bynumber /etc/rpc Contains the program number and
the names of RPCs known to your
system. The RPC program number
is the key used in the map. After the
NIS map is created, /etc/rpc is
never consulted.
480 Chapter 22 Managing NIS and LDAP
Table 22.1 A Summary of NIS Maps
Map Name NIS Filename Description
services.byname /etc/services Lists Internet services known to
your network. The service name is
the key in the map. After the NIS
map is created, /etc/services is
never consulted.
ypservers An NIS-created file Lists NIS servers known to your
network.
NIS is configurable to distribute any or all of the files shown in Table 22.1 as NIS maps, as
well as any others you might want.
So now you know about master servers, slave servers, and clients and files chosen on a master
server by a map. What are your teams? How do you identify which clients receive the map of
files from which servers? You do so by using domains.
NIS Domains
In this book, the word domain is used fairly often, and not as interchangeably as it might first
appear because domain can be used in two different areas of UNIX systems management:
& DNS (Domain Name Services)ÐA DNS domain is defined as a logical entity or
organization that represents a part of a network identifiable by DNS. A DNS domain
is often referred to by the registered portion of the DNS name; for example,
redhat.com is a domain.
& NIS (Network Information Services)ÐAn NIS domain is a group of systems that
use the same set of NIS maps and therefore use the same defined configurations. For
example, every system that is a member of an NIS domain likely shares the same
defined user and password lists, issued on request from that system by the NIS master
server.
The term domain can also refer to a group of Microsoft Windows computers, but defining a
Windows domain is well outside the scope of this book. If you're dealing with Microsoft
products or documentation, you might want to keep in mind that the term domain may not
mean what you think it means at first glance.
In summary, NIS is a tool: It saves you time by copying configuration files to other UNIX
systems. This capability is greatÐif you have the setup correct the first time. If your master
files are sloppy, all your systems will have the same sloppy, incorrect, or possibly dangerous
information. Before you configure NIS on the first system as a master server, you should
spend some extra time checking the accuracy of your initial files.
481
LDAP: What Is It, and Why Is It Better Than NIS?
NIS domains are standalone and do not communicate with each other. Each master
Part
server is a separate entity unto itself. If you have several NIS domains, you have to
administer and make changes to each one separately. III
Ch
? See ``Removing a User,'' p. 253
22
NIS Security Concerns
Although NIS simplifies the task of system administration, it also presents numerous security
problems when it is not correctly configured. For example, any user can obtain copies of the
databases exported by an NIS server. As a result, a malicious user might acquire a copy of the
distributed password file, as well as all the other information contained in the NIS database.
Another security concern includes NIS having no provisions for safeguarding against
spoofing of network addresses. For this and other reasons, the acronym NIS is sometimes
referred to as meaning Network Intruder Service. NIS's successor NIS+ addresses some of
these security concerns and is also more scalable, but it is not a perfect replacement.
LDAP: What Is It, and Why Is It Better Than NIS?
NIS is an older directory service and is really useful only if every system is a UNIX-based
machine. However, it's not scalable, and it also has security concerns, as presented in the
preceding section.
Another attempt to define a hierarchical and global directory service was the X.500 protocol.
The X.500 protocol defined a Directory Information Tree (DIT), with each entry in an
X.500 DIT made up of a collection of attributes. In a manner like the mail message header
definitions given in RFC 822, each attribute is defined as a two-part item containing a type
element and a corresponding value attribute element.
The X.500 standard defines several object classes for directories and supports the capacity for
locally defined extensions to the system. The basic object classes include various categories,
including name, alias, country, locality, and organization. An object is defined by its
attributes. Approximately 40 basic attribute types are in use, such as Common Name (CN),
Organization Name (ON), Street Address (SA), and Country . X.500 defined a Directory
Access Protocol (DAP) as the way to access the information in the databases.
However, although the X.500 protocol was powerful, it was also unwieldy and difficult to
configure, administer, and maintain. Enter the LDAP as a less cumbersome and easier way to
work with X.500 directories.
The Lightweight Directory Access Protocol (LDAP) is an open-standard protocol for accessing
information and directory services. The protocol runs over Internet transport protocols, such
as TCP, and can be used to access either standalone directory servers or X.500 directories.
482 Chapter 22 Managing NIS and LDAP
The full definition of LDAP is in the RFC at http://www.umich.edu/~dirsvcs/ldap/doc/rfc/
rfc1777.txt.
You can find some Internet-based LDAP directories, such as Bigfoot.com and 411.com.
Also, if you use Microsoft's Outlook, Microsoft has released an update that allows
the application to use LDAP directories. For more information, look at
http://officeupdate.microsoft.com/downloadDetails/o98ldap.htm .
Project: Installing an LDAP Server
This project presents the basic steps of how to download, install, and configure an LDAP
server on your Linux system, and also how to configure a common mail client (included with
Netscape) to query your LDAP server. This is only an introduction and will not address
issues such as location of the LDAP server on your network, or what the design of the LDAP
database should be.
Downloading and Installing OpenLDAP
To begin this project, download the latest copy of the LDAP software from OpenLDAP's
Web site at http://www.openldap.org/. OpenLDAP is based on the University of
Michigan's LDAP 3.3 release, providing compatibility with applications written to use the
university's software.
As of this writing, the version is 1.2.6, but new versions may be available by the time
you read this chapter. You should look at the INSTALL file for the current release for
any changes.
Before you compile the software, you must first set a few configuration options.
LDAP is similar in architecture to DNS. You start at the very top and then work your way
down to countries, followed by organizations. Your LDAP server starts its hierarchy tree at a
base DN. The base DN is usually a combination of the organization name and the ISO name
of the country; it is defined by prefacing the organization with o= and the country with c=,
,
separating the pair with a comma ( ). Thus, you could have base DNs of o=Boston, c=US or
o=Marketing Group, c=FI. No central organization or registry is responsible for assigning base
DNs, so you can make up a base DN for yourself.
After you have defined the base DN, it's time to compile OpenLDAP. For the more recent
versions of OpenLDAP, you can use a configure script instead of having to directly edit
makefiles. Before you start, read the install notes; OpenLDAP includes some nice optional
features such as cryptographic support.
To start the process, enter the following at the command line:
./configure
483
Project: Installing an LDAP Server
By default, the installation puts the following kinds of files in the following directories:
Part
& the daemons are placed in /usr/local/libexec/
III
& a sample client is placed in /usr/local/bin
&
Ch
programs to convert source files to LDAP format are placed in /usr/local/sbin/
22
& configuration files are placed in /usr/local/etc/ldap
Next, you need to edit the /usr/local/etc/openldap/slapd.conf file to configure the base
DN that you created earlier. The slapd.conf file has better comments than the average
system or service configuration file, so search in the file for the line that begins with suffix,
and edit it appropriately.
Populating the Database
After you've installed LDAP, it needs data in its database with which to respond to queries.
For this example, you can just move the /etc/passwd file.
Go to the Web site http://www.padl.com/tools.html and download the LDAP migration
tools. Configure the tools for your particular setup according to the guidelines in the README
file. Then enter the following command:
./migrate_passwd.pl /etc/passwd ./passwd.ldif
This command creates a passwd.ldif file that should have a set of entries that look like this:
dn: cn=Dave Ross,ou=marketing,dc=afakecompany,dc=com
cn: Dave Ross
objectclass: top
objectclass: person
objectclass: organizationalPerson
objectclass: inetOrgPerson
objectclass: posixAccount
objectclass: account
mail: [email protected]
givenname: David
sn: Ross
uid: dross
userPassword: {crypt}FGgkUelSHwaaI
loginShell: /bin/tcsh
uidNumber: 31
gidNumber: 2031
homeDirectory: /home/dross
Each entry in the file starts with a unique DN. You might not need all the categories, so you
can add and delete categories in your LDAP database as you like.
Before creating the database, you need to add a header to tell slapd (the LDAP daemon) what
the base DN is.
484 Chapter 22 Managing NIS and LDAP
At the head of the file, put in something like the following:
dn: o=A Fake Company, c=US
o: A Fake Company
objectclass: company
To create the database, issue the following command at the command line:
/usr/local/sbin/ldif2ldbm -i passwd.ldif
Now that the database is populated, start the server with the following command and let it
run:
/usr/local/libexec/slapd
For a simple test of the database, issue this command to see all the entries present:
/usr/local/bin/ldapsearch 'objectclass=*'
Configuring a Client
So you have the data in there. Now what? It's time to configure a client to use your new
LDAP database.
Using Netscape 4.6 on Linux, open the Address Book. By default, you should see a local
Address Book and three remote ones (Netcenter, InfoSpace, and Verisign). Right-click the
Netcenter icon and select the New Directory command from the pop-up menu. A dialog box
appears, prompting you to enter the following:
& A directory name (you can call it whatever you like)
& The host name of the LDAP server
& The base DN
Leave the check boxes unchecked for now and close the dialog box. That's all.
CHAPTER
23
Using Samba
In this chapter by Jack Tackett, Jr.
What Is Samba? 486
Installing Samba 487
Configuring Samba on Linux 487
Running the Samba Server 494
Using smbclient 495
Case Study: OpenLinux and swat 497
486 Chapter 23 Using Samba
What Is Samba?
This chapter gives you the information you need to install, configure, and use the Session
Message Block (SMB or Samba) protocol services under Linux. With Samba, you can do the
following:
& Share a Linux file system with Windows 95, 98, or NT
& Share a Windows 95, 98, or NT file system with Linux
& Share a printer connected to a Linux system with Windows 95, 98, or NT systems
& Share a Windows 95, 98, or NT printer with Linux
Samba is the client/server protocol used by Microsoft's operating systems to share files and
printer services. Microsoft and Intel developed the SMB protocol system in 1987, and later
Andrew Tridgell ported the system to various UNIX systems and then Linux.
For a quick introduction to Samba see
http://us1.samba.org/samba/docs/SambaIntro.html
The Samba suite is made up of several components. All the components are controlled via
one configuration file called smb.conf. The smbd daemon provides the file and print services
to SMB clients, such as Windows for Workgroups, Windows NT, or LanManager. The
configuration file for this daemon is described in smb.conf. The nmbd daemon provides
NetBIOS nameserving and browsing support. It can also be run interactively to query other
?
name service daemons.
See ``Understanding Multitasking,'' p. 366
The smbclient program implements a simple FTP-like client. It is useful for accessing SMB
shares on other compatible servers, such as Windows machines, and it can also be used to
allow a UNIX box to print to a printer attached to any SMB server, such as a PC running
Windows 98.
The testparm utility allows you to test your smb.conf configuration file. The smbstatus utility
allows you to tell who is currently using the smbd server.
As stated, all these programs are controlled from one configuration file, called smb.conf,
located in /etc. The file is composed of various named sections specified in brackets [ ].
Within each section, the parameters are specified by key = value pairs, for example,
workgroup = MYGROUP.
Samba is a large topic that can fill a book in its own right. Fortunately, in addition to this
chapter, there is a large amount of documentation included with the program, including the
487
Configuring Samba on Linux
man pages and lots of .txt files with hints and useful information. There are several Samba
mailing lists with discussions on a variety of topicsÐsee http://lists.samba.org/ for full
information. The Usenet newsgroup comp.protocols.smb also contains lots of useful
information and is frequented by lots of Samba users, even though the group is not strictly
aimed at Samba but instead any system using the SMB protocol. The newsgroup was initially
setup by people on the Samba mailing list. Finally, the main Web site for Samba is
http://samba.org/samba/.
Part
III
Installing Samba
Ch
You can install Samba during regular installation or later using RPM. If you need to
23
install the package, first download the current version from Red Hat's Web site ( http://
www.redhat.com). You can then install the package (the current version from Red Hat is
samba-2.0.5a-19990721.i386.rpm and the current version from the Samba organization can
be found at http://us1.samba.org/samba/ftp/Binary_packages/redhat/RPMS/6.0/ ) by
using the following command:
?
rpm -Uvh samba-1.9.18p5-1.i386.rpm
See ``Installing Packages with RPM,'' p. 169
The package should contain all the files you need to run Samba, including the two primary
programs smbd and nmbd. However, you might have to recompile the various programs if
you are using a different distribution.
Configuring Samba on Linux
The main configuration file, called smb.conf, is located in the /etc directory. Listing 23.1
provides the default listing shipped with Red Hat 5.1.
A semicolon character ( ;) at the beginning of a line indicates that the line is a
comment and is to be ignored when processed by the Samba server.
Listing 23.1 The Sample smb.conf Samba Configuration File
; The global setting for a RedHat default install
; smbd re-reads this file regularly, but if in doubt stop and restart it:
; /etc/rc.d/init.d/smb stop
; /etc/rc.d/init.d/smb start
;======================= Global Settings =====================================
[global]
; workgroup = NT-Domain-Name or Workgroup-Name, eg: REDHAT4
workgroup = WORKGROUP
488 Chapter 23 Using Samba
Listing 23.1 continued
; comment is the equivalent of the NT Description field
comment = RedHat Samba Server
; volume = used to emulate a CDRom label (can be set on a per share basis)
volume = RedHat4
; printing = BSD or SYSV or AIX, etc.
printing = bsd
printcap name = /etc/printcap
load printers = yes
; Uncomment this if you want a guest account
; guest account = pcguest
log file = /var/log/samba-log.%m
; Put a capping on the size of the log files (in Kb)
max log size = 50
; Options for handling file name case sensitivity and / or preservation
; Case Sensitivity breaks many WfW and Win95 apps
; case sensitive = yes
short preserve case = yes
preserve case = yes
; Security and file integrity related options
lock directory = /var/lock/samba
locking = yes
strict locking = yes
; fake oplocks = yes
share modes = yes
; Security modes: USER uses Unix username/passwd, SHARE uses WfW type
Apasswords
; SERVER uses a Windows NT Server to provide authentication services
security = user
; Use password server option only with security = server
; password server = <NT-Server-Name>
; Configuration Options*****Watch location in smb.conf for side-effects*****
; Where %m is any SMBName (machine name, or computer name) for which a custom
; configuration is desired
; include = /etc/smb.conf.%m
; Performance Related Options
; Before setting socket options read the smb.conf man page!!
socket options = TCP_NODELAY
; Socket Address is used to specify which socket Samba
; will listen on (good for aliased systems)
; socket address = aaa.bbb.ccc.ddd
; Use keep alive only if really needed!!!!
; keep alive = 60
; Domain Control Options
; OS Level gives Samba the power to rule the roost. Windows NT = 32
; Any value < 32 means NT wins as Master Browser, > 32 Samba gets it
; os level = 33
; specifies Samba to be the Domain Master Browser
; domain master = yes
489
Configuring Samba on Linux
Listing 23.1 continued
; Use with care only if you have an NT server on your network that has been
; configured at install time to be a primary domain controller.
; domain controller = <NT-Domain-Controller-SMBName>
; Domain logon control can be a good thing! See [netlogon] share section
Abelow!
; domain logons = yes
; run a specific logon batch file per workstation (machine)
; logon script = %m.bat Part
; run a specific logon batch file per username
; logon script = %u.bat III
; Windows Internet Name Serving Support Section
; WINS Support - Tells the NMBD component of Samba to enable its WINS Server Ch
; the default is NO. 23
; wins support = yes
; WINS Server - Tells the NMBD components of Samba to be a WINS Client
; Note: Samba can be either a WINS Server, or a WINS Client, but NOT
Aboth
; wins server = w.x.y.z
; WINS Proxy - Tells Samba to answer name resolution queries on behalf of a
Anon
; WINS Client capable client, for this to work there must be at least
Aone
; WINS Server on the network. The default is NO.
; wins proxy = yes
;============================ Share Declarations ==============================
[homes]
comment = Home Directories
browseable = no
read only = no
preserve case = yes
short preserve case = yes
create mode = 0750
; Un-comment the following and create the netlogon directory for Domain Logons
; [netlogon]
; comment = Samba Network Logon Service
; path = /home/netlogon
; Case sensitivity breaks logon script processing!!!
; case sensitive = no
; guest ok = yes
; locking = no
; read only = yes
; browseable = yes ; say NO if you want to hide the NETLOGON share
; admin users = @wheel
; NOTE: There is NO need to specifically define each individual printer
[printers]
comment = All Printers
path = /var/spool/samba
browseable = no
printable = yes
; Set public = yes to allow user 'guest account' to print
public = no
writable = no
490 Chapter 23 Using Samba
Listing 23.1 continued
create mode = 0700
;[tmp]
; comment = Temporary file space
; path = /tmp
; read only = no
; public = yes
; A publicly accessible directory, but read only, except for people in
; the staff group
;[public]
; comment = Public Stuff
; path = /home/samba
; public = yes
; writable = yes
; printable = no
; write list = @users
; Other examples.
;
; A private printer, usable only by fred. Spool data will be placed in fred's
; home directory. Note that fred must have write access to the spool
Adirectory
; wherever it is.
;[fredsprn]
; comment = Fred's Printer
; valid users = fred
; path = /homes/fred
; printer = freds_printer
; public = no
; writable = no
; printable = yes
;
; A private directory, usable only by fred. Note that fred requires write
; access to the directory.
;[fredsdir]
; comment = Fred's Service
; path = /usr/somewhere/private
; valid users = fred
; public = no
; writable = yes
; printable = no
;
; a service which has a different directory for each machine that connects
; this allows you to tailor configurations to incoming machines. You could
; also use the %u option to tailor it by user name.
; The %m gets replaced with the machine name that is connecting.
;[pchome]
; comment = PC Directories
; path = /usr/pc/%m
; public = no
; writeable = yes
;
;
491
Configuring Samba on Linux
Listing 23.1 continued
; A publicly accessible directory, read/write to all users. Note that all
files
; created in the directory by users will be owned by the default user, so
; any user with access can delete any other user's files. Obviously this
; directory must be writable by the default user. Another user could of
course
; be specified, in which case all files would be owned by that user instead.
;[public] Part
; path = /usr/somewhere/else/public
; public = yes III
; only guest = yes
; writable = yes Ch
; printable = no 23
;
;
; The following two entries demonstrate how to share a directory so that two
; users can place files there that will be owned by the specific users. In
this
; setup, the directory should be writable by both users and should have the
; sticky bit set on it to prevent abuse. Obviously this could be extended to
; as many users as required.
;[myshare]
; comment = Mary's and Fred's stuff
; path = /usr/somewhere/shared
; valid users = mary fred
; public = no
; writable = yes
; printable = no
; create mask = 0765
The smb.conf file layout consists of a series of named sections. Each section starts with its
name in brackets, such as [global]. Within each section, the parameters are specified by
key = value pairs, such as comment = RedHat Samba Server.
smb.conf contains three special sections and one or more custom sections. The special
sections are [global], [homes], and [printers].
The [global] Section
The [global] section of smb.conf controls parameters for the entire SMB server. The section
also provides default values for the other sections.
The first line from the [global] section in Listing 23.1 defines the workgroup that this
machine will belong to on your network:
[global]
; workgroup = NT-Domain-Name or Workgroup-Name, eg: REDHAT4
workgroup = WORKGROUP
Next, the file specifies a comment for the system and identifies a volume label as follows:
; comment is the equivalent of the NT Description field
comment = RedHat Samba Server
492 Chapter 23 Using Samba
; volume = used to emulate a CDRom label (can be set on a per share basis)
volume = RedHat4
The next entry tells the Samba server what type of printing system is available on your server,
and the line after that indicates where the printer configuration file is located:
; printing = BSD or SYSV or AIX, etc.
?
printing = bsd
See ``Understanding the /etc/printcap File,'' p. 397
The next line instructs Samba to make available on the network all the printers defined in the
printcap file:
printcap name = /etc/printcap
load printers = yes
The next entry indicates a username for a guest account on your sever. This account is used
to authenticate users for Samba services available to guest connections:
; Uncomment this if you want a guest account
; guest account = pcguest
The log file entry specifies the location of the log file for each client accessing Samba services.
The %m parameter tells the Samba server to create a separate log file for each client. The max
log size entry sets a maximum file size for the logs created:
log file = /var/log/samba-log.%m
; Put a capping on the size of the log files (in Kb)
max log size = 50
The [homes] Section
The [homes] section of smb.conf allows network clients to connect to a user's home directory
on your server without having an explicit entry in the smb.conf file. When a service request is
made, the Samba server searches the smb.conf file for the specific section corresponding to
the service request. If it does not find the service, Samba checks to see whether a [homes]
section exists. If the [homes] section does exist, Samba searches the password file to find the
home directory for the user making the request. When it's found, this directory is shared with
the network.
The comment entry is displayed to the clients to let them know which shares are available. The
browseable entry instructs Samba how to display this share in a network browse list. The
read-only parameter controls whether a user can create and change files in his or her home
directory when it is shared across the network. The preserve case and short preserve case
parameters instruct the server to preserve the case of any information written to the server.
This step is important because Windows filenames are not typically case sensitive, but Linux
filenames are case sensitive. The final entry sets the file permissions for any files created on
the shared directory:
493
Configuring Samba on Linux
[homes]
comment = Home Directories
browseable = no
read only = no
preserve case = yes
short preserve case = yes
?
create mode = 0750
See ``File Permissions,'' p. 414
Part
The [printers] Section III
The [printers] section of smb.conf defines how printing services are controlled if no Ch
smb.conf file. Thus, with this section, like the [homes]
specific entries are found in the
23
section, if no specific entry is found for a printing service, Samba uses the [printers] section
(if it's present) to allow a user to connect to any printer defined in /etc/printcap.
The comment, browseable, and create mode entries mean the same as discussed in ``The
[homes] Section.'' The path entry indicates the location of the spool file to be used when
servicing a print request via SMB:
[printers]
comment = All Printers
path = /var/spool/samba
browseable = no
printable = yes
; Set public = yes to allow user 'guest account' to print
public = no
writable = no
?
create mode = 0700
See ``Selecting a Printer to Work with Linux,'' p. 392
The printable value, if set to yes, indicates that this printer resource can be used to print.
The public entry controls whether the guest account can print.
Sharing Directories
After configuring your defaults for the Samba server, you can create specific shared
directories limited to just certain groups of people or to everyone. For example, suppose you
want to make a directory available to only one user. To do so, you create a new section and fill
in the needed information. Typically, you need to specify the user, the directory path, and the
configuration information to the SMB server, as shown here:
[jacksdir]
comment = Jack's remote source code directory
path = /usr/local/src
valid users = tackett
browsable = yes
public = no
writable = yes
create mode = 0700
494 Chapter 23 Using Samba
jacksdir. The path to the directory on
This sample section creates a shared directory called
the local server is /usr/local/src. browsable entry is set to yes, jacksdir will
Because the
show up in the network browse list. However, because the public entry is set to no, only the
user named tackett can access this directory using Samba. You can grant access to other
users by listing them in the valid users entry.
Testing the smb.conf File
After creating the configuration file, you should test it for correctness by using the testparm
program. testparm is a very simple test program to check the /etc/smb.conf configuration
file for internal correctness. If this program reports no problems, you can use the
configuration file with confidence that smbd will successfully load the configuration file.
Using testparm is not a guarantee that the services specified in the configuration file
will be available or will operate as expected.
testparm has the following command line:
testparm [configfile [hostnamehostip]]
configfile indicates the location of the smb.conf file if it is not in the default location
(/etc/smb.conf). The hostnamehostIP optional parameter instructs testparm to see whether
the host has access to the services provided in the smb.conf file.
The following example shows sample output from running testparm. If the smb.conf file
contains any errors, the program reports them along with a specific error message:
# testparm
Load smb config files from /etc/smb.conf
Processing section ``[homes]''
Processing section ``[printers]''
Loaded services file OK.
Press enter to see a dump of your service definitions
When you press Enter, testparm begins evaluating each section defined in the configuration
file.
Running the Samba Server
The Samba server consists of two daemons, smbd and nmbd. The smbd daemon provides the
file and print sharing services. The nmbd daemon provides NetBIOS name server support.
You can run the Samba server either from the init scripts as detailed in Chapter 10, ``Booting
?
and Shutting Down,'' or from inetd as a system service.
See ``Understanding the Boot Process,'' p. 234
495
Using smbclient
Because Red Hat and Caldera both start SMB services from the init scripts instead of as a
service from inetd, you can use the following command to start or stop the SMB server:
/etc/rc.d/init.d/samba start|stop
Using smbclient
The smbclient program allows Linux users to access SMB shares on other, typically Part
Windows, machines. If you want to access files on other Linux boxes, you can use a variety of
III
?
methods including FTP, NFS, and the r- commands (such as rcp).
Ch
See ``Using the r- Commands,'' p. 695
23
smbclient provides an FTP-like interface that allows you to transfer files with a network share
on another computer running an SMB server. Unfortunately, unlike NFS, smbclient does not
allow you to mount another share as a local directory.
smbclient provides command-line options to query a server for the shared directories
available or to exchange files. For more information on the available command-line options,
consult the man page for smbclient. Use the following command to list all available shares on
the machine win.netwharf.com:
smbclient -L -I win.netwharf.com
The -L parameter requests the list. The -I parameter instructs smbclient to treat the
following machine name as a DNS-specified entry instead of a NetBIOS entry.
To transfer a file, you must first connect to the Samba server by using the following
command:
smbclient '\\WORKGROUP\PUBLIC' -I win.netwharf.com -U tackett
The parameter '\\WORKGROUP\PUBLIC' specifies the remote service on the other machine. It is
typically either a file system directory or a printer. The -U option allows you to specify the
username with which you want to connect. Samba prompts you for a password (if this account
requires one) and then places you at the prompt
smb: \
where \ indicates the current working directory.
From this command line, you can issue any of the various commands shown in Table 23.1 to
transfer and work with files.
Table 23.1 smbclient Commands
Command Parameters Description
? or help [command] Provides a help message on command, or
in general if no command is specified.
496 Chapter 23 Using Samba
Table 23.1 smbclient Commands
Command Parameters Description
! [shell command] Executes the specified shell command
or drops the user to a shell prompt.
cd [directory] Changes to the specified directory on
the server machine (not the local
machine). If no directory is specified,
smbclient reports the current working
directory.
lcd [directory] Changes to the specified directory on
the local machine. If no directory is
specified, smbclient reports the current
working directory on the local ma-
chine.
Del [files] The specified files on the server are
deleted if the user has permission to do
so. Files can include wildcard charac-
ters.
dir or ls [files] Lists the indicated files. You can also
use the command ls to get a list of
files.
exit or quit none Exits from the smbclient program.
get [remote file] or Retrieves the specified remote file and
[local name] saves the file on the local server. If
local name is specified, the copied file is
saved with this filename instead of with
the filename on the remote server.
mget [files] Copies all the indicated filesÐ
including those matching any
wildcardsÐto the local machine.
md or mkdir [directory] Creates the specified directory on the
remote machine.
rd or rmdir [directory] Removes the specified directory from
the remote machine.
put [file] Copies the specified file from the local
machine to the server.
mput [files] Copies all the specified files from the
local machine to the server.
print [file] Prints the specified file on the remote
machine.
queue none Displays all the print jobs queued on
the remote server.
497
Case Study: OpenLinux and swat
Case Study: OpenLinux and swat
Configuring Samba can be a complicated task, but fortunately the Samba 2.0 released with
OpenLinux provides a Web administration tool called swat (Samba Web Administration
Tool), as shown in Figure 23.1. You can run swat from either inetd as a service or from your
Apache Web browser to help create and update your smb.conf file. Use swat also to start and
stop the various Samba daemons and check the status of the system from the Web.
Part
Figure 23.1 III
Caldera pro-
vides a Web- Ch
based interface
for configuring
23
Samba.
Table 23.2 describes the purposes for the various buttons.
Table 23.2 The swat Command Buttons
Button Description
Home Returns you to the default home page for swat.
Globals Allows you to modify the components of the [global] section in smb.conf.
Shares Allows you to modify the components of the [shares] section in smb.conf.
Printers Allows you to modify the components of the [printers] section of smb.conf.
Status Displays the current status of the smb system.
View Displays what will be written to smb.conf by swat.
Password Allows you to change the password used to access swat.
Any changes made using swat requires you to restart smbd and nmbd in order for the changes
to take effect.
498 Chapter 23 Using Samba
Tip #124 from
Jack swat does not create a backup file, so you should copy smb.conf to a backup file
before using swat. Use the following command to create a backup:
cp smb.conf smb.conf.bak
Running swat from inetd
You need to configure the /etc/services and /etc/inetd.conf files to run swat as a service.
Use a text editor and, logged in as root, add the following line to /etc/services:
Swat 901/tcp
This line assigns port 901 to the swat program as a TCP port. Next, add the following line to
/etc/inetd.conf:
swat stream tcp nowait.400 root /usr/local/samba/bin/swat swat
Make sure /usr/local/samba/bin/swat is the correct path to the installed program; if not,
enter the correct path. This entry in inetd.conf instructs inetd to execute swat as a TCP
stream program listening on port 901 as indicated in /etc/services. After making the
necessary changes, you need to restart inetd with the command:
killall -HUP inetd
Tip #125 from
Jack If you want to provide a more secure connection, have the tcp wrapper's daemon
/etc/inetd.conf:
spawn swat by using the following line in
swat stream tcp nowait.400 root /usr/sbin/tcpd /usr/sbin/swat
Make sure to use the correct path to swat, which you can find using the command
locate swat.
You can now run swat by pointing your Web browser to http://localhost:901. swat
prompts for a username and password and while you can use any valid username and
password to access swat, you can only make changes to the system when logged in as root.
Running swat via the Web
Apache runs swat as a cgi program and so requires a bit more detail to configure. First,
you need to create a directory called swat as a subdirectory under your Apache document
root. The document root is identified by the line <Directory /home/httpd/html> in the
/etc/httpd/conf/access.conf file. Use the following commands as root:
mkdir /home/httpd/html/swat
cp -R samba/swat /home/httpd/html/swat
Make sure to use the correct samba/swat directory. Make sure the following subdirectories
under /home/httpd/html/swat are created:
499
Case Study: OpenLinux and swat
help
images
include
Next, copy the swat executable file to the cgi-bin directory, typically /etc/httpd/cgi-bin.
Because anyone can typically reach a Web site from the Internet, you need to be concerned
with security. After all, you don't want anyone to be able to reconfigure your Samba
Part
installation. To insure proper security, you need to password protect your Web site. To
password protect a Web site, you need to create an .htaccess file and a set a password with
III
the htpasswd program. Ch
First, create the .htaccess file in the /home/httpd/html/swat directory with the following 23
command:
vi /home/httpd/home/swat/.htpasswd
and then add the following lines:
AuthName ``swat restricted''
AuthType Basic
AuthUserFile /etc/swat.users
require valid-user
Next, you need to create the password file specified by the line authUserFile in .htaccess:
htpasswd -c /etc/swat.user root
The program htpasswd prompts for a password to assign to the user root.
Next, change the line beginning with AllowOveride in /etc/httpd/conf/access.conf from
None to AuthConfig. The line should read:
AllowOveride AuthConfig
Finally, you need to restart the Web server with either of the following:
apachectl restart
or
/etc/rc.d/init.d/httpd stop
/etc/rc.d/init.d/httpd start
Then you can access swat by using the following URL:
http://localhost/swat/cgi-bin/swat
Finally, if you want to run swat on a Red Hat system, see FAQ entry http://us1.samba.org/
samba/samba/docs/FAQ/#21 on the samba.org Web site.
PART
IV
Using X Windows
24 Installing the X Window System 503
25 Using the X Window System 529
26 Working with KDE 557
27 Working with GNOME 577
CHAPTER
24
Installing the X Window System
In this chapter by Rob Napier
What Is the X Window System? 504
Understanding the X Window System 504
Installing the XFree86 System 508
Configuring XFree86 514
Troubleshooting 524
Case Study: The X Window System Across a Network 525
504 Chapter 24 Installing the X Window System
What Is the X Window System?
Linux provides the capability to run windowed applications across a heterogeneous network
by incorporating the XFree86 implementation of the X11 standard of the X Window System
created at the Massachusetts Institute of Technology (MIT). This system is much more than
a graphical interface used to run applications; it's a powerful client/server system that allows
applications to be run and shared across a network. Although XFree86 is meant to run in a
networked environment, it runs fine on a single machine. You don't need a network to run
XFree86 or X applications.
Historically, XFree86 has been one of the most complex parts of Linux to install and
configure. This is no longer the case for many standard hardware configurations because the
most popular Linux distributions now install and configure it automatically.
Like most parts of Linux, XFree86 also has a How-To document. The ``XFree86 How-To''
is maintained by Eric S. Raymond [email protected] and can be found on the World Wide
Web at http://metalab.unc.edu/LDP/.
Typically, you don't have to worry about software damaging your hardware. However,
because XFree86 deals directly with your video card and monitor, it can cause
physical damage, especially if you try to use XFree86 with an unsupported video card.
Make sure that you have the necessary hardware before trying to run XFree86.
Reading the documentation that comes with the XFree86 system is strongly suggested.
Always refer to the documentation that comes with your installed version of XFree86
because older or newer documentation may not be accurate for your installation. For
Red Hat 6.0, the documentation can be found at /usr/X11R6/lib/X11/doc,
particularly theREADME file, and the ``XFree86 How-To'' by Eric S. Raymond is at
/usr/doc/HOWTO/Xfree86-HOWTO. The most up-to-date version of XFree86 is
available at http://www.xfree86.org, so you should look there before trying to
work with an unlisted video board or chipset.
Understanding the X Window System
The X Window System is a powerful graphical operating environment that supports many
applications across a network, and it can be freely distributed. The version of the X Window
System discussed in this chapter is X11R6. However, Linux and XFree86 are constantly
evolving, and a newer version of X may be available on the Net. In fact, at the time of this
?
printing, work is underway to incorporate X11R6.4 into XFree86-4.0.
See ``Using FTP for Remote File Transfer,'' page 686
505
Understanding the X Window System
XFree86, the X server commonly used by Linux, is the X11R6 standard ported to Intel-based
systems. XFree86 supports a wide range of standard PC hardware. The version of XFree86
discussed in this chapter is 3.3.3.1. This version ships with Red Hat 6.0 and Caldera
OpenLinux 2.2.
The X Window System originally grew out of a cooperative effort between two sections at
MIT: the section responsible for a networking program called Project Athena and a section
called the Laboratory for Computer Science. Both used many UNIX workstations and soon
realized they were each reinventing the wheel when it came to programming graphical user
interfaces (GUIs) for UNIX workstations. To cut down on the amount of code both groups
were writing, they decided to create one robust, extensible windowing systemÐthe X
Window System.
In 1987, several vendorsÐin hopes of creating a single windowing system for UNIX
Part
workstationsÐformed an organization called the X Consortiumto promote and standardize the
X Window System. Thanks to this effort, open computing became a reality. The X IV
Consortium is composed of entities such as IBM, Digital Equipment, and MIT. This group
Ch
of large organizations oversees the construction and release of new versions of X11.
24
XFree86 is a trademark of the XFree86 Project, Inc. The original programmers who ported
the X Window System to the 80386 platform decided to found the project so that they could
gain membership in the X Consortium. By becoming a member of the X Consortium, the
XFree86 Project gained access to works-in-progress and could thus port the new features to
XFree86 while the features were being implemented for the X Window System, rather than
wait until after the official release to make the port. As of January 1, 1997, the X Consortium
turned the X Window System over to the Open Group.
The X Window System is actually a series of pieces working together to present users with a
GUI:
& The base window system is a program providing services to the X Window System.
& The next piece is a protocol for communicating across the networkÐthe X Network
Protocol.
& On top of the program implementing the X Network Protocol is the low-level
interface, named Xlib, between the network/base system and higher-level programs.
Application programs typically use functions in Xlib instead of the lower-level
functions.
& Tying these pieces together is a window manager. The window manager is an X
application whose purpose is to control how windows are presented to users.
& Many modern distributions of Linux, particularly Red Hat 6.0 and OpenLinux 2.2,
also include a desktop. Although the desktop is not technically part of the X Window
System, it does provide an integral part to the users' graphical environment and works
very closely with the window manager. Desktops generally handle features such as
customizable window styles (themes), session management, and interprogram
communications. They often provide a simpler environment for application
506 Chapter 24 Installing the X Window System
developers. The two most popular Linux desktops are GNOME and KDE. They will be
discussed in detail in Chapter 26, ``Working with KDE,'' and Chapter 27, ``Working with
GNOME.''
Unlike most other window systems, the base window system of the X Window System
doesn't provide user interface objects such as scrollbars, command buttons, or menus. The
user interface items are left to the higher-layer components and the window manager.
Therefore, users have much greater flexibility in how things appear and what interface they
prefer.
X applications include not only window managers, but also games, graphics utilities,
programming tools, and many other tidbits. Just about any application that you need has
either been written for or ported to the X Window System. The setup and use of several of
the standard X applications are covered in more detail in Chapter 25, ``Using the X Window
System.''
The X Window System implements a window manager to handle the task of creating and
controlling the interface that makes up the visual portion of the X Window System. It isn't to
be confused with the OS/2 Presentation Manager or the Microsoft Windows Program
Manager. Both of them are closer to the idea of a desktop rather than a window manager.
For the not-so-faint-of-heart, XFree86 also includes programming libraries and files for
programmers who want to develop their own applications under XFree86 or compile X
applications available on the Internet. Although the topic of creating X applications is beyond
the scope of this book, ample documentation is available on any number of Internet
distribution sites such as www.xfree86.org and on many CD-ROM distributions to help you
gain the foothold necessary to create applications for XFree86.
What Is a Client/Server System?
The X Window System is a client/server system controlled by two individual pieces of
software. One piece runs on the client, and the other runs on the server. The client and server
pieces of this puzzle can be on different systems or, as is the case with most personal
computers, both pieces can reside on the same machine.
Client/server is one of the major buzzwords used in the computer industry today. Like most
basic concepts in the industry, client/server has been overplayed and overused to the point of
confusing the average computer user. In the traditional sense, a server is a machine that just
provides resourcesÐdisk drive space, printers, modems, and so onÐto other computers over
a network. A client is the consumer of these services; in other words, a client uses the disk
space, printer, or modems provided by a server.
Now that you understand what a client is and what a server is, it's time to reverse everything
you know. In the X Window System, the client/server relationship is the opposite of what
you've come to know in the PC world. The accepted or common notion of a server is that it
provides services to a client using them. In the most basic form, a client displays the
application that's running on the server.
507
Understanding the X Window System
Under the X Window System, the server displays the application that's running on the client.
This concept might seem a bit confusing at first, but it will make sense when you become
more familiar with the X Window System.
In the X Window System, a client provides the programs and resources necessary to run an
applicationÐwhat in the traditional sense would be called a server. The resources reside on
the client system (remember that the client and server systems can be on the same machine),
whereas the application is displayed and interacted with on the server system.
For example, if you run xcalc on your Linux system using XFree86, then XFree86 (specifically
a program such as /usr/X11R6/bin/XF86_SVGA) is the X server, and xcalc is the X client. The X
server must be running on your local machine (so that it can display your information), but
the X client might be on any machine that you can connect to.
The capability of an X application, which is the client, to run under a server located on either Part
the same computer or on another computer is called network transparency. Thus, it doesn't
IV
matter whether an X application runs on a local or remote machine. This capability can be
Ch
used to run time-consuming tasks on another machine, leaving the local machine
unencumbered to perform other tasks. It can also be used to run applications on other 24
operating systems. Using a non-UNIX X server such as WQR's Reflection X, you can run
your favorite X clients on your Linux workstation but display them on a connected machine
running Microsoft's NT. The reverse is not true, however, because NT does not know how
to run X clients and cannot easily display its own clients on remote servers.
Output Capabilities
The base window system provides the X Window System with plenty of bitmapped graphical
operations. The X Window System and X applications use these operations to present
information graphically to the users. XFree86 offers overlapping windows, immediate
graphics drawings, high-resolution bitmapped graphics and images, and high-quality text.
Whereas early implementations of the X Window System were mostly monochrome-based,
modern implementations like XFree86 support a wide range of color systems.
The X Window System also supports the multiprocessing capabilities of UNIX; thus,
XFree86 supports the multiprocessing capabilities of Linux. Each window displayed under
the X Window System can be a separate task running under Linux.
User Interface Capabilities
The X Consortium did not define standards for user interfaces. At the time very little research
had been done on user interface technology, so no clear interface was considered the best. In
fact, even today, unilaterally declaring one interface the best can alienate many people. The
preferred look and feel presented by the user interface is a very personal decision.
The X Consortium wanted to make the X Window System a standard across UNIX
workstations, which is one reason it is available freely on the Internet. Making the X Window
System freely available fosters interoperability, which is the cornerstone of open systems. Had
508 Chapter 24 Installing the X Window System
the X Consortium dictated a user interface, the X Window System may not have gained its current
level of popularity.
Input Capabilities
Systems running the X Window System typically have some form of pointing device, usually
a mouse. XFree86 requires a mouse or a device, such as a trackball, that emulates a mouse. If
you don't have such a device, you can't use the XFree86 system with Linux. The X Window
System converts signals from the pointing device and from the keyboard into events. The X
Window System then responds to these events, performing appropriate actions.
If your mouse or other hardware pointing device isn't among those supported by
Linux, you'll have problems using XFree86.
Installing the XFree86 System
You probably installed the XFree86 system while installing the entire Linux package from the
accompanying CD-ROM. The X Window System is contained in the XFree86-* RPMs. If
you didn't install the X Window System System at that time, you can use RPM to install X
Window System. First, however, you must verify that you have the appropriate hardware for
?
XFree86.
See ``Installing Packages with RPM,'' p. 169
Ensuring Hardware Support for XFree86
You must make sure that you have the proper hardware to run X Window System, the proper
amount of memory, and the necessary disk space.
You need about 50MB of disk space to install the XFree86 system and the X applications
provided. You need at least 16MB of virtual memory to run XFree86. Virtual memory is the
combination of the physical RAM on your system and the amount of swap space you've
allocated for Linux. You must have at least 4MB of physical RAM to run XFree86 under
Linux, thus requiring a 12MB swap file. The more physical RAM you have, the better the
?
performance of your XFree86 system will be.
See ``Creating the Swap Partition,'' p. 77 for Red Hat and p. 93 for OpenLinux
Next, you need a video card containing a video-driver chipset supported by XFree86.
According to the January 2, 1999, release of Eric S. Raymond's ``XFree86 How-To,'' the
video cards with the chipsets listed in Table 24.1 are supported by XFree86.
509
Installing the XFree86 System
Table 24.1 Chipsets Supported by XFree86
Manufacturer Chipset(s)
Ark Logic ARK1000PV, ARK1000VL, ARK2000PV,
ARK2000MT
Alliance AP6422, AT24
ATI 18800, 18800-1, 28800-2, 28800-4, 28800-5, 28800-6,
68800-3, 68800-6, 68800AX, 68800LX, 88800GX-C,
88800GX-D, 88800GX-E, 88800GX-F, 88800CX,
264CT, 264ET, 264VT, 264GT, 264VT-B, 264VT3,
264GT-B, 264GT3 (including the Mach8, Mach32,
Mach64, 3D Rage, 3D Rage II, and 3D Rage Pro)
Avance Logic ALG2101, ALG2228, ALG2301, ALG2302, ALG2308,
ALG2401 Part
Chips & Technologies 65520, 65525, 65530, 65535, 65540, 65545, 65546, IV
65548, 65550, 65554, 65555, 68554, 69000, 64200,
Ch
64300
Cirrus Logic CLGD5420, CLGD5422, CLGD5424, CLGD5426, 24
CLGD5428, CLGD5429, CLGD5430, CLGD5434,
CLGD5436, CLGD5440, CLGD5446, CLGD5462,
CLGD5464, CLGD5465, CLGD5480, CLGD6205,
CLGD6215, CLGD6225, CLGD6235, CLGD6410,
CLGD6412, CLGD6420, CLGD6440, CLGD7541,
CLGD7543, CLGD7548(*), CLGD7555
Cyrix MediaGX, MediaGXm
Compaq AVGA
DEC TGA
Epson SPC8110
Genoa GVGA
IBM 8514/A (and true clones), XGA-2
IIT AGX-014, AGX-015, AGX-016
Matrox MGA2064W (Millennium), MGA1064SG (Mystique
and Mystique 220), MGA2164W (Millennium II PCI
and AGP), G100, G200
MX MX68000, MX680010
NCR 77C22, 77C22E, 77C22E+
NeoMagic 2200, 2160, 2097, 2093, 2090, 2070
Number Nine I128 (series I and II), Revolution 3D (T2R)
NVidia/SGS Thomson NVidia/SGS Thomson
OAK OTI067, OTI077, OTI087
RealTek RTG3106
Rendition V1000, V2x00
510 Chapter 24 Installing the X Window System
Table 24.1 Chipsets Supported by XFree86
Manufacturer Chipset(s)
S3 86C911, 86C924, 86C801, 86C805, 86C805i, 86C928,
86C864, 86C964, 86C732, 86C764, 86C765, 86C767,
86C775, 86C785, 86C868, 86C968, 86C325, 86C357,
86C375, 86C375, 86C385, 86C988, 86CM65, 86C260
SiS 86C201, 86C202, 86C205, 86C215, 86C225, 5597,
5598, 6326
3DLabs GLINT 500TX, GLINT MX, Permedia, Permedia 2,
Permedia 2v
Tseng ET3000, ET4000AX, ET4000/W32, ET4000/W32i,
ET4000/W32p, ET6000, ET6100
Trident TVGA8800CS, TVGA8900B, TVGA8900C,
TVGA8900CL, TVGA9000, TVGA9000i,
TVGA9100B, TVGA9200CXR, Cyber9320,
TVGA9400CXi, TVGA9420, TGUI9420DGi,
TGUI9430DGi, TGUI9440AGi, TGUI9660XGi,
TGUI9680, ProVidia 9682, ProVidia 9685, Cyber
9382, Cyber 9385, Cyber 9388, 3DImage975,
3DImage985, Cyber 9397, Cyber 9520
Video 7/Headland Technologies HT216-32
Weitek P9000, P9100
Western Digital/Paradise PVGA1
Western Digital WD90C00, WD90C10, WD90C11, WD90C24,
WD90C24A, WD90C30, WD90C31, WD90C33
Understanding the XFree86 RPMs
XFree86 is split into a large number of RPMs, some of which are required, whereas others
are optional. If you installed XFree86 when you installed Red Hat or OpenLinux, everything
was probably done for you. If you didn't install XFree86 at that time, or you're trying to
upgrade XFree86, then you need to decide what packages to install. See Table 24.2 for
the list of recommended packages, Table 24.3 for a list of X servers (you need one, plus
XFree86-VGA16 as a backup), and Table 24.4 for the optional packages that ship with
Red Hat 6.0. OpenLinux 2.2 RPMs have very similar names for the most common packages.
Table 24.2 Recommended XFree86 RPMs
RPM Description
XFree86 Base XFree86 system.
XFree86-libs Shared libraries for most X applications.
X11R6-contrib Many useful X applications.
511
Installing the XFree86 System
Table 24.2 Recommended XFree86 RPMs
RPM Description
Xconfigurator Easy-to-use front end for configuring X Window System. This RPM is
generally shipped only with Red Hat. For other distributions, use
XF86Setup (see ``Using XF86Setup'').
Table 24.3 XFree86 X Server RPMs
RPM Description
XFree86-VGA16 Generic VGA 16-color server. For most configurations, you'll never
Part
want to use this server, but it's good to have around because it works
with just about everything.
IV
XFree86-SVGA SVGA server. Most nonaccelerated chipsets use this server.
Ch
XFree86-Mono SVGA monochrome server.
24
XFree86-S3 S3 accelerated server (except ViRGE).
XFree86-S3V S3 ViRGE accelerated server.
XFree86-Mach32 ATI Mach32 accelerated server.
XFree86-Mach64 ATI Mach64 accelerated server.
XFree86-Mach8 ATI Mach8 accelerated server.
XFree86-8514 8514/A accelerated server.
XFree86-P9000 P9000 accelerated server.
XFree86-AGX AGX accelerated server.
XFree86-W32 ET4000/W32 and ET6000 accelerated server.
XFree86-3DLabs 3D Labs GLINT and Permedia accelerated server.
XFree86-FBDev Amiga, Atari and Macintosh/m68k server.
XFree86-I128 #9 Imagine 128 server.
Table 24.4 Optional XFree86 RPMs
RPM Description
XFree86-75dpi-fonts Standard 75dpi fonts.
XFree86-ISO8859-2 Central European (ISO 8859-2) fonts.
XFree86-ISO8859-9 Turkish language (ISO 8859-9) fonts.
XFree85-Xnest A ``nested'' X server. This X server can run in a window under
another X server, generally for testing purposes.
512 Chapter 24 Installing the X Window System
Table 24.4 Optional XFree86 RPMs
RPM Description
XFree86-XF86Setup XFree86 configuration front end. For Red Hat 6.0, Xconfigurator
(refer to Table 24.2) is generally preferred.
XFree86-Xvfb X server that does not require display hardware. It is very useful
for development but generally not used by end users.
XFree86-cyrillic-fonts Cyrillic fonts.
XFree86-devel Development package. It is required if you plan to develop or
compile X applications or servers.
XFree86-doc Low-level documentation for X developers.
XFree86-xfs Font server for XFree86. It is used to provide the fonts on your
machine to other machines on the network.
? See ``Installing Packages with RPM,'' p. 169
Font packages often come in several flavors. The ``base'' version (as listed in Table 24.4) is
generally terminal fonts and keyboard remaps for the languages in question. The 75dpi fonts
version contains 75dpi (dots per inch) fonts, which can be used by most monitors. The 100dpi
fonts version contains 100dpi fonts, which are good for high-end (usually large) monitors that
can handle 100dpi. Finally, some fonts come in a Type1 fonts version, which are Type 1
scalable fonts. Installing at least the base, 75dpi, and Type1 versions is highly recommended.
If your monitor can support the 100dpi versions, having them can be quite nice.
Remember that you have to install the appropriate programs from each package. Although
not all packages are required, if you install XFree86 after installing Linux, you should review
the full details on the packages to install. If you have the 50MB needed for a full installation,
go ahead and install each package, with the exception of the X server. Install only one X server
for your chipset.
The XFree86 RPMs that ship with Red Hat 6.0 have serious performance, functionality,
and security bugs. Upgrading these packages from ftp://ftp.redhat.com/
updates/6.0 is highly recommended.
Installing XFree86 for Red Hat 6.0
The easiest way to install XFree86 for Red Hat 6.0 is to let the Red Hat installer do it for you.
This will save you a lot of trouble in determining which RPMs you require and will
automatically detect your video hardware. Follow these steps:
513
Installing the XFree86 System
?
1. Insert your Boot disk and Red Hat 6.0 CD and reboot your system.
See ``Shutting Down Linux,'' p. 243
2. At the main boot prompt, press Enter to install Red Hat 2.0 or later, and press Enter
again at the welcome screen. Select your language, keyboard, and install device. Then
select Upgrade and answer the question about whether you have any SCSI devices.
3. The installer will now inform you that it has determined what you should install and
will ask whether you would like to customize the list of packages. Select Yes.
4. Look through the list and deselect everything except the three User Interface sections.
The installer has probably selected these if you have not yet installed XFree86. If it
has, then you can probably accept its selections (check over them if you like). If it has
not selected the User Interface sections, do so now. Unless you are low on disk space,
just select everything.
Part
5. Select OK and wait for the installer to complete. Skip the boot disk creation and boot
IV
loader installation if you've done them before. Finally, your system will reboot.
Ch
6. The RPMs that ship with Red Hat 6.0 have serious problems with functionality,
security, and performance, so you shouldn't stop here. You should now go to 24
ftp.redhat.com and get the updates.
7. When your system reboots, log in as root and collect a list of your installed RPMs:
# rpm -qa | grep XFree86
8. xterm so you can remember what RPMs
You should write these down or open another
you need. Now ftp to ftp.redhat.com:
# ncftp ftp://ftp.redhat.com/redhat/updates/6.0/i386
9. Get a list of all the updated XFree86 RPMs:
ncftp /redhat/updates/6.0/i386 > ls XFree86*
10. Now look at your list of installed RPMs and make a note of any that have been
updated (there will be several). Queue the updated RPMs for download:
ncftp /redhat/updates/6.0/i386 > bgget XFree86-3.3.3.1-52.i386.rpm
+ Spooled: get XFree86-3.3.3.1-52.i386.rpm
ncftp /redhat/updates/6.0/i386 > bgget XFree86-75dpi-fonts-3.3.3.1-52.i386.rpm
+ Spooled: get XFree86-75dpi-fonts-3.3.3.1-52.i386.rpm
[...]
Tip #126 from
Rob ncftp has filename completion, so you don't have to type the entire filename. Just
type enough to make it unique and press the Tab key. If you press Tab when what
you've already typed is not unique, ncftp will expand as much as it can.
For example, you can type XF < Tab >1< Tab > to get XFree86-100dpi-fonts-
3.3.3.1-52.i386.rpm.
514 Chapter 24 Installing the X Window System
You want to queue the downloads because there are too many RPMs to fit on one line.
11. When you've queued all the files you want, quit ncftp and the download will
automatically start. You can watch the progress of the download by watching
~/.ncftp/batchlog:
# tail -f ~/.ncftp/batchlog
12. When the download is complete, update all the RPMs with the following command:
# rpm -U XFree86*
If you receive dependency errors while upgrading, see ``Dependency errors while
upgrading with rpm'' in the Troubleshooting section of this chapter.
? See ``Updating Packages with RPM,'' p. 171
XFree86 has historically been one of the most complex parts of Linux to configure. This is no
longer the case, however, for most common hardware. There are two cases where installation
can still be difficult, however.
First, cutting edge hardware may or may not be supported by XFree86 at all. When XFree86
does support it, you may have to use beta versions of XFree86 or even hacks on XFree86.
These will not be supported by the new configuration tools.
Second, some vendors do not publish the specifications for their boards. In order for XFree86
to support these boards, developers have to reverse engineer them, which takes a lot of time
and effort. Unless the board is extremely popular (e.g. the S3 ViRGE), there may not be
XFree86 support for a very long time. Getting unsupported boards to work can be difficult or
impossible if you are unable to write the drivers yourself.
These are just the worst cases. For most popular hardware you can just use Xconfigurator,
and it should work without any trouble.
Using Xconfigurator
Xconfigurator is written by Red Hat. It is a text-based, menu-driven program which walks
you through setting up your X server. It provides defaults for a wide variety of hardware and
is much easier to use than the configuration tools that historically came with XFree86.
After you install XFree86, you can log in as root and run Xconfigurator. If your video card
and monitor are listed, select the appropriate entries, and with luck, you will be done. Then
you can check out your configuration by running startx.
515
Installing the XFree86 System
If something does go wrong (and your monitor doesn't explode), pressing
Ctrl+Alt+Backspace should terminate the X server and return you to a shell prompt.
If you don't find your hardware listed, you can check the information in the /usr/X11/lib/
X11/doc directory. In particular, check out the files named README.Config and README.Linux,
as well any of the README files that refer to your hardware.
If you find enough information here to get you through Xconfigurator, then you're done.
Otherwise, it's time to use some other tools.
Using XF86Setup
Part
The version of XF86Setup that comes with Red Hat 6.0 does not run. To use it, you
IV
need to update your XFree86 RPMs from http://updates.redhat.com.
Ch
24
XF86Setup can handle some cards that Xconfigurator doesn't know about. After you read the
files in /usr/X11/lib/X11/doc, try running it. Then select your card and monitor
specifications and run startx. If that still doesn't work, it's time to start digging into
XF86Config.
You should never use an XF86Config file from someone else, or even one verbatim
from this book or any other source, without looking over the file for improper values.
For example, driving your monitor at unsupported frequencies may damage your
equipment.
Running the SuperProbe Program
If the preceding installation procedures don't work, you can run a program to configure your
system. XFree86 provides a program called xf86config to help you configure your XFree86
system, but this program requires you to answer several questions. These questions deal with
the type of hardware you have on your system, and incorrect information can cause X to
damage that hardware.
You should read these document files located in the /usr/X11R6/lib/X11/doc directory:
QuickStart.doc and README.Config. You can use the following command to read the files:
less filename
You should also gather any manufacturer's manuals for your video card and monitor.
516 Chapter 24 Installing the X Window System
Next, run the SuperProbe utility:
/usr/X11R6/bin/SuperProbe
This utility scans your system, trying to identify the installed video hardware. You should
write down the information reported for later use with the xf86config program. You should
also double-check the information generated by SuperProbe with your hardware's
documentation. The SuperProbe program generates information that will be placed in the
various sections of the XF86Config file.
Understanding the XF86Config Sections
The XF86Config file is a normal ASCII text file read by XFree86 and used to configure the X
server to run properly under your hardware system. The file is formatted into the sections
shown in Table 24.5.
Table 24.5 XF86Config File Sections
Section Description
Files Lists directories for the font and rgb files.
ServerFlags Specifies special flags for the X server.
Keyboard Describes the type of keyboard.
Pointer Describes your pointing device, typically your mouse.
Monitor Provides detailed descriptions about your monitor. This section is very
important because incorrect information can severely damage the monitor.
Device Describes your video card.
Screen Uses the information from the Monitor and Device sections to describe your
physical screen area, including such items as number of colors and size of the
screen in pixels.
Each section in the file has the following general form:
Section ``Name''
data entry values
data entry values
more values as needed...
#this is a comment line and is ignored by XFree86
EndSection
You should build such a configuration file using a text editor such as vi, following the
examples given. After creating the file, you can run the xf86config program to generate an
XF86Config file for comparison. Finally, you can run the X server in a special mode to probe
for your system's settings, which you might not be able to determine from the examples, the
generated file, or the documentation. These precautions are necessary because of the real
threat of damage to your system.
517
Installing the XFree86 System
? See ``Using vi,'' p. 207
The Files Section
The Files section lists the various fonts installed on your system in the /usr/X11R6/lib/X11/
fonts directory. Each font series has its own subdirectory here, so you can use the following
command to determine which ones are loaded:
ls /usr/X11R6/lib/X11/fonts
Each directory listed should have a corresponding entry in the Files section.
Depending on your selections during installation, your font files should go into standard
directories, and your Files section should appear as in the sample section here:
Section ``Files''
RgbPath ``/usr/X11R6/lib/X11/rgb'' Part
fontPath ``/usr/X11R6/lib/X11/misc/''
fontPath ``/usr/X11R6/lib/X11/Type1/'' IV
fontPath ``/usr/X11R6/lib/X11/speedo/''
fontPath ``/usr/X11R6/lib/X11/75dpi/''
Ch
fontPath ``/usr/X11R6/lib/X11/100dpi/'' 24
EndSection
The ServerFlags Section
You'll rarely need to edit the default ServerFlags section. This section controls the following
three flags used by the X server to control its operation:
Flag Description
NoTrapSignals This advanced flag causes the X server to dump
coreÐcreate a debugging fileÐwhen an operating
system software signal is received by the X server.
DontZap This flag disables the use of the Ctrl+Alt+Backspace key
combination to terminate the X server.
DontZoom This flag disables switching between various graphics
modes.
In the following sample section, each flag is commented out and thus disabled:
Section ``ServerFlags''
#NoTrapSignals
#DontZap
#DontZoom
EndSection
The Keyboard Section
In the Keyboard section, you can specify several options for your keyboard, such as key
mappings. The minimal Keyboard section looks like this:
518 Chapter 24 Installing the X Window System
Section ``Keyboard''
Protocol ``Standard''
AutoRepeat 500 5
ServerNumLock
EndSection
Many more options are available, as shown in Table 24.6, but many aren't required for proper
operation of your keyboard. You can type man XF86Config at a shell prompt to see a full
description of the various parameters for each section of the XF86Config file.
Table 24.6 Keyboard Section Options
Option Parameter/Description
Protocol Is Standard or Xqueue (Standard is the default)
AutoRepeat delay rate Sets the delay before repeating the key at the specified rate
ServerNumLock Tells the X server to handle the response to the NumLock key
internally
VTSysReq Specifies that the X server will handle switching between virtual
terminals by using the SysRq key instead of the Ctrl key
Typically, you use the Alt+F x method to switch between the various virtual terminals under
Linux (where F x indicates any function key). But when you're working in XFree86, you must
use Ctrl+Alt+F x to access the virtual terminal. Of course, if you're questioning the need for
virtual terminals when running a GUI, consider what happens if your X session locks; you can
?
then use a virtual terminal to kill your X session.
See ``Display Managers and Logging In,'' p. 538
The Pointer Section
The Pointer section deals with your mouse or other pointing device. XFree86 uses the
information here to configure your mouse. Minimally, you should specify the protocol used
by your mouse and the device type. If you have a serial mouse, the device is the serial port
used by the mouse. The following is a sample Pointer section:
Section ``Pointer''
Protocol ``Auto''
Device ``/dev/mouse''
EndSection
The various protocols supported by Linux are as follows:
Auto Microsoft NetScrollPS/2
BusMouse MMHitTab OSMouse
GlidePoint MMSeries PS/2
GlidePointPS/2 MouseMan SysMouse
519
Installing the XFree86 System
IntelliMouse MouseManPlusPS/2 ThinkingMouse
IMPS/2 MouseSystems ThinkingMousePS/2
Logitech NetMousePS/2 Xqueue
You can use Auto for newer serial mouse devices and some PS/2 and bus mouse devices.
Logitech refers to the old serial Logitech protocol. Most newer Logitech serial mouse devices
use the Microsoft or MouseMan protocols. PS/2 should work with any PS/2-style mouse.
Some of the other options available in the Pointer section are shown in Table 24.7, but you
shouldn't add them to your XF86Config file unless you're absolutely sure what effect they'll
have on your system.
Table 24.7 Pointer Section Options
Part
Option Description
IV
BaudRate rate Specifies the baud rate for a serial mouse.
Ch
SampleRate rate Needed by some Logitech mouse devices.
24
ClearDTR or ClearRTS Required by some mouse devices using the MouseSystem
protocol.
ChordMiddle Needed by some Logitech mouse devices. If your middle mouse
button doesn't work, you should try this option.
Emulate3Buttons Allows a two-button mouse, such as some Microsoft mouse
devices, to emulate a three-button mouse. The third button is
emulated when you press both buttons at once. Many X
applications need a three-button mouse for proper operation.
The Monitor Section
The Monitor section is probably the most important section of the XF86Config fileÐand
probably the most dangerous. Misinformation in this file can cause catastrophic damage to
your system, so be careful!
The SuperProbe program and your manufacturer's documentation will help greatly in
creating this section. You can also use the file /usr/X11R6/lib/X11/doc/Monitors to search
for information on your particular monitor.
The following is a typical Monitor section:
Section ``Monitor''
Identifier ``Sanyo 1450 NI''
VendorName ``Sanyo''
ModelName ``My 14 inch monitor''
Bandwidth 60
HorizSync 30-60
VertRefresh 50-90
#Modes: Name dotclock Horizontal Timing Vertical Timing
520 Chapter 24 Installing the X Window System
ModeLine ``640x480'' 25 640 672 768 800 480 490 492 525
ModeLine ``800x600'' 36 800 840 912 1024 600 600 602 625
ModeLine ``1024x768i'' 45 1024 1024 1224 1264 768 768 776 816
EndSection
Your Monitor section can have more than one monitor defined, so for each monitor, you can
supply the information shown in Table 24.8.
Table 24.8 Monitor Section Options
Option Description
Identifier string Identifies the monitor.
VendorName string Identifies the manufacturer.
ModelName string Identifies the make and model.
HorizSync range Specifies the valid horizontal sync frequencies (in kHz). They can be a
range if you have a multisync monitor or a series of single values for a
fixed-frequency monitor.
VertRefresh range Specifies the vertical refresh frequencies. They can be listed as a range
or a series of single values, like the HorizSync value.
Gamma value Specifies the Gamma correction value for your monitor.
ModeLine values Specifies a series of values for each resolution to be displayed on the
monitor. You can also enter this information in a Mode block, which is
more descriptive, but most configuration programs use the terse
ModeLine, so that's what we'll show here.
Each monitor entry must have Identifier, HorizSync, VertRefresh, and one or more
Modeline entries.
For each resolution, you need a ModeLine entry in the Monitor section. The entry has the
following format:
ModeLine '' name'' dotclockhorizontalverticalmodifiers
name is a descriptive name for the mode. dotclock (or pixel clock) is your video adapter's
driving clock frequency. horizontal and vertical are each a list of four frequencies that
describe how your monitor actually displays the information it receives from the video
adapter. modifiers are entries such as Interlace, +VSync, or Composite that modify the entire
description.
All these values should be determined by running the xf86config program (discussed later in
the section ``Running the xf86config Program'') or from the various documentation files
included with the XFree86 package. For your initial test, it's best to enter a standard
configuration from the documentation and then let XFree86 probe your system for more
appropriate values.
521
Installing the XFree86 System
A full description of these values is beyond the scope of this book, but if you have a monitor
that is not supported, you can refer to the ``Video Timings How-To'' by Eric S. Raymond
([email protected]). You can find this document in /usr/X11R6/lib/X11/doc/VideoModes.-
doc.
Creating a ModeLine can be very challenging and may result in damage to your
monitor and yourself. Before trying to create a ModeLine, make sure that you have
carefully read the most up-to-date version of the ``Video Timings How-To,'' available
at http://metalab.unc.edu/LDP. At the time of this writing, the most up-to-date
version was v3.6 from June 13, 1999. The version that ships with Red Hat 6.0 is about
two years out of date.
Part
IV
The Device Section
Ch
The Device section describes the system's video card to XFree86. The Device section for
Standard VGA looks like the following:
24
Section ``Device''
Identifier ``SVGA''
VendorName ``Trident''
BoardName ``TVG89''
Chipset ``tvga8900c''
VideoRam 1024
Clocks 25.30 28.32 45.00 36.00 57.30 65.10 50.40 39.90
Option ...
EndSection
The only values that might be hard to come by are the clock values. Your video card uses
these values to generate the clock signals that, in turn, provide the various frequencies needed
to display information on your monitor. If you get these values really wrong, you can blow
your monitor! You can get this value by running XFree86 with a special parameter,
-probeonly, which allows XFree86 to scan your system without much chance of physical
damage to your system ( -probeonly is discussed later in this chapter). XFree86 then generates
a report with most of the values needed for your configuration.
Your server might also require additional parameters. These optional entries in the Device
section are detailed in the appropriate man page for your server.
The Screen Section
Your XF86Config file can contain many Monitor and Device section entries. These entries
are tied together in the Screen section to create your X desktop for your X server. A sample
Screen section is as follows:
Section ``Screen''
Driver ``vga2''
Device ``SVGA''
522 Chapter 24 Installing the X Window System
Monitor ``Sanyo 1450 NI''
Subsection ``Display''
Depth 8
Modes ``1024x768'' ``800x600'' ``640x480''
ViewPort 0 0
Virtual 1024 768
EndSubsection
EndSection
The Screen section uses the identifier names from the Device and Monitor sections. The
Driver value tells what X server you're running and can have one of the following values:
& Accel
& SVGA
& VGA16
& VGA2
& Mono
Most configurations use either the Accel or SVGA drivers.
Within the Screen section are display subsections, which describe the various modes available
for a particular resolution. Each Mode value refers back to each Modeline value defined in the
Monitor section.
XFree86 starts at the position specified by the ViewPort value. A value of 0,0 tells XFree86 to
start with position 0,0 in the upper-left corner of the display.
With the Virtual value, you can define a virtual screen that's larger than your physical
screen. If you specify a larger screen, XFree86 automatically scrolls the screen as needed
when you move the pointer to positions outside the range of your physical screen.
Tip #127 from
Rob Many programs found on the Internet assume a three-button mouse and a screen
size of 1152 Á 900. This screen size is a typical screen size found on a Sun workstation.
So to emulate such a system, you would need to specify the Emulate3Buttons in
thePointer section and a Virtual 1152 900 in a Display subsection of the
Screen section.
Perhaps one of the most modified values is the Depth entry. It indicates the color bit depth of
your display. Eight-bit is 256 colors, 16-bit is approximately 65,000 colors, and 24-bit is
approximately 16 million colors. Your video card must be able to handle the bit depth you
request. In many cases, you need to set your resolution (that is, Mode) to a lower setting to get
more colors. Some high-end video cards support 32-bit color (approximately 4 billion colors).
Some cards ignore a request for 32-bit color and simply provide 24-bit color instead.
523
Installing the XFree86 System
Tip #128 from
Rob The Mode line is supposed to accept multiple modes, and you should be able to
switch between them by pressing Ctl+Alt+Keypad-Plus. Unfortunately, this key
combination doesn't always work, and you'll find yourself stuck with the first mode in
your list. If this is the case, you will have to choose the resolution that you want and
remove the other modes from the Mode line.
Running the xf86config Program
After running SuperProbe and building a basic XF86Config file, you then can run the
xf86config program to generate a configuration file for your system. First, you should make
sure that you aren't in the /usr/X11R6/lib/X11 directory because XFree86 looks for the
Part
XF86Config file there first, and you don't want to overwrite the file you just created. To run
the xf86config program, issue the following command: IV
/etc/X11/bin/xf86config Ch
The xf86config program asks many questions about your system, which it uses to fill in the 24
various sections of the XF86Config file. After the program finishes, you must check to make
sure that the values are similar to the ones you collected while creating your version of the
file. The only items you'll need help with are the clock values for your monitor. You can get
XFree86 itself to help with those values.
Running XFree86 in Probe-Only Mode
If you run XFree86 in a special mode, the program generates a file with information about
your entire system. You can use the information in this file to complete your XF86Config file.
To run XFree86 in the special probe-only mode, simply enter this command:
X -probeonly > /tmp/x.value 2]]&1
The command redirects the output of XFree86 into a file named /tmp/x.value. You can edit
this text file with any text editor, such as vi. Using the text editor, you can cut the clock
information from this file and paste the information into your XF86Config file, thus
?
completing your configuration file for XFree86.
See ``Copying, Cutting, and Pasting,'' p. xxx (Chapter 9, ``Using the vi Editor.'')
Now you can copy the file you've created into one of the directories XFree86 looks through.
More than likely, you can copy the file by using this command:
cp XF86Config /etc/X11/
You're now ready to start your X server by using the startx command.
Tip #129 from
Rob If something does go wrong, pressing Ctrl+Alt+Backspace will terminate the X server
and return you to a shell prompt.
524 Chapter 24 Installing the X Window System
Troubleshooting
When XFree86 starts up, text flashes on the screen, the monitor clicks and then repeats too fast to
stop it.
In almost all cases, this indicates an error in your XF86Config file when your system is at run
level 5.
On Red Hat 6.0 systems and many others, run level 5 means that XFree86 automatically
starts up. This is due to the following entry in your /etc/inittab:
x:5:respawn:/etc/X11/prefdm -nodaemon
The respawn flag tells init to restart this program if it ever dies. So when XFree86 starts up,
it reads its configuration file, terminates with an error, and then init immediately restarts it.
Since XFree86 switches your monitor's video mode when it starts up, you won't be able to
type any commands at the console while this is happening.
You can get control of your machine by rebooting into single user mode. Press
Ctrl+Alt+Delete to reboot your machine. At the boot: prompt, type linux -s and press
Enter. The system will boot into single-user mode, and you will be logged in as root. Switch
to run level 3 by typing the following:
# telinit 3
This will bring your system up without XFree86 so you can work on the configuration file. In
order to test your configuration file, run startx. This will show you where the errors are so
you can fix them.
Tip #130 from
Rob If your problem machine is on a network, you can avoid rebooting. Just go to another
machine and telnet in as a normal user (you can't telnet as root). Then su to root
(type su, press Enter, and type the root password). Finally, type telinit 3 and your
machine will be ok again.
I get dependency errors while upgrading with rpm.
Many of the XFree86 RPMs depend on other XFree86 RPMs. If you try to upgrade them in
the wrong order, you'll get errors.
Figuring out the correct order can be quite difficult and is seldom worthwhile. Just run the
upgrade command again and everything should work. During the first pass, all the
independent RPMs will install, so during the second pass, the dependent RPMs will be fine.
XFree86 displays the wrong resolution.
XF86Config probably lists video modes that you didn't intend.
525
Case Study: The X Window System Across a Network
You need to edit your XF86Config file. Before doing this, you need to switch to run level 3. If
you don't and you make any errors in this file, it will be difficult to get control of your
machine again. To switch to run level 3, become root and type telinit 3 at the command
prompt.
Now edit XF86Config and look for the Screen sections. They will look something like this:
Section ``Screen''
Driver ``svga''
# Use Device ``Generic VGA'' for Standard VGA 320x200x256
#Device ``Generic VGA''
Device ``My Video Card''
Monitor ``NEC MultiSync XV14''
Subsection ``Display''
Depth 16
Modes ``320x200'' ``800x600''
ViewPort 0 0 Part
Virtual 800 600
EndSubsection IV
EndSection Ch
There may be several Screen sections. Make sure you are using the one for your driver. 24
Notice the 320x200 in the Modes line. XFree86 will start with the first resolution it finds in the
Modes line. Unless you really need to switch video modes (it is usually less useful than it would
seem), just list one mode here. Type startx and make sure it works. Then switch back to run
level 5 with telinit 5 at the command prompt.
XFree86 doesn't start when Linux boots.
You are in run level 3 instead of run level 5. Edit your /etc/inittab and change the id line to
this:
id:5:initdefault:
When you restart your machine, XFree86 will start automatically.
Case Study: The X Window System Across a
Network
The X Window System is an incredibly flexible, portable system. Because it is network
transparent, it is easy to run applications on remote machines and display them on your
desktop. Figure 24.1 shows my personal desktop.
This desktop is noteworthy because of how complex it really is. The X Window System hides
so much of that complexity that it is easy to take it for granted. Here are some facts to
consider:
526 Chapter 24 Installing the X Window System
Figure 24.1
The X Window
System makes
remote ma-
chines seem
local.
& Mutt (the mail program) is running on a remote Solaris machine via an ISDN line. All
the text data is encrypted and compressed as it is sent between the two machines (this
is done with a program called ssh). The graphical part of the xterm, however, is
actually running on my local Linux machine, so the ISDN line only has to handle the
text.
& Emacs (the editor) is also running on the local Linux machine. But the file being
edited is actually on the same remote machine that Mutt is running on.
& Meeting Maker (the calendar program) is running completely on the remote Solaris
machine, sending all of its graphics information back to my Linux machine (once
again compressed and encrypted). This allows me to run Meeting Maker even though
it isn't ported to Linux.
& The buttons at the bottom will launch these remote applications automatically. Once
set up, I can generally ignore which machine is running what application. They all
display on the same machine, and I can treat them all the same.
& What isn't obvious from the screen shot is that I'm actually sitting at a Windows 98
machine running Reflection X. The Linux machine and Windows 98 machine are
connected with a fast local network. The Linux machine displays my desktop on the
Windows machine. I still get all the advantages of Linux, but I can run Windows
applications without leaving my chair. Besides that, another user (my wife) can log into
the Linux console and use Netscape there while I'm reading mail, running my own
copy of Netscape, etc. It's all transparent.
527
Case Study: The X Window System Across a Network
I've used a lot of different operating systems and graphical user interfaces and have yet to find
one that can offer the incredible flexibility and power of the X Window System.
Part
IV
Ch
24
CHAPTER
25
Using the X Window System
In this chapter by Rob Napier
Navigating the X Window System 530
Using Window Managers for Linux 531
Choosing a Window Manager 536
Themes 537
Display Managers and Logging In 538
Choosing Your Display Manager 539
XFree86 Startup 540
Using X Applications 541
Troubleshooting 551
Project: Making Yourself at Home with fvwm2 552
530 Chapter 25 Using the X Window System
Navigating the X Window System
The X Window System presents to the user several windows, each showing the output of an
X application called a client. The client can be running on the user's PC, which is more than
?
likely with Linux, or on another workstation on the network.
See ``What Is a Client/Server System?'' p. 506
How you move around in the X Window System very much depends on your window
manager. Most window managers use an onscreen pointer called a cursor to indicate where
you're working. The cursor can take on many shapes, depending on what you're doing and
what window manager you're running.
The X Window System, like most graphical user interfaces (GUIs), allows input from the
keyboard and a pointing device, which is usually a mouse. Typically, for a window to accept
input, it must be the active window. An active window normally has a different appearance
(for example, a highlighted border) than inactive windows. When a window is selected, it is
said to have gained focus.
Getting Focus
Making a window active depends on how you've configured your window manager. Some
configurations allow the window to become active when you merely move the cursor into the
window; others require you to click the window with the mouse, like you do in Microsoft
Windows. Table 25.1 lists the most common focus schemes.
Table 25.1 Common Focus Schemes
Name Description
Click You must click a window for it to receive focus. This focus scheme is used by
Microsoft Windows.
Mouse Whatever window the pointer is currently in has focus. If the pointer is not
currently in a window, then no window has focus.
Sloppy This scheme is very similar to Mouse, except that a window does not give up focus
until the pointer moves into a new window. Using this scheme is often much more
convenient than using Mouse focus.
Mouse and Sloppy focus can take some getting used to if you are familiar with GUIs that
allow only Click focus. When you get used to them, though, these types of focus let you do
things you couldn't do with Click focus. For example, if you are reading one window and
typing in another, you can have the window you're reading appear on top, even partially
obscuring the active window, while the window you are typing in retains focus.
531
Using Window Managers for Linux
Using Menus
Many GUIs on PCs today provide drop-down and pop-up menus. Again, the availability of
such items depends on the window manager, including the types of menu choices provided.
Most X window managers don't have a main menu bar across the top or bottom of the
monitor; instead, they use a floating menu. You typically invoke this floating menu by
clicking over an empty area of the desktop. You hold down the mouse button and drag the
cursor through the various menu selections. When you find the desired menu choice, you
simply release the button, which is very much like how you navigate menus on a Macintosh
and very unlike how you navigate menus under Microsoft Windows.
Using Virtual Terminals in XFree86
Your X server runs on a virtual terminal assigned by Linux. This terminal is assigned to the
seventh virtual terminal, which you can reach by pressing Ctrl+Alt+F7 from a character
terminal. From XFree86, you can reach the other terminals by pressing the Ctrl+Alt+F x key
combination, where x represents the number of the virtual terminal you want to access.
Although accessing the other virtual terminals can be handy, XFree86 does allow you to start
character terminal emulators, called xterm sessions.
Part
If your X server is running, you must use the Ctrl+Alt+F x combination to move from IV
the X server to a virtual terminal. You can still use the Alt+F x combination to move
Ch
among the virtual terminals.
25
Using Window Managers for Linux
As stated earlier in the chapter, the X Window System doesn't specify a window manager.
The look and feel of the X Window System is left up to the userÐcompletely up to the user.
Almost every aspect of the behavior of the GUI is in your control. In this spirit, Linux doesn't
provide just one window manager for XFree86. Table 25.2 lists some of the various window
managers available for Linux.
Table 25.2 Some Window Managers Available for Linux
Name Description
Twm Tab Window Manager (sometimes called Tom's Window Manager). It is
the ancestor of many other window managers, such as vtwm, ctwm,
tvtwm, piewm, fvwm (and its descendants), and was the first ICCCM
window manager. (ICCCM is a standard for how programs communicate
with each other.)
532 Chapter 25 Using the X Window System
Table 25.2 Some Window Managers Available for Linux
Name Description
fvwm2 F Virtual window manager (no one remembers what the F actually stands
for, including Rob Nation, the author). It is one of the most popular
window managers for Linux (and elsewhere). Note that many Linux
systems have an fvwm and an fvwm2 executable. fvwm is usually version
1.24r, whereas fvwm2 is some 2.x version. You almost always want
fvwm2.
fvwm95 Hack of fvwm2 that looks a lot like Microsoft's Windows 95.
AfterStep Emulation of the NeXT interface written on top of fvwm2.
Window Maker Window Maker emulates the NeXT interface. It is very popular and has
a lot of extensions available. It is now somewhat integrated with
GNOME and KDE.
Blackbox Blackbox is also based on the NeXT interface, but it is designed to be
small and fast, without lots of fancy features.
Enlightenment The primary window manager for GNOME. It is quite large but is very
pretty. It is the default window manager for Red Hat 6.0. Enlightenment
can also be run without GNOME, but this practice is becoming less
common than it once was.
kwm The primary window manager for KDE. kwm is almost never run
without KDE. It is somewhat smaller than Enlightenment but doesn't
have some of the bells and whistles.
twm
Although few new users use twm, studying this window manager is very valuable because it
set the groundwork for many of the most popular window managers today. In particular, the
very popular fvwm2 is based on twm code.
The twm window manager for the X Windows System provides title bars, shaped windows,
several forms of icon management, user-defined macro functions, click-to-type and pointer-
driven keyboard focus, and user-specified key and mouse button bindings. This program is
usually started by the user's session manager or startup script. When used from xdm or xinit
without a session manager, twm is frequently executed in the foreground as the last client.
When it is run this way, exiting twm causes the session to be terminated (that is, logged out).
By default, an application window is surrounded by a frame with a title bar at the top and a
special border around the window. The title bar contains the window's name, a rectangle
that's lit when the window is receiving keyboard input, and function boxes known as title
buttons at the left and right edges of the title bar. Clicking Button1 (usually the leftmost
mouse button, unless it has been changed with xmodmap) on a title button invokes the
function associated with the button. In the default interface, windows are iconified (minimized
to an icon) when you click the left title button, which looks like a dot. Conversely, windows
are deiconified, or restored, when you click the associated icon or entry in the icon manager.
533
Using Window Managers for Linux
You can resize windows by clicking the right title button (which resembles a group of nested
squares), dragging the pointer over the edge that's to be moved, and releasing the pointer
when the outline of the window is the size you want. Similarly, you can move windows by
clicking the title bar, dragging a window outline to the new location, and then releasing when
the outline is in the position you want. Just clicking the title bar raises the window without
moving it.
When you create new windows, twm honors any size and location information you request.
Otherwise, you see an outline of the window's default size, its title bar, and lines dividing the
window into a three-by-three grid that track the pointer. Each mouse button performs a
different operation:
& Clicking Button1 positions the window at the current position and gives it the default
size.
& Clicking Button2 (usually the middle mouse button) and dragging the outline gives
the window its current position but allows you to resize the sides as described
previously.
& Clicking Button3 (usually the right mouse button) gives the window its current
position but attempts to make it long enough to touch the bottom of the screen.
Part
IV
fvwm
Ch
The fvwm window manager is a derivative of twm, redesigned to minimize memory
consumption, provide a three-dimensional look to window frames, and provide a simple 25
virtual desktop. Memory consumption is estimated at about a half to a third the memory
consumption of twm, due primarily to a redesign of twm's inefficient method of storing
mouse bindings (associating commands to mouse buttons). Also, many of the configurable
options of twm have been removed.
XFree86 provides a virtual screen whose operation can be confusing when used with the
fvwm virtual window manager. With XFree86, windows that appear on the virtual screen
actually get drawn into video memory, so the virtual screen size is limited by available video
memory.
With fvwm's virtual desktop, windows that don't appear onscreen don't actually get drawn
into video RAM. The size of the virtual desktop is limited to 32,000 632,000 pixels. Using a
virtual desktop of more than five times the size of the visible screen in each direction is
impractical.
Memory usage with the virtual desktop is a function of the number of windows that
exist. The size of the desktop makes little difference.
534 Chapter 25 Using the X Window System
When you're becoming familiar with fvwm, disabling XFree86's virtual screen by setting the
virtual screen size to the physical screen size is recommended. When you become familiar
with fvwm, you might want to re-enable XFree86's virtual screen.
fvwm provides multiple virtual desktops for users who want to use them. The screen is a
viewport onto a desktop that's larger than (or the same size as) the screen. Several distinct
desktops can be accessed. The basic concept is one desktop for each project or one desktop
for each application. Because each desktop can be larger than the physical screen, windows
that are larger than the screen or large groups of related windows can be viewed easily.
The size of each virtual desktop must be specified at startup; the default is three times the
physical size of the screen. All virtual desktops must be the same size. The total number of
distinct desktops doesn't need to be specified but is limited to approximately 4 billion total.
All windows on the current desktop can be displayed in a pager, miniature view, or the
current desktop. Windows that aren't on the current desktop can be listed, with their
geometries, in a window list, accessible as a pop-up menu. (The term geometries specifies the
coordinates and number of pixels needed for the window under an X window manager.)
Sticky windows are windows that float above the virtual desktop by ``sticking to the screen's
glass.'' They always stay put onscreen. Using this type of window is convenient for clocks and
xbiffs, for example, so you need to run only one such utility, and it always stays with you.
The xbiff application notifies you when new mail arrives.
Window geometries are specified relative to the current viewport; that is, xterm-geometry
+0+0 always appears in the upper-left corner of the visible portion of the screen. You can
specify geometries that place windows on the virtual desktop but offscreen. For example, if
the visible screen is 1,000 61,000 pixels, the desktop size is three-by-three, and the current
viewport is at the upper-left corner of the desktop, invoking xterm-geometry +1000+1000
places the window just off the lower-right corner of the screen. You can find it by moving the
mouse to the lower-right corner of the screen and waiting for it to scroll into view. Keep in
mind that you can map a window only onto the active desktop, not an inactive desktop.
A geometry specified as xterm-geometry 7575 generally places the window's lower-right
corner five pixels from the lower-right corner of the visible portion of the screen. Not all
applications support window geometries with negative offsets.
Many Linux systems have an executable called fvwm, which is version 1.24r of fvwm.
To get version 2.x, you need to run fvwm2.
535
Using Window Managers for Linux
fvwm95
The fvwm95 window manager is a hack based on fvwm2.x. The developers' goals were to
simulate the major features of a well-known operating system's GUI, to make the users more
comfortable in a UNIX environment, and to avoid bloating the simple and clean GUI code of
fvwm. For more information, go to http://mitac11.uia.ac.be/html-test/fvwm95.html.
AfterStep
AfterStep started life as a package for fvwm called BowMan. BowMan gave fvwm a NeXT
look and feel. It was renamed AfterStep when it started picking up new features beyond
simple emulation. The most notable of AfterStep's features is the wharf (see Figure 25.1). It is
based on NeXT's dock but was renamed to avoid copyright problems.
Figure 25.1
AfterStep's
wharf allows
quick access to
programs and
can even hold
running appli-
cations. Part
IV
Ch
25
The individual buttons on the wharf can launch other applications or can be applications
themselves. For example, the picture of the penguin in Figure 25.1 launches an xterm session,
and the clock is a real, running clock. Many programs are available for the AfterStep wharf.
Window Maker
Window Maker is another window manager based on the NeXT interface. It is original code,
however, and is not based on fvwm. Several of the AfterStep developers have moved over to
the Window Maker project. This window manager, which includes some integration into
?
KDE and GNOME, also provides themes.
See ``Themes,'' p. 537
536 Chapter 25 Using the X Window System
Blackbox
Blackbox is a fairly new window manager being developed by Brad Hughes
[email protected]).
( In the words of its author, ``from the time the first line of
code was written, Blackbox has evolved around one premise, minimalism.'' It is also
based on the NeXT interface but is not as flashy as Window Maker. For that reason,
it is also much faster and requires less memory.
Enlightenment
Enlightenment, which is quickly becoming a very popular window manager, is the default
window manager for Red Hat 6.0. Although it was initially based on fvwm, its newer versions
have been written from scratch. Enlightenment provides more bells and whistles than perhaps
any other window manager. It is also one of the larger window managers and requires a
significant amount of memory and CPU speed to operate properly. Enlightenment is covered
in more detail in Chapter 27, ``Working with GNOME.''
kwm
The K Desktop Environment (KDE) is a large freeware project that was designed to create an
integrated desktop environment similar to the CDE but developed and released entirely
?
under the GNU General Public License (GPL).
See ``The GNU General Public License,'' p.861
The primary window manager for KDE is kwm. For more information on KDE and kwm,
see Chapter 26, ``Working with KDE."
Choosing a Window Manager
To set the default window manager for all users in Red Hat 6.0, you need to edit /etc/
sysconfig/desktop. See Table 25.3 for what to put in this file for various window managers.
Table 25.3 /etc/sysconfig/desktop Settings
Window Manager /etc/sysconfig/desktop Value
Enlightenment GNOME
kwm KDE
AnotherLevel AnotherLevel
Others Path to executable
If this file doesn't exist, Red Hat 6.0 tries to run whichever it finds first of the following list:
Enlightenment, kwm, AnotherLevel, fvwm2, fvwm, and twm.
AnotherLevel isn't really a window manager; it's an older system for picking a window
manager. It includes AfterStep, Window Maker, fvwm95, and mwm. If the default window
537
Themes
manager is AnotherLevel, each user's $HOME/.wm_style is consulted to determine which of these
window managers to run, and it defaults to fvwm95 if no such file exists.
Each user can override the default window manager by creating a $HOME/.Xclients file with
one of the values shown in Table 25.4.
Table 25.4 $HOME/.Xclients Settings
Window Manager /etc/sysconfig/desktop Value
Enlightenment gnome-session
kwm start-kde
Others Path to executable
? See ``XFree86 Startup,'' p. 540
Themes
Using themes is perhaps one of the most fun parts of using window managers. Themes allow
Part
you to define the look and feel of the window manager in a central, consistent way (see
Figures 25.2 and 25.3). IV
Ch
Figure 25.2
Enlightenment 25
using the EMac
theme.
For example, the only difference between the two desktops in Figure 25.2 and Figure 25.3 is
the theme. The first desktop uses the EMac theme by Jon Rista, whereas the second desktop
uses the Clean theme that comes with Enlightenment. Switching between themes does not
require shutting down the currently running applications.
538 Chapter 25 Using the X Window System
Figure 25.3
Enlightenment
using the Clean
theme.
Exactly what a theme can do and how to install it are dependent on the window manager.
Many window managers now support themes in various ways. Check out the site http://
themes.org, which has become the definitive repository for themes, including themes for
AfterStep, Blackbox, Enlightenment, fvwm, kwm, and Window Maker. For more information
on using themes in kwm, see Chapter 26, ``Working with KDE.'' For more information on
using themes in Enlightenment, see Chapter 27, ``Working with GNOME."
Display Managers and Logging In
Before you can use XFree86, you're going to have to log in to Linux. This step is generally
handled by a display manager such as xdm (the original X display manager), gdm (the
GNOME display manager), or kdm (the KDE display manager).
Generally, you don't need to worry about your display manager, but you might be interested
in taking advantage of some new features in gdm and kdm.
xdm
xdm was the original X display manager. It provides user authentication, XDMCP support,
and a host chooser. XDMCP, or X Display Manager Control Protocol, is the protocol used by
remote X serversÐtypically X terminals or X servers running on non-UNIX platformsÐto
log in to your local machine. A host chooser allows remote X servers to find out what other
hosts are willing to let them log in. Often information such as number of current users on
each host is also given; with this information, the end user can choose a host with fewer
people currently logged in.
539
Choosing Your Display Manager
gdm
gdm, which is the GNOME display manager, is a reimplementation of xdm with some extra
features. In particular, it is somewhat more secure than xdm and provides a face browser if
desired. A face browser allows you to select a picture of yourself from a list rather than type in
your username. The photo is stored in ~user/.gnome/photo, where ~user is your home
directory.
The version of gdm that ships with Red Hat 6.0 has a bug in it that prevents remote
hosts from logging in.
kdm
kdm, which is the KDE display manager, provides most of the same features as gdm. It has
the added benefit of allowing you to choose what window manager you want to use at login
time, which is nice if you switch window managers a lot. Typically, the choice between gdm
and kdm is based on whether you use GNOME or KDE.
Part
IV
Choosing Your Display Manager
Ch
In Red Hat 6.0, init starts your display manager via the following entry in /etc/inittab:
25
?
x:5:respawn:/etc/X11/prefdm -nodaemon
For more information about init, see ``Understanding the Boot Process,'' p. 234
This entry tellsinit to start /etc/X11/prefdm if Linux is in run level 5. If prefdm stops for
?
any reason, init should restart it. -nodaemon tells prefdm not to put itself in the background.
See ``Understanding Multitasking,'' p. 366
But what is prefdm? At this point, we've discussed xdm, gdm, and kdm, but not prefdm.
That's because prefdm isn't really a program. It's a symbolic link to the display manager that
you would like to use (your preferred display manager).
To change your default display manager, you should make a link from /etc/X11/prefdm to
the appropriate executable: /usr/X11R6/bin/xdm, /usr/bin/gdm, or /usr/bin/kdm. Then you
can restart your display manager either by pressing Ctrl+Alt+Backspace or by killing the
prefdm process:
ln -s /usr/X11R6/bin/xdm /etc/X11/prefdm
kill Ápidof prefdmÁ
? See ``Links,'' p. 412
540 Chapter 25 Using the X Window System
XFree86 Startup
The startup of XFree86 is one of the most complex sequences that you are likely to have to
deal with. The number of configuration files can be dizzying. Table 25.5 covers most of the
configuration files for Red Hat 6.0, given a default installation (GNOME/Enlightenment), in
roughly the order they are used. KDE or AnotherLevel is quite similar. Other distributions
are mostly similar, but you will find significant differences. The configuration files on other
versions of UNIX can be radically different, particularly regarding in which directory
configuration files are located.
Table 25.5 Red Hat 6.0 X Configuration Files
File Description
/etc/inittab This file read by /sbin/init tells init to run
/etc/X11/prefdm.
/etc/X11/prefdm This file points to the preferred display
manager. It defaults to /usr/bin/gdm.
/etc/X11/gdm/gdm.conf In the Servers section of this gdm configura-
tion file, the 0= line tells gdm which X server to
start. It points to /usr/bin/X11/X.
/usr/bin/X11/X This file points to /usr/bin/X11/Xwrapper (this
is a security feature).
/usr/bin/X11/Xwrapper This file runs /etc/X11/X.
/etc/X11/X This file points to /usr/X11R6/bin/XF86_SVGA
(for example). It is the X server.
/etc/X11/gdm/gdm.conf (again) DisplayInitDir determines which
configuration file is read next.
/etc/X11/gdm/Init/Default This initialization script for gdm points to
/etc/X11/xdm/Xsetup_0.
/etc/X11/xdm/Xsetup_0 This Display Manager setup script runs
xsetroot, which sets the background color,
and xsri, which displays the Red Hat logo.
/etc/X11/gdm/gdm.conf (again) PreSessionScriptDir determines which
configuration file is read next. This directory
doesn't exist by default. SessionDir determines
which configuration file is read next.
/etc/X11/gdm/Sessions/Default This file runs /etc/X11/xdm/Xsession.
/etc/X11/xdm/Xsession This script first runs when a user's session
begins. It merges in the system and user's X
resources and keyboard mappings. By default,
there are no system X resources or keyboard
mappings.
541
Using X Applications
Table 25.5 Red Hat 6.0 X Configuration Files
File Description
$HOME/.Xresources These files are the user's X resources. X
resources are a centralized way of storing
program configurations without using separate
configuration files. On some systems, this file is
called .Xdefaults, but this filename isn't
automatically read by Red Hat 6.0.
$HOME/.Xmodmap This file lets the user remap the keyboard in
various ways.
/etc/X11/xdm/Xsession (again) This script now checks for scripts to run
$HOME/.xsession and
in the user's directory (see
$HOME/.Xclients below). If they aren't found, it
runs /etc/X11/xinit/Xclients.
$HOME/.xsession This file is the first user script to try to run
instead of Xclients.
$HOME/.Xclients This file is the second user script to try to run
instead of Xclients. Using this script is the
normal way to override the default window
manager. Part
/etc/X11/xinit/Xclients This file is the script to run if neither.xsession IV
nor .Xclients exists. It runs gnome-session.
For more information on gnome-session's Ch
configuration files, see Chapter 27, ``Working
25
with GNOME.''
/etc/X11/gdm/gdm.conf (again) PostSessionScriptDir determines
which configuration file is read next. This
directory doesn't exist by default.
Table 25.5 provides a simplified overview of how XFree86 starts up. Among other things, it
does not try to go into X authentication, the greeter, or the chooser. In general, these issues
should not concern users who log in to their Linux machines locally. If you are planning to
use XDMCP and log in to your machine remotely, you might need to investigate this startup
sequence in more depth. We hope that the preceding description will point you in the right
direction.
Using X Applications
Most Linux distributions come with a large number of applications. The following sections
describe some of the most common and useful applications.
xterm
xterm is a common X application that simulates a common video terminal, such as the DEC
vt100 series. When you start an xterm session, you can run any command-line program or
542 Chapter 25 Using the X Window System
execute any Linux command just as you do on any of the virtual terminals supplied by Linux (see
Figure 25.4).
Figure 25.4
xterms provide
convenient ac-
cess to a
command-line
shell.
The xterm program is a terminal emulator for the X Window System. It provides terminals
compatible with DEC vt102 and Tektronix 4014 for programs that can't use the window
system directly. If the underlying operating system supports terminal resizing capabilities,
xterm uses the facilities to notify programs running in the window whenever it's resized.
The vt102 and Tektronix 4014 terminals each have their own windows, so you can edit text in
one and look at graphics in the other at the same time. So you can maintain the correct aspect
ratioÐthe height of the screen in pixels divided by the width of the screen in pixelsÐ
Tektronix graphics are restricted to the largest box with a Tektronix 4014 aspect ratio that
fits in the window. This box is located in the upper-left area of the window.
Although the text and graphics windows might be displayed at the same time, the window
containing the text cursor is considered the ``active'' window for receiving keyboard input and
terminal output. The active window can be chosen through escape sequences, the vt Options
menu in the vt102 window, and the Tek Options menu in the 4014 window.
Emulations
$TERMCAP entries that work with xterm include xterm, vt102, vt100, and ANSI. The $TERMCAP
environment variable specifies the type of terminal your system emulates. xterm automatically
searches the termcap database file in this order for these entries and then sets the TERM and
$TERMCAP environment variables.
For more information on the termcap entries and the escape sequences supported,
see the man page for termcap by typing man termcap at the command line prompt.
Many of the special xterm features can be modified under program control through a set of
escape sequences different from the standard vt102 escape sequences.
543
Using X Applications
The Tektronix 4014 emulation is also fairly good. Four different font sizes and five different
line types are supported. The Tektronix text and graphics commands are recorded internally
by xterm and can be written to a file by sending the Tektronix COPY escape sequence.
Other xterm Features
xterm automatically selects the text cursor when the pointer enters the window and deselects
it when the pointer leaves the window. If the window has the focus, the text cursor is selected
no matter where the pointer is.
In vt102 mode are escape sequences to activate and deactivate an alternate screen buffer,
which is the same size as the display area of the window. When it is activated, the current
screen is saved and replaced with the alternate screen. xterm's capability to save lines scrolled
off the top of the window is disabled until the normal screen is restored. The termcap entry
for xterm allows the visual editor vi to switch to the alternate screen for editing and to restore
the screen on exit.
In vt102 or Tektronix mode are escape sequences to change the names of the windows.
Mouse Usage with xterm
Part
When the vt102 window is created, xterm lets you select text and copy it within the same or
other windows. IV
Ch
The selection functions are invoked when you use the pointer buttons with no modifiers and
when you use them with the Shift key. The assignment of the functions to keys and buttons 25
can be changed through the resource database.
Mouse Button1 (usually the left button) is used to save text into the cut buffer. You move the
cursor to the beginning of the text and then press the button while moving the cursor to the
end of the region; then you release the button. The selected text is highlighted and saved in
the global cut buffer. This selected text is then made the primary selection when you release
the button. Double-clicking selects entire words, triple-clicking selects lines, quadruple-
clicking goes back to characters, and so on.
Mouse Button2 (usually the middle button) pastes the text from the primary selection, if any.
Otherwise, text is inserted from the cut buffer, inserting it as keyboard input.
By cutting and pasting pieces of text without trailing new lines, you can take text from several
places in different windows and form a command to the shell, for example, or take output
from a program and insert it into your favorite editor. Because the cut buffer is shared
globally among different applications, you should regard it as a file whose contents you know.
The terminal emulator and other text programs should be treating the cut buffer as if it were
a text file; that is, the text is delimited by new lines.
The scroll region within the window displaying xterm displays the position and amount of
text now showing in the window relative to the amount of text actually saved. As more text is
saved (up to the system-determined maximum), the size of the highlighted area decreases.
544 Chapter 25 Using the X Window System
Clicking Button1 with the pointer in the scroll region moves the next line to the top of the
display window. Clicking Button2 moves the display to a position in the saved text that
corresponds to the pointer's position in the scrollbar. Clicking Button3 moves the top line of
the display window down to the pointer position.
Unlike the vt102 window, the Tektronix window doesn't allow you to copy text. It does,
however, allow Tektronix GIN mode, in which the cursor changes from an arrow to a cross.
Pressing any key sends that key and the current coordinate of the cross cursor. Clicking
Button1, Button2, or Button3 returns the letters l, m, or r, respectively. If you press the Shift
key when you press a button, the corresponding uppercase letter is sent. So that a pointer
button is distinguished from a key, the high bit of the character is set.
xv
xv is a screen-capture program. Unlike most Linux applications, this program is shareware.
Figure 25.5 shows the main dialog box for the xv application.
Figure 25.5
xv provides a
complete
screen capture
and graphics
file format
conversion
program.
Shareware programs are those that you can download for free. If you find the
programs useful, within a certain time period you are asked to pay the creator of the
program. Shareware programs are usually fairly inexpensive.
The buttons along the right side of the main dialog box are the most useful. Table 25.6
describes their functions. The main file list provides the filenames of each graphics file now
available to the program.
545
Using X Applications
Table 25.6 xv's Command Buttons
Button Description
Next Selects the next file in the file list box.
Prev Selects the previous file in the file list box.
Load Loads a file from disk into the program.
Save Saves the currently captured image to a disk file. You can choose from the
following image types: GIF, JPEG, TIFF, PostScript, PBM (raw), PBM (ASCII),
X11 bitmap, XPM, BMP, Sun raster file, IRIS RGB, Targa (24-bit), Fits, and PM.
Print Prints the currently selected image file.
Delete Deletes the currently selected image file.
By using the Grab button in the lower-right corner of the dialog box, you can capture any
area of the desktop. Clicking this button brings up the xv grab dialog box (see Figure 25.6).
Figure 25.6
You can use a
variety of
Part
methods to
grab any part IV
of the screen
under xv. Ch
25
You use the mouse to select onscreen the object that you want to capture. To capture a
window, you can click the Grab button and then click the left mouse button in the window
you want to capture. You can also set a delay value, click the AutoGrab button, and then
position the mouse cursor in the window. Either way, xv captures the image and displays it in
a window of its own. You can then use the main dialog controls to manipulate and save the
image.
xcalc
Figure 25.7 shows xcalc, a scientific calculator desktop accessory that emulates a TI-30 or
HP-10C calculator. Operations can be performed with mouse Button1 or, in some cases, with
the keyboard.
Many common calculator operations have keyboard accelerators. To quit, you can click the
AC key of the TI calculator with mouse Button3 or click the OFF key of the HP calculator
with mouse Button3. In TI mode, the number keys; the +/- key; and the +, -, *, /, and = keys
all do exactly what you expect them to.
The operators obey the standard rules of precedence. Thus, entering 3+4*5= results
in 23, not 35. You can use the parentheses to override operator precedence. For
example, entering (1+2+3)*(4+5+6)= results in 90 (6615).
546 Chapter 25 Using the X Window System
Figure 25.7
XFree86 pro-
vides various
calculators, in-
cluding TI (pic-
tured) and HP
emulators.
The entire number in the calculator display can be selected for pasting the result of a
calculation into text. Table 25.7 lists the various functions for TI emulation.
Table 25.7 TI Emulation
Key/Function Description
1/x Replaces the number in the display with its reciprocal.
x^2 Squares the number in the display.
SQR Takes the square root of the number in the display.
CE/C When clicked once, clears the number in the display without clearing the
state of the machine, allowing you to re-enter a number if you make a
mistake. Clicking it twice clears the state. (Clicking AC also clears the
display, state, and memory.) Clicking CE/C with Button3 turns off the
calculator, exiting xcalc.
INV Inverts the function. See the individual function keys for details of their
inverse function.
sin Computes the sine of the number in the display, as interpreted by the
current DRG mode (see DRG). If inverted, it computes the arcsine.
cos Computes the cosine. When inverted with the INV key, computes the
arccosine.
tan Computes the tangent. When inverted, computes the arctangent.
DRG Changes the DRG mode, as indicated by DEG, RAD, or GRAD at the
bottom of the calculator display. When in DEG mode, numbers in the
display are assumed to be degrees; in RAD mode, numbers are in radians; in
GRAD mode, numbers are in grads. When inverted, the DRG key has a
feature of converting degrees to radians to grads and vice versa. For example,
put the calculator into DEG mode, and enter 45 INV DRG. xcalc displays
.2285398, which is 45 degrees converted to radians.
e Indicates the constant e, which is 2.22182818.
EE Used for entering exponential numbers. For example, to get -2.3E-4, you
enter 2 . 3 +/- EE 4 +/-.
log Calculates the log (base 10) of the number in the display. When inverted, it
raises 10.0 to the number in the display. For example, entering 3 INV log
results in 1000.
547
Using X Applications
Table 25.7 TI Emulation
Key/Function Description
ln Calculates the log (base e) of the number in the display. When inverted, it
raises e to the number in the display. For example, entering e ln results in 1.
y^x Raises the number on the left to the power of the number on the right. For
example, entering 2 y^x 3 = results in 8, which is 2^3.
PI Indicates the constant pi, which is 3.14159222.
x! Computes the factorial of the number in the display. The number in the
display must be an integer in the range 0 to 500; depending on your math
library, however, it might overflow long before that.
( Left parenthesis.
) Right parenthesis.
/ Division.
* Multiplication.
- Subtraction.
+ Addition.
= Performs calculation. Part
STO Copies the number in the display to the memory location.
IV
RCL Copies the number from the memory location to the display.
Ch
SUM Adds the number in the display to the number in the memory location.
25
EXC Swaps the number in the display with the number in the memory location.
+/- Negate; change sign.
. Decimal point.
In RPN or HP, mode, the numbered keys; CHS (change sign); and +, -, *, /, and Enter keys
all do exactly what you expect. Many of the remaining keys are the same as in TI mode. The
differences are detailed in Table 25.8.
Table 25.8 HP Emulation
Key/Function Description
< Erases digits from the display; you can use this backspace key if you make a
mistake while entering a number. If you invert backspace, the x register is
cleared.
ON Clears the display, state, and memory. Clicking it with Button3 turns off the
calculator, exiting xcalc.
INV Inverts the meaning of the function keys. It is the f key on an HP calculator,
but xcalc doesn't display multiple legends on each key. See the individual
function keys for details.
10^x Raises 10.0 to the number in the top of the stack. When inverted, it
calculates the log (base 10) of the number in the display.
548 Chapter 25 Using the X Window System
Table 25.8 HP Emulation
Key/Function Description
e^x Raises e to the number in the top of the stack. When inverted, it calculates
the log (base e) of the number in the display.
STO Copies the number in the top of the stack to a memory location. There are
10 memory locations. You can specify the memory you want by following
this key with a digit key.
RCL Pushes the number from the specified memory location onto the stack.
SUM Adds the number on top of the stack to the number in the specified memory
location.
x:y Exchanges the numbers in the top two stack positions, the x and y registers.
R v Rolls the stack downward. When inverted, it rolls the stack upward.
(blank keys) These keys were used for programming functions on the HP-10C. Their
functionality hasn't been duplicated in xcalc.
The GIMP
In many ways, GIMP has revolutionized the Linux world. GIMP, which stands for GNU
Image Manipulation Program, is freely distributed software suitable for photo retouching,
image modification, and image creation (see Figure 25.8). The user interface is closely related
to Photoshop.
Figure 25.8
The GIMP is
powerful
image-
manipulation
software.
The GIMP revolutionized the Linux world by being one of the first attempts at a ``mass
market'' Linux application. Although it does not have some of the advanced features of
Photoshop, it has a focus on Web graphics that makes it very powerful in that market. In
549
Using X Applications
particular, it has extremely flexible scripting language, which allows you to script any action that
can be done through the user interface. Therefore, applying complex manipulations to a changing
original is much easier. For example, you can create a Web page that dynamically generates text
with drop shadows by writing a short script and running GIMP to generate the image.
The second way that The GIMP revolutionized the Linux world was by providing the GIMP
Toolkit, or Gtk. When the original authors of the GIMP, Peter Mattis and Spencer Kimball,
needed to create graphics routines, they chose to do so in a very general way. That is, they did
not tie their graphics routines tightly to The GIMP. In doing so, they created a toolkit that
could be used by many other projects that needed graphics routines, most notably GNOME.
GNOME and gtk will be covered in more detail in Chapter 27, ``Working with GNOME.''
The GIMP supports a wide variety of features, including the following:
& Standard painting tools such as Brush, Pencil, Airbrush, and Clone
& Advanced memory management for working with very large images and large numbers
of layers
& Full alpha channel support
& Layers and channels
& A range of very powerful scripting languages including Script-fu and a Perl interface
Part
& Multiple Undo/Redo
IV
& Gradient editor and blend tool
Ch
& Various animation tools
25
& A wide range of file formats, including GIF, JPG, PNG, XPM, TIFF, TGA, MPEG,
PS, PDF, PCX, BMP, and others.
& Many plug-ins to handle more complex transformations
Tip #131 from
Rob The GIMP is a handy way to make and manipulate screenshots. Just choose Xtns,
Screen Shot, and then select whether you wish a full screen shot or just a window.
Click Grab, and you'll have your image ready for manipulation.
Seyon
Seyon is a complete full-featured telecommunications package for the X Window System (see
Figure 25.9).
The following are some of Seyon's features:
& A dialing directoryÐThe dialing directory supports an unlimited number of entries.
The directory is fully mouse-driven and features call-progress monitoring, dial
timeout, automatic redial, multi-number dialing, and a circular redial queue. Each
550 Chapter 25 Using the X Window System
item in the dialing directory can be configured with its own baud rate, bit mask, and script
file. The dialing directory uses a plain ASCII text phone book that you can edit from within
Seyon. Seyon also supports manual dialing.
& Terminal emulationÐTerminal emulation supports DEC vt102, Tektronix 4014,
and ANSI. Seyon delegates its terminal emulation to xterm, so all the familiar xterm
functionsÐsuch as the scroll-back buffer, cut-and-paste utility, and visual bellÐare
available through Seyon's terminal emulation window.
Figure 25.9
Although ac-
cessing the
Internet is im-
portant today,
many users still
need access to
bulletin boards
via their mod-
ems.
Using xterm also means that Seyon has a more complete emulation of vt102 than any
other UNIX or DOS telecommunications program. You also can use other terminal
emulation programs with Seyon to suit the user's need; for example, color_xterm can
be used to provide emulation for color ANSI (popular on many BBS systems), and xvt
can be used if memory is a bit tight.
& A scripting languageÐYou can use scripting language to automate tedious tasks such
as logging in to remote hosts. Seyon's script interpreter uses plain-text files and has a
syntax similar to that of sh, with a few extra additions. It supports many familiar
statements such as conditional branching by if...else and looping by goto. Scripts
can be assigned to items in the dialing directory for automatic execution after a
connection is made.
& A variety of download protocols, including Zmodem ÐSeyon supports an
unlimited number of slots for external file transfer protocols. Protocols are activated
from a mouse-driven transfer console that uses a plain ASCII text file, editable from
within Seyon, for protocol configuration. Seyon prompts the user for filenames only if
the chosen protocol requires filenames or if the transfer operation is an upload for
which Seyon also accepts wildcards. Multiple download directories can be specified for
the different transfer slots.
551
Troubleshooting
Seyon detects incoming Zmodem signatures and automatically activates a user-
specified Zmodem protocol to receive incoming files. Zmodem transfers are thus
completely automatic and require no user intervention.
& Various translation modesÐSeyon can perform useful translations with the user's
input. For example, Seyon can translate Backspace to Delete, a newline marker to a
carriage-return marker, and meta-key translation; that is, you can switch your Esc
meta key to the Alt key. The latter mode simulates the meta key on hosts that don't
support 8-bit-clean connections and makes possible the use of the meta key in
programs such as Emacs.
As for other features, Seyon allows you to interactively set program parameters, online help,
and software (XONN/XOFF) and hardware (RTS/CTS) flow control; capture a session to a
file; and temporarily run a local shell in the terminal emulation window.
Seyon is intended to be simple yet extensively configurable. Almost every aspect of Seyon can
be configured via the built-in resources to suit the user's taste.
xlock
Patrick J. Naughton ( [email protected]) wrote xlock and released it to the world. The
Part
xlock program locks the local X display until you enter your password at the keyboard. While
IV
xlock is running, all new server connections are refused, the screen saver is disabled, the
mouse cursor is turned off, the screen is blanked, and a changing pattern is put onscreen. If a Ch
key or a mouse button is pressed, you are prompted for the password of the user who started
25
xlock.
If you enter the correct password, the screen is unlocked, and the X server is restored. When
you're typing the password, Ctrl+Shift+u and Ctrl+Shift+h are active as kill and erase
commands, respectively. To return to the locked screen, click the small icon version of the
changing pattern.
Troubleshooting
The wrong window manager starts.
You have probably created configuration files that you didn't mean to. There are a lot of X
configuration files, and they all have to be consistent.
Depending on what window manager you want, there are a few solutions. First you should
run switchdesk to get your files back to a known state. If you don't want KDE or GNOME/
?
Enlightenment, choose AnotherLevel.
See ``Using the switchdesk tool for Red Hat 6.0'' p. 561
If you chose KDE or GNOME/Enlightenment, you should be done. AnotherLevel is actually
a collection of window managers. By default you will see fvwm95 when you login. If this isn't
552 Chapter 25 Using the X Window System
what you wanted, click the Start button and select Exit Fvwm, Switch To. Then select the window
manager that you want.
If the window manger you want isn't available on the Switch To menu, you can edit
~/.Xclients and replace whatever you find there with the full path name to the window
manager you want to use.
gdm isn't accepting XDMCP connections from other machines.
The version of gdm that ships with Red Hat 6.0 has a known bug in it that prevents it from
accepting XDMCP connections.
Use kdm or xdm by changing the /etc/X11/prefdm link as root. To select xdm, do this:
# ln -s /usr/X11/bin/xdm /etc/X11/prefdm
For kdm, do the following:
# ln -s /usr/bin/kdm /etc/X11/prefdm
When you restart your machine, your display manger will change and you will be able to
accept XDMCP connections.
The font server isn't accepting TCP connections.
In Red Hat 6.0, the font server, xfs, is configured to reject TCP connections by default. To
turn this back on, as root edit /etc/X11/fs/config and add the following line:
port = 7100
You can then restart xfs by going to the /etc/rc.d/init.d directory and typing the following:
./xfs restart
Project: Making Yourself at Home with fvwm2
fvwm2 is one of the most popular window managers available for Linux. It is also one of the
most configurable. Unfortunately, all the flexibility comes with a price: a very complex
configuration file. In this project, you learn some of the most useful configuration options in
fvwm2, and how you can use them to your advantage.
Choosing fvwm2
Red Hat 6.0 does not provide a convenient way to choose ``normal'' fvwm2. They do provide
a way to choose fvwm95, which is really just fvwm2 with a lot of customization to make it
look like Windows 95. Unfortunately, much of the power of fvwm2 is lost when you put this
interface on it.
To fix this, just override Red Hat's normal window manager selection routine by editing
~/.Xclients and replacing whatever is there with fvwm2. Log out and log back in. You
should then see a window manager that looks something like Figure 25.10. The root menu
shown in the figure is activated by pressing the left mouse button.
553
Project: Making Yourself at Home with fvwm2
Figure 25.10
The fvwm2 en-
vironment fo-
cuses on
functionality
rather than
fanciness.
Preparing to Customize fvwm2 Part
Before you can customize fvwm2, you need to create your own configuration file. Because IV
you'll want to start with the default configuration file, just copy it to your own directory:
Ch
$ cp /etc/X11/fvwm2/system.fvwm2rc ~/.fvwm2rc 25
If~/.fvwm2rc exists, then fvwm2 will use it. If not, it will default to /etc/X11/fvwm2/
system.fvwm2rc.
Look over the configuration file by editing ~/.fvwm2rc with any text editor, such as vi. The
full configuration file is too long to cover in detail here, but you can get complete
documentation by typing man fvwm2 at the command prompt.
Tip #132 from
Rob If you use Emacs to edit .fvwm2rc, you may want to put the following line at the top
of the file:
# -*- fvwm-generic-mode -*-
This will tell Emacs to color-code keywords in the configuration file, making it much
easier to edit.
Customizing the Menu
The first thing you will probably want to do is modify the root menu to include the programs
you will use most. The root menu supports simple and complex actions, as well as sub-menus,
titles and dividers.
554 Chapter 25 Using the X Window System
To edit the menu, look for the AddToMenu lines in .fvwm2rc. Perhaps you would like to put
Netscape on your Utilities menu before Emacs. The current menu looks like this:
AddToMenu Utilities ``Utilities'' Title
+ ``Top'' Exec exec xterm -T Top -n Top -e top
+ ``Calculator'' Exec exec xcalc
+ ``Xman'' Exec exec xman
+ ``Xmag'' Exec exec xmag
+ ``Editres'' Exec exec editres
+ ``'' Nop
+ ``Emacs'' Exec exec emacs
+ ``Mail'' MailFunction xmh ``-font fixed''
+ ``'' Nop
+ ``XLock'' Exec exec xlock -mode random
+ ``'' Nop
+ ``Reset X defaults'' Exec xrdb -load $HOME/.Xdefaults
Above the Emacs line, add the following:
+ ``Netscape'' Exec exec netscape
Here's what this line means:
+ This is a line-continuation character. It indicates that
this is still part of the previous AddToMenu.
"Netscape" This is the label to put on the menu.
Exec This is what fvwm2 should do when you select this
entry. Exec tells fvwm2 to execute the rest of this line as
if you typed it from the command prompt.
exec netscape This is the command to execute. The leading exec helps
save some memory by getting rid of the shell that fvwm2
creates. When you are calling programs from fvwm2,
you almost always want to use the construct Exec exec
program.
Save the file, left-click the desktop, and choose Exit fvwm. Then choose Restart fvwm2. This
will re-read your configuration file, which will now include Netscape in the menu.
Specifying Start-Up Programs
Some programs are so useful that you want them to execute every time you login. Perhaps
you like xeyes, shown in figure 25.11. xeyes is a cute program that displays a pair of eyes that
watch your mouse pointer as it moves around the screen.
To add this as a start-up program, you need to add it to the InitFunction. Edit .fvwm2rc and
look for InitFunction. It will look something like this:
AddToFunc InitFunction "I" Module FvwmButtons
+ "I" exec xsetroot -mod 2 2 -fg \#554055 -bg \#705070
The first line runs the button bar that you see at the bottom of the screen. The second line
sets the background color.
555
Project: Making Yourself at Home with fvwm2
Figure 25.11
xeyes is an old
amusement
program that
many people
still like to keep
around.
Adding your own line is very similar to what you did in the menus. Just add a line that looks Part
like this:
IV
+ "I" Exec exec xeyes
Ch
As in the previous example, the + indicates that this is still part of the previous AddToFunc. 25
The I indicates that this is to be done immediately, rather than waiting for the user to click
something. Then the Exec exec xeyes line is exactly as in the Netscape example and runs
xeyes.
Further Customizations
Complete customization of fvwm2 is beyond the scope of this section, but there is a lot more
to customize. fvwm2 gives you a lot of control over keyboard mappings, mouse behavior, and
what windows look like. For more information, type man fvwm2 at the command prompt.
CHAPTER
26
Working with KDE
In this chapter by Rob Napier
What Is KDE? 558
Installing KDE 560
Navigating KDE 562
Configuring KDE 570
The Pros and Cons of KDE 574
558 Chapter 26 Working with KDE
What Is KDE?
KDE stands for K Desktop Environment. KDE is a UNIX desktop, which means that it
provides a user-friendly GUI that can be integrated with applications. It is very similar to
what the Macintosh and Windows interfaces provide. It also provides a framework for
developers and a way for applications to communicate with each other. Figure 26.1 shows
KDE in action.
Figure 26.1
KDE provides
many of the
same features
that you find
on Macintosh
and Windows
machines and
many features
you don't.
From a user's point of view, KDE provides many features, such as the following:
& Files directly on the desktopÐIn KDE, just like a Macintosh or Windows machine,
you can drag files to the desktop and store them there as icons. Applications can be
automatically launched by links placed on the desktop as well.
& Session managementÐSession management means that applications can remember
the state they were in when you logged out and restart the application in exactly the
same state when you log back in. For example, you can restart in the same file you
were working on and where you were in the file.
& Desktop panelÐThe desktop panel provides quick access to applications by
providing the Application Starter (or K-Menu) and panel buttons to start your favorite
applications. You can also swallow applications, which means that what looks like a
button can actually be a running application. This capability is very useful for
programs such as xbiff, which checks for new mail.
559
What Is KDE?
& TaskbarÐThe taskbar shows all the currently available windows, which allows you to
quickly switch between applications, even if they're minimized.
& PagerÐA pager allows you to maintain several desktops and easily switch between
them. For example, you might put your mail program on one desktop, all your
programming tools on another, and a game of Solitaire on a third.
KDE is made up of many parts, only a few of which are readily apparent to you. Figure 26.2
shows how the various layers of KDE interact, and Table 26.1 describes the layers.
Figure 26.2
Applications
OpenParts
Here you can
see the devel-
KDE Libraries/Qt KOM
oper's view of
X11 CORBA
KDE.
Linux Libraries
Table 26.1 KDE Components
Component Purpose
Linux Libraries Low-level routines that are outside KDE. They manage hardware and
Linux system interactions.
X11 Low-level routines that are outside KDE. They manage the graphics
display and hardware. See Chapter 24, ``Installing the X Windows
System,'' for more information.
KDE Libraries/Qt Wrappers around the X11 graphics system. Qt in particular is a set of
C++ classes that simplify GUI development. It is produced by Troll
Tech AS and was once the center of a very large debate in the KDE Part
community (see the following sidebar).
IV
CORBA Stands for Common Object Request Broker Architecture. An industry
standard way for applications to communicate with each other. Ch
KOM Stands for KDE Object Model. Wrappers around CORBA that provide 26
additional features. KOM provides similar functionality to Micro-
soft's COM.
OpenParts Object classes or servers that provide functionality to other object
classes or clients. For example, an application can include an image
viewer OpenPart to display its images. Then the application doesn't
have to worry about what format the image is in. OpenParts provides
similar functionality to Microsoft's ActiveX.
Applications KDE applications that are instances of a KApplication object. If you
implement all KDE applications with the same object class, all KDE
applications automatically conform to KDE standards.
If all this information seems complicated, don't worry too much about it. Users seldom have
to deal with OpenParts, CORBA, or X11. Knowing what these words mean is useful so that
you can understand the documentation and so that you can understand how KDE buzzwords
map to similar buzzwords in the Windows and Macintosh worlds. For example, KDE's
Qt/KOM/OpenParts provides a very similar framework to Microsoft's MFC/COM/ActiveX.
560 Chapter 26 Working with KDE
The Qt Controversy
When KDE was started, the developers wanted to use an existing cross-platform GUI toolkit to provide widgets.
Widgets are GUI components such as minimize buttons or window frames. Because the KDE project developers
wanted to deliver a product quickly, they didn't want to take a lot of time to develop their own toolkit when a
good one already existed. They chose what they felt was the best freely available toolkit on the market: Qt by
Troll Tech.
Troll Tech (http://www.troll.no) is a Norwegian company that began developing Qt in 1992. It is a very
good product, runs on many platforms, is freely available, and at 80,000 lines of code, would have taken quite
awhile to reproduce. It does not, however, use the same license as KDE, the GNU General Public License, or GPL
(see Appendix C, ``The GNU General Public License''). Although Qt does distribute its source code, users are not
permitted to make modifications to it and redistribute it. They must send all modifications back to Troll Tech to
be incorporated into the base product. GPL allows users to create derivative products and redistribute them.
The other major difference is that users can purchase a license to redistribute Qt in commercial products
without distributing their source code. GPL products and GPL-derived products must always provide source
code.
This small difference of licensing led to one of the great ``holy wars'' of the Linux community. Those who believe
that the end users should always have the right to modify and redistribute code claimed that the entire KDE
Project could not be considered Open Source. This was a great rallying cry for supporters of GNOME. GNOME is
another Linux desktop that is licensed entirely under GPL. (See Chapter 27, ``Working with GNOME,'' for more
information about GNOME.)
One group attempted to bridge the gap with the Harmony Project (http://harmony.ruhr.de). The Harmony
Project was originally designed to create an LGPL version of Qt. LGPL is the Library (or Lesser) General Public
License, which is very close to GPL, and is acceptable to most camps, including the Open Source Initiative. The
developers of Harmony are not KDE developers. The KDE developers are too familiar with Qt to create a clone
without violating copyright.
Much of this information is now irrelevant. Troll Tech released a free version of Qt 2.0 under the Qt Public
License, or QPL (http://www.troll.no/qpl), which has been endorsed by leaders in the community as
meeting the criteria of Open Source. Although QPL is not precisely the same as GPL, it is close enough to
resolve the issue for the vast majority of the community.
Harmony isn't dead, though. The Harmony Project has moved beyond just cloning Qt and now is trying to make
a true competitor with many new and exciting features.
From the user's point of view, the most important parts of KDE are kwm, kpanel, and kfm.
kwm, the KDE Window Manager, provides all the normal window manager functionality,
such as displaying, minimizing, moving, and resizing windows. kpanel provides a convenient
panel, usually at the bottom of the screen. It has quick-launch buttons, menus, a clock, and
much more. kpanel also provides the taskbar, usually at the top of the screen. It shows all the
currently open windows and allows you to switch between them easily. kfm is the KDE File
Manager, which not only provides access to your files, but also acts as a Web and document
browser.
Installing KDE
Caldera OpenLinux provides KDE as its default desktop, so getting KDE up and running is
as easy as installing the distribution. Red Hat defaults to GNOME, so you need to make
some modifications.
Installing KDE for Red Hat 6.0
Table 26.2 lists the KDE packages for Red Hat 6.0.
561
Installing KDE
Table 26.2 KDE Packages
RPM Purpose
qt (Required) Qt toolkit.
kdesupport (Recommended) Various non-KDE libraries that KDE requires.
kdelibs (Required) KDE shared libraries.
kdebase (Required) The core of KDE, such as the window manager.
kdegames (Optional) Several games.
kdegraphics (Optional) Various programs to view and draw graphics.
kdeutils (Optional) Calculator, editor, and so on.
kdemultimedia (Optional) CD player and other multimedia applications.
kdenetwork (Optional) Internet applications including a mail reader and a news reader.
kdeadmin (Optional) System administration programs.
kdetoys (Optional) Fun programs that aren't games. For example, kmoon puts the
phase of the moon on your panel.
korganizer (Optional) Personal organizer.
?
The required and recommended packages should be installed in the order listed using rpm.
See ``Installing Packages with RPM,'' p. 169
The version of KDE that ships with Red Hat 6.0 is 1.1.1Ðpre2. It is a pre-release beta
Part
of version 1.1.1, and you should upgrade to the released version to avoid known
bugs. You can get updated RPMs from http://updates.redhat.com. IV
Ch
26
Using the switchdesk Tool for Red Hat 6.0
The easiest way to switch to KDE is to use the switchdesk tool. switchdesk may not be
installed by default, so you might have to install the switchdesk-1.7.0-1 package by using
?
rpm.
See ``Installing Packages with RPM,'' p. 169
After you've installed switchdesk, launching it is simple:
$ switchdesk
This command produces a dialog box asking whether you prefer GNOME, KDE, or
AnotherLevel. Click KDE, choose OK, and then log out. When you log back in, your default
desktop will be KDE.
562 Chapter 26 Working with KDE
Choosing KDE Without switchdesk
If you're using a distribution without switchdesk, you can make KDE your default desktop
by modifying your $HOME/.Xclients file to contain the following:
?
exec startkde
?
See ``Using vi,'' p. 207
See ``Choosing a Window Manager,'' p. 536
Navigating KDE
If you're familiar with other GUIs, such as the Macintosh interface or Microsoft Windows,
KDE should be quite familiar. Figure 26.3 shows the major components of the desktop.
Figure 26.3
KDE's desktop
brings your
most-used
functions to
your fingertips.
The Panel
The panel is probably the most important part of the KDE desktop. The default settings give
plenty of functionality, but the real power is in configuring the panel. Figure 26.4 shows each
of the default buttons.
Application Starter (K-Menu)
The Application Starter, which is also called the K-Menu, holds links to all your programs
and configuration menus. It is similar to the Windows 95 Start menu. Press Alt+F1 to open
the Application Starter.
563
Navigating KDE
Figure 26.4
Using the KDE
panel well can
greatly improve
your productiv-
ity.
Home Directory
The Home Directory launches the file manager on your home directory. For more
information about the file manager, see ``KDE File Manager'' later in this chapter.
KDE Control Center
Using the Control Center, you can configure various parts of KDE, including the window
manager, file manager, keyboard, sound, and much more. The Control Center is also the
?
place you can choose themes, which define a complete look and feel.
See ``Themes,'' p. 537
Part
Find Utility
IV
KDE has a powerful tool for finding files. You can search by name pattern, modification time, Ch
file type, size, or contents. This tool is modeled very closely on Microsoft Windows 98's Find 26
utility.
Help Browser
If you're a new KDE user, the Help Browser can be a great asset. It is essentially a simple
Web browser. In fact, the help files are written in HTML, and hyperlinks in the help system
may point to Web pages on the Internet. Links outside the help system are not displayed in
?
the same window, though. The Help Browser can launch kfm to handle those requests.
See ``KDE File Manager,'' p. 568
Window List
The Window List provides instant access to all your current windows, no matter what
desktop they are running on. Being able to access these windows can be a real time-saver
when you have many windows open across numerous desktops. You can also get the Window
List by clicking the middle mouse button on the desktop.
564 Chapter 26 Working with KDE
Logout
Logout logs you out of KDE and returns to the login screen. Before logging out, it warns you
of any non-KDE applications that are currently running. Most KDE applications allow
session management, which means that when you restart KDE, they will return to exactly
where they left off. Non-KDE applications can't restart this way, so KDE lists those
applications that will lose unsaved data and won't be resumed at the same point. KDE still
generally restarts the applications when you log back in.
Lock Screen
KDE provides various screensavers that activate after a configurable time of inactivity. If you
want to start the screensaver immediately, you can press the Lock button. To unlock the
screen, you need to enter your login password. The screensaver is configurable through the
Control Center (see the previous section ``KDE Control Center''). Be careful with the Lock
button because it is very close to the Logout button. This layout is traditional for the
Common Desktop Environment (CDE), so if you are used to Sun or HP workstations, it
should be familiar.
Virtual Desktops
Virtual desktops really differentiate the Linux desktops from the Windows and Macintosh
environments. Each of the four buttons in this block provides you with a completely different
desktop, each displaying different windows. When screen real estate is scarce, having separate
desktops can be very convenient.
Generally, you put similar applications together so that you can easily move between desktops
as you move between tasks. A common layout is xterminals on one desktop, a Web browser
on another, mail on a third, and diversions (games) on a fourth. Other users choose to put all
their programming tools on one desktop, documentation on a second, and mail on a third. It's
up to you. KDE remembers what programs are running on which desktops when you log out,
so when you log in, everything restarts in the right place.
Virtual desktops are highly configurable. To configure your desktops, right-click the panel in
an area with no buttons, and then click Configure to bring up the KPanel Configuration
dialog box (see Figure 26.5). Click the Desktops tab to configure your desktop.
Using the numbered blocks in this dialog, you can change the names of the desktops. You
should name them meaningfully so that you can remember what is on each desktop. The
Visible slider sets the number of virtual desktops available. You can have two, four, six, or
eight desktops. Four is the default. The Width slider sets the width of the virtual desktop
buttons on the panel. You should set the buttons wide enough to see the names you have
chosen but narrow enough not to take unneeded space on the panel. KDE remembers your
desktop names between sessions.
565
Navigating KDE
Figure 26.5
You can con-
figure the KDE
virtual desk-
tops to match
the way you
want to work.
Tip #133 from
Rob If you want to change only the name of a desktop, you don't need to use the KPanel
Configuration dialog box. Just click the desktop button to go to that desktop, and then
click it again. You then are presented with a standard I-beam text cursor to change the
name.
You can move windows between virtual desktops by clicking the menu button (the far left
button on the title bar), clicking To Desktop, and then clicking the desktop that the window
Part
should live on. The shortcut for the menu button is Alt+F3.
IV
Windows can also be made sticky. Sticky windows appear on all desktops. To make a window
sticky, click the thumbtack button on the left side of the title bar. Clicking the thumbtack Ch
again makes the window stay only on the current desktop. 26
Clock
The clock appears on the far right side of the panel. You can configure the clock to display
24-hour time or Internet time. Internet time is a new time standard proposed by Swatch, a
Swiss watchmaker. It has 1,000 beats per day measured from midnight in Biel, Switzerland,
which happens to be the place where the Swatch company headquarters are located and is
UTC+1 (that is, Internet time is one time zone off the rest of the world). For example, @500"
Internet time is noon in Biel, or 7:00 a.m. in New York. It is very unlikely that Internet time
will become a serious time standard, but the KDE clock can display it if you like.
To configure the clock, right-click the panel in an empty area. Next, click Configure, and
then click the Options tab. The two options for the clock are Clock Shows Time in AM/PM
Format and Clock Shows Time in Internet Beats. If you choose Internet Beats, the AM/PM
selection is ignored.
566 Chapter 26 Working with KDE
Panel Hide
On either side of the panel is a textured tab with an arrow on it. When you click one of these
buttons, the panel slides off the screen in the direction indicated. This capability is handy if
?
you want more of the screen available for working.
See ``Getting More Desktop,'' p. 573
Tip #134 from
Rob You can use the KDE panel with non-KDE-compliant window managers (like fvwm2).
You won't get all the integration that you get with KDE, but you will get a useful
application launcher. From the command line, run the following:
kpanel -no-KDE-compliant-window-manager
The Taskbar
The taskbar is generally at the top of the screen and shows all the windows on all the
desktops. Names of minimized windows are in parentheses, and windows on different
desktops are separated by a small break.
When you click a button in the taskbar, you are taken to the appropriate desktop, and the
window receives focus. The button appears pressed to indicate the currently active window.
Clicking the active button minimizes the window.
Right-clicking a button provides a local menu with various options. Table 26.3 lists these
options and what they do. Where two options are listed, only one of them is available at a
time.
Table 26.3 Taskbar Local Menu Options
Option Purpose
Maximize/Restore Maximize increases the size of the window to full screen. If the
panel and taskbar are visible, the window does not overlap them.
Restore restores the window to its previous size.
Iconify/Deiconify Iconify removes the window from the screen so that it is accessible
only from the taskbar. This action is the same as minimizing.
Deiconify restores the window to its previous size.
Sticky/Unsticky Sticky makes a window appear on all desktops. Unsticky turns off
this feature.
Onto current desktop This option is available for buttons that represent windows on
other desktops. Clicking it brings that window to the current
desktop.
Iconify other windows This option iconifies all the windows on this desktop except for
the selected one. This capability is useful for when you are dealing
with just one window and want to clear your desktop of everything
else.
567
Navigating KDE
Table 26.3 Taskbar Local Menu Options
Option Purpose
Close Close sends a close request to the program. This request generally
does not kill a hung program.
? See ``Getting More Desktop,'' p. 573
Tip #135 from
Rob You can maximize, iconify, or close windows by clicking the buttons in the upper right
corner of the window just as you would expect. You can also maximize a window only
horizontally or vertically. Click the maximize button with the middle mouse button to
vertically maximize a window. Click the maximize button with the right mouse button
to horizontally maximize a window.
The Root Menu
To get the Root menu, right-click the background. A menu then pops up with various
options. These options are explained in Table 26.4.
Table 26.4 KDE Root Menu Options
Option Purpose
Part
New New creates various types of objects on the desktop. For example, you IV
can create a new folder or a shortcut to a program or a link to a Web
page. Ch
Bookmarks Any file can be bookmarked and then can be easily accessed from this 26
menu. For more information about bookmarks, see ``Bookmarks''
later in this chapter.
Help on desktop This option opens the Help Browser.
Execute Command This option pops up a dialog box that accepts one-line UNIX
commands. This capability can be very convenient when you want to
start programs that don't have links to them yet. You can also enter
URLs to quickly view a Web page. Finally, to get man pages on
UNIX commands, type man:command, where command is the command
for which you want information. The shortcut for Execute Command
is Alt+F2.
Display Properties Display Properties customizes the appearance of KDE. See ``Making
It Pretty'' later in this chapter for more information.
Refresh Desktop KDE can become confused and fail to draw something correctly. For
instance, a dying program may not properly erase itself from the
screen. If this happens, select Refresh Desktop, and KDE redraws all
the windows to make sure they are correct.
568 Chapter 26 Working with KDE
Table 26.4 KDE Root Menu Options
Option Purpose
Unclutter Windows If you get a lot of windows overlapping each other, seeing what you
want to see might be difficult. Unclutter Windows tries to make sure
that all the windows are minimally overlapping.
Cascade Windows Similar to Unclutter Windows, this option stacks all the open
windows so that you can find them all.
Arrange Icons Arrange Icons lines up all the icons on the desktop onto the left side
of the screen.
Lock Screen This option is the same as the Lock Screen button described earlier in
this chapter.
Logout This option is the same as the Logout button described earlier in this
chapter.
Trash
Like most GUIs designed since the original Macintosh, KDE provides a trash can for
throwing away files, folders, and links. With the KDE trash can, like other trash cans, you
empty it by right-clicking the Trash icon and clicking Empty Trash Bin. Clicking the Trash
opens the file manager and allows you to drag things back out.
The Trash icon changes depending on whether things are currently in the Trash.
Templates
The Templates folder contains templates of kdelnk files. KDE uses kdelnk files to point to
applications, printers, mountable devices, and Internet resources. Bookmarks are imple-
?
mented this way.
?
See ``Bookmarks,'' p. 571
See ``Templates,'' p. 570
Autostart
Anything in the Autostart folder is run every time KDE starts up. Generally, this folder
contains links to applications, but you can put anything in the Autostart folder, including
documents and folders. Documents are opened with their default application, and folders are
opened with kfm. Generally, it's best to put links to what you want rather than the file itself.
KDE File Manager
kfm is much more than just a file manager. It is also a fairly full-featured Web browser that
understands a wide variety of UNIX file formats including archive formats such as tar and
zip (see Figure 26.6).
569
Navigating KDE
Figure 26.6
The KDE File
Manager
makes it easy
to manage
your files.
The different icons in Figure 26.6 indicate different kinds of files. The file folders are
directories. Folders with locks on them are directories you don't have permission to view.
The icons with paint cans are graphics images. The plug icons are UNIX domain sockets
(you will almost never have to deal with them directly, but kfm understands them). kfm can
also detect executables, text files, HTML files, backups, videos, scripts, and much more.
Particularly useful is kfm's handling of archive file formats such as tar and zip. kfm treats
most archives just like directories, so you can click them and view their contents directly.
Some archives are passed to Archiver, which also allows you to view the contents easily.
Other file types are handled as you would expect: Images are passed to kview, text files are
passed to kedit, and executables are executed.
Part
Tip #136 from IV
Rob KDE treats FTP sites just like local file systems. For example, if you open ftp://
[email protected]/~bob, kfm requests bob's password and then displays the Ch
directory as if it were local. For anonymous FTP, leave off the bob@.
26
kfm also reads Web pages. It doesn't do all the tricks of Netscape, but that is often considered
a benefit. Just enter the URL on the Location line, and then use kfm as you would any other
Web browser.
kfm handles cookies better than the mainstream browsers. By default, kfm asks permission
before accepting cookies. It then allows you to choose how to handle cookies on a
domain-by-domain basis. To set up this feature, either answer the questions the first time
you visit a site, or click Options, Configure Browser. Then click the Cookies tab and add
your domain specific settings.
kfm handles frames, but many Web sites don't send it frame information because many Web
sites decide whether browsers can handle frames by checking their User-Agent String.
Browsers use this string to identify themselves. Many Web sites check only for Netscape
570 Chapter 26 Working with KDE
Navigator, which identifies itself as Mozilla, or Internet Explorer. kfm identifies itself as
Konqueror, which very few Web sites expect. To trick sites into behaving properly, you might
need to tell kfm to lie about who it is. You can do so by clicking Options, Configure Browser.
Then click the User Agent tab and add entries for each server you want to lie to.
"Konqueror'' comes from the following sequence: Navigator, Explorer, Conqueror.
Configuring KDE
KDE is extremely configurable, but only in specific ways. Much like Windows 98, KDE can
be configured only to perform tasks that the developers thought you might like to do.
Although KDE is still more configurable than Windows 98, tasks as simple as removing the
clock from the panel (or replacing it with your own) can be very difficult or impossible. The
developers thought that everyone would want a simple digital clock with the date on the right
?
side of the panel.
See ``The Pros and Cons of KDE,'' p. 574
Templates
The Templates folder on your desktop contains various kdelnk files that you can use to create
new objects on your desktop or in folders. All the entries in the Root menu's New menu come
from this folder (except Folder, which is built in). To create a new instance of the template,
right-click the background to pull up the Root menu. Then click New and the item you want
to create. The following sections describe items that can be created.
Folder
Folders are directories and can hold other files or folders. Don't be afraid to use spaces or
punctuation other than / (the slash) in the folder names. In KDE, / is a special character and
cannot be put into a folder name.
File System Device
Using the File System Device, you can mount devices such as CDs or floppies directly from
the desktop. Before you can create this kind of link, you need to make sure you are able to
?
mount the device.
See ``Mounting and Unmounting File Systems,'' p. 444
When you are able to mount the device normally, create a File System Device by using the
New menu. Then right-click the new icon and click Properties. On the General tab, change
the name to something meaningful like Floppy.kdelnk or cdrom.kdelnk. On the Device tab,
enter the device name (such as /dev/fd0 or /dev/cdrom) as the Device. Then select mounted
571
Configuring KDE
and unmounted icons. Icons of floppy disks and CDs with and without small green lights beside
them are available for your convenience.
URLs
URLs includes FTP URLs, Internet addresses (URLs), and World Wide Web URLs. The
only difference between them is their icons. When you create a new URL link, just enter the
desired URL in the dialog that pops up. Now you have a shortcut to your favorite sites.
MIME Type
MIMEstands for Multipurpose Internet Mail Extensions. Originally, it was intended to encode
files in email messages, but it has moved on to handle a wide variety of tasks. KDE uses
normal MIME mechanisms to identify and handle files. If KDE applications need new
MIME types, the applications install them. Generally, you set up your own MIME types to
associate a type of file with a non-KDE application. This topic is beyond the scope of this
book, but you can find a very good tutorial for it in the KDE help pages. Point your Help
Browser tofile:/usr/share/doc/HTML/en/kfm/app.html and file:/usr/share/doc/HTML/
en/kfm/mime.html.
Bookmarks
Any file in KDE can have a bookmark pointing to it. From within kfm, right-click the file or
folder and select Add to Bookmarks. A bookmark to the file or folder is then placed in your
Bookmarks menus, both on the Root menu and in kfm.
To edit or remove bookmarks, click Bookmarks on the kfm toolbar, and then choose Edit
Part
Bookmarks. When you do so, you open another kfm window looking at your bookmarks. All
normal kfm functions work here. To give the bookmark a more convenient name, right-click IV
the bookmark and click Properties. Then change the filename to whatever you would like it
Ch
to be. Changing the name doesn't affect the file the bookmark points to in any way.
26
Tip #137 from
Rob Remember that KDE doesn't care if a file is local or remote. If you view a web page
using kfm, you can right-click the background of the web page and select Add to
Bookmarks. The same is true of files on an FTP server or any other network resource.
Making It Pretty
You can customize KDE in many ways to make your desktop more attractive. Few of these
changes will boost your performance, but they may make you happier while you're working.
All the settings described in the following sections are found on the KDE Control Center
under Desktop.
572 Chapter 26 Working with KDE
Background
One of the most personal things about a desktop is the background. Some users like simple,
flat backgrounds, whereas others like textures or even images. KDE supports all of them.
Each desktop can have its own background, so you need to select which desktop you're
working on. Having different backgrounds can be very useful in helping you determine which
desktop you're on at any time.
The One Color option does exactly what it sounds like. You select the color bar to open a
color chooser. The Two Color option allows you to blend between two colors or create
various patterns based on two colors. You can find the two colors used on the color bar under
One Color and the color bar under Two Color.
Wallpaper is also fairly obvious if you're used to other desktops. KDE has a wide selection of
arrangement types such as tiled, centered warp, and symmetrical tiled. Just experiment with
them to see which works best with your image. KDE also can modify your background as you
work. To use this feature, select Random, and after a set time, KDE switches images based on
a list that you provide (or just all images in a directory). This feature can really break up the
monotony. It can also be distracting, depending on your mood.
The Dock Into the Panel selection puts an icon on the right side of your panel that you can
use to quickly bring up this dialog box. If you're using the Random selection just described,
clicking this icon cycles you to the next background without waiting for the timeout.
Tip #138 from
Rob It is possible to use a screensaver as your background. This takes up some CPU
resources, but can look quite good on a fast machine. Just run the screensaver with
the-inroot parameter. For example, from the command prompt, type the following:
$ kmatrix.kss -inroot
For a list of all the available screensavers, type the following:
ls ${KDEDIR}/bin/*.kss
Colors
If you just want another color scheme, KDE provides a wide variety. If you like changing the
colors, check out the section ``Themes'' later in this chapter.
Fonts
Using the Fonts feature, you can set your default fonts, which will be honored by most KDE
apps and some other apps. One of the best uses of this feature is to set a larger font size for
really high-resolution screens on small monitors.
573
Configuring KDE
Screensaver
What modern desktop would be complete without 20 or so screensavers? If you are a science
fiction movie buff, check out the Matrix screen saver by Jamie Zawinski ( [email protected]).
Using Themes in KDE
For an explanation of what themes are, see ``Themes'' in Chapter 25, ``Using the X Window
System.'' This section describes how to get them and how to set them up.
KDE comes with four themes, which are not nearly enough for a real theme lover. So, you
can start by going to the source of KDE themes: http://kde.themes.org. At this time, KDE
has 80 themes available. Just select the theme you want from the gallery and download it. It is
saved as a tar.gz file. Theoretically, themes are supposed to use a new .themerc file format,
but I've never seen this format except in the stock themes. Generally, you need to use
kinstall by Dmitry Zakharov ([email protected]) from http://kde.themes.org/ar-
chive/ktinstall.tar.gz. Once you have your theme file and kinstall installed, type the
following at the command line:
$ kinstall theme
theme is the name of your theme file (including the tar.gz file). You don't even have to restart
KDE. Your theme comes up immediately.
Tip #139 from
Rob The themes available are of variable quality. kde.themes.org rates themes from
72 to +2. From the theme gallery, select the lowest rating that you wish to see, and
Part
you will remove a lot of the less interesting themes. If you prefer to make your own
decisions, though, set the filter down to 72 and you can see everything. You can also IV
submit and read comments on the themes.
Ch
26
Getting More Desktop
KDE offers many options for maximizing your available screen real estate. If you're interested
in reorganizing your desktop, try some of the following:
& Make your panel smallerÐFrom the Panel tab of the KPanel Configuration dialog,
click Tiny.
& Move the panel off the screen by clicking the left or right hide buttons ÐKDE
puts the Application Starter, Window List, and a disk navigator on your taskbar
whenever you move panels off the screen.
574 Chapter 26 Working with KDE
& Autohide the panel and the taskbarÐFrom the Options tab of the KPanel
Configuration dialog, click Auto Hide Panel and Auto Hide Taskbar. Choosing these
options causes the panel and taskbar to hide until you touch the mouse cursor to their
border. For instance, when you move the mouse cursor to the bottom of the screen,
the panel reappears.
& Get rid of the taskbar altogetherÐYou don't really need the taskbar because the
Window List provides similar functionality. From the Panel tab of the KPanel
Configuration dialog, select Hidden.
The Pros and Cons of KDE
Should you use KDE? Should you use GNOME? Should you avoid them both and use a
conventional Linux window manager? The choice is up to you, but the following sections
describe some of the advantages and disadvantages of KDE, particularly in relation to
GNOME.
Resources
Determining exactly how much memory one desktop uses versus another can be very difficult
given how shared libraries can skew the numbers, but in general terms, KDE uses about the
same memory as GNOME, possibly a bit more in their default configurations. Both use three
to four times the memory of AfterStep, which is certainly not the smallest of the window
managers. Note that GNOME has the potential to use a lot more memory if all of
Enlightenment's features are turned on.
KDE should run in as little as 16M of memory, but you will generally need 24M-32M of
memory for good performance. With very complex themes and many running applications,
you may require as much as 64M-128M of memory.
Performance
KDE's performance is usually quite good. On reasonably high-end hardware, detecting
serious differences between the performance of one window manager over another can be
difficult. In general, however, KDE is faster than GNOME and slower than simple window
managers like fvwm2.
Configuration
KDE is quite configurable, but not nearly as configurable as GNOME. This is a problem and
a benefit. The problem is that you cannot configure anything for which the programmers did
not decide to provide a hook. On the other hand, GNOME is so configurable that it can be
quite confusing, especially for new users.
575
The Pros and Cons of KDE
Integration
The parts of KDE are highly integrated. This is again a benefit and a problem. The benefit is
that the parts tend to work together very well and in very reliable ways. The problem is that
the parts of KDE sometimes know so much about each other that it is hard to replace any part
of KDE with an improved version.
Stability
KDE is quite stable and seldom crashes on reasonable hardware. Most of the tools work
together, although some of the documentation has broken links in the 1.1.1 distribution.
Final Analysis
Essentially, KDE is a strong, highly integrated desktop environment in the tradition of
Windows 98. It is more stable and configurable than Windows 98, but the design
philosophies are quite similar. If you're looking for a more full-featured, stable version of the
Windows interface, KDE is a great choice.
If, on the other hand, you're interested in a much more configurable but also much more
chaotic environment, read on in Chapter 27, ``Working with GNOME."
Part
IV
Ch
26
CHAPTER
27
Working with GNOME
In this chapter by Rob Napier
What Is GNOME? 578
Installing GNOME 580
Navigating GNOME 582
Configuring GNOME 587
The Pros and Cons of GNOME 592
578 Chapter 27 Working with GNOME
What Is GNOME?
GNOME stands for the GNU Network Object Model Environment. (For more information on
the GNU project, see Appendix C, ``The GNU General Public License.'') GNOME intends
to be a full-featured, user-friendly desktop built entirely of Open Source components. Here,
Open Source is a very specific definition, given by the Open Source Initiative at http://
www.opensource.org/osd.html. This definition is important because it formed the reasoning
behind creating GNOME. Had KDE been fully Open Source, GNOME may never have
?
been created (though some people may debate this point).
See ``The Qt Controversy,'' p. 560
Figure 27.1 shows GNOME in action, but it is somewhat misleading. What Figure 27.1
really shows is the Enlightenment window manager, panel, the GNOME Midnight
Commander File Manager, gnome-help-browser, gnomepager_applet, and
gnome_util_applet. The last two are running on the panel.
Figure 27.1
GNOME is
more of a fed-
eration of like-
minded com-
ponents than a
single inte-
grated system
like KDE.
Although KDE also implements its parts as many separate programs, for GNOME it is a
deep philosophical (almost religious) issue. For instance, whereas the KDE's panel knows all
about its pager, the GNOME pager is just an applet that runs on the panel. If you don't want
a pager, you can just not run it (virtual desktops still work without it). If someone develops a
better pager, you can swap it in. You can move the pager around on the panel if you prefer it
somewhere else. You can even run two pagers. The same is true for the clock on the far right.
At least five different clocks come with GNOME on a Red Hat 6.0 system. It's all completely
configurable. These capabilities are both a blessing and a curse.
From a user's point of view, the GNOME environment provides many features, such as the
following:
579
What Is GNOME?
& Files directly on the desktopÐWith GNOME, just like a Macintosh or Windows
machine, you can drag files to the desktop and store them there as icons. Applications
can be automatically launched by links placed on the desktop as well.
& Session managementÐSession management means that applications can remember
the state they were in when you logged out and restart the application in exactly the
same state when you log back in. For example, you can restart in the same file you
were working on and where you were in the file.
& Desktop panelÐThe desktop panel provides quick access to applications by
providing the Main Menu button and panel buttons to start your favorite applications.
You can also swallow applications, which means that what looks like a button can
actually be a running application. This capability is used extensively by GNOME to
handle pagers, clocks, load monitors, and much more.
& PagerÐA pager allows you to maintain several desktops and easily switch between
them. For example, you might put your mail program on one desktop, all your
programming tools on another, and a game of Solitaire on a third.
Like KDE, GNOME is based on CORBA, the Common Object Request Broker
Architecture. It is an industry standard way for applications to communicate with each other.
The GNOME implementation of CORBA is called ORBit. KDE is generally programmed in
C++, but GNOME provides bindings for a wide variety of languages. Bindings are interfaces
that allow a programmer to communicate with a system in a particular programming
language. Currently, GNOME has bindings for C, C++, Objective C, TOM, and Guile.
GNOME is also heavily dependent on Gtk+, the GIMP toolkit. Gtk+ also has a wide variety
of language bindings, including the ubiquitous C, C++, and Perl, but it also binds to unusual
languages such as Eiffel, Dylan, Pike, and Haskell, among others. This means that if you want
to create a small GUI script in Perl, you can use Gtk+ to get attractive widgets that conform to
your current theme. A theme defines the appearance of various parts of your desktop, such as
?
backgrounds and textures.
See ``The GIMP,'' p. 548
Part
IV
The GNOME Manifesto
Ch
The GNOME Manifesto was written to explain the philosophy of GNOME. Unlike the more pragmatic KDE,
GNOME is as much a way of doing things as a set of programs. The following is the GNOME Manifesto, which 27
you can find at http://www.gnome.org/about/manifesto.shtml.
What Is GNOME?
GNOME is an Open Source desktop environment built from components that meet the Open Source guidelines
in full.
No Compromises
GNOME is open in the full sense of the word. It seeks to impose only that order necessary for consistency.
Window Manager
GNOME defines a set of ``hints'' for window managers. If you use a GNOME-aware window manager, it will
cooperate nicely with GNOME. If you don't, then GNOME works just fine. The ``hint'' interface is published in full
for anyone to use.
580 Chapter 27 Working with GNOME
No religionÐpick any window manager.
Commercial Use
GNOME is the key to the desktop. Its authors recognize that it is not appropriate to ``control'' that interface or
require that a commercial vendor pays some third party for the ability to write GNOME-compliant applications.
All the core GNOME software is distributed under the GNU Library General Public License, a license that permits
the software to be used so long as it is dynamically linked or the user can relink it to new versions of the
libraries. This is the same license used by the Linux C libraries.
You need to purchase no expensive software licenses to make your commercial application GNOME compliant.
Vendor Neutral
No component of the interface is controlled solely by one company or restricted from modification. Any
organization or person, however large or small, can contribute to GNOME. Furthermore, if you don't happen to
agree with a decision, the license enshrines rights to distribute modified versions.
Truly open. No core component is non Open Source.
Language Bindings
GNOME enforces no programming language restrictions on a developer. The core libraries are written in C, so
they are fast and efficient. The external interface is currently available in C, C++, Objective C, TOM, and Guile.
If you want to add bindings for another language we will be delighted to help.
Themes
The low-level GNOME/Gtk interface is currently being extended to support themes. A mechanism is already
used in Enlightenment and Windowmaker window managers to allow a user to freely control the look and feel
of the base interface components without forcing the original program author to do the work.
User-driven look and feel.
Portability
The toolkit and libraries used in GNOME are intended to be portable to all UNIX-like platforms, and if people
contribute code beyond. GNOME seeks to avoid ties with any platform-specific interfaces where possible, and
when not possible to provide code for all platforms.
Multiple operating systems.
Installing GNOME
Red Hat 6.0 now provides GNOME as the default desktop, so you don't need to do anything
to it. This version is even relatively stable, unlike the rest of Red Hat 6.0's X and KDE
packages.
For other distributions (including Red Hat 5.2 and earlier), GNOME installation can be
quite complicated. Rather than try to give all the permutations here, I recommend that you
go to the Getting GNOME page at http://www.gnome.org/start/getting.shtml. If you
have Caldera OpenLinux, check out the FAQ at http://www.gnome.org/gnomefaq/html/
gettingotherrpm.html. This FAQ entry is somewhat out of date but should still work.
GNOME seems to have become more and more focused on Red Hat during its life. Much of
this focus is due to Red Hat Advanced Development Labs (RHAD) providing so much
support for GNOME. In fact, the Enlightenment window manager is primarily developed by
Mandrake ( [email protected]) and Rasterman ([email protected]). Until recently,
Rasterman was a member of RHAD (he now works with Mandrake at VA Research).
581
Installing GNOME
With GNOME, as with most Open Source projects, the only way that things happen is for
people to choose to work on them. Because Red Hat works a lot on GNOME, GNOME
tends to install easily on Red Hat. Because Caldera focuses on KDE (and does not even ship
GNOME with OpenLinux 2.2), installing it on OpenLinux systems is not very easy. I don't
mean to imply a bias on the part of the GNOME team. I simply mean that if no one works on
something, then it doesn't get done. The power of Open Source is that if anyone chooses to
make good OpenLinux RPMs for GNOME, that person is absolutely free to do so (and even
encouraged to do so).
Over 60 packages are required or recommended for GNOME, all with their own versions.
Whereas all KDE packages are released together and are numbered with the version of KDE
(1.1.1, for instance), each GNOME package is independent, and new versions may be
released whenever necessary. For example, gnome-core may be at version 1.0.7, whereas
enlightenment may be at version 0.15.5.
Luckily, a set of packages collectively called GNOME 1.0 is considered fairly stable. You can
find these packages at ftp://ftp.gnome.org/pub/GNOME/gnome-1.0.Packages are available for
Debian, Red Hat, Slackware, and S.u.S.E. For other distributions, you can pull down the
sources and compile them.
Using the switchdesk Tool for Red Hat 6.0
If you have switched to another desktop or window manager, the easiest way to switch back to
GNOME is to use the switchdesk tool. switchdesk may not be installed by default, so you
?
might have to install the switchdesk-1.7.0-1 package by using rpm.
See ``Installing Packages with RPM,'' p. 169
After you've installed switchdesk, launching it is simple:
$ switchdesk
This command produces a dialog box asking whether you prefer GNOME, KDE, or
AnotherLevel. Select GNOME, choose OK, and then log out. When you log back in, your
Part
default desktop will be GNOME.
IV
Choosing GNOME Without switchdesk Ch
If you're using a distribution without switchdesk, you can make GNOME your default
27
desktop by modifying your $HOME/.Xclients file to contain the following:
?
exec gnome-session
?
See ``Using vi,'' p. 207
See ``Choosing a Window Manager,'' p. 536
582 Chapter 27 Working with GNOME
Navigating GNOME
If you're familiar with other GUIs, such as the Macintosh interface or Microsoft Windows,
GNOME should be quite familiar. Figure 27.2 shows the major components of the desktop.
Figure 27.2
GNOME's
desktop is in-
credibly config-
urable and
personal.
The Panel
The panel is probably the most important part of the GNOME desktop. The default settings
give plenty of functionality, but the real power lies in configuring the panel (see Figure 27.3).
Figure 27.3
The GNOME
panel holds all
your most-
used features.
Main Menu Button
The Main Menu holds links to all your programs and configuration menus. It is similar to the
Windows 95/98 Start menu but adds a few more options. For instance, if you right-click a
menu option, you can add a launcher for that option to your panel or to your personal menu.
You can also drag and drop menu items to the panel or to the desktop to make shortcuts.
No shortcut key is available to bring up the Main Menu like there is in KDE. This lack of a
shortcut is mostly due to how GNOME works. The panel itself is just an application, so it
can't watch for keystrokes. The window manager would have to do that and then send that
information on to the panel. But Enlightenment doesn't allow you to add that kind of
accelerator key easily and doesn't specifically know about the panel in any case. This problem
583
Navigating GNOME
is surmountable (and may be addressed in future versions of Enlightenment), but it is indicative of
the issues in a loosely integrated system like GNOME.
Integrated Help System
The help system has all the online GNOME documentation, including a great deal of
documentation for GNOME applications. Although the GNOME Help Browser does use
URLs to locate files, it is not a Web browser and does not understand http addresses.
Tip #140 from
Rob GNOME's Integrated Help System does not include help for Enlightenment. To get
help for Enlightenment, click the middle mouse button on the desktop and click Help.
Enlightenment also provides ToolTips. By pointing the mouse pointer to a window
button or frame, Enlightenment will display help text about that item.
Control Center
Much like KDE's Control Center, GNOME's Control Center allows you to configure much
of GNOME's behavior. To configure Enlightenment's behavior, you need to go to the
Window Manager section and run Enlightenment's configuration tool.
Terminal and Netscape
The Terminal and Netscape buttons are normal launch buttons that run a terminal emulator
and Netscape, respectively. You can add many more buttons if you like.
Pager
Virtual screens and desktops really differentiate the Linux desktops from the Windows and
Macintosh environments. Each of the four areas in the pager provides you with a completely
different virtual screen, each displaying different windows. When screen real estate is scarce,
Part
having these different areas can be very convenient. Inside each block are small
IV
representations of the windows currently displayed on that virtual screen. Enlightenment also
allows you to have virtual desktops, each of which can have virtual screens. The pager shows Ch
these virtual desktops as additional blocks of screens. Enlightenment can handle up to 32
27
desktops of up to 8 68 (64) screens per desktop for a total of 2,048 screens. You should never
need even a fraction of this number. The default is 2 62 virtual screens and one virtual
desktop. For the remainder of this chapter, I will use the term desktop to mean both virtual
screen and virtual desktop unless the distinction is important.
Generally, you put like applications together so that you can easily move between desktops as
you move between tasks. A common layout is xterminals on one screen, a Web browser on
another, mail on a third, and diversions (games) on a fourth. Other users choose to put all
their programming tools on one desktop, documentation on a second, and mail on a third. It's
584 Chapter 27 Working with GNOME
up to you. GNOME remembers what programs are running on which desktops when you log out
if you request it, so when you log in, everything restarts in the right place.
Configuring virtual desktops can be somewhat confusing because this feature is shared by the
pager and the window manager. To change how the pager displays these desktops (for
example, to make the pager smaller or to turn off the task list), click the ? button on the panel.
To configure the number of virtual screens and desktops, their backgrounds, and other
?
options, you need to use the Enlightenment configuration tool described previously.
See ``Making It Pretty,'' p. 590
You can move windows between virtual desktops by clicking their menu button (usually the
far left button on the title bar), clicking Desktop, and then clicking the direction the window
should be moved (right, left, above, below).
Tip #141 from
Rob To move a window between distant virtual desktops, you can make the window sticky
by clicking Stick/Unstick, moving the desired desktop, and then clicking Stick/Unstick
again. This action leaves the window on the new desktop.
Task List
Technically part of the pager, the task list shows buttons for each window on the current
desktop. Visible windows are shown with a GNOME foot icon beside them. The active
window has yellow rays coming out of the GNOME foot icon. Hidden windows have a small
box icon beside them. You can access a full list of all windows on all desktops by clicking the
arrow icon between the pager and the task list.
Clock
You can configure the default GNOME clock to show 12-hour or 24-hour time and display
the date if you request it. The clock can also display UNIX time, which is the number of
seconds since the epoch (00:00:00 UTC, January 1, 1970). It is unlikely that you will want to
display UNIX time.
Panel Hide
On either side of the panel is a tab with an arrow on it. When you click one of these buttons,
the panel slides off the screen in the direction indicated. This feature is handy if you want
?
more of the screen available for working.
See ``Getting More Desktop,'' p. 573
585
Navigating GNOME
The Root Menu
To open the Root menu, right-click the desktop. A menu then pops up with various options.
These options are explained in Table 27.1.
Table 27.1 GNOME Root Menu Options
Option Purpose
New Using this option is a quick way to start various programs
or create folders on the desktop. Terminal starts gnome-
terminal, Directory allows you to create a new folder, and
Launcher lets you create a new program launcher. The
GIMP and Gnumeric spreadsheets start these programs.
Because the New menu is completely customizable, you can
use it to replace the Main Menu.
Arrange Icons This option orders the icons on your desktop on the left
side of your screen.
Create New Window This option launches Midnight Commander, the GNOME
File Manager.
Re-create Desktop Shortcuts If you accidentally remove any of the default desktop
shortcuts, this option puts them back for you.
Rescan Desktop If the GNOME desktop ever gets confused, or if you
manually reconfigure it behind its back, this option gets all
your desktop icons back where they should be.
Configure Background Image This option is a shortcut to the Control Center's Back-
ground section. See ``Making It Pretty'' later in this chapter
for more information on setting background images.
Setting images is not always as simple as it sounds.
The Enlightenment Menu
You can access the Enlightenment menu by pressing the middle mouse button on the
background. The Enlightenment menu is independent of the Root menu because Part
Enlightenment can be run without GNOME. If you don't want to use the panel, you can IV
configure the Enlightenment menu to have everything you need on it. Table 27.2 details the
Ch
options from the Enlightenment menu.
27
Table 27.2 Enlightenment Menu Options
Option Purpose
Gnome Apps This menu is the same as the System section of the
GNOME Main Menu.
Other Programs These menus list other preloaded programs that aren't on
the Gnome Apps menu.
586 Chapter 27 Working with GNOME
Table 27.2 Enlightenment Menu Options
Option Purpose
Desktop This menu provides various desktop functions. Cleanup
Desktop attempts to unclutter the desktop by moving
overlapping windows apart. Goto Next Desktop moves one
desktop (not screen) down the stack. Goto Previous
Desktop moves one desktop up the stack. FX - Ripples
really shows what Enlightenment can do. It makes the
bottom of your screen look like a pool of water, reflecting
everything above it with ripples. On a fast machine, it can
be quite pretty.
Themes All installed Enlightenment themes are listed here, and you
can quickly switch between them. See ``Themes'' later in
this chapter for more information.
Enlightenment Configuration This option launches the Enlightenment Configuration
dialog.
About Enlightenment This option shows version and contact information for
Enlightenment.
Help This option brings up the Enlightenment Help Browser
(which is separate from the GNOME Help Browser).
Restart Enlightenment If you have reconfigured Enlightenment manually, this
option lets you restart it to pick up the new configurations.
Midnight Commander File Manager
GNOME comes with the Midnight Commander File Manager (see Figure 27.4). Although it
is not a full Web browser like kfm, it does offer a number of additional benefits.
Figure 27.4
Midnight Com-
mander is an
extremely flex-
ible file man-
ager for power
users.
Using the Icons, Brief, Detailed, and Custom buttons across the top of the window, you can
quickly switch between different views of the files in your directory. For power users who
often want to see the file details, this capability can be a great help. You can configure the
Custom button by using the Edit, Preferences dialog.
587
Configuring GNOME
Other options in the Edit, Preferences dialog that are of particular interest to power users are
the Show Hidden Files and Use Shell Patterns Instead of Regular Expressions options.
Turning off the latter is very nice if you like to have the full power of regular expressions
when you're searching.
For remote sites, Midnight Commander handles FTP-style URLs, just like kfm.
For powerful filtering capabilities, click Commands, Run Command in Panel to set up
complex filters using find or any other UNIX tool that creates a list of filenames. One of the
default expressions is ``Find SUID and SGID Programs'' for the security-conscious. SUID
and SGID programs generally run with unrestricted privileges, making them a potential
security hole. The ``Find SUID and SGID Programs'' expression automatically runs the
following command:
find . \(\(-perm -04000 -a -perm +011 \) \
-o \(-perm -02000 -a -perm +01 \) \) -print
?
The results are then displayed in the file browser, just as if they were all in a directory.
See ``Searching for Files,'' p. 432
Tear-Off Menus
In Figure 27.4, notice the small tabs on the left side of the three menu bars at the top of the
window. They are tear-off menus, which means you can shuffle them around by dragging
them up or down. You can even tear them off the window entirely and place them on the
desktop if you prefer. Most GNOME applications support tear-off menus.
Configuring GNOME
GNOME is incredibly configurable. Almost every part of GNOME can be reconfigured,
replaced, or removed. If you're comfortable with GNOME, environments like KDE,
Macintosh, or Windows 95/98 are extremely limited and confining. The other side of this
flexibility is that GNOME users are expected to take care of themselves and know how all the Part
parts of GNOME interact with each other. Sometimes two different components try to
IV
configure the same thing (such as the background), and it's up to you to establish who
Ch
controls what.
27
You can handle many configuration options in two or three unrelated (and possibly
conflicting) ways. Often you can choose a GNOME or Gtk+ way or an Enlightenment way.
The following sections focus on the most convenient and useful ways to configure your
system, which for GNOME users is usually the GNOME or Gtk+ way (but not always).
Making GNOME Useful
GNOME is very useful right out of the box. By reconfiguring it, however, you can make it
more compatible with the way you work. For example, GNOME allows you to create
shortcuts to the tools and files you use most. It also allows you to put tools right on your
588 Chapter 27 Working with GNOME
panel so they are always available. This section focuses on how to improve your productivity by
better configuring GNOME.
Putting Applications on the Main Menu
Putting options on the Main Menu is usually the first thing you want to do. To get started,
click Main Menu, Settings, Menu Editor to open the Menu Editor shown in Figure 27.5. If,
for some reason, this option isn't on the menu yet, you can run the Menu Editor directly by
clicking Main Menu, Run Program, and then entering gmenu.
Figure 27.5
The Menu Edi-
tor is extremely
visual and in-
tuitive.
The Menu Editor should be fairly straightforward for most users. Unless you are running as
root, you can put things only on the User menu because the System menu is used by all users.
In the Menu Editor, select User Menus and click New Item to open an untitled application.
Enter the name you want to call it, a comment for the pop-up hints, and the command to run.
Type is usually an application, although if you want to put a directory in your menu, you can
choose Directory. Midnight Commander shows you the directory when you select this item.
The Run in Terminal check is handy if the program is text-based rather than a GUI. Finally,
choose an icon and click Save. For the vast majority of programs, that's all you need to do.
Tip #142 from
Rob The Menu Editor supports drag and drop just like most GNOME applications. To move
elements around, just drag them to where you want them. If you want to create a
launcher on the desktop or the panel, you can just drag an entry there.
589
Configuring GNOME
Putting New Launchers on the Panel
After you've put a menu entry on the Main Menu, adding it to the panel is easy. Just open the
Main Menu, find the launcher you want, and drag it to the panel. You can then use the middle
mouse button to move the launcher around. Right-click the launcher to modify it or
remove it.
Putting Applets on the Panel
Much of the power of the panel is in applets. The clock is an applet, but many more are
available. For instance, you can add a CD player, load monitor, mail monitor, and even a
Slashdot headline viewer.
Tip #143 from
Rob Slashdot (http://slashdot.org) is the premier site for up-to-the-minute high-tech
information. Its slogan is ``News for Nerds. Stuff that matters.''
http://freshmeat.net for the latest software
Other important Linux sites are
updates, and http://userfriendly.org for influential Linux humor.
To put a new applet on your panel, click Main Menu, Panel, Add Applet, and then choose the
applet you want. After it's on the panel, you can use the middle mouse button to move it
around. The right mouse button pulls up a menu so that you can set the applet's properties or
remove it from the panel.
Getting More Panel Room
After you've added all the elements that you're sure to want, you might find that your panel is
overloaded. What you need is more panels. In GNOME, the panel is just an application, so
having more than one is not a problem. You can reasonably have up to four panels, though
GNOME won't stop you from creating more. To create a new panel, right-click the Hide
button and click Add New Panel. Then click Edge Panel or Corner Panel. Part
Corner panels start in a corner and extend horizontally or vertically for as far as they need to.
IV
Edge panels cover an entire horizontal or vertical edge, regardless of what is on them. You Ch
can convert each panel to corner or edge by right-clicking the Hide button and clicking
27
Convert to Corner Panel (or Edge Panel).
Making a Drawer
Often it is convenient to create a subpanel to contain related tools. To create such a subpanel,
right-click the Hide button and click Add Drawer. By doing so, you create a drawer on the
panel that you can move around with the middle mouse button. You can then add launchers
and applets to it just like a panel. You can even add drawers to drawers.
590 Chapter 27 Working with GNOME
Making It Pretty
Enlightenment is arguably the prettiest (or at least fanciest) window manager ever developed.
The look and feel are fantastically configurable. Sometimes configuring can be a bit
confusing, so the following sections focus on the most common changes.
Setting the Background
Setting the background should be very easy, but GNOME doesn't always do what you might
expect. The problem is that GNOME can set the background, and Enlightenment can set the
background. If both of them try to set the background, then they will likely fight about it, and
your background may change suddenly when you're not expecting a change.
The secret is to pick one of them to handle the background and tell the other one to leave it
alone. The Red Hat 6.0 default is to let GNOME set the background and to turn off
Enlightenment's background. This approach is good if you want the same background for
every desktop, but if you want different backgrounds for each desktop (not screen), you need
?
to let Enlightenment handle the job.
See ``Pager,'' p. 583
To use GNOME's background properties, follow these steps:
1. Go to the Control Center and click Background.
2. From there, select whether you want a solid, gradient, or image background.
Choose OK.
3. Click Window Manager, Enlightenment to bring up the Enlightenment Configura-
tion Editor.
4. Click Backgrounds and No Background to let GNOME do all the background
handling.
To use Enlightenment's background properties, follow these steps:
1. Go to the Control Center and click Background.
2. From there, click Disable Background Selection and choose OK.
3. Click Window Manger, Enlightenment to bring up the Enlightenment Configuration
Editor.
4. Click Backgrounds.
5. From the list, select the desktop that you want to work on.
6. Select the background from the pictures to the right. If you don't like any of them, you
can click Add New. This option allows you to use your own images as backdrops.
Modifying Window Behavior
In Enlightenment, windows can behave in numerous ways when they are created, resized, or
moved. To configure this behavior, go to the Enlightenment Configuration Editor by
choosing Control Center, Window Manager, Enlightenment.
591
Configuring GNOME
To change how windows move, click Basic Options. Your options are Opaque, Lined, Box,
Shaded, Semi-Solid, and Translucent. Experimenting with these options is worthwhile, but
keep these points in mind: Opaque is a good default; Lined and Box work very well on slow
machines; and Shaded, Semi-Solid, and Translucent are extremely slow unless you have very
good hardware.
You also handle window resizing in the Basic Options dialog. The same options exist as for
moving, except that you can't choose Translucent.
In the Basic Options dialog, you also can set how window focus works. Your choices are
?
Mouse, Sloppy, and Click.
See ``Getting Focus,'' p. 530
From the Enlightenment Configuration Editor, click Audio if you want to turn on sounds.
From here, you can turn on Enlightenment sound events. Most things that happen in
Enlightenment can generate sounds. If you like sounds, enjoy. Sometimes, though, these
sounds can become quite annoying after a very short time.
Screensavers
GNOME uses xscreensaver, which was first developed by Jamie Zawinski ( [email protected]) in
late 1991. Given that long history, it is not surprising that GNOME ships with over 70
different screensavers. To choose one, go to the Control Center and click Screensaver.
Browse through the list until you find one you like.
Tip #144 from
Rob If you feel you need more screensavers, you can look at http://www.jwz.org/
xscreensaver and pick up the latest version. Several new screensavers have been
added since the version that ships with Red Hat 6.0.
Part
Themes IV
As mentioned previously, themes allow you to define a common appearance for your Ch
environment. For example, themes allow you to define how your windows and buttons look
27
and what color your background is.
Both Gtk+ and Enlightenment have themes, but they handle different features. Gtk+'s themes
generally handle things within windows, such as the window background or how buttons are
drawn. Enlightenment's themes generally handle features outside windows such as their
frames and widgets. To get a complete look, you need to use both.
592 Chapter 27 Working with GNOME
Gtk+ Themes
Gtk+ comes with 12 themesÐwhich is pretty good, but there's always room for more. You
can start by going to http://gtk.themes.org. At this time, 66 themes are available for Gtk+.
You can select the theme you want from the gallery and download it. It is saved as a tar.gz
file. Then you can go to the Control Center and click Theme Selector. Next, click Install
New Theme and select your theme file. Finally, select the theme from the list and click OK.
The theme takes effect immediately.
Enlightenment Themes
Enlightenment comes with seven themes, so you'll probably want more if you like themes.
The best place to get Enlightenment themes is http://e.themes.org. At this time,
Enlightenment has 129 themes available. You can select the theme you want from the gallery
and download it. It is saved as an .etheme file (which is really just a tar.gz file). Put this file in
your .enlightenment/themes/ directory (create this directory if it doesn't exist yet).
Now open the Enlightenment Configuration tool (by choosing Control Center, Window
Manager, Enlightenment), click Themes, and select your theme. Select Apply, and you're
done.
Getting More Desktop
GNOME offers many options for maximizing your available screen real estate. If you're
interested in reorganizing your desktop, try some of the following:
& Move the panel off the screen by clicking the left or right hide buttons.
& Autohide the panel. To do so, right-click the Hide button and click This Panel
Properties. Then click Auto Hide. Choosing these options causes the panel to hide
until you touch the mouse pointer to their border. For instance, when you move the
mouse cursor to the bottom of the screen, the panel reappears.
& Convert your panel to a corner panel. By doing so, you can give yourself some more
room on the side of the panel. To convert the panel, right-click the Hide button and
click Convert to Corner Panel.
The Pros and Cons of GNOME
Should you use GNOME? Should you use KDE? Should you avoid them both and use a
conventional Linux window manager? The choice is up to you, but the following sections
describe some of the advantages and disadvantages of GNOME, particularly in relation to
KDE. You also have the option of running Enlightenment without GNOME, which has
some advantages and disadvantages of its own.
593
The Pros and Cons of GNOME
Resources
Determining exactly how much memory one desktop uses versus another can be very difficult
given how shared libraries can skew the numbers, but in general terms, GNOME uses about
the same memory as KDE, possibly a bit less in its default configuration. Both use three to
four times the memory of AfterStep, which is certainly not the smallest of the window
managers. Enlightenment without GNOME obviously uses quite a bit less memory than with
GNOME, though it is still larger than window managers like AfterStep. If all of
Enlightenment's features are turned on, GNOME can take quite a bit more memory than
KDE. GNOME claims to run in 16M of memory, but you will generally need closer to 64M
of memory to get good performance.
Performance
GNOME's performance, particularly with several of Enlightenment's features turned on, can
be a problem. With many of the fancy features turned off, though, GNOME's performance is
pretty good. It is still generally a bit slower than KDE and can be quite a bit slower than other
window managers like AfterStep. In particular, you may find that windows and menus take
longer to appear with GNOME. This is especially true if you are displaying your desktop
across a network.
Configuration
GNOME is incredibly configurable. This is a benefit and a problem. If you like to have
complete control over what your environment looks like and how it works, GNOME and
Enlightenment are definitely for you. With this flexibility, however, comes a lot of complexity
and some very confusing conflicts. Some features don't always work together unless you know
how to configure them properly. If you want an environment that is stable, even if it is also
more rigid, KDE may be for you.
Using Enlightenment without GNOME still gives you several configuration options without
all the conflicts. Enlightenment is just a window manager, so you lose the advantages of
having a desktop, such as being able to put shortcuts directly on the desktop and manage Part
sessions.
IV
Ch
Integration
GNOME is more like a federation than an integration. Many of the parts of GNOME work
27
just fine by themselves. This aspect is wonderful if you want to swap out individual
components, but it also means that the components don't always work together well. When
they do work, though, they work better than anything else.
Stability
Enlightenment in particular is only mostly stable. In general, a GNOME/Enlightenment
environment will crash more than KDE. Although you can often just restart the part that
crashed (even Enlightenment can often be restarted without losing your running programs),
594 Chapter 27 Working with GNOME
it can be quite frustrating. If you are used to a Windows 95 environment, GNOME is still much
more stable (particularly because crashes do not bring down the whole system), but as a whole,
GNOME is only somewhat stable in Linux terms.
Final Words
GNOME is the ideal desktop for people who like the cutting edge. GNOME and
Enlightenment are breaking new ground in how desktops work. In many ways, Enlight-
enment is the first truly innovative windowing interface since NeXT. GNOME is pushing
the idea of really open software in directions it's probably never gone before. If you're a
technophile, GNOME is a wonderful collection of parts that you can use to build your own
environment. Just browse through e.themes.org to see how differently users have expressed
themselves. If you like to tinker with your machine, GNOME provides no end of
opportunities.
On the other hand, if you don't see your desktop as an important form of self-expression,
then you probably won't be very comfortable in GNOME or Enlightenment and should
seriously consider KDE. If you administer other people's machines, you might find
GNOME's endless variations a nightmare to support. Once again, KDE is probably a better
choice. The power of Linux is that it is your choice to make.
PART
V
Network Administration
28 Understanding the TCP/IP Protocol Suite 597
29 Configuring a TCP/IP Network 619
30 IP Firewalling and Masquerading 635
31 Connecting to the Internet 663
CHAPTER
28
Understanding the TCP/IP
Protocol Suite
In this chapter by Steve Burnett
The History of TCP/IP 598
Internet Terminology 598
The Open Systems Interconnection Model 600
The TCP/IP Protocol Stack 603
IP Addresses 604
Subnetworks and Subnet Masks 608
Routing 610
Internet Network Setup 611
Troubleshooting 617
598 Chapter 28 Understanding the TCP/IP Protocol Suite
The History of TCP/IP
The suite of widely used protocols known as Transmission Control Protocol/Internet
Protocol (TCP/IP) has become critical as networks of all sizes, including the Internet, depend
on it for their communications.
TCP/IP has grown from its initial development as a government-sponsored project to
widespread use, connecting networks of all sizes. Recognized for its capability to enable
communication among dissimilar machines, it's found on virtually all workstations,
minicomputers, and mainframes.
In the late 1960s, the U.S. Department of Defense (DOD) recognized an electronic
communication problem developing within the department. Communicating the ever-
increasing volume of electronic information among DOD staff, research labs, universities,
and contractors had hit a major obstacle. The various entities had computer systems from
different computer manufacturers, running different operating systems, and using different
networking topologies and protocols. How could information be shared?
The Advanced Research Projects Agency (ARPA) was assigned to resolve the problem of
dealing with different networking equipment and topologies. ARPA formed an alliance with
universities and computer manufacturers to develop communication standards. This alliance
specified and built a four-node network that's the foundation of today's Internet. During the
1970s, this network migrated to a new, core protocol design that became the basis for
TCP/IP.
The mention of TCP/IP requires a brief introduction to the Internet, a huge network of
networks that allows computers all over the world to communicate. The Internet grows at
such a phenomenal rate that any estimate of the number of computers and users on the
Internet would be out of date by the time this book went to print! Nodes include universities,
major corporations, research labs in the United States and abroad, schools, businesses both
large and small, and individually owned computers. The explosion in past years of the World
Wide Web has driven the Internet's expansion. In addition, the Internet is also a repository
for millions of shareware programs, news on any topic, public forums and information
exchanges, and email. Another feature is remote login to any computer system on the network
by using the Telnet protocol. Because of the number of systems that are interconnected,
massive computer resources can be shared, enabling large programs to be executed on remote
systems. Massively distributed processing projects such as the 1997 decryption of the Data
Encryption Standard are possible only with the ``everything is connected to everything else''
behavior of the Internet.
Internet Terminology
The Internet Protocol suite is composed of many related protocols based on the foundation
formed by TCP and IP. To clarify the relationship of these components, Table 28.1 provides
some definitions and notations.
599
Internet Terminology
Table 28.1 Networking Terms
Term Definition
datagram A unit of information that's exchanged; this term is used interchangeably with
the words data packet and network message.
DNS (Domain Name Service) A service provided by one or more computers in a
network to help locate a path to a desired node. This service saves every system
on a network from having to keep a list of every system it wants to talk to. DNS
is used by mail gateways.
GOSIP (Government Open Systems Interconnection Profile) A collection of OSI
protocols used in U.S. government computer networks and projects.
Internet A computer network based on TCP/IP and related protocols. It is a public
network of networks interconnecting businesses, universities, government
facilities, and research centers.
FTAM (File Transfer, Access, and Management) A file transfer and management
protocol as specified by OSI.
FTP (File Transfer Protocol) A protocol that enables file transfer between systems.
IP (Internet Protocol) A protocol responsible for transporting datagrams across
the Internet. The current standard version of IP is IP version 4, or IPv4.
IPv6 (Internet Protocol version 6) Also referred to as Ipng, this evolutionary upgrade
of Ipv4 is not discussed in this chapter.
NFS (Network File System) A network virtual disk system that enables a client
computer to mount remote file systems and directories. It was originally
developed by Sun Microsystems.
NIC (Network Information Center) A group responsible for administering Internet
and TCP/IP addresses, as well as network names.
node A computer on a network.
OSI (Open System Interconnection) The ISO standard model for defining data
communication.
RFC (Request For Comments) The documentation maintained by NIC relating to
Internet protocols, addressing, routing, configuration, and other related
Internet topics.
RIP (Routing Information Protocol) A protocol used to exchange information
between routers.
RMON (Remote monitor) A remote network monitor that enables the collection of
information about network traffic.
Part
RPC (Remote Procedure Call) A type of call that enables procedures to be executed
on a server.
V
SMTP (Simple Mail Transfer Protocol) A protocol used to transfer electronic mail
Ch
between systems.
SNMP (Simple Network Management Protocol) A protocol used to manage remote
28
network devices and to collect information from remote devices related to
configuration, errors, and alarms.
600 Chapter 28 Understanding the TCP/IP Protocol Suite
Table 28.1 Networking Terms
Term Definition
TCP (Transmission Control Protocol) The protocol between a pair of applications
responsible for reliable, connection-oriented data transmission.
Telnet The protocol used to establish remote terminal connections.
UDP (User Datagram Protocol) A connectionless protocol used to transfer data
between agents.
VT (Virtual terminal) A method for using Telnet to log in to remote systems
through the network.
The Open Systems Interconnection Model
Many different types of computers are used today, varying in operating systems, CPUs,
network interfaces, and many other qualities. These differences make the problem of
communication between diverse computer systems important. In 1977, the International
Standards Organization (ISO) created a subcommittee to develop data communication
standards to promote multivendor interoperability. The result is the Open Systems
Interconnection (OSI) model.
The OSI model doesn't specify any communication standards or protocols; instead, it
provides guidelines that communication tasks follow.
It's important to understand that the OSI model is simply a modelÐa frameworkÐthat
specifies the functions to be performed. The model doesn't detail how these functions
are performed. ISO, however, does certify specific protocols that meet OSI standards
for parts of the OSI model. For example, the CCITT X.25 protocol is accepted by ISO as
an implementation that provides most of the services of the Network layer of the OSI
model.
To simplify matters, the ISO subcommittees took the divide-and-conquer approach. Because
the complex communication process is divided into smaller subtasks, the problem becomes
more manageable, and each subtask can be optimized individually. The OSI model is divided
into seven layers:
& Application
& Presentation
& Session
& Transport
& Network
601
The Open Systems Interconnection Model
& Data Link
& Physical
Tip #145 from
Steve One easy way to remember the order of the layers (from the top down) is
mnemonically, by making a sentence from the first letters of the layer names: All
People Seem To Need Data Processing.
Each layer is assigned a specific set of functions. Each layer uses the services of the layer
beneath it and provides services to the layer above it. For example, the Network layer uses
services from the Data Link layer and provides network-related services to the Transport
layer. Table 28.2 explains the services offered at each layer.
The concept of a layer making use of services and providing services to its adjacent
layers is simple. Consider how a company operates: An assistant provides services to
the president (the next layer up) to write a memo. The assistant uses the services of a
messenger (the next layer down) to deliver the message. Because these services are
separated, the assistant (application) doesn't have to know how the message is
actually carried to its recipient. The assistant merely has to ask the messenger
(network) to deliver it. Just as many assistants can send memos in this way by using a
standard messenger service, a layered network can send packets by handing them to
the network layer for delivery.
Table 28.2 Services Provided at Each OSI Layer
Layer Description
Physical (Layer 1) This layer provides the physical connection between a computer
system and the network. It specifies connector and pin assign-
ments, voltage levels, and so on.
Data Link (Layer 2) This layer ``packages'' and ``unpackages'' data for transmission. It
forms the information into frames. A frame represents the exact
structure of the data physically transmitted across the wire or
Part
other medium.
V
Network (Layer 3) This layer provides routing of data through the network.
Ch
Transport (Layer 4) This layer provides sequencing and acknowledgment of trans-
mission.
28
Session (Layer 5) This layer establishes and terminates communication links.
Presentation (Layer 6) This layer converts data and ensures that data is exchanged in a
universal format.
602 Chapter 28 Understanding the TCP/IP Protocol Suite
Table 28.2 Services Provided at Each OSI Layer
Layer Description
Application (Layer 7) This layer provides an interface to the application that a user
executesÐa ``gateway'' between user applications and the network
communication process.
Don't confuse the Application layer with application programs you execute on the
computer. Remember that the Application layer is part of the OSI model that doesn't
specify how the interface between a user and the communication pathway happens;
an application program is a specific implementation of this interface. A real
application typically performs Application, Session, and Presentation layer services
and leaves Transport, Network, Data Link, and Physical layer services to the network
operating system.
Each layer communicates with its peer in other computers. For example, layer 3 in one
system communicates with layer 3 in another computer system.
When information is passed from one layer down to the next, a header is added to the data to
indicate where the information is coming from and going to. The header-plus-data block of
information from one layer becomes the data for the next. For example, when layer 4 passes
data to layer 3, layer 4 adds its own header. When layer 3 passes the information to layer 2,
layer 3 considers the header-plus-data from layer 4 as data and adds its own header before
passing that combination down.
In each layer, the information units are given different names (see Table 28.3). Therefore, by
knowing the terms used to reference the data, you know which layer of the model is being
discussed.
Table 28.3 Terms Used by OSI Layers to Refer to Information Units
OSI Layer Information Unit Name
Application Message
Transport Segment
Network Datagram
Data Link Frame (also called packet)
Physical Bit
Before the advent of the OSI model, the U.S. Department of Defense defined its own
networking model, known as the DOD model. The DOD model is closely related to the TCP/
IP suite of protocols, as explained in the following section.
603
The TCP/IP Protocol Stack
The TCP/IP Protocol Stack
The TCP/IP protocol stack represents a network architecture that's similar to the ISO OSI
networking model. Figure 28.1 shows the mapping of TCP/IP layers onto the ISO protocol
stack.
OSI INTERNET
Figure 28.1
OSI and TCP/IP
APPLICATION
compared.
TELNET NFS
FTP SNMP
PRESENTATION SMTP DNS
SESSION
TCP UDP
TRANSPORT
NETWORK IP
DATA LINK
PHYSICAL
TCP/IP doesn't make as fine distinctions between the top layers of the protocol stack as does
OSI. The top three OSI layers are roughly equivalent to the Internet process protocols. Some
examples of process protocols are Telnet, FTP, SMTP, NFS, SNMP, and DNS.
The Transport layer of the OSI model is responsible for reliable data delivery. In the Internet
protocol stack, this layer corresponds to the host-to-host protocols. Examples are TCP and
UDP. TCP is used to translate variable-length messages from upper-layer protocols and
provides the necessary acknowledgment and connection-oriented flow control between
remote systems.
UDP is similar to TCP, except that it's not connection-oriented and doesn't acknowledge
data receipt. UDP only receives messages and passes them along to the upper-level protocols.
Because UDP doesn't have any of the overhead related to TCP, it provides a much more
efficient interface for such actions as remote disk services.
The Internet Protocol (IP) is responsible for connectionless communications between
systems. It maps onto the OSI model as part of the Network layer, which is responsible for Part
moving information around the network. This communication is accomplished by examining
V
the Network layer address, which determines the systems and the path to send the message.
Ch
IP provides the same functionality as the Network layer and helps get the messages between
28
systems, but it doesn't guarantee the delivery of these messages. IP may also fragment the
messages into chunks and then reassemble them at the destination. Each fragment may take a
different network path between systems. If the fragments arrive out of order, IP reassembles
the packets into the correct sequence at the destination.
604 Chapter 28 Understanding the TCP/IP Protocol Suite
IP Addresses
The Internet Protocol requires that an address be assigned to every device on the network.
This address, known as the IP address, is organized as a series of four octets. These octets each
define a unique address, with part of the address representing a network (and optionally a
subnetwork) and another part representing a particular node on the network.
Several addresses have special meanings on the Internet:
& An address starting with a zero references the local node within its current network.
For example, 0.0.0.23 references workstation 23 on the current network. Address
0.0.0.0 references the current workstation.
& The loopback address, 127, is important in troubleshooting and network diagnoses.
The network address 127.0.0.0 is the local loopback inside a workstation.
& The ALL address is represented by turning on all bits, giving a value of 255.
Therefore, 192.18.255.255 sends a message to all nodes on network 192.18; similarly,
255.255.255.255 sends a message to every node on the Internet. These addresses are
important to use for multicast messages and service announcements.
When you assign node numbers to your workstations, you should not use 0, 127, or
255 because they are reserved numbers and have special meanings.
IP Address Classes
The IP addresses are assigned in ranges referred to as classes, depending on the application
and the size of an organization. The three most common classes are A, B, and C. These three
classes represent the number of locally assignable bits available for the local network. Table
28.4 shows the relationships among the different address classes, the available number of
nodes, and the initial address settings.
Table 28.4 IP Address Classes
Class Available Nodes Initial Bits Starting Address
24
A 2 =167,772 0xxx 0-127
16
B 2 =65,536 10xx 128-191
8
C 2 =256 110x 192-223
D 1110 224-239
E 1111 240-255
605
IP Addresses
Class A addresses are used for very large networks or collections of related networks. Class B
addresses are used for large networks having more than 256 nodes (but fewer than 65,536
nodes). Class C addresses are used by most organizations. It's a better idea for an organization
to get several class C addresses because the number of class B addresses is limited. Class D is
reserved for multicast messages on the network, and class E is reserved for experimentation
and development.
Obtaining IP Addresses
The administration of Internet addresses is currently handled by the Network Information
Center (NIC):
Network Solutions
ATTN: InterNIC Registration Services
505 Huntmar Park Drive
Herndon, VA 22070
703-742-4777
ON THE WEB
You also can reach the InterNIC on the Web at the following address:
http://www.internic.net
The Internet addressing and domain name registration process is in a state of flux as
of this writing. To see what's happened since the publication of this book, visit
http://www.networksolutions.com/ and http://www.icann.org/ for the
most current information, or contact your Internet service provider.
When you connect a computer or a network to the Internet, in most cases your Internet
service provider can arrange for your network IP address registration.
Obtaining RFCs
In addition to assigning addresses, the NIC can provide other information of value. It's a
repository for all technical documentation related to the Internet. It has a collection of
documents that describe all the associated protocols, routing methodologies, network Part
management guidelines, and methods for using different networking technologies.
V
As mentioned in Table 28.1, RFC stands for Request For Comments. You can obtain RFCs Ch
from the Internet by using the FTP protocol to connect to several different repositories. The
28
RFC series is available on the Internet via anonymous FTP from various sites, such as
ftp.internic.net in the /rfc directory, and can also be accessed via Telnet at
rs.internic.net.
606 Chapter 28 Understanding the TCP/IP Protocol Suite
Table 28.5 lists the pertinent RFCs for establishing a network. Some of these documents go
into great detail about how the different protocols function and the underlying specifications
and theory. Others are more general and provide key information that can be useful to a
network manager. At a minimum, an Internet network manager should know where these
documents are located and how to obtain them. They provide information that can help in
planning and expanding an organization's network.
Table 28.5 RFCs of Interest
RFC Name Description
RFC791.txt Internet Protocol DARPA Internet Program Protocol Specification
RFC792.txt Internet Control Message Protocol
RFC793.txt Transmission Control Protocol DARPA Internet Program Protocol
Specification
RFC950.txt Internet Standard Subnetting Procedure
RFC1058.txt Routing Information Protocol
RFC1178.txt Choosing a Name for Your Computer
RFC1180.txt A TCP/IP Tutorial
RFC1208.txt A Glossary of Networking Terms
RFC1219.txt On the Assignment of Subnet Numbers
RFC1234.txt Tunneling IPX Traffic Through IP Networks
Network Naming
The naming of network nodes requires some planning. When you select names, you should
keep network management and user acceptance in mind. Many organizations have network-
naming standards. If your organization has such standards in place, it's best to follow them to
prevent confusion. If not, there's plenty of room for imagination. Computer and network
names can be as simple as naming the workstations after the users, such as Diane, Beth, or
John.
If you have many similar computers, numbering them (for example, PC1, PC2, and PC128)
may be appropriate. Naming must be done in a way that gives unique names to computer
systems. Don't name a computer thecomputerinthenorthoffice and expect users not to complain.
After all, even the system administrator must type the names of computers from time to time.
Also, you should avoid names like oiiomfw932kk. Although such a name may prevent
network intruders from connecting to your computer, it may also prevent you from
connecting to your workstation.
Tip #146 from
Steve Names that are distinctive and follow a theme work well, helping the coordination of
future expansion and giving the users a sense of connection with their machines. After
607
IP Addresses
all, having a good relationship with a machine called sparky is a lot easier than having
a relationship with a machine called OF1284.
Remember the following points when you're selecting a naming scheme:
& Keep names simple and shortÐsix to eight characters at most. Although the Internet
Protocol allows names up to 255 characters long, you should avoid going to such
lengths because some systems can't handle long names. (Each label can be up to 63
characters long. Each part of a period-separated full domain name for a node is a
label.)
& Consider using a theme such as stars, flowers, or colors, unless other naming standards
are required at your site.
& Don't begin the name with numbers.
& Don't use special characters in the name.
& Don't duplicate names.
& Be consistent in your naming policy.
If you follow these guidelines, you can establish a successful naming methodology.
Internet names represent the organizations and the functionality of the systems within the
network. The following are examples of names that you can use:
spanky.engineering.mycompany.com
nic.ddn.mil
The following are examples of names that are difficult to use or remember:
thisismyworkstation.thelongwindeddepartment.longcompnam.com
34556nx.m3422.mycompany.com
The latter of these examples could be encoded information about a workstation in room 345
on network 56 with network executive functions, but this type of naming scheme is usually
considered poor practice because it can lead to confusion and misdirected messages.
By using an Internet name such as [email protected], you can
reference a user on a particular node.
Part
NIC Naming Tree
V
The NIC maintains a network naming tree. This tree is used to group similar organizations
Ch
under similar branches of the tree. Figure 28.2 shows the naming tree. Major organizations
are grouped under similar branches. This is the source for Internet labels, such as com, edu, 28
and gov, that are seen in Internet names.
608 Chapter 28 Understanding the TCP/IP Protocol Suite
Figure 28.2 INTERNET
The NIC nam-
NAMING TREE
ing tree.
EDU COM GOV MIL NET NATO
Table 28.6 shows some of the common leaf names and definitions for the NIC tree. Many
other leaves are under the tree, but the ones shown here are the most common.
Table 28.6 Common NIC Names
Name Types of Organizations
edu Educational facilities (such as universities and colleges)
com Commercial (most corporations)
gov United States nonmilitary government bodies (White House, Department of
Agriculture)
mil Military (military users and their contractors)
net Internet network management and administration
org Other types of organizations (usually nonprofit)
Subnetworks and Subnet Masks
Subnetting is the process of dividing a large logical network into smaller physical networks.
Reasons for dividing a network may include electrical limitations of the networking
technology, a desire to segment for simplicity by putting a separate network on each floor of a
building (or in each department or for each application), or a need for remote locations
connected with a high-speed line.
The resulting networks are smaller chunks of the whole and are easier to manage. Smaller
subnets communicate with one another through gateways and routers. Also, an organization
may have several subnetworks that are physically on the same network to logically divide the
network functions into workgroups.
The individual subnets are a division of the whole. Suppose that a class B network is divided
into 64 separate subnets. To subnet the Class B network, the IP address is viewed in two
parts: network and host (see Figure 28.3). The network part becomes the assigned IP address
and the subnet information bits. These bits are, in essence, removed from the host's part of
the address. The assigned number of bits for a class B network is 16. The subnet part adds 6
bits, for a total of 22 bits to distinguish the subnetwork. This division results in 64 networks
with 1,024 nodes in each. The network part can be larger or smaller, depending on the
number of networks desired or the number of nodes per network.
609
Subnetworks and Subnet Masks
Figure 28.3 NETWORK HOSTS
An example of
31 15 0
class B subnet-
work masking. 6-BIT CLASS B SUBNETWORK
NETWORK SUB HOSTS
31 15 9 0
NUMBER OF SUBNETS = 64
NUMBER OF HOSTS/NET = 1024
SUBNET MASKING
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0
31 15 9 0
USING DOT CONVENTION
MASK - 255.255.252.0
Setting a subnet mask is a matter of determining where the network address ends and the host
address begins. The subnet mask contains all 1s in the network field and 0s in the host field.
Suppose a class C network is composed of the following:
N = network
H = Host
NNNNNNNN.NNNNNNNN.NNNNNNNN.HHHHHHHH
Each position represents a single bit out of the 32-bit address space. If this class C network is
to be divided into four class C networks, the pattern resembles the following:
NNNNNNNN.NNNNNNNN.NNNNNNNN.NNHHHHHH
The subnet mask looks like the following:
11111111.11111111.11111111.11000000
If this address is written in base-10 dot notation, the subnet mask is 255.255.255.192. This
mask is used to communicate among nodes on all subnetworks within this particular network.
If three bits are taken from the host field, eight networks can be formed, and the resulting
network mask is as follows:
11111111.11111111.11111111.11100000
This subnet mask is 255.255.255.224. Each of the eight networks would have 29 nodes
because five address bits are available. (It would be 32, except that all 1s, all 0s, and 127 aren't Part
legal addresses.)
V
This concept can be extended to class B and class A networks. The only difference is that the Ch
remaining fields are 0 (zero).
28
Consider a class B network. The address space is divided as follows:
NNNNNNNN.NNNNNNNN.HHHHHHHH.HHHHHHHH
610 Chapter 28 Understanding the TCP/IP Protocol Suite
If two bits are taken from the host field and added to the network part, the following subnet
mask is used:
11111111.11111111.11000000.00000000
The mask is written as 255.255.192.0.
Tip #147 from
Steve The bits needed for the subnet mask can be taken from any of the bit positions within
the host field, but using bits from random bit positions in the host field leads to
complex subnet masks and address exclusions. You should avoid this situation if at all
possible.
Routing
Routingis a method of transferring information between networks. A router works at the
Network layer of network protocols. Data can be routed by several different means. The
routing method implemented for an Internet network is the Routing Information Protocol
(RIP).
Routing Information Protocol (RIP)
RIP is designed to be used in small- to medium-sized networks and is based on Xerox
Network Systems (XNS) routing protocols. RIP determines a message route by using a
distance-vector routing algorithm. This algorithm assumes that each path is assigned a cost.
This cost can be representative of network throughput, type of line, or desirability of the
path. The protocol then determines the lowest cost path over which to transmit the message.
(You can obtain information about routing from several RFCs.)
How a Routing Protocol Works
A TCP packet must ``hop'' from node to node until the packet reaches its destination. To maintain a list of hops
to adjacent nodes, a RIP router keeps a routing table in the router or computer memory. This table is updated at
30-second intervals with information from neighboring routers. The information is used to recalculate the
lowest cost path between systems. Each router on a network sends out (advertises) and receives routing
information.
The routing protocol is limited in the distance a message can be routed. Each router can route a message only
to a cost of 16. If the message sent out on a wire costs more than 16, the host is deemed unreachable. Cost is a
method of assigning values to different paths through the network and is a way of ensuring an efficient route to
a destination when there's more than one way to get there.
When a network break occurs, the routers must relearn least-cost paths. This process takes time and can result
in messages being transmitted at a higher cost for a period of time. When a node goes down, all routers must
readjust their respective routing tables. During this time, messages can be lost in the network. After a period of
time, the routers are again synchronized and routing continues.
Router crashes are also a concern. In the event of a crash, adjacent routers update their adjacency to a crashed
router in 180 seconds. After that period of time, if no routing information is received from the crashed router,
that path is removed from the local router's database.
611
Internet Network Setup
RIP doesn't manage routing distances, just cost. As a result, RIP might not use the shortest
physical path between two points. Work and modifications have been made to the protocols
to help correct this problem. A newer routing protocol being developed and tested is Open
Shortest Path First (OSPF), which is beginning to gain acceptance and use.
Network Segmentation
Internet networks are divided into segments for various reasons. Some of these reasons are
related to the underlying networking technologies; others are related to geographical
locations. Some of the best reasons to isolate network segments are based on network usage.
If a lot of traffic in a network occurs between a few nodes, it's best to isolate those nodes. This
isolation drops the usage and provides a more responsive network for the other network
users.
Other reasons to segment are to change networking technologies or to communicate between
different networking technologies. For example, an office area may be running Token Ring,
and the shop floor area may be running Ethernet. Each has a distinct function. The office
may require Token Ring to communicate with an AS/400. The shop floor may have Ethernet
to enable shop floor controllers and computers to communicate. The shop floor information
then may be uploaded to the office network for order tracking. The connection between the
technologies is usually through routers. The routers forward only information that must be
exchanged from one network to the other. This information can then be shared between
nodes on the respective networks.
Excessive use of routers in a network can become a burden to the network, thus outweighing
their benefits. The use of a router is of little benefit if all the nodes on one network must get
to all the nodes on another network, and vice versa. In this instance, the advantages of routing
would be diminished because of the overhead in the routing protocols. In that kind of
situation, a bridge is a better alternative.
A bridge enables all information from two networks to be shared. The access is at the Physical
layer and not at the Network layer, so address translation and routing overhead aren't
incurred. A bridge enables all information, including system broadcast messages, to be
transmitted. If two networks rarely share information, a router is a better choice; otherwise, a
bridge is the proper choice.
Internet Network Setup
Part
The design and configuration of an Internet network is similar to the design of any computer
network. It encompasses many types of nodes, including workstations, servers, printers,
V
mainframes, routers, bridges, gateways, print servers, and terminals. The Internet requires Ch
that each device has a unique IP address. A device can have more than one address, depending
28
on its function, but at least one address is required for communication with the other devices.
612 Chapter 28 Understanding the TCP/IP Protocol Suite
Understanding the Types of Connections
A TCP/IP network can consist of several systems connected to a local area network (LAN) or
hundreds of systems with connections to thousands of systems on the Internet. Each
organization can create the type of network appropriate for its needs.
Figure 28.4 shows a simple network that consists of several workstations and a file server.
Each station on the network is assigned the network address of 194.62.23. Each device is
assigned an individual node address. This network is typical of most departments within a
company or even a small office. It has room to connect printers and more workstations to the
network. The network has no provisions for connections to other local or wide area networks
(WANs).
Figure 28.4
SERVER 194.62.23.1
A simple net-
work.
NETWORK BACKBONE
WKST_1 WKST_2 WKST_3
194.62.23.2 194.62.23.3 194.62.23.4
The network illustrated in Figure 28.5 is more complex. It includes three separate networks
interconnected through a combination of routers and servers. Each workstation and
computer on each segment may or may not be isolated from using information on one of the
other two networks. This variable isolation is a characteristic of the subnet mask and security
enabled on the servers and routers.
Information from one network is routed to one of the other networks on an as-needed basis.
This type of configuration is typical of most large corporate networks. It may be chosen based
on physical-length limitations of the underlying network technology or individual network
loading. One or more of the networks may experience high traffic that must be distributed
across several networks.
Router 1 between networks 1 and 2 provides for routing information between the two
networks. If server 1, connecting networks 2 and 3, has routing enabled, information from
network 3 to network 2 is routed. Also, information can be routed from network 3 to network
2 by means of server 1 and from network 2 to network 1 by means of router 1. Server 1,
connecting networks 2 and 3, has two IP addresses: one IP address on network 2 and another
address on network 3. The same is true for router 1, with addresses on network 2 and
network 1.
Tip #148 from
Steve Consider a situation in which a lot of Internet network traffic takes place between
network 3 and network 1. In this case, placing an additional router between network
1 and network 3 might be worthwhile. The additional router can eliminate some of
613
Internet Network Setup
the routing overhead on server 1 and enable information to be passed between
networks when server 1 is down.
The additional router can add a level of fault tolerance to the network. This fault
tolerance is based on the fact that information can still be routed to network 2 from
network 3, even when server 1 is down. The path between network 3 and network 2
would be through network 1 and router 1. Figure 28.6 shows the addition of router 2.
Figure 28.5
MAC_1 SERVER_2
A more com-
plex network.
194.54.21.2 194.54.21.3
NETWORK ONE
194.54.21.1
PC_1 WS_1 ROUTER_1
194.54.22.3 194.54.22.2 194.54.22.1
194.54.22.4 NETWORK TWO
SERVER_1
194.54.26.1 NETWORK THREE
194.54.26.2 194.54.26.3 194.54.26.4
PC_2
MAC_2
PRINTER_1
The fault tolerance of a network improves its integrity and can be of particular importance in
certain applications. If time-critical information must be shared between two networks, an
alternative path should be provided between the networks. This path could be provided
through the use of additional routers. Because these paths may be indirect (through a third
network), a configuration parameter should be used.
This parameter is usually referred to as network cost. The cost of a hop can be increased by
increasing the value a packet takes across a network path. The default preferred path is the
low-cost path; the alternative path is the high-cost path. This arrangement prevents
information from being routed over the high-cost path on a regular basis.
Part
Figure 28.6 shows an additional router added between networks 1 and 3. The desired path for V
information from network 3 to network 2 is through server 1. Because router 2 connects
Ch
network 3 and network 1, information can be routed between those two networks. Also,
because router 2 is between network 1 and network 2, information is routed through that 28
path. Information from network 3 that's bound for network 2 can go over one of the two
paths: either through server 1, or through router 2 and router 1. The latter isn't the preferred
path because information can be routed directly over server 1. Therefore, a higher cost is
614 Chapter 28 Understanding the TCP/IP Protocol Suite
assigned to the path of router 2 and router 1 from network 2. This type of path analysis must be
performed in a multiple-segment network.
Figure 28.6
MAC_1 SERVER_2
The network
after adding a
second router 194.54.21.2 194.54.21.3
for fault toler- NETWORK ONE
ance. 194.54.21.1
PC_1 WS_1 ROUTER_1
194.54.22.3 194.54.22.2 194.54.22.1
NETWORK TWO 194.54.22.4 194.54.21.4
SERVER_1 ROUTER_2
194.54.26.5
NETWORK THREE 194.54.26.1
194.54.26.2 194.54.26.3 194.54.26.4
PC_2
MAC_2
PRINTER_1
Choosing a Networking Configuration
The physical media used by an Internet network can be almost any network technology in
current use. Internet network traffic isn't limited to Ethernet, ARCnet, or Token Ring. It can
travel over asynchronous RS-232, T1 lines, and through frame relay. Whatever networking
topology is selected for the network, the configuration, installation, and operation rules
associated with that networking technology must be followed.
Tip #149 from
Steve Keep in mind the bandwidth that an application requires. Many applications require
megabytes of data to be transferred, so bandwidth becomes a prime consideration.
You can usually save bandwidth by compressing files before sending them over the
network. Some applications have data compression built in to automatically shrink
and expand the size of a file being transferred between the server and client
applications.
Another consideration is the physical location of the network. If all nodes are in the same
building, a single LAN can be used. However, if the networks are located across town, a T1
connection might be needed. If the nodes are located in different geographic locations, a
frame relay or a packet-switched network can be used.
615
Internet Network Setup
In laying out a network, you must consider the type of information to be carried over the
network, the physical location, and network loading. To help determine the capacity of the
network, you can examine the types of workstations, servers, and applications.
If diskless workstations are used in a network, a higher network load is placed on the network
for each node. The reason for this is that each remote diskless workstation requires all
operating system code to be downloaded through the network. Because all applications,
utilities, and data files are stored remotely, every action on that workstation requires network
access.
Also of concern is the amount of NFS traffic that will occur on the network. NFS provides
remote virtual disk services, so information retrieved and stored on these remote disks is
constantly used on the network.
Other considerations are large graphics images, swapping and page files used for virtual
memory, distributed database applications, printer traffic, and terminal traffic. They are all
considerations in any network, but the designers and users of PC-based LANs don't usually
have to contend with them. When a network is connected into a general user community, all
aspects of the networking environment come into play.
Other items to be examined are the need for dial-up and remote access. If this access is
related to terminal and screen traffic, a serial port from an existing system may suffice. If a
Point-to-Point Protocol (PPP) connection is made, you must consider how much overhead it
will impose on the network when users are loading software utilities, programs, and databases
over the phone lines. This issue is of concern because IP isn't limited to a high-speed link
such as Novell IPX and other networking protocols.
Understanding Network Configuration Guidelines
A network must be designed based on guidelines and rules. You should consider the following
questions when planning a network:
& How will the network be used today?
& How will the network be used for the next several years?
& What applications are going to be used on the network?
& Will workgroups within the organization require networking resources in the future?
& What types and numbers of workstations will be on the network?
& How many servers, minicomputers, and other hosts will be on the network?
&
Part
What other network devices, such as printers and plotters, will be on the network?
V
& Will shared disk arrays and optical jukeboxes be necessary?
& Will management of the network be centralized?
Ch
28
& Will the network be connected to the Internet or other corporate networks, or will it
be the basis for a wide area network?
616 Chapter 28 Understanding the TCP/IP Protocol Suite
& What other protocols will use the networking technology (IPX, DECNET, LAT,
OSI protocols, and TCP/IP)?
& Where will critical data be interchanged (determine several different paths)?
& How will the network grow and change?
After you address all these questions, you can define the network. The number of nodes
indicates how many class C address spaces are needed or whether a class B is needed.
Tip #150 from
Steve Connection to remote facilities should also be addressed. The load can be distributed
across multiple network segments. You should try to minimize the traffic that has to
go across networks. For example, if you have two systems that exchange a lot of
information, and hops across three networks are required for them to communicate,
consider moving the systems to the same network.
You need to determine the best networking topology to meet the requirements specified in
the network analysis. The best approach to allow for growth in the network is to determine
the maximum load and to develop a network in which that load is at a minimum.
Using Routers, Bridges, and Switches
Special-purpose devices are used to provide connections between networks and systems.
Sometimes the terms gateway and router are used interchangeably. Strictly speaking, gateway
describes a system that sends messages between different types of networks; a router sends
messages between networks of the same type.
In this text, router is routinely used to describe any device that takes messages from one
network and passes them through to another network. The router contains enough
intelligence to know whether the message received must be forwarded to another network or
a router.
Routers operate at the Network layer and are usually associated with a protocol, such as IP or
IPX. Most routers that route IPX traffic can route IP traffic as well. A router is used to
connect multiple local and wide area networks. It provides a method of sharing data between
networks. Also, because a router works at the Network layer, it can help reduce broadcast
traffic.
If one network uses a lot of different protocols, and another network uses only IP, a router
that routes only IP messages is needed if those two networks are to communicate. The router
prevents messages from being placed on a network that can't manage them.
617
Troubleshooting
Bridges, on the other hand, can be used to interconnect local and wide area networks; they
share information regardless of protocol. A bridge allows two interconnected networks to
have many different protocols on them at the same time. The messages forwarded by a bridge
usually don't contain any further routing information. The messages are usually left
undisturbed.
One drawback of using bridges is that all network broadcast and multicast messages from all
interconnected networks are seen on all legs connected by a bridge. The result is a lot of
overhead related to network update messages. Also, a bridge forwards messages only to
network addresses on the other side of the bridge, but it can forward all network protocols
and broadcast messages.
Conceptually, switches are multiport bridges. Because bridging is an OSI layer 2 function, all
the common networking standards such as Ethernet, Token Ring, or FDDI can be bridged or
switched. Switches are generally used in an existing network to divide a larger local area
network into many smaller ones.
Performance of LAN switches can vary greatly and is often based on how they handle the
forwarding of packets (many switches are configured as bridges). Methods for packet
forwarding can be one of the following:
& Store-and-forward
& Cut-through
& Modified cut-through
Routers, bridges, and switches are used to share information between networks. The
appropriateness of each is determined by networking requirements, the protocols involved,
network capacity, and user demands. The proper selection of components can help a network
operate efficiently, allow for future growth, and help ensure continued reliability.
Tip #151 from
Steve You should use bridges only if multiple protocol packets are to be shared. Otherwise,
a router is a better choice because it helps reduce network overhead.
Troubleshooting Part
When you're troubleshooting, you frequently can trace a TCP/IP-related issue to other V
related issues. Say you want to see whether your coworker Nathan Bradley is still logged in to
Ch
his workstation (named hawthorne), and you're both on the same private network. To do so,
you issue the following command:
28
finger nathanb@haw thorne
618 Chapter 28 Understanding the TCP/IP Protocol Suite
You might get the response
finger: unknown host: hawthorne
indicating his workstation's name isn't hawthorne. The problem could be with the Domain
Name Service's configuration of the network. (Chapter 38, ``Configuring Domain Name
Service (DNS),'' explains the Internet name resolution system.) After some investigation, you
find the problem isn't with the DNS configuration; it's misinformation. The workstation
assigned to Nathan is actually named wilde. So you try again:
finger nath anb@wilde
You might get the response
finger: nathanb: no such user
indicating his login name isn't nathanb. So you check the company directory and see his login
name is simply nathan. You try again by entering
finger nathan@wilde
and get the response
finger: connect: connection refused
indicating the finger command has been disabled for the workstation wilde. At this point, if
not before, you simplify your life and use the telephone to call Nathan.
CHAPTER
29
Configuring a TCP/IP Network
In this chapter by Steve Burnett
Understanding the TCP/IP Configuration Files 620
Initializing Ethernet Interfaces 622
Understanding TCP/IP Routing 626
Project: Monitoring a TCP/IP Network with netstat 634
620 Chapter 29 Configuring a TCP/IP Network
Understanding the TCP/IP Configuration Files
Configuring a TCP/IP network is one of the more common tasks you'll face when
administering Linux machines. In the most basic cases, it's not very complex, but it does
require a bit of thought on the design of your network and knowledge of a small number of
programs and configuration files.
TCP/IP networking in Linux is controlled by a set of configuration files in the /etc
directory. These files tell Linux what its IP address, host name, and domain name are and also
control the network interfaces. Table 29.1 shows you what each file does; the following
sections describe these files in detail.
Table 29.1 Linux TCP/IP Networking Configuration Files
File Description
/etc/hosts Maps host names to IP addresses
/etc/networks Maps domain names to network addresses
/etc/rc.d/rc3.d/S10network Configures and activates your Ethernet interfaces at boot
time
The /etc/hosts File
Every computer on a TCP/IP network has an IP address, canonical host name, and zero or
more host name aliases. The /etc/hosts file is the original method for mapping host names
to IP addresses.
All host names, domain names, and IP addresses used in this chapter are fictitious
and don't reflect any true network on the Internet.
For illustrative purposes, look at the network that A Fake Company, Inc., has built. This
network consists of the single class B network address assigned to A Fake Company by
InterNIC (the organization that controls Internet addresses); this network has been split into
?
two class C subnetworks.
See ``IP Address Classes,'' p. 604
The format for the hosts file is as follows:
# /etc/hosts for linux1.afakecompany.com
#
# For loopbacking.
127.0.0.1 localhost
# This machine
166.82.1.21 linux1.afakecompany.com linux1 # the local machine
# Other hosts on our network
621
Understanding the TCP/IP Configuration Files
166.82.1.20 server.afakecompany.com server # the server
166.82.1.22 wk1.afakecompany.com # workstation 1 Part
166.82.1.10 netpr1.afakecompany.com netpr1 # networked printer
166.82.1.1 gateway.afakecompany.com gateway # the router V
166.82.1.1 gate-if1 # 1st interface on gateway
166.82.2.1 gate-if2 # 2nd interface on gateway Ch
166.82.1.30 linux2.afakecompany.com linux2 # Laptop via PLIP 29
# end of hosts file
Tip #152 from
Steve Notice that the preceding gateway has two host names for the IP address 166.82.1.1.
Giving a unique name to each network interface on a machine is a good idea. Doing
so makes it easier to see what's going on when you use the ifconfig, route, and
similar commands.
The format of the hosts file consists of one IP address per line beginning in the first column,
the canonical host name associated with that address, and then zero or more aliases. The
fields are separated by spaces or tabs. Empty lines and text following a # character are treated
as comments and are ignored.
The IP address 127.0.0.1 is known as the local loopback address and is reserved for this purpose.
It's normally assigned the name localhost. If you're going to use your machine only as a
standalone system or use SLIP or PPP to connect to the outside world, you need only the
localhost address in your hosts file.
The function of the /etc/hosts file has been mostly taken over by Domain Name
Service (DNS) on machines connected to the Internet or large internal networks. DNS
isn't available during boot or when you're running in single-user mode, however, so
it's a good idea to place the information for essential machines such as servers and
gateways in /etc/hosts.
On a network with only a few machines that aren't connected to the Internet, keeping
a complete listing of all hosts in /etc/hosts is easier than configuring and
maintaining DNS.
Tip #153 from
Steve Naming your networks makes it convenient to perform tasks such as static routing
that take a host name or network name. You don't have to remember the subnets by
their IP addresses, just their names.
622 Chapter 29 Configuring a TCP/IP Network
The /etc/networks File
Just as hosts have names and IP addresses, networks and subnets can be named. This naming
is handled by the /etc/networks file. The IP addresses in the networks file include only the
network address portion plus the subnetwork byte. The following is a sample file for
afakecompany.com:
# /etc/networks for afakecompany.com
localnet 127.0.0.0 # software loopback network
afakecompany-c1 166.82.1 # Development Group Network, Class C
afakecompany-c2 166.82.2 # MIS Network, Class C
# end of networks file
First is the localnet name and IP address, 127.0.0.0. If you aren't connecting your Linux
machine to a TCP/IP network or are using only SLIP or PPP, all you need to put in this file
is the localnet name and IP address.
The next lines identify the two class C subnetworks that A Fake Company has made from its
class B network.
Initializing Ethernet Interfaces
The ifconfig program makes network interfaces such as the software loopback and Ethernet
cards known to the Linux kernel so that Linux can use them. You also can use the ifconfig
program to monitor and change the state of network interfaces. A simple invocation of
ifconfig is the following:
ifconfig interfaceaddress
This command activates the specified network interface and assigns an IP address to it. This
process is called bringing up an interface. The generalized calling syntax for ifconfig is as
follows:
ifconfig interface [aftype] [options] | address
Table 29.2 lists the command-line arguments for ifconfig.
Table 29.2 Command-Line Arguments for ifconfig
Argument Description
interface Specifies the name of the network interface, usually the name of
the device driver followed by an identification number. This
argument is required.
aftype Specifies the address family that should be used for decoding and
displaying all protocol addresses. Currently, the inet (TCP/IP),
ddp (Appletalk Phase 2), ipx (Novell), and AX.25 and netrom
(both amateur packet radio) address families are supported. The
inet family is the default.
623
Initializing Ethernet Interfaces
Table 29.2 Command-Line Arguments for ifconfig
Part
Argument Description
V
up Activates the specified interface.
Ch
down Deactivates the specified interface.
29
[-]arp Turns on or off the use of the ARP protocol on the specified
interface. The minus sign is used to turn off the flag.
[-]trailers Turns on or off trailers on Ethernet frames. This argument isn't
currently implemented in the Linux networking system.
[-]allmulti Turns on or off the promiscuous mode of the interface. Turning
this mode on tells the interface to send all traffic on the network
to the kernel, not just traffic addressed to your machine.
metric N Sets the interface metric to the integer value N. The metric value
represents the ``cost'' of sending a packet on this route. Route
costing isn't currently used by the Linux kernel but is to be
implemented at a future date.
mtu N Sets the maximum number of bytes the interface can handle in
one transfer to the integer value N. The current networking code
in the kernel doesn't handle IP fragmentation, so you must make
sure that the Maximum Transmission Unit (MTU) value is set
large enough.
dstaddr addr Sets the IP address of the other end of a point-to-point link.
This argument has been made obsolete by the pointopoint
keyword.
netmask addr Sets the IP network mask for the specified interface.
irq addr Sets the interrupt line used by this device. Remember that many
devices don't support dynamic IRQ setting.
[-]broadcast[addr] Sets the broadcast address for the interface when an address is
included. If no address is given, the IFF_BROADCAST flag for the
specified interface is turned on. A leading minus sign turns off
the flag.
[-]pointopoint[addr] Turns on point-to-point mode on the specified interface. This
argument tells the kernel that this interface is a direct link to
another machine. The address, when included, is assigned to the
machine on the other end of the list. If no address is given, the
IFF_POINTOPOINT flag for the interface is turned on. A leading
minus sign turns off the flag.
hw Sets the hardware address for the specified interface. The name
of the hardware class and the ASCII equivalent of the hardware
address must follow this keyword. Ethernet ( ether), AMPR
AX.25 ( ax25), and PPP ( ppp) are now supported.
address Indicates the host name or IP address to be assigned to the
specified interface. Host names used here are resolved to their IP
address equivalents. This parameter is required.
624 Chapter 29 Configuring a TCP/IP Network
Tip #154 from
Steve You normally don't need to use all the options. ifconfig can set everything needed
from just the interface name, netmask, and IP address assigned. You need to explicitly
set most parameters only when ifconfig misses or you have a complex network.
If your Linux machine is on a network, you must keep the ifconfig program secure
from unauthorized use. Setting a network interface to promiscuous mode allows a
person to snoop in your network and get sensitive data such as passwords. This is a
serious breach of security.
? See ``Handling Physical Security,'' p. 276
Using ifconfig to Inspect a Network Interface
Running ifconfig with no arguments causes it to output the status of all network interfaces
the kernel knows about. Running ifconfig with just an interface name on the command line
prints the status of the interface, as shown here:
$ ifconfig lo
lo Link encap Local Loopback
inet addr 127.0.0.1 Bcast 127.255.255.255 Mask 255.0.0.0
UP LOOPBACK RUNNING MTU 2000 Metric 1
RX packets 0 errors 0 dropped 0 overruns 0
TX packets 1658 errors 10 dropped 0 overruns 0
This example uses lo, the software loopback interface. You can see the assigned IP address
(inetaddr), broadcast address (Bcast), and netmask (Mask). The interface is UP with an MTU of
2000 and a Metric of 1. The last two lines give statistics on the number of packets received
(RX) and transmitted (TX), along with packet error, dropped, and overrun counts.
Configuring the Software Loopback Interface
All Linux machines with the networking layer installed in the kernel have a software loopback
interface. This interface is used to test networking applications and to provide a network for
local TCP/IP services when the machine isn't connected to a real network.
The network interface name for the loopback system is lo. You enter the following to run
ifconfig:
ifconfig lo 127.0.0.1
This command activates the loopback interface and assigns the address 127.0.0.1 to it. This
address is traditionally used for the loopback because InterNIC will never assign the class A
network, 127.0.0.0, to anyone.
625
Initializing Ethernet Interfaces
To make the loopback system fully operational, you need to add a route for it by using the
route
Part
command, which is discussed later in the section ``Understanding TCP/IP Routing."
V
Configuring a Network Interface Ch
Configuring an Ethernet network interface takes a little bit more work, especially if you're 29
using subnetworks. The basic call to ifconfig looks like this for linux1.afakecompany.com:
ifconfig eth0 linux1
This command causes ifconfig to activate Ethernet interface 0, look up the IP address for
linux1 in the /etc/hosts file, and assign it to this interface. Examining the eth0 interface at
this point reveals the following code:
$ ifconfig eth0
eth0 Link encap 10Mbps Ethernet HWaddr 00:00:E1:54:3B:82
inet addr 166.82.1.21Bcast166.82.1.255 Mask 255.255.255.0
UP BROADCAST RUNNING MTU 1500 Metric 0
RX packets 3136 errors 217 dropped 7 overrun 26
TX packets 1752 errors 25 dropped 0 overrun 0
Interrupt:10 Base address:0x300
Note that the broadcast address and netmask were set automatically by ifconfig based on the
IP address it found in /etc/hosts. If you're using subnetworks, you need to specify the
broadcast address and netmask explicitly. For example, if you have a class C network and are
using the first bit in the host portion of the address to make two subnetworks, you must
specify the broadcast address and netmask when running ifconfig:
ifconfig eth0 linux1 broadcast 166.82.1.127 netmask 255.255.255.128
Configuring Parallel IP Interfaces
The Parallel IP (PLIP), Serial Line IP (SLIP), and Point-to-Point Protocol (PPP) interfaces
are managed by ifconfig somewhat differently. To bring up a PLIP interface, you add the
pointopoint option to the ifconfig command line. Assume that the A Fake Company laptop
linux2 is attached to the first parallel port on linux1. You call ifconfig as follows to activate
the PLIP link:
ifconfig plip0 linux1 pointopoint linux2
This command activates the plip0 interface with the IP address for linux1, sets the
pointopoint flag, and tells the interface that the IP address for the other end of the link is
linux2. ifconfig looks up the IP addresses for linux1 and linux2 in /etc/hosts and assigns
the addresses appropriately. On a laptop, you use the following analogous call:
?
ifconfig plip0 linux2 pointopoint linux1
See ``Understanding the Requirements for SLIP and PPP,'' p. 664
626 Chapter 29 Configuring a TCP/IP Network
Understanding TCP/IP Routing
Routing determines the path a packet takes from its source through the network to its
destination. This path is determined by matching the destination IP address against the
kernel routing tables and transmitting the packet to the indicated machine, which may or may
not be the destination of the packet. The kernel routing table contains information in the
form ``To get to network X from machine Y, send the packet to machine Z with a cost of 1,''
along with time-to-live and reliability values for that route.
Deciding On a Routing Policy
The first step in setting up routing on your network is deciding on a routing policy. For
small, unconnected networks, using the route command to set up static routes on each
machine at boot time is sufficient. Large networks with many subnets or networks connected
to the Internet need to use dynamic routing. The routing program provides dynamic routing
by communicating with routing programs on other machines and installing routes based on
what it learns about the topology of the network.
A very common strategy combines static and dynamic routing. Machines on each subnet use
static routing to reach their immediate neighbors. The default routeÐthe route used for
packets that match no other route in the routing tableÐis set to a gateway machine that's
doing dynamic routing and knows about the rest of the world. Large networks can be
constructed this way, minimizing the hassle of configuration files and the amount of
bandwidth used by the dynamic routing programs.
Using the /sbin/route Program
The /sbin/route program manipulates the kernel routing table and is used to set static routes
to other computers or networks via interfaces that have been configured and activated by
ifconfig. This is normally done at boot time by the /etc/rc.d/rc3.d/S10network script.
Table 29.3 describes the command-line arguments for /sbin/route.
Table 29.3 Command-Line Arguments for /sbin/route
Argument Description
(None) Giving no option to /sbin/route causes it to output the current routing table.
-n This argument causes the same output as giving no option but replaces host
names with their numerical IP addresses.
del This argument deletes the route for the specified destination address from the
routing table.
add This argument adds to the routing table a route to the specified address or
network.
627
Understanding TCP/IP Routing
Examining the Kernel Routing Table
Part
Running /sbin/route without any command-line arguments or just -n outputs the routing
V
table:
Ch
/sbin/route
Kernel routing table 29
Destination Gateway Genmask Flags Metric Ref UseIface
127.0.0.0 * 255.0.0.0 U 0 0 100 lo
The preceding output is from a machine with just the loopback interface activated. Table
29.4 describes the fields in the routing table report.
Table 29.4 The Fields in the Routing Table Report
Field Description
Destination The destination IP address of the route.
Gateway The host name or IP address of the gateway the route uses. If no gateway
exists, an asterisk is output.
Genmask The netmask for the route. The kernel uses this field to set the generality of
a route by bitwise ANDing the Genmask against a packet's IP address before
comparing it to the destination IP address of the route.
Flags The flags for the route ( U means up, H means host, G means gateway, D
means dynamic route, and M means modified).
Metric The metric cost for the route. This field isn't currently supported in the
kernel networking layer.
Ref The number of other routes that rely on the presence of this route.
Use The number of times the routing table entry has been used.
Iface The network interface to which this route delivers packets.
Returning to the A Fake Company network, the following is an example from the laptop
linux2, with a SLIP link up and running:
$ /sbin/route
Kernel routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
slip.afakecompany.c * 255.255.255.255 UH 0 0 0 sl0
127.0.0.0 * 255.0.0.0 U 0 0 100 lo
default slip.afakecompany.c * UG 0 0 1 sl0
The table entry for the loopback is the same as before, and you have two new entries. The
slip.afakecompany.com. The
first specifies a route to other new entry specifies a default
route by using slip.afakecompany.com as a gateway.
Every machine connected to a network must have a default route in its routing table.
The default route is used when no other routing table entry matches the destination
for a packet.
628 Chapter 29 Configuring a TCP/IP Network
Adding Static Routes
You add routes to the routing table by running the route program with the add argument.
The command-line argument syntax for the route add command is as follows:
route add [ -net | -host ] addr [gw gateway] [metric cost]
A[netmask mask] [dev device]
Table 29.5 describes the command-line arguments that the route add command uses.
Table 29.5 Command-Line Arguments Used by route add
Argument Description
-net | -host Forces the specified address to be treated as a network or host address.
addr Specifies the destination address for the new route. It can be an IP address,
host name, or network name.
gw gateway Specifies that any packets for this address be routed through the specified
gateway.
metric cost Sets the metric field in the routing table.
netmask mask Specifies the netmask of the route being added. The route program will
guess what it is, so you don't need to specify it under normal circumstances.
dev device Forces route to associate the new route with the specified network interface
device. Again, route usually guesses correctly what device to use for the new
route, so you don't have to use this argument often.
When adding a gateway route to the routing table, you must make sure that the
specified gateway is reachable. You usually have to add a static route for the gateway
before adding the route by using the gateway.
Looking at Routing Examples
Now you're ready for some examples, starting with the loopback interface. After configuring
the loopback interface with ifconfig, you need to add a route to it, as in the following:
# route add 127.0.0.1
Nothing else is needed because route compares the address given to it with the addresses for
the known interfaces and assigns the loopback interface to the new route. The following
example shows how to set the routing for the SLIP link on the A Fake Company linux2
machine after the SLIP link is established and ifconfig is used to activate the interface:
# route add slip.afakecompany.com
# route add default gw slip.afakecompany.com
629
Understanding TCP/IP Routing
The first command adds a static route for the host slip.afakecompany.com; the second one
slip.afakecompany.com
Part
tells the kernel to use as a gateway for all packets with unknown
destinations. V
Ch
29
Make sure that any host names you use with the route command are in the /etc/
hosts file so that route can find the IP addresses for them; otherwise, route fails.
If you're subnetting your network by splitting the IP address in the middle of an octet, you
have to specify the required netmask when running route. For example, if you have a class C
network and have four subnets using the first two bits of the last octet, you need to run route
like this:
# route add hostname netmask 255.255.255.192
Tip #155 from
Steve An octet is the name for each of the four portions of an IP address. The fourth octet in
the example command above is 192.
This command ensures that route puts the right netmask in the routing table entry.
For Ethernet and other broadcast network interfaces, you need to add routes that tell the
kernel what network can be reached via each configured interface. After using ifconfig to
bring up the eth0 network interface on linux1.afakecompany.com as you did previously, you
need to run route to install the route to the network on that interface:
# route add -net 166.82.1.0
That command might not look like enough to set the routing table entry correctly because no
interface is indicated; however, route manages to find the interface by comparing the IP
address on the command line to the IP address of each network interface. It assigns the route
to the interface that matches it. In this case, eth0 has been assigned the address 166.82.1.21
with a netmask of 255.255.255.0. This address matches the network address given in the
route command, so route installs a route to the network 166.82.1.0 by using interface eth0,
as follows:
$ route
Kernel routing table
Destination Gateway Genmask Flags Metric Ref UseIface
166.82.1.0 * 255.255.255.0 UN 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 100 lo
To tell linux1 how to reach the other subnet, you need two more routing table entries to be
safe:
630 Chapter 29 Configuring a TCP/IP Network
# route add gateway.afakecompany.com
# route add -net 166.82.2.0 gw gateway.afakecompany.com
These lines add a static route to gateway.afakecompany.com and then add a network route for
166.82.2.0 by using gateway.afakecompany.com as the gateway for that network, as shown in
the following:
$ route
Kernel routing table
Destination Gateway Genmask Flags Metric Ref UseIface
gateway.afakecompany * 255.255.255.0 UH 0 0 0 eth0
166.82.1.0 * 255.255.255.0 UN 0 0 0 eth0
166.82.2.0 gateway.afakecompany 255.255.255.0 UN 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 100 lo
This output shows the static route you added for gateway.afakecompany.com and the
gatewayed route to the 166.82.2.0 network.
Deleting Routes with the route Command
You delete routes by calling route with the del option and specifying the destination address
of the route you want to delete. For example, the following command deletes the network
route for network 166.82.2.0:
# route del -net 166.82.2.0
Project: Monitoring a TCP/IP Network with
netstat
The netstat program is an invaluable tool in monitoring your TCP/IP network. It can
display the kernel routing table, the status of active network connections, and useful statistics
about each network interface. Table 29.6 describes the common command-line arguments
for netstat; a few additional arguments are targeted for advanced users. Refer to the man
page for more information.
Table 29.6 Common Command-Line Arguments for the netstat Program
Argument Description
-a Shows information about all Internet connections, including those that are just
listening.
-i Shows statistics for all network devices.
-c Shows continually updating network status. This argument makes netstat
output a network status listing once per second until it's interrupted.
-n Shows remote and local addresses and port information in numeric/raw form
rather than resolves host names and service names.
-o Shows the timer state expiration time and backoff state of each network
connection.
-r Shows the kernel routing table.
631
Project: Monitoring a TCP/IP Network with netstat
Table 29.6 Common Command-Line Arguments for the netstat Program
Part
Argument Description
V
-t Shows only TCP socket information, including those that are just listening.
Ch
-u Shows only UDP socket information.
29
-v Shows the version information for netstat.
-w Shows raw socket information.
-x Shows UNIX domain socket information.
Displaying Active Network Connections
Running netstat with no command-line arguments generates a listing of the active network
connections on your machine. The following demonstrates the default output from netstat:
$ netstat
Active Internet connections
Proto Recv-Q Send-Q Local Address Foreign Address (State)
tcp 0 0 linux1.afakecompany.com:71 server.afakecompany.:telnet ESTABLISHED
Active UNIX domain sockets
Proto RefCnt Flags Type State Path
unix 1 [ ACC ] SOCK_STREAM LISTENING /dev/printer
unix 2 [ ] SOCK_STREAM CONNECTED /dev/log
unix 2 [ ] SOCK_STREAM CONNECTED
unix 1 [ ACC ] SOCK_STREAM LISTENING /dev/log
The first section shows an active TCP protocol connection from port 1266 on
linux1.afakecompany.com to the telnet port on server.afakecompany.com. Table 29.7
describes the fields in the Active Internet Connections listing.
Table 29.7 Active Internet Connection Fields
Field Description
Proto The protocol used by this connection, TCP, or UDP.
Recv-Q The number of bytes received on this socket but not yet copied by the
user program.
Send-Q The number of bytes sent to the remote host that haven't been
acknowledged.
Local Address Local host name and port number assigned to this connection. The
socket IP address is resolved to the canonical host name for that address,
and the port number is translated into the service name unless the -n flag
is used.
Foreign Address The foreign host name and port number assigned to this connection.
The -n flag affects this field as it does the Local Address field.
632 Chapter 29 Configuring a TCP/IP Network
Table 29.7 Active Internet Connection Fields
Field Description
State The current state of the socket. It can be in one of the following states:
ESTABLISHEDÐThe connection is fully established.
SYN_SENTÐThe socket is now trying to make a connection to a remote
host.
SYN_RECVÐThe connection is being initialized.
FIN_WAIT1ÐThe socket has been closed and is waiting for the connection
to shut down.
FIN_WAIT2ÐThe connection has been closed. The socket is waiting for a
shutdown from the remote host.
TIME_WAITÐThe socket is closed and is waiting for a remote host
shutdown retransmission.
CLOSEDÐThe socket isn't in use.
CLOSE_WAITÐThe remote host has shut down its connection. The local
host is waiting for the socket to close.
LAST_ACKÐThe remote connection is shut down, and the socket is
closed. The local host is waiting for an acknowledgment.
LISTENÐThe socket is listening for the incoming connection attempt.
UNKNOWNÐThe state of the socket isn't known.
User The login ID of the user who owns the socket.s
The second section displays active UNIX domain sockets. UNIX domain sockets are an
interprocess communication (IPC) mechanism that uses the UNIX file system as the
rendezvous system. Processes create special files in the file system that are then opened by
other processes on the machine that wants to communicate. The preceding netstat listing
shows two sockets listening: one on /dev/printer and the other on /dev/log. Two sockets
are also currently connected: one to /dev/log and one that has no specified path associated
with it. Table 29.8 describes the fields in the Active UNIX Domain Sockets listing.
Table 29.8 Fields in the Active UNIX Domain Sockets Listings
Field Description
Proto The protocol in use on this socket. It is usually unix.
RefCnt The number of processes attached to this socket.
Flags The flags for this socket. Currently, the only known flag is SO_ACCEPTON ( ACC),
which indicates that the socket is unconnected and the process that made the socket
is waiting for a connection request.
Type The mode in which the socket is accessed. This field contains one of the following
keywords:
OCK_DGRAMÐDatagram, connectionless mode
OCK_STREAMÐConnection-oriented stream mode
OCK_RAWÐRaw mode
OCK_RDMÐReliably delivered message mode
OCK_SEQPACKETÐSequential packet mode
NKNOWNÐMode not known to netstat program
633
Project: Monitoring a TCP/IP Network with netstat
Table 29.8 Fields in the Active UNIX Domain Sockets Listings
Part
Field Description
V
State The current state of the socket. The following keywords are used:
Ch
FREEÐThe socket isn't allocated.
LISTENINGÐThe socket is waiting for a connection request. 29
UNCONNECTEDÐThe socket doesn't have a current connection.
CONNECTINGÐThe socket is attempting to make a connection.
CONNECTEDÐThe socket has a current connection.
DISCONNECTINGÐThe socket is attempting to shut down a connection.
UNKNOWNÐThe state of the socket is unknown. You don't see this keyword under
normal operating conditions.
Path The pathname used by other processes to connect to the socket.
Tip #156 from
Steve If network interfaces drop many packets or get many overrun errors, they can be
symptoms of an overloaded machine or network. Checking the network interface
statistics is a quick way of diagnosing this problem.
Invoking netstat with the -o option adds the internal state information to the Active Internet
Connections listing. The following is an example:
$ netstat -o
Active Internet connections
Proto Recv-Q Send-Q Local Address Foreign Address (State)
tcp 0 0 localhost:1121 localhost:telnet ESTABLISHED off (0.00/0)
tcp 0 0 localhost:telnet localhost:1121 ESTABLISHED on (673.69/0)
The added data is at the end of each line and includes receiver retransmission byte count,
transmitter retransmission byte count, timer state (on/off), and time/backoff values (in
parentheses). The time displayed is the time left before the timer expires. The backoff value is
the retry count for the current data transmission. This data is useful in diagnosing network
problems because you can easily see which connection is having problems.
Because the -o option outputs the state of internal TCP/IP data, the format of this
data may change, or this option may be removed in a later release of the networking
software.
Examining the Kernel Routing Table
Invoking netstat with the -r option prints the kernel routing table. The format is the same
as for the route command.
634 Chapter 29 Configuring a TCP/IP Network
Displaying Network Interface Statistics
Invoking netstat with the -i option prints usage statistics for each active network
interfaceÐanother excellent tool for debugging network problems. By using this command,
you can easily see when packets are being dropped, overrun, and so on.
The following is an example of using the -i option, and Table 29.9 explains each field in the
listing:
$ netstat -i
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flags
lo 2000 0 0 0 0 0 1558 1 0 0 LRU
Table 29.9 Fields in the Kernel Interface Table
Field Description
Iface The name of the network interface.
MTU The largest number of bytes that can be sent in one transmission by this interface.
Met The metric value for this interface.
RX-OK The number of packets received with no errors.
RX-ERR The number of packets received with errors.
RX-DRP The number of packets dropped.
RX-OVR The number of packet, overrun with errors.
TX-OK The number of packets transmitted with no errors.
TX-ERR The number of packets transmitted with errors.
TX-DRP The number of packets dropped during transmission.
TX-OVR The number of packets dropped due to overrun errors.
Flags
The following flags can be shown in this field:
AÐThe interface receives packets for multicast addresses.
BÐThe interface receives broadcast packets.
DÐThe interface debugging feature is now activated.
LÐThis is the loopback interface.
MÐThe interface is in promiscuous mode.
NÐThe interface doesn't process trailers on packets.
OÐThe Address Resolution Protocol is turned off on this interface.
PÐThis interface is being used as a point-to-point connection.
RÐThe interface is running.
UÐThe interface has been activated.
CHAPTER
30
IP Firewalling and Masquerading
In this chapter by Jack Tackett, Jr.
IP Firewalling and Masquerading 636
Introduction to Firewalls 636
Port Forwarding 644
A Simple Packet Filtering Firewall 645
Monitoring 654
Under Attack 655
Network Security Policy 655
Configuring IP Masquerading 655
Troubleshooting IP Masquerading 660
636 Chapter 30 IP Firewalling and Masquerading
IP Firewalling and Masquerading
This chapter discusses a little about firewalls in LinuxÐwhat they are, how they work, and
how to begin building one. Although this chapter will not make anyone an expert on firewalls,
it does introduce the basic concepts.
The term firewall comes from the firewall that is used in cars (and other motorized vehicles),
which protects the occupants in the cabin from a fire in the engine compartment. A firewall
on a network protects both users and the data that is behind it on the local network from the
savagery of the Internet (or extranet). It can also be used to prevent users on the local network
from connecting to prohibited sites, and it can be used to compartmentalize the internal
network.
Tip #157 from
Jack Some people learn too late that a firewall does not protect against insiders. Internal
firewalls can be used to isolate departments so that damage is not widespread; for
example, it can be used to protect accounting from engineering, and engineering
from sales, each of which likely has little or no business browsing through files that
belong to the others.
Introduction to Firewalls
Basically, there are two kinds of firewalls in Linux. Each of these two basic types has two
subtypes. The two basic types are packet filters and proxy firewalls. The packet filter firewalls
can be one of the following types:
& ForwardingÐIt is in this type of packet filter firewall that decisions are made whether
to forward or not.
& MasqueradingÐThese firewalls rewrite the source and destination addresses.
Proxy firewalls can be one of the following types:
& StandardÐWith this type of proxy firewall, a client connects to a special port and is
redirected out through another port.
& TransparentÐWith this firewall, the client doesn't use a special port, but the firewall
software proxies the connection through transparently.
Packet Filters
Packet filtering firewalls work on the following principle: The information that is needed to
make a decision about what to do with a packet is contained in the header. The header
contains information regarding the source and destination addresses, time to live (TTL),
protocol, and much more. It also contains a header checksum that recounts the size of the
637
Introduction to Firewalls
payload and whether the header has been corrupted. In all, some thirteen separate fields of
information are contained in an IP header, some of which contain multiple pieces of information.
For complete details, refer to RFC 791 (available from http://www.rfc-editor.org/rfc.html).
IP does not check the payload other than to tell if the payload is the correct size. The
transport control protocol (TCP) is responsible for ensuring the integrity of the data
payload.
Part
Ch
OpenLinux uses ipchainsto provide packet filtering. To implement a packet filtering
firewall, decisions must be made regarding the types of packets to address specifically, and 30
what to do with those packets when they are encountered.
The ipchains software permits a number of different criteria to be applied to packets. The
criteria can be applied to incoming packets, outgoing packets, or packets that will pass
through the firewall. These decisions can be based on where the packets came from by
address, where they are going by address, or where they are going by port. Different rules can
be applied, depending on whether these are TCP packets, UDP packets, or ICMP packets.
Finally, for any packets that are not specifically addressed, the overall policy determines the
fate of the packets.
Proxy Firewalls
Proxy firewalls work differently than packet filters do. All traffic is received on the firewall,
whether it is incoming or outgoing. But proxies redirect permitted traffic through the firewall
by rewriting the headers. To be redirected, the traffic has to log in to the firewall. In fact,
much of the discussion in the preceding section is applicable to a proxy. The difference is
subtle; because packet filtering software is rewriting the headers when it is masquerading, it is
difficult to explain that there is a difference between a transparent proxy and a masquerading
(packet filter) firewall. But the main difference is that the proxy redirects (locally) the traffic
that is arriving at one interface and leaving by another, and a packet filter normally does not
redirect traffic. Looking at it from a different perspective, proxies work higher up in the OSI
model than packet filters.
Any router, gateway, or host, that transports a networking packet from one network
to another rewrites the header. But this rewriting doesn't alter the source or
destination addresses; it only alters the TTL and checksum, andÐon occasionÐthe
total length and fragment offset (among others), if the packet needs to be fragmented
to continue. When header rewriting is discussed in this text, it refers mostly to address
rewriting.
638 Chapter 30 IP Firewalling and Masquerading
The Open Source Interconnect (OSI) model is one of the popular models that is used to explain
how packets move from the Application layer to the Physical layer (it might not be completely
accurate, but it's a good theoretical paradigm nevertheless). The seven layers are used to
explain where certain software works. For the purposes of this text, it is only important to
note that the level where each works is different, and that this is one of the distinguishing
characteristics between proxies and packet filters.
But this difference is important. Proxies are more overhead intensive, but they can inspect
entire packets more thoroughly. They also tend to be a little more difficult to set up initially.
Which to Use?
Packet filtering firewalls and proxying firewalls perform similar functions. They act as shields
to protect trusted network segments from untrusted ones. In this regard, each works equally
well. Both require monitoring and occasional reconfiguring; both, if they are misconfigured,
provide only a false sense of security; and both, when implemented in a methodical,
well-thought-out manner, can provide a modicum of security.
Tip #158 from
Jack If you remember that the only secure system is the one that is not assembled and
powered on (and therefore not of much use, either), you'll have a more realistic idea
of what firewalls can buy you: time to react to an attack. It remains your responsibility
to reconfigure the firewall to protect against this attack. I want to emphasize that: A
firewall buys you time to react to an attack.
Basically, the decision to use packet filtering or proxying comes down to an individual
decision, and might be based on the prior experiences of those who are involved in
maintaining the firewall. If you are implementing a firewall and have worked with proxies
before and are comfortable with them, by all means, continue. The use of one type doesn't
preclude use of the other. Some proxies that are designed to work specifically with Web
(HTTP) traffic can complement a packet filter nicely. For example, the use of junkbuster to
block particular Web sites or advertising banners is often easier than writing packet filtering
rules to deny or reject the banner sites. Conversely, ipchains rules can be created solely for
the purpose of logging traffic, and can be used in conjunction with a proxy to track specific
kinds of traffic. So the best option might be to mix and match, depending on your overall
objectives and level of comfort.
From a security standpoint, therefore, neither is better. The one place in which ipchains
might tip the balance is in situations in which you want to rewrite the Type of Service (TOS)
field to optimize traffic flow.
639
Introduction to Firewalls
Tip #159 from
Jack You can use ipchains to modify the TOS field to specify one of minimum delay,
maximum reliability, maximum throughput, or minimum cost.
Physical Configurations
Part
When discussing physical configurations, you need to look at both the hardware and the
software as it applies to the firewall. Remembering that one of the reasons the firewall exists is
V
to protect the trusted network from the untrusted network, this host must necessarily be both Ch
the funnel for network traffic that is moving between the two networks, making it a possible
30
choke point, and the focus for those who are looking to penetrate your security. If they want
to come in, they've got to pass through here first.
The Firewall Host
You need to consider the type of host you want to install. You can install a firewall that has
only one interface, and that uses that interface for both trusted and untrusted connections;
but you'll want to consider whether this is wise, taking into account the cost of a second
interface versus the weakened security posture that this configuration entails. It is better to
have a host with two interfaces, which completely isolates one network from the other. With
this configuration, bypassing the firewall becomes more difficult.
A host that isolates an untrusted network from a trusted network using two
interfacesÐone for each networkÐis termed aBastion Host.
The question of how powerful a system needs to be depends on your decisions; if you are
going to use the firewall to connect two 10Mb Ethernet cards, and if you plan to use packet
filtering and no proxies, an 80486-33 processor with 16MB RAM is sufficient for low to
moderate traffic loads. However, if you plan to use 100Mhz Ethernet cards in a high traffic
route, this CPU will not be able to keep up with the demand and you'll experience significant
packet loss.
The Firewall Kernel
In building a firewall, you'll need to reconfigure the kernel. Several parameters must be set in
the kernel to permit packet filtering. Some of these parameters are subjective and are based
on your hardware. One must be turned off; others are required. For more information on
building a custom kernel, see Chapter 14, ``Configuring the Linux Kernel.'' The first
parameter follows:
CONFIG_EXPERIMENTAL=y
640 Chapter 30 IP Firewalling and Masquerading
One of the items that kernel hackers typically review is the experimental status on a number
of the kernel parameters. When it is reviewed, the experimental status will, most likely, better
reflect reality. Therefore, enabling this is a judgement call, although you might need it for
some drivers.
It is strongly recommended that you consider compiling the kernel as a monolithic kernel as
opposed to a modular kernel (for more information on building a custom kernel, see Chapter
?
14).
See ``Configuring a New Kernel,'' p. 297
This is an exception to the rule, ``Always build a modular kernel.'' The reasoning behind this
is that anyone can build a module. If someone manages to crack your security, they'll want
access again. Rather than amateurishly adding a user with root privileges, they can drop a
loadable module on your system, have it inserted, and then erase the other traces of entry.
Although this is extreme and a rather sophisticated way to go about it, it might enable nearly
undetectable access to your system. So kernel modules can be dangerous. On the other hand,
anyone this sophisticated can, most likely, easily find other means to enter the system.
Unfortunately, to enable things such as IP port forwarding and some other parameters,
module support is required. If you don't see a parameter that you need, it might either be
experimental or only available as a module.
In other sections, give careful thought to the parameters that you install. You need to support
your hardware (disk drives), the file system, Ethernet drivers, other communications drivers
(modems, ISDN devices, and protocols such as PPP), and ELF formats. But sound and other
unnecessary parameters need to be disabled.
If this is a home network or part of a small, low-profile business with low bandwidth,
you might not need or want to go to the extremes detailed in this chapter. Only you
can perform a proper risk assessment for your situation.
With the other sections appropriately set, the section that needs to be detailed is
``Networking Options.'' Although it is not the last section in the kernel configuration, it is
certainly the most important for the purposes of this chapter.
Networking Options
The Linux 2.2.x kernel adds significant complexity to the networking options section from
the 2.0.x kernels. The additional options can be daunting, and the help is not always helpful.
Plan to spend some time getting acquainted with this section. Items that are of interest are
highlighted here; some of these items are required, others are recommended, and still others
are optional.
641
Introduction to Firewalls
Items that are marked as not recommended have the potential for weakening your
firewall. If you know that you won't use it, don't install it.
The following parameter is required for programs such as tcpdump:
CONFIG_NETLINK= recommended Part
This first kernel parameter is required for such programs as tcpdump. However, tcpdump V
puts your Ethernet card in promiscuous mode:
Ch
CONFIG_PACKET= not recommended 30
The following parameter requires devices with major number 36 to communicate. If this
option is chosen, Routing messages needs to also be chosen, as does IP: firewall packet
netlink device, which can be used to warn of possible attacks:
CONFIG_NETLINK= recommended
Use of the following parameter requires /dev/route to be created with major 36 so that you
can read routing information:
CONFIG_RTNETLINK= optional
CONFIG_NETLINK_DEV= required
The following is required for packet filters or masquerading, but not for proxy firewalls:
CONFIG_FIREWALL= required/optional
The following is only required to configure an Ethernet card with multiple IP addresses; it
might put your NIC into promiscuous mode:
CONFIG_NET_ALIAS= not recommended
CONFIG_FILTER= not recommended
CONFIG_UNIX= required
CONFIG_INET= required
CONFIG_IP_MULTICAST= optional
To use TOS, verbose route monitoring, or large routing tables, this is required. The next six
parameters depend on this one:
CONFIG_IP_ADVANCED_ROUTER optional
CONFIG_IP_MULTIPLE_TABLES optional
CONFIG_IP_ROUTE_MULTIPATH optional
CONFIG_IP_ROUTE_TOS optional
CONFIG_IP_ROUTE_VERBOSE recommended
CONFIG_IP_ROUTE_LARGE_TABLES optional
Following is the network address translation parameter for routers:
CONFIG_IP_ROUTE_NAT recommended/required
The next two parameters depend on this one:
642 Chapter 30 IP Firewalling and Masquerading
CONFIG_IP_PNP not recommended
CONFIG_IP_PNP_BOOTP not recommended
CONFIG_IP_PNP_RARP not recommended
CONFIG_IP_FIREWALL required
The following parameter requires a device with major 36 and minor 3, plus a program to read
the device:
CONFIG_IP_FIREWALL_NETLINK optional
Only choose this parameter for firewalls, but always choose it for firewalls. It is required for
masquerading. For non-masquerading firewalls, packet filters act on the first packet only, and
others are passed on. Hosts receiving fragments cannot reassemble them without the first
packet, and therefore are discarded in time. However, some hostsÐmost notably Microsoft
Windows and NTÐare susceptible to ``big ping'' attacks, even if they only receive the last
packet. Choose yes for this option:
CONFIG_IP_ALWAYS_DEFRAG recommended/required
The following parameter is required for ipchains REDIRECT targets and for transparent
proxies; otherwise, it is not needed. If in doubt, include
CONFIG_IP_TRANSPARENT_PROXY optional/required
The following is required for masquerading firewalls:
CONFIG_IP_MASQUERADE optional/required
This is required only if you chose CONFIG_IP_MASQUERADE in the preceding example and want
to masquerade ICMP. Without this, ping does not work. Furthermore, MS traceroute,
which uses ICMP rather than UDP, also does not work:
CONFIG_IP_MASQUERADE_ICMP optional/recommended
This parameter requires the ipmasqadm program. The next three options depend on enabling
this. The following require kernel module support:
CONFIG_IP_MASQUERADE_MOD optional/required
CONFIG_IP_MASQUERADE_IPAUTOFW optional
CONFIG_IP_MASQUERADE_IPPORTFW optional
CONFIG_IP_MASQUERADE_MFW optional
CONFIG_IP_ROUTER optional
The following also requires kernel module support:
CONFIG_NET_IPIP optional
The following is useful for multicast, or if the remote end is a Cisco router:
CONFIG_NET_IPGRE optional
CONFIG_NET_IPGRE_BROADCAST optional
The following requires CONFIG_IP_MROUTE:
CONFIG_IP_MROUTE optional
CONFIG_IP_PIMSM_V1 optional
CONFIG_IP_PIMSM_V2 optional
CONFIG_IP_ALIAS not recommended
643
Introduction to Firewalls
This is needed only if you are directly connected to more than 256 hosts; ARPD is also
required:
CONFIG_ARPD optional
CONFIG_SYSCTL
The following requires and CONFIG_PROC_FS, as well as putting a 1 in
/proc/sys/net/ipv4/tcp_syncookies:
CONFIG_SYN_COOKIES recommended
CONFIG_INET_RARP not recommended Part
CONFIG_IP_NOSR recommended (highly) V
CONFIG_SKB_LARGE optional
CONFIG_IPV6 optional Ch
CONFIG_IPV6_EUI64 optional
CONFIG_IPV6_NO_PB optional 30
The ipchains program currently only works for IP, not IPX:
CONFIG_IPX not recommended
CONFIG_IPX_INTERN not recommended
CONFIG_SPX not recommended
CONFIG_ATALK not recommended
CONFIG_X25 not recommended
CONFIG_LAPB not recommended
CONFIG_BRIDGE optional
CONFIG_LLC not recommended
CONFIG_ECONET optional
CONFIG_ECONET_AUNUDP optional
CONFIG_ECONET_NATIVE not recommended
CONFIG_WAN_ROUTER optional
CONFIG_NET_FASTROUTE DO NOT USE
The following is of limited availability:
CONFIG_NET_HW_FLOWCONTROL optional
Following is an alternative to CONFIG_NET_HW_FLOWCONTROL, if you think that your system will
be saturated by high volume traffic:
CONFIG_CPU_IS_SLOW optional
Sixteen options depend on the following parameter, but are omitted for the sake of brevity:
CONFIG_NET_SCHED not recommended
Software Considerations
After the kernel is built, you can look over the system for software that isn't required for
operation. Extraneous software needs to be removed. This is especially true of compilers,
games, and other unnecessary software. The use of the X Window software is discouraged
because this binds to port 6000 and 6010. If you feel that it is needed, consider using
ipchains to deny output on the untrusted network side. This includes nfs and other services
that are not used or needed. If an intruder breaches the firewall, it doesn't make sense to
ipchains see
provide tools to use or services to activate. For more information on the
ipchains How-To at http://metalab.unc.edu/mdw/HOWTO/IPCHAINS-HOWTO.html.
644 Chapter 30 IP Firewalling and Masquerading
What you probably want to have is softwareÐsuch as ipmasq or ipfwadmÐto help manage
the ipchains and other firewall rules.
Tip #160 from
Jack To download a pre-compiled copy of ipfwadm, see:
http://members.home.net/ipmasq/ipfwadm.gz.
This program might or might not be included on the Linux CDs, but it is worth downloading
and installing if you have many rules to track. The secure shell ( ssh) program is also highly
recommended. Running TCP wrappers on ports that are not forwarded and not used (normal
services do not run on a firewall), as well as using tripwire to watch files, is a good idea.
Another good program to add to your arsenal is a Perl program called courtney. This
program watches for port scans.
Other Considerations
A firewall is not to be considered a normal network host, and it is not to be treated like one.
This system should not enable normal users to log in or share files or directories on the
network. The need to use good passwords for the accounts on the firewall, along with the
necessity of using shadow passwords, goes without saying. What might need to be said is that
the firewall is not to have the same password as any other host on your network. The fact that
the host is broken because the attacker broke the password should not automatically provide
access to other hosts on the internal network.
This host needs to also be physically separated from the rest of the hosts and placed in a
secure area where unauthorized individuals cannot gain physical access to it. Any machine to
which a knowledgeable individual has access can be ``broken,'' often in minutes. The case
needs to be locked, and access to the system setup password must be protected.
Port Forwarding
Port forwarding is redirecting a connection from one host to another; this is what proxy
firewalls do well. If you connect from host foo to host bar on port 80, and that port is
redirectedÐby softwareÐto host baz on port 80, host foo thinks it's connected to host bar on
port 80 even though it is actually accessing host baz on port 80. Host baz sees a connection
from host bar. This can work in either direction, permitting inside clients out and outside
clients inÐbut in a controlled environment. Either way, all connections appear to be to and
from host bar.
The ipchains software does not do port forwarding. Although one of the ipchains targets is
REDIRECT, this target is for local redirection, not redirection to another host. If you use
ipchains and want to do port forwarding, you'll need to useipportfw. The ipmasqadm
645
A Simple Packet Filtering Firewall
program is a useful wrapper to ipportfw. Outside connections coming in might require redirection,
particularly if the internal (trusted) network is being masqueraded.
A Simple Packet Filtering Firewall
The next few sections step you through building a very simple firewall with ipchains. This
firewall is not adequate for use as isÐyou'll need to determine if this is what you need. But it
Part
does give you a good idea about how to plan and implement a simple firewall, including how
to write the ipchains rules. In the real world, it's just not this easy; this is only a chapter, V
although it deserves a book.
Ch
To begin, you'll need to know something about the network from which you're connecting, 30
and about the network to which you're connecting. The following assumptions are valid for
the rest of this section:
& Internal (trusted) NetworkÐ209.191.169.128/25
& External (untrusted) NetworkÐ209.191.169.0/25
& Bastion hostÐfoo, with foo1/foo2 interfaces, 209.191.169.1/209.191.169.129.
Planning
You can start from one of two general policies. The overall policy can be either ``Permit
everything that is not specifically prohibited'' or ``Prohibit everything that is not specifically
permitted.'' Because the former is easier, your general firewall policy will be ``Prohibit."
The network is set up internally as trusted, and no services will be run from inside for now.
All the services that the company wants to provide to the Internet will reside on the untrusted
network: anonymous FTP, HTTP, and so on. You might see the untrusted portion of the
company's network referred to as the DMZ, the demilitarized zone, in some textbooks. This
is because it is similar to the front lines in a battle; if the bad guys are going to show their
faces, this is where they will try to penetrate. This has the disadvantage of having more hosts
to monitor for intrusions, but it has the advantage of not allowing an intruder into your
internal net via port forwarding.
Because this network is considered low risk, the decision has been made to run mail on the
firewall, with a pop server for users to get mail whether they are at the office or home ( imap
will not be run for security reasons). The smtp and popd services will later be moved inside via
port forwarding. DNS will be run from inside, but will only service the internal network.
Primary DNS will be provided externally by the Internet provider.
Allow internal clients to use standard services on the Internet, except nntp. The following
summarizes the policy you're going to implement:
Summary:
Default policy: prohibit
Anon FTP: external (deny incoming in to the firewall)
http: external (deny incoming in to the firewall)
ssh will be used: deny incoming telnet
646 Chapter 30 IP Firewalling and Masquerading
smtp: firewall (future: port forward to internal machine)
popd: firewall (future: port forward to internal machine)
DNS: internal (no external access)
also stop incoming pings
ipchains General
In order to understand how to proceed with ipchains, you need to understand how ipchains
works. The next few sections walk you through some of the finer points. Most ipchains text
makes the assumption that all packets run through the chains, or list of rules. In fact, however,
ipchains only sees a packet if that packet is the first or only packet. Subsequent packet
fragments do not traverse the chains. The reason for this is simpleÐa host cannot reassemble
the fragments into a packet until it has the first packet. If this packet is denied, the others time
out and are dropped.
Tip #161 from
Jack When you think chain, think of a logically grouped list of rules. Each rule in a chain is
a test to apply against the IP header for a match.
The chains contain rules, numbered from one. As you will see, some rules can be referred to
either by rule specification or by rule number.
A rule specification is the set of conditions that the packet must meetÐthe test. The same basic
rule can exist in multiple chains, so the chain argument is normally required.
There are seven variations on the ipchains command line. The first six contain a command
as the first argument. All six variations accept options as a final argument.
The commands are as follows (all commands are preceded by a hyphen):
& -A (append)ÐThis takes a chain name and a rule specification as mandatory
arguments.
& -D (delete)ÐThis takes a chain name and a rule specification as mandatory arguments.
& -C (test/check)Ð -s, -d, -p, and -i are required. This takes a chain name and a rule
specification as mandatory arguments.
& -I (insert)ÐAn extension of append, but is placed ahead of the rule that is referenced.
& -R (replace)ÐAn Insert and Delete. This takes as mandatory arguments a chain name,
a rule number, and a rule specification.
& -D (delete)ÐThis takes as mandatory arguments a chain name and rule number (this is
a variation on the preceding delete command, where the rule number is known).
& -F (flush)ÐDelete all rules.
& -Z (zero)ÐZero counters.
647
A Simple Packet Filtering Firewall
& -N (new)ÐCreate a user-defined chain. This requires a chain name but otherwise
works on all chains.
& -X (delete a user-defined chain)ÐThis requires a chain name, and the chain must be
empty, but otherwise works on all chains.
& -P (policy)ÐThis takes as mandatory arguments a chain and a target.
& -M (masquerade)ÐThis requires the following mandatory arguments:
. -LÐList Part
. -SÐSet V
tcp, tcpfin, udp
The masquerade command, as opposed to the MASQ target, requires either -L or -S. Ch
The -S command requires three arguments: the tcp (TCP session), tcpfin (TCP 30
session after receiving a FIN packet), and UDP timeout values in seconds.
& -h (help)ÐThe seventh variation on the ipchains command line is help, which takes
no commands, only accepts one option, and optionally accepts one argument.
This option lists the usage argument (it can take the argument icmp to provide a list of
ICMP code and type names that it knows can be used as arguments).
ipchains Options
A number of options are available for ipchains. These include some options to save
mistyping a second rule when it is the same as the first but in the opposite direction, as well as
a way to reverse the meaning of a parameter. Where address masks are specified, the mask can
be either of the following types: /N or N.N.N.N.. Addresses can also be hostnames. Ports can
be either numbers or service names.
The -b option enables you to specify one rule with a source and a destination address, but to
have ipchains also build a rule with the addresses reversed.
The ! can be used with a number of options to reverse the meaning. The options include the
following:
& -p proto ÐProtocol. Can accept ! (as in -p ! icmp) to match all but icmp messages; or
it can accept all to match all protocols.
& -s address ÐSource address. Can optionally take !, a netmask, or a port. Note that an
-s is not specified. Because
address of 0/0 matches all addresses and is the default if
ICMP doesn't use ports, you can follow -s with either an ICMP name, as listed by
ipchains -h icmp, or a type number. If you use a name, you cannot also use -d code.
& -d address ÐDestination address. Same criteria as for -s. If you use -s and specify an
ICMP type number, you can use -d and specify the code.
& -i name ÐInterface name. Can accept !. Also accepts a +suffix on the interface name to
ppp+ is all PPP interfaces (ppp0-pppN).
signify all interfaces of that type; that is,
& -j target ÐTarget for rule (user-defined chain name or special value), if it matches. If
special value is REDIRECT, port can be included.
& -m mark ÐNumber to mark on matching packets.
648 Chapter 30 IP Firewalling and Masquerading
& -nÐNumeric output of addresses and ports. By default, ipchains tries to resolve
them.
& -lÐLog matching packets. These are logged by the kernel.
& -oÐOutput matches to netdev, the userspace device.
& -t and xorÐMasks for TOS field. Used to manipulate the TOS field.
& -vÐVerbose mode. Outputs the interface address, rule options (if any), TOS masks,
and packet and byte counters.
& -xÐExpand numbers. When packet and byte counters are displayed, do not use the
abbreviations K, M, or G, but display all zeros.
& -fÐSecond and further fragments. Can be preceded by not.
& -yÐMatches TCP packets that have the SYN bit set. Can be preceded by !.
The following are valid ICMP types and subtypes (indented under the main type):
Echo -- reply (pong)
Destination -- unreachable
Network -- unreachable
Host -- unreachable
Protocol -- unreachable
Port -- unreachable
Fragmentation -- needed
Source -- route -- failed
Network -- unknown
Host -- unknown
Network -- prohibited
Host -- prohibited
TOS -- network -- unreachable
TOS -- host -- unreachable
Communication -- prohibited
Host -- precedence -- violation
Precedence -- cutoff
Source -- quench
redirect
network -- redirect
host -- redirect
TOS -- network -- redirect
TOS -- host -- redirect
649
A Simple Packet Filtering Firewall
Echo -- request (ping)
Router -- advertisement
Router -- solicitation
Time -- exceeded (ttl ± exceeded)
Ttl -- zero -- during -- transit
Ttl -- zero -- during -- reassembly
Part
Parameter -- problem
V
Ip -- header -- bad
Ch
Required -- option -- missing
30
Timestamp -- request
Timestamp -- reply
Address -- mask -- request
Address -- mask -- reply
Table 30.1 lists the values that you need to use if you want to implement routing priorities
based on the Type of Service (TOS).
Table 30.1 TOS Masks
TOS Name Value Example Uses
Minimum Delay 0x01 0x10 ftp, telnet, ssh
Maximum Throughput 0x01 0x08 FTP-data
Maximum Reliability 0x01 0x04 snmp, DNS
Minimum Cost 0x01 0x02 nntp, email
The TOS is only usable if you compiled support into the kernel ( CONFIG_IP_ROUTE_TOS).
The -o option requires kernel support ( CONFIG_IP_FIREWALL_NETLINK) and a device with
major 36 and minor 3.
Built-in Chains
The three built-in chains in ipchains are input, forward, and output. Other user-defined
chains can be created and destroyed; these three, however, cannot be destroyed, and must
always contain at least one rule. By default, these rules are all DENY.
As packets are received, they traverse these chains, rule by rule, in the following order: input,
forward, output. They continue in the chain until a match is encountered. When a match is
encountered, the chain is interrupted until the target is evaluated. If no target exists, the chain
continues with the next rule.
650 Chapter 30 IP Firewalling and Masquerading
A rule does not have to have a target. Perhaps you want to know how many packets match a
certain rule. As the rule is matched, the rule counter is incremented. Combined with the
counter for the chain, you can see how many of the packets that traversed the chain matched
any particular rule.
If there is a target, ipchains evaluates the target. The target can either be a user-defined
chain name or a special value. If it is a user-defined chain name, ipchains immediately
transfers to that chain and begins traversing it. If no matches are found in the user-defined
chain, ipchains returns to the chain that sent it and continues with the next rule in the chain.
Tip #162 from
Jack Think of user-defined chains as subroutines, and of the targets sending them as
GOSUBS.
If there is a target and it is not a user-defined chain name, it must be one of the following
special values:
& ACCEPTÐThis match is okay; jump to the next chain.
& DENYÐQuietly drops the packet on the floor.
& REJECTÐSame as DENY, but generates an ICMP destination not reachable response.
& MASQÐMasquerade: Forward and user-defined chains only.
& REDIRECTÐInput and user-defined chains only; performs local redirection.
& RETURNÐJumps immediately to the end of the chain.
Tip #163 from
Jack Think of the special values as a GOTO. This terminates the current chain and either
starts on the next packet (DENY, REJECT) or sends the packet to the next built-in chain
(or the calling chain, if you've jumped to a user-defined chain).
User-Defined Chains
User-defined chains provide a way to group rules logically. These chains are called from
built-in chains as targets. At any point in the chain, you can call a user-defined chain. When a
user-defined chain terminates with no matches, it returns to the next argument in the calling
chain.
When creating user-defined chains, names can be up to eight characters long. Names are
lowercase because uppercase is not used, but is rather reserved for future use. The name
cannot be one of the built-in names or special values.
651
A Simple Packet Filtering Firewall
How ipchains Works
When ipchains looks at an IP header, the following occurs, in this order:
& checksum performedÐThe packet is accepted and passed or denied and dropped.
& sanity check ÐThis step looks for malformed packets and drops them.
& input chain ÐIf it is not DENY or REJECT, continue to the next step.
& demasqueradeÐResponses to masqueraded hosts have address rewritten; otherwise, Part
skip.
V
& routingÐSends the packet to local process or forward chain.
Ch
& local process ÐThe interface changes to lo and, if it is destined for a local process,
30
traverses the output and input chains; otherwise, it only traverses the output chain
where the local process handles it.
& localÐIf the packet went through the local process, but did not originate
locallyÐthat is, if it came from a remote host but was processed locally for forwarding
(proxy processing, port forward, and so on)Ðand the final destination is remote, local
sends it to forward chain; otherwise it is sent to output chain, where, if it is notDENY
or REJECT, it is passed to the local host.
& forward chain ÐThis is the chain for all packets that are using this host as a gateway
to another remote host.
& output chain ÐThis is for all the packets that are leaving this host.
local process is confusing; but if you think of HOST
The preceding description of
and LOCALHOST as two different hosts, it makes more sense. To go from HOST to
LOCALHOST, you must traverse all the rules (except forward), leaving HOST and
entering LOCALHOST, and then going the other way for processes that are destined
for remote hosts.
Simple Firewall Policies
Now you're ready to get down to specifying what you want to filter. There are a few things to
keep in mind. While you are making changes to rules, you can change /proc/sys/net/ipv4/
ip_forward from 1 to 0 to turn off forwarding. This prevents things from slipping through
while you're making changes. This is also the first place to look if nothing is passing through
your firewall when you expect it to.
If you change all the built-in chain policies to DENY or REJECT, make sure that you do
not specify rules that require lookups. Use IP addresses, not hostnames.
652 Chapter 30 IP Firewalling and Masquerading
Keep in mind, also, that rules are matched in order. The first rule to match with a special
value terminates that chain (except as explained previously), so be careful about which rules
come first. Take a look at the following rule: ipchains -I input 1 -j REJECT (insert, as the first
rule,REJECT). Because this rule has no -s, it applies to all addresses. Furthermore, because it
has no -i, it applies to all interfaces. Finally, with no -p, it applies to all protocols. Essentially,
this rule rejects everything, even messages from localhost.
So to start, always keep your policies simple and build on them from there.
What to Filter and Where
Sometimes you need to think about not only what you want to filter, but where. Suppose you
don't want to answer ping packets for any host. You can handle this in two waysÐbut only
one makes good sense. The first way is to deny or reject echo-requests as follows:
ipchains -A input -s echo-request -j DENY.
The second way is to deny or reject the echo-response before it goes out:
ipchains -A output -s echo-response -j DENY.
Although both of these work, they have very different effects. You probably want to use the
first method. Normally, your first response is correct. But be aware that both of these prevent
the sender from receiving a reply. If the ping packet happens to be a big-ping and is being
sent to a vulnerable host inside your network, the first method works (if you compiled
CONFIG_IP_ALWAYS_DEFRAG). The second method does not.
What Not to Filter
Some administrators believe that ICMP packets are not that important. They equate icmp
with ping. Unfortunately, a number of other important network messages use ICMP. The
destination-not-reachable messages travel this way, so you won't receive them. Although
TCP normally times out, these ICMP messages still need to be passed. Your OpenLinux
system, for example, uses ICMP messages to set the maximum transmission unit (MTU). For
Ethernet, this is normally 1500 for maximum throughput. Fragmenting causes more delays
than dropping the MTU. So Linux sets the Don't Fragment (DF) bit. If a host or router
needs to fragment the packets, it can't because the DF bit is set, so it drops the packet and
sends an ICMP message. Linux drops the MTU and tries again until it can pass packets. If
you don't accept ICMP messages, your connections with some hosts might be excruciatingly
slow.
Most administrators are also aware that DNS uses UDP, and they want to block TCP on port
53 (the DNS port). But when DNS needs to do a zone transfer or other large data transfer, it
switches to TCP.
The bottom line is this: If you experience network problems after implementing certain rules,
back the rules out until you stop experiencing the problem, and then reimplement them one
at a time, with logging turned on, until you can isolate the problem.
653
A Simple Packet Filtering Firewall
Implementing the Policies
Now all that's left is to implement the policies. Declare a few variables because this reduces
errors. Follow these steps:
1. Use the following names, which are similar to the ones in the previous summary:
foo1int=209.191.169.1/25 # this is eth0
foo2int=209.191.169.129/25 # this is eth1
fooall=209.191.169.0/24 # so you can specify both subnets Part
2. Stop pings that come from outside: V
ipchains -A input -s echo-request -i eth1 -j DENY #incoming
Apings from outside denied Ch
30
3. Now pass all traffic from inside out (exceptnntp):
ipchains -A input -s foo1int ! 119 -d 0/0 -i eth0 -j ACCEPT
4. Block those pesky services that are a common security problem or that you just don't
want, such as telnet, ftp, http, and imap:
ipchains -A input -i eth1 -s 0/0 -d fooall 23 -j DENY
ipchains -A input -i eth1 -s 0/0 -d fooall telnet -j DENY
ipchains -A input -i eth1 -s 0/0 -d fooall 80 -j DENY
ipchains -A input -i eth1 -s 0/0 -d fooall imap -j DENY
In the previous step, you really don't need to specify eth1 because if it came from
eth0, you've already accepted it previously and, therefore, you won't get the chance
to deny it.
5. Be sure to accept DNS (on foo1int only), ssh, and returns on most upper ports (and
on both networks) as well as smtp and pop-3:
ipchains -A input -p all -s 0/0 -d foo1int domain -j ACCEPT
ipchains -A input -p tcp -s 0/0 -d fooall 22 -j ACCEPT
ipchains -A input -p tcp -s 0/0 -d fooall 1024:5999 -j ACCEPT
ipchains -A input -p tcp -s 0/0 -d fooall 6010: -j ACCEPT
ipchains -A input -p tcp -s 0/0 -d fooall 25 -j ACCEPT
ipchains -A input -p tcp -s 0/0 -d fooall pop-3 -j ACCEPT
6. Verify that localhost packets are Okay:
ipchains -A input -i lo -j ACCEPT
7. Block the rest:
ipchains -P input DENY
8. You want to forward both waysÐthe input rules are taking care of most of the work:
ipchains -P forward ACCEPT
9. Make some optimizations:
654 Chapter 30 IP Firewalling and Masquerading
. Minimum delay for Web, telnet, and ssh traffic:
ipchains -A output -p tcp -d 0/0 80 -t 0x01 0x10
ipchains -A output -p tcp -d 0/0 telnet -t 0x01 0x10
ipchains -A output -p tcp -d 0/0 22 -t 0x01 0x10
. Maximum throughput for FTP-data:
ipchains -A output -p tcp -d 0/0 ftp-data 0x01 0x08
. Maximum reliability for smtp:
ipchains -A output -d tcp -d 0/0 smtp 0x01 0x04
. Minimum cost for pop-3:
ipchains -A output -d tcp -d 0/0 pop-3 0x01 0x02
. To finish up (if you got this far, just let it go), enter the following rule:
ipchains -P output ACCEPT
There's just one more rule that you might want in order to stop IP spoofing (by someone
from the outside who is pretending to be you). No one should connect to the external
interface, claiming to be from the inside. The following rule enables you to log these attempts
and reject them:
ipchains -I input 1 -i eth1 -s foo1int -l -j REJECT
Testing the Policies
The easiest way to test the policies is to make up a few cases that you do and don't want to get
through and use the ipchains check option (-C). You'll probably want to use the -v option to
get a verbose listing of the check. This tells you if the packet is passed. Remember that the-C
option requires -s with address and port, -d with address and port, -p, and -i in addition to
the chain name.
Often, enabling logging for some rules can help. This needs to be used judiciously or you will
have very large logs, very quickly. One rule at a time is a good idea.
Monitoring
Remember to look through the logs from time to time, particularly if you put in rules that are
designed to detect attacks. Furthermore, just in case someone does break in, you might want
to have a trusted internal host doing your syslogging for you.
tcpd, tripwire, courtney, and all the other tools don't do any good if they are not properly
used and checked. The first thing an attacker does is look for these things. Time is what your
firewall is buying you. However, time works for the attacker and against you. The best time to
catch an attack is before the penetration occurs, when your network is being probed. To
enter, an attacker must find your weaknesses. This way, you will have warning. It might not
be much, though.
655
Configuring IP Masquerading
Under Attack
After you've been probed or attacked, you need to have a plan to deal with the situation: Do
you allow it to continue in an attempt to track it, or do you stop it cold? Do you alert the
authorities? (It is a crime.) Do you have the authority to contact the police? If not, who does?
TheseÐand many moreÐquestions are all part of a good network security policy. But it is
just as important to practice what you'll do when the time comes. Note that I said when, not
Part
if. Some Internet sites that carry tools for crackers are a constant source of new exploits, so
V
finding out how the attacker entered, if possible, is essential to preventing a recurrence.
Ch
30
Network Security Policy
Possibly the greatest failing of most companies is the lack of a coherent network security
policy. A good policy explains clearly the network policy, penalties for violation of the
policies, and enforcement guidelines (what happens to violators). This policy must apply to
allÐequally. A synthesis of this policy (two or three sentences) needs to be posted on Web
and FTP sites to warn guests.
But the document should not focus on the prohibitive/punitive side. Rather, a good policy
needs to cover the actions that are to be taken, as well as when and by whom, when your
security is at risk. Furthermore, it needs to cover a reasonable timeline for the eventuality that
someone, somewhere will at least attempt to penetrate your security. When an attack is
discovered, what actions are taken? This includes discovery of the attempt after the fact as
well as during an ongoing attack. A number of responses are possible in each situation,
but because time is often the key element, those who are involved must know their part.
In many larger companies, emergency response teams have been designated. A good
reference is RFC-2196, ``Site Security Handbook''. All current RFCs are available from
http://www.rfc-editor.org/rfc.html.
Configuring IP Masquerading
In a nutshell, IP masquerading enables you to provide access to a TCP/IP network outside
your own network through a single server using a single IP address.
The ``single IP address'' phrase is what makes this statement so significant. With a block of IP
addresses, any network can contain multiple machines, each with its own legitimate IP
address, and have a gateway machine provide each machine behind it with access to other
networks (see Figure 30.1).
One common networking scheme does not follow this recipe, however: home and small
business networks, most of which do not have legitimate blocks of IP addresses to work with,
and most of which have only dial-up access to other TCP/IP networks, in particular the
Internet (see Figure 30.2).
656 Chapter 30 IP Firewalling and Masquerading
All of these machines must use a
Figure 30.1 valid IP address to access the Internet
This network
207.179.18.3
has a block of
IP addresses.
207.179.18.4 207.179.18.2
Frame Relay/
xDSL/
Fiber Optic
Connection Internet
Service Internet
207.179.18.5 Hub 207.179.18.7 Central Provider
Server/
Gateway
207.179.18.1
207.179.18.6
These machines can all use bogus
Figure 30.2
IP addresses and still access the Internet
This network
192.168.1.3
has one IP ad-
dress and IP
masquerading.
192.168.1.4 192.168.1.2
Modem/
PPP
192.168.1.1 Connection Internet
Service Internet
192.168.1.5 Hub 192.168.1.7 Central 207.179.18.3 Provider
Server/
Gateway
(Masquerades
192.168.1.2-7
192.168.1.6 as if they are
all 207.179.18.3)
Using IP masquerading, a home or small business network can offer all connected machines
access to an outside network such as the Internet using a single IP address on a single server
machine. All Internet-bound network packets are masqueraded as if they were being sent
from the server that is running IP masquerading. The server maintains the information
necessary to route the returning network packets back to the machines that are supposed to
receive them.
Note that although IP masquerading is commonly seen on servers connected to the Internet
through a modem via PPP, nothing says that you cannot do the same thing using Ethernet
657
Configuring IP Masquerading
connections to the Internet. For example, a company might want to have a training room full of
computers networked to the Internet, but it might not want to give up a whole block of IP
addresses for the task. If you give the machines bogus IP addresses and masquerade them through
one machine, only one IP address is necessary.
Required Kernel Components
The Linux kernel that ships with OpenLinux is precompiled with everything you need to
Part
utilize IP masquerading. The kernel for Red Hat does not have support; therefore, it requires
you to configure a custom kernel as described in this chapter. Debian 2.1 also requires a new V
kernel to support IP masquerading and firewalling.
Ch
30
Tip #164 from
Jack For detailed information on IP masquerading, see the official Web site at http://
members.home.net/ipmasq/.
All you need to do to implement it is load some kernel modules and set up some simple
firewall rules.
If you want to compile your own kernels, check out the following list of items that need to be
compiled into the kernel for IP masquerading to work (the names are listed as they are seen
during a make config, make xconfig, or make menuconfig procedure). Figure 30.3 provides a
view of the X-based kernel configuration screen.
Figure 30.3
Using the
Xconfig pro-
gram eases
your work load
when creating
a new kernel.
& Prompt for development and/or incomplete code/drivers
& Loadable module support
& Networking support
& Network firewalls
& TCP/IP networking
658 Chapter 30 IP Firewalling and Masquerading
& IP: forwarding/gatewaying
& IP: firewalling
& IP: masquerading (experimental)
& IP: ipautofw masquerade support (recommended)
& IP: ICMP masquerading (recommended)
& IP: always defragment (recommended)
& Dummy net driver support (recommended)
Of course, you can compile in other options as either modules or built-in support; the
preceding list is just for masquerading features.
Setting Up
IP masquerading is accomplished through the packet filtering firewall capabilities of modern
Linux kernels. When an outgoing network packet hits the firewall machine (the server with
IP masquerading set up on it), the firewall rewrites elements of each package to make them
look as if they are emanating from the firewall and not the machine behind the firewall. The
return packets are modified to go back to the machine that sent the original outgoing packets.
The machine on the Internet to which the packets were sent thinks that they were sent from
the firewall machine, and the host behind the firewall thinks that the return packets were sent
from the machine on the Internet. To both ends of the transaction, nothing odd seems to be
going on at all.
Some services cannot be accessed without special handling. The following modules were
?
made for just such services. These modules are all offered on default installations of Linux:
See ``Working with Kernel Modules,'' p. 303
& ip_masq_ftp.oÐGives the machines on your network the capability to use the File
Transfer Protocol (FTP) in nonpassive mode through the server that is doing the
masquerading.
& ip_masq_irc.oÐGives the capability to use Internet Relay Chat (IRC) clients through
the masquerading server.
& ip_masq_quake.oÐAllows you to participate in three-dimensional Internet gore-fests
as typified by the game Quake.
& ip_masq_raudio.oÐAllows masqueraded machines behind your server to receive
audio and video streams over the Internet through Real Networks clients.
& ip_masq_vdolive.oÐPermits audio/video streams used by the various VDO-Live
clients to pass through the masquerading server.
Your first task is to create a network with bogus IP addresses. The blocks of addresses
allocated in RFC1597 as being reserved for private networks are listed here in Table 30.2.
659
Configuring IP Masquerading
Table 30.2 Private Network Address Blocks
Class Address Start Address End Mask
A 10.0.0.0 10.255.255.255 8
B 172.16.0.0 172.31.255.255 16
C 192.168.0.0 192.168.255.255 24
Part
Unless literally hundreds of computers are to be set up on the network, odds are that the V
Class C block is the one that is used. Ch
You use the ipchains utility to set up the forwarding rules for IP masquerading by using the 30
following syntax:
ipchains -P forward DENY
ipchains -A forward -j MASQ -s startIP/mask -d 0.0.0.0/0
The first command denies everything from being forwarded through the firewall (a security
measure). In the second command, you replace startIP with the beginning IP address from
Table 30.1 for the class of addresses you are using and replace mask with the corresponding
mask value.
For example, if a Class C address block is being used by the computers on the network, you
use the following commands:
ipchains -P forward DENY
ipchains -A forward -j MASQ -s 192.168.0.0/24 -d 0.0.0.0/0
In most cases, this is all that you need to do to set up the firewall rules. Combined with the
modules listed earlier, most of the common Internet services are covered by this
configuration.
On occasion, you might need to set up special forwarding of specific ports on the
masquerading server. One case in particular is when Cu-SeeMe video conferencing is being
used. The ports on the masquerading server through which Cu-SeeMe operates need to be
forwarded through a specific ``control'' port; otherwise, its network packets cannot be
exchanged with the outside world.
This special form of port forwarding can be implemented using a utility called ipmasqadm.
This utility takes general network ports and funnels them through specified control ports.
The ipmasqadm utility does not care what it is forwarding; it just takes network packets from
a range of ports and sends them out a different port.
For instance, even though ipmasqadm does not care that Cu-SeeMe is running through ports
7648 and 7649, it nonetheless forwards all the traffic that is moving through them to port
7648 if the following command is executed:
# Set up Cu-SeeMe firewall hole
ipmasqadm autofw -A -r udp 7648 7649 -c udp 7648 -u
660 Chapter 30 IP Firewalling and Masquerading
This command line tells ipmasqadm to add ( -A) an autoforward rule (autofw) that takes a
range (-r) of UDP ports, starting at 7648 and ending at 7649, and sends their network traffic
to UDP port 7648, which is specified as their new control port ( -c). It also says not to require
that remote hosts connect within 15 seconds of triggering this new control port ( -u).
If you have this requirement (if you plan on masqueraded machines being able to use
Cu-SeeMe through the masquerading server), you need to execute this command with
?
/etc/rc.d/rc.local or some other startup script.
See ``Understanding the Boot Process,'' p. 234
The preceding command, which sets up port 7648 as a control port for ports 7648-
7649, essentially opens that range of ports to the outside world. The risk in doing so
might not be critical, but any access from the outside is a potential risk and needs to
be implemented carefully. Do not open ports to the outside world with ipmasqadm
unless you have a real need to do so.
One point that is important to note about IP masquerading is that, for the most part, it is a
one-way street. You can go from the machines on your network out to the Internet, but
without special packet forwarding rules, you cannot get into the systems behind the
masquerading server from the Internet.
What does this mean? It means that a machine that has a bogus IP address and is sitting
behind a masquerading server cannot serve Web pages out on the Internet, nor can it accept
direct FTP connections from the outside, nor Telnet, nor ssh/secure transactions, and so on.
Thus services that are to be accessible from outside your network probably need to be set up
on the same server machine that is running IP masquerading.
Troubleshooting IP Masquerading
My application doesn't work through IP masquerading!
Make sure your application is supported by IP masquerading. The following programs are
known to work: HTTP Web browsers like Netscape and MSIE; ftp, Real Audio, telnet, SSH,
POP3 and SMTP (email), NNTP (USENET news), ping, traceroute, IRC, CU-SeeMe (with
special modules loaded). Many Internet phone programs either don't work or work only
partially.
I don't have a permanent Internet connection, can I still use IP masquerading?
Yes, IP masquerading works just as well with dynamically assigned IPs as with static IPs.
661
Troubleshooting IP Masquerading
What type of network connection/device do I need?
IP masquerading does not depend on your connection, just so long as you can establish a
TCP/IP connection to the net. You can use standard analog modems, cable modems, ISDN
modems or routers, bascially anything as long as you can establish TCP/IP.
Does IP maquerading work with just OpenLinux?
No, IP masquerading is not tied to a single Linux distribution, or to Linux in general.
See the IP masquerading HOWTO, at http://www.redhat.com/mirrors/LDP/HOWTO/mini/ Part
IP-Masquerade.html, for more information. V
Ch
IP masguerading doesn't work.
An open ended questionÐthings to check is to make sure you've updated the kernel to work 30
with IP masquerading. If you've upgraded the kernel recently then check your configurations.
Make sure you have enable IP forwarding. Check the HOWTO. Search through the IP
Masquerade Mailing list archive at http://home.indyramp.com/lists/masq/, for more
information.
CHAPTER
31
Connecting to the Internet
In this chapter by Steve Burnett
Understanding the Requirements for SLIP and PPP 664
Using dip to Automate SLIP Operations 664
Using diplogin to Provide SLIP Service 670
Using PPP 671
Project: Configuring PPP with KDE 677
664 Chapter 31 Connecting to the Internet
Understanding the Requirements for SLIP and
PPP
The Linux kernel supports two serial line protocols for transmitting Internet Protocol (IP)
traffic: Serial Line Internet Protocol (SLIP) and Point-to-Point Protocol (PPP). These
protocols were developed as a poor man's alternative to expensive leased-line setups for
getting Internet connectivity. Anyone with a reasonably high-speed modem and a service
provider that supports these protocols can get his or her Linux machine IP-connected for a
very low cost compared to leased-line systems. SLIP drivers for Linux were available soon
after Linux was first released, and PPP support was added shortly thereafter. Although PPP
has come to dominate the industry, the SLIP configuration is still helpful for supporting
older systems.
You need to make sure that a few features are set up in your Linux kernel or configuration
?
files. TCP/IP networking must be enabled, and the loopback interface should be configured.
See ``Configuring the Software Loopback Interface,'' in p. 624
You'll want the IP address of your Domain Name Service (DNS) server to be included in
your /etc/resolv.conf file to make accessing other machines besides your dial-up host
convenient. If your dial-up link is slow or error-prone, you might want to run a name server
on your Linux box to cache any DNS lookups and decrease the amount of DNS IP traffic on
?
your dial-up link.
/etc/resolv.conf
?
See ``The File,'' p. 797
See ``Using the named Daemon to Set Up the Server,'' p. 798
Using dip to Automate SLIP Operations
Linux offers a number of programs to manage your SLIP operations. dip, the Dial-Up IP
Protocol driver, is one of the most versatile tools. It provides a scripting language for
automating control of the modem and automatically sets up the SLIP network interface and
kernel routing tables. You can use dip to initiate SLIP connections or provide dial-up SLIP
service to other machines. The syntax for dip is as follows:
dip [-tvi] [-m mtu] [scriptfile]
Table 31.1 describes dip's most common command-line arguments.
Table 31.1 dip Common Command-Line Arguments
Argument Description
-a Prompts you for username and password.
-t Runs dip in command mode. Command mode gives you full access to
everything dip can do, allowing you to initiate a SLIP connection manually.
Using dip to Automate SLIP Operations
665
Table 31.1 dip Common Command-Line Arguments
Argument Description
-v Displays the current error level when used with -t.
-i Tells dip to operate in input mode. This flag is used when dip provides SLIP
service for others dialing into your machine.
-m mtu Forces dip to use the specified MTU value.
scriptfile Specifies the name of the dip script to run.
Using dip in Command Mode
Invoking dip with the -t option places it in command mode. This mode lets you control dip
directly and is an excellent tool for developing and debugging dip scripts. The following
Part
shows you what dip's command mode looks like:
V
$ /sbin/dip -t
DIP: Dialup IP Protocol Driver version 3.3.7i-uri (17 Apr 95) Ch
Written by Fred N. van Kempen, MicroWalt Corporation. 31
DIP>
From theDIP> prompt, you can run any dip command by typing it and pressing Enter. The
help command displays a list of the available commands. Invoking a command with incorrect
arguments displays a brief usage statement for that command. Table 31.2 describes the
commands available for use at the command mode prompt or in dip scripts.
Table 31.2 Commands Available in dip
Command Description
chatkey keyword [code] Adds a keyword and error-level code to the set of error codes
returned by the dial command. You can use the chatkey
command to detect when your modem returns BUSY, VOICE, or
other specific messages.
config [arguments] Allows you to directly manipulate the SLIP interface dip
provides. This command normally is disabled because it's a
severe security risk. The source code file command.c must be
modified slightly to enable this command.
databits bits Sets the number of bits that can be used as data in each byte.
This command accommodates 6- and 7-bit dial-up connections.
default Causes dip to set a default route in the kernel routing table
pointed at the remote host.
dial num Dials the specified telephone number.
echo on|off Turns echo on or off. Echo on makes dip display what it's
sending to and receiving from the modem.
flush Throws away any responses from the modem that haven't been
read yet.
666 Chapter 31 Connecting to the Internet
Table 31.2 Commands Available in dip
Command Description
get $var Sets the variable $var to either the constant ask or remote
constant specified, prompts you for a value, or takes the next
word from the serial line and assigns it to $var.
goto label Jumps to the specified label in the dip script.
help Displays a listing of available commands in command mode.
if $var op number Performs a conditional branch in a goto label script. $var must
be one of $errlvl, $locip, or $rmtip. The number must be an
integer, and the following operators are available and have their
traditional C language meanings: ==, !=, << , >>, << =, and >> =.
init initstring Sets the initialization string sent to the modem by the reset
command to initstring.
mode SLIP|CSLIP Sets the protocol mode for the connection and makes dip go
into daemon mode. This command normally causes dip to go
into daemon mode and not return control to the script or the
DIP>> command line.
modem HAYES Sets the modem type. Only the HAYES modem type is now
supported. ( HAYES must be capitalized.)
netmask mask Sets the netmask for the routes dip installs to mask.
parity E|O|N Sets the parity of the serial line to even, odd, or none.
password Prompts you for a password and retrieves it in a secure manner.
This command doesn't echo the password as you type it.
print Echoes text to the console dip started on. Variables included in
the text are replaced with their values.
portdev Sets the device dip uses.
quit Exits the dip program.
reset Sends the init string to the serial line.
send text Sends the specified text to the serial line. The traditional C-style
backslash sequences are properly handled.
sleep num Delays processing for the specified number of seconds.
speed num Sets the serial line speed.
stopbits bits Sets the number of stop bits used by the serial port.
timeout num Sets the default timeout to the integer value num. This value is
measured in seconds.
term Makes dip go into terminal emulation mode. This mode allows
you to interface directly with the serial link. Pressing Ctrl+>
returns you to the DIP>> prompt.
wait word num Makes dip wait for the specified word to arrive on the serial line
with a timeout of num seconds.
Using dip to Automate SLIP Operations
667
dip also provides a number of variables for your use. Some of them, such as the local and
remote IP addresses, you can set; others are read-only and are used for diagnostic and
informational purposes. Each variable begins with a dollar sign and must be typed in
lowercase letters. Table 31.3 lists these variables and their uses.
Table 31.3 Variables Provided by dip
Variable Description
$local The host name of the local machine.
$locip The IP address assigned to the local machine.
$remote The host name of the remote machine.
$rmtip The IP address of the remote machine.
Part
$mtu The MTU value for the connection.
$modem The modem type being used (read-only).
V
$port The name of the serial device dip is using (read-only). Ch
$speed The speed setting of the serial device (read-only). 31
$errlvl The result code of the last command (read-only) executed. Zero indicates success;
any other value is an error.
Tip #165 from
Steve Setting the $local or $remote variable to a host name causes dip to resolve the
host name to its IP address and store that in the respective IP address variable. Setting
the variable this way saves a step in the scripts you write.
You can't set the read-only variables directly by using the get command.
Using dip with Static IP Addresses
Assigning individual IP addresses to each machine that uses a SLIP provider is very common.
When your machine initiates a SLIP link to the remote host, dip configures the SLIP
interface with this known address. Listing 31.1 shows a dip script using static IP addresses for
initiating a SLIP link from notebook.afakecompany.com to server.afakecompany.com.
Listing 31.1 A Sample dip Script for Using Static IP Addresses over SLIP
# Connect notebook to server using static IP Addresses
# Configure Communication Parameters
668 Chapter 31 Connecting to the Internet
Listing 31.1 continued
port /dev/cua1 # use modem on /dev/cua1 serial line
speed 38400
modem HAYES
reset # Send initialization string to modem
flush # Throw away modem response
get $local notebook # Set local IP address
get $remote server # Set remote IP address
# Dial number for server modem
dial 555-1234
if $errlvl != 0 goto error # If the dial command fails, error out
wait CONNECT 75
if $errlvl != 0 goto error # If we don't get a CONNECT string
# from the modem, error out
send \r\n # Wake up login program
wait ogin: 30 # Wait 30 seconds for login prompt
if $errlvl != 0 goto error # Error out if we don't get login prompt
send Snotebook\n # Send SLIP login name for notebook
wait ssword: 5 # Wait 5 seconds for password prompt
if $errlvl != 0 goto error # Error out if we don't get password
send be4me\n # Send password
wait running 30 # Wait for indication that SLIP is up
if $errlvl != 0 goto error # Otherwise error out
# We're in, print out useful information
print Connected to $remote with address $rmtip
default # Make this link our default route
mode SLIP # Turn on SLIP mode on our end
# Error routine in case things don't work
error:
print SLIP to $remote failed.
Tip #166 from
Steve Tracking SLIP accounts separately from other types of accounts can be difficult.
Traditionally, UNIX user accounts are assigned login names with all lowercase letters.
Using the client machine name with a capital S added to the front as the login name
for that machine's SLIP account makes tracking it easier and avoids login name
collisions with normal user accounts.
The script in Listing 31.1 initializes the modem and sets the local and remote IP addresses for
the SLIP link. If you use host names here, dip resolves them to their IP address equivalents.
The script then dials the modem and works its way through the login sequence. When the
script is used to log in and ensure that the SLIP link is up on the remote host, it has dip
configure the routing table and then switch the serial line into SLIP mode.
If an error occurs, the error routine at the end of the script prints a warning message and
aborts the script. dip is excellent about leaving the serial line in a reasonable state when done
with it.
Using dip to Automate SLIP Operations
669
Using dip with Dynamic IP Addresses
As SLIP became more popular, the task of managing IP addressees for SLIP clients got more
and more difficult. This problem got worse when terminal servers supporting SLIP came into
use. At that point, you might be assigned any one of a range of IP addresses, depending on
which port the terminal server received your call, requiring changes in dip that captured IP
address information from the incoming data on the serial line. Listing 31.2 shows a dip script
that captures the local and remote IP addresses from the serial line.
Listing 31.2 A Sample dip Script for Dynamic IP Addresses
# Connection script for SLIP to server with dynamic IP address
# assignment. The terminal server prints out:
#
# remote address is XXX.XXX.XXX.XXX the local address is YYY.YYY.YYY.YYY Part
# Set the desired serial port and speed. V
port /dev/cua1
speed 38400 Ch
31
# Reset the modem and terminal line.
Reset
flush
# Prepare for dialing.
dial 555-1234
if $errlvl != 0 goto error
wait CONNECT 60
if $errlvl != 0 goto error
# We are connected. Login to the system.
login:
wait name: 10 # Log in to system
if $errlvl != 0 goto error
send Snotebook\n # Send user ID
wait ord: 10
if $errlvl != 0 goto error
send be4me\n # Send password
if $errlvl != 0 goto error
get $remote remote 10 # Get remote IP address
if $errlvl != 0 goto error
get $local remote 10 # Get local IP address
if $errlvl != 0 goto error
done:
print CONNECTED to $remote with address $rmtip we are $local
default # Set routing
mode SLIP # Go to SLIP mode
goto exit
error:
print SLIP to $host failed.
exit:
670 Chapter 31 Connecting to the Internet
The script in Listing 31.2 uses get $remote remote 10 to watch the serial line and to capture
the first thing that looks like an IP address in the $remote variable. The command times out
in 10 seconds with an error if it doesn't see an IP address.
Using diplogin to Provide SLIP Service
The dip program automates starting SLIP links from the client machine. Linux also supports
incoming dial-up SLIP links. A few packages are available for doing this job as well. Here,
you'll use the diplogin program, which is really just another name for dip.
Providing SLIP service to others requires that you create a specific account for each
person on your Linux box and configure that account correctly. You also need to write an
/etc/diphosts file with appropriate information for each host you're providing SLIP service
for.
Creating SLIP Accounts
You can manually create the SLIP account or use the adduser script with appropriate
responses to each question. The following is a sample /etc/passwd entry for
notebook.afakecompany.com in the passwd file on server.afakecompany.com:
Snotebook:IdR4gDZ7K7D82:505:100:notebook SLIP Account:/tmp:/sbin/diplogin
/tmp is recommended for use as the home directory for SLIP accounts to minimize security
risks by preventing SLIP users from writing files into sensitive areas of your file system by
default. Make sure that you use the correct path to the diplogin program.
Using the /etc/diphosts File
The /etc/diphosts file controls access to SLIP on your machine and contains the connection
parameters for each account allowed to use SLIP. It contains lines that look similar to the
following:
Snotebook::notebook.afakecompany.com:notebook SLIP:SLIP,296
The fields in this file are the user ID, secondary password, host name or IP address of the
calling machine, an informational field not currently used, and the connection parameters for
this account. The connection parameters field contains the protocol (SLIP or CSLIP) and the
Maximum Transmission Unit (MTU) value for this account.
If the second field isn't empty, diplogin prompts for an external security password when the
specified account logs in to your machine. If the response from the remote host doesn't match
the string in this field, the login attempt is aborted.
Thediplogin program requires root privileges to modify the kernel routing table. If
dip setuid root, you can't use a link between dip and
you aren't running
diplogin. You must make a separate copy of dip called diplogin and have its suid
root.
671
Using PPP
That's all it takes. Setting up SLIP accounts and the /etc/diphosts file completely
configures your system to support incoming SLIP links.
Using PPP
Point-to-Point Protocol (PPP) is another protocol for sending datagrams across a serial link.
Developed after SLIP, PPP contains a number of features SLIP lacks: PPP can automatically
negotiate options such as the following:
& IP addresses
& Datagram sizes
& Client authorization
Part
It can also transport packets from protocols other than IP.
V
Automating PPP Links with pppd and chat
Ch
PPP operates in two parts: the PPP driver in the Linux kernel and a program called pppd that 31
the user must run. The most basic means of using PPP is to log in manually to the remote
host by using a communications program and then manually start pppd on the remote and
local hosts. It's much more convenient to use a chat script with pppd that handles the modem,
logging in to the remote host, and starting the remote pppd. Before you dive into pppd, take a
quick look at chat.
Using the chat Program
chatis a program for automating the interaction between your computer and a modem. It's
used mainly to establish the modem connection between the local and remote pppd daemon
processes. The syntax for chat is as follows:
chat [options] script
Table 31.4 lists the command-line options for the chat program.
Table 31.4 chat Command-Line Options
Option Description
-f filename Uses the chat script in the specified file
-l lockfile Makes a UUCP-style lock file by using the specified lock file
-t num Uses the specified number as the timeout in seconds for each expected string
-v Makes a chat log of everything it sends and receives to syslog
script Specifies the chat script to use
672 Chapter 31 Connecting to the Internet
You can't use the -f option and specify a chat script at the same time; they're mutually
exclusive. If you use the -l option for chat, don't use the lock option with pppd because the
lock file created by chat causes pppd to fail, thinking that the modem device is already in use.
Tip #167 from
Steve When you're debugging chat scripts, run tail -f /var/adm/messages on one
-v option when you run chat in another. You can then
virtual console and use the
watch the conversation chat is having as it comes up on the first virtual console.
Creating chat Scripts
chatscripts consist of one or more expect-reply pairs of strings separated by spaces. The chat
program waits for the expected text and sends the reply text when it receives it. Optional
subexpect-subreply pairs can be included in the expect portion, separated by hyphens.
The following is a typical chat script for logging in to a Linux machine:
ogin:-\r\n-ogin: abbet1 word: costello
chat should wait for the string ogin: to appear. If chat times out before
This script says that
chat should send a carriage return and linefeed and wait for the string
receiving this string,
ogin: again. When chat sees the ogin: string, it sends abbet1, waits for the word:, and sends
costello in response.
Tip #168 from
Steve Include only the text necessary in expect strings to positively identify what you're
looking for. This will minimize the chance of getting a mismatch or having your script
blow up because of garbled text. For example, use ogin: instead of login: and
word: instead of password:.
chat normally sends a carriage return after each reply string unless a \c character sequence
ends the string. Carriage returns aren't looked for in expect strings unless explicitly requested
with the \r character sequence in the expect string.
Most modems can report why a call failed when they get a busy signal or can't detect a carrier.
You can use the abort expect string to tell chat to fail if it receives the specified strings.
Multiple abort pairs are cumulative. The following script is an example of using the abort
expect string:
abort 'NO CARRIER' abort 'BUSY' ogin:±ogin: ppp word: be4me
This chat script makes chat fail if it receives NO CARRIER or BUSY at any point during the
script.
673
Using PPP
chat recognizes a number of character and escape sequences, as outlined in Table 31.5.
Table 31.5 Character and Escape Sequences Recognized by chat
Sequence Description
BREAK Makes chat send a break to the modem when used as a reply string. This special
signal normally causes the remote host to change its transmission speed.
'' Sends a null string with a single carriage return.
\b Indicates the backspace character.
\c Suppresses the newline character sent after a reply string and must be at the end
of the reply string.
\d Makes chat wait for one second.
\K Provides another means of specifying a break signal. Part
\n Sends a newline character. V
\N Sends a null character. Ch
\p Pauses for 1/10th of one second.
31
\q Prevents the string it's included in from showing in the syslog file.
\r Sends or expects a carriage return.
\s Sends or expects a space character.
\t Sends or expects a tab character.
\\ Sends or expects a backslash character.
\ddd Specifies an ASCII character in octal.
^C Specifies the control character represented by C.
Tip #169 from
Steve You can use theabort string to prevent low-speed calls on your high-speed modem.
To do so, configure your modem to return a CARRIER 28800 string when it makes a
connection and add abort CARRIER 14400 to your chat script. This way, chat fails
if your modem connects at 14400 bps instead of 28800 bps.
Using PPP with chat
The pppd program has command-line options that control all aspects of the PPP link. The
syntax for the pppd command is as follows:
pppd [options] [tty_name] [speed]
Table 31.6 describes the most commonly used options.
674 Chapter 31 Connecting to the Internet
Table 31.6 Frequently Used pppd Command-Line Options
Option Description
device Uses the specified device. pppd adds /dev/ to the string if needed.
When no device is given, pppd uses the controlling terminal.
speed Sets the modem speed.
asyncmap map Sets the async character map. This map specifies which control
characters can't be sent through the connection and need to be escaped.
The map is a 32-bit hex number in which each bit represents a
character. The 0th bit (00000001) represents character 0 600.
auth Requires the remote host to authenticate itself.
connect program Uses the program or shell command to set up the connection. chat is
used here.
crtscts Uses hardware flow control.
xonxoff Uses software flow control.
defaultroute Makes pppd set a default route to the remote host in your kernel
routing table.
disconnect program Runs the specified program after pppd terminates its link.
escape c1,c2,... Causes the specified characters to be escaped when transmitted. The
characters are specified by using the ASCII hex equivalent.
file filename Reads pppd options from the specified file.
lock Uses UUCP-style locking on the serial device.
mru num Sets the maximum receive unit to the specified number.
netmask mask Sets the PPP network interface netmask.
passive Makes pppd wait for a valid connection rather than fail when it can't
initiate a connection immediately.
silent Keeps pppd from initiating a connection. pppd waits for a connection
attempt from the remote host instead.
More than 40 other command-line arguments control all aspects of PPP at all levels. Refer to
the man page for information about them.
Thepppd program demands that the file /etc/ppp/options exists, even if it's
pppd, is an excellent place to put options you want
empty. This file, which is read by
pppd to use every time it runs.
You can combine pppd and chat in a number of ways. You can specify all the command-line
arguments for both programs on the command line, put the pppd options in a file, or put the
chat script in a file. The following is a simple example with everything on the command line:
675
Using PPP
$ pppd connect 'chat ``'' ATDT5551234 ogin: notebook word: be4me' \
/dev/cua1 38400 mru 296 lock debug crtscts modem defaultroute
This example runs pppd with a simple chat script that dials a phone number and logs the user
notebook in to the remote host. The device, speed, MRU, and a number of other options are
included.
At the other extreme, you can place most of the options for pppd in a file and have chat read a
script file. The following is the call to pppd:
pppd /dev/cua1 38400 connect 'chat -f server.chat'
The following lines display the contents of the reference file:
# Global PPP Options File
mru 296 # Set MRU value
lock # Use UUCP locking Part
crtscts # Use hardware handshaking V
modem # Use modem control lines
defaultroute # Make PPP set up default route Ch
pppd reads this file and processes the options it finds within. Any text following a # character 31
is treated as a comment and ignored.
The following chat script sets a number of abort strings, dials the phone number, waits for a
login prompt, and logs the ppp user in to the remote host with the password ppp-word:
abort 'NO CARRIER'
abort 'BUSY'
abort 'VOICE'
abort 'CARRIER 2400'
"'' ATDT555-1234
CONNECT '\c'
ogin:-BREAK-ogin: ppp
word: ppp-word
Providing PPP Service
Configuring your Linux machine to be a PPP server is even easier than setting up a SLIP
server. It requires only one new account and a shell script that properly runs the pppd
program.
To start, create an account called ppp with an /etc/passwd entry that looks like this:
$ ppp:*:501:300:PPP Account:/tmp:/etc/ppp/ppplogin
Then set the password appropriately. The UID (501) and GID (300) numbers need not be
the same. You can also assign one account to each PPP client you have, if you want. The
/etc/ppp/ppplogin file should be an executable script such as the following:
#!/bin/sh
# PPP Server Login Script
# Turn off messages to this terminal
mesg n
# Turn off echoing
676 Chapter 31 Connecting to the Internet
stty -echo
# Run pppd on top of this sh process
exec pppd -detach silent modem crtscts
pppd with the -detach argument to keep pppd from detaching itself from
This script executes
pppd detaches, the script exits, causing the dial-up connection to close. The
the tty it's on. If
silent option makes pppd wait for the remote pppd daemon to initiate the link. The modem
options make pppd monitor the modem control lines, and crtscts makes pppd use hardware
flow control.
That's all there is to it. When a user logs in to your machine with the proper user ID and
password, the PPP link is established automatically on your box.
Keeping Your PPP Link Secure
Keeping your PPP link secure is very important. Allowing anyone to connect your machine to
a PPP server or allowing anyone to connect to your PPP server is as bad as letting anyone put
a machine directly on your network. PPP provides a direct IP connection, effectively putting
the machines on both ends of the link on the same network.
Two authentication protocols have been developed to make PPP more secure: the Password
Authentication Protocol (PAP) and the Challenge Handshake Authentication Protocol
(CHAP). While a PPP connection is being established, each machine can request the other to
authenticate itself. This process allows complete control of who can use your PPP service.
CHAP is the more secure protocol and is discussed here.
CHAP uses a set of secret keys, which are text strings kept secret by the owners of the machines
using CHAP, and an encrypted challenge system to authenticate each other. A useful feature
of CHAP is that it periodically issues challenge requests as long as the PPP link is up. This
feature, for example, can detect intruders who have replaced the legitimate user by switching
phone lines.
The secret keys for CHAP are stored in /etc/ppp/chap-secrets. To use authentication on
your PPP link, add the auth pppd and add the appropriate information
option to the call to
for the host being authenticated into the chap-secrets file. The following is a sample
chap-secrets file for notebook.afakecompany.com:
# notebook.afakecompany.com CHAP secrets file
# client/server/secret/IP addr
notebook.afakecompany.com server.afakecompany.com ``It's Full of Stars''
Anotebook.afakecompany.com
server.afakecompany.com notebook.afakecompany.com ``three stars''
Aserver.afakecompany.com
notebook.afakecompany.com ``three stars'' afakecompany.com
Each line contains up to four fields: the client host name, the server host name, the secret key,
and an optional list of IP addresses that this client can request be assigned to it. The client
and server designations in this file are determined by the host that makes the authentication
request (the server). The client has to respond to the request.
677
Project: Configuring PPP with KDE
This file defines three different CHAP secrets. The first line is used when
server.afakecompany.com requests CHAP authentication from notebook.afakecompany.com;
the second is used for the reverse situation. The last line defines a wildcard situation for the
client. This situation allows any machine that knows the proper secret key to make a PPP link
to notebook.afakecompany.com. *
The wildcard designator ( ) can be used in the client or
server field.
Careful management of the chap-secrets file gives you complete control over the machines
that can access your PPP server and the machines that you can access with PPP.
Project: Configuring PPP with KDE
KDE is another graphical user interface, presented in Chapter 26, ``Working with KDE.''
Among the tools included as part of KDE is the application, which allows you to use a GUI to Part
set up your dial-up configuration instead of the command line. To use , you need to have the V
X Windows System running on your Linux system and have KDE installed and running as
Ch
well. You also need a dial-up PPP account somewhere and the configuration information for
that account. 31
For this project, you're going to define a PPP dial-up for Mike James, an employee of A Fake
Company. He'll be calling into his company's network from home. Just follow these steps:
1. To start kppp, open a terminal and enter the following at the command prompt:
kppp
The kppp window appears, as shown in Figure 31.1.
Figure 31.1
In the kppp
main window,
you can
choose which
PPP configura-
tion to use.
In the kppp main window, you can choose from multiple PPP configurations, such as
to an ISP or to an office network. For the two dial-up providers, you are unlikely to
have the same name and password, and the modems for the two will be different. For a
home system, two users might have separate accounts for the same ISP.
2. Click the Setup button to display the kppp Configuration window (see Figure 31.2).
3. Click the New button to display the New Account window (see Figure 31.3).
4. Click the Dial tab to display the Dial pane of the New Account window.
5. For Mike's work configuration, enter the following information:
Connection Name: mikework
Phone Number: 555-5555
678 Chapter 31 Connecting to the Internet
Figure 31.2
In the kppp
Configuration
window, you
can choose
your PPP con-
figuration.
Figure 31.3
In the New Ac-
count window
of kppp, you
can define new
accounts.
Authentication: Terminal-based (click the downward-pointing arrow and choose
Terminal-based from the drop-down menu).
Leave the IP pane alone because A Fake Company uses Dynamic IP Address
allocation.
6. Click the DNS tab to display the DNS pane of the New Account window, as shown in
Figure 31.4.
7. Enter the following information:
Domain Name: afakecompany.com
DNS IP Address: 192.168.0.1
8. Click the Add button to save the IP address and move it to the DNS Address List field.
9. Click OK. The New Account window closes, and the kppp Configuration window
displays the name mikework (see Figure 31.5).
10. Click OK to close the kppp Configuration window and return to the main kppp
window, which now displays the mikework configuration name in the Connect To
field, as shown in Figure 31.6.
679
Project: Configuring PPP with KDE
Figure 31.4
In the DNS
pane of the
New Account
window, you
can set the
DNS server ad-
dress(es).
Part
Figure 31.5
V
The kppp Con-
Ch
figuration win-
dow displays 31
created ac-
counts.
Figure 31.6
You can
choose created
configurations
from the main
kppp window.
11. Click the Connect button to have your computer dial the defined number and start a
PPP session with the parameters given.
PART
VI
Using the Internet
32 Accessing the Network with telnet, ftp, and the r- Commands 683
33 Surfing the Internet with the World Wide Web 701
34 Using Electronic Mail 717
35 Surviving Usenet News 741
CHAPTER
32
Accessing the Network with
telnet, ftp, and the r- Commands
In this chapter by Steve Burnett
Using telnet to Access Remote Computers 684
Using ftp for Remote File Transfer 686
Using the r- Commands 695
The ssh Command 698
Troubleshooting 699
684 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Using telnet to Access Remote Computers
The main advantage of computer networking is the capability to share resources and
information and to access that information from remote locations. Linux provides a robust set
of tools for doing just that. Whereas the World Wide Web lets you access lots of information
in a hypertext format, additional tools allow you to log in to remote computers, transfer files,
and execute remote commands.
The telnet command is a common tool for remote login under most operating systems,
including Linux. telnet gives you a terminal session on the remote computer that allows you
to execute commands as though you were logged in locally.
To log in to a computer via telnet, you must know a valid username and password on the
remote machine. Although some systems do provide guest login capabilities, such capabilities
are very rare due to security concerns. When guest logins are allowed, they almost always
place users in a restricted shell or in a menu system. The idea behind these guest
environments is to provide computer security and protect the system from malicious or
careless unknown users. A restricted shell prevents the users from executing specific
commands; a menu system allows choices from only a predefined set of menus, blocking out
shell access entirely.
telnet also allows users to log in to their own computers from a remote location by entering
their usernames and passwords. This way, users can check email, edit files, and run programs
on their normal computers as though they were logged in locally. However, you have to make
do with a terminal-based environment instead of the X Windows system. telnet provides
only terminal emulation for common terminals such as the DEC VT-100, which doesn't
support graphical environments such as X Windows.
telnet is less common than it was even a couple of years ago, especially with the increased
availability of Web-based front ends for email and other applications. However, for some
situations, telnet is still useful. However, telnet is not a secure or safe tool to use in an open
network such as the Internet. For more information, see Chapter 13, ``Improving System
Security."
telnet Command Summary
The basic syntax for telnet is as follows:
telnet [hostname]
hostname is the name of a remote computer. If you don't specify a remote host, telnet starts
in its interactive command mode. If you give a remote host name, telnet tries to initiate a
session immediately.
telnet accepts several command-line arguments, as listed in Table 32.1.
Using telnet to Access Remote Computers
685
Table 32.1 Command-Line Arguments for the telnet Command
Argument Description
-d Turns on debugging.
-a Attempts automatic login.
-n tracefile Turns on tracing and saves trace data in tracefile.
-e escape_char Sets the escape character for the session to be escape_char. If the
escape_char character is omitted from the argument, this telnet session
doesn't have an escape character.
-l user Sends the username user to the remote system for automatic login. This
argument automatically includes the -a argument.
port Indicates the port number to connect to on the remote system. This
argument is used for specifying different network programs. If it's not
specified, telnet connects to the default telnet port.
Sample telnet Session
It's time to take a walk through a sample telnet session. You start the telnet session by
typing telnet, followed by the host name of the computer you want to connect to. telnet Part
then returns with the message Trying some IP address (where some IP address is the address VI
of the computer you specified). If telnet successfully connects to the computer (that is, the Ch
computer is up and running and the network isn't down), Linux reports Connected to
32
computer name and then tells you that the escape character is some specific character
sequence, almost always Ctrl+]. The escape character specifies the character sequence that
you type to drop from your terminal session into the telnet command interpreter. You do so
if you want to send commands directly to the telnet program and not to your remote
computer session.
After telnet successfully connects to the remote system, the login information is displayed,
and the system prompts you for your login ID and password. Assuming that you have a valid
username and password for the remote system, you successfully log in and can now work
interactively on the remote system.
The following is an example of a telnet session from a Linux computer that connects to
another Linux computer:
$ telnet server.somewhere.com
Trying 127.0.0.1...
Connected to server.somewhere.com.
Escape character is '^]'.
Red Hat Linux release 5.0 (Hurricane)
kernel 2.0.36 on an I686
login:bubba
Password: password
Last login: Mon Nov 15 20:50:43 from localhost
server:~$
server:~$ logout
686 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Connection closed by foreign host.
$
When you're finished with the remote session, be sure to log out. telnet then reports that the
remote session is closed, and you return to your local shell prompt.
Using ftp for Remote File Transfer
File Transfer Protocol (FTP) is a simple and effective means of transferring files between
computers that are connected on a TCP/IP network. FTP allows users to transfer ASCII and
binary files.
During an FTP session, you connect to another computer by using the FTP client program.
From this point, you can move up and down through the directory tree, list directory
contents, copy files from the remote computer to your computer, and transfer files from your
computer to the remote system. Normal file protections apply; you can't get or put a file on
the remote system if you don't have the proper permissions for that file.
To use FTP to transfer files, you must know a valid username and password on the remote
computer. This username/password combination is used to validate your FTP session and to
determine what access you have to files for transfer. Also, you obviously need to know the
name of the system the FTP site you want to access is located on.
You should be aware that FTP clients have different command sets, depending on the
operating system in question. This chapter covers the Linux FTP client; however, when you
start an FTP session with a remote computer, the commands that the remote system expects
might be different. It's rare for FTP systems to be completely incompatible with each other.
Typically, the commands that you normally use are either slightly different or unavailable.
Anonymous FTP
Due to the explosive growth of the Internet, many organizations have made huge repositories
of information available via FTP. These FTP sites have everything from text files to software
of every conceivable type available. But how do you access this enormous storehouse of data if
you don't have an account on the remote computer? Do you need to get an account on every
FTP site to be able to access these files? Not really.
A common convention on the Internet allows guest FTP access to file repositories so that
users can transfer files. This guest access is called anonymous FTP. To use anonymous FTP,
you start an FTP session to the remote system and use anonymous as the username and your
email address as the password. For example, in the following sample, the user named smith on
linux.somewhere.com wants to initiate an FTP session with a common FTP site:
$ ftp ftp.uu.net
ftp.uu.net (login:smith): anonymous
Password: [email protected]
Using ftp for Remote File Transfer
687
Many sites don't allow anonymous FTP. Allowing guest users to connect to your
computer does involve some risk. In cases in which anonymous FTP isn't allowed, the
ftp command fails with a message similar to Login failed - User ``anonymous''
unknown. Sites that do permit anonymous FTP typically place the users in a restricted
directory tree with read-only access. If you're allowed to place files on the remote
computer, you usually can put them in only one directory.
Also, several of the Web browsers (Netscape Navigator, for example) support the FTP
protocol as well as the HTTP protocol. Because several Web browsers also support
email, they can support anonymous FTP connections automatically. Check the user
guide for your Web browser for more information.
ftp Command Summary
The Linux ftp command provides a verbose set of command options in interactive mode. As
mentioned earlier, some remote hosts might not support all these commands. However, you
might not need to use many of them. Table 32.2 lists the commands available while in FTP.
Part
Table 32.2 ftp Commands Available in Interactive Mode
VI
Command Description
Ch
! Escapes to the shell
32
$ Executes a macro
account Sends the account command to a remote server
append Appends to a file
ascii Sets the file transfer type to ASCII mode
bell Beeps when a command is completed
binary Sets the file transfer type to binary mode
bye Terminates the FTP session and exits
case Toggles mget upper- or lowercase filename mapping
cd Changes the working directory on the remote computer
cdup Changes the remote working directory to the parent directory
chmod Changes file permissions of the remote file
close Terminates the FTP session
cr Toggles carriage return stripping when receiving an ASCII file
delete Deletes the remote file
debug Toggles debugging mode
dir Lists the contents of the remote directory (gives size and permissions)
disconnect Terminates the FTP session (same as close)
688 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Table 32.2 ftp Commands Available in Interactive Mode
Command Description
exit Terminates the FTP session and exits
form Sets the file transfer format
get Gets a file from the remote computer
glob Toggles wildcard expansion of local filenames
hash Toggles printing the # character for each buffer transferred
help Prints local help information
idle Gets or sets the idle timer on the remote computer
image Sets the file transfer type to binary mode (same as binary)
lcd Changes the local working directory
ls Lists the contents of the remote directory (gives size and permissions)
macdef Defines a macro
mdelete Deletes multiple files on the remote computer
mdir Lists the contents of multiple remote directories
mget Gets multiple files from the remote computer
mkdir Makes a directory on the remote machine
mls Lists the contents of multiple remote directories
mode Sets the file transfer mode
modtime Shows the last modification time of the remote file
mput Sends multiple files to the remote computer
newer Gets a remote file if the remote file is newer than the corresponding local file
nmap Sets templates for default filename mapping
nlist Lists the contents of the remote directory
ntrans Sets the translation table for default filename mapping
open Connects to the remote FTP site. op is a shorter version of the same command
passive Enters passive transfer mode
prompt Forces interactive prompting on multiple commands
proxy Issues the command on an alternate connection
put Sends one file to the remote computer
pwd Prints the working directory on the remote machine
quit Terminates the FTP session and exits
quote Sends an arbitrary ftp command
recv Receives a file
reget Gets the file restarting at the end of the local file
rstatus Shows the status of the remote machine
Using ftp for Remote File Transfer
689
Table 32.2 ftp Commands Available in Interactive Mode
Command Description
rhelp Gets help from the remote server
rename Renames a file
reset Clears queued command replies
restart Restarts the file transfer at the specified byte count
rmdir Removes a directory on the remote machine
runique Assigns a unique filename to each file received when retrieving multiple files
with the same filename to the same directory
send Sends one file to the remote computer
site Sends a site-specific command to the remote server, one of umask, idle, chmod,
help, group, gpass, newer, or minfo
size Shows the size of the remote file
status Shows the current status
struct Sets the file transfer structure
system Shows the remote system type
Part
sunique Assigns a unique filename to each file sent when sending multiple files with
the same filename to the same directory VI
tenex Sets the tenex file transfer type Ch
tick Toggles printing byte size counter during transfers 32
trace Toggles packet tracing
type Sets the file transfer type
user Sends new user information
umask Gets or sets the umask on the remote computer
verbose Toggles verbose mode
? Prints local help information
As you can see, ftp has quite a few commands. However, you really need to look at only the
ones that you use most frequently.
Starting an FTP Session
You can use the open command to open an FTP session with a remote host. Its syntax is as
follows:
open hostname
You usually need this command only if you're going to connect to more than one site during
an FTP session. If you want to connect to only one computer during the session, just specify
the remote host name on the command line as an argument to the ftp command.
690 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Ending an FTP Session
You can use the close, disconnect, quit, and bye commands to end an FTP session with a
remote computer. The identical close and disconnect commands close your connection to
the remote computer but leave you in the FTP program on your local computer. If you want
to stop what you are doing, but stay connected, the Escape key cancels your last command.
The quit, exit, and bye commands close your connection to the remote computer if one is
active; then they exit the FTP program on your computer. Pressing Ctrl+c quits the FTP
session without waiting for the other system to acknowledge you.
Changing Directories
You can use the cd [directory] command to change directories on the remote computer
during your FTP session. The cdup command takes you to the parent directory of the current
directory. The lcd command changes your local directory so that you can specify where to
find or put local files.
Remote Directory Listing
The ls command lists the contents of a remote directory, just like ls from an interactive
shell. The syntax for ls is as follows:
ls [directory] [local_file]
If a directory is specified as an argument, ls lists the contents of that directory. If a local
filename is given, the directory listing is put into the file you specified on your local
computer.
The dir and ls commands provide a long listing, giving protections, size, owner, and date.
The syntax of the dir command is as follows:
dir [directory] [local_file]
The following is an example of a dir directory listing:
-rw-r±r± 1 root archive 2928 May 17 1993 README
-rw-r±r± 1 root archive 1723 Jun 29 1993 README.NFS
dr-xr-xr-x 2 root wheel 8192 Jun 6 12:16 bind
-rwxr-xr-x 5 root wheel 8192 Aug 2 06:11 decus
drwxr-xr-x 19 root archive 8192 Feb 7 1994 doc
drwxr-xr-x 6 root wheel 8192 Jun 15 15:45 edu
?
dr-xr-xr-x 7 root wheel 8192 Sep 28 09:33 etc
See Chapter 20, ``Managing File Systems'' p. 439
Getting Files from a Remote System
You can use the get and mget commands to retrieve files from a remote computer. The get
command retrieves the file that you specify as an argument ( filename). The following is the
get command's syntax:
get filename [remote_filename]
Using ftp for Remote File Transfer
691
You can also give a local filename, which is the name of the file when it's created on your local
computer. If you don't give a local filename, remote_filename is used.
The mget command retrieves multiple files at once. The syntax for mget is as follows:
mget filename_list
You specify these files by giving a list of filenames separated by spaces or by using a wildcard
pattern to mget. You're prompted for each file. To turn off prompting, use the prompt
command before using mget. In both cases, the files are transferred as ASCII files, unless
you've set the transfer mode to something else.
Sending Files to a Remote System
You can use the put and mput commands to send files to a remote computer. The put
command sends the local file that you specify as an argument. The syntax is as follows:
put filename
The mput command sends a series of local files. The syntax for mput is shown here:
mput filename_list
You specify these files by giving a list of filenames separated by spaces or by using a wildcard Part
pattern to mput. When using mput, you're prompted for each file. To turn off prompting, use
VI
the prompt command. In both cases, the files are transferred as ASCII files, unless you've set
Ch
the transfer mode to something else.
32
Changing the File Transfer Mode
FTP transfers files as ASCII files unless you specify something else. Transferring files this
way is fine for plain text but renders any binary data useless. The ascii and binary
commands set the transfer mode so that you can prevent damage to your binary files.
.tar are
Many files that you'll want to transfer are in binary format. Files ending with
tar command. Files ending in .z and .gz are compressed
archives created with the
with either the compress command or the GNU gzip command, respectively. Files
ending in .zip are compressed archives created with PKZIP. When in doubt, use
binary transfer mode. Using ASCII mode corrupts binary data files.
Checking Transfer Status
When transferring a large file, you might find it useful to have ftp give you feedback on how
far along the transfer is. The hash command causes ftp to print a # character onscreen each
time the transmission of a data buffer is completed. This command works for sending and
receiving files.
692 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Local Commands While in FTP
You can use the ! character to pass a command to the command shell on your local computer
while you're in FTP. This capability can be very useful if you need to do something while
you're in the midst of an FTP session. Suppose that you need to create a directory to hold
received files. If you enter !mkdir new_dir, Linux makes a directory named new_dir in your
current local directory.
A Sample FTP Session
Listing 32.1 shows a short FTP session.
Listing 32.1 Making an FTP Connection and Getting a Directory Listing
$ ftp opus
Connected to opus.
220 opus FTP server (Linux opus 2.0.6 #4 Mon Nov 15 16:01:33 CDT 1999) ready.
Name (opus:smith): smith
Password (opus:smith): password
331 Password required for smith.
230 User smith logged in.
Remote system type is UNIX.
Using ASCII mode to transfer files.
ftp >> dir
200 PORT command successful.
150 Opening ASCII mode data connection for /bin/ls.
total 8
-rw-r±r± 1 root daemon 1525 Sep 29 15:37 README
dr-xr-xr-x 2 root wheel 512 Jun 24 11:35 bin
dr±r±r± 2 root wheel 512 Jun 24 11:18 dev
dr±r±r± 2 root wheel 512 Jun 24 11:24 etc
dr-xr-xr-x 4 root wheel 512 Sep 29 15:37 pub
dr-xr-xr-x 3 root wheel 512 Jun 24 11:15 usr
-r±r±r± 1 root daemon 461 Jun 24 13:46 welcome.msg
226 Transfer complete.
433 bytes received in 0.027 seconds (16 Kbytes/s)
ftp >> get README
200 PORT command successful.
150 Opening ASCII mode data connection for README (1525 bytes).
226 Transfer complete.
local: README remote: README
1561 bytes received in 0.0038 seconds (4e+02 Kbytes/s)
ftp >> quit
221 Goodbye.
$
In Listing 32.1, a user opens an FTP session to the host opus and logs in as smith. The
remote FTP server prompts for the password, which the user types (the password doesn' t
appear onscreen). ftp then logs smith in to the remote system and displays the ftp]] prompt
for interactive mode commands. The user tells ftp to list the remote directory with the dir
command and then transfers the file README with the get command. When finished with the
FTP session, the intrepid user then logs out by using the quit command and is returned to
the local Linux shell prompt.
Using ftp for Remote File Transfer
693
A Sample Anonymous FTP Session
In the preceding section, you saw a user initiate an FTP session with a system and look at
some directories. The user had a valid username and password on the remote system. Now
look at an anonymous FTP session to a major software archive site on the Internet. Listing
32.2 is similar to Listing 32.1, but it has some interesting differences.
Listing 32.2 Performing an Anonymous FTP Connection
$ ftp ftp.uu.net
Connected to ftp.uu.net.
220 ftp.UU.NET FTP server
(Version wu-2.4(1) Wed Nov 17 15:45:10 EST 1999) ready.
Name (ftp.uu.net:bubba): anonymous
331 Guest login ok, send your complete email address as password.
Password: your_email_address
230-
230- Welcome to the UUNET archive.
230- A service of UUNET Technologies Inc, Falls Church, Virginia
230- For information about UUNET, call +1 703 204 8000,
230- or see the files in /uunet-info
230-
230- Access is allowed all day. Part
230- Local time is Wed Nov 17 15:53:02 1999.
230- VI
230- All transfers are logged with your host name and email address.
230- If you don't like this policy, disconnect now! Ch
230- 32
230- If your FTP client crashes or hangs shortly
230- after login, try using a
230- dash (-) as the first character of your password.
230- This will turn off the informational messages which may
230- be confusing your ftp client.
230-
230-Please read the file /info/README.ftp
230- it was last modified on Mon Nov 15 17:39:53 1999 - 2 days ago
230 Guest login ok, access restrictions apply.
ftp >
ftp > dir
200 PORT command successful.
150 Opening ASCII mode data connection for /bin/ls.
total 4149
drwxr-sr-x 2 34 0 512 Jul 26 1992 .forward
-rw-r±r± 1 34 uucp 0 Jul 26 1992 .hushlogin
-rw-r±r± 1 34 archive 59 Jul 31 1992 .kermrc
-rw-r±r± 1 34 archive 0 Jul 26 1992 .notar
drwx±s±x 5 34 archive 512 Jul 23 19:00 admin
lrwxrwxrwx 1 34 archive 1 Jul 26 1992 archive ->
drwxrws±x 4 0 archive 512 Apr 20 16:29 bin
lrwxrwxrwx 1 34 archive 23 Sep 14 1993 by-name.gz ->
Aindex/master/by-name.gz
lrwxrwxrwx 1 34 archive 23 Sep 14 1993 by-time.gz ->
Aindex/master/by-time.gz
-rw-r±r± 1 34 archive 90112 Apr 26 1991 compress.tar
lrwxrwxrwx 1 0 archive 9 Jul 23 18:50 core -> /dev/null
694 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Listing 32.2 continued
drwxrws±x 2 0 archive 512 Jul 26 1992 dev
drwxrwsr-x 21 34 archive 1024 Sep 29 15:18 doc
drwxrws±x 6 0 archive 512 Apr 14 16:42 etc
lrwxrwxrwx 1 34 archive 31 Dec 8 1993 faces ->
A/archive/published/usenix/faces
drwxrwsr-x 2 34 archive 512 Jul 26 1992 ftp
drwxrwsr-x 4 34 archive 512 Sep 29 10:34 government
drwxrwsr-x 18 34 archive 1024 Sep 29 10:28 graphics
-rw-rw-r± 1 27 archive 798720 Jul 11 20:54 gzip.tar
lrwxrwxrwx 1 34 archive 17 Jul 26 1992 help -> info/archive-
help
drwxrwsr-x 20 34 archive 1024 Dec 2 1993 index
drwxrwsr-x 19 34 archive 512 Sep 29 10:30 inet
drwxrwsr-x 4 34 archive 512 Sep 29 15:36 info
drwxrwsr-x 25 34 archive 512 Sep 29 10:29 languages
drwxrwsr-x 4 34 archive 512 Sep 29 10:28 library
drwx±s±x 2 0 0 8192 Jul 26 1992 lost+found
lrwxrwxrwx 1 34 archive 20 Aug 2 1992 ls-lR.Z ->
Aindex/master/ls-lR.Z
lrwxrwxrwx 1 34 archive 21 Sep 14 1993 ls-lR.gz ->
Aindex/master/ls-lR.gz
lrwxrwxrwx 1 34 archive 21 Aug 2 1992 ls-ltR.Z ->
Aindex/master/ls-ltR.Z
lrwxrwxrwx 1 34 archive 22 Sep 14 1993 ls-ltR.gz ->
Aindex/master/ls-ltR.gz
drwxrwsr-x 24 34 archive 1024 Sep 29 15:10 networking
drwxrwsr-x 2 34 archive 512 Aug 10 09:26 packages
d±xrws±x 17 34 archive 512 Sep 26 12:29 private
drwxrwsr-x 25 34 archive 1536 Sep 29 15:30 pub
drwxrwsr-x 17 34 archive 1024 Sep 29 15:38 published
lrwxrwxrwx 1 34 archive 10 Jul 26 1992 sco-archive -> vendor/sco
drwxrwsr-x 20 34 archive 512 Sep 29 04:18 systems
drwxrwxrwx 14 34 archive 1536 Sep 29 15:36 tmp
lrwxrwxrwx 1 34 archive 17 Jul 26 1992 unix-today ->
Avendor/unix-today
lrwxrwxrwx 1 34 archive 17 Jul 26 1992 unix-world ->
Avendor/unix-world
drwxrwsr-x 36 34 archive 1024 Sep 29 15:29 usenet
drwxrws±x 6 0 archive 512 Oct 22 1992 usr
lrwxrwxrwx 1 34 archive 16 Aug 2 1992 uumap -> networking/
uumap
-rw-rw-r± 1 34 archive 3279895 Sep 28 21:05 uumap.tar.Z
drwxrwsr-x 3 210 archive 2560 Sep 29 15:36 uunet-info
drwxrwsr-x 64 34 archive 1536 Sep 29 10:29 vendor
226 Transfer complete.
3257 bytes received in 0.76 seconds (4.2 Kbytes/s)
ftp >
ftp > cd systems/unix/linux
250-Files within this subtree are automatically mirrored from
250-tsx-11.mit.edu:/pub/linux
250-
250 CWD command successful.
ftp >
ftp > binary
200 Type set to I.
695
Using the r- Commands
Listing 32.2 continued
ftp >> get sum.Z
200 PORT command successful.
150 Opening BINARY mode data connection for sum.Z (80959 bytes).
226 Transfer complete.
local: sum.Z remote: sum.Z
80959 bytes received in 5.6 seconds (14 Kbytes/s)
ftp >> quit
221 Goodbye.
$
In Listing 32.2, an FTP session is initiated with ftp.uu.net, which is a major FTP archive
site on the Internet. The username given at the login prompt is anonymous because this is an
anonymous FTP. For the password, the full email address is used. ftp.uu.net then displays a
welcome banner that gives some information about the archive. In this example, you can see
that the user changes directories, sets the file mode to binary, gets a compressed binary file,
and exits.
Using the r- Commands
ftp and telnet, several other commands allow you to access remote computers
In addition to Part
and exchange files over a network. These commands are known collectively as the r- VI
commands. These commands are generally UNIX-only: The telnet and ssh applications
Ch
were developed to provide more generic cross-platform solutions.
32
The r- commands deserve special notice because one of their features can cause a severe
security loophole if you aren't careful. When you issue an r- command, the remote system
checks a file named /etc/hosts.equiv to see whether your local host is listed. If it doesn't
.rhosts in your home directory on the remote
find your local host, it checks for a file named
machine. The r- command then checks to see whether your local host name is in the .rhosts
file. If your local host is listed in either place, the command is executed without checking for a
password.
Although not having to type your password every time you need to access a remote computer
can be very convenient, it can obviously cause severe security problems. We recommend that
you carefully consider the security implications of using the r- commands before setting up
/etc/hosts.equiv and .rhosts files on your local system.
The rlogin Command
The rlogin command is similar to the telnet command because it allows you to start an
interactive command session on a remote system. The syntax of rlogin is as follows:
rlogin [-8EKLdx] [-e char] [-k realm] [-l user_name] hostname
However, the most common usage is simply this:
rlogin hostname
696 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Table 32.3 explains the various options for rlogin.
Table 32.3 Command-Line Options for the rlogin Command
Option Description
-8 Allows an 8-bit input data path at all times, which allows for formatted ANSI
characters and other special codes to be sent. If this option isn't used, parity bits are
stripped except when the remote stop and start characters are other than Ctrl+s and
Ctrl+q.
-E Stops any character from being recognized as an escape character. When this
option is used with the -8 option, it provides a completely transparent connection.
-K Turns off all Kerberos authentication. It's used only when connecting to a host that
uses the Kerberos authentication protocol.
-L Allows the rlogin session to be run in litout mode. Refer to the tty man page for
more information.
-d Turns on socket debugging on the TCP sockets used for communication with the
remote host. Refer to the setsockopt man page for more information.
-e Used to set the escape character for the rlogin session. The escape character is ~ by
default. You can specify a literal character or an octal value in the form \nnn.
-k Requests rlogin to obtain Kerberos tickets for the remote host in the specified
realm instead of the remote host's realm as determined by krb_realmofhost(3).
-l Allows the remote name to be specified. If available, Kerberos authentication is
used.
-x Turns on DES encryption for all data passed via the rlogin session. Encryption
can affect response time and CPU usage, but it provides increased security.
The rsh Command
The rsh command, an abbreviation for remote shell, starts a shell on the specified remote host
and executes the command, if any, that you specify on the rsh command line. If you don't
give a command to execute, you're logged in to the remote machine by using rlogin.
The syntax of the rsh command is as follows:
rsh [-Kdnx] [-k realm] [-l user_name] hostname [command]
However, the most common usage is this:
rsh hostname [command]
The command argument can be virtually any Linux command that you can enter from the shell
prompt. Table 32.4 explains the command-line options for rsh.
697
Using the r- Commands
Table 32.4 Command-Line Options for the rsh Command
Option Description
-K Turns off all Kerberos authentication. It's used only when connecting to a host that
uses Kerberos.
-d Turns on socket debugging on the TCP sockets used for communication with the
remote host. See the setsockopt man page for more information.
-k Requests rsh to obtain Kerberos tickets for the remote host in the specified realm
instead of the remote host's realm as determined by krb_realmofhost(3).
-l Allows the remote name to be specified. If available, Kerberos authentication is
used, and authorization is determined as with the rlogin command.
-n Redirects input from the special device /dev/null.
-x Turns on DES encryption for all data passed. Encryption can affect response time
and CPU usage, but it provides increased security.
Linux takes the standard input to the rsh command and copies it to the standard input of the
remotely executed command. It copies the standard output of the remote command to
standard output for rsh. It also copies the remote standard error to the local standard error
Part
file descriptor. Any quit, terminate, and interrupt signals are sent to the remote command.
Also, any special shell characters that aren't enclosed within quotation marks, as in >>, are VI
handled locally. If they are enclosed within quotation marks, these characters are handled by Ch
the remote command.
32
The rcp Command
The rcp command, which stands for remote copy, is the last of the r- commands that you
might need to know. It's used to copy files between computers. You can use rcp to copy files
from one remote computer to another, without either the source or destination being on the
local machine.
The rcp command has two forms. The first form is used to copy a file to a file. The second
form is used when copying files or a directory to a directory. The syntax for the rcp command
can be either of the following:
rcp [-px] [-k realm] filename1filename2
rcp [-px] [-r] [-k realm] file(s) directory
Each file or directory argument is either a remote filename or a local filename. Remote
filenames have the form rname@rhost:path, where rname is the remote username, rhost is the
remote computer, and path is the path to the file. The filename must contain a colon.
Table 32.5 explains the arguments for rcp.
698 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
Table 32.5 Command-Line Arguments for the rcp Command
Option Description
-r Recursively copies the source directory tree into the destination directory. For you
to be able to use this option, the destination must be a directory.
-p Tries to preserve the modification times and modes of the source files, ignoring the
umask.
-k Requests rcp to obtain Kerberos tickets for the remote host in the specified realm
instead of the remote host's realm, as determined by krb_realmofhost(3).
-x Turns on DES encryption for all data passed by rcp. Encryption can affect
response time and CPU usage, but it provides increased security.
If the path specified in the filename isn't a full pathname, it's interpreted as being relative to
the login directory of the specified user on the remote computer. If no remote username is
given, the current username is used. If a path on a remote host contains special shell
characters, you can quote it by using \, '', or ' as appropriate. Using these special characters
causes all the shell metacharacters to be interpreted remotely.
rcp doesn't prompt for passwords. It performs its copies via the rsh command.
The ssh Command
ssh (short for secure shell ), like the rsh command, is a program for logging into a remote
machine and executing commands on that remote machine. ssh is designed to replace both
rsh and rlogin by providing the capability to define an encrypted session between two
untrusted systems over an insecure network. One problem with telnet is that, when you log
in to the remote system, the password is sent as ASCII over the network. By watching the
Ethernet packets, someone could collect your login name and password for the remote
system. ssh prevents this from happening by using RSA-based authentication for the
connection. Because of its security, ssh is commonly used by system administrators today.
ssh clients are available for almost all UNIX-related systems in command-line or GUI forms,
and are also available for other operating systems, including Macintosh OS and Microsoft
Windows.
The ssh command is similar to the telnet command because it allows you to start an
interactive command session on a remote system. The syntax of ssh is as follows:
ssh [-a] [-c idea|blowfish|des|3des|arcfour|tss|none] [-e escape_char]
A[-I identity_file] [-l login_name] [-n] [-k] [-V] [=o option] [-p port]
A[-q] [-P] [-t] [-v] [-x] [-C] [-L port"host:hostport] [-R port:host:hostport]
Ahostname [command]
699
Troubleshooting
However, the most common usage is this:
ssh hostname
Table 32.6 explains the various options for ssh.
Table 32.6 Command-Line Options for the ssh Command
Option Description
-a Disables forwarding of the authentication agent.
-c Selects the cipher to use for encrypting the session. idea is the default, arcfour is
the fastest, and none is the equivalent of using rlogin or rsh (no encryption).
-e Sets the escape character for the session.
-f Sets ssh in the background after authentication and forwardings are established.
-i Selects the identity file from which the private key for RSA authentication is read.
-k Disables forwarding of Kerberos tickets.
-l Sets the login name for use to the remote machine.
-n Redirects stdin from /dev/nulls used when ssh runs in the background.
-o Used for user-defined options following the format in the configuration file. Part
-p Sets the port to connect to on the remote host. VI
-q Activates quiet mode, which suppresses all messages except fatal errors.
Ch
-P Uses a nonprivileged port.
32
-t Forces pseudo-tty allocation.
-v Activates verbose mode (useful for debugging).
-x Disables X11 forwarding.
-C Requests compression of all data.
-L Specifies the local port to forward to the designated remote host and port.
-R Specifies the remote port to be forwarded to the local host and designated port.
For more information on security issues, see Chapter 13, ``Improving System Security."
Troubleshooting
I transferred a binary file, but it doesn't work properly. I can't unzip it, un- tar it, uncompress it, or
anything. What should I do?
Make sure that you set the transfer mode to binary. You can do so by using the binary
command at the ftp prompt.
700 Chapter 32 Accessing the Network with telnet, ftp, and the r- Commands
I'm in the process of transferring a large file and want to check the progress.
Use the hash command. ftp prints the # character onscreen after every data buffer that's
processed. The data buffer may vary depending on your version of Linux, but it's typically
1,024, 4,096, or 8,192 bytes.
I was trying to use anonymous FTP, but the site told me that the user anonymous was unknown and
that the login failed.
Either you misspelled anonymous, or the site doesn't allow anonymous FTP. In the latter case,
you must have a valid username and password on the remote computer.
I want to transfer several files, but I don't want FTP to prompt me for each one.
Use the prompt command, which toggles prompting on and off.
I tried to use anonymous FTP, but the site told me that I didn't enter a valid email address as the
password.
In the past, the convention during an anonymous FTP connection was to enter guest as the
password. Now the convention is to enter your email address. Many FTP sites run special
FTP server software that checks the password and makes sure that it's in the form
[email protected]. Try again and make sure that you enter your full email address
correctly.
CHAPTER
33
Surfing the Internet with the
World Wide Web
In this chapter by Steve Burnett
Introducing the World Wide Web 702
Using FTP with a Web Browser 706
Using Archie with a Web Browser 708
Using telnet with a Web Browser 709
Using gopher with a Web Browser 710
Accessing Usenet News with a Web Browser 711
Getting on Mailing Lists 711
majordomo 712
Using Wide Area Information Servers (WAIS) 714
Case Study: A Mailing List Subscription Process 714
702 Chapter 33 Surfing the Internet with the World Wide Web
Introducing the World Wide Web
The Internet is a completely distributed network, which means that your computer is
connected directly not only with the computer down the hallway, but with thousands of
others all over the world. Your computer connects to another computer, which is connected
to other computers, and so on.
To make matters more complex, the Internet is international in scope. Virtually every
country in the world has some form of access to the Internet. For years, many services were
available to get to information (FTP, Gopher, and so on), but none were easy to use. You
might want to use FTP, Gopher, Telnet, WAIS, Archie, or another service. To do so, you
had to have all the appropriate software. Then you had to know what service to use and when.
And so on. Something like the World Wide Web was needed as a form of ``information
navigator'' to make it easier for users to get to information on the Internet.
The Web began as a network and hypertext project at CERN, a European physics research
lab, in 1989. Researchers saw a need for people from any location to be able to share and
exchange information and documents in real-time from any type of computer. They also
wanted a simple and consistent way to handle this information. From this need, the World
Wide Web was born.
The Web uses a set ofhypertext links that allow users to easily navigate among documents,
graphics, files, audio clips, and so on from sites anywhere on the Internet. When you select a
hypertext link in a document, whatever item the link points to is automatically retrieved. One
link at a time, Internet users quickly find their way to the various bits of information they
want.
Understanding the Web's Structure
The Web is based on a client/server model. The client software package (Web browser
software on your computer) contacts a server computer (Web server software) and exchanges
messages with that computer through a set of rules that both client and server understand.
This set of rules is known as a protocol. Web servers and clients communicate through a
protocol known as the Hypertext Transfer Protocol (HTTP). When a Web client program
retrieves a document from a Web server, the programs are probably communicating by using
HTTP. As you'll see later in this chapter, other Internet protocols also may be supported by
the Web server.
Of Clients and Servers
The client/server relationship is an important concept in networking and especially in navigation of the Web. A
server is a computer that offers services for other computers to use. Services can be any kind of program,
routine, or data provided by the server. For example, a server might return information from a database to
which you don't have direct access.
A client is a computer that uses services from a server. The client contacts the server and requests some sort of
service. Many times, a client computer uses special software designed to interact with a specially designed
server program on the server computer.
703
Introducing the World Wide Web
Under this client/server model, people with different computers in different locations can access information on
the same server. You can set up different server computers with different types of data. Because people are
using a client software program to communicate with the server, you can develop a different client program for
each computer platform that they use. That way, people using Windows or a Macintosh can use client software
to access information on a UNIX or Linux server just as easily as UNIX or Linux users can.
To access the Web, you need client software known as a Web browser. A Web browser is a
program that understands how to communicate to a Web server via the HTTP protocol; it
displays information and provides a way to represent hypertext links. Many browsers are
available. The most commonly used browsers are Netscape's Navigator and Microsoft's
Internet Explorer. You can get a browser in any number of ways: You can get it from your
Internet service provider (ISP), buy it in store, download it from the Internet, and so on. After
you install the browser and configure the software with your Internet access information,
you're ready to go.
Understanding URLs
You get information on the Web by using a descriptive address known as a uniform resource
locator (URL, pronounced earl). Think of an URL as a pointer to an object on the Internet that
tells you not only where the object is located, but also what it is named and how to access it.
Everything you access through the Web has an URL.
The syntax of URLs may look intimidating, but it is really quite straightforward. The
following is an example:
http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/Docs/whats-new.html
:
Scary? It's really not that bad. The part to the left of the colon ( ) specifies the access method Part
to get to the data. This access method defines the protocol used to communicate with the
VI
server and also gives a good clue as to the type of interaction that will take place. Table 33.1
Ch
lists several valid access methods.
33
Table 33.1 Valid Access Methods for URLs
Method Description
http Provides interactive hypermedia links to pages written in Hypertext Markup
Language (HTML). This protocol is used to access most Web pages.
wais Accesses a wide area information server (WAIS) site.
gopher Accesses a Gopher server.
ftp Provides an anonymous FTP connection.
telnet Opens a Telnet connection to a site.
news Allows you to read Usenet news.
704 Chapter 33 Surfing the Internet with the World Wide Web
Before the Web
Many services and sources of information existed before the Web. These services use protocols other than
HTTP. However, many Web clients such as Netscape Navigator allow you to access these services from within
the browser. For example, you can transfer files to your computer by using FTP, retrieve documents from
Gopher servers, perform text searches with WAIS (wide area information server), and read Usenet news.
Following the :// in the URL is the host name of the server computer you want to contact.
After the server name is the directory path to the document you want to view or retrieve. This
path depends totally on where the file is located on the remote server. (You might not have a
path in some cases, if the file is in a default directory.) Finally, the filename of the document is
given. This document can be text, a hypermedia document, a sound file, a graphic, or some
other type of file.
So, look again at the example. The URL
http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/Docs/whats-new.html
uses the HTTP protocol to contact the server computer www.ncsa.uiuc.edu and says
that you're interested in the document named whats-new.html located in the directory
/SDG/Software/Mosaic/Docs. The .html extension on the document name tells your Web
client (for example, Netscape Navigator) that this document is written in Hypertext Markup
Language (HTML). HTML is a special syntax used to write hypertext pages for the Web.
Searching the Web
The Web is huge, and it's getting bigger every day. Rather than click through thousands of
pages, you can use search engines to help you find information faster. A search engine is a
program that looks through its database for information that matches your request. Some
search engines, such as AltaVista and Infoseek, search the entire Web and store their
information in huge databases. Other search engines search only a specific Web site.
When you see a Search button at a typical Web site, it's usually only for that Web site. When
you want to search the entire Web, you need a more general search tool. The following list
describes some of the many search engines that scan Web sites across the Internet. Some even
let you search other Internet information sources, like those on Usenet or FTP sites.
& AltaVista (http://www.altavista.com)ÐFor Web and Usenet. You can find
anything, anywhere on the Web or Usenet, but you should narrow your search as
much as possible; you can easily get too many results back!
& Yahoo! (http://www.yahoo.com)ÐFor Web, Usenet, email addresses, current news,
people searches, city maps, and stocks. Yahoo! isn't really a search engine. It's basically
a huge list of Web sites, sorted into categories, that have been submitted by users. It's
useful for common information and for getting an idea of just how muchÐand
variedÐthe information on the Web is. Yahoo! also provides links to search engines.
705
Introducing the World Wide Web
& Infoseek (http://www.infoseek.com)ÐFor Web, Usenet, FAQs, current news, email
addresses, maps, stocks, and company listings. Infoseek contains a search engine and
listing service and is good when you want to search more than the Web or Usenet.
Infoseek has a different search language than many of the other search engines.
& Excite (http://www.excite.com)ÐFor Web, Usenet, and Excite Web site reviews.
Excite does conceptual searching of the Web and is good when you're not sure of the
exact term you need to search for. Because Excite uses a single-site search engine on
many Web sites, it's free.
& Lycos (http://www.lycos.com/)ÐFor Web, FTP sites, and Gopher sites. Lycos has
Yahoo!-like features. It's good for simple searches on common topics. You can search
for sounds, graphics, or subjects.
& Search.Com (http:// search.cnet.com/)ÐFor Web and Usenet. This search engine
also lets you search other search engines such as AltaVista, HotBot, or Infoseek.
Search.Com provides an A-to-Z listing of other search engines and has a handy utility
to suggest which search engines will find what you need.
& Inference Find! (http://www.inference.com/)ÐFor Web only. Not a search engine
itself, Inference Find! groups the results of other search engines and eliminates
repeats. As of this writing, it calls WebCrawler, Yahoo!, Lycos, AltaVista, Infoseek,
and Excite.
& HotBot (http://www.hotbot.com)ÐFor Web and Usenet. HotBot is good for finding
sites that use a particular technology, such as JavaScript or VRML. You can also
narrow your search to a specific geographic location (such as Europe), class of
domains (such as edu), or a single Web site (such as www.apple.com).
Part
Good keywords make your search more effective. You should come up with words that are
unique to what you really want to find. Try to avoid heavily used terms, such as www, Internet,
VI
computer, and so on. If you do need them, combine them with other more specific terms and Ch
Boolean operators to help narrow your search, such as WWW and Search Engines.
33
Most search engines also allow you to use quotation marks (``'') to search for phrases.
Putting quotation marks around a phrase keeps the words together as a phrase.
Without quotation marks, each word in a phrase will be searched for individually.
Check your search engine's help for specific details.
You'll probably find that even a search engine lists too many sites to look through. You can
reduce the number of sites you find by narrowing your search. Correctly using some simple
termsÐAND, OR, and NOTÐcan help you narrow down thousands of sites to just a few.
These terms aren't your everyday AND, OR, and NOT. They come from the symbolic logic
system developed by a nineteenth-century mathematician named George Boole. Boolean
searches use a basic syntax made up of operators and search terms. Because the terms don't
706 Chapter 33 Surfing the Internet with the World Wide Web
work quite the same as in English grammar, make sure that you get them straight. Table 33.2
shows examples of how to use AND, OR, and NOT.
Table 33.2 Useful Boolean Expressions
Expression Description
AND or + Returns pages that contain all your search terms. If all words aren't on the
page, the page isn't displayed. Use AND or + when you have dissimilar terms
and want to narrow the results to a few precise hits. For example, BMW AND
roadster or BMW + roadster displays only pages that contain both BMW and
roadster on the page.
OR Returns pages that contain any of your search terms. Use OR to return pages
with any of the terms listed in your search. For example, BMW OR roadster
displays all pages that contain either BMW, roadster, or both.
NOT Returns pages that don't contain words specified in your search (not
supported by all search engines).
Don't be afraid to experiment. Try several different searches with the same goal in mind so
that you can get a better feel for the results that some of these expressions and your search
words or phrases return. You'll find that some experimentation with search terms will help
you become more adept at narrowing your selections to a manageable size.
Using FTP with a Web Browser
FTP, or File Transfer Protocol, is the method that the Internet uses to exchange files
between computers. No matter what you're searching forÐsoftware, documentation, FAQ
lists, programs, or just about anything elseÐyou probably can get a copy through anonymous
FTP.
Anonymous FTP is a service that lets you retrieve data from around the Internet without
having an account on that machine. By using anonymous FTP, you can access any files that
?
the system administrators on the remote system have made publicly available.
See ``Using ftp for Remote File Transfer,'' p. 686
FTP supports ASCII mode transfers for text files and binary mode transfers for other types of
files. Fortunately, most Web clients automatically determine the file type for you, so you
don't have to worry about it. You usually can determine the type of archive or compression
program that was used on the file by looking at the file extension. Table 33.3 lists the most
common file extensions you'll encounter.
707
Using FTP with a Web Browser
Table 33.3 Common File Extensions on Binary Files Available Through FTP
Extension Description
.Z A file compressed with the UNIX compress program
.z A file probably compressed with the GNU gzip program or the UNIX
compress program
.gz A file compressed with the GNU gzip program
.tar An archive of several files created by the UNIX tar program
.zip An archive of several files created by pkzip
Sometimes you might find files that have been created by more than one of these methods.
For example, the file programs.tar.Z is an archive of several files created by the tar utility
and then compressed with the compress utility.
To use a Web client such as Netscape (which has built-in FTP support) to perform
anonymous FTP transfers, you replace the protocol portion of the URL with ftp. For
example, to start an anonymous FTP session to metalab.unc.edu, you use the following
URL:
ftp://metalab.unc.edu
This URL causes your Web client to try to make an FTP connection to metalab.unc.edu and
log you in as an anonymous FTP session. After your FTP session is established, you can
navigate through directories and transfer files by clicking the hyperlinks displayed.
Part
Tip #170 from VI
Steve Most anonymous FTP servers request that you use your email address as your
Ch
password. If you have problems, check that your email preferences are set correctly in
your browser. 33
To specify a nonanonymous FTP session in Netscape, enter the following:
ftp://[email protected]
username is your username, and ftp.startup.com is the place you want to go. You then are
prompted to enter your password.
When you select a text file to transfer from a remote server in an FTP session, most
Web clients display the file onscreen. You need to save the file to disk via a menu
selection. Some Web browsers allow you to specify loading a file to disk rather than
to the screen.
708 Chapter 33 Surfing the Internet with the World Wide Web
ON THE WEB
For a listing of FTP sites available via the Web, see the following page:
http://hoohoo.ncsa.uiuc.edu/ftp/
Using Archie with a Web Browser
Just like with the Web, one major problem you might have with anonymous FTP is figuring
out where the files that you're interested in are located on the Internet. To help users locate
files, the archie system was created. archie is basically a search engine for anonymous FTP
sites.
Archie is a database query program that contacts anonymous FTP sites around the world and
asks each site for a complete list of all its files. archie then indexes this information in its own
internal database. You can search this database for the location of files on the Internet.
Because updating Archie databases is obviously a time-consuming process, the databases are
updated usually only about once a month. Thus, it's possibleÐalthough unlikelyÐthat the
location archie gives you is incorrect.
Archie is a popular service. The various Archie servers around the world can get very heavily
loaded, and requests can take awhile to complete. Some sites place limits on the number of
simultaneous connections to keep the servers from becoming too slow to use. If you try an
Archie server and find that it's fully loaded, you can either try a different server or wait a few
minutes and try again.
Tip #171 from
Steve Many Archie servers now support inquiries via Web forms. To conduct Archie
searches, go to the following URL for a list of Archie gateways to the Web. From this
page, you can link to many of the mirror sites of the Archie database. Linking to the
site closest to you is usually quickest:
http://archie.emnet.co.uk/
A list of current Archie servers is maintained at:
http://archie.emnet.co.uk/services.html
To connect to one of these servers, telnet to it and log in as archie. Each server is slightly
different, but most are basically the same. After you log in to a server, you get a prompt such
as the following where you can enter your search commands:
archie>
Different servers have different default search values. To determine what the default setup is
for the server that you connect to, use the show search command. The show search command
returns one of the following values:
regex archie interprets your search string as a UNIX regular
expression.
Using telnet with a Web Browser
709
exact Your search string must exactly match a filename.
sub Your search string matches if a filename contains it as a
substring. This search is not case sensitive.
subcase This search is similar to the sub search type, except that
the case of the letters in the string must match.
You can set the desired search type by using the set search command as shown here:
archie > set search search-type
When you have your search set up the way you want it, you use the prog command to search
by filename. For example, the following pair of commands
archie > set search sub
archie > prog linux
performs a search regardless of the case of the Archie database for all files that contain the
substring linux. For each match that Archie finds, it reports the host computer that has the
file, along with the full pathname of the file on that host.
If you get confused or just need some assistance when you're using Archie, type help at the
archie > prompt. You then get information on how to get help in Archie. From the help >
prompt, type a ? to see a list of subtopics on which you can get help.
After you find the information you're looking for, you need to exit Archie by typing exit or
quit at the archie > prompt.
Using telnet with a Web Browser
Part
VI
telnet has been around almost as long as the Internet. By using telnet, you can connect to
Ch
databases, library catalogs, and other information resources around the world. Want to see
what the weather's like in Vermont? Check on crop conditions in Azerbaijan? Get more 33
information about somebody whose name you've seen online? Telnet lets you do this and
more. When you telnet to another computer, you're going across the Internet and logging in
to that machine. You won't find graphics as you do on the Web; telnet is text only.
gopher is another early Internet tool, and many Telnet sites are most easily found
through Gopher menus. See the following section on Gopher.
To start telnet from your browser, enter the URL of the Telnet site you want to go to. For
example, the following command starts a telnet program and takes you to the location you
entered:
telnet://pac.carl.org
From there, you're out of the browser and have entered ``menu land."
710 Chapter 33 Surfing the Internet with the World Wide Web
? See ``Using telnet to Access Remote Computers,'' p. 684
Configuring Netscape to Work with telnet
telnet probably isn't built into your browser. If not, you need to get a telnet program, install it on your
computer, and then configure the browser to use it. The following is a sample set of instructions for configuring
telnet for Netscape:
1. From the Netscape Navigator Options menu, choose Preferences.
2. Select Applications and Directories from the available tabs.
3. Select Browse next to the Telnet Application window.
4. Find and select the telnet executable.
5. Press Enter. Netscape is now configured.
Most Telnet sites are fairly easy to use and have online help systems. Most also work
bestÐand, in some cases, onlyÐwith VT100 emulation. You also might find that many of the
resources are now available on the Web.
Tip #172 from
Steve telnet is no longer quite as common as it once was, because your login name and
password are sent unencrypted and are readable by anyone using a packet sniffer. A
program called ssh is very similar to telnet, but is much more secure.
Using gopher with a Web Browser
gopher is an Internet service that allows you to access information by making selections from
a series of menus. gopher was one of the first Internet services that made a serious attempt at
offering a user-friendly interface.
When you connect to a site that offers Gopher services, you get a menu of available choices.
Each menu is either a file or another menu. You can select your choice from the menu
without having to know the name or IP address of the destination site or the directory and
filenames of the particular information you're asking for. Gopher handles the details for you.
No information resources on the Internet are actually ``gopher-specific.'' Anything you
can get through gopher can be accessed by other means, such as an HTML Web
page, FTP, or telnet. In some cases, sites may have chosen to make resources
available only via gopher for security reasons.
711
Getting on Mailing Lists
To access a Gopher server with a Web browser, change the protocol part of the URL so that
gopher instead of http. For example, the
it says URL for the Gopher server at
metalab.unc.edu is gopher://metalab.unc.edu.
Gopher provides an easy means to navigate the Internet. Unfortunately, the information that
Gopher can retrieve may not be well organized, so finding what you want can be a bit of an
adventure. Because the items in Gopherspace are presented as a set of menus, you sometimes
have to wade through many different menus to get to the file you're searching for. This
problem aside, however, a lot of good information is available through gopher.
One disadvantage of gopher is the lack of a standard subject list for the various gopher servers.
The administrators for each gopher server have organized their information in their own
manner. This means that each gopher server you access has different subjects. If gopher
servers do happen to have some of the same subjects, chances are they aren't named the same
way.
ON THE WEB
For a listing of gopher sites, see the following:
http://dir.yahoo.com/Computers_and_Internet/Internet/Gopher/
Accessing Usenet News with a Web Browser
In the simplest definition, Usenet news (also called netnews or simply news) is a forum for
online discussion. Many computers around the world exchange chunks of information, called
Part
articles, on almost every subject imaginable. These computers aren't physically connected to
the same network; they're logically connected in their capability to exchange data. See
VI
Chapter 35, ``Surviving Usenet News,'' for a complete discussion of Usenet news. Ch
News articles on Usenet are divided into newsgroups by subject. These groups are then 33
?
divided into hierarchies based on very general subject distinctions.
See ``How Usenet Is Structured,'' p. 745
Usenet news has discussion on almost any topic that you can think of. It's a great way to find
and exchange information.
Getting on Mailing Lists
Another avenue for discussion on the Internet comes from email mailing lists. Mailing lists
vary from Usenet news in that the various messages and discussion articles are sent via email
rather than via the Usenet news medium.
Why use a mailing list instead of a Usenet newsgroup? Usually, mailing lists are targeted at a
smaller group of people. Setting up a new newsgroup on Usenet is fairly difficult because
proposal, discussion, and voting periods are required. On the other hand, any system
712 Chapter 33 Surfing the Internet with the World Wide Web
administrator can set up a mailing list. Also, because each mailing list is maintained on one
computer, the system administrator has more control over who can be on the list and can deal with
problem users more effectively. Some mailing lists, such as those that discuss computer security
issues, are restricted to certain people. If you need to be on one of these lists, you have to apply
with the list manager to be allowed to subscribe.
Finding Mailing Lists
As with Usenet news, mailing lists exist on a wide variety of subjects. A complete list of
publicly available mailing lists is posted regularly to the Usenet newsgroup news.answers.
ON THE WEB
You can search for mailing lists via the Web at the following site:
http://www.liszt.com/
Using Mailing Lists
Mailing lists are typically set up by using a mail reflector. A mail reflector is a special email
address that's set up to reflect any mail sent to it back out to a group of people. Usually, two
email addresses are associated with a mailing list: that of the list maintainer and that of the list
itself.
Suppose that there's an email address for the users of widgets. The email address for the list
might be something like [email protected]. If you send an email message to this list
address, it's reflected to all the people who subscribe to the list.
By convention, Internet mailing lists use a special email address for administrative requests,
such as subscribing to the list. This address is usually constructed by adding -request to the
name of the list. So for the imaginary widgets mailing list, the administrative email address
would be [email protected]. All mail that addresses administrative topics
should be sent to the administrative address.
Each mailing list (and Usenet newsgroup) has its own rules and culture. You should become
familiar with the local customs before sending mail out to the list. Usually, you get an
introduction message and possibly a list of frequently asked questions (FAQ) when you
subscribe to a list. The introduction message contains any special rules that apply to the list.
Make sure that you read the FAQ first so that you don't ask the same questions as hundreds of
?
other people.
See ``Netiquette on Usenet,'' p. 752
majordomo
One of the most common mailing list programs available is called majordomo. majordomo is a
Perl script that is commonly used on UNIX-based mail servers. The main Web site for
majordomo is http://www.greatcircle.com/majordomo/, where you can download the script
majordomo 713
and read installation instructions; the FAQ file is located at http://www.greatcircle.com/
majordomo/majordomo-faq.html.
Tip #173 from
Steve Another popular mailing list application is LISTSERV. A FAQ file for the LISTSERV
program is available on the Web at http://www.ou.edu/research/electron/
internet/list-faq.htm.
majordomo, as well as other mailing list managers, does more than simply add and remove
names from a subscription list and send mail to everyone on the subscription list: Each has
several commands that allow you to query the mailing list manager about the lists. Some of
these commands are disabled by the list owner, but they might not be. The best source of
information about how a given mailing list is configured is the list owner; many mailing lists
maintain a Web site about the list.
Table 33.5 provides a summary of commands majordomo responds to. Remember, you should
send these commands to the mailing list manager address for a given mailing list, and not the
mailing list address itself. Also, if you use a signature file in your email, adding a hyphen on
the line just before your signature file, like the example here, prevents the mailing list
manager from trying to read your signature file as a set of commands:
-
This is my signature file. There are many like it....
You can also substitute the word end for the hyphen, like this: Part
end VI
This is my signature file. There are many like it....
Ch
The commands in Table 35.5 should be in the body of the email you send to the majordomo
33
address, not the subject. (If they're in the subject, they don't do anything.) For each
command, items contained in brackets ( []) are optional; be sure to leave out the brackets
around the items when you enter them. An item in angle brackets, such as < address>, is a
meta-symbol that you should replace with appropriate text (and as with options, without the
angle brackets).
Table 33.5 majordomo Command Summary
Command Meaning
subscribe <list> <address> Subscribes you (or your <address> if specified) to
the named <list>.
unsubscribe <list> < address> Unsubscribes you (or your <address> if specified)
from the named <list>.
714 Chapter 33 Surfing the Internet with the World Wide Web
Table 33.5 majordomo Command Summary
Command Meaning
"unsubscribe *'' Unsubscribes you (or your < address>) from all
lists. This command might not work if you have
subscribed using multiple addresses.
get <list> <filename> Gets a file related to < list>.
index < list> Returns an index of files you can ``get'' for
< list>.
which <address> Finds out which lists you (or your <address> if
specified) are on.
who < list> Finds out who is on the named < list>.
info < list> Retrieves the general introductory information
for the named < list>.
intro < list> Retrieves the introductory message sent to new
users. Nonsubscribers might not be able to
retrieve this message.
lists Shows the lists served by this majordomo server.
help Retrieves a general message for how to use
majordomo.
end Stops processing commands (useful if your mailer
adds a signature).
Using Wide Area Information Servers (WAIS)
WAIS (which stands for wide area information servers) is a system for searching a large set of
databases for information. The term wide area implies being able to use a large network, such
as the Internet, to conduct searches by using client/server software.
By using WAIS, you can retrieve text or multimedia documents that are stored on databases
throughout the Internet. You can think of WAIS as being similar to Gopher, except that
WAIS does the searching for you.
Like Gopher, to use WAIS, you need client software or you have to use telnet to connect to
a site that provides public access to a WAIS client. You can use an interactive UNIX WAIS
client known as swais. To use this system, you can telnet to metalab.unc.edu and log in as
swais. You then get a menu of databases that you can search. (which stands for wide area
information servers) is a system for
Case Study: A Mailing List Subscription Process
The most common error for someone new to a mailing list is using the wrong address when
attempting to subscribe. Let's look at an example. Brad has an email account through the
fictitious company justabody.com. He finds a Web site operated by afakecompany.com
715
Case Study: A Mailing List Subscription Process
focusing on a product he's interested in, so he wants to subscribe to the mailing list and find out
more about the product.
The Web site information says the following:
The address to send information to the mailing list is [email protected]. If
you want to subscribe to the list, send email to [email protected] with
``subscribe'' in the body of the message, followed by the list's name, like so:
subscribe coolstuff
Brad isn't paying attention, so he sends the following email:
From: [email protected]
To [email protected]
Subject: subscribe coolstuff
Message content: Please add me to the coolstuff list, thanks.
Brad
The coolstuff mailing list is set up to allow only subscribing members to send mail to the list.
This decision was made to keep spam, or unsolicited commercial email, off the list. So James,
the person at A Fake Company who runs the coolstuff mailing list, gets the following email
from majordomo:
Date: Sun, 25 Jul 1999 21:52:57 -0400
From: [email protected]
To: [email protected]
Subject: BOUNCE [email protected]: Non-member submission from [Brad
Davison < [email protected]>]
]]From [email protected] Sun Jul 25 21:52:56 1999
Received: from mail.justabody.com ([email protected] [348.33.456.112]
(may be forged)) Part
by mail.afakecompany.com (8.8.7/8.8.7) with ESMTP id RWA24521
for < [email protected]>; Sun, 25 Jul 1999 21:52:55 -0400 VI
Received: from localhost (brad@localhost)
by mail.justabody.com (8.8.7/8.8.7) with ESMTP id RWA24521 Ch
for < [email protected]>; Sun, 25 Jul 1999 21:51:25 -0400
Date: Sun, 25 Jul 1999 21:51:25 -0400 (EDT) 33
From: Brad Davison < [email protected]>
To: [email protected]
Subject: subscribe
Message-ID: < [email protected]>
Subject: subscribe
Message-ID: < [email protected]>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Please add me to the coolstuff list, thanks.
Brad
716 Chapter 33 Surfing the Internet with the World Wide Web
James sends Brad email suggesting he read the Web site again on how to subscribe to the
mailing list. Brad tries again:
From: [email protected]
To [email protected]
Subject: subscribe coolstuff
Message content: subscribe
end
Brad receives a Welcome message from majordomo welcoming him to the coolstuff list.
CHAPTER
34
Using Electronic Mail
In this chapter by Steve Burnett
Understanding Email 718
Sending Email with mail 720
Reading Your Mail 723
Printing Mail Messages 726
Getting Help with mail 727
Saving Email to Files with mail 728
Deleting and Undeleting Messages with mail 729
Replying to Email with mail 729
Routing Mail to Others 731
Customizing Your mail Environment 734
Quitting the mail Program 736
Using the elm Mailer 737
Project: Using the Mutt Email Client 739
718 Chapter 34 Using Electronic Mail
Understanding Email
Electronic mail, or email, has taken over a significant chunk of the technological world. Tens of
millions of computer users worldwide have access to electronic mail. A large number of
commercial networks or Internet service providers (ISPs) can give you or your organization
access to electronic mail around the world.
Email is any program that users on a single computer system or a network of systems use to
send and receive electronic messages. At a minimum, you provide the program with the
address of the recipient and the message you want to send. The address includes the login
name of the person who is to receive the mail. If that user is on another system in a network,
the address also includes a means of identifying the target computer system. You either
prepare the message while you're using your email program, or you prepare it beforehand by
?
using a text editor such as vi.
See ``Using vi,'' p. 207
Using electronic mail has several advantages:
& You can send reports, data, and documents that can reach their destination in a matter
of seconds or minutes.
& You don't have to worry about interrupting someone when you send a message, nor
are you necessarily interrupted when you receive messagesÐthat's handled by the
computer system.
& You don't need to play phone tag or make an appointment to communicate with
someone.
& You can send and receive at a convenient time.
When you send email, it's up to the computer system to make the delivery, which can involve
putting your message out on a network to be delivered at some other site. At this point, you
say that the mail has been sent. Soon after that, the message arrives at the recipient's machine.
If the sender and the receiver are on the same computer system, the sending and receiving
take place on one machine. The email system on the target computer verifies that the
addressee exists, and the message is added to a file that holds all the email for that user. (If no
network is involved, the local computer system verifies the addressee.) The mail-storage file is
called the user's system mailbox and usually has the same name as the user who's receiving the
mail. For example, if your login name is george, your system mailbox is the file named george
in the directory /var/spool/mail. When the message has been ``delivered'' to the mailbox,
you say that the mail has been received.
In a common kind of email called Post Office Protocol (POP) mail, email is stored on a
remote system and is then retrieved as you read mail. This chapter reflects a full mail
system, utilizing the sendmail program, which handles the background jobs of
sending and receiving email.
719
Understanding Email
Figure 34.1 shows the relationship between sending and receiving email.
Figure 34.1 System
SENDER Mailbox
Sending and
receiving Message 1 Mail
email.
Message 2 Program
....
Program
RECEIVER
Mail System
Mail System
Personal
Mailbox File Trash!
Send
mbox
Receive
Does the Mail Always Get Through?
When you send email, you might see a message onscreen that says Mail Sent! This message means that the
mail has been sentÐnot that it has been received or delivered. Usually, your email system notifies you if your
message can't be delivered.
Email messages might not go through for several reasons. If mail is going out to a network, the network address
may be correct, but the name of the user on that network may not be correct. Or perhaps the complete address
is correct, but because of problems with permissions or quotas, the message cannot be placed in the user's
system mailbox. In both cases, the mail is sent but is undeliverable. Another scenario is that the email is
delivered but the user's mailbox is corrupted or destroyed. A final possibility is that the recipient ignores email
or doesn't log in for several days, weeks, or more.
Your computer system notifies you when you have mail. When you read your email, you can treat it on a
message-by-message basis. Some of the things you can do with your mail are the following:
. Delete individual messages after you read themÐor without bothering to read them (using email doesn't
mean that you won't get junk mail)
. Keep some messages in the system mailbox
. Keep some messages in a personal mailbox
. Keep other messages in individual files or folders
. Reply directly to the sender of a message Part
. Do a ``group reply'' to a group of users who all received the same message
VI
. Forward mail to others
. Print your mail Ch
34
You are responsible for managing your mail so that it doesn't take up any more disk space
than necessary. You most certainly don't have to save every piece of email you get. You'll also
see that reading your incoming mail is easier if you regularly delete or remove messages from
your system mailbox.
Several different email programs are available for Linux, including email programs that are
integrated with Web browsers such as Netscape. The most common email interface, available
on virtually every UNIX environment, is mail. With the mail program, you can do the
following:
720 Chapter 34 Using Electronic Mail
& Manage and view your email
& Include a subject header on email you send
& Include a cc header for sending copies of email to others
& Forward email to others
& Set up mailing lists
The following sections show examples of mail. Later in the chapter, you will be introduced to
another mail program for LinuxÐthe elm mailer.
Sending Email with mail
You can send email to an individual, a group of individuals, or a mailing list. Just as when you
want to send a paper letter, you must specify the address of the recipient with email.
Sometimes you can compose or write a message while you're sending email; at other times,
you can send a prepared message; you might even send the output of a command or program
with email. When you're using mail or elm, the message you send has to be a text fileÐthat is,
an ASCII file.
The Simple Mail Transport Protocol (SMTP) is used to transfer mail between
computers. It now supports only ASCII files. To send a binary file via email, you have
to convert the file to ASCII by using the uuencode utility. Many more modern GUI-
based email clients handle the file conversion and packaging requirements for you.
Regardless of how the message is prepared, you send mail by using a command of the
following form:
mail address
This command starts the mail system. You can then compose the mail message and send it to
the specified address. In this syntax, address is the email address of the person who is to
receive the message. An address can have several different forms. To send email to someone
who has a login ID on the machine you're using, you can use the login ID of that person. For
example, to send email to someone on your system whose login name is george, you enter the
following command:
mail george
If george is on another system that you can access through some network or collection of
networks, you must include the name by which that system is known on the network. Suppose
thatgeorge is the name of a user on a computer system whose network name is
apples.startup.com. You can send email by entering this command:
mail [email protected]
721
Sending Email with mail
The exact form of the address depends on the type of network being used and any local
conventions or rules. Ask a local expert or your system administrator about the form of
addresses on a network in your company.
To send the same message to several users, you can include each of their addresses on the line
with the mail command, as in this example:
mail fred bill [email protected]
Writing a Message While Sending Email
Many users compose or write messages while they're in the email program rather than
compose a message beforehand. This method is usually the quickestÐbut not the
neatestÐway to send mail. It's not neat because you have limited editing capabilities while
composing your message. Generally, you can deal with only one line at a time. First, you type
the command to send email, specify the address(es), and then press Return. Then you type the
message, indicating that you're done by typing a period on a line by itself. You can also press
Ctrl+d to end the message. For an example of how to send email to a user named lynn, enter
this command to start the mail system and specify lynn's address on your system:
mail lynn
Subject: Congratulations! Lunch Thursday?
Now type the message, pressing Return when you want to end a line. The following is a
sample message that you might want to send to lynn (press Return at the end of each line to
space the paragraphs of the message):
Lynn,
Just wanted to tell you that I thought you did a great
job at the meeting yesterday! It seems as if we're
finally turning this problem around.
Want to get together for lunch Thursday?
Give me a call.
joe
Part
.
VI
You can also end the message with Ctrl+d instead of a period. The computer responds by
Ch
displaying EOT, which means end of transmission.
34
Canceling a Message
You can cancel a message while you're writing it, but you can't cancel it after it's sent. To
cancel a message while you're writing it, you press whatever key is configured on your system
as the interrupt key (usually Ctrl+c or Del). When a message is canceled, it's saved in a file
named dead.letter, usually in your home directory. You can delete this file or edit it later for
another message. When using mail, you must press Ctrl+c twice to cancel (in case you press
Ctrl+c or Del by mistake). After canceling your mail message, you see the command-line
prompt. The following example shows how the cancel function works:
722 Chapter 34 Using Electronic Mail
mail lynn
Subject: Congratulations! Lunch Thursday?
Lynn,
Just wanted to tell that I thought you did a great
job< Ctrl-c >
(Interrupt -- one more to kill letter)
You now must decide whether you want to continue the letter or kill it. If you decide to
continue, you just keep typing the text of the letter as follows:
at the meeting yesterday! It seems as if we're finally
turning this problem around.
At this point, you decide to cancel the letter again, so you press Ctrl+c or Del. The system
responds with (Interrupt -- one more to kill letter). Because you want to kill the message,
press Ctrl+c or Del a second time; mail quits, and you see the shell prompt.
Sending a Prepared Message
You might want to use a text editor such as vi to compose a message to be sent by email. If
you use a text editor, you have the tools to format the text and check your spelling, for
example. What program you use to create the text doesn't matter, as long as you end up with
a text or ASCII file.
Suppose that the file you want to send is named report.txt and the recipient's address is
[email protected]. Essentially, you can send the file in three ways, as outlined in the
following list. In the following examples, the mail command uses the option -s, and the
string that serves as the subject heading is surrounded by quotation marks:
& Use a pipeÐTo send report.txt with the mail command, you enter the following:
cat report.txt | mail -s'' Sales Report'' [email protected]
& Redirect inputÐTo send report.txt with the mail command and the -s option, you
enter the following:
mail -s'' Sales Report'' [email protected] < report.txt
& Use ~r to include a file in a messageÐTo use mail to send the file (by using the
default Subject prompt), you enter these commands:
mail [email protected]
Subject: Sales Report
~r report.txt
~.
EOT
You see the system prompt after you complete any of these three methods; the result is the
same in any case.
In the third example, you use ~r to read, or include, the file report.txt in the
email message. This example shows the use of a tilde command. To use such
commands, you precede a command with the tilde character (~) while you're reading
723
Reading Your Mail
or sending mail. You might find several other tilde commands useful; they're
discussed at appropriate points throughout the chapter.
Sending the Result of a Command or Program by Email
If you run a command or program that produces results to the screen (known as stdout), you
can pipe that output to a mail command. Suppose that you have some information in a file
called contrib.lst, and you want to use the sort command to sort the file and then send the
results to yourself (login name bkorn) and Tom (whom you met earlier in this chapter). To do
all that, you enter this command:
sort contrib.lst | mail -s ``Sorted Contrib Info'' bkorn [email protected]
Reading Your Mail
Most Linux systems notify you when you log in that you have email. It's up to you to read and
act on that email. You can use mail or another email program to read any mail you have. As
you read your mail, the email program marks each message as read. Depending on what
commands you use and how you quit the email program, the messages you've read are kept
either in your system mailbox, /var/spool/mail/$LOGNAME, or in your login directory in the
file named mbox.
Using mail to Read Mail
To read your mail with mail, you enter mail. If your login name is bkorn, for example, you
see a display similar to this (what you type is shown in bold):
mail
mail Type ? for help.
"/var/spool/mail/bkorn'': 5 messages 2 new 1 unread
1 sarah Wed Jan 8 09:17 15/363
2 [email protected] Thu Jan 9 10:18 26/657 Meeting on Friday Part
U 3 fred_Fri Jan 10 08:09 32/900 New Orders
> N 4 jones Fri Jan 10 13:22 35/1347 Draft Report VI
N 5 [email protected] Sat Jan 11 13:21 76/3103 Excerpt from book
? Ch
Note the following points about this display:
34
& The first line identifies the program and says to type a question mark for help.
& The second line indicates that mail is reading your system mailbox, /var/spool/mail/
bkorn, and that you have five messages. Two have arrived since you last checked your
mail; one appeared previously, but you haven't yet read it; and two messages have
already been read.
724 Chapter 34 Using Electronic Mail
& The next five lines give information about your mail. You can ignore the first few
characters for now. Each line holds a message number, the address of the sender, the
date the message was sent, the number of lines and characters in the message, and the
subject (if one was given). Consider the following line:
2 [email protected] Thu Jan 9 10:18 26/657 Meeting on Friday
This line indicates that message number 2 is from [email protected]Ðan address
that indicates the message came to your machine from another network (mail from a
local user is marked with just the user's login ID). The message was sent on Thursday,
January 9, at 10:18; it consists of 26 lines and 657 characters. The subject is Meeting on
Friday.
& A message line starting with N indicates new mailÐmail received since you last
checked your email. A message line starting with U indicates unread mail. A message
line without N or U indicates mail you've read and saved in your system mailbox.
& >
The greater-than character ( ) on a message line marks the current messageÐthe
message you'll act on next.
& ?
The question mark ( ) on the last line is the command prompt from mail.
Reading the Current Message
The current message is marked by the greater-than character (>). To read that message, you
can press Return. When you open it, you see something like the following:
Message 4:
From jones Fri, Jan 10 13:22 EST 1997
Received: by your.system.com
Date: Fri, 10 Jan 1997 13:22:01 -0500
From: Carol Jones [[jones]]
Return-Path: [[jones]]
To: aborat, lynn, oackerm, bkorn
Subject: Draft Report
Here is a draft of the report I intend to submit next week.
Please take a look at it and let me know your comments.
Thanks.
-------------------------------------Report Starts Here-----------------------------------------
Opportunities for Expansion
Prepared by Carol Jones
Over the past 6 months, we've seen an indication of an increase in the
demand for our services. Current market trends indicate that the demand
will continue for at least 18 months and possibly longer. The manager of
our service staff states ``We're up to our necks in new customers and
:
The message is displayed one screen at a time. Any time you see a colon, you can press Return
to see the next screen or q to quit viewing the message. In this case, press Return to see the
next screen of the message.
On the last screen, you see EOF: (for end of file). You can press q or Return to get back to the ?
prompt. Notice that the greater-than character still points to the message you've just read.
The message that was the current message is still the current message.
725
Reading Your Mail
Some lines are displayed before the message itself begins. These lines make up the header
information, which can be useful. Typically, header information includes the following:
& The message number
& The name of the person who sent the message
& The time the message was sent
& The name of the system that received the message
& The date the message was received
& The ``real name'' of the sender, as well as his or her login ID
& The return path
& The message recipient(s)
& The subject
All this information is passed on with each email message. The sender is always identified,
making forgeries difficult. The real name that appears in the From line is taken from a field
from the sender's entry in the password file. The mail system uses the Return-Path or
Reply-To information if you generate a reply (as discussed later in this chapter). The To line
contains the address or list of addresses of the recipients of this message. (This sample
message is a group message.) Here, the sender filled in the Subject line.
Reading the Next Message
You can read the next message (the message following the current message in your mailbox)
in two ways. You can press Return or n to display the next message. It becomes the current
message after you read it. You read the next message in the same way you read the current
message. After you read the last message in the list, you see the message 'At EOF'.
Reading Any Message
All the messages in your mailbox are numbered. You can read messages in any order by
entering the message number when you see the ? prompt. For example, to read message Part
number 2, type 2 and press Return. Message number 2 then becomes the current message. VI
Ch
Reading Email from Other Files
34
When you start mail, you read messages kept in your system mailbox, which has the path
/var/spool/mail/$LOGNAME. Recall from Chapter 16, ``Understanding Linux Shells,'' that
LOGNAME is the shell variable that holds your login name. If you log in as bkorn, your mail is
held in /var/spool/mail/bkorn. You can read mail from other files that hold complete email
messagesÐthat is, messages with the headers and text of the messages. Naturally, you must
have read permission for those files.
726 Chapter 34 Using Electronic Mail
To read messages from a file, you type the command to start the email program followed by
-f filename and then press Return. For example, to read the email in the file mbox, you enter
this command:
mail -f mbox|
You can read the mail in that file in the same way you read email from your system mailbox.
The mbox file is located in your home directory and automatically contains messages
you've already read but haven't deleted. These messages are saved to mbox when
you exit mail.
Sending Mail While Reading
You can send email while you're using the mail program to read your messages. To do so,
you enter m address at the ? prompt. For an example, follow these steps:
1. Start the mail program (type mail and press Return).
2. Read some messages or perform other tasks, but at the ? prompt, enter the following
to send email to a user whose login name is ernie:
m ernie
3. At the prompt for a subject, type a subject heading, as follows:
Subject: Game Time
4. Type the message, and end it with a period on the last line, as in the following
example:
Don't forget we're playing V-ball at 6:30
.
The computer responds with the following lines:
EOT
?
5. Continue using mail.
Printing Mail Messages
When you're using mail, you can print the current message to a printer connected to your
system. First, you need to make the message you want to print the current message. Then you
can enter | lpr at the ? prompt. You are, in effect, piping the current message to the lpr
program.
To print a collection of messages, you can save them in a file and then print the file. See the
section ``Saving Email to Files with mail'' later in this chapter for information on effective
ways to save messages.
Getting Help with mail 727
Getting Help with mail
When you type the command to start your email program, you see a ? prompt. The mail
program tells you to type ? for help. To get a list of commands and some information about
each command, you type ? and press Return.
After you type ? and press Return, you see a display similar to the following:
Mail Commands
t < message list > type messages
n goto and type next message
e < message list > edit messages
f < message list > give head lines of messages
d < message list > delete messages
s < message list > file append messages to file
u < message list > undelete messages
R < message list > reply to message senders
r < message list > reply to message senders and all recipients
pre < message list > make messages go back to /usr/spool/mail
p < message list > print message
m < user list > mail to specific users
q quit, saving unresolved messages in mbox
x quit, do not remove system mailbox
h print out active message headers
! shell escape
cd [directory] chdir to directory or home if none given
A < message list > consists of integers, ranges of same, or user names
separated by spaces. If omitted, Mail uses the last message typed.
A < user list > consists of user names or aliases separated by spaces.
Aliases are defined in .mailrc in your home directory.
&
This listing shows you the commands you can use from the ? prompt. Although some of these
commands are explained later in this chapter, the following are some points to note right
now:
& In each case, you can use the first letter of the command or type the entire command.
& Items in [] and <> are optional; you don't type the brackets as part of the command. Part
& You can make the term message list refer to all messages by using *. To save all VI
messages in a file named allmail, for example, type s * allmail and press Enter.
&
Ch
You can make the term message list refer to a single message number. To save
34
message number 2 to a file named meeting, type s 2 meeting and press Return.
& You can make the term message list refer to a range of message numbers by
separating the two message numbers with a hyphen. For example, 2-4 refers to
messages numbered 2, 3, and 4. To save messages 2, 3, and 4 in a file named memos,
type s 2-4 memos and press Return.
& The term print in the line print message doesn't mean to print messages on a printer.
It means to display the messages.
& The edit command is useful for modifying messages before forwarding them to
someone else or saving them in a file.
728 Chapter 34 Using Electronic Mail
Saving Email to Files with mail
You'll probably want to save some of the email you receive. Keeping all your mail in your
system mailbox is not practical for these reasons:
& You'll have too many messages to wade through when you want to read your mail.
& System administrators often limit the size of your system mailbox. This size limit
depends on how your system administrator set up your mailbox. If you reach that
limit, you might be prevented from receiving any new mail.
& Your mail won't be organized, and finding important messages or all messages relating
to a specific project or topic can be difficult.
Earlier in this chapter, you learned that the messages you've read are saved (unless you say
otherwise) in the file mbox. You also know that you can read these messages by typing mail -f
mbox and pressing Return. You can also read messages from other files by using the mail
command's -f option.
You can save the current message in a file in two primary ways (with and without a header)
when you use mail. With both methods, you can specify a file to hold the message, and the
message is added to that file. If you don't specify a file, the message is added to the file mbox
(your personal mailbox) in your home directory. If you use q to quit the mail program, the
messages are removed from your system mailbox.
When you see the ? prompt, you can use any of the following methods to save a message:
& Type s to add the current message to mbox in your home directory.
& Type s filename to add the text of the current message to the named file with the
headers intact. (This method is useful if you want to use your email program to read
the messages later.)
& Type w filename to add the text of the current message to the named file without the
header information. (This method is useful when you want to use only the text of the
messages in a file that may be processed by some other program.)
Tip #174 from
Steve To keep messages in your system mailbox rather than the mbox file after you read
them, use the preserve command, pre. You can use this command with a message
list.
You know that messages you've already read are automatically saved to mbox unless you use
the preserve command.
It's a good idea to get in the habit of specifying a filename when you use the save command, s.
If you don't specify a filename, the current message is added to the file mbox. If you include a
message list but don't specify a file, mail uses the message list as the name of the file to which
Replying to Email with mail 729
it saves the current message. If you use q to quit the email program, the saved messages are
removed from your system mailbox.
Deleting and Undeleting Messages with mail
To delete a message from a file of messages you're reading, you use the d command. If you
quit the mail program by using q, any messages you deleted with the d command are removed
from the file.
You use the d or delete command to mark messages for deletion when you use mail to read
your email. If you then quit the program by using q, the marked messages are removed from
your mailbox. Unless you've saved them, they're gone for good. For some messages, deleting
without saving them is a very good idea.
To delete the current message, type d and press Return. You can also specify a message list.
If you mark a message or a group of messages to be deleted, you can change your mind and
undelete the message or messages by using the u command. You must use the u command
q to quit; when you enter q, the messages are gone for good. You use the u or
before you enter
undelete command in the same way you use d or delete.
Tip #175 from
Steve To undelete all the messages you marked for deletion, enter u * at the ? prompt.
Replying to Email with mail
To reply to email, you use the address specified in the Reply-To header field. If that field isn't
present, you use the information in the Return-Path header field. Following are partial
headers of two messages; one has both header fields, and the other has only the Return-Path Part
header field. The pertinent fields are in bold in each example.
VI
Message 1: Ch
From [email protected] Mon Nov 8 18:31 EST 1993 34
Received: from MALTE.ABC.COM by s850.mwc.edu with SMTP
Return-Path: < [email protected] >
Date: Mon, 9 Nov 98 18:17:15 -0500
Comment: From the DuJour List
Originator: [email protected]
Errors-To: [email protected]
Reply-To: < [email protected] >
Sender: [email protected]
Message 2:
From jones Fri, Jan 8 13:22 EST 1999
730 Chapter 34 Using Electronic Mail
Received: by your.system.com
Date: Fri, 8 Jan 1999 13:22:01 -0500
From: Carol Jones < jones >
Return-Path: < jones >
To: aborat, lynn, oackerm, bkorn
Subject: Draft Report
To reply to the first message, you use the Reply-To address [email protected]. Note that
the Reply-To and Return-Path fields are different. For the second example, you use jones to
respond to the sender of the message.
You should always use theReply-To address if it's included in the header because it
represents the specific address of the sender. When the Reply-To address isn't
available, the Return-Path address usually provides an adequate address back to
the sender.
You can let the mail program determine the address to use to reply to an electronic mail
message. To do so, you can use either of the following commands:
R Addresses a reply to the sender of the message
r Addresses a reply to the sender and all recipients of an
email message
With either command, you can specify a message list, as explained earlier in this chapter.
Otherwise, the R or r command applies to the current message.
The following partial header shows how to use these two commands. This header is
excerpted from a message from Carol Jones, in which she asks a group to comment on a draft
of a report she has prepared:
From jones Fri, Jan 8 13:22 EST 1999
Received: by your.system.com
Date: Fri, 8 Jan 1999 13:22:01 -0500
From: Carol Jones < jones >
Return-Path: < jones >
To: aborat, lynn, oackerm, bkorn
Subject: Draft Report
To respond to jones only, you enter R at the ? prompt. You see the following response:
To: jones
Subject: Re: Draft Report
The To line tells you that the reply is going to one person. The Subject header indicates that
the message is a reply to the one originally sent.
To make comments for everyone on the distribution list to see, you enter r at the ? prompt.
You then see the following response lines:
To: jones, aborat, lynn, oackerm, bkorn
Subject: Re: Draft Report
731
Routing Mail to Others
The To line tells you that the reply is going to everyone on the original distribution list, as
well as the author. The Subject header indicates that the message is a reply to the one
originally sent.
From here on, you enter your message in the manner described earlier in the section
``Sending Email with mail."
Be careful about using r to reply to a message. Whatever you send is sent to
everyone who got a copy of the original message. Because Linux is case sensitive, and
most people aren't used to typing capital letters as commands, using r is a very
common mistake and can sometimes be embarrassing.
Think about what you write and who will read your message before you send a reply.
Being sarcastic or scathing doesn't work very well with email; you usually end up
sounding like a bully. Using email isn't the same as talking with someone: You don't
get a chance to see or hear the person's reactions, and he or she doesn't get a chance
to see or hear you either. When you use email, it's a lot easier and more effective to
be polite and direct.
You can see how easy it is to forward mail; as soon as you send something to one
person, you can never tell where the message will end up or how many people will
see it. Think, and be considerate.
? See ``Lack of Visual Reference,'' p. 748
Part
Routing Mail to Others
VI
Email is distributed by addresses. Tasks such as forwarding a message, sending copies (cc:) of
Ch
a message, creating aliases or simpler forms of addresses, and creating mailing lists all involve
manipulating addresses. You don't have to do the manipulation directly; the mail program 34
has these capabilities built in.
Forwarding Messages
To forward a message (actually, you're including the message with a message you compose),
you must first start mail in the same way that you start it to read your messages. Then you use
the m, r, or R command to send a message. As you compose your message, you use a tilde
command, ~f, to forward one or several messages. The general form of the ~f command is
~f msglist. The following is a step-by-step example of how to forward a message:
732 Chapter 34 Using Electronic Mail
1. Start mail (type mail and press Return). The system responds with something similar
to the following:
mail Type ? for help.
"/var/spool/mail/bkorn'': 5 messages 2 new 1 unread
1 sarah Wed Jan 6 09:17 15/363
2 [email protected] Thu Jan 7 10:18 26/657 Meeting on Friday
U 3 fred Fri Jan 8 08:09 32/900 New Orders
> N 4 jones Fri Jan 8 13:22 35/1347 Draft Report
N 5 [email protected] Sat Jan 9 13:21 76/3103 Excerpt from book
?
2. Read message 5 by typing 5 and pressing Return. (The text of that message isn't
shown here.) Suppose that you want to forward it to your friends whose addresses are
sarah, [email protected], and [email protected].
3. Use the m command to send mail to the addresses listed in step 2, type a subject, and
type a beginning for your message, as shown here:
? m sarah [email protected] [email protected]
Subject: Forwarding an excerpt from new Que Linux book
Hi!
I'm forwarding an excerpt I came across from a new book by Que.
It's Special Edition Using Linux, Fifth Edition. I'll be
getting my own copy tomorrow.
Do you want me to pick up a copy for you, too?
4. Use the ~f command to forward message number 5 (type ~f 5 and press Return). mail
responds with the following message:
Interpolating: 5
(continue)
5. The cursor is now under the word continue. You can continue adding text to your mail
message, or you can end it by typing ~. and pressing Return. If you end it, the ?
prompt appears.
Sending a Copy with mail
You can send a copy of an email message to one or more addresses by putting those addresses
on what's known as the cc: list. The cc: list works as you expect it to: The mail is sent to the
primary address or addresses (those in the To header) and also to the address or addresses in
the Cc header. To include addresses in the cc: list, you use the tilde command ~c address
while you're sending the message.
The following example shows how to send a brief memo to a primary address ( wjones) and a
copy of it to yourself and another address (your address is bkorn, and the other user's address
is ecarlst). You send one to yourself so that you have a copy of the memo. Follow these steps
to add a cc: list to the list of recipients:
1. Start mail to send email to the primary address, wjones, and give a subject header.
Enter the following commands to achieve this:
$ mail fred
Subject: Memo - Sales Agreement with Framistan
733
Routing Mail to Others
2. Enter the text of the memo you want to send. For example, type the following:
TO: Fred Jones
Date: Oct 31, 1999
From: Henry Charleston
RE: Sales Agreement With Framistan Motors
On October 27,1999, I held a meeting with the CEO of Framistan Motors.
We concluded and initialed a sales agreement by which Framistan would
purchase 10,000 units of our thermo-embryonic carthurators. The agreement
has been forwarded to the appropriate parties in our organization and
we intend to formally complete the agreement within two weeks.
3. Give the ~c address command to add addresses to the cc: list. For example, type the
following to send copies to yourself ( bkorn) and to ecarlst:
~c ecarlst bkorn
4. To send the message, enter a tilde and period ( ~.) and press Return. The EOT message
appears, followed by the shell prompt.
Tip #176 from
Steve To review and possibly modify the headers on an outgoing message, you can enter
~h while you're composing the message. You're shown the headers one at a time,
and you can modify them.
When a message is sent this way, all the recipients can see the headers To and Cc. Anyone who
replies to the message with the r command will send the reply to every address in the To and
Cc lists, as well as to the author.
You can customize mail so that it always prompts you for a Cc header in the same way that it
prompts you for the Subject header (this topic is discussed later in the section ``Customizing
Your mail Environment''). Of course, you can keep from entering anything in the Cc list by
pressing Return.
Part
Using Aliases and Mailing Lists
VI
The mail program, like most email programs, allows you to create an alias for an address and a
Ch
group alias for a list of addresses. You can treat the group alias as a mailing list. Using an alias
for an individual address is easier than using the regular address because the alias is typically 34
shorter and easier to remember.
To set an individual or group alias for one mail session, you use the alias command at the ?
prompt while you're reading your email. To make the aliases more useful, put the aliases in a
file named .mailrc in your home directory (as described in the following section).
The following is an example of setting and using aliases with the mail program:
1. Start mail by entering mail at the prompt. After the headers are presented, you see the
? prompt:
734 Chapter 34 Using Electronic Mail
mail Type ? for help.
"/var/spool/mail/bkorn'': 5 messages 2 new 1 unread
1 sarah Wed Jan 5 09:17 15/363
2 [email protected] Thu Jan 6 10:18 26/657 Meeting on Friday
U 3 wjones Fri Jan 7 08:09 32/900 Framistan Order
> N 4 chendric Fri Jan 7 13:22 35/1347 Draft Report
N 5 [email protected] Sat Jan 8 13:21 76/3103 Excerpt from GREAT new Linux
?
2. To set up an individual alias, use the alias command followed by the alias
for the address. The following example creates the alias ros for the address
[email protected]:
alias ros [email protected]
3. Use theros alias in an address; mail expands it to its complete form. For example, you
can enter the command m ros to start a message you want to mail to
[email protected].
To set up a group alias, you use the alias command followed by the alias for the addresses.
The following creates an alias called friends and then forwards some mail to the group:
alias friends chendric karlack abc.com!homebase!fran [email protected]
m friends
Subject: Excerpts from new Linux book - get a copy!
~f 5
Interpolating: 5
~.
EOT
?
Customizing Your mail Environment
You can customize your mail environment by putting commands or set-environment
variables in the .mailrc file in your home directory. The mail program checks that file
whenever you use the program. You can set quite a few environment variables and commands
in .mailrc, and different mail programs use different commands. Check your man page for
your mail program for a list of all the .mailrc options. Some of the commands mail
recognizes are given earlier in the section ``Getting Help with mail''; this section describes a
subset of the commands and variables that can be used in the .mailrc file. Table 34.1 lists
these commands; Table 34.2 lists the environment variables.
Table 34.1 mail Commands
Command Definition
# Denotes a comment. No action is taken.
alias Sets an individual or group alias. Used as alias alias-nameaddress-list.
set Sets an environment variable. Used as set variable-name or set
variable-name=string.
Customizing Your mail Environment
735
Tip #177 from
Steve You can issue any of the commands in Table 34.1 from the ? prompt any time you
use mail; these commands are active only for that session.
Table 34.2 mail Environment Variables
Variable Definition
askcc Prompts for the cc: list after the message is entered. Default is noaskcc.
asksub Prompts for the Subject list before the message is entered. Enabled by default.
noheader Doesn't print header information on available messages when you start mail.
Disabled by default.
ignore Ignores interrupt characters when you enter messages. This variable is useful if
you have a ``noisy'' connection over some telephone or other communication
lines. Default is noignore.
metoo Allows you to receive messages sent to a group alias that contains your address.
When you have your name in a group alias, a message normally isn't sent to
you. Default is nometoo.
Tip #178 from
Steve You can set a systemwide environment by putting the commands or set variables in
the /etc/mail.rc file.
The following example sets up the .mailrc file so that you use the commands and
#
environment variables listed in Tables 34.1 and 34.2. The pound sign ( ) is used to document
the work. You can create this file by using vi or any other editor that can produce a text or
ASCII file. Part
# .mailrc file for D. Wayne Love VI
# make sure interrupts are NOT ignored
set noignore Ch
# set variables so that prompts for Subject and Cc always appear 34
set asksub
set askcc
# individual aliases
alias billy wcuth
alias ben [email protected]
alias me dwlove
# group aliases, mailing list
alias mercs [email protected] quinn taura
alias research jones brown smith
alias googol djames bkorn [email protected] bkorn
736 Chapter 34 Using Electronic Mail
Place these statements in the .mailrc file. Now whenever you start mail, these command
statements are processed.
Quitting the mail Program
As you read email in a mailbox, you can read, skip, or delete messages. (You'll learn about
deleting messages later in this chapter.) These actions don't take place in the mailbox itself,
but in a temporary copy of the mailbox. You can quit the email program so that your mailbox
is changed by your actions (the modified temporary copy replaces the original mailbox), or
you can quit so that your mailbox is unchanged regardless of what you did during your email
session.
Quitting and Saving Changes
To quit the mail program and save the changes that occur, press q and then Return at the ?
prompt. You see the shell prompt again. When you quit mail this way, messages you read but
didn't delete are saved in a file named mbox in your home directory.
Suppose that you use mail to read your mail. Your login name is bkorn, and your home
directory is /home/bkorn. When you enter mail to start the mail program, you see the
following screen of information:
mail Type ? for help.
"/var/spool/mail/bkorn'': 5 messages 2 new 1 unread
1 sarah Wed Jan 8 09:17 15/363
2 [email protected] Thu Jan 9 10:18 26/657 Meeting on Friday
U 3 fred Fri Jan 10 08:09 32/900 New Order
> N 4 jones Fri Jan 10 13:22 35/1347 Draft Report
N 5 [email protected] Sat Jan 11 13:21 76/3103 Excerpt from book
?
Now suppose that you read the current message by pressing Return, and then you read
message 1 by typing 1 and pressing Return at the ? prompt. If you press q and then Return to
quit, you see the following information:
Saved 2 messages in /home/bkorn/mbox
Held 3 messages in /var/spool/mail/bkorn
The two messages you read are saved in the file mbox in your home directory; the other three
messages are saved in your system mailbox, /var/spool/mail/bkorn.
If you save read messages like this often, mbox can become quite large. You might want to
print that file occasionally and delete it. You can also read the mail from that file as though it
were your system mailbox, as described later in this chapter.
You can read mail and indicate that the current message is to be kept in your system
mailbox, /var/spool/mail/bkorn, and not in the file mbox. To do so after you
read a message, you enterpre (for preserve) at the ? prompt.
Using the elm Mailer
737
Quitting and Not Saving Changes
The other way to quit the mail program is to press x and then Return at the ? prompt. When
you do that, you exit the program with no changes to your system mailbox or any other
fileÐas if you didn't read your mail at all. You then see the shell prompt. You might want to
exit the mail program in this way when you want to leave the program but save the mail in
your system mailbox.
Using the elm Mailer
As stated earlier in this chapter, several different mail programs are available for Linux. Each
has its own advantages and disadvantages.
One mail reader that comes with the Red Hat distribution of Linux is the elm mailer. This
mail program is a screen-oriented mailer rather than a line-oriented one. It provides a set of
interactive menu prompts and is easy to use. Virtually everything that you can do with mail
can be done under elm, and usually much more easily!
Because elm is easy to use, the following sections just touch on the highlights of using it. You
can find more in-depth information by using elm's online help or by reading its man page.
Starting elm
To start a mail session with elm, type elm at the command prompt. If this is the first time
you've used elm, it prompts you for permission to set up a configuration directory in your
account and create an mbox mail file if one doesn't exist. You see the following as you start elm
for the first time:
$ elm
Notice:
This version of ELM requires the use of a .elm directory in your home
directory to store your elmrc and alias files. Shall I create the
directory .elm for you and set it up (y/n/q)? y
Great! I'll do it now.
Part
Notice:
VI
ELM requires the use of a folders directory to store your mail folders in.
Shall I create the directory /home/gunter/Mail for you (y/n/q)? y Ch
Great! I'll do it now.
34
After elm creates its directory and mbox file, it runs the main mail program. It is a full-
screen-oriented mailer. Your screen clears, and you see a display similar to the following:
Mailbox is '/var/spool/mail/gunter' with 2 messages [ELM 2.4 PL25]
N 1 Nov 11 Jack Tackett Linux book
N 2 Nov 11 Jack Tackett more ideas
You can use any of the following commands
by pressing the first character;
d)elete or u)ndelete mail, m)ail a message,
r)eply or f)orward mail, q)uit
To read a message, press <return>. j = move down, k = move up, ? = help
Command:
738 Chapter 34 Using Electronic Mail
At the top of the screen, elm tells you where your system mailbox is located, how many
messages are in it, and what version of elm you're running. elm then lists one line for each
message in your mailbox. It places the letter N before each new message, just like the mail
program. The summary line for each message tells you the message status, the message date,
the sender, and the subject. (As always, your display may vary slightly depending on your
version of elm.) The current message is highlighted in the list (in the preceding listing, the
current message appears in boldface).
Using elm Commands
At the bottom of the screen is a command summary that tells you what commands you have
available for the current screen. As you can see in the preceding example, you can delete or
undelete mail, mail a message, reply to a message, forward mail, or quit. Pressing the j key
moves the message selection to the previous message; the k key moves it to the next message.
Help is available if you press the ? key. The Command: prompt at the bottom of the screen tells
you to press a command key for elm to do something.
As you can see, elm is very easy to use because of the large number of prompts and onscreen
help that's available. Table 34.3 lists all the commands that can be executed from within elm.
Table 34.3 Command Summary for elm
Command/Keystroke Description
Return, Spacebar Displays the current message
| Pipes the current message or tagged messages to a system
command
! Escapes the shell
$ Resynchronizes the folder
? Displays online help
+, [rpa] Displays the next index page
-, [lpa] Displays the previous index page
= Sets the current message to the first message
* Sets the current message to the last message
number+Return Sets the current message to number
/ Searches subject lines for a pattern
// Searches entire message texts for a pattern
]] Saves the current message or tagged messages to a folder
[[ Scans the current message for calendar entries
a Changes to ``alias'' mode
b Bounces (remails) the current message
C Copies the current message or tagged messages to a folder
c Changes to another folder
739
Project: Using the Mutt Email Client
Table 34.3 Command Summary for elm
Command/Keystroke Description
d Deletes the current message
Ctrl+d Deletes messages with a specified pattern
e Edits the current folder
f Forwards the current message
g Sends a group (all recipients) reply to the current message
h Displays the header with a message
J Increments the current message by one
j, [dpa] Advances to the next undeleted message
K Decrements the current message by one
k, [upa] Advances to the previous undeleted message
l Limits messages by specified criteria
Ctrl+l Redraws the screen
m Mails a message
n Moves to the next message, displaying the current message and
then incrementing
o Changes elm options
p Prints the current message or tagged messages
q Quits, maybe prompting for deleting, storing, and keeping
messages
Q Acts as a quick quitÐno prompting
r Replies to the current message
s Saves the current message or tagged messages to a folder
t Tags the current message for further operations
T Tags the current message and goes to the next message
Part
Ctrl+t Tags messages with a specified pattern
VI
u Undeletes the current message
Ch
Ctrl+u Undeletes messages with a specified pattern
34
x, Ctrl+q Exits, leaving the folder untouched; asks whether you want to exit
if you have changed the folder
X Exits, leaving the folder untouched, unconditionally
Project: Using the Mutt Email Client
Mutt is a freeware mail client that is rapidly gaining in popularity. Although comparative
novices can use it perfectly well (by default, Mutt looks and works much like the elm client),
740 Chapter 34 Using Electronic Mail
Mutt is especially popular among more knowledgeable users because of its extreme configurability.
Some of the primary features of Mutt include the following:
& Color support
& Support for message threading
& MIME supportÐincluding RFC2047 support for encoded headers and PGP/MIME
(RFC2015)
& POP3 support
& Support for multiple mailbox formats including mbox, MMDF, MH, and maildir
& Key bindings (by default, identical to elm)
& Capability to search using regular expression
& Support for Delivery Status Notification (DSN)
& Capability to include attachments from the command line when composing
& Capability to reply to or forward multiple messages at once
& .mailrc-style configuration files
& Installation process using GNU autoconf
Where to Get Mutt
Mutt is distributed under the GNU public license terms on the Red Hat 6.0 installation
CD-ROM, in the Mail directory of the Applications section. An international version of Mutt
that contains support for PGP is available from several FTP servers, including ftp://
ftp.gbnet.net/pub/mutt-international/ among others.
For More Information on Mutt
The Mutt home page is at http://www.mutt.org/index.html. It contains links to the online
Mutt manual by Michael Elkins (at http://www.mutt.org/doc/manual/) and the Mutt FAQ
by Felix von Leitner (at http://www.math.fu-berlin.de/~leitner/mutt/faq.html). The
Mutt home page also contains information on several mailing lists devoted to the mail client.
CHAPTER
35
Surviving Usenet News
In this chapter by Steve Burnett
What Is Usenet News? 742
A Usenet Glossary 742
A Brief History 744
How Usenet Is Structured 745
No Central Authorit 747
Usenet Culture 747
Reading and Posting News 749
Netiquette on Usenet 752
Project: Using the rn Newsreader 753
742 Chapter 35 Surviving Usenet News
What Is Usenet News?
With the explosive growth of the Internet, Usenet news has attracted lots of attention. Many
online services now offer access to Usenet. But what's Usenet? UsenetÐshort for user
networkÐis a proto-network of machines that exchange information grouped into subject
hierarchies. The term proto-network is used because Usenet isn't a physical network in the
normal sense. It's made up of all the computers that exchange Usenet news.
In the simplest definition, Usenet news, netnews, or simply news is a forum for online
discussion. Many computers around the world exchange chunks of information, called articles,
on almost every subject imaginable. These computers aren't physically connected to the same
network; they're logically connected in their capability to exchange data. Thus, they form the
logical network referred to as Usenet. In this chapter, the terms Usenet, news, and netnews are
used interchangeably.
The software that drives Usenet is divided into two parts: newsreaders (the software
that users use to read and post news articles) and the software that processes articles
and transfers them between systems.
Many people initially think of a PC bulletin board system (BBS) when trying to understand
Usenet. Although Usenet news does bear some similarity to a BBS at first glance, some very
substantial and important differences are apparent upon closer inspection:
& The various news articles on different subjects don't reside on one computer, as with a
BBS. They're sent from computer to computer via a store-and-forward mechanism.
Each site that receives news exchanges articles with one or more neighbors in
transactions that are known as news feeds. As a result, news articles take time to
propagate from place to place.
& No one is in charge. Yes, you read that right. Usenet has no overall manager, such as a
BBS sysop (system operator). Each site has a good deal of autonomy. Usenet news has
been described, very accurately, as ``organized anarchy.''
In general, Usenet news is divided into two logical parts: the programs and protocols that
make up the mechanism for posting articles and transferring news articles between
computers, and the user programs for reading and posting news articles. This chapter deals
primarily with the user portion.
A Usenet Glossary
Usenet news has its own structure and culture, which are discussed later in the section
``Usenet Culture.'' Usenet also has a terminology all its own. These ``buzzwords'' tend to
743
A Usenet Glossary
confuse new users, especially those who use BBS systems. Table 35.1 provides a brief glossary of
common terms found on Usenet.
Table 35.1 Common Terms Encountered on Usenet
Term Definition
Article A single message posted to a newsgroup.
AFAIK Acronym for As Far As I Know.
Bandwidth An engineering term referring to the amount of data a given transmission
medium can hold. Commonly used as in the phrase waste of bandwidth for
articles that contain little useful information.
BTW Acronym for By the Way.
FAQ An acronym for the Frequently Asked Questions list. Many newsgroups post a
FAQ on a regular basis. It's usually considered impolite to post a question
with the answer already in the FAQ for a group.
Flame An article that's full of rude, angry, insulting statements directed at
another person.
FYI Acronym for For Your Information.
Hierarchy Usenet's system of grouping newsgroups into a tree structure based on
subject.
IMHO Acronym for In My Humble Opinion.
Newsgroup A logical group of articles that are about one general subject.
Newsreader A user program, such as rn, that's used to read and post articles to Usenet.
net.personality Someone who's famous within the Usenet or Internet community.
net.police A mythical organization responsible for enforcing the rules on Usenet.
Typically used as sarcasm.
Netiquette The etiquette of Usenet.
Newbie Someone who's new to using Usenet news.
Quoting Including parts of a message to which you're responding. Most news-
readers allow you to quote articles. You should quote only relevant
portions of an article to save bandwidth. Sometimes also referred to as
quotebacks.
ROFL Acronym for Rolling On the Floor, Laughing.
RTFM An acronym for Read The Forgotten Manual. Typically used as in ``Here's a
short answer to your question. RTFM for more info.''
Part
Post
Signal-to-noise
To submit an article to a newsgroup.
Engineering term referring to the ratio of the amount of data to the
VI
amount of background noise. On Usenet, it refers to how much useful Ch
information is in a newsgroup versus the amount of off-topic background
chatter. A high signal-to-noise ratio refers to a newsgroup that has lots of
35
useful information and very little off-topic chatter. Signal-to-noise can also
be used as a descriptive term for a specific person.
744 Chapter 35 Surviving Usenet News
Table 35.1 Common Terms Encountered on Usenet
Term Definition
Signature A short file that's included at the end of all your posts. Typically includes
your name, email address, and possibly a witty quote of some sort.
Sigfile See signature.
Smileys Common symbols for denoting emotion in a post or email message. For
example, :-) and :-( are a happy face and a sad face. (Tilt your head toward
your left shoulder and look at them sideways.)
TIA Acronym for Thanks In Advance.
A Brief History
In late 1979, two graduate students at Duke University began considering how to connect
UNIX computers so that they could exchange text messages. Another graduate student at the
University of North Carolina became involved in this effort and wrote the first news transfer
system, which consisted of a collection of shell scripts. This software was installed on the first
two Usenet sites, unc and duke. In early 1980, another computer at Duke, phs, was added.
The news software was eventually rewritten in C for public distribution. It became known as
the A News software.
As the news software grew in popularity, it quickly became obvious that the current news
transport software couldn't handle the increasing flow of news. Programmers at the
University of California at Berkeley began to rewrite the current A News software to increase
its capabilities. This new version, known as B News, was released in 1982.
Throughout this time, news articles were being transferred by using the UNIX-to-UNIX Copy
Program (UUCP) protocol. As more sites joined the news network, the network load grew to
unmanageable levels. Soon realizing that UUCP no longer worked as the main transport
protocol for news, people began looking to the Internet and the TCP/IP protocols for help.
In 1986, a software package that implemented the Network News Transport Protocol (NNTP)
was released. This protocol is defined in RFC 977. NNTP allowed news articles to be
exchanged by using TCP/IP instead of the slower UUCP protocol. It also allowed users to
read and post news from remote machines so that the main news processing software didn't
have to be installed on every computer.
When NNTP became available on the Net, the already rapid growth of the Usenet system
exploded. The current news-processing software, B News, quickly became too slow to handle
the increasing news flow. In 1987, Henry Spencer and Geoff Collyer of the University of
Toronto developed a new news-processing software, C News. Then Rich Salz developed a
news transport system known as INN, one of the most widely used news servers on the
Internet.
745
How Usenet Is Structured
The Usenet news system continues to grow at a rapid pace. Other commercial information
service providers are now carrying Usenet news as part of their online services. Several BBS
networks, such as FidoNet, also carry Usenet news.
An excellent reference for the history of Usenet news is the news article ``Usenet
Software: History and Sources,'' by Dr. Gene Spafford. You can find this article on the
World Wide Web at http://www.faqs.org/faqs/usenet/software/part1/.
How Usenet Is Structured
Usenet is made up of literally many thousands of newsgroups. How many, exactly? Well,
nobody knows for sureÐwell over 20,000. You can check out thousands of groups on
virtually every topic, and that number is growing every day. The topics range from silly or
pointless to focused and precisely helpful.
Group Hierarchies
With so many different newsgroups, trying to find information on the subjects that you are
interested in would be a nightmare if the newsgroups weren't organized in some way. Usenet
newsgroups are organized in a hierarchy based on subject. The names of the newsgroups are
made up of subnames, each separated by a period. These names go from a general category to
a specific category as you read the name from left to right. At the top of the hierarchy are
several standard group categories, plus lots of specialized categories. These standard
categories are well established. Table 35.2 lists the top-level group standard categories in the
Usenet news system.
Table 35.2 Top-Level Group Standard Categories in the Usenet Hierarchy
Class Description
comp Many different computer-related topics
misc Miscellaneous topics that don't easily fit into another category
news Various topics that relate to the Usenet news system itself
rec Recreational and hobby subjects
Part
soc
sci
Social issues
Various scientific topics
VI
talk Subjects designed for ongoing conversations Ch
35
As with everything else on the Internet, there are exceptions to the rules in Table 35.2. Many
other top-level hierarchies exist; most are devoted to different regions of the world. For
746 Chapter 35 Surviving Usenet News
example, the ba and triangle group hierarchies are concerned with topics of interest to the
San Francisco Bay area and the North Carolina Research Triangle Park area, respectively.
One of these additional group hierarchies deserves special discussion. The alt hierarchy has
very relaxed rules for newsgroup creation. Virtually anyone can create a group under the alt
hierarchy; however, creating a newsgroup under any other top-level group is extremely
difficult. The alt hierarchy carries many newsgroups that discuss topics that are out of the
mainstream of society. In fact, many people find some of the topics in the alt hierarchy to be
objectionable. Many Net debates on censorship have started because sites decided to ban part
or all of the newsgroups in the alt hierarchy.
News Distributions
In addition to grouping articles in hierarchies, Usenet also provides a feature for limiting the
spread of an article within the news system. New distributions provide a mechanism for
limiting articles to a particular geographic area. If a distribution is set to a particular area,
only sites within that distribution area receive the article. The system administrator at each
site decides what distributions apply to that site.
Why limit the distribution of an article? Suppose that you live in North Carolina and you're
posting a meeting announcement for a local user group meeting. It's unlikely that Usenet
readers in Australia are interested in your meeting. By limiting the distribution of your article
to the appropriate geographical area, you can save network bandwidth, reduce the cost of
sending your message, and reduce the aggravation of users around the world who have to read
your message.
You can limit the distribution of your article by including a Distribution: line in the header
of your article as you post it. Most newsreaders ask you for the distribution when you post an
article. After the colon in the Distribution: line, enter the appropriate geographical
distribution. Table 35.3 lists some commonly used news distribution areas.
Table 35.3 Commonly Used News Distribution Values
Value Explanation
local Typically, articles with a local distribution are limited to a group of local news
servers within your organization. This distribution is often used for local
organizational newsgroups.
nc Every state has a statewide distribution that's the same as the postal abbreviation
for the state. The Distribution:nc used in this example limits the article to
machines within the state of North Carolina.
us Sends the article to all Usenet sites in the United States.
na Sends the article to all Usenet sites in North America.
world Sends the article to every reachable Usenet site in the world. Typically, this
distribution is the default if no other distribution value is specified.
747
Usenet Culture
Your site may have some additional distributions that apply. You might be able to use
organization-wide or regional distributions to determine the scope of your article. In general,
you should try to pick a distribution that sends your article only to the areas where it will be
of interest.
No Central Authority
That Usenet has no central authority mystifies many people. Your local system administrator
really has authority over only the local system. No central group or organization dictates
policy or takes complaints. Despite this glaring lack of regulated structure, Usenet works
remarkably well. In fact, many people argue that it works better than if some central authority
existed.
How do things keep working in an orderly manner? Usenet is run by cooperation between
sites and by customs that have evolved over its life.
Usenet tends to be very good at policing itself. If a user starts to abuse the network, you can
rest assured that the user and his or her system administrator will get thousands of email
messages and several phone calls about the problem. This response usually results in quick
problem resolution.
Usenet Culture
Usenet has a particular culture all its own. You should take some time to try to become
familiar with the facets of this culture before diving in. Life on Usenet will be much easier if
you do.
In the past few years, many online communication services have added Usenet news as a
feature. As a result, tens of thousands of people who are new to Usenet have started reading
and posting Usenet news. Many of these users have complained about Usenet participants
being rude or generally unlike the users of their online service. Well, the culture of Usenet is
different from almost any other information service that you'll find. It's not better or
worseÐjust different. If you try to make allowances for differences in Net culture, you'll
probably find that your experiences on Usenet are a bit easier to handle.
Well over 1 million peopleÐprobably several million, though no one knows for sureÐread
and post Usenet news articles daily. These people are from all occupations, all walks of life,
and many different countries around the world. Because Usenet news is carried on computers Part
all over the world, it truly forms an international community. Many of the people that you
encounter on Usenet don't speak English as a primary language. You can't assume that the
VI
Ch
people reading your articles share your cultural background, ethnic group, religion, or social
values. The most that you can assume is that whoever reads your article is probably very 35
different from you in several ways.
748 Chapter 35 Surviving Usenet News
Tip #179 from
Steve One aspect of Usenet culture, the flame, is usually an unpleasant experience for new
users. A flame is a rude message, usually degrading and filled with insults that
someone posts in response to one of your articles. Unfortunately, as you'll see, you
can do very little about flames other than ignore them. Usenet is far too large a place
for you to try to make everyone happy, and some people really seem to like flaming
other people just for the fun of it. Perhaps they find it cheaper than psychotherapy. ...
Lack of Visual Reference
One problem with electronic communications is that you lack any kind of visual input during
the conversation. When you talk to another person face to face, you constantly receive
information on a conscious and subconscious level from the other person's body language.
Because you can't see the other people who read and post on Usenet, these visual cues are
missing. Because you typically use body language and visual cues to determine emotion and
feeling, you can easily misunderstand someone's post.
Fortunately, you can use several conventions on Usenet to replace part of the missing visual
cues. You can place added emphasis on a particular phrase by surrounding it in asterisks, as in
I *really* mean it!. Also, the use of all capital letters is considered shouting. If you
accidentally post an article with your Caps Lock key on, several people will probably tell you
about it.
You can also express emotions by writing them into your message. For example, if you make a
sarcastic statement, you can make sure that it's understood as such by adding < sarcasm > at
the end of the line. Smileys, also known as emoticons, also work to add emotion to your post. A
smiley is an ASCII representation of a face, which you look at sideways to see clearly. For
example, :-) is a happy, smiling face, and :-( is a sad face.
ON THE WEB
A canonical list of smileys is available at the following site. Have a look at it if you're really interested;
some of them are quite funny and original, but the happy and sad faces are the most commonly used.
Using some of the longer or rarer ones will likely result in the same effect as using obscure words in
common conversationÐyou'll just confuse people.
http://www.eff.org/pub/Net_culture/Folklore/Arts/smiley2.list
Newsgroup Culture
Just as people are different, each newsgroup on Usenet has a different culture. Each
newsgroup has a different subject focus and attracts different types of people. In some groups,
you may find large numbers of college students, whereas in others, you may find primarily
research scientists.
749
Reading and Posting News
Some of the more technical hierarchies, such as comp and sci, tend to be more oriented
toward factual discussion, although heated debates do take place. Members of these groups
are usually interested in discussing facts and issues related to some technical subject. When
you post here, make sure that you take time to carefully compose your article and have
references for the various points that you make.
The less technical hierarchies, such as rec, tend to be somewhat more opinion-oriented.
Remember, you'll probably get replies to your articles that reflect other people's opinions
that are quite different from your own. Groups in the talk hierarchy, along with some of the
misc groups, get into some very heated discussions. Many of these groups discuss very
sensitive topics such as abortion and gun control. Be careful in these groups if you're new to
Usenet. Make sure that you take time to get familiar with the group before posting. Be
prepared to receive strongly worded replies and email about your articles. Many of the people
here hold very strong beliefs.
When you first start reading a newsgroup, you should take some time to familiarize
yourself with the culture of the particular group before posting. Read the group for at
least a few days, and try to get a feel for the tone of the articles and the things that
are considered to be acceptable and unacceptable behavior. Look for a FAQ to get a
feel for the group (if one hasn't been posted recently in the newsgroup, a search
using one of the search engines on the Web can help).
In a few newsgroups, posting articles is restricted. These groups are known as moderated
newsgroups. Moderated newsgroups are managed by a person known as a moderator. All articles
posted to the group must be approved by the moderator before posting. The moderator
decides whether the content of the article is appropriate to the group and, if so, posts the
article to the group. Most news software automatically detects whether a newsgroup is
moderated, and if so, it emails your article to the moderator instead of posting it directly.
Reading and Posting News
Now that you're familiar with Usenet, you're ready to look at the basic process for reading
and posting news articles. The following sections describe reading and posting news in
general terms; the exact details depend on the news-reading software that you're using. Many
different software packages are available for interacting with news, and each of them is Part
different. Many people use a Web browser with an integrated newsreader, such as Netscape.
Others prefer to use a line-oriented tool, such as rn. These general concepts should apply
VI
Ch
35
across all news reading software.
Subscribing to Newsgroups
The first thing you should do when you start reading news is to decide which newsgroups you
want to read. The process of selecting the newsgroups is known as subscribing.
750 Chapter 35 Surviving Usenet News
Most newsreaders offer you a list of available newsgroups so that you can select the ones in
which you're interested. The actual process of subscribing varies among news-reading
software packages, but it usually involves selecting a series of newsgroups from a list. From
then on, only the groups that you've subscribed to are visible when you read news. You can
always subscribe to additional groups anytime you want or unsubscribe from a group in which
you're no longer interested.
Remember the earlier mention of over 20,000 newsgroups? If your newsreader is set to
download the entire list of all newsgroups carried by your news server, receiving this list may
take awhile.
Reading News
After you subscribe to your newsgroups, you can begin reading news. You select a newsgroup
from a list of your subscribed groups. Your newsreader displays a list of article subjects for the
various articles in the newsgroup. These subjects may be sorted in some order, or they may be
unsorted, depending on your newsreader. Some newsreaders can sort articles based on
subject, showing which articles are replies to other articles. This process is known as
threading.
When you select an article to read, you see several lines of information at the top of the
article. These lines make up the article header. The header contains lots of information about
the article, including the author, the date it was written, the subject, the newsgroups that the
article was posted to, and the path the article took to get to your site. You also might see
additional information, such as the organization the author is affiliated with and a set of
keywords that identify the content of an article.
Under most newsreaders, an article is marked as read when you look at it. Usually, only new
articles are displayed when you select a newsgroup. This means that after you look at an
article, it probably won't show up in your article list again. If you want to keep the article, you
can save it to disk or print it. You can also usually mark the article as unread so that your
newsreader displays it again the next time you go into the newsgroup. Many newsreaders also
allow you to list old articles; this way, you have a list of old news articles in a newsgroup that
are marked as having been read but haven't yet been deleted by the news system.
Replying via Email
After you read an article, you might decide that you want to comment on the topic under
discussion. If your information isn't of general interest to everyone in the newsgroup, you
might want to reply to the article via email, which most newsreaders allow you the option of
doing.
If you choose to reply through email, the newsreader software uses the information in the
article header to figure out the email address of the author and then invokes an email editor
for you to edit your message. You usually also have the option of including the original article
in your reply. If you do include the original article, make sure that you edit the original
751
Reading and Posting News
message to include only the relevant portions. After you finish editing your reply, you can send
your email message to the article's author.
Because of the common use of email, especially email addresses gleaned from Usenet
postings, many Usenet posters munge, or modify, their email addresses to stop automated
address collectors from being able to send them unsolicited commercial email, known as
spam. The poster's email address may have an obvious false entry and look something like
[email protected]. Alternatively, the poster may have instructions in his or
her sigfile on what to change in the reply-to address to actually reach him or her. For
example, a poster with a false address of [email protected] might have
instructions in the sigfile to ``Replace tails with wings to reply."
Posting an Article
The act of creating a news article and sending it out through the Usenet system is known as
posting an article. When you decide to post an article, you can either post a follow-up article to
another article or create a new article on a new subject. Your newsreader typically has
different commands for the different types of posts that it can perform.
Posting a Follow-Up
A follow-up article is a reply to another article. This article stays in the same subject thread as
the original article and is shown as a reply by threaded newsreaders.
When you post a follow-up, you can choose to include the original post. Including parts of
the original post is a good way to provide a frame of reference for your reply. Remember that
several days may pass between the time some sites see the original article and your reply. If
you do choose to include the original article, try to include, or quote, only the parts of the
article that are relevant to your reply. Trying to wade through several levels of included files
and quotes looking for the new information gets tedious. Also, some news servers will reject
your reply if the quoted material is more than a certain percentage of your entire message,
depending on the policies of the administrators.
You should check the Subject line to make sure that the subject still accurately reflects the
content of your post, and change it if you're now discussing a new topic. Also, take a look at
the Newsgroup line to make sure that your follow-up is going to the appropriate newsgroups.
In particular, consider whether sending the message to multiple newsgroups is appropriate or
whether the topic is relevant to only one or two of the original newsgroups.
Part
VI
Posting a New Article
If you decide to start a thread of discussion on a new subject, you want to post a new article
instead of a follow-up. The mechanics of posting the article are very similar to those of Ch
posting a follow-up. You give the appropriate command to your newsreader; your newsreader 35
asks for some information, such as the destination newsgroups, subject, and distribution; and
you're placed into an editor. The major difference is that you're creating a subject thread
instead of replying to one.
752 Chapter 35 Surviving Usenet News
Tip #180 from
Steve A complete document on Usenet writing style is posted regularly to the newsgroup
news.announce.newusers.
You should think about several points as you write your article. You can think of them as
``Usenet Style Tips'' if you want. These tips cover the format of your article and its content.
You should keep your lines fewer than 80 characters long. Many terminals can't display lines
that are more than 80 characters. Similarly, you should try to keep the length of your article
under 1,000 lines or so. Some sites are still running old versions of the news transport
software, and long articles can cause them problems.
Tip #181 from
Steve You should not post a file such as a graphic or application to a newsgroup not
intended for such files. Most newsgroups that are intended for files are have the word
.binaries at the end of the newsgroup name.
You probably want to create a signature file that's automatically included at the end of every
post. Most newsreaders support signature files, although the exact mechanism varies
depending on your software. Most people put their names and email addresses in their
signature files, along with their geographical locations. Some people add witty quotes or
small ASCII pictures. Try to avoid having a large signature file. It is considered bad
netiquette to include your full name, nicknames, an inspirational quote consisting of
Whitman's Leaves of Grass, and a 20-line ASCII art drawing of your car. A good rule is to limit
yourself to four lines. Some news software automatically limits your signature to four lines
or so.
You need to give a subject to your article when you post it. Try to pick a subject line that's
short yet descriptive. Thousands of people scan the subjects in any particular newsgroup, and
you want them to be able to pick out your article if it's of interest to them. Also, carefully
consider which newsgroups you're going to post your article to. Most newsreaders allow you
to post an article to more than one newsgroup. You should post to only the smallest number
of groups that you need. Remember that thousands of people are reading each newsgroup.
Netiquette on Usenet
Throughout this chapter, the importance of being aware of how the tone and content of your
message are interpreted has been stressed. This general consideration of behavior on Usenet,
and the Internet in general, has its own termÐnetiquette. Netiquette applies to all areas of the
Internet, including electronic mail.
753
Project: Using the rn Newsreader
The term netiquette simply refers to ``proper and polite'' behavior as it applies to Usenet news.
Most of the time, you should have no real problems on Usenet as long as you remember that
it's a very big and diverse place. Not everyone on Usenet shares your background, beliefs, or
values, and you should try to remember this fact as you post articles.
Make sure that you clearly communicate your ideas in your posts. Because of the lack of body
language and the delay between posts and replies, interpreting someone's meaning
incorrectly is surprisingly easy. Also, remember that many participants don't speak English as
a native language and may be unaware of local idioms and sarcasm.
Blatant commercial advertising is frowned on in Usenet news. Appropriate newsgroups do
exist for advertising products and services. Similarly, you should not post chain articles, such
as the infamous MAKE.MONEY.FAST or Craig Shergold get-well card article. These
articles have been circulating around Usenet for years, and you (and your system
administrator) will incur the wrath of thousands of people instantly if you post one of them.
Resist the urge to post flames, especially spelling and grammar flames. Even though flames
seem to be a permanent part of the Usenet ``landscape,'' these personal attacks and raving
messages accomplish little. If someone should flame you for one of your posts, take time to
calm down and carefully consider how to respond; the best solution might be not to respond
at all. Sometimes you may receive a flame, but a calm response from you may elicit an
apology from the person who flamed you. If you just zip off another flame in anger, you only
escalate the problem. Remember that the person on the other end is really a person, not a
computer.
If a user is causing a real problem, you can add him or her to your kill file, a
configuration file for your newsreader that contains a list of users or subjects.
Anything that appears in your kill file is automatically not displayed when you read
news. Most newsreaders support some version of a kill file. Using this file is a fairly
painless way to cut down the noise from really annoying users.
In general, a little common sense and courtesy go a long way in avoiding any problems on
Usenet. However, remember that Usenet is a huge place. Simply too many people are out
there for you to try to make everyone happy. Eventually, someone will get angry over one of
your posts, and you'll probably be flamed.
Part
Project: Using the rn Newsreader
VI
Ch
35
Many different types of news-reading software are availableÐfar too many to describe in this
chapter. The rn newsreader is a very common news-reading program that can be found on
almost every UNIX variant. It was developed by Larry Wall and is widely available. Although
rn isn't the easiest newsreader to use, nor does it have some of the fanciest features, it's still
754 Chapter 35 Surviving Usenet News
one of the most popular newsreaders in existence. rn allows you to read news via an ASCII
interface that's suitable for local work on a terminal or from a remote network session.
Another newsreader, trn, is quite popular and is distributed with many distributions
of Linux. The trn newsreader is almost identical to rn except for the threading
support. For compatibility with a wide variety of UNIX systems, only the rn
newsreader is discussed in this chapter. For more information on the threading
capabilities of trn, refer to its Linux man page.
When you start rn for the first time, you see a message welcoming you to the program,
followed by a list of newsgroups. You have the opportunity to subscribe to different groups at
this point. If your site carries a large number of groups, setting up your initial subscription
information can be quite time-consuming. rn saves your subscription information in your
home directory in a file named .newsrc.
After you complete your subscriptions, rn places you in a newsgroup selection mode. The
name of each of your subscribed newsgroups is displayed one at a time. You can enter the
newsgroup and start reading articles by pressing the y key, skip to the next group by pressing
the n key, or go to the previous newsgroup by pressing the q key. You can also get a list of
subjects in the newsgroup by pressing the = key at the newsgroup prompt. Most of the
commands in rn and trn are one-character commands, and help is available at every
command prompt by pressing the h key.
After you select a newsgroup to read, you enter article-selection mode. In this mode, several
commands can help you navigate the articles in the newsgroup. Table 35.4 lists some of the
commands available in article-selection mode.
Table 35.4 Some Commands Available in Article-Selection Mode
Command Description
n+spacebar Scans forward for the next unread article. The spacebar does this only at the
end of the article, at the article-selection prompt.
spacebar Shows the next page of the current article if not at the article-selection
prompt.
Shift+n Goes to the next article.
Ctrl+Shift+n Goes to the next article with the same subject as the current article.
p Scans backward for the previous unread article; stays at the current article if
none is found.
Shift+p Goes to the previous article.
Ctrl+Shift+r Goes to the last previous article with the subject that's the same as the
current article.
h Displays help for article-selection mode.
755
Project: Using the rn Newsreader
Table 35.4 Some Commands Available in Article-Selection Mode
Command Description
r Replies to the article author via email.
Shift+r Replies to the article author via email and includes the current article.
f Posts a follow-up article.
Shift+f Posts a follow-up article and includes the original article in the new article.
s+ filename Saves the current article to a file named filename.
q Quits the current group and returns to newsgroup selection mode.
Table 35.4 shows only some of the options available within rn and trn. These feature-rich
programs allow lots of user customization. Refer to the man pages and the online help for
more information.
Part
VI
Ch
35
PART
VII
Setting Up Linux Internet Servers
36 Getting Started with Apache 759
37 Configuring an FTP Server 781
38 Configuring Domain Name Service (DNS) 793
39 Configuring Email 809
40 Configuring a Usenet News Service 825
CHAPTER
36
Getting Started with Apache
In this chapter by Steve Burnett
Installing Apache 760
Establishing the File Hierarchy 760
Basic Configuration 761
Starting Apache 765
Debugging Server Startup 766
Secure Transactions with SSL 767
Special Modules 768
Understanding Security Issues 777
Project: Adding Customized Error Messages 780
760 Chapter 36 Getting Started with Apache
Installing Apache
To use a Linux system as a Web server, you must install special server software on your
system. Two of the most popular UNIX Web server packages are Apache and NCSA's httpd.
In fact, a May 1999 survey showed that Apache and its derivatives accounted for more than 60
percent of all installed Web servers, and it is used on approximately 3 million Web servers.
Although this chapter is specific to the Apache server, the vocabulary is certainly applicable to
other Web servers. The NCSA family of servers has much in common with Apache with
respect to configuration files because Apache was derived originally from the NCSA 1.3
server, and maintaining backward compatibility with existing NCSA servers was originally a
mandate with the development team. Such compatibility is less complete now that NCSA and
Apache development paths have diverged.
Apache is known to compile on just about every UNIX variant: Solaris 2.x, SunOS 4.1.x, Irix
5.x and 6.x, Linux, FreeBSD/NetBSD/BSDI, HP-UX, AIX, Ultrix, OSF1, NeXT, Sequent,
A/UX, SCO, UTS, Apollo Domain/OS, QNX, as well as OS/2; a Windows NT 4.0 port has
been completed as of this writing. Portability has been a high priority for the development
team.
Apache binaries and their sources are included with most distributions of Linux. The
complete source code for Apache is also provided. Because Apache binaries are included on
the CD-ROMs, you can skip the compilation process and move on to the next section if
you're in a hurry to get Apache up and running. However, if you ever want to add new
modules or tweak the functionality provided by Apache, you should know how to compile it.
You can find instructions on compiling Apache on the Apache Software Foundation site at
http://www.apache.org/; the instructions are not covered in this book.
Establishing the File Hierarchy
The next step in the process of setting up a server is to make some fundamental decisions
regarding where on the file system different parts of the server will reside. You should write
down your decisions for each of these locations; you will need them in the next section, ``Basic
Configuration'':
& Where is the server root? It is the subdirectory in which the server will reside and from
which the conf/ directory, the logs/ and cgi-bin/ subdirectories, and other server-
related directories lead. The default suggestion is to use /usr/local/apache, although
the usual server root is /pub/htdocs. You can have your configuration files and log
files in other locations. The server root was designed to be a convenient place to keep
everything server-related together. Also, if the server crashes and leaves a core file, that
file will be found in the server root directory.
& Where is the document root? The document root is the directory in which all your
HTML and other media reside. A file called myfile.html in the document root would
be referenced as http://host.com/index.html. This directory can be a subdirectory
of the server root, or it can be outside the server root and in its own directory. It's
761
Basic Configuration
commonly located as a subdirectory of the server root and named htdocs. If, for more disk
Part
space or other reasons, you choose to move the document root out of the server root
directory, you should give it a short nameÐfor example, /home/www or /www/htdocs. If VII
you're implementing a Web server on top of an FTP server, for example, you might
Ch
&
want to point the document root to /home/ftp/pub.
Where will the Web server log files be kept? This space should have a fairly large
36
working area, depending on how busy you estimate your server will be. For a point of
reference, a site with 100,000 hits per day (which would fall under moderate traffic,
relatively speaking) can expect to generate 15MB of log file information per day. For
performance reasons, it's usually best to have the log directory on a separate disk
partition, or even a separate disk drive and drive controller altogether, because on even
a moderately busy server the access log can be written to several times per second.
Basic Configuration
Apache has three separate configuration files. This model goes back to NCSA, and the
reasoning is sound: Administrative configuration falls largely into three main areas, so setting
them up as separate files allows Webmasters to give different write permissions to each if they
so desire.
You can find the configuration files for Apache in the conf/ subdirectory of the server root
directory. Each has been provided with a -dist filename suffix; it's recommended that you
make a copy without the -dist and edit those new files, keeping the -dist versions as backups
and reference.
Tip #182 from
Steve Although Apache comes with and uses the following three different configuration
files, it treats them all identically:
. httpd.conf
. srm.conf
. access.conf
Because httpd.conf is the first file read by Apache, you can copy the contents of the
other two into httpd.conf and delete them if you like. If you delete the two
unneeded files, you should add the following directives to httpd.conf to prevent
Apache from generating error messages about the missing srm.conf and
access.conf files:
AccessConfig /dev/null
ResourceConfig /dev/null
This chapter presents the three configuration files separately.
762 Chapter 36 Getting Started with Apache
The basic format of the configuration files is a combination of a shell-like interface and
pseudo-HTML. The elemental unit is the directive, which can take a number of arguments, as
shown in the following syntax:
Directiveargumentargument....
For example, the directive might look like one of the following:
Port 80
AddIcon /icons/back.gif ..
You can also group directives together inside certain pseudo-HTML tags. Unlike HTML,
these tags should be on their own lines, as in the following example of the Virtualhost
directive:
<Virtualhost www.myhost.com>
DocumentRoot /www/htdocs/myhost.com
ServerName www.myhost.com
</Virtualhost>
httpd.conf
The default configuration file Apache reads is httpd.conf. This file sets the basic system-
level information about the server, such as what port it binds to, which users it runs under,
and so on. If you aren't the system administrator of the site at which you're installing the
server, you might want to ask the administrator to help you with these questions.
The essential items to cover in this file include the following:
& Port number
For example:
Port 80
This number indicates the TCP/IP port number to which the Web server binds. Port
80 is the default port in http: URLs. In other words, http://www.myhost.com/ is
equivalent to http://www.myhost.com:80/.
Tip #183 from
Steve For a number of reasons, however, you might want to run your server on a different
port; for example, a server might already be running on port 80 (not secrecy,
however; port scanners are too common for security through obscurity to be a valid
option here).
763
Basic Configuration
& User #number_or_uid
Part
Group #number_or_uid
For example:
VII
User nobody Ch
Group nogroup 36
You need to launch Apache as root to bind to a port lower than 1024. Immediately
after grabbing the port, Apache changes its effective user ID to something else,
typically as user nobody. Apache's changing of its user ID is very important for security
reasons.
This user ID needs to be able to read files in the document root, and it must have read
permission on the configuration files. The argument should be the actual username;
however, if you want to give a numeric user ID, you can prepend the number with a
#
pound sign ( ). The Group directive follows the same principle: You must decide
which group ID you want the server to run with.
Tip #184 from
Steve Running your Web server as root means that any hole in the server (be it through the
server itself or through a CGI script, which is much more likely) could be exploited by
an outside user trying to run a command on your machine. Thus, setting the user to
nobody, www, or some other reasonably innocuous user ID is the safest bet.
& ServerAdmin email_address
You need to set the email address of a user who can receive mail related to the actions
of the server. In the case of a server error, the browser visiting your site receives a
message to the effect of ``please report this problem to [email protected].'' In the
future, Apache might send warning email to the ServerAdmin user if it encounters a
major systems-related problem.
& ServerRoot directory
For example:
ServerRoot /usr/local/apache
You can set the server root you decided on earlier. Here, you give the full path and
don't end it with a slash.
& ErrorLog directory/filename
TransferLog directory/filename
You can specify exactly where to log errors and Web accesses. If the filename you give
doesn't start with a slash, it's presumed to be relative to the server root directory. I
suggested earlier that the log files be sent to a separate directory outside the server
root; here, you specify the logging directory and the name of the log files within that
directory.
764 Chapter 36 Getting Started with Apache
& ServerName DNS_hostname
At times, the Web server needs to know the host name it's being referred to as, which
can be different from its real host name. For example, the name www.myhost.com
might actually be a DNS alias for gateway.myhost.com. In this case, you don't want
the URLs generated by the server to be http://gateway.myhost.com/. ServerName
allows you to set that name precisely.
srm.conf
The second configuration file to cover before launch is srm.conf. The important elements to
set in this file include the following:
& DocumentRoot directory
As described before, this directory is the root level of your tree of documents, which
could be either /usr/local/apache/htdocs or /www/htdocs. This directory must exist
and be readable by the user (usually nobody) the Web server runs as.
& ScriptAlias request_path_alias directory
ScriptAlias lets you specify that a particular directory outside the document root can
be aliased to a path in the request and that objects in that directory are executed rather
than simply read from the file system. For example, the default offering
ScriptAlias /cgi-bin/ /usr/local/apache/cgi-bin/
means that a request forhttp://www.myhost.com/cgi-bin/fortune executes the
program /usr/local/apache/cgi-bin/fortune. Apache comes bundled with a
number of useful beginner CGI scripts, simple shell scripts that illustrate CGI
programming.
Finally, the directory containing the CGI scripts should not be under the document
root. Bizarre interactions between the code that handles ScriptAlias and the code
that handles request/pathname resolution could cause problems.
access.conf
access.conf is structured more rigidly than the other configuration files; the content is
contained within <Directory></Directory>, pseudo-HTML tags that define the scope of the
directives listed within.
So, for example, the directives between the following two code lines affect everything located
under the /www/htdocs directory:
<Directory /www/htdocs>
</Directory>
Furthermore, wildcards can be used. For example, the following:
765
Starting Apache
<Directory /www/htdocs/*/archives/>
.... Part
</Directory>
VII
applies to /www/htdocs/list1/archives/, /www/htdocs/list2/archives/, and so on.
Ch
36
Tip #185 from
Steve Although this chapter is still useful, release 1.3.0 and higher of Apache include a GNU
Autoconf-style front end, which supports all previous configuration options as well as
the enhancements in 1.3.0 and above. Use of that interface for configuration is
recommended in general.
User Directories
Sometimes sites with many users enable their users to manage their own parts of the Web
tree in their own directories. The users do so by using the URL semantics
http://myhost.com/~user/
where ~user is actually an alias to a directory in the users' home directories. This approach is
different from using the Alias directive, which can map only a particular pseudo-directory
into an actual directory. In this case, you want ~user to map to something like /home/user/
public_html. Because the number of ``users'' can be very high, some sort of macro is useful
here. That macro is the UserDir directive.
With UserDir, you specify the subdirectories within the users' home directories where they
can put content, and those subdirectories are mapped to the ~user URL. So, in other words,
the default
UserDir public_html
causes a request for
http://myhost.com/~dave/index.html
to cause a lookup for the UNIX file
/home/dave/public_html/index.html
presuming that /home/dave is Dave's home directory.
Starting Apache
To start Apache, simply run the binary you compiled earlier (or your precompiled binary)
with the -f flag pointing to the httpd.conf file also created earlier, as in this example:
/usr/local/apache/src/httpd -f /usr/local/apache/conf/httpd.conf
Several Linux distributions (Red Hat Linux included) include a startup script for the Web
server if you install Apache.
766 Chapter 36 Getting Started with Apache
At this point, you would be wise to use the ps command to see whether httpd is running.
Typically, something like ps -aux | grep will suffice. You might be surprised to see a number
of simultaneous httpd processes running. What's going on?
The first Web servers, such as CERN and NCSA, used the model of one main Web server
cloning itself with every single request that came in. The clone would respond to the request,
while the original server returned to listening to the port for another request. Although this
design was certainly simple and robust, the act of cloning (or, in UNIX terms, forking) was an
expensive operation under UNIX, so loads above a couple hits per second were quite
punishing even on the nicest hardware. It was also difficult to implement any sort of throttling,
which reduced the amount of cloning that took place. When the number of clones was very
high, it was hard for the original server to know how many clones were still around. Thus,
servers had no easy way to refuse or delay connections based on a lack of resources.
Apache, like some of the other UNIX-based Web servers, instead uses the model of a group
of persistent children running in parallel. The children are coordinated by a parent process,
which can tell how many children are alive, spawn new children (if necessary), and even
terminate old children if many are idle, depending on the situation. ( Parent and child are the
actual UNIX terms.)
Now, let's get back to the server. At this point, you can start your Web browser and point it
to your local server. (You should use the usual http:// format and add the ServerName
parameter you defined in the httpd.conf file.) Does it work? If all goes well, you should be
able to see a directory index listing of everything in the document root directory, or if an
index.html is located in that directory, you see the contents of that file. By default, Apache
installs a Web page in the document root telling the Webmaster that the Apache installation
was successful.
Debugging Server Startup
Apache is usually pretty good about giving meaningful error messages, but some are
explained in more detail in the following sections.
Server Startup Error Messages
You might see the following messages upon server startup. Note that the descriptions follow
the messages.
httpd: could not open document config file .....
fopen: No such file or directory
These open file error messages are usually the result of giving just a relative path to the -f
argument, so Apache looks for the file or files relative to the compiled-in server root (what's
set in src/httpd.h) instead of those relative to the directory you're in. You must give the full
path or the path relative to the compiled-in server root.
httpd: could not bind to port [X]
bind: Operation not permitted
767
Debugging Server Startup
The port and bind error messages are most likely caused by attempting to run the server on a
Part
VII
port below 1024 without launching it as root. Most UNIX operating systems, including
Linux, prevent people without root access from trying to launch any type of server on a port
less than 1024. If you launch the server as root, the error message should disappear.
Ch
httpd: could not bind to port
bind: Address already in use
36
These port and bind error messages mean that something is already running on your machine
at the port you've specified. Do you have another Web server running? No standard UNIX
mechanism is available for determining what's running on what ports; on most systems, the
file /etc/services can tell you what the most common daemons are, but it's not a complete
list. You could also try using the netstat command, with various options such as -a.
httpd: bad user name ....
httpd: bad group name ....
Bad user or group name error messages mean that the user or the group you specified in
httpd.conf doesn't actually exist on your system. You might see errors telling you that
particular files or directories don't exist. If the files appear to be there, make sure that they're
readable by the user IDs that the server runs as (that is, both root and nobody).
Initial Server Startup Error Messages
Suppose that Apache has started up and, according to ps, it's actually running. When you go
to the site, however, you experience the following problems or error messages:
No connection at all.
First, you should make sure that no firewalls that would filter out packets to the server are
between you and the server. Second, you can try using telnet to the port you launched the
Web server onÐfor example, telnet myhost.com 80. Connected to
If you don't get a
myhost.com message back, your connection isn't even making it to the server in the first place.
403 Access Forbidden
Your document root directory may be unreadable, or you might have something in your
access.conf file that prevents access to your site from the machine where your Web browser
is running.
500 Server Error
Is your front page a CGI script? The script might be failing.
The errors listed here are the most common errors made in initial server startups. If you can
confirm that contact with the server is actually being made, the next best place to look for
error information is in the ErrorLog.
768 Chapter 36 Getting Started with Apache
Secure Transactions with SSL
This section takes a slight detour and discusses a variant of the Apache Web server, Apache-
SSL, which can conduct secure transactions over the Secure Sockets Layer (SSL) protocol.
SSL is an RSA public-key-based encryption protocol developed by Netscape Communica-
tions for use in the Netscape Navigator browser and Netscape Web servers.
Eric Young, author of the widely used libdes package, with Tim Hudson wrote a library that
implements SSL, eponymously named SSLeay. The SSLeay package has since expanded to
become an all-purpose cryptography and certificate-handling library, while retaining the
same name, SSLeay.
Ben Laurie, a member of the Apache Group, then took the SSLeay library and interfaced it
with the Apache server, making his patches available to people on the Net. Sameer Parekh of
Community ConneXion, Inc. (hereafter referred to as C2) then took Ben Laurie's patches
and built a package legal for use within the United States.
Because the RSA technology used by SSL in the United States is covered by patents owned by
RSA Data Security, Inc. (RSADSI) ( www.rsa.com), it isn't legal to use the SSLeay package
``out-of-the-box'' within the United States. C2 licensed the RSA technology to make use of
the package legal within the United States by using the RSAREF package, produced by
RSADSI and Consensus Development Corporation ( www.consensus.com).
Due to export restrictions, someone outside the United States cannot legally download and
install the C2 Apache-SSL package. In fact, the SSL patches could not be included on the
CD-ROM with this book because the book would suddenly have earned the label
``munition,'' and clearance from the U.S. government to export the book would have been
required!
To learn more about SSL and Apache, you can go to http://www.apache-ssl.org.
As an alternative to Apache-SSL, you can use a module called mod_ssl, which is based on
Apache-SSL and provides a more supported implementation. You can find the Apache
Interface to the SSLeay module on the Web at http://www.engelschall.com/sw/mod_ssl/.
Special Modules
Most of the functionality that distinguishes Apache from the competition has been
implemented as modules to the Apache API. The implementation of these modules has been
extremely useful in allowing functionality to evolve separately from the rest of the server and
for allowing for performance tuning. The following sections cover that extra functionality in
detail.
Server-Side Includes
Server-side includes are best described as a preprocessing language for HTML. The
``processing'' takes place on the server side. As such, visitors to your site never need to know
769
Special Modules
that you use server-side includes, and thus they require no special client software. The format of
Part
these includes looks something like the following:
<!--#directiveattribute="value'' --> VII
Ch
Sometimes a given ``directive'' can have more than one attribute at the same time. The funky
syntax is due to the desire to hide this functionality within an SGML comment; that way, 36
your regular HTML validation tools work without your having to learn new tags or anything.
The syntax is important; leaving off the final --, for example, will result in errors.
#include
The #include directive is probably the most commonly used. You use it to insert another
HTML file into the HTML document. The allowed attributes for #include are virtual and
file. The functionality of the file attribute is a subset of that provided by the virtual
attribute, and it exists mostly for backward compatibility, so its use isn't recommended.
The virtual attribute tells the server to treat the value of the attribute as a request for a
relative linkÐmeaning that you can use ../ to locate objects above the directory and that
other transformations, such as Alias, will apply. The following is an example of such:
<!--#include virtual="quote.txt'' -->
<!--#include virtual="/toolbar/footer.html'' -->
<!--#include virtual=''../footer.html'' -->
#exec
The #exec directive is used to run a script on the server side and insert its output into the SSI
(server-side includes) document being processed. You have two choices: executing a CGI
script by using the cgi attribute or executing a shell command by using the cmd attribute. For
example,
<!--#exec cgi="counter.cgi'' -->
takes the output of the CGI program counter.cgi and inserts it into the document.
The CGI output still has to include the "text/html" content-type header; otherwise,
an error occurs.
Likewise,
<!--#exec cmd="ls -l'' -->
takes the output of a call to ls -l in the document's directory and inserts it into the output
page as a replacement for the #exec command. Like the file attribute for the #include
directive, this type of #exec command is mostly for backward compatibility because it's
something of a security hole in an untrusted environment.
770 Chapter 36 Getting Started with Apache
There are definitely security concerns with allowing users access to CGI functionality
and even greater concerns with using #exec cmd, such as
cmd="cat /etc/passwd''
If, as site administrator, you want to let users use server-side includes but not use the
#exec directive, you can set IncludesNOEXEC as an option for the directory in the
access configurations.
#echo
The #echo directive has one attributeÐvarÐwhose value is any CGI environment variable as
well as a small list of other variables, as shown in Table 36.1.
Table 36.1 Values for the var Attribute
Attribute Definition
DATE_GMT The current date in Greenwich mean time.
DATE_LOCAL The current date in the local time zone.
DOCUMENT_NAME The file system name of the SSI document, not including the directories
below it.
DOCUMENT_URI The file system URI of the SSI document. URI stands for Uniform Resource
Identifier. In a Uniform Resource Locator (URL) of the format http://
host/path/file, the URI is the /path/file part.
LAST_MODIFIED The date the SSI document was modified.
For example, the command
<!--#echo var="DATE_LOCAL''-->
inserts something along the lines of Wednesday, 03-Mar-99 10:44:54 GMT into the document.
#fsize, #flastmod
The #fsize and #flastmod directives print the size and the last-modified date, respectively, of
any object given by the URI listed in the file or virtual attribute, as in the #include
directive. For example, the command
<!--#fsize file="index.html''-->
returns the size of the index.html file in that directory.
#config
You can modify the rendering of certain SSI directives by using the #config directive. The
sizefmt attribute controls the rendering of the #fsize directive with values of bytes or
771
Special Modules
abbrev. The exact number of bytes is printed when bytes is given, whereas an abbreviated
version of the size (in KB for kilobytes or MB for megabytes) is given when abbrev (the
Part
default) is set. For example, a snippet of SSI HTML like VII
<!--#config sizefmt="bytes''--> Ch
The index.html file is <!--#fsize virtual="index.html'' --> bytes
36
returns The index.html file is 4,522 bytes. Meanwhile,
<!--#config sizefmt="abbrev'' -->
returns The index.html file is 4K bytes.
The timefmt directive controls the rendering of the date in the DATE_LOCAL, DATE_GMT, and
LAST_MODIFIED values for the #echo directive. It uses the same format as the strftime call. (In
fact, the server does call strftime, a system call that formats the time in a string of specified
length.) The string format consists of variables that begin with %. For example, %H is the hour
of the day, in 24-hour format. For directions on how to construct a strftime-format date
string, you can consult strftime's man page for a list of variables.
An example might be
<!--#config timefmt="%Y/%m/%d-%H:%M:%S'' -->
with the resulting date string for January 2, 1999, at 12:30 in the afternoon as
1999/01/02-12:30:00
Finally, the last attribute the #config directive can take is errmsg, which is simply the error to
print if any problems occur parsing the document. For example, the right default is the
following:
<!--#config errmsg="An error occurred while processing this directive'' -->
Cookies
The use of HTTP cookies are a method for maintaining statefulness in a stateless protocol.
What does this mean? In HTTP, a session between a client and a server typically spans many
separate actual TCP connections, thus making it difficult to tie together accesses into an
application that requires state, such as a shopping-cart application. Cookies provide a
solution to that problem. As implemented by Netscape in its browser and subsequently by
many others, servers can assign clients a cookie, meaning some sort of opaque string whose
meaning is significant only to the server itself, and then the client can give that cookie back to
the server on subsequent requests.
The mod_cookies module nicely handles the details of assigning unique cookies to every
visitor, based on the visitor's host name and a random number. This cookie can be accessed
from the CGI environment as the HTTP_COOKIE environment variable, for the same reason
that all HTTP headers are accessible to CGI applications. The CGI scripts can use this
cookie as a key in a session-tracking database, or cookies can be logged and tallied up to get a
good, if undercounted, estimate of the total number of users that visited a site, not just the
number of hits or even number of unique domains.
772 Chapter 36 Getting Started with Apache
Happily, there are no configuration issues here. You can simply compile with mod_cookies,
and away you go. The process couldn't be easier.
Tip #186 from
Steve For security reasons, some users severely dislike cookie configurations. Think carefully
if you need cookies before using them.
Configurable Logging
For most folks, the default log file format (also known as CommonLogfileFormat, or CLF)
doesn't provide enough information when it comes to doing a serious analysis of the efficacy
of a Web site. It provides basic numbers in terms of raw hits, pages accessed, hosts accessing,
timestamps, and so forth, but it fails to capture the ``referring'' URL, the browser being used,
and any cookies being used. So you can get more data for your log files in two ways: by using
the NCSA-compatibility directives for logging certain bits of information to separate
browsers or by using Apache's own totally configurable log file format.
NCSA Compatibility
For compatibility with the NCSA 1.4 Web server, two modules were added. These modules
log the User-Agent and Referer headers from the HTTP request stream.
User-Agent, which is the header most browsers send, identifies what software the browser is
using. Logging of this header can be activated by an AgentLog directive in the srm.conf file or
in a virtual-host-specific section. This directive takes one argument, the name of the file to
which the user-agents are logged, as in this example:
AgentLog logs/agent_log
To use the AgentLog directive, you need to ensure that the mod_log_agent module has been
compiled and linked to the server.
Similarly, the Referer header is sent by the browser to indicate the tail end of a link. In other
words, when you're on a page with an URL of ``A,'' and a link on that page has an URL of
``B,'' and you follow that link, the request for page ``B'' includes a Referer header with the
URL of ``A.'' This header is very useful for finding what sites link to your site and what
proportion of traffic they account for.
The logging of the Referer header is activated by a RefererLog directive, which points to the
file to which the referers get logged:
RefererLog logs/referer_log
One other option the Referer logging module provides is RefererIgnore, a directive that
allows you to ignore Referer headers. RefererIgnoreis useful for weeding out the referers
773
Special Modules
from your own site, if all you're interested in is links to you from other sites. For example, if your
site is www.myhost.com, you might want to use the following: Part
RefererIgnore www.myhost.com VII
Remember that logging of the Referer header requires compiling and linking in Ch
mod_log_referer. 36
Totally Configurable Logging
The previous modules were provided, like many Apache features, for backward compatibility.
They have some problems, though. Because they don't contain any other information about
the request they're logging from, telling which Referer fields went to which specific objects
on your site is nearly impossible. Ideally, all the information about a transaction with the
server can be logged into one file, extending the Common Logfile Format or replacing it
altogether. Well, such a beast exists in the mod_log_config module.
The mod_log_config module implements the LogFormat directive, which takes as its
argument a string, with variables beginning with % to indicate different pieces of data from the
request. Table 36.2 lists the variables.
Table 36.2 Variables for the LogFormat Directive
Variable Definition
%h Remote host.
%l Remote identification via identd.
%u Remote user, as determined by any user authentication that may take place. If
the user wasn't authenticated and the status of the request is a 401
(authorization error), this field may be bogus.
%t Common Logfile Format for time.
%r First line of request.
%s Status. For requests that are internally redirected, this is the status of the
original request; %>s gives the last request.
%b Bytes sent.
%{STUFF}i Contents of STUFF: header line(s) in the request from the client to the server.
%{STUFF}o Contents of STUFF: header line(s) in the response from the server to the client.
For example, if you want to capture in your log just the remote host name, the object
requested, and the timestamp, you would use the following:
LogFormat ``%h \"%r\'' %t''
This command logs things that look like the following:
host.outsider.com ``GET / HTTP/1.0'' [06/Mar/1996:10:15:17]
774 Chapter 36 Getting Started with Apache
You Can Quote Me on This
You really have to use quotation marks around the request variable. The configurable logging module
automatically interprets the values of the variables rather than just reads the variable name. You use a slash-
quote, \", to indicate that you want an actual quotation mark character rather than the end of the string. For
example, if you want to add logging of the User-Agent string, your log format becomes
LogFormat ``%h \"%r\'' %t \"%{User-Agent}i\"''
Because the User-Agent field typically has spaces in it, it too should be quoted. Suppose that you want to
capture the Referer field:
LogFormat ``%h \"%r\'' %t %{Referer}i''
You don't need the escaping quotation marks because Referer headers, as URLs, don't have spaces in them.
However, if you're building a mission-critical application, you might as well quote it also, because the Referer
header is supplied by the client and, thus, you have no guarantees about its format.
The default log file format is the Common Logfile Format (CLF), which is expressed as
follows:
LogFormat ``%h %l %u %t \"%r\'' %s %b''
In fact, most existing log file analysis tools for CLF ignore extra fields tacked onto the end.
To capture the most important extra information and yet have that information still be
parsable by those tools, you might want to use this format:
LogFormat ``%h %l %u %t \"%r\'' %s %b %{Referer}i \"%{User-Agent}i\"''
Tip #187 from
Steve If you want even more control over what gets logged, you can use the configurable
logging module to implement a simple conditional test for variables. This way, you
can configure it to log variables only when a particular status code isÐor
isn'tÐreturned. The format for trapping a status code is to insert a comma-separated
list of those codes between the % and the letter of the variable:
%404,403{Referer}i
This example means that the Referer header is logged only if the status returned by
the server is 404 Not Found or 403 Access Denied. All other times, just a - is
logged. Having only 403 or 404 errors logged might be useful if you care about using
Referer only to find old links that point to resources no longer available.
To negate the Referer !
status code, you put an exclamation point ( ) at the beginning of the
list of status codes. For example,
%!401u
logs the user in any user authentication transaction, unless the authentication failed, in which
case you probably don't want to see the name of the bogus user anyway.
Remember that, like many functions, logging functions can be configured per virtual host.
Thus, if you want all logs from all virtual hosts on the same server to go to the same log, you
might want to do something like
775
Special Modules
LogFormat ``hosta ....''
Part
in the <VirtualHost>
LogFormat ``hostb ....''
sections for hosta and
VII
Ch
in the <VirtualHost>
section ``Virtual Hosts."
sections for hostb. More details about virtual hosts appear later in the
36
You have to compile in mod_log_config to configure logging on a ``per-virtual-host''
basis. You must also make sure that the default logging module, mod_log_common,
isn't compiled in; otherwise, the server gets confused.
Host-Based Access Control
You can control access to the server, or even a subdirectory of the server, based on the host
name, domain, or IP number of the client's machine. You do so by using the directives allow
and deny, which can be used together with order. allow and denycan take multiple hosts:
deny from badguys.com otherbadguys.com
Typically, you want to do one of two things: You want to deny access to your server from
everyone but a few other machines, or you want to grant access to everyone except a few
hosts. You deny access from all but a few machines by using these commands:
order deny,allow
allow from mydomain.com
deny from all
This directive means ``Grant access only to hosts in the domain mydomain.com.'' This domain
could include host1.mydomain.com, ppp.mydomain.com, and the-boss.mydomain.com.
The preceding directive tells the server to evaluate the deny conditions before the allow
conditions when determining whether to grant access. Likewise, you can handle the ``exclude
only a couple of sites'' case described earlier by using the following:
order allow,deny
allow from all
deny from badguys.com
order is needed becauseÐagainÐthe server needs to know which rule to apply first. The
default for order is deny,allow.
In a third argument to order, called mutual-failure, a condition has to pass the allow and
deny rules to succeed. In other words, it has to appear in the allow list, and it must not appear
in the deny list, as in the following example:
order mutual-failure
allow from mydomain.com
deny from the-boss.mydomain.com
776 Chapter 36 Getting Started with Apache
In this example, the-boss.mydomain.com is prevented from accessing this resource, but every
other machine at mydomain.com can access it.
Protecting resources by host name is dangerous. A determined person who controls
the reverse-DNS mapping for his or her IP number can relatively easily spoof any host
name he or she wants. Thus, it's strongly recommended that you use IP numbers to
protect anything sensitive. In the same way, you can simply list the domain name to
refer to any machine in that domain. You also can give fragments of IP numbers, as
shown here:
allow from 204.62.129
This example allows only hosts whose IP numbers match, such as 204.62.129.1 or
204.62.129.130.
Typically, these directives are used within a<Limit> container, and even that within
a <Directory> container, usually in an access.conf configuration file. The
following example is a good template for most protections; it protects the /www/
htdocs/private directory from any host except those in the 204.62.129 IP space:
<Directory /www/htdocs/private>
Options Includes
AllowOverride None
<Limit GET POST>
order allow,deny
deny from all
allow from 204.62.129
</Limit>
</Directory>
Using .htaccess Files
Apache uses special files, known as .htaccess files, for controlling access to directories.
Searching directories for .htaccess .htaccess files work
files is fairly painful. Because
hierarchically, when a request is made for /path/path2/dir1/dir2/foo, Apache looks for an
.htaccess file in every subdirectory. In the example of /path/path2/dir1/dir2/foo , it looks
through at least five subdirectoriesÐa significant disk access load that's best to avoid if
possible.
To solve the problem of too many disk hits, you should put anything controlled via your
.htaccess access.conf configuration file or even the srm.conf file. If you have
files into the
to look for .htaccess files in subdirectories and can narrow it down to a specific subdirectory,
777
Understanding Security Issues
you can have the server look only for .htaccess files in that subdirectory by using
AllowOverride. Part
Suppose that your document root is in /www/htdocs, and you want to turn off the searching
VII
for all .htaccess files except those in /www/htdocs/dir1/dir2 and everywhere below. To do Ch
so, you would put something like the following into your access.conf configuration file: 36
<Directory /www/htdocs>
Options All
AllowOverride None
</Directory>
<Directory /www/htdocs/dir1/dir2>
Options All
AllowOverride All
</Directory>
Listing the directories in that order is important so that the second <Directory> doesn't take
precedence over the first.
Understanding Security Issues
The security of your server is, no doubt, one of your biggest concerns as a Web site
administrator. Running a Web server is, by nature, a security risk. For that matter, so is
plugging your machine into a network at all. However, you can do a lot to make your Web
server more secure from external forces (people trying to break into your site) and internal
forces (your own Web site users mistakenly or willingly opening up holes).
CGI Issues
The biggest cause for concern about protecting your site from external threats is CGI scripts.
Most CGI scripts are shell-based, using Perl or C-shell interpreted programs rather than
compiled programs. Thus, many attacks have occurred by exploiting ``features'' in those
shells. This section doesn't go into too much detail about how to make CGI scripts
themselves safe. As an administrator, you should know a couple of important points, however.
A CGI script runs with the user ID of the server child process. In the default case, this is
nobody. To adequately protect yourself, you might want to consider the nobody user an
untrustworthy user on your site, making sure that this user doesn't have read permission to
files you want to keep private and doesn't have write permission anywhere sensitive. Certain
CGI scriptsÐfor example, a guestbook application that allows users to record comments
about your Web siteÐdemand write access to certain files. So if you want to enable those
types of applications, it's best to specify a directory to which CGI scripts can write without
worrying about a malicious or misdirected script overwriting data that it shouldn't.
Furthermore, as a site administrator, you can limit the use of CGI to specific directories by
using the ScriptAlias directive. Alternatively, if you have turned on .cgi as a file extension
for CGI scripts, you can use the Options ExecCGI directive in access.conf to further control
the use of CGI files.
778 Chapter 36 Getting Started with Apache
Let me give you an example of controlling access with ExecCGI: If you want to allow for CGI
to be used everywhere on the site (with a document root of /home/htdocs) except for the
``users'' subdirectory because you don't trust your users with CGI scripts, your access.conf
should look something like Listing 36.1.
Listing 36.1 A Sample access.conf File Showing Directory Configuration
Information
<Directory /home/htdocs/>
Options Indexes FollowSymLinks Includes Multiviews ExecCGI
AllowOverride None
</Directory>
<Directory /home/htdocs/users/>
Options Indexes SymLinksIfOwnerMatch IncludesNOEXEC Multiviews
AllowOverride None
</Directory>
Because ExecCGI isn't in the Options list for the second directory, no one can use CGI scripts
there.
Unfortunately, there really is no middle ground between allowing CGI scripts and
disallowing them. Now, most languages used for CGI programs don't have security concepts
built into them, so you need to deal with rules such as ``don't touch the hard disk'' or ``don't
send the /etc/passwd file in email to an outside user'' as though you had actual Linux users
who needed the same restrictions applied to them. Maybe this will change when Sun's Java
language gets more use on the server side, or when people use raw interpreted languages less
and higher-level programming tools more often.
Trust Issues with Server-Side Includes
As you can see from Listing 36.1, another change was made between the trusted part of the
server and the untrusted part: the Includes argument to Options was changed to
IncludesNOEXEC. IncludesNOEXEC allows your untrusted users to use server-side includes
without allowing the #include of CGI scripts or the #exec command to be run. The #exec
command is particularly troublesome in an untrusted environment because it basically gives
shell-level access to an HTML author.
Symbolic Links
In an untrusted environment, UNIX symbolic links (which enable linking across file system
boundaries) also are a concern for Web site administrators. Malicious users could very easily
create symbolic links from directories where they have write permission to an object or
resource, even outside the document root, to which all they need is read permission. For
example, a user could create a link to the /etc/passwd file and then release it onto the Web,
exposing your site to potential crack attemptsÐparticularly if your operating system doesn't
use shadow passwords.
779
Understanding Security Issues
www.altavista.com), a
In a recent incident involving the AltaVista search engine (
Part
bin, root, ftp, and so on) turned up
VII
search for words common to password files (
references to actual password files that had, intentionally or not, been left public.
These password files included a few with encrypted passwords, which were easy
Ch
36
enough to break with a few hours of CPU time on most workstations.
To protect against symbolic link security breaches, you have two options as the site
administrator: to allow only symbolic linking if the owner of the link and the owner of the
linked-to resource are the same by using SymLinksIfOwnerMatch, or to disallow symbolic links
altogether by not specifying FollowSymLinks or SymLinksIfOwnerMatch.
Also note that both <Directory> segments in Listing 36.1 included AllowOverride None. Not
allowing symbolic links is the most conservative setting; if you want to allow certain elements
in those directories to be tuned by using .htaccess files, you can specify them by using the
AllowOverride directive. However, stating None is the safest policy.
Publicly Writable Spaces
The last security threat that's specific to Web servers is that of allowing publicly writable
spaces to be served up via HTTP. For example, many sites allow their FTP ``incoming''
directory to be accessed via the Web directly. This can be a security hole if someone were to
place there a malicious CGI script or a server-side include file that calls #exec to do some
damage. If you decide you need to take the risk of providing public writable spaces, you can
take some steps to protect yourself:
& The most conservative setting you should set for the Options directive is this:
Options Indexes
You could use None, but Indexes really doesn't introduce any additional security
problems, as long as you're comfortable with others being able to download anything
that has been submitted. In the light of recent legislation by the U.S. government
regarding ``indecent'' materials, you might not want to take this risk either.
& Make sure that you set AllowOverride None so that people can't upload an .htaccess
file into your directory and modify all your settings and security policies.
& Make sure that the FTP daemon you're using doesn't allow the execute bit to be set.
By preventing the execute bit to be set, you prevent the execution of uploaded CGI
scripts. If you're using XBitHack to activate your server-side includes, you can prevent
those includes from being run as well. This approach is mainly a backup for setting the
Options, as in Listing 36.1, which should protect you against these threats anyway.
These same rules apply if you have CGI scripts that generate their own uniquely addressable
HTML or CGI files. For example, if the guestbook.cgi program constantly appends the
submitted personal information to a guestbook.html file, all the same rules apply; the
contents of that HTML file must be considered unsafe. This possible security breach can be
780 Chapter 36 Getting Started with Apache
plugged if the CGI script double-checks what's getting written and removes ``dangerous'' code,
such as server-side includes.
Project: Adding Customized Error Messages
Apache can give customized responses in the event of an error. You can control these
responses by using the ErrorDocument directive in the configuration files. The syntax is as
follows:
ErrorDocument HTTP_response_codeaction
HTTP_response_code is the event that triggers the action. The action can be any of the
following
& A local URI to which the server is internally redirected
& An external URL to which the client is redirected
& A text string that starts with a " character and where the %s variable contains any extra
information, if available
The following is an example:
ErrorDocument 500 ``Ack! We have a problem here: %s.
ErrorDocument 500 /errors/500.cgi
ErrorDocument 500 http://backup.myhost.com/
ErrorDocument 401 /subscribe.html
ErrorDocument 404 /debug/record-broken-links.cgi
Two extra CGI variables are passed to any redirected resource: REDIRECT_URL contains the
original URL requested, and REDIRECT_STATUS gives the original status that caused the
redirection. This information will help the script if its job is to try to figure out what caused
the error response.
CHAPTER
37
Configuring FTP Servers
In this chapter by Steve Burnett
Installing WU-FTP 782
Anonymous FTP 782
Configuring WU-FTP 789
Project: Setting Up an Anonymous FTP Server 790
782 Chapter 37 Configuring FTP Servers
Installing WU-FTP
File Transfer Protocol (FTP) is a simple and effective means of transferring files between
computers that are connected on a TCP/IP network. FTP allows users to transfer both
ASCII and binary files and is documented in RFC 959 and RFC 1759.
During an FTP session, you connect to another computer by using the FTP client program.
From this point, you can move up and down through the directory tree, list directory
contents, copy files from the remote computer to your computer, and transfer files from your
computer to the remote system. Normal file protections apply; you can't get or put a file on
the remote system if you don't have the proper permissions for that file.
wu-ftpd is the common name for wuarchive-ftpd. Developed by Bryan D. O'Connor, wu-ftpd
is a replacement FTP service. Its name is derived from its origin: wu-ftpd was originally
developed at Washington University ( *.wustl.edu). Immensely popular, wu-ftpd is used on
many FTP sites across the globe. This chapter will present how to install and configure wu-
ftpd.
wu-ftpd is included with Red Hat Linux, so you should be able to select it during installation.
You should know that many Linux distributions make a service operable when it is installed as
part of the system setup. You should review the chapter on system security and disable
services you do not want until you have time to configure those services. The following are
good sources of information for WU-FTP:
& http://www.landfield.com/wu-ftpd/ÐA resource center
& http://www.cetis.hvu.nl/~koos/wu-ftpd-faq.htmlÐThe WU-FTP Frequently
Asked Questions file
& http://sunsite.doc.ic.ac.uk/sun/sunsite-sun-info/sun-faq/FAQs/
SettingUpSecureFTP.faqÐSome advice for securing an FTP site
You can always find the latest version of wu-ftpd 9 with the latest security patches, bug fixes,
and so on at ftp.wu-ftpd.org in ftp://ftp.wu-ftpd.org/pub/wu-ftpd/ (please use FTP to
access this directory). You can access this system via anonymous FTP.
Anonymous FTP
Many organizations have made huge repositories of information available via FTP. These
FTP sites hold everything from text files to software of every conceivable type available. But
how do you access this enormous storehouse of data if you don't have an account on the
remote computer? Do you need to get an account on every FTP site to be able to access these
files? Not really.
A common convention on the Internet allows guest FTP access to file repositories so that
users can transfer files. This guest access is called anonymous FTP. To use anonymous FTP,
you start an FTP session to the remote system and use anonymous as your username and your
783
Anonymous FTP
email address as the password. For example, in the following sample, the user named smith on
linux.somewhere.com wants to initiate an FTP session with a common FTP site:
$ ftp ftp.uu.net
ftp.uu.net (login:smith): anonymous
Password: [email protected]
Part
Many sites don't allow anonymous FTP. Allowing guest users to connect to your
computer does involve some risk. In cases in which anonymous FTP isn't allowed, the
VII
ftp command fails with a message similar to Login failed - User ``anonymous'' Ch
unknown. Sites that do permit anonymous FTP typically place the users in a restricted
directory tree with read-only access. If you're allowed to place files on the remote
37
computer, you usually can put them in only one directory.
Also, several of the Web browsers (Netscape Navigator, for example) support the FTP
protocol as well as the HTTP protocol. Because several Web browsers also support
email, they can support anonymous FTP connections automatically. Check the user
guide for your Web browser for more information.
To begin an anonymous FTP session, follow these steps:
1. Open a terminal window and enter the following at the command prompt:
[burnett@ns burnett]$ ftp
The terminal prompt changes to the following, indicating that the FTP program is
running:
ftp>
2. Enter the following command:
ftp> open ftp.wu-ftpd.org
Connected to ftp.wu-ftpd.org.
220 ftp.wu-ftpd.org FTP server ready.
Name (ftp.wu-ftpd.org:burnett):
3. Enter the name anonymous. The system responds with the following
331 Guest login ok, send your complete e-mail address as password.
Password:
4. Enter your complete email address as the password. The FTP server uses this
password to track how many unique users are logging into the server over a given time
period.
If you don't want to give your email address, and instead you simply type a random
string of garbage, such as rgjlrlk, the system might reject your anonymous login.
Some servers are configured to test the anonymous passwords for proper email
construction, so they reject passwords that are not formed like email addresses.
784 Chapter 37 Configuring FTP Servers
After you've entered your email address as your password, the server responds with the
following message:
230-Welcome to the FTP server for the WU-FTPD Development Group
230-
230-This server is the primary distribution site for the WU-FTPD daemon.
230-
230-The pub directory contains the distribution and supporting files.
230-
230-If you are uploading contributions, please place them in the incoming
230-directory and email [email protected] announcing your upload.
230-
230 Guest login ok, access restrictions apply.
Remote system type is UNIX.
Using binary mode to transfer files.
ftp>
5. Enter pwd, as shown here, and press Enter to see the current directory:
ftp> pwd
257 ``/'' is current directory.
6. Enter ls, as shown here, to list the files in the current directory:
ftp> ls
200 PORT command successful.
150 Opening ASCII mode data connection for directory listing.
total 16
d--x--x--x 2 root root 1024 Jul 19 12:19 bin
d--x--x--x 2 root root 1024 Jan 12 1999 dev
d--x--x--x 2 root root 1024 May 25 10:30 etc
drwxrwx-wx 2 root wuftpd 1024 Jul 19 20:15 incoming
drwxrwx--x 3 root wuftpd 2048 Jul 28 19:46 private
drwxrwxr-x 6 root wuftpd 2048 Jul 9 07:43 pub
226 Transfer complete.
ftp>
7. Use cd to change to the /pub directory, as shown here, and you see the following
response:
ftp> cd pub
250-This directory contains the following:
250-
250-ANNOUNCE-WUFTPD
250- The original email announcing the formation of the WU-FTPD Development
250- Group.
250-
250-BeroFTPD
250- Directory containing the current versions of BeroFTPD. The Development
250- Group has determined a primary goal for future releases is to merge many
250- of the features of BeroFTPD back into the base release of the daemon.
250- Until that work is completed, the most-current version of BeroFTPD will
250- remain available here.
250-
250-README-MIRRORS
250- The original email announcing the currently-known mirrors of the WU-FTPD
250- Development Group's FTP site.
250-
250-README
785
Anonymous FTP
250- This file.
250-
250-support
250- Directory containing many of the support packages either needed to
250- create a working FTP server, or to assist in the operation of existing
250- sites. The WU-FTPD Development Group has stated the goal of including
250- many of these support facilities into future releases of the daemon.
250- Until that work is completed, these supporting packages will remain
250- available here.
250- Part
250-wu-ftpd
250- Directory containing the base distribution of the WU-FTPD daemon and VII
250- supporting documentation. Ch
250-
250-
250-Other sources of information
37
250-
250-wu-ftpd Resource Center: http://www.landfield.com/wu-ftpd/
250-wu-ftpd FAQ: http://www.cetis.hvu.nl/~koos/wu-ftpd-faq.html
250-wu-ftpd list archive: http://www.landfield.com/wu-ftpd/mail-archive/
250-
250-Please read the file README
250- it was last modified on Fri May 21 23:30:39 1999 - 70 days ago
250-Please read the file README-MIRRORS
250- it was last modified on Thu Jul 22 02:36:57 1999 - 8 days ago
250 CWD command successful.
ftp>
Notice that the README file in the /pub directory is piped to the terminal when you
change directories to that directory.
8. Use cd to change to the wu-ftpd directory, as follows:
ftp> cd wu-ftpd
250-This directory contains the base distribution of the WU-FTPD daemon and
250-supporting documentation.
250-
250-ANNOUNCE-RELEASE
250- The original email announcing the public release of the current version
250- of the daemon.
250-
250-FIXES
250- Directory containing the FIXES-* files documenting the changes the daemon
250- has gone through since the release of version 2 by the Washington
250- University at St. Louis. These may be incomplete, inaccurate or down-
250- right misleading, but they're all we have for now.
250-
250- This directory is available for on-the-fly tar and compression. If you
250- want to read all the FIXES, please retrieve FIXES.tar.gz or FIXES.tar.Z
250- rather than each individual file.
250-
250-README
250- This file.
250-
250-attic
250- Directory containing historical versions of the daemon, patches or other
250- historical information. You don't want to run anything you find here;
250- it's only for research purposes.
786 Chapter 37 Configuring FTP Servers
250-
250-binaries
250- Directory containing by-machine pre-compiled binary distributions of the
250- daemon. These are kept as current as possible. Browse arround to see
250- if there's a binary for your system; it could save you a lot of time
250- and trouble.
250-
250-examples
250- Directory containing an example of a fully-functional, minimal-install
250- FTP site. Many of your questions can be answered by the examples you'll
250- see in here.
250-
250-quickfixes
250- Directory containing by-version patches released to fix problems between
250- releases.
250-
250-rfc0959.txt
250- The definition of the current FTP protocol implemented by the WU-FTPD
250- daemon. Not for the faint-of-heart.
250-
250-telnet.testing.HOWTO
250- A short document explaining how to test the operation of your daemon
250- using the telnet command. Often times problems are masked by the
250- operation of your ftp client (command); this procedure allows you to see
250- the inner-workings of the protocol.
250-
250-unsupported
250- Patches to the daemon which are too old or too crufty to be included in
250- the base release of the daemon. These are here because the WU-FTPD
250- Development Group likes the ideas and may, at some time, include the
250- features in a future version of the daemon.
250-
250-upload.configuration.HOWTO
250- A longish document describing how to safely configure your FTP site to
250- allow remote users to upload files. This is one of the thorniest issues
250- for running a secure site; the document attempts to make the
250- considerations more clear. This document is included in the base release.
250- The version here may be more up-to-date that what you received, so check
250- the date.
250-
250-wu-ftpd-current.tar.Z
250-wu-ftpd-current.tar.gz
250- The current public-release version of the daemon. These are symbolic
250- links to the actual files. Both GNU Zip (gz) and UNIX compress (Z)
250- versions are available. Download the GNU format unless your system only
250- supports UNIX compress.
250-
250-wu-ftpd-faq.txt
250- Koos van den Hout's WU-FTPD Frequently Asked Questions (FAQ) .. if you
250- have a problem, the answer is almost always here .. honest. Please read
250- this document completely before posting your questions to the support
250- mailing lists. This copy may be out-of-date; check the URL given inside
250- for the most-current version.
250-
250-Please read the file README
250- it was last modified on Wed May 26 09:01:36 1999 - 65 days ago
250-Please read the file README-MIRRORS
787
Anonymous FTP
250- it was last modified on Thu Jul 22 02:36:57 1999 - 8 days ago
250 CWD command successful.
ftp>
9. Use cd to change to the binaries/intel/linux/redhat directory:
ftp> cd binaries/intel/linux/redhat
250-Offical Redhat RPMs
250------------------------------------------
250-
250- Binaries Part
250- --------------
250- RH 4.2: wu-ftpd-2.5.0-0.4.2.i386.rpm
VII
250- RH 5.2: wu-ftpd-2.5.0-0.5.2.i386.rpm Ch
250- RH 6.0: wu-ftpd-2.5.0-2.i386.rpm
250- 37
250- Source
250- --------------
250- RH 4.2: wu-ftpd-2.5.0-0.4.2.src.rpm
250- RH 5.2: wu-ftpd-2.5.0-0.5.2.src.rpm
250- RH 6.0: wu-ftpd-2.5.0-2.src.rpm
250-
250-WU-FTPD Development Group
250------------------------------------------
250-The source RPMs should be usable on any system supporting RPM.
250-
250-This first group uses the default Redhat SPEC file and is built using all
250-defaults.
250-
250- For Redhat 4.2 or later
250- ------------------------------------------
250- wu-ftpd-2.5.0-1.RH4-2.i386.rpm
250- wu-ftpd-2.5.0-1.RH4-2.src.rpm
250-
250- For Redhat 5.1 or later
250- ------------------------------------------
250- wu-ftpd-2.5.0-1.RH5-1.i386.rpm
250- wu-ftpd-2.5.0-1.RH5-1.src.rpm
250-
250- For Redhat 6.0 or later
250- ------------------------------------------
250- wu-ftpd-2.5.0-1.RH6-0.i386.rpm
250- wu-ftpd-2.5.0-1.RH6-0.src.rpm
250-
250-The second group uses experimental SPEC files developed by a member of the
250-WU-FTPD Development Group. Installation should be quite a bit easier as
250-more conditions are checked for. The binaries are built for Redhat 4.2
250-using statically linked binaries so they should work on all Redhat systems.
250-
250- For classical operation (using inetd)
250- ------------------------------------------
250- wu-ftpd-2.5.0-inet-1.i386.rpm
250- wu-ftpd-2.5.0-inet-1.src.rpm
250-
250- For operation as a standalone daemon
250- ------------------------------------------
250- wu-ftpd-2.5.0-standalone-1.i386.rpm
250- wu-ftpd-2.5.0-standalone-1.src.rpm
788 Chapter 37 Configuring FTP Servers
250-
250-Please read the file README
250- it was last modified on Fri Jun 11 10:13:26 1999 - 49 days ago
250 CWD command successful.
ftp>
Ls the current directory:
ftp> ls
200 PORT command successful.
150 Opening ASCII mode data connection for directory listing.
total 8212
-r--r--r-- 1 wuftpd wuftpd 1500 Jun 11 10:13 README
-r--r--r-- 1 wuftpd wuftpd 163730 Jun 10 16:41 wu-ftpd-2.5.0-0.4.2.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 278405 Jun 10 16:42 wu-ftpd-2.5.0-0.4.2.src.rpm
-r--r--r-- 1 wuftpd wuftpd 164117 Jun 10 16:42 wu-ftpd-2.5.0-0.5.2.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 278099 Jun 10 16:43 wu-ftpd-2.5.0-0.5.2.src.rpm
-r--r--r-- 1 wuftpd wuftpd 161701 May 21 23:39 wu-ftpd-2.5.0-1.RH4-2.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 270140 May 21 23:39 wu-ftpd-2.5.0-1.RH4-2.src.rpm
-r--r--r-- 1 wuftpd wuftpd 161940 May 21 23:39 wu-ftpd-2.5.0-1.RH5-1.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 269939 May 21 23:39 wu-ftpd-2.5.0-1.RH5-1.src.rpm
-r--r--r-- 1 wuftpd wuftpd 165732 May 21 23:39 wu-ftpd-2.5.0-1.RH6-0.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 270340 May 21 23:39 wu-ftpd-2.5.0-1.RH6-0.src.rpm
-r--r--r-- 1 wuftpd wuftpd 168048 Jun 10 16:44 wu-ftpd-2.5.0-2.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 278093 Jun 10 16:46 wu-ftpd-2.5.0-2.src.rpm
-r--r--r-- 1 wuftpd wuftpd 493153 May 25 14:25 wu-ftpd-2.5.0-inet-1.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 272183 May 25 14:25 wu-ftpd-2.5.0-inet-1.src.rpm
-r--r--r-- 1 wuftpd wuftpd 498847 May 25 14:25 wu-ftpd-2.5.0-standalone-1.i386.rpm
-r--r--r-- 1 wuftpd wuftpd 273546 May 25 14:25 wu-ftpd-2.5.0-standalone-1.src.rpm
226 Transfer complete.
ftp>
Assuming you're installing for Red Hat Linux 6.0 on an Intel processor, three options
are present in this directory as of this writing:
. wu-ftpd-2.5.0-2.i386.rpmÐBuilt using the default Red Hat SPEC file.
. wu-ftpd-2.5.0-inet-1.i386.rpmÐBuilt using experimental and enhanced
SPEC files. The binaries are built for Red Hat 4.2 using statically linked
binaries, so they should work on all Red Hat systems. This file was built for
classical operation, using the inetd service.
. wu-ftpd-2.5.0-standalone-1.i386.rpmÐBuilt using experimental and en-
hanced SPEC files. Installation should be quite a bit easier as more conditions
are checked for. The binaries are built for Red Hat 4.2 using statically linked
binaries, so they should work on all Red Hat systems. This file was built for
operation as a standalone daemon.
In this case, you should choose the first file.
10. Enter the get command for the file you want to retrieve, as shown here:
789
Configuring wu-ftpd
ftp> get wu-ftpd-2.5.0-1.RH6-0.src.rpm
local: wu-ftpd-2.5.0-1.RH6-0.src.rpm remote: wu-ftpd-2.5.0-1.RH6-0.src.rpm
200 PORT command successful.
150 Opening BINARY mode data connection for wu-ftpd-2.5.0-1.RH6-0.src.rpm (270340
bytes).
226 Transfer complete.
270340 bytes received in 157 secs (1.7 Kbytes/sec)
ftp>
11. Enter quit, as shown here, to quit FTP and return to your terminal window. The Part
remote FTP server gives a summary report of your activity in this session:
ftp> quit
VII
221-You have transferred 270340 bytes in 1 files. Ch
221-Total traffic for this session was 276748 bytes in 1 transfers.
221-Thank you for using the FTP service on ftp.wu-ftpd.org. 37
221 Goodbye.
You have new mail in /var/spool/mail/burnett
[burnett@ns burnett]$
12. Do an extended ls to ensure the file is present:
[burnett@ns burnett]$ ls -al wu-ftpd-2.5.0-1.RH6-0.src.rpm
-rw-rw-r-- 1 burnett burnett 270340 Jul 30 21:22 wu-ftpd-2.5.0-1.
RH6-0.src.rpm
You can then use the Red Hat Package Manager (RPM) to install the WU-FTP
package. Use of the Red Hat Package Management system is presented in Chapter 7,
``Upgrading and Installing Software."
Configuring wu-ftpd
Before you start configuring wu-ftpd, you should read the man page for it. Carefully review
the possible startup options, such as -a and -L. Table 37.1 shows some options for invoking
wu-ftpd.
Table 37.1 Examples of wu-ftpd Command-Line Options
Field Description
-l Provides additional logging information
-i Logs uploads
-o Logs downloads
To start the configuration, you need to find the existing FTP entry in the /etc/inetd.conf
file and edit it to refer to wu-ftpd, as shown in the following example:
ftp stream tcp nowait root /usr/sbin/ftpd ftpd -lio
If you want to install wu-ftpd on a nonstandard port, you need to edit both the /etc/
services and /etc/inetd.conf files. This technique can be useful for debugging wu-ftpd
without interrupting the existing FTP service.
790 Chapter 37 Configuring FTP Servers
Next, add a line to /etc/services indicating the service name, port number, and protocol, as
shown here:
wuftp 3010/tcp # command port to test wu-ftpd
You also might want to add a second line like this to /etc/services:
wuftp-data 3009/tcp # data port to test wu-ftpd
Now you can add or edit the appropriate FTP line to the /etc/inetd.conf file, the key part
being to reference the same service name as is in the /etc/services file:
wuftp stream tcp nowait root /usr/sbin/ftpd ftpd -lio
Your next step is to make sure that the name of the service wu-ftpd matches in both the /etc/
inetd.conf and the /etc/services files.
Finally, you can run bin/ckconfig from the distribution to verify your setup at this point and
then try logging in to your FTP server as a real user.
Project: Setting Up an Anonymous FTP Server
If you want to set up an anonymous FTP server, you should use only the most recent version
of the preferred FTP service. Older versions may contain exploits that can be used by the
unscrupulous to gain access to your server or its contents.
Setting Permissions
When you're ready to set permissions for your anonymous FTP server, the permissions
system of Linux can be helpful. For example, the anonymous FTP root directory ( ~ftp) and
its subdirectories should not be owned by the FTP account or be in the same group as the
FTP account. If any of these directories are owned by FTP or are in the same group as the
FTP account and are not write-protected, an intruder can add files or modify other files. One
way to secure the FTP service is to make the directories owned by the root account; after all,
if the intruder has root access, he or she can do anything. Making the FTP root directory and
its subdirectories owned by root, part of the system group, and protected so that only root has
write permission helps to secure the anonymous FTP service.
The following is an example of an anonymous FTP directory setup:
drwxr-xr-x 7 root system 512 Apr 7 12:33 ./
drwxr-xr-x 25 root system 512 Jan 18 10:40 ../
drwxr-xr-x 2 root system 512 Mar 21 11:46 bin/
drwxr-xr-x 2 root system 512 May 23 18:49 etc/
drwxr-xr-x 10 root system 512 Oct 16 11:34 pub/
All files and libraries involved, specifically those used by the FTP daemon as well as those in
~ftp/bin and ~ftp/etc, should have the same protections as these directories.
791
Project: Setting Up an Anonymous FTP Server
Password and Group Files
When you're setting up your anonymous FTP server, do not use the system's /etc/passwd file
as the password file or use the system's /etc/group as the group file in the ~ftp/etc directory.
Placing these system files in the ~ftp/etc directory allows intruders access to the files in
question.
A more safe method is to use a dummy version of both the ~ftp/etc/passwd and ~ftp/etc/
group files. These files should be owned by root. The dir command uses these dummy
Part
versions to show owner and group names of the files and directories instead of displaying VII
arbitrary numbers. The following are some other tips to consider:
Ch
& The ~/ftp/etc/passwd file should contain no account names that are the same as 37
those in the system's /etc/passwd file. These files should include only those entries
that are relevant to the FTP hierarchy or needed to show owner and group names.
& Make sure that the password field has been cleared. The following examples show the
*
use of asterisks ( ) to clear the password field.
The following is an example of entries in a ~ftp/etc/passwd file from an anonymous
FTP area on afakecompany.com:
toys:*:3822:20:Software Toys::
ftp:*:3823:90:Anonymous FTP::
doc:*:3824:90:User Documentation::
The following is an example group file from the anonymous FTP area on cert.org:
toys:*:20: ftp:*:90:
CHAPTER
38
Configuring Domain Name Service
(DNS)
In this chapter by Steve Burnett
How The Net Began 794
Introducing DNS 794
Configuring the Resolver 795
Using the named Daemon to Set Up the Server 798
794 Chapter 38 Configuring Domain Name Service (DNS)
How The Net Began
When the Internet was first formed, the number of hosts on the Net was very small.
Maintaining the name/address mapping was fairly easy: Each host simply kept a complete list
of all host names and addresses in a local file. As the growth of the Internet accelerated, this
system quickly became unwieldy. When a new host was added, it was necessary to update
every host file on every computer. Also, because each new computer resulted in a new line in
every host file, the size of the host files began to grow to quite a large size. Clearly, a new
solution was needed.
Mapping Internet system names to IP addresses is a task that requires a good degree of
consideration. With the explosive growth of the Internet over the past few years, the original
system of maintaining host name to IP address mappings in a local flat ASCII file quickly
proved impractical. With thousands of computers on the Net and more being added daily, a
new system was needed. That new system was a network-wide distributed database known as
BIND, the Berkeley Internet Domain server. Also referred to variously as the Domain Name
Service, the Domain Name System, or DNS, this system provides an effective, relatively
transparent host name to the IP address mapping mechanism.
DNS is notoriously hard to configure, but when you're successful, maintaining it is fairly
easy.
Introducing DNS
DNS provides a mechanism for converting IP addresses into mnemonic names that represent
\hosts, networks, and mail aliases. It does so by dividing the entire Internet IP and name
space into different logical groups. Each group has authority for its own computers and other
information.
Because DNS is a complicated topic, it has its own specialized set of terms. Table 38.1 lists
the definitions of some commonly used DNS terms.
Table 38.1 Commonly Used DNS Terms
Term Definition
Domain The logical entity or organization that represents a part of a network.
For example, afakecompany.com is the name of the primary domain
for the fictitious business A Fake Company.
Domain name The name portion of a host name that represents the domain that
contains the host. For example, in the address quakeserver.afake-
company.com, the domain name is afakecompany.com. This term is
also used interchangeably with domain.
Host A computer on a network.
Node A computer on a network.
795
Configuring the Resolver
Table 38.1 Commonly Used DNS Terms
Term Definition
Name server A computer that provides DNS services to map DNS names to IP
addresses.
Resolve The act of translating a DNS name into its corresponding IP address.
Resolver A program or library routine that extracts DNS information from a
name server.
Reverse resolution The act of matching a given IP address to its DNS name. This
process is also called reverse DNS.
Spoof The act of appearing to the network as having a different IP address
or domain name.
Part
VII
DNS can be conceptually divided into the following three parts:
& Domain name space ÐThis part is a specification for a tree structure that identifies a
Ch
38
set of hosts and provides information about them. Conceptually, each node in the tree
has a database of information about the hosts under its authority. Queries attempt to
extract the appropriate information from this database. In simple terms, the domain
name space is just the listing of all different types of information, names, IP addresses,
mail aliases, and such that are available for lookup in the DNS system.
& Name servers ÐThese programs hold and maintain the data located in the domain
name space. Each name server has complete information about a subset of the domain
name space and cached information about other portions.
A name server has complete information for its area of authority. This authoritative
information is divided into areas known as zones, which can be divided among different
name servers to provide redundant service for a zone. Each name server knows about
other name servers that are responsible for different zones. If a request comes in for
information from the zone that a given name server is responsible for, the name server
simply returns the information. However, if a request comes in for information from a
different zone, the name server contacts the appropriate server with authority for that
zone.
& Resolvers ÐThese programs or library routines extract information from the name
servers in response to a query about a host in the domain name space.
Configuring the Resolver
The first step in using DNS is to configure the resolver library on your computer. You must
configure your local resolver if you intend to use DNS name resolution, even if you're not
going to run a local domain name server.
796 Chapter 38 Configuring Domain Name Service (DNS)
The /etc/host.conf File
The local resolver libraries are configured via a file named host.conf that is located in the
/etc directory. This file tells the resolver what services to use and in what order. This file is a
plain ASCII file that lists resolver options, one per line. Fields can be separated by either
spaces or tabs. The # character indicates the start of a comment.
You can specify several options in the host.conf file, as shown in Table 38.2.
Table 38.2 Configuration Options for the /etc/host.conf File
Option Description
order Specifies in what order different name resolution mechanisms are tried. The
specified resolving services are tried in the order listed. The following name
resolution mechanisms are supported: hosts (attempts to resolve the name by
looking in the local /etc/host file), bind (queries a DNS name server to resolve
the name), and nis (uses the Network Information Service [NIS] protocol to try
to resolve the host name).
alert Takes off or on as arguments. If this option is turned on, any attempt to spoof an
IP address is logged via the syslog facility.
nospoof If reverse resolution is used to match a host name to a specified address,
nospoofresolves the host name that's returned to verify that it does match the
address that you queried. This option prevents spoofing of IP addresses. You
enable it by specifying nospoof on. Caution: Using this option can cause a
noticeably additional load on the server.
trim Takes a domain name as an argument. trim removes the domain name before
performing an /etc/hosts lookup on the name. This way, you can put just the
base host name in /etc/hosts without specifying the domain name.
multi Takes off or on as arguments. It is used only with host queries to determine
whether a host is allowed to have more than one IP address specified in
/etc/hosts. This option has no effect on NIS or DNS queries.
The following is an example of an /etc/host.conf configuration file that uses these options:
# Sample /etc/host.conf file
#
# Lookup names via DNS first then fall back to /etc/hosts
order bind hosts
# We don't have machines with multiple addresses
multi off
# check for IP address spoofing
nospoof on
# and warn us if someone attempts to spoof
alert on
# Trim the afakecompany.com domain name for host lookups
trim afakecompany.com
This example shows a general resolver configuration for the domain afakecompany.com. The
resolver looks up the host names by using DNS first and then tries the local /etc/hosts file.
797
Configuring the Resolver
Specifying the local /etc/hosts file in the resolution search is a good idea. If, for
some reason, your name servers should be unavailable, you can still resolve the
names for hosts listed in your local hosts file. You should also keep a list of all your
local hosts in your /etc/hosts files on each of your local computers.
Multiple IP addresses for a single machine are disabled. This host checks for IP address
spoofing by re-resolving the host name that a reverse IP address lookup returns. This
procedure is a bit of a performance hit, but it helps make sure that no host is pretending to be
a different host than it really is. Also, you've set up the resolver to warn you if an attempt to
spoof is detected. Finally, the resolver trims the domain afakecompany.com from any host
names that are looked up in the local /etc/hosts file.
Part
The /etc/resolv.conf File
VII
Now that you've configured the basic behavior of the resolver library, you need to set up
Ch
some information for the DNS portion of the resolver. You need to do so only if you're using
DNS for host name resolutionÐthat is, by specifying bind in the order statement of the
38
/etc/host.conf file. But then you wouldn't be reading this chapter if you weren't going to
use DNS, would you?
The /etc/resolv.conf file controls the way the resolver uses DNS to resolve host names. It
specifies the DNS name servers to contact when resolving a host name and in what order to
contact them. It also provides the local domain name and some clues as to how to guess at the
domain name of hosts that are specified without a domain name.
Table 38.3 lists the valid options for the /etc/resolv.conf file.
Table 38.3 Configuration Options for the /etc/resolv.conf File
Option Description
domain Specifies the local domain name of this host. If it's not given, the resolver tries
to get the local domain name from the getdomainname() system call.
nameserver Specifies the IP address of a DNS name server to contact for name resolution.
You can list up to three name servers by using the nameserver option multiple
times. The name servers are tried in the order listed. You should put your
most reliable name server first so that queries don't time out on a server that's
likely to be down.
search Lists domains to try if no domain name is specified as part of a query host
name. If no search option is given, the list of domains is created by using the
local domain plus each parent domain of the local domain.
The following is a sample /etc/resolv.conf file for afakecompany.com:
# /etc/resolv.conf for afakecompany.com
#
798 Chapter 38 Configuring Domain Name Service (DNS)
# Set our local domain name
domain afakecompany.com
# Specify our primary name server
nameserver 166.82.1.3
In this example, you specify the local domain via the domain option and list one name server
to use for resolving host names.
You need to specify the IP address of the DNS name server as an argument to the
nameserver optionÐnot the host name. If you specify the host name, DNS doesn't
know which host to contact to look up the host name of the name server.
The preceding example doesn't use the search option to specify the search order. This means
that if you try to query the address of a machineÐfor example, skippyÐthe resolver tries to
look up skippy first. If this lookup fails, the resolver looks up skippy.afakecompany.com and
then skippy.com.
DNS servers can and do go down unexpectedly. If you rely solely on a DNS server for name
resolution, you might find yourself unable to work if it crashes. Make sure that you specify
multiple servers and keep a good list of hosts in your local /etc/hosts file, just in case.
Using the named Daemon to Set Up the Server
Now the real magic starts. You've seen how to set up the basics of resolver configuration and
how to tell your resolver which name servers to contact. In the following sections, you'll learn
the mechanics of setting up a name server.
The DNS name server under Linux is provided by the named (pronounced name-de) daemon.
This daemon is typically started at boot time and reads its configuration information from
a set of configuration files. named typically runs until the machine is shut down. After named
starts and is initialized with its configuration information, it writes its process ID to the
/etc/named.pid ASCII file. It then starts listening for DNS requests on the default network
port specified in /etc/services.
The named.boot File
named reads when it starts is typically /etc/named.boot. This very small file
The first file that
named; it contains pointers to the various
is the key to all the other configuration files used by
configuration files and to other name servers. In the named.boot file, comments start with a
semicolon and continue to the end of the line. Several options can be listed in the named.boot
file; Table 38.4 lists these options.
799
Using the named Daemon to Set Up the Server
Table 38.4 Configuration Options for the named.boot File
Option Description
directory Specifies the directory where the DNS zone files are located. You can specify
several different directories by using the directory option repeatedly. You
can give file pathnames as being relative to these directories.
primary Takes a domain name and filename as arguments. The primary option
declares named to be authoritative for the specified domain and causes named to
load the zone information from the specified file.
secondary Tells named to act as a secondary server for the specified domain. It takes a
domain name, a list of addresses, and a filename as arguments. named tries to
transfer the zone information from the hosts specified in the address list and
then stores the zone information in the file specified on the option line. If
named can't contact any of the hosts, it tries to retrieve the information from
Part
VII
the secondary zone file.
cache Sets up caching information for named. The cache option takes a domain name
and a filename as arguments. The domain name is typically specified as .
Ch
(dot). The file contains a set of records, known as server hints, which list
information about the root name servers. 38
forwarders Takes a list of name servers as arguments. This option tells the local name
server to try to contact the servers in this list if it can't resolve an address from
its local information.
slave Turns the local name server into a slave server. If the slave option is given,
the local server tries to resolve DNS names via recursive queries. It simply
forwards the request to one of the servers listed in the forwarders option line.
In addition to the options listed here, a few additional options aren't commonly used. Refer to
the named man page for more information on these options.
Because afakecompany.com isn't attached to the Internet, many of the IP host and
network addresses in these examples are fake. When you're setting up your own
name server, make sure that you use the correct addresses assigned to you.
The following is a sample named.boot file:
; named.boot file
; A sample named.boot for afakecompany.com
;
directory /var/named
;
cache . named.ca
primary afakecompany.com named.hosts
primary 197.198.199.in-addr.arpa named.rev
This example sets up the primary name server for afakecompany.com. As you can see,
comments start with the ; character. The directory statement in the file tells named that all
800 Chapter 38 Configuring Domain Name Service (DNS)
its working files are located in the /var/named directory. Because none of the other files listed in
the named.boot file have directory paths associated with them, they're located in /var/named.
Tip #188 from
Steve /var/named is the default directory. The name server can be installed anywhere, and
the configuration files may be located anywhere, as long as you define the directory
path in named.boot.
The next line sets up the caching information for this name server. This option should be
present on almost every machine running as a name server. It tells named to enable caching
and load the root server information from the file named.ca.
Tip #189 from
Steve The cache entry is very important. Without it, no caching is enabled on the local
name server, which can cause severe performance problems for name lookups. Also,
the local server can't contact any root name servers and, as a result, can't resolve any
non-local host names unless it's set up as a forwarding name server.
The next line in the named.boot file tells named that this server has primary authority for
the domain afakecompany.com. The zone and host information records are in the file
named.hosts. You'll learn about these zone authority records in detail in the following
section.
A second primary named.hosts file shows that you also have primary zone
line in the
authority for the zone 197.198.199.in-addr.arpa with zone information in the named.rev
file. This strange syntax is named's way of getting information to match IP addresses to DNS
names. Because DNS was originally set up to match DNS names to IP addresses, a different
primary line is needed to do reverse resolution.
The in-addr.arpa domain is used to specify reverse, or IP address, to DNS name
resolution.
Database Files and Resource Records
All information in the various named database files is stored in a format known as a resource
record. Each resource record has a type associated with it, which tells the record's function. A
resource record is the smallest piece of information that named uses.
801
Using the named Daemon to Set Up the Server
Most people find the syntax for resource records and master database files in general to be a
bit arcane and obscure. It doesn't help matters that some resource records have to appear in
certain places in certain files. Most DNS configuration problems can be traced to errors in
these master configuration files. All this said, it's time to dive in and look at the resource
record syntax and the various master files.
Within the master configuration files, you have the option of specifying absolute host
names or host names relative to this domain. Host names are considered absolute if
they end in a dot character ( .), as in foo., afakecompany., or com.. Host names
that don't end with a dot are considered relative to the local domain, also known as
the origin. You can refer to the origin itself by using the @ character.
Part
Resource records use a general syntax that's consistent across all types of resource records. To VII
add to the confusion, however, several parts of the record are optional depending on the
Ch
record type and may assume a default value if not specified. The basic format of a resource
record is the following:
38
[owner] [ttl] [class] typedata
Fields are separated by whitespace such as spaces or tabs. Table 38.5 describes what the
various fields mean.
Table 38.5 Fields in the Resource Record Data Format
Field Description
owner Identifies the domain or host name that the record applies to. If no name is given,
the domain name of the previous resource record is assumed.
ttl Identifies the time-to-live field, which tells how long, in seconds, the information
in this record is valid after it's retrieved from a DNS server. If no ttl value is given,
the minimum ttl of the last Start of Authority (SOA) record is used.
class Specifies a networking address class. For TCP/IP networks, you use the value IN. If
the class isn't given, the class of the previous resource record is used.
type Lists the type of the resource record. This value is required. The various resource
record types are listed in the next section.
data Specifies the data associated with this resource record. This value is required. The
format of the data field depends on the content of the type field.
As you can see, the format of a resource record can get quite confusing. Several fields are
optional, and the data field depends on the type of the resource record. To make matters
worse, resource records come in several different types. Table 38.6 lists the most common
resource record types; a few additional types are rarely used. If you're interested in the
additional types, refer to the appropriate RFCs and the man pages for named.
802 Chapter 38 Configuring Domain Name Service (DNS)
Table 38.6 Commonly Used Resource Record Types
Type Description
A Associates a host name with an address. The data field holds the address in dotted
decimal format. Any given host can have only one A record, which is an address
record, because this record is considered authoritative information. Any additional
host name or address mappings for this host must be given by using the CNAME type.
CNAME Associates an alias for a host with its canonical name, the name specified in the A
record for this host.
HINFO Provides information about a host. The data field holds the hardware and software
information for a particular host. It's just a free-format text string, so you can put in
whatever makes sense for your hardware.
MX Sets up a mail exchanger record. The data field holds an integer preference value
followed by a host name. MX records tell a mail transport to send mail to another
system that knows how to deliver it to its final destination.
NS Points to a name server for another zone. The data field of the NS resource record
contains the DNS name of the name server. You need to specify an A record as well
to match the host name with the address of the name server.
PTR Maps addresses to names, as in the in-addr.arpa domain. The host name must be
the canonical host name.
SOA Tells the name server that all the resource records following it are authoritative for
this domain. (SOA stands for Start of Authority.) The data field is enclosed by
parentheses and is typically a multiline field. The data field of the SOA record
contains the following entries:
originÐThe canonical name of the primary name server for this domain. It's
usually given as an absolute domain name ending with a . (dot), so it's not modified
by the named daemon.
contactÐThe email contact of the person who's responsible for maintaining this
domain. Because the @ character has special meaning in resource records, it's
replaced by a . (dot). If the person responsible for maintaining zone information
about afakecompany.com is Mike, the contact address is listed in the file as
mike.afakecompany.com.
serialÐThe version number of the zone information file, which is given as an
integer. It's used by secondary name servers to determine when the zone
information file has changed. You should increment this number by one every time
you change the information file so that the name server will notice the changes.
refreshÐThe length of time in seconds that a secondary server should wait before
trying to check the SOA record of the primary name server. The SOA records don't
change very often, so you can usually set this value to be on the order of one day or
so.
retryÐThe time in seconds that a secondary server waits to retry a request to a
primary server if the primary server isn't available. Typically, it should be set to a
few minutes or so.
expireÐThe time in seconds that the secondary server should wait before throwing
away the zone information if it has been unable to contact the primary server. This
number should typically be very large, on the order of 30 days or so.
minimumÐThe default ttl value for resource records that don't specify a ttl. If your
network doesn't change very much, this number can be set to a fairly large value,
such as a couple of weeks. You can always override it by specifying a ttl value in
your resource records.
803
Using the named Daemon to Set Up the Server
Tip #190 from
Steve The MX record does not have to be another local system, but can be another system
name.afakecompany.com can host the Web site for the
entirely. The name server
domain nothinghere.com, but an MX record can forward all mail addressed to
nothinghere.com to another server not associated with afakecompany.com.
As you can see, the format of the resource records gets complicated in a hurry. Things should
get clearer as you look at a few of the master configuration files used by named.
Part
Tip #191 from
Steve Some of the options for the nslookup command can redefine the default settings for VII
the command, such as the root option. If you consistently find yourself specifying
Ch
custom settings for
directory named
nslookup, you can put the custom settings in a file in your home
.nslookup (note the . at the beginning of the file name). 38
The named.hosts File
In thenamed.boot file you created earlier in this chapter, you listed named.hosts as being the
file that contains information about your local domain, afakecompany.com. You could have
named the file anything you wanted by listing the name on the primary line of named.boot.
The named.hosts file contains authoritative information about the hosts in the zone of
authorityÐafakecompany.com. Listing 38.1 shows a sample named.hosts file that uses several
of the resource record types.
Listing 38.1 A Sample named.hosts File
; named.hosts file for afakecompany.com
;
@ IN SOA ns.afakecompany.com. mike.afakecompany.com. (
6 ; serial number
86400 ;refresh 24 hrs
300 ; retry 5 minutes
2592000 ; expire 30 days
86400 ; minimum 24 hrs
)
IN NS ns.afakecompany.com.
;
; your domain itself afakecompany.com
;
@ IN A 184.122.110.1
IN MX 100 mailhost.afakecompany.com
IN HINFO PC-486 Linux
804 Chapter 38 Configuring Domain Name Service (DNS)
Listing 38.1 continued
;
; your primary nameserver
;
ns IN A 184.122.110.1
nameserver IN CNAME ns.afakecompany.com.
;
; other hosts
;
mailhost IN A 184.122.110.2
jarre IN A 184.122.110.3
IN MX 100 mailhost.afakecompany.com
skippy IN A 184.122.110.4
IN MX 100 mailhost.afakecompany.com
;
; the localhost
;
localhost IN A 127.0.0.1
Host names in resource records that end with a . (dot) aren't translated any further.
named assumes that the host name
If the dot isn't the last character in the host name,
you gave is relative to the origin domain name referred to by @ and appends the
domain name to the host name.
Look at the named.hosts file in Listing 38.1 in detail. The first record that you come to
SOA (Start of Authority) record for the sample domain. The first line of
in this file is the
this record starts with the @ character, which indicates the current origin or domain
(afakecompany.com). The definition of the origin comes from the domain listed on the
corresponding primary line in named.boot. After that, you see the codes IN and SOA, which tell
named that this resource record uses Internet (TCP/IP) addressing and is a start-of-authority
record.
The next two entries on the line are the canonical name of the primary name server for
ns.afakecompany.com) and the email contact with the @
this domain ( replaced by a dot
mike.afakecompany.com). You then list the various fields of the data
( required by an SOA
record, one per line. (Refer to Table 38.6 for a complete explanation of each of these entries.)
After theSOA record, the next line is a name-server resource record, which lists
ns.afakecompany.com as being a name server for the domain. Because no domain is listed
in the domain field, it's assumed to be the last domain specified, which was @, listed in the
SOA record. And, of course, the @ character really expands to be the local domain,
afakecompany.com. What could possibly be easier to understand?
The next three lines set up some information about the afakecompany.com domain itself.
Although you've listed the domain name as @ for clarity because it was the last domain name
listed in the file, these resource records would still apply to it by default if you had left the
805
Using the named Daemon to Set Up the Server
domain field blank. The following line allows users to refer to afakecompany.com as though it
were a real machine:
@ IN A 184.122.110.1
It has been assigned the IP address of 184.122.110.1, which, as you'll see, is really the IP
address of ns.afakecompany.com. The next line sets up a mail exchanger MX record for
afakecompany.com so that all mail going to it gets forwarded to mailhost.afakecompany.com
instead. The last line in this group sets up a host information HINFO record for
afakecompany.com, which tells the world that it's a PC-486 running Linux.
ns.afakecompany.com as being your name server via
A few lines earlier in the file, you listed
an NS named to work correctly, you must provide an address or A record
resource record. For
that gives the address of ns.afakecompany.com. The next line in your file does just that.
Following the ``glue record'' that gives the address of the name server, you have a CNAME
Part
resource record. This record tells you that nameserver.afakecompany.com is an alias for
ns.afakecompany.com. VII
mailhost, Ch
38
You then proceed to set up address records for three other hosts in your domain:
jarre, and skippy. A records for jarre and skippy are MX records that
Notice that after the
route any mail received by jarre or skippy to mailhost.afakecompany.com . Because no name
was specified in the first field of these MX records, they apply to the previous nameÐjarre or
skippy.
Because the owner field of a resource record defaults to the last one specified if it's
left blank, you can easily group records that apply to one host. However, you must be
careful if you add new records for new hosts to a file. If you add them to the middle
of a file, you might cause the default host to change for some of the existing resource
records. Look carefully before you add resource records to an existing file.
Finally, the last host in this named.hosts file is the localhost, which is mapped to address
127.0.0.1. As you can see, the syntax for these files gets quite complicated and gives you lots
of room for errors.
The named.rev File
The named.rev file is very similar to the named.hosts file, except that it essentially works in
reverse; it maps addresses to host names. Listing 38.2 shows a sample named.rev file for
afakecompany.com.
Listing 38.2 A Sample named.rev File
; named.rev file for afakecompany.com
;
@ IN SOA ns.afakecompany.com. mike.afakecompany.com. (
6 ; serial number
806 Chapter 38 Configuring Domain Name Service (DNS)
Listing 38.2 continued
86400 ;refresh 24 hrs
300 ; retry 5 minutes
2592000 ; expire 30 days
86400 ; minimum 24 hrs
)
IN NS ns.afakecompany.com.
;
; reverse map your IP addresses
;
1 IN PTR ns.afakecompany.com.
2 IN PTR mailhost.afakecompany.com.
3 IN PTR jarre.afakecompany.com.
4 IN PTR skippy.afakecompany.com.
In this example, you have the same SOA record that you saw in the named.hosts file. This
record just sets up the authority information for the domain. In this case, @, the value of the
origin, is set to 197.198.199.in-addr.arpa from the primary line in the named.boot file.
Recall that the in-addr.arpa domain refers to reverse mapping of addresses to names.
The address listed as part of your in-addr.arpa line is your network address
backward. Your sample network for this chapter has the address 184.122.110.0.
When you list it in the reverse mapping data files, you list it as the following:
197.198.199.in-addr.arpa
You have the NS record that lists the name server for your domain. Following that are the
records that make up the reverse address resolution records. They are PTR records and give
the host number (the part of the IP address not listed in the in-addr.arpa value) and the
canonical host name that matches it. You must use the canonical host name here instead of a
relative host name. For example, the following line tells named to map the host address
184.122.110.2 to the host name mailhost.afakecompany.com:
2 IN PTR mailhost.afakecompany.com.
The named.ca File
As stated earlier in this chapter, the caching operation of named is very important.
Fortunately, the named.ca file that sets up caching is also usually the simplest of the named
configuration files. It just lists the root name servers for the various domains with their IP
addresses. It contains a couple of special field indicators that tell named that they are root
servers.
You can probably just copy the format of the sample named.ca file in Listing 38.3. To get a
complete current list of the root name servers, use the nslookup utility.
807
Troubleshooting
Listing 38.3 A Sample named.ca File
; named.ca file
;
. 99999999 IN NS NS.NIC.DDN.MIL.
99999999 IN NS NS.NASA.GOV.
99999999 IN NS KAVA.NISC.SRI.COM.
99999999 IN NS TERP.UMD.EDU.
99999999 IN NS C.NYSER.NET.
99999999 IN NS NS.INTERNIC.NET.
;
NS.NIC.DDN.MIL. 99999999 IN A 192.112.36.4
NS.NASA.GOV. 99999999 IN A 128.102.16.10
KAVA.NISC.SRI.COM. 99999999 IN A 192.33.33.24
TERP.UMD.EDU. 99999999 IN A 128.8.10.90
C.NYSER.NET. 99999999 IN A 192.33.4.12
NS.INTERNIC.NET. 99999999 IN A 198.41.0.4
Part
As you can see, the named.ca file simply maps NS name server records to the appropriate VII
addresses for them.
Ch
38
Another way to assign addresses to systems in a network is to use the Dynamic Host
Configuration Protocol (DHCP), which is very popular for systems that may not be
attached to a given network at all times, such as notebook computers. You can find
the ISC's (the Internet Software Consortium) Web site for the DHCP service on the
Web at http://www.isc.org/view.cgi?/products/DHCP/index.phtml.
Troubleshooting
DNS is a very complex system. You can do many things wrong that will cause your system
not to behave properly. Many of the problems that occur with a DNS setup may appear to be
identical but come from different causes. However, most of the problems result from syntax
errors in your configuration files. The following are a few troubleshooting tips:
& Make sure that you specify the host names correctly in your DNS configuration files.
If you're using an absolute host name, be sure to end it with a dot.
& Be especially careful with the names used in SOA and CNAME records. If you make errors
here, these resource records can redirect host name queries to computers that don't
exist.
& Be sure to increment the serial number in your configuration files when making
changes. If you forget, DNS will not reread the file.
808 Chapter 38 Configuring Domain Name Service (DNS)
& Be sure to enter the correct IP address for A records and check to see that it matches
your /etc/hosts file (if you have one). Also, make sure that the DNS name and IP
address match the corresponding reverse resolution information in named.rev.
& Your best tool for figuring out errors is the nslookup command. Use nslookup to test
your DNS server thoroughly. Do regular and reverse resolution for every address
in your DNS database to make sure that all the names and addresses are correct.
CHAPTER
39
Configuring Email
In this chapter by Steve Burnett
An Overview of Electronic Mail 810
sendmail 816
Project: Creating a Mailing List with sendmail and majordomo 822
810 Chapter 39 Configuring Email
An Overview of Electronic Mail 822
The following sections present a broad overview of electronic messaging. First is a discussion
of some of the general concepts of electronic mail, including two basic kinds of mail software
and where sendmail belongs in that division. Next, you'll find a description of the RFCs
(Requests for Comment), where the protocols used to communicate within and across
networks are defined. Finally, you'll find explanations of some of the protocols used to define
electronic messages.
History and General Concepts
One of the first widely used office mail systems was IBM's PROFS. A mainframe-based
system, PROFS had features similar to modern email systems such as Microsoft Exchange
and Lotus Notes. Such features included the following:
& Strong administration and management tools
& Security customization
& Scheduling capabilities
PROFS and other messaging systems of the time shared several similarities. Whether
mainframe- or UNIX-based, they were text-based and were considered host-based
centralized messaging systems. Because PROFS was scalable and adaptable, IBM only
recently switched away from it for its enterprise mail use.
As personal computers grew in acceptance and became widely used throughout corporations,
people started to take advantage of the shift in computing power from the mainframe to the
desktop. An early application of personal computer networks was file sharing, which made use
of a central file server and a shared universally accessible network drive. Shortly thereafter,
messaging systems began to take advantage of the new power on users' desktops. So host-
based messaging shifted (in some cases) to LAN-based messaging.
The Shared-File Messaging Model
cc:Mail is an example of LAN-based messaging, which is also called shared-file messaging. In the
shared-file messaging model, the desktop client has all the power and all the control. A client
sends messages to a mailbox on a server and polls the server to retrieve mail from its specified
mailbox directory. The server is passive, only storing messages. It performs no processing or
sorting and has no provisions for setting rules to control message flow. Shared-file messaging
provided the following gains over host-based messaging:
& Added attachments to text-only messages
& Required lower-cost servers
& Simplified setup
& Improved performance for some client actions
811
An Overview of Electronic Mail 822
However, shared-file messaging systems introduced new problems. Because each user needed
full access to the file system, including other users' mailboxes, security was an issue. Also,
because each client had to poll a server to get new mail, network traffic increased. Network
bandwidth is a bottleneck more often than server or client bandwidths are.
The Client/Server Messaging Model
The client/server messaging system divided the tasks of message processing between the
desktop workstations and the servers. Using a push model for messages, mail clients no
longer clogged the network by constantly polling for new messages. Client/server messaging
also improved on shared-file messaging by improving security so that users would have more
difficulty reading others' mail. With the more intelligent server, the sorting and processing of
messages could be performed before messages were transferred across the network to a client.
MUAs, MTAs, and MDAs
An electronic mail system can be broken down into three elements: the Mail User Agent
(MUA), the Mail Transport Agent (MTA), and the Mail Delivery Agent (MDA).
The MUA is the user interfaceÐthe software with which the user reads mail, organizes mail
into directories or folders, and sends mail. People prefer different features in their MUAs, and Part
not all MUAs are available on all platforms. Many MUAs can coexist on the same machine.
For example, a UNIX workstation can have any or all of the following MUAs available for
VII
Ch
39
use: mailx, elm, pine, mutt, mailtool, and dtmail. A given user can use any MUA present on
his or her system because the MUAs are simply local applications. In addition, MUA
functionality is often included in multipurpose software such as Lotus Notes and Netscape
Mail.
The MTA isn't used to write a mail message; it's used to route the mail from a local MUA to
another MTA on another system. (sendmail is an example of an MTA that is not used to read
or write mail directly. sendmail is intended only to deliver preformatted messages.) Mail
routing may occur either locally or remotely. In a local mail transfer in which both the sender
and destination have accounts on the same machine, the MTA is responsible for transporting
mail from itself to a local MDA. In the process, the MTA may possibly edit the protocols,
addresses, and routing of the mail message. A message created on a UNIX-to-UNIX Copy
Protocol (UUCP) network requires some transformation before that message can be received
by a person on a Transmission Control Protocol/Internet Protocol (TCP/IP) network. The
MTA acts as a gateway for mail to get a message from one network to another network that
uses different protocols. In the vast majority of situations, a given machine has only a single
MTA.
The MDA is the third component of the mail-handling routine. Although sendmail handles
SMTP mail transfer between MTAs directly, sendmail relies on Mail Delivery Agents
(MDAs) to handle local delivery from the sendmail queue to a queue used by an MUA. Two
common MDAs that sendmail is often configured to use are /bin/mail and procmail. /bin/
mail is almost universally available on UNIX systems; procmail is widely available and is both
812 Chapter 39 Configuring Email
faster and much more capable than the standard /bin/mail, providing strong capabilities for the
presorting and preprocessing of mail.
Tip #192 from
Steve To better understand the MUA/MTA/MDA relationship, consider a real-world example
of a person sending a letter. The MUA represents the person sending the letter: That
person writes a letter, places it in an envelope, puts an address and stamp on it, and
then delivers it to a post office. The MTA is like the post office staff: They accept the
letter, examine the address, reformat the address if necessary, and route the letter
either to a mailbox in the same post office (if the letter is local) or to another post
office (for a remote destination). The MDA corresponds to the postal worker who
delivers the mail from the post office to the intended location. If a gateway is used,
this analogy can be extended: An MTA that receives a letter for a destination in
another state has to transfer that message to another MTA that knows how to deliver
letters in the target state.
The IETF Requests for Comment
A Request for Comment, or RFC, is a formal description of protocol formats used on the
Internet. These protocols are also adhered to by many non-Internet systems. The RFCs are
issued by the Internet Engineering Task Force (IETF). RFCs are identified and referred to
by numbers for clarity; it's easier to refer to RFC822 than to refer to the ``Standard for the
Format of ARPA Internet Text Messages.'' As of this writing, more than 2,000 RFCs have
been published, some of which have been made obsolete by later RFCs. To find a given RFC,
look for the IETF on the World Wide Web at http://www.ietf.org/.
Because mail is such a commonly used function of the Internet, many of the RFCs set
standards for mail exchange. sendmail and other MTAs address the needs and definitions of
many of these protocols. However, attempting to describe in detail all the RFCs relevant to
mail transport and format could take years of time and thousands of pages.
Table 39.1 presents, in chronological order, the RFCs relevant to sendmail.
Table 39.1 RFCs Dealing with Electronic Mail Messaging
Number Title Comment
RFC819 Domain Naming Contains convention
for Internet user ap-
plications.
RFC821 Simple Mail Transfer Defines SMTP.
Protocol
813
An Overview of Electronic Mail 822
Table 39.1 RFCs Dealing with Electronic Mail Messaging
Number Title Comment
RFC822 Standard for the Format of Defines the format
ARPA Internet Text Mes- (headers, body, and
sages how to separate the
two) for Internet text
mail messages.
RFC976 UUCP Mail Interchange
Format Standard Defines the UNIX-to-UNIX
Copy Protocol (UUCP) for-
mat of mail messages be-
tween two systems.
RFC1123 Requirements for Internet Extends and updates
HostsÐApplication and RFC822,mostly by
Support clarifying ambiguous
issues in the original
document.
RFC1327 Mapping between X.400 Updates RFC822.
(1988)/ISO
10021 and RFC822 Part
RFC1521 and RFC1522 MIME (Multipurpose Inter-
net Mail Extensions) Parts
Provides another ex-
tension to the mail
VII
One and Two format defined in Ch
RFC822 by defining
Multipurpose Internet
39
Mail Extensions
(MIME), which,
among other things,
allows insertion of
binary files such as
graphics and sound to
mail messages. These
two were made obso-
lete by
RFC2045_2049.
RFC1651 SMTP Service Extensions Introduces ESMTP
(Extended Simple
Mail Transfer Proto-
col).
RFC1652 SMTP Service Extension for
8-bit MIME Transport
RFC1653 MTP Service Extension for
Message
RFC1869 SMTP Service Extensions Makes RFC1651
obsolete.
RFC1870 SMTP Service Extension for Makes RFC1653
Message Size Declaration obsolete.
814 Chapter 39 Configuring Email
Table 39.1 RFCs Dealing with Electronic Mail Messaging
Number Title Comment
RFC1891 SMTP Service Extension for
Delivery Status Notifications
RFC1892 The Multipart/Report Con-
tent Type for the Reporting
of Mail System Administra-
tive Messages
RFC1893 Enhanced Mail System
Status Codes
RFC1894 An Extensible Message For-
mat for Delivery Status No-
tifications
RFC2045_2049 Multipurpose Internet Mail Make RFC1521 and
Extensions (MIME) Parts RFC1522 obsolete.
One through Five
Internet Protocols
sendmail uses the Simple Mail Transfer Protocol (SMTP) to move messages between two mail
servers. Acting as a server-to-server protocol, SMTP requires another protocol such as POP3
to collect and process messages locally and deliver the messages to specific users. SMTP is the
communications protocol generally used in UNIX-based networks for mail over TCP/IP
connections. Unlike the UUCP protocol, which must have a ``map'' of which machines exist
between the sender and the destination, TCP/IP allows one system on a network to talk
``directly'' to another by passing packets of information back and forth between the two.
SMTP, which is a TCP-based client/server protocol, is defined in the IETF's RFC821, titled
``Simple Mail Transfer Protocol.''
SMTP is complex in details but is fundamentally simple. After a reliable connection is
established, the mail client (MUA) initiates a brief handshaking sequence with the mail server
(MTA). The client then sends one or more messages to the MTA for delivery. Before each
message is sent, the mail client sends a list of the message's local recipients and the sender's
address. In an obvious paper mail parallel, this information is referred to as the message's
envelope.
The handshaking sequence and message content exchange take place in a formal language
made up of four-character commands and three-digit reply codes. For example, an Extended
Simple Mail Transport Protocol (ESMTP) mail exchange log might look like this:
$ /usr/sbin/sendmail -v [email protected] < message
[email protected]... Connecting to localmail.mail.fake.com. via smtp...
220 localmail.mail.fake.com ESMTP Sendmail 8.9/8.9/; Sat, 22 May 1999 08:06:22
230_0700
>>> EHLO gateway.oppositemail.com
250 localmail.mail.fake.com Hello [email protected] [192.168.0.5],
250 pleased to meet you
815
An Overview of Electronic Mail 822
>>> MAIL From:[email protected]
250 <[email protected]>... Sender ok
>>> RCPT To:[email protected]
250 Recipient ok
>>> DATA
354 Enter mail, end with ''.'' on a line by itself
>>> .
250 WAA11745 Message accepted for delivery
[email protected]... Sent (WAA11745 Message accepted for delivery)
Closing connection to localmail.mail.fake.com.
>>> QUIT
221 localmail.mail.fake.com closing connection
A framework for additional features in electronic mail is called Extended Simple Mail Transport
Protocol (ESMTP). ESMTP is a mechanism by which any extensions used with traditional
SMTP can be negotiated between the client and server. The mechanism, as described in
RFC1651, is open ended: Two possible extensions were defined in RFC1652 and RFC1653.
RFC1652 defines 8-bit MIME encoding, which enables a user to send 8-bit data in mail
messages without having to recode the data using base64, quoted-printable, or some other
encoding method. This type of encoding also eliminates the breakage that can result from
sending 8-bit data to an RFC821-compliant SMTP server that doesn't know what to do with
the components it receives. Part
Message size declaration (defined in RFC1653) offers a method for a server to limit the size of VII
a message it is prepared to accept. With RFC821 SMTP, the only possibility is for the server
Ch
to discard the message after it has been sent in its entirety and after the message has crossed
the network onto the server. Unfortunately, deleting the message after the message has
39
arrived is a waste of bandwidth, and the mail client has no way of knowing that the message
was discarded because of its size.
Other extensions possible with ESMTP include requesting a delivery status notification on
outgoing messages (so that senders can be notified when messages arrive at their destination)
and negotiating encryption between secure mail servers for more secure mail.
Mail Message Formatting
SMTP defined how to transfer a mail message across the Internet but did not define how to
recognize a mail message. RFC822 defines the format of Internet electronic mail messages.
The format is simple, as befits a standard:
& A header containing various required and optional message attributes
& A blank line
& The message contents
The header fields are much longer than the content in the sample message given here:
Return-Path: [email protected]
Received: from localmail.mail.fake.com (localmail.mail.fake.com [168.9.100.10])
by gateway.oppositemail.com (8.9/8.9) with ESMTP id WAA01322 for
<[email protected]>; Sat, 22 May 1999 18:17:06 _0500
816 Chapter 39 Configuring Email
Received: from beta.mail.fake.com (beta.mail.fake.com [207.266.47.2]) by
localmail.mail.fake.com (8.9/8.9) with SMTP id WAA13732 for
<[email protected]> Sat, 22 May 1999 18:22:06 _0500
Message-Id: [email protected]
X-Sender: [email protected]
X-Mailer: Amiga Eudora Lite Version 2.1.2
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii''
Date: Sat, 22 May 1999 18:22:08 _0500
To: [email protected]
From: David Wylie [email protected]
Subject: Test message
This is a test message.
David
The blank line after the Subject line divides the header from the message body that follows.
Any subsequent blank line is part of the message body and has no structural significance.
Most header fields are brief and have a fairly obvious meaning (such as Subject), whereas
some others are lengthy and not readily understood (such as Received). For a detailed
explanation of the many standard and less-standard header fields, see Chapter 35 of Bryan
Costales's and Eric Allman's sendmail, 2nd Edition (O'Reilly & Associates).
Each header line consists of a ``keyword-value'' pair that defines a characteristic of that
message. For example, a required characteristic of a mail message is a message recipient. This
characteristic is defined by the keyword To:, one or more spaces or tabs, and then the value
that specifies the mailing address of the recipient. In the preceding message, this
characteristic is defined by the following line:
sendmail
sendmail is generally considered one of the few true nightmares of UNIX system
administration. sendmail is difficult to configure and can be approached in much the same
way that novices approach UNIX. When someone once complained to Eric Allman (the
creator of sendmail) that sendmail administration was complicated, he replied, ``Configuring
sendmail is complex because the world is complex.'' Although sendmail can do just about
anything you can think of, instructing it how to do what you want it to can be a chore.
However, although sendmail is difficult to work with, recent versions have improved the tasks
of configuring and administering sendmail substantially. The addition of a large set of M4
macros and the ability to use intelligible names for options plus the single-character switches
in the configuration file has made sendmail configuration an easier task. sendmail has also
become a reasonably mature product. Although flaws are still found almost monthly,
sendmail is used in enterprise networks for mail delivery across a wide set of networks and in
high-volume environments.
817
sendmail
sendmail's History
In the late 1970s, Eric Allman was at the University of California at Berkeley. He wrote the
predecessor to sendmail, called delivermail, which was released in 1979 to solve the problem
of transferring mail among the three networks on campus at that time. Those three networks
were ARPAnet (which was using NCP, or Network Control Protocol), a UUCP mail system,
and an internal network called BerkNet.
The next year, ARPAnet started to convert from NCP to TCP (Transmission Control
Protocol). Previously, mail was delivered using FTP (File Transfer Protocol), but SMTP was
developed to plan for the possible growth of the network's mail traffic by several orders of
magnitude.
In response to these changes, Allman adopted an inclusive approach to formats of electronic
mail messages. If a message didn't match the preferred format, sendmail attempted to fix the
message format instead of immediately rejecting the message. Allman also chose to limit the
functional goal of sendmail to be a router of mail, instead of including the functionality of an
end-user mail application. The 4.1c version of BSD (Berkeley Software Distribution) UNIX
included the first public release of sendmail.
Meanwhile, others were busy extending sendmail's capabilities separately from Allman. In
Part
addition to various private efforts, several commercial vendors such as Sun and Hewlett-
Packard developed their own versions of sendmail as they saw needs for improvements not
VII
included in the current versions. Out of these parallel developments came several versions of Ch
sendmail with varying levels of compatibility. In 1998, Allman took sendmail to commercial 39
status as of version 8.9, leaving the 8.8.x version the freeware it always was.
Tip #193 from
Steve A reference to ``V5,'' ``V7,'' ``V8,'' or something similar is referring to the version
number of sendmail. V5 is version 5.x, V7 is version 7.x, V8 is version 8.x, and so on.
sendmail's Architecture
In general, compilation and installation of the sendmail distribution are often simpler than
they first appear. The source package includes make-description files tailored for many
different systems and a ``build'' script that chooses the correct one for the local environment.
At times, administrators might need to make minor changes to the make-description file
that's most similar to their particular environment in order to match the specific local system.
sendmail the Daemon
sendmail itself is normally configured to run on a UNIX system as a daemon to listen for
incoming mail.
818 Chapter 39 Configuring Email
A daemon is a UNIX system program that runs in the background without a
controlling terminal window.
When sendmail is run as a daemon, unless it's instructed not to on startup, it forks and runs in
the background, listening on socket 25 for incoming SMTP connections. The command to
run sendmail as a daemon on a Berkeley UNIX-based system might look something like this:
/usr/lib/sendmail -bd -q30m
This command can be defined as one of the startup commands executed when the UNIX
system boots. The following sample command is taken from a startup script named
sendmail.init found in the /etc/rc.d/init.d directory on a Linux system:
# Start daemons.
echo -n ``Starting sendmail: ''
daemon sendmail -bd -q1h
echo
touch /var/lock/subsys/sendmail
;;
The -bd flag launches sendmail as a daemon, and the -q1h switch instructs sendmail to check
the queue once every hour. In contrast, the sample command preceding this one has the -q
switch instructing sendmail to check the queue every 30 minutes.
The first action sendmail takes when it's started is to read the /etc/sendmail.cf
configuration file.
sendmail Configuration and Control Using the sendmail.cf File
Part of sendmail's power is derived from the access provided to sendmail's underlying
configuration files. As mail messages are funneled through sendmail's configuration files,
sendmail performs all message routing functions, including parsing, forwarding, delivering,
returning, and queuing.
The core of sendmail's configuration is the sendmail.cf file. A complex configuration file
read only once at sendmail's initial runtime, sendmail.cf contains three important types of
information:
& Options such as operational control switches, mailer definitions, and the locations of
other sendmail subconfiguration files
& Rulesets for rewriting addresses on incoming and outgoing messages
& Macros for use in rulesets
Tip #194 from
Steve No one bothers to write a sendmail.cf file starting with a blank page in a text
editor. If you are configuring sendmail for a new network, you can almost certainly
819
sendmail
find a sendmail.cf file that requires minimal changes. Of course, after you have a
functioning mail server, you should make a backup of the working configuration and
put it somewhere safe.
Version 8 and above of sendmail added the use of the m4 macro preprocessor, which is used
to generate sendmail.cf files containing the features you select. A sendmail m4 creation file
typically should be given an .mc (macro configuration) file suffix, but it is not required for the
process to work. Many sample .mc scripts are supplied with the standard sendmail
distributions.
For example, a minimum .mc file for a Linux workstation (without appropriate comments)
could be something like this:
OSTYPE(linux)dnl
MAILER(local)dnl
The preceding are the only two required macros in an .mc file. You are likely to want more
features, but this fileÐnamed smallest_linux.mcÐcould be run with the following
command (assuming you are in the /usr/lib/sendmail/cf/cf directory, where the standard
Part
sendmail distribution places m4 files):
m4 ../m4/cf.m4 smallest_linux.mc > sendmail.cf VII
Ch
The following list provides a breakdown of the elements in the preceding command:
m4 Calls the m4 preprocessor
39
../m4/cf.m4 Identifies m4's default configuration file
smallest_linux.mc Is the two-line macro configuration file
> sendmail.cf Specifies that output is to be placed in the sendmail.cf
file
Now that you have used m4 to generate a sendmail.cf file containing exactly the features you
requested, you still need to customize the sendmail.cf file for use at your site. Using m4 for
sendmail.cf generation, however, is fast and accurate. In addition to using the many m4
macros that ship with the sendmail distribution, you can also write your own as you feel
necessary and include them for use.
Tip #195 from
Steve For a quick sendmail configuration file, you can fill out the World Wide Web form
interface to the m4 configuration tool (for version 8 sendmail) at http://
www.completeis.com/sendmail/sendmail.cgi. Select your desired options with
the Web form, and a sendmail.cf file with the chosen options will be returned to
you.
820 Chapter 39 Configuring Email
sendmail Configuration Files and Their Locations
sendmail.cf is the first file that sendmail reads on startup. The sendmail.cf file contains the
locations of all other subconfiguration files used by sendmail, which are listed in Table 39.2.
Table 39.2 sendmail Configuration Files
Filename and Location Description
/etc/aliases ASCII text list of defined aliases to names
/etc/aliases.db Database of aliases compiled from /etc/aliases
/etc/sendmail.hf Help file
/var/log/sendmail.st Collected statistics
/var/spool/mqueue/* Temporary files for mail queue
/var/run/sendmail.pid The process ID of the daemon
The locations listed in Table 39.2 are only the default locations of the files. Because their
locations are defined within sendmail.cf, you can set them to whatever name and directory
path you like.
sendmail contains far too many configuration options to list in this book. The syntax for
options comes in two types: very cryptic and a bit less cryptic. In the cryptic version of option
syntax, the O (capital o, not zero) command starts an option command in the sendmail.cf file.
So the following two sample commands from a sendmail.cf file perform the same function:
O8pass8
O EightBitMode=pass8
These commands tell sendmail to pass 8-bit formatted data as 8 bits, and not to truncate it to
7 bits. Notice the syntax change: The single-character version ( O8) does not contain a space
between the O and the number signifying which option is being set, whereas the name version
O EightBitMode) must contain a space between the O command and the name of the option.
(
As with all other sendmail commands, the O must be in the leftmost position on the line,
which is also column 1.
This restriction prevents misinterpretation of commands, such as this next line that also can
be found in a sendmail.cf file:
DMMONGO
M
This command defines , a macro ( ), to have the value MONGO so that you can use $M when
rewriting rules instead of typing MONGO. Without the restriction that a command is identified
by an O in the leftmost column, the O in MONGO might be interpreted as a command.
The options just presented illustrate the form of the option command for use within a
configuration file. However, you can define options either in an m4 macro file or on the
command line. The command-line versions of the preceding options use a dash before the
821
sendmail
option, a lowercase o to indicate a single-character option command, and an uppercase O to
indicate a named option command, as shown in these examples:
-o8pass8
-O EightBitMode=pass8
-O EightBitMode=pass8
sendmail Rulesets
sendmail uses rules to rewrite addresses on incoming and outgoing mail. These rules are the
center of sendmail's capability, as well as its complexity: sendmail's rewriting rules are a
specialized text-oriented programming language. Eric Allman designed sendmail so that the
ruleset performs two core tasks:
& Examines each recipient's address to determine which MDA should be used to send
the message to (or nearer to) the recipient
& Transforms addresses in both the envelope and the message header to facilitate
delivery or reply
Rewriting rules are organized into rulesets. A ruleset is a subroutine or module consisting of a
sequence of rules. When an address is passed to a ruleset, the subroutine passes the address to Part
each of its rules in order. If the matching clause matches the investigated address, the rule is VII
applied, the address is transformed, and the result is passed to the next rule. If the address
Ch
39
does not match the current rule, the address is not transformed, and the next rule in the set is
tried.
sendmail Ruleset Syntax
Each ruleset is identified by a number, and each new ruleset begins with an S in the leftmost
column, followed by its identifying number. Rules begin with the letter R and are not
numbered. A non -R command ends the ruleset, for example:
######################################
### Ruleset 0 -- Parse Address ###
######################################
S0
R$* $: $>98 $1 handle local hacks
Rule syntax is cryptic but fairly simple. Each rule has a left-hand side and a right-hand side. A
comment portion is optional. The two sides and the optional comment are separated by tabs.
The left-hand side is compared to the address as a string pattern. If the pattern matches the
left-hand side, the address is transformed by the rule's right-hand side and is passed on to the
next rule.
In sendmail.cf, #
an octothorp ( ) begins a comment line. Empty lines are ignored. The S0
defines the beginning of Ruleset 0. The R on the next line defines the beginning of a rule. The
$* accepts every address that is passed to it, and the $: $>98 $1 passes the address to Ruleset
98 for further processing. The text handle local hacks is a comment. Because rules are tab-
delimited, the comment portion does not require a comment marker ( #)at the beginning.
822 Chapter 39 Configuring Email
The Core sendmail Rulesets
Several standard rulesets exist, and they may appear in any order in sendmail.cf. When
sendmail reads the configuration file, it sorts the rules appropriately. A ruleset that is expected
but is not present is treated as if it were present but empty. The following are the main
rulesets:
& Ruleset 0 resolves an MDA by reading the address.
& Ruleset 1 processes the sender's address.
& Ruleset 2 processes the recipient's address.
& Ruleset 3 preprocesses all addresses.
& Ruleset 4 post-processes all addresses.
& Ruleset 5 rewrites unaliased local users.
Aliasing in sendmail
An alias is an abbreviation for one or more full mailing addresses. Although an alias may be
merely a nickname for a longer address you don't want to type every time (such as john for
[email protected]), an alias can also be the name of a list of several
recipients.
Many MUAs maintain their own alias lists, but these alias lists are normally in formats that
cannot be shared with other MUAs. If you typically use pine on a Linux workstation, its alias
file is not available to your Lotus Notes client on your Windows 95 or 98 workstation when
you write a letter with that tool. In contrast, the many possible alias lists contained in aliases
maintained in sendmail's alias file are recognized and expanded when a message is processed
by sendmail, regardless of the MUA used to create that message. sendmail allows for multiple
alias filesÐup to 12 by default.
You can learn more details about sendmail from the Web at http://www.sendmail.org/ or
from Bryan Costales's and Eric Allman's book sendmail, 2nd Edition (O'Reilly & Associates).
Project: Creating a Mailing List with sendmail and
majordomo
majordomo is one of the most common mailing list programs on the Internet. It allows you to
create and manage mailing lists. After the mailing list is created, the list can be maintained
entirely through email, without root access to the mail server.
Before you can create a mailing list, you need the following:
& A Linux (or other UNIX) system
& sendmail or another mail transport agent installed
& Perl (because majordomo is a Perl script)
823
Project: Creating a Mailing List with sendmail and majordomo
Go to http://www.greatcircle.com/majordomo/ and download and install majordomo. After
you have majordomo installed, you can make a mailing list by following these steps:
1. Make sure you have a secure connection (either local or over ssh).
2. Log in.
3. Change to root by entering su root and the root password (unless you were already
root).
4. Enter the following to move to the /etc/ directory:
cd /etc/
5. Enter the following to edit the aliases file:
vi aliases
6. Write a block that looks like this:
# my friends
friends: ``|/usr/local/majordomo/wrapper resend -l friends friends-list''
friends-list: :include:/usr/local/majordomo/lists/friends
owner-friends: yourusername
friends-owner: yourusername
friends-approval: yourusername
friends-request: ``|/usr/local/majordomo/wrapper request-answer friends'' Part
7. When you're done, remember to save and exit vi: VII
Esc :wq
Ch
8. Enter the following command to compile the aliases:
newaliases
39
You should see a reply that looks like this:
/etc/aliases: 25 aliases, longest 39 bytes, 528 bytes total
/etc/aliases.majordomo: 63 aliases, longest 94 bytes, 2966 bytes total
9. Enter the following:
cd /usr/local/majordomo/lists
10. Enter the following to create a new file that will contain the email addresses of the
subscribers:
vi friends
11. Type your email address; then save and quit vi.
12. Enter the following to change the file's ownership from root to majordom:
chown majordom friends
13. Enter the following to change the file's group membership from root to daemon:
chgrp daemon friends
14. Enter the following to get out of root:
exit
824 Chapter 39 Configuring Email
15. In your mail program, send mail to the listname. In this example, it's
[email protected]. majordomo then creates a configuration file named
friends.config in /usr/local/majordomo/lists.
16. Customize the config settings. Some suggested changes to the default settings for a
new list would be the following:
From: admin_passwd = friends.admin
To: admin_passwd to [whatever you want]
From: description =
To: description = The mailing list for my friends
From: mungedomain = no
To: mungedomain = yes
From: reply_to =
To: reply_to = friends-list
From: subscribe_policy = closed
To: subscribe_policy = open
From: subject_prefix =
To: subject_prefix = [FRIENDS]
17. Enjoy the list!
CHAPTER
40
Configuring a Usenet News
Service
In this chapter by Steve Burnett
A Usenet Primer 826
Usenet Servers 828
Configuring Usenet Clients 829
Project: Configuring Pine as a Newsreader 831
826 Chapter 40 Configuring a Usenet News Service
A Usenet Primer
Usenet is often confused with the Internet, but Usenet isn't the Internet. Usenet is not a
network, but instead is a service carried over the Internet, as well as many computers not
directly part of the Internet. The best way I've found to describe Usenet is to say that it is
20,000 (or so) bulletin boards, each with a different title describing what the topic for that
board is supposed to be. You can look for a bulletin board with a topic you think you might
be interested in and read some or all of the messages on the board that day. If you want to,
you can put a message up to either reply publicly to someone else's message or to start a new
discussion. You can also copy a person's address and send him or her a private letter that
doesn't appear on the board. Later, you can come back and see whether the board has any
new and interesting messages from other people.
Usenet is unlike a party telephone line because you don't deal with other people in real-time.
You can't interrupt someone while he or she is thinking of what to write on the board. (You
could, however, repeat that person's message afterward and quote it out of context; but
besides being rude, that's not the same as interrupting the person and preventing others from
hearing his or her words.) Usenet is very like a party, though, because very little control is
placed on who can say something. If a person insists, for example, that squirrels are the only
warm-blooded animals that cannot carry rabies, he or she can post that message. Of course,
people who know this statement is false can reply with the correct information. Then, while
the first person can continue to insist that he or she is right, the rest of the readers of that
board are likely to start ignoring messages from the first person.
For information on the history of Usenet, you can use Netscape to go to http://
www.yahoo.com/Reference/FAQs/ for several FAQs (Frequently Asked Questions)
concerning Usenet.
History and Origins of Usenet
Back in the dark ages of computing (circa late 1970s), a version of UNIX labeled V7 was
released. One of the applications included was UUCP, which stands for UNIX-to-UNIX-
Copy. In 1979, two graduate students at Duke University started using UUCP to exchange
messages between two systems at the university. Next, a set of shell scripts was developed to
exchange messages between Duke and the network at the University of North Carolina at
Chapel Hill. Later, the shell scripts were rewritten in C, and they have been rewritten and
extended many times since then. In general, Usenet works by news servers connecting to each
other and asking ``Do you have this posting #37590576 to newsgroup alt.binaries.misc?
No? Okay, here it is.'' Or alternatively, ``Do you have this posting #37590576 to newsgroup
alt.binaries.misc? Yes? Okay, how about this one?'' until the first server has passed every
message on topics the second server will accept. The server software was first called A News,
827
A Usenet Primer
then B News, then C News, and then INN (Internet News). Other news server software includes
Netscape's Collabra and Backplane's Diablo.
Usenet Structure
To quote Douglas Adams in The Hitchhikers' Guide to the Galaxy, ``Space is BIG. Really BIG.''
Usenet (as of this writing) is approximately 20,000 different newsgroups, with several million
total participants. Some of these newsgroups are dead, and no one ever posts to them. Some
newsgroups are highly active and are likely to split into multiple newsgroups soon (either
because no one can keep up with the sheer volume or because a large segment of that
newsgroup is interested in a narrower subset of topics than the other readers of the
newsgroup).
An example of newsgroup spawning happened in comp.sys.powerpc, a newsgroup devoted to
discussions of the PowerPC RISC processor. When Be, Inc. announced the BeBox, a dual-
processor workstation running a new operating system, a substantial fraction of the
newsgroup focused exclusively on Be's hardware and software. To accommodate both the
interests of the people who wanted to discuss the BeOS and the wishes of the people with no
interest in the BeOS, a newsgroup called comp.sys.be was formed. Alternatively, a topic
discussion such as ``undocumented features of the PowerPC processor family'' would be of
interest to the entire newsgroup, and the information generated from this discussion might
become a large section of the newsgroup's FAQ or a separate FAQ altogether.
Tip #196 from
Steve If you are new to Usenet, you should make sure the newsgroup
news.announce.newusers is Part
VII
on your subscription list. Basic informational guides
about aspects of Usenet are periodically reposted to this newsgroup every two weeks.
Ch
40
Although the sheer volume of Usenet can appear overwhelming, its structure is based on
some logic. Table 40.1 presents some of the first-level divisions of Usenet. A first-level
identifier appears as the leftmost part of every Usenet name.
Table 40.1 Usenet Hierarchy Names
Hierarchy Name Description
biz Business
comp Anything to do with computers
misc Miscellaneous
news Usenet issues, general information
rec Recreational (sports, crafts, hobbies)
sci Scientific
828 Chapter 40 Configuring a Usenet News Service
Table 40.1 Usenet Hierarchy Names
Hierarchy Name Description
soc Social (personal, cultural)
talk Conversation on anything
alt Everything else
Other first-level identifiers for newsgroups tend to be regional. For example, a newsgroup
with a name starting with de.* is generally populated by speakers of the German language,
and most of the de.* hierarchy deals with German and European issues.
The alt.* hierarchy of newsgroups is an enormous part of Usenet. The requirements for
creating an alt.* newsgroup are easier than the requirements for creating a major hierarchy
newsgroup. In addition, the alt.* newsgroups are not always carried by every Internet access
provider, for two reasons. The first reason is bandwidth: The alt.* newsgroups are a
substantial portion of all Usenet newsgroups, and some of themÐespecially the newsgroups
devoted to binary files of either applications or imagesÐcan take up enormous amounts of
bandwidth. The second reason for restricting distribution of the alt.* newsgroups is
offensiveness: The alt.* newsgroups tend to tolerate more extreme or obnoxious language
and topics than the mainstream newsgroups, and people are more likely to be offended by
their content.
Some Internet access providers have a ``user-defined'' policy for Usenet access: They
provide groups from the full Usenet newsgroups list, but they carry only newsgroups
the users of that network have asked for. This ``a la carte'' policy substantially reduces
the actual bandwidth of the Usenet feed required for the network, but the users'
interests are not restricted or censored. Other Internet access providers have a mixed
policy: They carry only groups that they approve of and that their users ask for.
? See Chapter 35, ``Surviving Usenet News,'' p. 741
Usenet Servers
Many Usenet news server programs have been written. A few of them are described here:
& INN is one of the most common news server applications. The cornerstone of the
package is innd, an NNTP server service daemon that is descended from C News.
INN was maintained by Rich Salz until he turned over ongoing development and
maintenance to the Internet Software Consortium (ISC) in 1996. If you're familiar
with older news servers, innd can be characterized as a C News relaying news service
that can read multiple NNTP streams. Access to the news posts in the news server's
database is handled by a separate server named nnrpd. A new nnrpd is started or
829
Configuring Usenet Clients
spawned for each client newsreader accessing the local database. The latest version is INN
2.2 available from servers as indicated on http://www.isc.org/inn.html. Dave Barr
maintains an excellent informational INN Web page at http://www.visi.com/~barr/
INN.html.
& Diablo, available from Matt Dillon's Backplane Systems, is a backbone news feed
transit server. Diablo is designed to accept news feeds from other news hosts and route
them to other news servers quickly and efficiently. You cannot run the transit server
portion of Diablo on a machine that needs to accept nntp post commands or
newsreading-related nntp commands. It's a router for news feeds only.
& Leafnode is a news server package designed for small sites only. It's intended to serve
tens of clients and to operate over a slow link to the Internet. Leafnode uses very little
space on disk and comparatively little bandwidth, and it tends to recover rapidly from
errors. Unfortunately, its recovery strategy is generally to delete a news post if it's
corrupted or otherwise malformed, so you might lose a news post you're reading if a
problem occurs. You can find the software and more information at http://
wpxx02.toxi.uni-wuerzburg.de/~krasel/leafnode.html .
Configuring Usenet Clients
Usenet operates on the familiar client/server relationship: A server exchanges messages with
another server and stores the messages on the local system. To read a Usenet newsgroup, you
need to contact your network or Internet service provider (ISP) and ask for the name of an
NNTP server. When you have a server name (which normally looks something like either
test.fake.com or 192.168.2.221), you can start. Part
TIN and NN
VII
Ch
40
TIN and NN are two very similar newsreaders with similar configuration requirements, and
both are included in many common Linux distributions.
If you're using a version of TIN compiled with the NNTP options from a UNIX shell
account, you can try one of these commands:
& If you are using the ksh or bash shell, use the following:
$ NNTPSERVER= test.fake.com tin -r -f .fakenewsrc I .newsnet/.index
& If you are using the C shell or tcsh shell, use the following:
% setenv NNTPSERVER test.fake.com; tin -r -f .fakenewsrc I .newsnet/.index
For NN, the configuration is similar:
& If you are using the ksh or bash shell, use the following:
$ NNTPSERVER=test.fake.com nn newsrc=~ .fakenewsrc
& If you're using the C shell or tcsh shell, use the following:
% setenv NNTPSERVER test.fake.com nn newsrc=~/.fakenewsrc
830 Chapter 40 Configuring a Usenet News Service
At some point while reading newsgroups, you either might want to contribute to a current
discussion or start a new conversational topic. Before you start actively participating in
Usenet, you need to understand the rules of the Usenet subculture. Follow this general advice
when you start posting messages to a newsgroup:
& Don't post until you've read the group for at least a week and have a feel for the tone
of the newsgroup. Do you consider the average discussion too rude and brutal, or
would the newsgroup's current residents consider you an irritant?
& Look for and read the FAQ (if one exists). Many of the Usenet's FAQs are archived at
ftp://rtfm.mit.edu/pub/usenet/.
& Post only to newsgroups in which your message is relevant. If you are trying to sell a
waterbed in California, you should probably advertise only in .forsale newsgroups in
the immediate area. Readers of the triangle.forsale newsgroup serving the eastern
North Carolina area are unlikely to want to ship your waterbed across the country.
Similarly, if you have a question about Amiga computers, don't post your query to a
Macintosh-oriented newsgroup.
You might post a message to a newsgroup but look at that newsgroup later and not see your
message. Some newsgroups are moderated, which means that all posts to that newsgroup are
read by a person or group of persons who weed out inappropriate messages and don't send
them to the newsgroup. You can find out whether a newsgroup is moderated by reading the
FAQ for that newsgroup, by reading the charter for that newsgroup, or by reading messages
on that newsgroup for a week or two before posting, and noticing whether anyone describes
himself or herself as a moderator of that newsgroup.
If you don't see a FAQ immediately visible on the newsgroup when you first log in, try
http://
using one of the Usenet-capable search engines such as
www.dejanews.com or http://www.altavista.digital.com and searching
for the newsgroup name and FAQ as keywords. Alternatively, you might ask ``Is
there a FAQ for this group?'' in your first message to save you a world of angst.
In general, if you use common sense, you'll probably be fine. For more advice on the
sometimes tricky topic of Netiquette, you can use your Web browser to go to http://
www.fau.edu/netiquette/netiquette.html, a Web site with many good resources for
newcomers to the Internet in general.
831
Project: Configuring Pine as a Newsreader
Project: Configuring Pine as a Newsreader
Pine is most commonly used as a mail reader, but it can also be used as a newsreader. To set
up your pine mail client for accessing Usenet, press S (Setup) and then C (Config). Then edit
the line news collections to read as follows:
*{test.fake.com/NNTP}[]
Next, press E to exit pine and restart it. Then press L (List Folders), go down to the news
folders, and select A (Add) to subscribe to the newsgroups you want.
The resulting screen might look similar to this:
PINE 3.96 FOLDER LIST Folder: INBOX 313 Messages
---------------------------------------------------------------------------------
Folder-collection <mail/[]> ** Default for Saves ** (Local)
---------------------------------------------------------------------------------
[ Select Here to See Expanded List ]
---------------------------------------------------------------------------------
News-collection <News on test.fake,com> (Remote)
---------------------------------------------------------------------------------
[ Select Here to See Expanded List ]
Although reading news can be fun and informative, only reading (and not posting) messages
in newsgroups is a behavior pattern called lurking, which may be frowned upon by long-time
Usenet people. If you want to post new Usenet messages to newsgroups or to send replies by
electronic mail, you must typically also fill out the following fields:
& Your name (first and last name)
& Your email address ( [email protected]) Part
& The computer that forwards your mail (ask your system administrator for the SMTP
VII
mail server's name)
Ch
This information is typically already set up for your mail reader software. You should be able
40
to use the same entries. Check with a system administrator to make sure, though.
Most system administrators have an information sheet already prepared with this
information. If they don't, keep this information summary where you can get to it so
that the next person to ask can benefit from your research.
With pine and other newsreaders, as well as most email applications, you can define a
signature file. A signature file is a block of information you want included with every posting
you make to a newsgroup. Typically, people include their names, email addresses, and (if a
work-related account) their titles or ranks. Sometimes people might also include a short
quote they consider witty.
832 Chapter 40 Configuring a Usenet News Service
Tip #197 from
Steve Information you would rarely want to put in a signature file are your home address
and telephone number. Without even considering the possibility of malicious or
prankish behavior, there is always the possibility that the person making a legitimate
business call from Singapore to you (in New York City) can forget the time zone
difference. Remember that if a person can read your email or news post over the
Internet, he or she can almost always reply the same way.
Another caution if you set up a signature file is to remember that you have your signature file
defined. Manually pasting your signature file onto the end of your outgoing message or
attaching the signature file as an attached file when the signature file is automatically included
will make you look silly.
PART
VIII
Appendixes
A Sources of Information 835
B The Linux How-To Index 843
C The GNU General Public License 861
D The Open Source Definition 871
APPENDIX
A
Sources of Information
In this appendix
Linux Web Sites 836
Usenet Newsgroups 837
Online Documents 839
Magazines 840
Linux FTP Sites 840
For Linux Developers 841
836 Appendix a Sources of Information
Linux Web Sites
Because Linux is a child of the Internet, you can find a great many Web sites related to Linux.
In fact, Linux is a pretty popular subject on the Web. Table A.1 lists the URLs that contain
most of the Linux information on the Web.
Table A.1 Major Linux Web Sites
URL Description
http://metalab.unc.edu/mdw The site for Linux information; the
home of the Linux Documentation
Project (LDP)
http://metalab.unc.edu/linux/links.html Home to a plethora of links to many
Linux related sites; one of the best on
the Net
http://www.Linux.org.uk The Web site for European Linux
users
http://www.li.org The Linux International Web site
http://www.linuxworld.com The Linux World Web site
http://freshmeat.net A site that provides recent announce-
ments for a wide variety of Linux
projects (Not for newbies)
http://slashdot.org The place to get all the news a geek
could ever want; not 100 percent Linux
but great nonetheless
http://www.redhat.com The Red Hat Linux Web site
http://www.caldera.com Caldera's Web site
http://www.linuxgames.com A site-tracking development of games
for the Linux platform
http://www.linux.org The Linux Organization Web site
http://www.debian.org The Official Debian Linux Organiza-
tion Web site
http://linuxtoday.com A site dedicated to providing relevant
news about Linux and the open source
industry to the business community
http://www.corel.com Corel's Web site with links to all their
Linux related products, including
WordPerfect for Linux
http://www.stardivision.com The official Web site for
StarDivisionÐwhose products
include StarOffice
http://www.slackware.org The official Web site for Slackware
837
Usenet Newsgroups
Table A.1 Major Linux Web Sites
URL Description
http://sunsite.unc.edu/linux-source The Linux Source Navigator, which
allows you to view Linux source code in
hypertext
http://www.yahoo.com/Computers_and_Internet/ The Yahoo! site pointing to many
Operating_Systems/Unix/Linux current sites
Usenet Newsgroups
If you have access to Usenet newsgroups, you might enjoy the following newsgroups, which
provide a variety of information about Linux. Only two, comp.os.linux.announce and
?
comp.os.linux.answers, are moderated.
See ``Usenet Culture,'' p. 747
& comp.os.linux.announceÐThis moderated newsgroup is used for important
announcements, such as bug fixes.
& comp.os.linux.answersÐThis moderated newsgroup provides answers to any of your
Linux questions, especially about setting up Linux. Please read the appropriate Linux
documentation and FAQs before posting a question to this group.
& comp.os.linux.development.systemÐThis newsgroup is devoted to the many
programmers around the world who are developing the Linux system.
& comp.os.linux.development.appsÐThis newsgroup is devoted to the many pro-
grammers around the world who are developing applications for Linux.
& comp.os.linux.hardwareÐThis newsgroup provides answers to hardware compat-
ibility questions.
& comp.os.linux.setupÐThis newsgroup provides help with Linux setup and
installation problems. Part
& comp.os.linux.advocacyÐThis newsgroup provides a medium to discuss why Linux VIII
is the greatest operating system.
App
& comp.os.linux.networkingÐThis newsgroup provides answers to networking Linux
A
with the rest of the world.
& comp.os.linux.xÐThis newsgroup provides answers to installing and running X
under Linux.
& comp.os.linux.m68kÐThe purpose of this newsgroup is to promote interest and
development of the port of Linux to Motorola's 680x0 architecture.
A newsgroup named comp.os.linux.misc serves as a catchall for any Linux topic not suited to
the other newsgroups. Also, more than 170 other Usenet newsgroups contain the word Linux.
A sample of the more common Linux newsgroups are listed as follows. Go exploring!
838 Appendix a Sources of Information
alt.linux.sux alt.os.linux
alt.uu.comp.os.linux.questions alt.os.linux.slackware
aus.computers.linux dc.org.linux-users
de.comp.os.linux.hardware de.comp.os.linux.misc
de.comp.os.linux.networking de.comp.os.linux.x
de.alt.sources.linux.patches uk.comp.os.linux
fj.os.linux fr.comp.os.linux
han.sys.linux linux.apps.bbsdev
linux.apps.linux-bbs linux.apps.seyon
linux.apps.seyon.development linux.apps.flexfax
linux.debian linux.debian.announce
linux.debian.user linux.dev.gcc
linux.dev.680x0 linux.dev.admin
linux.dev.apps linux.dev.bbs
linux.dev.c-programming linux.dev.config
linux.dev.debian linux.dev.doc
linux.dev.fido linux.dev.fsf
linux.dev.fsstnd linux.dev.ftp
linux.dev.hams linux.dev.ibcs2
linux.dev.interviews linux.dev.japanese
linux.dev.laptop linux.dev.linuxbsd
linux.dev.linuxnews linux.dev.linuxss
linux.dev.localbus linux.dev.lugnuts
linux.dev.mca linux.dev.mgr
linux.dev.msdos linux.dev.net
linux.dev.new-lists linux.dev.newbie
linux.dev.normal linux.dev.nys
linux.dev.oasg linux.dev.oi
linux.dev.pkg linux.dev.ppp
Linux.dev.qag linux.dev.scsi
linux.dev.serial linux.dev.seyon
linux.dev.sound linux.dev.standards
linux.dev.svgalib linux.dev.tape
linux.dev.term linux.dev.uucp
linux.dev.wabi linux.dev.word
839
Online Documents
linux.dev.kernel linux.dev.x11
linux.fido.ifmail linux.free-widgets.announce
linux.free-widgets.bugs linux.free-widgets.
development
linux.local.chicago linux.local.nova-scotia
linux.local.silicon-valley linux.motif.clone
linux.new-tty linux.news.groups
linux.ports.alpha linux.samba
linux.samba.announce linux.sdk
linux.wine.users linux.test
DejaNews, now located at www.deja.com, archives most, if not all, of the Linux newsgroups.
LinuxWorld also provides an open news server for its site located at forum.linuxworld.com.
Online Documents
Matt Welsh spearheads a dedicated group of Linux enthusiasts who are systematically writing
a complete set of Linux manuals that are made available on the Internet. You can find the
latest versions of the documentation at sunsite.unc.edu in the /pub/Linux/docs directory.
You can also find earlier versions of these documents in your version of Linux's /docs
directory. The current home for the LDP is located at this address:
http://sunsite.unc.edu/mdw
Available documents include the following:
& ``Linux Installation and Getting Started,'' by Matt Welsh.
& ``The Linux System Administrators' Guide,'' by Lars Wirzenius.
& ``The Linux Network Administrators' Guide,'' by Olaf Kirch.
& ``The Linux Kernel Hackers' Guide,'' by Michael K. Johnson. Part
& ``The Linux Frequently Asked Questions (FAQ) List,'' maintained by Ian Jackson. It's VIII
composed of questions and answers on myriad Linux topics.
&
App
&
``The Linux META-FAQ,'' maintained by Michael K. Johnson.
``The Linux INFO-SHEET,'' maintained by Michael K. Johnson.
A
& ``The Linux Software Map,'' maintained by Aaron Schab. This document contains
information about each of the software packages available for Linux via FTP.
Linux How-Tos
Appendix B, ``The Linux How-To Index,'' provides an index to all the available How-To
documents. These How-To documents provide detailed explanations of their topics. Some of
the titles include the following:
840 Appendix a Sources of Information
& ``The Linux Installation How-To''
& ``The Linux Hardware How-To''
& ``The Linux Printing How-To''
See Appendix B for a complete list of Linux How-To and mini-How-To site addresses.
These files are located in the /usr/doc/HOWTO directory on your local drive. Most are archived
with gzip to save disk space. To read these or other compressed files, you can use thezless
command.
Many FAQs about Linux topics and GNU programs are shipped with Linux and can be
found in the /usr/info directory.
man Pages
The Linux operating system itself provides plenty of online help via the man command. To
access online help, you can enter man followed by the topic for which you want information.
Magazines
Linux Journal is the leading U.S. periodical devoted explicitly to Linux. You can request more
information about this publication from the following address:
Linux Journal
P.O. Box 85867
Seattle, WA 98145
206-527-3385
http://www.linuxjournal.com
Linux MagazineÐThe Chronicle of the Revolution
Linux Magazine
(510) 665-7847
[email protected]
http://www.linux-mag.com/index.html
Linux FTP Sites
You can find a great deal of up-to-date information regarding Linux on the Internet. Table
A.2 lists the FTP sites that maintain Linux archives. The main archive site, located at the
University of North Carolina-Chapel Hill, is named sunsite.unc.edu.
841
For Linux Developers
Table A.2 FTP Sites with Linux Archives
Site Name Directory
tsx-11.mit.edu /pub/linux
sunsite.unc.edu /pub/Linux
nic.funet.fi /pub/Linux
ftp.mcc.ac.uk /pub/linux
ftp.dfv.rwth-aachen.de /pub/linux
ftp.informatik.rwth-aachen.de /pub/Linux
ftp.ibp.fr /pub/linux
kirk.bond.edu.au /pub/OS/Linux
ftp.uu.net /systems/unix/linux
wuarchive.wustl.edu /systems/linux
ftp.win.tue.nl /pub/linux
ftp.stack.nl /pub/Linux
ftp.ibr.cs.tu-bs.de /pub/os/linux
ftp.denet.dk /pub/OS/Linux
? See ``Using ftp for Remote File Transfer,'' p. 686
For Linux Developers
So you think that Linux is the greatest thing to come along in quite some time, and you want
to help develop future releases? Well, you're in luck. An active set of mailing lists on the
Internet is devoted to various topics and issues surrounding Linux development. This set is a
multichannel mailing list, meaning that messages on different topics are sent to different
groups of people. You must subscribe to each channel that you're interested in. If you think
Part
you want to get involved in a Linux development project, you can get more information by
sending an email message to
VIII
[email protected] App
with lists in the body to get a list of the lists there. You can add a line with help in the body
A
to get the standard Majordomo help file, which has instructions for subscribing and
unsubscribing.
APPENDIX
B
The Linux How-To Index
In this appendix
What Are Linux How-Tos? 844
Where Can I Get Linux How-Tos? 844
How-To Index 845
Writing and Submitting a How-To 858
Copyright 859
844 Appendix b The Linux How-To Index
What Are Linux How-Tos?
This appendix contains an index to the Linux How-Tos and mini-How-Tos, as well as other
information about the How-To project. Linux How-Tos are documents that describe in
detail certain aspects of configuring or using Linux. For example, the ``Installation How-To''
gives instructions on installing Linux, and the ``Mail How-To'' describes how to set up and
configure mail under Linux. Other examples include the ``NET-3 How-To'' and the
``Printing How-To.''
How-Tos are comprehensive documents, much like a FAQ, but generally not in
question-and-answer format. However, many How-Tos contain a FAQ section at the end.
Several How-To formats are available: plain text, PostScript, DVI, and HTML.
In addition to the How-Tos, you can find a multitude of mini-How-Tos on short, specific
subjects. They are available only in plain text and HTML format.
Where Can I Get Linux How-Tos?
You can retrieve How-Tos via anonymous FTP from the following sites:
ftp://metalab.unc.edu/pub/Linux/docs/HOWTO
ftp://tsx-11.mit.edu/pub/linux/docs/HOWTO
You also can retrieve them from many mirror sites, such as the following:
ftp://metalab.unc.edu/pub/Linux/MIRRORS.html
You can browse How-Tos in HTML format from http://metalab.unc.edu/LDP/HOWTO/
on the World Wide Web. Many mirror sites, such as http://metalab.unc.edu/LDP/
mirrors.html, also mirror the HTML files. Because metalab.unc.edu is heavily used,
try to use a mirror site if possible.
Near the beginning of the month, How-Tos are also posted to the Usenet newsgroup
comp.os.linux.answers. A tool called NewstoHOWTO assembles the postings.
How-To Translations
How-To translations are available on metalab.unc.edu and mirrors around the world. So far,
the following are available:
Chinese ( zh) Croatian ( hr)
French ( fr) German ( de)
Hellenic ( el) Indonesian ( id)
Italian ( it) Japanese ( ja)
Korean ( ko) Polish ( pl)
845
How-To Index
Slovenian ( sl) Spanish ( es)
Swedish ( sv) Turkish ( tr)
If you know of any other translation projects, please let me know, and I will add them to this
list. If you are interested in getting your translations archived on metalab.unc.edu, please
read the directory structure specification at the following Web site and get in touch with me:
http://metalab.unc.edu/pub/Linux/docs/HOWTO/translations/Directory-Structure
How-To Index
The following Linux How-Tos are currently available:
``3Dfx How-To,'' by Bernd Kreimeier, [email protected]. How to use 3Dfx graphics
accelerator chip support. Updated 6 February 1998.
``AX25 How-To,'' by Terry Dawson, [email protected]. How to
configure AX25 networking for Linux. Updated 17 October 1997.
``Access How-To,'' by Michael De La Rue, [email protected]. How to use
adaptive technology with Linux. Updated 28 March 1997.
``Alpha How-To,'' by David Mosberger, [email protected]. Overview of Alpha
systems and processors. Updated 6 June 1997.
``Assembly How-To,'' by Francois-Ren Rideau, [email protected]. Information on
programming in x86 assembly. Updated 6 June 1999.
``Bash Prompt How-To,'' by Giles Orr, [email protected]. How to create and
control terminal and xterm prompts. Updated 7 January 1999.
``Benchmarking How-To,'' by Andre D. Balsa, [email protected]. How to do
basic benchmarking. Updated 15 August 1997.
``Beowulf How-To,'' by Jacek Radajewski, [email protected], and Douglas Eadline,
plogic.com. Introduces the Beowulf Supercomputer architecture and provides
background information on parallel programming. Updated 22 November 1998.
``BootPrompt How-To,'' by Paul Gortmaker, [email protected]. List of
boot time arguments and overview of booting software. Updated 15 May 1999.
``Bootdisk How-To,'' by Tom Fawcett, [email protected]. How to create a
boot/root maintenance disk for Linux. Updated 15 May 1999.
``Busmouse How-To,'' by Chris Bagwell, [email protected]. Information on bus Part
mouse compatibility with Linux. Updated 4 May 1998.
``CD Writing How-To,'' by Winfried Trumper, [email protected]. How to write
VIII
App
CDs. Updated 11 April 1999.
``CDROM How-To,'' by Jeff Tranter, [email protected]. Information on B
CD-ROM drive compatibility for Linux. Updated 24 March 1999.
``Chinese How-To,'' by Chih-Wei Huang, [email protected]. How to
configure Linux for use with the Chinese character set. Updated June 1998.
846 Appendix b The Linux How-To Index
``Commercial How-To,'' by Mr.Poet, [email protected]. Listing of commercial
software products for Linux. Updated 7 March 1999.
``Config How-To,'' by Guido Gonzato, [email protected]. How to fine-tune
and customize your Linux system. Updated 19 January 1999.
``Consultants How-To,'' by Mr. Poet, [email protected]. Listing of Linux
consultants. Updated 19 May 1999.
``Cyrillic How-To,'' by Alexander L. Belikoff, [email protected]. How to configure
Linux for use with the Cyrillic character set. Updated 23 January 1998.
``DNS How-To,'' by Nicolai Langfeldt, [email protected]. How to set up DNS.
Updated 11 February 1999.
``DOS/Win to Linux How-To,'' by Guido Gonzato, [email protected]. How
to move from DOS/Windows to Linux. Updated 22 February 1999.
``DOSEMU How-To,'' by Uwe Bonnes, [email protected]
stadt.de. How-To about the Linux MS-DOS Emulator, DOSEMU. Updated 24
April 1999.
``Danish How-To,'' by Niels Kristian Bech Jensen, [email protected]. How to configure
Linux for use with the Danish character set. Updated 20 February 1999.
``Diskless How-To,'' by Robert Nemkin, [email protected]. How to set up a
diskless Linux box. Updated 13 May 1999.
``Distribution How-To,'' by Eric S. Raymond, [email protected]. A list of
Linux distributions. Updated 8 May 1999.
``Emacspeak How-To,'' by Jim Van Zandt, [email protected]. How to use
``emacspeak'' with Linux. Updated 10 April 1999.
``Esparanto How-To,'' by Wolfram Diestel, [email protected].
How to use Esperanto in general and ISO-8859-3 in particular with Linux. Updated
June 1998.
``Ethernet How-To,'' by Paul Gortmaker, [email protected]. Information
on Ethernet hardware compatibility for Linux. Updated 5 May 1999.
``Finnish How-To,'' by Pekka Taipale, [email protected]. How to configure Linux for use
with the Finnish character set. Updated 14 February 1996.
``Firewall How-To,'' by Mark Grennan, [email protected]. How to set up a firewall
using Linux. Updated 8 November 1996.
``French How-To,'' by Guylhem Aznar, [email protected]. How to
configure Linux for use with the French character set.
``Ftape How-To,'' by Kevin Johnson, [email protected]. Information on ftape drive
compatibility with Linux. Updated August 1998.
``GCC How-To,'' by Daniel Barlow, [email protected]. How to set up the
GNU C compiler and development libraries. Updated 28 February 1996.
847
How-To Index
``German How-To,'' by Winfried Trmper, [email protected]. Information on using
Linux with German-specific features. Updated 19 March 1997.
``Glibc2 How-To,'' by Eric Green, [email protected]. How to install and migrate to
the glibc2 library. Updated 8 February 1998.
``HAM How-To,'' by Terry Dawson, [email protected]. How to
configure amateur radio software for Linux. Updated 1 April 1997.
``How-To Index,'' by Tim Bynum, [email protected]. Index of How-To
documents about Linux. Updated 19 June 1999.
``Hardware Compatibility How-To,'' by Patrick Reijnen,
[email protected]. A list of hardware known to work
with Linux. Updated 20 March 1999.
``Hebrew How-To,'' by Yair G. Rajwan, [email protected]. How to configure
Linux for use with the Hebrew character set. Updated 12 September 1995.
``INFO-SHEET,'' by Michael K. Johnson, [email protected]. Generic intro-
duction to the Linux operating system. Updated 1 September 1998.
``IPCHAINS How-To,'' by Paul Russell, [email protected]. How to
install and configure the enhanced IP firewalling chains software. Updated 12 March
1999.
``IPX How-To,'' by Terry Dawson, [email protected]. How to
install and configure IPX networking. Updated 6 May 1998.
``IR How-To,'' by Werner Heuser, [email protected]. An introduction
to the software provided by the Linux/IR project. Updated 9 February 1999.
``ISP Hookup How-To,'' by Egil Kvaleberg, [email protected]. Basic introduction
to hooking up to an ISP. Updated 5 March 1998.
``Installation How-To,'' by Eric S. Raymond, [email protected]. How to obtain
and install Linux. Updated 20 November 1998.
``Intranet Server How-To,'' by Pramod Karnad, [email protected]. How to set
up a Linux Intranet server. Updated 7 August 1997.
``Italian How-To,'' by Marco ``Gaio'' Gaiarin, [email protected]. How to configure
Linux for use with the Italian character set. Updated 3 November 1998.
``Java-CGI How-To,'' by David H. Silber, [email protected]. How to set up
Java-capable CGI bin. Updated 1 December 1998.
``Kernel How-To,'' by Brian Ward, [email protected]. How to upgrade Part
and compile the Linux kernel. Updated 5 June 1999. VIII
``Keyboard and Console How-To,'' by Andries Brouwer, [email protected]. Information App
about the Linux keyboard, console, and non-ASCII characters. Updated 25 February
1998.
B
848 Appendix b The Linux How-To Index
``KickStart How-To,'' by Martin Hamilton, [email protected]. Briefly describes how
to use the Red Hat Linux KickStart system to rapidly install large numbers of identical
Linux boxes. Updated 11 January 1999.
``LinuxDoc+Emacs+Ispell How-To,'' by Philippe Martin, [email protected]. Assists
writers and translators of Linux How-Tos or any other paper for the Linux
Documentation Project. Updated 27 February 1998.
``META-FAQ,'' by Michael K. Johnson, [email protected]. A listing of Linux
sources of information. Updated 25 October 1997.
``MGR How-To,'' by Vincent Broman, [email protected]. Information on the MGR
graphics interface for Linux. Updated 30 May 1996.
``MILO How-To,'' by David A. Rusling, [email protected]. How to
use the Alpha Linux Miniloader (MILO). Updated 6 December 1996.
``MIPS How-To,'' by Ralf Baechle, [email protected]. Describes the MIPS port of the
Linux operating system, common problems and their solutions, availability, and more.
Updated 31 March 1999.
``Mail How-To,'' by Guylhem Aznar, [email protected]. Information
on electronic mail servers and clients. Updated January 1999.
``Modem How-To,'' by David S. Lawyer, [email protected]. Help with selecting,
connecting, configuring, troubleshooting, and understanding modems for a PC.
Updated May 1999.
``Multi-Disk How-To,'' by Stein Gjoen, [email protected]. How to set up multiple
hard disk drives. Updated 27 May 1999.
``Multicast How-To,'' by Juan-Mariano de Goyeneche, [email protected]. Tries to
cover most aspects related to multicast over TCP/IP networks. Updated 20 March
1998.
``NET-3 How-To,'' by Terry Dawson, [email protected]. How to
configure TCP/IP networking under Linux. Updated August 1998.
``NFS How-To,'' by Nicolai Langfeldt, [email protected]. How to set up NFS
clients and servers. Updated 3 November 1997.
``NIS How-To,'' by Thorsten Kukuk, [email protected]. Information on using NIS/YP
on Linux systems. Updated 9 March 1999.
``Networking Overview How-To,'' by Daniel Lpez Ridruejo, [email protected].
Gives an overview of the networking capabilities of the Linux operating system,
providing pointers for further information and implementation details. Updated 10
July 1998.
``Optical Disk How-To,'' by Skip Rye, [email protected]. How to use optical
disk drives with Linux. Updated 11 December 1998.
``Oracle How-To,'' by Paul Haigh, [email protected]. How to set up Oracle
as a database server. Updated 4 August 1998.
849
How-To Index
``PCI How-To,'' by Michael Will, [email protected]. In-
formation on PCI-architecture compatibility with Linux. Updated 30 March 1997.
``PCMCIA How-To,'' by Dave Hinds, [email protected]. How to install
and use PCMCIA Card Services. Updated 15 May 1999.
``PPP How-To,'' by Robert Hart, [email protected]. Information on using
PPP networking with Linux. Updated 31 March 1997.
``PalmOS How-To,'' by David H. Silber, [email protected]. How to use your Palm
OS device with a Linux system. Updated 20 September 1998.
``Parallel Processing How-To,'' by Hank Dietz, [email protected]. Discussion
of parallel processing approaches for Linux. Updated 5 January 1998.
``Plug and Play How-To,'' by David Lawyer, [email protected]. How to get your Linux
system to support Plug-and-Play. Updated June 1999.
``Polish How-To,'' by Sergiusz Pawlowicz, [email protected]. Information on
using Linux with Polish-specific features. Updated 8 February 1999.
``Portuguese-How-To,'' by Carlos Augusto Moreira dos Santos,
[email protected]. Este documento pretende ser um guia de
referncia de configuraao do Linux e seus programas. ... Updated 24 May 1999.
``PostgreSQL How-To,'' by Al Dev (Alavoor Vasudevan), [email protected]. How
to set up PostgreSQL as a database server. Updated 8 January 1999.
``Printing How-To,'' by Grant Taylor, [email protected]. How-To on
printing software for Linux. Updated 9 March 1999.
``Printing Usage How-To,'' by Mark Komarinski, [email protected]. How to use
the printing system for a variety of file types and options. Updated 6 February 1998.
``Quake How-To,'' by Bob Zimbinski, [email protected], and Thomas Mike Hallock,
medina.net. This document explains how to install, run, and troubleshoot Quake,
QuakeWorld, and Quake II on an Intel Linux system. Updated 30 August 1998.
``RPM How-To,'' by Donnie Barnes, [email protected]. How to use the Red Hat
Package Manager (.rpm). Updated 8 April 1997.
``Reading List How-To,'' by Eric S. Raymond, [email protected]. Interesting
books pertaining to Linux subjects. Updated 20 April 1999.
``Root RAID How-To,'' by Michael A. Robinton, [email protected]. How to create a
root-mounted RAID file system. Updated 25 March 1998.
``SCSI Programming How-To,'' by Heiko Eissfeldt, [email protected]. Part
Information on programming the generic Linux SCSI interface. Updated 7 May 1996. VIII
``SMB How-To,'' by David Wood, [email protected]. How to use the Session App
Message Block (SMB) protocol with Linux. Updated 25 March 1999.
B
``SRM How-To,'' by David Mosberger, [email protected]. How to boot
Linux/Alpha using the SRM firmware. Updated 17 August 1996.
850 Appendix b The Linux How-To Index
``Security How-To,'' by Kevin Fenzi, [email protected]. General overview of security
issues. Updated 25 April 1999.
``Serial How-To,'' by David Lawyer, [email protected]. How to use serial devices
(modems, terminals, and so on) with Linux. Updated May 1999.
``Serial Programming How-To,'' by Peter H. Baumann, [email protected]. How
to use serial ports in programs. Updated 22 January 1998.
``Shadow Password How-To,'' by Michael H. Jackson, [email protected]. How to
obtain, install, and configure shadow passwords. Updated 3 April 1996.
``Slovenian How-To,'' by Primoz Peterlin, [email protected].
Information on using Linux with Slovenian-specific features. Updated 15 February
1999.
``Software Building-How-To,'' by Mendel Leo Cooper, [email protected].
How to build software packages. Updated 16 May 1999.
``Software Release Practice How-To,'' by Eric S. Raymond, [email protected].
Describes good release practices for Linux open-source projects. Updated 18 June
1999.
``Sound How-To,'' by Jeff Tranter, [email protected]. Sound hardware and
software for the Linux operating system. Updated 24 March 1999.
``Sound Playing How-To,'' by Yoo C. Chung, [email protected]. How to
play various sound formats under Linux. Updated 11 August 1998.
``Spanish How-To,'' by Gonzalo Garcia Agullo, [email protected].
Information on using Linux with Spanish-specific features. Updated 20 August 1996.
``TclTk How-To,'' by Luca Rossetti, [email protected]. Describes the Linux approach
to Tcl, a scripting language. Updated 7 November 1998.
``teTeX How-To,'' by Robert Kiesling, [email protected]. How to install the
teTeX package (TeX and LaTeX) under Linux. Updated 9 November 1998.
``Text-Terminal How-To,'' by David S. Lawyer, [email protected]. Explains what text
terminals are, how they work, and how to install and configure them. Updated June
1999.
``Thai How-To,'' by Poonlap Veeratanabutr, [email protected]. How to
configure Linux for use with the Thai character set. Updated 4 August 1998.
``Tips How-To,'' by Paul Anderson, [email protected]. How-To on
miscellaneous tips and tricks for Linux. Updated June 1998.
``UMSDOS How-To,'' by Jacques Gelinas, [email protected]. How to install
and use the UMSDOS file system. Updated 13 November 1995.
``UPS How-To,'' by Harvey J. Stein, [email protected]. Information on using a
UPS power supply with Linux. Updated 18 November 1997.
``UUCP How-To,'' by Guylhem Aznar, [email protected]. Informa-
tion on UUCP software for Linux. Updated 6 February 1998.
851
How-To Index
``Unix and Internet Fundamentals How-To,'' by Eric S. Raymond,
[email protected]. Describes the working basics of PC-class computers,
UNIX-like operating systems, and the Internet in nontechnical language.
Updated 3 December 1998.
``User Group How-To,'' by Kendall Grant Clark, [email protected]. Tips on
founding, maintaining, and growing a Linux user group. Updated 24 April 1998.
``VAR How-To,'' by Mr. Poet, [email protected]. Listing of Linux value-added
resellers. Updated 7 March 1999.
``VME How-To,'' by John Huggins and Michael Wyrick, [email protected]. How to
run Linux on your VMEbus Pentium and other PCI local bus-based VMEbus
processor designs. Updated 30 July 1998.
``VMS to Linux How-To,'' by Guido Gonzato, [email protected]. How to
move from VMS to Linux. Updated 20 April 1998.
``Virtual Services How-To,'' by Brian Ackerman, [email protected]. How to set up
virtual hosting services. Updated 15 August 1998.
``WWW How-To,'' by Wayne Leister, [email protected]. How to set up WWW clients
and servers. Updated 19 November 1997.
``WWW mSQL How-To,'' by Oliver Corff, [email protected]. How to set
up a Web server database with mSQL. Updated 17 September 1997.
``XFree86 How-To,'' by Eric S. Raymond, [email protected]. How to obtain,
install, and configure XFree86 3.2 (X11R6). Updated 12 May 1999.
``XFree86 Video Timings How-To,'' by Eric S. Raymond, [email protected].
How to compose a mode line for XFree86. Updated 13 June 1999.
``X Window User How-To,'' by Ray Brigleb, [email protected]. Information on
configuring the X Window environment for the Linux user. Updated 22 January
1999.
Mini-How-To Index
The following mini-How-Tos are available:
``3 Button Mouse mini-How-To,'' by Geoff Short, [email protected]. How
to configure your mouse to use three buttons. Updated 31 May 1998.
``ADSM Backup mini-How-To,'' by Thomas Koenig,
[email protected]. How to install and use the Part
ADSM backup program. Updated 15 January 1997.
VIII
``Asymmetric Digital Subscriber Loop (ADSL) mini-How-To,'' by David Fannin,
[email protected].
App
B
Addresses ordering, installation, and configuration. Updated 10
April 1999.
``AI-Alife mini-How-To,'' by John A. Eikenberry, [email protected]. Information
about AI software for Linux. Updated 13 January 1998.
852 Appendix b The Linux How-To Index
``Advocacy mini-How-To,'' by Paul L. Rogers, [email protected]. Suggestions
on how to advocate the use of Linux. Updated 7 May 1998.
``Alsa Sound mini-How-To,'' by Valentijn Sessink. Describes the installation of the
Alsa sound drivers for Linux. Updated 18 May 1999.
``Apache SSL PHP/FI frontpage mini-How-To,'' by Marcus Faure, [email protected].
How to build a multipurpose Web server. Updated July 1998.
``Automount mini-How-To,'' by Don, [email protected]. Describes the autofs
automounter, explains how to configure it, and points out some problems to avoid.
Updated 17 April 1999.
``Backup with MSDOS mini-How-To,'' by Christopher Neufeld,
[email protected]. How to back up Linux machines with
MS-DOS. Updated 5 August 1997.
``Battery Powered mini-How-To,'' by Hanno Mueller, [email protected]. How to
reduce a Linux system's power consumption. Updated 21 December 1997.
``Boca mini-How-To,'' by David H Dennis, [email protected]. How to install a
Boca 16-port serial card (Boca 2016). Updated 1 August 1997.
``BogoMips mini-How-To,'' by Wim C.A. van Dorst, [email protected].
Information about BogoMips. Updated 8 February 1999.
``Bridge mini-How-To,'' by Chris Cole, [email protected]. How to set up an
Ethernet bridge. Updated 7 September 1998.
``Bridge+Firewall mini-How-To,'' by Peter Breuer, [email protected]. How to set up
an Ethernet bridge and firewall. Updated 19 December 1997.
``Bzip2 mini-How-To,'' by David Fetter, [email protected]. How to use the new
bzip2 compression program. Updated 29 June 1998.
``Cable Modem mini-How-To,'' by Vladimir Vuksan, [email protected]. How to use a
cable modem with a cable ISP. Updated 9 June 1999.
``Cipe+Masquerading mini-How-To,'' by Anthony Ciaravalo, [email protected]. How to
set up a Virtual Private Network between your LAN and other LANs using cipe
through Linux-masquerading firewall machines. Updated 28 October 1998.
``Clock mini-How-To,'' by Ron Bean, [email protected]. How to set and keep your
clock on time. Updated December 1996.
``Coffee mini-How-To,'' by Georgatos Photis, [email protected]. Thoughts
about making coffee with Linux (humorous). Updated 15 January 1998.
``Colour ls mini-How-To,'' by Thorbjoern Ravn Andersen, [email protected]. How to
set up colors with ls. Updated 7 August 1997.
``Cyrus IMAP mini-How-To,'' by Kevin Mitchell, [email protected]. How to install
the Cyrus IMAP server. Updated 21 January 1998.
``DHCP mini-How-To,'' by Vladimir Vuksan, [email protected]. How to set up a
DHCP server and client. Updated 11 June 1999.
853
How-To Index
``DPT Hardware RAID mini-How-To,'' by Ram Samudrala, [email protected]. How to
configure hardware RAID. Updated 9 April 1999.
``Diald mini-How-To,'' by Harish Pillay, [email protected]. How to use diald to
dial an ISP. Updated 3 June 1996.
``Ext2fs Undeletion mini-How-To,'' by Aaron Crane, [email protected]. How to
retrieve deleted files from an ext2 file system. Updated 2 February 1999.
``Fax Server mini-How-To,'' by Erez Strauss, [email protected]. How to set up a
fax server. Updated 8 November 1997.
``Firewall Piercing mini-How-To,'' by Franois-Ren Rideau, [email protected]. How to
use PPP over Telnet transparently through an Internet firewall. Updated 27
November 1998.
``GIS-GRASS mini-How-To,'' by David A. Hastings, [email protected]. How to
install Geographic Information System (GIS) software. Updated 13 November 1997.
``GTEK BBS-550 mini-How-To,'' by Wajihuddin Ahmed, [email protected].
How to set up the GTEK BBS-550 multiport board with Linux. Updated 20 August
1997.
``Hard Disk Upgrade mini-How-To,'' by Yves Bellefeuille, [email protected]. How to
copy a Linux system from one hard disk to another. Updated 31 January 1998.
``IO Port Programming mini-How-To,'' by Riku Saikkonen, [email protected].
How to use I/O ports in C programs. Updated 28 December 1997.
``IP Alias mini-How-To,'' by Harish Pillay, [email protected]. How to use IP
aliasing. Updated 13 January 1997.
``IP Masquerade mini-How-To,'' by Ambrose Au, [email protected]. How to use
IP masquerading. Updated 7 February 1999.
``IP Subnetworking mini-How-To,'' by Robert Hart, [email protected]. Why
and how to subnetwork an IP network. Updated 31 March 1997.
``ISP Connectivity mini-How-To,'' by Michael Strates, [email protected]. How
to get mail and news over a dial-up connection. Updated 6 November 1997.
``Install from ZIP mini-How-To,'' by Kevin Snively, [email protected]. How to
install Linux from a parallel port zip drive. Updated 29 April 1998.
``Kerneld mini-How-To,'' by Henrik Storner, [email protected]. How to use
kerneld (dynamic module loading). Updated 19 July 1997.
``LBX mini-How-To,'' by Paul D. Smith, [email protected]. How to use Low- Part
Bandwidth X (LBX). Updated 11 December 1997. VIII
``LILO mini-How-To,'' by Alessandro Rubini, [email protected]. Examples of typical App
LILO installations. Updated 16 August 1998.
B
``Large Disk mini-How-To,'' by Andries Brouwer, [email protected]. How to use disks with
more than 1,024 cylinders. Updated 26 April 1999.
854 Appendix b The Linux How-To Index
``Leased Line mini-How-To,'' by Rob van der Putten, [email protected]. How to
set up leased-line modems. Updated July 1998.
``Linux+DOS+Win95+OS2 mini-How-To,'' by Mike Harlan, [email protected]. How
to use Linux, DOS, OS/2, and Windows 95 together. Updated 11 November 1997.
``Linux+FreeBSD mini-How-To,'' by Niels Kristian Bech Jensen, [email protected]
How to use Linux and FreeBSD together. Updated 15 January 1999.
``Linux+NT-Loader mini-How-To,'' by Bernd Reichert, [email protected].
How to use Linux and the Windows NT boot loader together. Updated 2 September
1997.
``Linux+Win95 mini-How-To,'' by Jonathan Katz, [email protected]. How to use Linux
and Windows 95 together. Updated 26 October 1996.
``Loadlin+Win95 mini-How-To,'' by Chris Fischer, [email protected]. How to
use Linux and Windows 95 together, using loadlin. Updated 13 March 1999.
``Mac Terminal mini-How-To,'' by Robert Kiesling, [email protected]. How
to use an Apple Macintosh as a serial terminal. Updated 9 November 1997.
``Mail Queue mini-How-To,'' by Leif Erlingsson, [email protected]. How to queue
remote mail and deliver local mail. Updated 3 September 1997.
``Mail2News mini-How-To,'' by Robert Hart, [email protected]. How to set up a
mail-to-news gateway. Updated 4 November 1996.
``Man Page mini-How-To,'' by Jens Schweikhardt, [email protected]. How to
write man pages. Updated July 1998.
``Modules mini-How-To,'' by Riley H. Williams, [email protected]. How to set up
and configure kernel modules. Updated 14 November 1997.
``Multiboot using LILO mini-How-To,'' by Renzo Zanelli, [email protected].
How to multiboot between Windows 95, Windows NT, and Linux. Updated 26
March 1998.
``NCD X Terminal mini-How-To,'' by Ian Hodge, [email protected]. Describes
how to connect an NCD X terminal to a UNIX host. Updated 3 April 1998.
``NFS-Root mini-How-To,'' by Andreas Kostyrka, [email protected]. How to set up
diskless Linux machines. Updated 8 August 1997.
``NFS-Root-Client mini-How-To,'' by Ofer Maor, [email protected]. How to set up
diskless Linux machines using NFS. Updated 2 February 1999.
``Netrom-Node mini-How-To,'' by Karl Larsen, [email protected]. How to set up the
ax25-utilities package for Amateur Radio such as making Netrom Nodes. Updated 19
October 1998.
``Netscape+Proxy mini-How-To,'' by Sarma Seetamraju, [email protected]. How to set
up a proxy server for Netscape. Updated 15 August 1997.
855
How-To Index
``Netstation mini-How-To,'' by Kris Buytaert, [email protected]. How to
hook up an IBM Netstation to your local network using a Linux box as a server.
Updated 22 February 1998.
``News Leafsite mini-How-To,'' by Florian Kuehnert, [email protected]. How to set up a
leaf news site. Updated 4 January 1998.
``Offline Mailing mini-How-To,'' by Gunther Voet, [email protected]. How to set
up email addresses without a dedicated Internet connection. Updated 4 June 1998.
``PLIP mini-How-To,'' by Andrea Controzzi, [email protected]. How to
set up PLIP (Parallel Line Interface Protocol). Updated 12 March 1998.
``Partition mini-How-To,'' by Kristian Koehntopp, [email protected]. How to
choose disk partitions. Updated 3 November 1997.
``Partition Rescue mini-How-To,'' by Rolf Klausen, [email protected]. How to
rescue deleted Linux partitions. Updated 22 October 1997.
``Path mini-How-To,'' by Esa Turtiainen, [email protected]. How to use the PATH
environment variable. Updated 15 November 1997.
``Pre-installation Checklist mini-How-To,'' by S. Parthasarathy, algo-
[email protected]. Preinstallation checklist and questionnaire. Updated 29 August
1998.
``Process Accounting mini-How-To,'' by Albert M.C. Tam, [email protected]. How to
set up process accounting. Updated 8 August 1997.
``Proxy ARP Subnet mini-How-To,'' by Bob Edwards, [email protected]. How
to use proxy ARP with subnetting. Updated August 1997.
``Public Web Browser mini-How-To,'' by Donald B. Marti Jr., [email protected].
How to set up a guest account to use a WWW browser. Updated 5 January 1998.
``Qmail+MH mini-How-To,'' by Christopher Richardson, [email protected].
How to install qmail and MH. Updated 5 March 1998.
``Quota mini-How-To,'' by Albert M.C. Tam, [email protected]. How to set up disk
quotas. Updated 8 August 1997.
``RCS mini-How-To,'' by Robert Kiesling, [email protected]. How to use RCS
(Revision Control System). Updated 14 August 1997.
``RPM+Slackware mini-How-To,'' by Dave Whitinger, [email protected]. How to
install the Red Hat Package Manager (RPM) under Slackware. Updated 13 April
1998. Part
``RedHat CD mini-How-To,'' by Morten Kjeldgaard, [email protected], and Peter von VIII
der Ah, [email protected]. How to make your own CDs from the Red Hat
App
Linux distribution equivalent to the ones commercially available from Red Hat.
Updated 9 September 1998.
B
``Remote Boot mini-How-To,'' by Marc Vuilleumier Stckelberg, Marc.Vuilleu-
[email protected]. How to set up a server-based boot selector. Updated
February 1999.
856 Appendix b The Linux How-To Index
``Remote X Apps mini-How-To,'' by Vincent Zweije, [email protected]. How to run
remote X applications. Updated 14 July 1998.
``SLIP-PPP Emulator mini-How-To,'' by Irish, [email protected]. How to use
SLIP-PPP emulators with Linux. Updated 7 August 1997.
``Sendmail Address Rewrite mini-How-To,'' by Thomas Roessler, [email protected].
How to set up sendmail's configuration file for the home user's dial-up access.
Updated 6 May 1998.
``Sendmail+UUCP mini-How-To,'' by Jamal Hadi Salim, [email protected]. How to
use sendmail and UUCP together. Updated August 1998.
``Secure POP via SSH mini-How-To,'' by Manish Singh, [email protected]. How to set
up secure POP connections using ssh. Updated 30 September 1998.
``Small Memory mini-How-To,'' by Todd Burgess, [email protected]. How to
run Linux on a system with a small amount of memory. Update 29 October 1997.
``Software RAID mini-How-To,'' by Linas Vepstas, [email protected]. How to configure
software RAID. Updated 21 November 1998.
``Soundblaster AWE mini-How-To,'' by Marcus Brinkmann, Marcus.Brinkmann@-
ruhr-uni-bochum.de. How to install the Sound Blaster AWE 32/64. Updated 11
January 1998.
``StarOffice mini-How-To,'' by Matthew Borowski, [email protected]. Information on
installing the StarOffice suite. Updated 2 June 1998.
``Term Firewall mini-How-To,'' by Barak Pearlmutter, [email protected]. How to use
term over a firewall. Updated 15 July 1997.
``TkRat mini-How-To,'' by Dave Whitinger, [email protected]. How to install
and use the TkRat mail program. Updated 2 February 1998.
``Token Ring mini-How-To,'' by Mike Eckhoff, [email protected]. How
to use token ring cards. Updated 7 January 1998.
``Ultra-DMA mini-How-To,'' by Brion Vibber, [email protected]. How to use Ultra-
DMA drives and controllers. Updated 27 May 1999.
``Update mini-How-To,'' by Stein Gjoen, [email protected]. How to stay updated
about Linux development. Updated 3 February 1998.
``Upgrade mini-How-To,'' by Greg Louis, [email protected]. How to
upgrade your Linux distribution. Updated 6 June 1996.
``VAIO mini-How-To,'' by Hideki Saito, [email protected]. Explains installation
of Linux on Sony VAIO computers. Updated 16 September 1998.
``Vesafb mini-How-To,'' by Alex Buell, [email protected]. How to
use the vesafb device. Updated 2 August 1998.
``VPN mini-How-To,'' by rpd Magosnyi, [email protected]. How to set up a
VPN (Virtual Private Network). Updated 7 August 1997.
857
How-To Index
``Visual Bell mini-How-To,'' by Alessandro Rubini, [email protected]. How to disable
audible bells and enable visual bells. Updated 11 November 1997.
``Windows Modem Sharing mini-How-To,'' by Friedemann Baitinger,
[email protected]. How to set up Windows to use a shared modem on a Linux
machine. Updated 2 November 1997.
``WordPerfect mini-How-To,'' by Wade Hampton, [email protected]. How to
set up WordPerfect for Linux. Updated 13 August 1997.
``X Big Cursor mini-How-To,'' by Joerg Schneider, [email protected]. How to use
enlarged cursors with X Windows. Updated 11 August 1997.
``XFree86-XInside mini-How-To,'' by Marco Melgazzi, [email protected]. How to
convert XFree86 to XInside modelines. Updated September 1997.
``xterm Title mini-How-To,'' by Ric Lister, [email protected]. How to put
strings into the title bar of an xterm. Updated 7 January 1998.
``ZIP Install mini-How-To,'' by John Wiggins, [email protected]. How to
install Linux onto a zip drive. Updated 10 January 1999.
``ZIP Drive mini-How-To,'' by Kyle Dansie, [email protected]. Provides a quick
reference guide on setting up and using the Iomega zip drive with Linux. Updated 10
January 1999.
Special How-To Index
``The High Availability How-To,'' by Harald Milz ( [email protected]) is available at the
following site:
http://metalab.unc.edu/pub/Linux/ALPHA/linux-ha/High-Availability-HOWTO.html
It is not included with the How-To collection because it relies on figures and cannot be
distributed in all supported formats.
``The Graphics mini-How-To,'' by Michael J. Hammel, ( [email protected]) is
available at the following Web site:
http://www.graphics-muse.org/linux/lgh.html
It is not included with the How-To collection because it needs to use a lot of images that
don't translate to other output formats.
Unmaintained How-Tos and Mini-How-Tos
Part
You can find a number of unmaintained documents at
docs/HOWTO/unmaintained.
http://metalab.unc.edu/pub/Linux/
VIII
App
B
They are kept around because old documentation is sometimes better than none. However,
you should be aware that you are reading old documentation.
858 Appendix b The Linux How-To Index
Writing and Submitting a How-To
If you are interested in writing a How-To or mini-How-To, please get in touch with me first
The following are a few guidelines that you should follow when writing a How-To or
mini-How-To:
& Try to use meaningful structure and organization, and write clearly. Remember that
many of the people reading How-Tos do not speak English as their first language.
& If you are writing a How-To, you must use the SGML-Tools package, available from
http://www.pobox.com/~cg/sgmltools, to format the How-To. This package enables
you to produce LaTeX (for DVI and PostScript), plain text, and HTML from a single
source document, and was designed specifically for the How-Tos. Using this package
also gives all the How-Tos a uniform look. It is very important that you format and
review the output of the formatting in PostScript, plain text, and HTML.
& If you are writing a mini-How-To, use either SGML (as described earlier) or HTML.
If you use SGML for your mini-How-To, it will be published along with the
How-Tos in Linux Documentation Project (LDP) books.
& Make sure that all the information is correct. I can't stress this point enough. When in
doubt, speculate but make it clear that you're only guessing.
& Make sure that you are covering the most recent version of the available software.
Also, be sure to include full instructions on where software can be downloaded from
(FTP site name, full pathname), and the current version number and release date of
the software.
& Include a FAQ section at the end, if appropriate. Many How-To documents need a
``FAQ'' or ``Common Problems'' section to cover information that can't be covered in
the regular text.
& Use other How-Tos or mini-How-Tos as models. The SGML source to the How-
Tos is available on Linux FTP sites. In addition, take a look at the LDP Style Guide
for some guidelines. Make sure that your name, email address, date, and a version
number appear near the beginning of the document. You also can include WWW
addresses and a snail mail address if you want. The standard header is
Title
Author's name and email address
Version number and date
For example:
The Linux How-To Index
by Tim Bynum
v2.10.29, 31 July 1997
& Be prepared to receive questions and comments about your writing. Several hundred
people access the How-To collection every day from around the world.
859
Copyright
After you have written the How-To, you can mail it to Tim Bynum at
[email protected]. If you used SGML-Tools, simply mail Tim the SGML
source; he'll take care of formatting the documents. He'll also take care of archiving the
How-Tos on metalab.unc.edu and posting them to the various newsgroups.
It is important that you go through Tim when submitting a How-To because he maintains
the archives and needs to keep track of what How-Tos are being written and who is doing
what.
All you have to do is send him periodic updates whenever appropriate.
Copyright
Copyright # 1995 71998 by Tim Bynum.
Unless otherwise stated, Linux How-To documents are copyrighted by their respective
authors. Linux How-To documents may be reproduced and distributed in whole or in part, in
any medium physical or electronic, as long as this copyright notice is retained on all copies.
Commercial redistribution is allowed and encouraged; however, the author would like to be
notified of any such distributions.
All translations, derivative works, or aggregate works incorporating any Linux How-To
documents must be covered under this copyright notice. That is, you may not produce a
derivative work from a How-To and impose additional restrictions on its distribution.
Exceptions to these rules may be granted under certain conditions; please contact the Linux
How-To coordinator at the address given below.
In short, we wish to promote dissemination of this information through as many channels as
possible. However, we do wish to retain copyright on the How-To documents and would like
to be notified of any plans to redistribute the How-Tos.
If you have questions, please contact Tim Bynum, the Linux How-To coordinator, at
[email protected] via email.
Part
VIII
App
B
APPENDIX
C
The GNU General Public License
862 Appendix c The GNU General Public License
What exactly is GNU? Many believe GNU software is public domain, and some believe
GNU software is shareware. Neither is true. Basically, GNU software is scopyrighted
software that authors have granted the permission to distribute under certain conditions.
Those conditions include the provision to provide source code and that no part of the
software may be placed under a copyright that restricts the further distribution of the
software; that is, you can't use source code copyrighted under the GNU License within your
program without making your source code freely available.
Although the GNU copyright specifies that you must make your source code available, it
doesn't mean you have to give your program away for free; you can charge a fee for your
program, but that fee must include the source code for both the GNU portions and your
portion. You can't charge a fee for the executable part of the program and then another fee
for the source code; you must charge one price for both. Thus, you can't withhold or charge
extra for what you might consider proprietary source code. This is the main objection many
software executives have with using GNU software within their programs; they don't want to
make their source code available to their competitors.
But the concept of GNU goes further, and perhaps the best source to explain this concept is
Richard Stallman, the patriarch of the GNU philosophy. Stallman is a founder and
proponent of the Free Software Foundation (FSF). He believes very strongly that all software
should be free and that computer systems should be open for use by anyone. The fact that
programs such as Linux and Emacs are freely available matches his philosophy. Anyone can
take them for his or her own use. Users are also encouraged to make modifications and share
those changes with others.
The GNU License is sometimes referred to as the GNU copyleft, as a play on the word
copyright. GNU is also a play on wordsÐGNU's Not UNIX. For more information on
the GNU copyleft, see http://www.fsf.org/copyleft/copyleft.html.
What does all this information have to do with Linux? Well, the various components of
Linux are distributed under GNU's General Public License. Thus, Linux is neither in the
public domain, nor is it shareware; Linus Torvalds and the others retain copyright to their
work under the GPL. The rest of this appendix is the GPL as published by the Free Software
Foundation.
ON THE WEB
The Web site for the Free Software Foundation is
http://www.fsf.org
863
Preamble
The GNU License Part
Version 2, June 1991 VIII
Copyright# 1989, 1991 Free Software Foundation, Inc., 675 Mass. Ave, Cambridge, MA App
02139 USA C
Everyone is permitted to copy and distribute verbatim copies of this license document, but
changing it is not allowed.
Preamble
The licenses for most software are designed to take away your freedom to share and change it.
By contrast, the GNU General Public License is intended to guarantee your freedom to share
and change free softwareÐto make sure the software is free for all its users. This General
Public License applies to most of the Free Software Foundation's software and to any other
program whose authors commit to using it. (Some other Free Software Foundation software
is covered by the GNU Library General Public License instead.) You can apply it to your
programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public
Licenses are designed to make sure that you have the freedom to distribute copies of free
software (and charge for this service if you wish), that you receive source code or can get it if
you want it, that you can change the software or use pieces of it in new free programs; and
that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these
rights or to ask you to surrender the rights. These restrictions translate to certain
responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must
give the recipients all the rights that you have. You must make sure that they, too, receive or
can get the source code. And you must show them these terms so they know their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this
license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone
understands that there is no warranty for this free software. If the software is modified by
someone else and passed on, we want its recipients to know that what they have is not the
original, so that any problems introduced by others will not reflect on the original authors'
reputations.
Finally, any free program is threatened constantly by software patents. We wish to avoid the
danger that redistributors of a free program will individually obtain patent licenses, in effect
making the program proprietary. To prevent this, we have made it clear that any patent must
be licensed for everyone's free use or not licensed at all.
864 Appendix c The GNU General Public License
The precise terms and conditions for copying, distribution and modification follow.
GNU General Public License Terms and
Conditions for Copying, Distribution, and
Modification
This license applies to any program or other work which contains a notice placed by the
copyright holder saying it may be distributed under the terms of this General Public License.
The ``Program,'' below, refers to any such program or work, and a ``work based on the
Program'' means either the Program or any derivative work under copyright law: that is to
say, a work containing the Program or a portion of it, either verbatim or with modifications
and/or translated into another language. (Hereinafter, translation is included without
limitation in the term ``modification.'') Each licensee is addressed as ``you.''
Activities other than copying, distribution and modification are not covered by this License;
they are outside its scope. The act of running the Program is not restricted, and the output
from the Program is covered only if its contents constitute a work based on the Program
(independent of having been made by running the Program). Whether that is true depends on
what the Program does.
1. You may copy and distribute verbatim copies of the Program's source code as you
receive it, in any medium, provided that you conspicuously and appropriately publish
on each copy an appropriate copyright notice and disclaimer of warranty; keep intact
all the notices that refer to this License and to the absence of any warranty; and give
any other recipients of the Program a copy of this License along with the Program.
2. You may charge a fee for the physical act of transferring a copy, and you may at your
option offer warranty protection in exchange for a fee.
You may modify your copy or copies of the Program or any portion of it, thus forming
a work based on the Program, and copy and distribute such modifications or work
under the terms of Section 1 above, provided that you also meet all of these
conditions:
. You must cause the modified files to carry prominent notices stating that you
changed the files and the date of any change.
. You must cause any work that you distribute or publish, that in whole or in
part contains or is derived from the Program or any part thereof, to be
licensed as a whole at no charge to all third parties under the terms of this
License.
. If the modified program normally reads commands interactively when run,
you must cause it, when started running for such interactive use in the most
ordinary way, to print or display an announcement including an appropriate
copyright notice and a notice that there is no warranty (or else, saying that
you provide a warranty) and that users may redistribute the program under
865
GNU General Public License Terms and Conditions for Copying, Distribution, and Modification
these conditions, and telling the user how to view a copy of this License.
Part
(Exception: if the Program itself is interactive but does not normally print such an
announcement, your work based on the Program is not required to print an
announcement.)
VIII
App
C
These requirements apply to the modified work as a whole. If identifiable sections of that
work are not derived from the Program, and can be reasonably considered independent and
separate works in themselves, then this License, and its terms, do not apply to those sections
when you distribute them as separate works. But when you distribute the same sections as
part of a whole which is a work based on the Program, the distribution of the whole must be
on the terms of this License, whose permissions for other licensees extend to the entire whole,
and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest your rights to work written
entirely by you; rather, the intent is to exercise the right to control the distribution of
derivative or collective works based on the Program.
In addition, mere aggregation of another work not based on the Program with the Program
(or with a work based on the Program) on a volume of a storage or distribution medium does
not bring the other work under the scope of this License.
3. You may copy and distribute the Program (or a work based on it, under Section 2) in
object code or executable form under the terms of Sections 1 and 2 above, provided
that you also do one of the following:
. Accompany it with the complete corresponding machine-readable source
code, which must be distributed under the terms of Sections 1 and 2 above on
a medium customarily used for software interchange; or,
. Accompany it with a written offer, valid for at least three years, to give any
third party, for a charge no more than your cost of physically performing
source distribution, a complete machine-readable copy of the corresponding
source code, to be distributed under the terms of Sections 1 and 2 above on a
medium customarily used for software interchange; or,
. Accompany it with the information you received as to the offer to distribute
corresponding source code. (This alternative is allowed only for noncom-
mercial distribution and only if you received the program in object code or
executable form with such an offer, in accord with Subsection b above.)
The source code for a work means the preferred form of the work for making
modifications to it. For an executable work, complete source code means all the source
code for all modules it contains, plus any associated interface definition files, plus the
scripts used to control compilation and installation of the executable. However, as a
special exception, the source code distributed need not include anything that is
normally distributed (in either source or binary form) with the major components
(compiler, kernel, and so on) of the operating system on which the executable runs,
unless that component itself accompanies the executable.
866 Appendix c The GNU General Public License
If distribution of executable or object code is made by offering access to copy from a
designated place, then offering equivalent access to copy the source code from the
same place counts as distribution of the source code, even though third parties are not
compelled to copy the source along with the object code.[/nl_list]
4. You may not copy, modify, sublicense, or distribute the Program except as expressly
provided under this License. Any attempt otherwise to copy, modify, sublicense or
distribute the Program is void, and will automatically terminate your rights under this
License. However, parties who have received copies, or rights, from you under this
License will not have their licenses terminated so long as such parties remain in full
compliance.
5. You are not required to accept this License, since you have not signed it. However,
nothing else grants you permission to modify or distribute the Program or its
derivative works. These actions are prohibited by law if you do not accept this
License. Therefore, by modifying or distributing the Program (or any work based on
the Program), you indicate your acceptance of this License to do so, and all its terms
and conditions for copying, distributing or modifying the Program or works based
on it.
6. Each time you redistribute the Program (or any work based on the Program), the
recipient automatically receives a license from the original licensor to copy, distribute
or modify the Program subject to these terms and conditions. You may not impose
any further restrictions on the recipients' exercise of the rights granted herein. You are
not responsible for enforcing compliance by third parties to this License.
7. If, as a consequence of a court judgment or allegation of patent infringement or for
any other reason (not limited to patent issues), conditions are imposed on you
(whether by court order, agreement or otherwise) that contradict the conditions of
this License, they do not excuse you from the conditions of this License. If you cannot
distribute so as to satisfy simultaneously your obligations under this License and any
other pertinent obligations, then as a consequence you may not distribute the
Program at all. For example, if a patent license would not permit royalty-free
redistribution of the Program by all those who receive copies directly or indirectly
through you, then the only way you could satisfy both it and this License would be to
refrain entirely from distribution of the Program.
If any portion of this section is held invalid or unenforceable under any particular
circumstance, the balance of the section is intended to apply and the section as a whole
is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any patents or other
property right claims or to contest validity of any such claims; this section has the sole
purpose of protecting the integrity of the free software distribution system, which is
implemented by public license practices. Many people have made generous
contributions to the wide range of software distributed through that system in reliance
on consistent application of that system; it is up to the author/donor to decide if he or
867
GNU General Public License Terms and Conditions for Copying, Distribution, and Modification
she is willing to distribute software through any other system, and a licensee cannot impose
Part
that choice.
This section is intended to make thoroughly clear what is believed to be a VIII
consequence of the rest of this License.
App
8. If the distribution and/or use of the Program is restricted in certain countries either by
patents or by copyrighted interfaces, the original copyright holder who places the
C
Program under this License may add an explicit geographical distribution limitation
excluding those countries, so that distribution is permitted only in or among countries
not thus excluded. In such case, this License incorporates the limitation as if written in
the body of this License.
9. The Free Software Foundation may publish revised and/or new versions of the
General Public License from time to time. Such new versions will be similar in spirit
to the present version, but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Program specifies a
version number of this License which applies to it and ``any later version,'' you have
the option of following the terms and conditions either of that version or of any later
version published by the Free Software Foundation. If the Program does not specify a
version number of this License, you may choose any version ever published by the
Free Software Foundation.
10. If you wish to incorporate parts of the Program into other free programs whose
distribution conditions are different, write to the author to ask for permission. For
software which is copyrighted by the Free Software Foundation, write to the Free
Software Foundation; we sometimes make exceptions for this. Our decision will be
guided by the two goals of preserving the free status of all derivatives of our free
software and of promoting the sharing and reuse of software generally.
NO WARRANTY
11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS
NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING
THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE
PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER
EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND
PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL
NECESSARY SERVICING, REPAIR OR CORRECTION.
12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED
TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER
PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS
PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING
ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DA-
868 Appendix c The GNU General Public License
MAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM
(INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING
RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY
OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN
ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
END OF TERMS AND CONDITIONS
How to Apply These Terms to Your New Programs
If you develop a new program, and you want it to be of the greatest possible use to the public,
the best way to achieve this is to make it free software which everyone can redistribute and
change under these terms.
To do so, attach the following notices to the program. It is safest to attach them to the start of
each source file to most effectively convey the exclusion of warranty; and each file should have
at least the ``copyright'' line and a pointer to where the full notice is found:
<one line to give the program's name and a brief idea of what it does.>
#
Copyright 3519yy <name of author>
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; either version 2 of the License, or (at your option) any later
version.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc.,
675 Mass. Ave, Cambridge MA 02139 USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this when it starts in an
interactive mode:
Gnomovision version 69, Copyright 19yy name of author
Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type ``show w''.
This is free software, and you are welcome to redistribute it
under certain conditions; type ``show c'' for details.
The hypothetical commands show w and show c should show the appropriate parts of the
General Public License. Of course, the commands you use may be called something other
than show w and show c; they could even be mouse-clicks or menu itemsÐwhatever suits your
program.
You should also get your employer (if you work as a programmer) or your school, if any, to
sign a ``copyright disclaimer'' for the program, if necessary. Here is a sample; however, you
need to alter the names:
869
How to Apply These Terms to Your New Programs
Yoyodyne, Inc., hereby disclaims all copyright interest in the program
'Gnomovision' (which makes passes at compilers) written by James Hacker. Part
<signature of Ty Coon> 1 April 1989
VIII
Ty Coon, President of Vice App
This General Public License does not permit incorporating your program into proprietary C
programs. If your program is a subroutine library, you may consider it more useful to permit
linking proprietary applications with the library. If this is what you want to do, use the GNU
Library General Public License instead of this license.
APPENDIX
D
The Open Source Definition
872 Appendix d The Open Source Definition
The Open Source movement exploded into public awareness during 1998 with the popularity
of Linux. This appendix provides the text of the Open Source definition, originally written as
part of the Debian distribution. Again this definition, like the GPL, emphasizes the right of
individual access to the source code for any program released into the Open, but also provides
criteria for releasing a program into the Open Source universe.
ON THE WEB
For more information on the Open Source Movement, check out its Web site:
http://www.opensource.org
And for the complete text of this definition, check out the following:
http://www.opensource.org/osd.html
The Open Source Definition
(Version 1.4)
Open source doesn't just mean access to the source code. The distribution terms of an open-
source program must comply with the following criteria:
1. Free Redistribution
The license may not restrict any party from selling or giving away the software as a
component of an aggregate software distribution containing programs from several
different sources. The license may not require a royalty or other fee for such sale.
(rationale)
2. Source Code
The program must include source code, and must allow distribution in source code as
well as compiled form. Where some form of a product is not distributed with source
code, there must be a well-publicized means of obtaining the source code for no more
than a reasonable reproduction costÐpreferably, downloading via the Internet
without charge. The source code must be the preferred form in which a programmer
would modify the program. Deliberately obfuscated source code is not allowed.
Intermediate forms such as the output of a preprocessor or translator are not allowed.
(rationale)
3. Derived Works
The license must allow modifications and derived works, and must allow them to be
distributed under the same terms as the license of the original software. (rationale)
4. Integrity of The Author's Source Code
The license may restrict source code from being distributed in modified form only if
the license allows the distribution of ``patch files'' with the source code for the purpose
of modifying the program at build time. The license must explicitly permit
distribution of software built from modified source code. The license may require
873
The Open Source Definition
derived works to carry a different name or version number from the original software.
(rationale)
5. No Discrimination Against Persons or Groups
The license must not discriminate against any person or group of persons. (rationale)
6. No Discrimination Against Fields of Endeavor
The license must not restrict anyone from making use of the program in a specific
field of endeavor. For example, it may not restrict the program from being used in a Part
business, or from being used for genetic research. (rationale) VIII
7. Distribution of License
App
The rights attached to the program must apply to all to whom the program is
redistributed without the need for execution of an additional license by those parties.
D
(rationale)
8. License Must Not Be Specific to a Product
The rights attached to the program must not depend on the program's being part of a
particular software distribution. If the program is extracted from that distribution and
used or distributed within the terms of the program's license, all parties to whom the
program is redistributed should have the same rights as those that are granted in
conjunction with the original software distribution. (rationale)
9. License Must Not Contaminate Other Software
The license must not place restrictions on other software that is distributed along with
the licensed software. For example, the license must not insist that all other programs
distributed on the same medium must be open-source software. (rationale)
10. Conforming Licenses and Certification
Any software that uses licenses that are certified conformant to the Open Source
Definition may use the Open Source trademark, as may source code explicitly placed
in the public domain. No other license or software is certified to use the Open Source
trademark.
(The following information is not part of the Open Source Definition, and it may change
from time to time.)
The GNU GPL, the LGPL, the BSD license, the X Consortium license, the Artistic, the
MPL, the QPL, the libpng license, the zlib license, and the IJG JPEG library license are
examples of licenses that we consider conformant to the Open Source Definition.
To have a license reviewed for certification, write [email protected]. We
strongly encourage use of already-certified licenses from the above list, since this allows use
of the Open Source mark without the need for review. Please report misuse of the Open
Source mark to [email protected].
874 Appendix d The Open Source Definition
Change history:
1.0Ðidentical to DFSG, except for addition of MPL and QPL to clause 10.
1.1Ðadded LGPL to clause 10.
1.2Ðadded public-domain to clause 10.
1.3Ðretitled clause 10 and split off the license list, adding material on procedures.
1.4Ðnow explicit about source code requirement for PD software.
1.5Ðallow ``reasonable reproduction cost'' to meet GPL terms.
Bruce Perens wrote the first draft of this document as The Debian Free Software Guidelines,
and refined it using the comments of the Debian developers in a month-long e-mail
conference in June, 1997. He removed the Debian-specific references from the document to
create the Open Source Definition.
Send questions or suggestions about this page to [email protected].
INDEX
Symbols 9 option (gzip PPP (Point-to-Point
command), 176 Protocol), 675-676
& (ampersand) security
operator, 343 A
accounts without
’ (apostrophe), 340 passwords, 279
* (asterisk), 341 a command (vi), 226 command accounts, 280
default accounts, 279
Usenet articles, 748 A News program, 744
group accounts, 280
wildcard character, 335 -a option
\ (backquote), 340 dip command, 664
guest accounts, 279-280
\ (backslash), 49, 354 fdisk command, 452
logins, 278
[ ] (brackets), 336, 341 fsck command, 450
passwd command, 291
{ } (braces), 341 gzip command, 176
root accounts, 284
^ (caret), 341 ipchains command, 646
unused accounts, 279
$ (dollar sign), 341 ls command, 421
SLIP (Serial Line Internet
FTP command, 687 Protocol), 670
mount command, 445
vi command, 226 netstat command, 630
activating swap files, 461
! (exclamation mark), od command, 431
Active Internet Connection
430, 687 ps command, 380
fields (netstat output),
/ (forward slash), 49, ssh command, 699
631-632
341, 408 telnet command, 685
active network connections,
> (greater than sign), 338 touch command, 434
631-633
- (hyphen), 332, 381 wget command, 274
Active UNIX Domain
. (period), 341, 409 A resource records, 802 Sockets fields (netstat
| (pipe character), 227, A/D (analog-to-digital) output), 632-633
369-370 converters, 148 -add option (route
# (pound sign), 734 abort string, 673 command), 626, 628
? (question mark) absolute pathnames, 409 addresses
FTP command, 689 access.conf file, 764-765 email, 733
wildcard character, 336 accessing Usenet IP (Internet Protocol)
" (quotation marks), newsgroups, 711 classes, 604-605
227, 340 account command dynamic, 669-670
; (semicolon), 226, 342 (FTP), 687 masquerading. See
~ (tilde), 227 Account PAM (Pluggable masquerading IP
:! command (vi), 227 Authentication Modules) addresses
0 option (kill module, 287 obtaining, 605
command), 389 accounts. See also passwords static, 667-668
1 option (gzip adding, 250-251, 255-257 private network blocks, 658
command), 176 deleting, 253, 257-258, 291 adduser command, 251
5 verification failure code editing, 290 administration
(RPM), 173 /etc/passwd files, 250 hardware/software issues,
8 option (rlogin group, 253-254 193-194
command), 696 logins, 250, 253 home directories, 254
passwords, 250-252 Linuxconf utility, 255
876
Administration
multiuser concepts, allow directive, 775 parent processes, 766
185-186, 192 ALSA (Advanced Linux protecting resources, 776
centralized-processing Sound Architecture), 146 secure transactions, 768
systems, 186-188 alt.* newsgroups, 828 security
client/server model, 192 AltaVista search engine, 704 CGI scripts, 777-778
distributed-processing ampersand (&), 368 publicly writable spaces,
systems, 188-190, 192 analog-to-digital (A/D) 779-780
peripherals, 195-196 converters, 148 server-side includes, 778
software upgrades, 198 Analyze Kernel symbolic links, 778-779
system administrators, 47, functions, 101 server-side includes,
184, 192 anonymous FTP (File 768-769
defined, 164 Transfer Protocol), #config directive, 770-771
role and duties, 184-185 686-687, 693-695, 706, #echo directive, 770
software installation 782-783 #exec directive, 769
duties, 167 archie, 708-709 #flastmod directive, 770
training, 198-199 group files, 791 #fsize directive, 770
system monitoring, password files, 791 #include directive, 769
196-198 server permissions, 790 starting, 765-766
system setup, 194-195 sessions, 707 apostrophe (’), 340
user accounts, 250 closing, 789 append command
adding, 250-251, directory navigation, (FTP), 687
255-257 784-788 Apple platforms. See
deleting, 253, 257-258 file retrieval, 788 Macintosh PowerPC
/etc/passwd files, 250 opening, 783 platforms
group accounts, 253-254 server response, 784 AppleTalk, 312-314
passwords, 252 support for, 783 Application layer (OSI
Advanced Interactive -ap option (vi), 228 model), 602
Executive (AIX), 27 Apache, 760 Application Starter
Advanced Linux Sound child processes, 766 (KDE), 562
Architecture (ALSA), 146 configurable logging applications. See commands;
Advanced Research Projects CLF (Common Logfile programs
Agency (ARPA), 598 Format), 774 archie, 708-709
advantages of Linux, 17-20 mod_log_config archives
applications, 18 module, 773 backups, 261
educational uses, 19 NCSA compatibility, copying, 268-269
portability, 18 772-773 creating, 267-268
professional/corporate virtual hosts, 774-775 arp commmand, 198
advantages, 18-19 configuring -arp option (ifconfig
AfterStep window manager, access.conf file, 764-765 command), 623
532, 535 basic setup, 761 ARPA (Advanced Research
AgentLog directive, 772 directives, 762 Projects Agency), 598
-ai option (vi), 228 httpd.conf file, 762-764 Arrange Icons command
AIX (Advanced Interactive srm.conf file, 764 (KDE Root menu), 568
Executive), 27 cookies, 771-772 articles (Usenet), 743-752
alias command, 734 customized error conventions, 748
aliases, 733 messages, 780 emoticons, 748
Apache config scripts, 764 debugging, 766-767 flames, 753
sendmail, 822 file hierarchy, 760-761 headers, 750
aliasing commands, 347-348 host-based access control, netiquette, 752-753
ALL address, 604 775-776 news distributions,
-allmulti option (ifconfig .htaccess files, 776-777 746-747
command), 623 indexes, 766
877
booting
posting, 743 crontab command, command-line options,
follow-up articles, 751 345-346 265-266
new articles, 751-752 nohup command, 344 examples, 266-268
reading, 750 monitoring, 379-383 verifying, 261
replying to, 750-751 priorities, setting bandwidth, 743, 614
threads, 750 nice command, 384-385 base system
ascii command (FTP), 687 renice command, 385-386 configuring, 142-143
ash command shell, 240-241 scheduling, 373 installing, 141-142
askcc variable, 735 sending signals to, 389-390 bash shell, 321, 829
asksub variable, 735 starting, 368-369 Bastion Host, 639
assigning variable values, stopping, 386 batch command, 345, 367,
351-353 normal termination, 372-373
asterisk (*), 341 387-388 batch processes, 366
Usenet articles, 748 terminating all BaudRate option (XFree86
wildcard character, 335 processes, 389 pointers), 519
at command, 344-345, 367, unconditional BBSes (Bulletin Board
370-372, 390 termination, 388-389 Systems), 742
AT&T UNIX development, backgrounds bell command (FTP), 687
25-26 GNOME (GNU Network Berkeley Software
-atime option (find Object Model Distribution (BSD), 26
command), 433 Environment), 590 /bin directories (UNIX),
audio. See sound KDE (K Desktop 419, 440
Auth PAM (Pluggable Environment), 572 binary command (FTP), 687
Authentication Modules) continuing after logout, BIND (Berkeley Internet
module, 287 383-384 Domain) server. See DNS
authenticating users, backquote (\), 340 (Domain Name Service)
287-289 backslash (\), 49, 354 bindings, 579
autodetecting backups, 260-263, 287 bit buckets, 414
hardware, 101 archives, 261 Blackbox window
-autoindent option (vi), 228 copying, 268-269 manager, 532
automating commnads, 266 creating, 267-268 block-special devices, 413
-autoprint option (vi), 228 commercial backup bookmarks, 571
AutoRepeat option programs, 262 Bookmarks command (KDE
(XFree86 keyboard), 518 cpio command Root menu), 567
Autostart folder (KDE), 568 advantages/ Boolean expressions,
disadvantages, 268 705-706
command-line boot disks
B options, 268 creating, 44, 66-67
examples, 269 Linux installation, 96-99
B News program, 744
-B option documentation, 260 boot managers. See LILO
cpio command, 268
dump command, 270-272 (Linux Loader)
ipchains command, 647
full, 260 boot options, 95
ls command, 422
incremental, 260 booting
kernel, 297, 301 boot process, 234-240
Backbox window
media types, 260, 263-265 /etc/inittab files, 238-239
manager, 536
planning, 261 init scripts, 237
background processes
scheduling, 260, 262-263 log file entries, 234
creating, 343-344
taper command, 269-270 rc.local files, 237-238
at command, 344-345
tar command rc3.d directories, 236
batch command, 345
advantages/ run levels, 235-236,
cron daemon, 344
disadvantages, 265 238-239
878
booting
shell scripts, 236 gzip command, 176 chains
troubleshooting, 244-247 ipchains command, 646 built-in, 649-650
file system mounting, ls command, 422 defined, 646
446-448 netstat command, 630 user-defined, 650
from floppy disks, 240-241 ps command, 380 Challenge Handshake
LILO (Linux Loader), ssh command, 699 Authentication Protocol
241-243 tar command, 265 (CHAP), 676
advantages/ touch command, 434 changing
disadvantages, 241 C shell (csh), 320, 829 file permissions, 415-416
configuration, 242-243 c verification failure code file time/date stamp, 434
installing, 242 (RPM), 173 run levels, 235
uninstalling, 245 calculators, xcalc, 545 CHAP (Challenge
bootparams map, 477 HP emulation, 547-548 Handshake Authentication
Bourne shell, 320 T1 emulation, 546-547 Protocol), 676
HOME variable, 326, 328 Caldera character-special
LOGNAME variable, OpenLinux, 25, 94 devices, 413
326, 330 WABI (Windows chat program
MAIL variable, 329 Application Binary abort string, 673
PATH variable, 326-329 Interface), 22 command-line options,
PS1 variable, 326, 329 Web site, 94 671-672
PWD variable, 326 canceling email messages, scripts, 672-673
SHELL variable, 326 721-722 chatkey command (dip), 665
TERM variable, 326, 329 carbon copies (email), chfn command, 251
TZ variable, 329 732-733 child processes, 369, 766
braces ({ }), 341 carriage returns, 342 chipsets, 509-510
brackets ([ ]), 336, 341 Cascade Windows chmod command, 281, 349,
breaches of security, 286 command (KDE Root 415-417, 687
bridges, 190, 611, 617 menu), 568 choosing passwords, 278
-broadcast option (ifconfig case command (FTP), 687 ChordMiddle option
command), 623 case structure, 355-356 (XFree86 pointers), 519
browsers, 703 cat command, 429 class field (resource
BSD (Berkeley Software cc: Mail, 810-811 records), 801
Distribution), 26 cd command, 48-49, 687, classes (IP addresses),
BSD386, 26 784, 787 604-605
buffers clean bit, 449
defined, 54 CD-ROMs clear command, 52
vi, 206 installing Linux from, 40,
built-in chains, 649-650 63-64, 96, 131 ClearDTR option (XFree86
bulletin board systems Linux requirements, 37 pointers), 519
(BBSes), 742 mounting, 117 clearing screen, 52
bus topology, 191 CDs, 155 CLF (Common Logfile
buses, 33 playing, 156 Format), 774
bye command (FTP), 687 saving to MP3 format, 156 client/server messaging, 811
signatures, 156 client/server model, 192
troubleshooting, 162 clients, 702
C cdup command (FTP), 687 defined, 506
C command (vi), 226 centralized-processing LDAP (Lightweight
C News program, 744 systems, 186-188 Directory Access
-c option CGI (Command Gateway Protocol), 484
fdisk command, 452
Interface) scripts, 777-778 NIS (Network Information
Service), 476
file command, 410
879
commands
swais, 714 cat, 429 close, 687
Usenet, 829 cd, 48-49 cr, 687
clock chat, 671-673 debug, 687
configuring, 85 chfn, 251 delete, 687
GNOME (GNU Network chmod, 281, 349, 415-417 dir, 687
Object Model clear, 52 disconnect, 687
Environment), 584 compress, 175, 435 exit, 688
KDE (K Desktop cp, 51, 425 form, 688
Environment), 565 cpio, 263, 268-269 ftp, 783
cloning, 766 cron, 367, 373 get, 688, 690, 788
close command (FTP), 687 crontab, 345-346, 367, glob, 688
closing 373-376, 390 hash, 688
FTP (File Transfer dbootstrap, 134 help, 688
Protocol) sessions, dd, 460 idle, 688
690, 789 delimiting, 342 image, 688
mail program, 736-737 depmod, 303 lcd, 688
Samba services, 495 df, 197, 411 ls, 688, 690, 784
vi, 210-212 dip, 664-666 macdef, 688
CNAME resource dpkg, 174 mdelete, 688
records, 802 dump, 263, 270-272 mdir, 688
color echo, 321, 350-351 mget, 688, 691
color highlighting, 333 editing, 347 mkdir, 688
KDE (K Desktop elm, 737-739 mls, 688
Environment), 572 entering, 45 mode, 688
command accounts, 280 env, 327 modtime, 688
command-completion error feedback, 346 mput, 688, 691
feature, 46-47 executing, 332 newer, 688
command groups, 342-343 exit status, 356 nlist, 688
command history, 45-46 fdisk, 72-77, 107-109, 126, nmap, 688
command line, parsing, 332 135-137, 451 ntrans, 688
command mode, command-line options, open, 688
dip driver, 665-666 107, 452-453 passive, 688
vi, 206, 208 starting, 452, 454 prompt, 688
command prompt, 45 file, 410-411 proxy, 688
command substitution, find, 266, 432-434 put, 688, 691
339-340 finger, 378-379 pwd, 688, 784
command-line interpreter. fsck, 450 quit, 688, 789
See shells FTP (File Transfer quote, 688
commands, 48 Protocol) recv, 688
!, 430 !, 687 reget, 688
adduser, 251 $, 687 rename, 689
aliasing, 347-348, 363 ?, 689 reset, 689
archie account, 687 restart, 689
set search, 709 append, 687 rhelp, 689
show search, 708 ascii, 687 rmdir, 689
arp, 198 bell, 687 rstatus, 688
ash command shell, binary, 687 runique, 689
240-241 bye, 687 send, 689
at, 344-345, 367, case, 687 site, 689
370-372, 390 cd, 687, 784, 787 size, 689
automating, 266 cdup, 687 status, 689
batch, 345, 367, 372-373 chmod, 687 struct, 689
880
commands
sunique, 689 taskbar commands, kernel routing table,
system, 689 566-567 printing, 633
tenex, 689 kill, 367, 386 nice, 367, 384-385
tick, 689 nontermination signals, nohup, 344, 367, 383-384
trace, 689 389-390 nslookup, 198
type, 689 normal termination, od, 430-432
umask, 689 387-388 passwd, 252, 291
user, 689 terminating all patch, 297
verbose, 689 processes, 389 ping, 200
grpck, 292 unconditional pppd, 673-675
gzip, 175-177, 435 termination, 388-389 printenv, 325
halt, 54, 245 killall, 387 printtool menu, 403
history, 347 last reboot, 246 ps, 197, 284, 331, 367,
history feature, 347 less, 52, 429-430 379-383
ifconfig, 197, 622 Linuxconf, 255 pwck, 291
command-line options, ln, 413 pwd, 409
622-623 lpc, 395-396 rawrite, 66-67, 99
network interface lpq, 395 rcp, 697-698
configuration, 625 lpr, 395, 726 rdist, 272-273
Parallel IP interface lprm, 395 read, 352
configuration, 625 ls, 49, 281, 332, 414, reboot, 54, 245
software loopback interface 421-424 renice, 367, 385-386
configuration, 624-625 lsmod, 302-303 restore, 272
status reports, 624 mail program, 734 rlogin, 695-696
init, 234-235-237 majordomo, 713-714 rm, 51, 426-428
insmod, 303-304 make, 177, 298 rmdir, 50
ipchains, 646 man, 48, 840 rmmod, 303-304
built-in chains, 649-650 mattrib, 53 rn newsreader, 754-755
command-line options, mcd, 53 route
646-649 mcopy, 53 command-line
ICMP types/subtypes, mdel, 53 options, 626
648-649 mdir, 53 routes, adding/deleting,
IP masquerading, 659 mformat, 53 628, 630
overview, 646-651 mkdir, 50, 424 routing tables,
packet filter firewalls, mkfs, 458-459 viewing, 627
637, 645-654 mkswap, 459 rpm, 168-172
port forwarding, 644 mlabel, 53 rsh, 696-697
TOS masks, 649 mmd, 53 scheduling
user-defined chains, 650 modprobe, 303 approximate times,
ipmasq, 644 more, 52, 429 372-373
ipmasqadm mount, 445, 471 priority, 384-386
firewall network mrd, 53 repeat execution, 373-376
options, 642 mren, 53 specified times, 370-372
forwarding ports, 659 mtype, 53 set, 325
ipportfw wrapper, 645 mv, 51, 426 shutdown, 54-55, 244
ipportfw, 644 netstat, 201 smbclient, 486, 495-496
ispell, 369 active network sndconfig, 150
jtest, 157 connections, listing, ssh, 698-699
KDE (K Desktop 631-633 stty, 324
Environment) command-line options, stty sane, 325
Root menu, 567-568 630-631 su, 283
interface statistics, 634
881
configuring
substituting output, config command (dip), 665 KDE (K Desktop
353-354 #config directive, 770-771 Environment)
swapoff, 461 configurable logging background, 572
swapon, 460 (Apache) bookmarks, 571
swat CLF (Common Logfile colors, 572
command buttons, 497 Format), 774 desktop organization,
running from inetd, 498 mod_log_config 573-574
running from Web, module, 773 fonts, 572
498-499 NCSA compatibility, screensaver, 573
switchdesk, 561 772-773 Templates folder,
taper, 263, 269-270 virtual hosts, 774-775 570-571
tar, 263 Configure Filter dialog themes, 573
advantages/ box, 1 kernel
disadvantages, 265 Configure Mouse dialog interactive text-based
command-line options, box, 82 program, 298
265-266 configuring menu-based program, 300
examples, 266-268 Apache options, 299
tcpdump, 641 access.conf file, 764-765 patch files, 297
telinit, 235 basic setup, 761 source code files, 297
telnet, 684-686 directives, 762 X Window system-based
terminating, 342 httpd.conf file, 762-764 program, 300-301
test, 358-360 srm.conf file, 764 LDAP (Lightweight
top, 196-197 base system, 142-143 Directory Access
touch, 434 clock, 85 Protocol) clients, 484
traceroute, 200-201 device drivers, 322 LILO (Linux Loader),
tty, 413 display managers, 539 242-243
umount, 448 DNS (Domain Name LISA program, 100-101
uptime, 247 Service) resolvers, mail program, 734, 736
useradd, 290 796-798 monitors, 83
userdel, 291 DNS (Domain Name mouse devices, 82
usermod, 290 Service) servers network interfaces, 625
vi, 226-229 named.boot files, 798-800 networks, 84
vmstat, 198 named.ca files, 806-807 bandwidth, 614
w, 367, 457 named.hosts files, bridges, 617
wget, 273-274 803-805 connection media, 614
who, 367, 376-378 named.rev files, 805-806 dial-up connections, 615
whois, 198 GNOME (GNU Network diskless workstations, 615
xf86config, 523 Object Model physical location, 614
zcat, 435 Environment), 587 remote connections, 616
zless, 436 applets, 589 routers, 616
comments, 351 background, 590 switches, 617
Common Logfile Format desktop, 592 TCP/IP, 84, 620-621
(CLF), 774 launchers, 589 NN newsreader, 829
comp newsgroups, 745 Main Menu options, 588 OpenLinux, 112-114
comparing numbers, 359 panel, 589 PAM (Pluggable
compiling kernel, 301 screensavers, 591 Authentication
completing commands, 348 sounds, 591 Modules), 288
Components to Install subpanels, 589 passwords, 87
dialog box, 80 themes, 591-592 PPP (Point-to-Point
compress command, window behavior, Protocol) configuration,
175, 435 590-591 677-679
compressing files, 435-436
882
configuring
printers, 401-402 xf86config command, 523 csh. See C shell
field values, 404 XF86Setup, 515 culture of Usenet
filters, 404-405 connections (network), newsgroups, 748-749
local/remote, 403 612-614 cursors
printtool menu continuing processes, vi, 215-217
commands, 403 383-384 xterm, 543
Samba Control Center (GNOME) customizing
default smb.conf file, (GNU Network Object Apache error
487-491 Model messages, 780
[global] settings, 491-492 Environment), 583 fvwm2 window manager
[homes] settings, 492-493 KDE (K Desktop menu, 553-554
[printers] settings, 493 Environment), 563 preparation, 553
shared direcories, control characters, 323 start-up programs,
493-494 control keys 554-555
testing configuration, 494 device drivers, 323-324 shells, 361-363
sendmail, 818-821 interrupt keys, 324 vi, 228-230
shell environment, 325 kill key, 325 cut/paste operations,
HOME variable, 326, parameters, 324 223-225
328 control structures, 350 cutting text, 348
LOGNAME variable, case, 355-356
326, 330 if, 357-358
MAIL variable, 329 iterative, 360-361
D
PATH variable, 326-329 cooked mode (device D command (vi), 226
PS1 variable, 326, 329 drivers), 323 -d option
PWD variable, 326 cookies, 771-772 fdisk command, 452
SHELL variable, 326 copying gzip command, 176
TERM variable, archives, 268-269
ipchains command,
326, 329 files, 51, 425
646-647
TZ variable, 329 rdist command, 272-273
ls command, 422
sound cards wget command, 273-274
rlogin command, 696
automated configuration Copyleft (GNU), 29 rsh command, 697
programs, 150 copyrights telnet command, 685
hardware GNU General Public
D verification failure code
information, 149 License, 28-29, 862
(RPM), 173
kernel modules, 151-152
D/A (digital-to-analog)
startup services, 85-87 applying to programs,
converters, 148
terminal environment 868-869
daemons, 414
control keys, 324-325 preamble, 863-864
cron, 344
device drivers, 322-323 terms/conditions, 864-868
inetd, 498
TIN newsreader How-To documents, 859
kerneld, 303-305
bash shell, 829 courtney program, 644 lpd, 394-395
C shell, 829 cp command, 51, 425 named
ksh shell, 829 cpio command, 263, named.boot file, 798-800
tcsh shell, 829 268-269 named.ca file, 806-807
window manager, 536-537 CPUs (central processing named.hosts file, 803-805
wu-ftpd, 789-790 units), 33 named.rev file, 805-806
X Window System, 83 cr command (FTP), 687 nmbd, 486
XFree86. See also crackers, 276 pppd, 673, 675
XF86Config file cron command, 344, printer daemons, 393
SuperProbe utility, 367, 373 processes, 366
515-516 crontab command, 345-346, rpc.mountd, 468
Xconfigurator, 514-515 367, 373-376, 390
883
direct variable assignment
rpc.nfsd, 468 delete command (FTP), 687 commands
sendmail, 817-818 deleting netstat, 201
smbd, 486 directories, 50, 426-428 ping, 200
sshd, 238 email, 729 traceroute, 200-201
data field (resource files, 51, 412, 426-428 dial command (dip), 665
records), 801 kernel modules, 304 dial-up connections, 615
Data Link layer (OSI partitions, 129-130 dial-up passwords, 292
model), 601 passwords, 252 dip (Dial-Up IP Protocol)
databases, populating, routes, 630 driver, 664
483-484 text, 220-221 command mode, 665-666
databits command (dip), 665 user accounts, 253-254, command-line options,
datagrams, 599 257-258, 291 664-665
date/time stamp, 434 delimiting commands, 342 diplogin, 670
dbootstrap program, 134 demons. See daemons dynamic IP addresses,
dd command, 460 deny directive, 775 669-670
de.* newsgroups, 828 dependencies, 164 /etc/diphosts file, 670-671
deactivating swap files, 461 depmod command, 303 static IP addresses,
Debian distribution, desktop, KDE (K Desktop 667-668
installing, 120 Environment), 558-559 variables, 667
base system, 141-143 advantages/disadvantages, dialing directory
fdisk command, 135-137 574-575 (Seyon), 549
from CD-ROM, 131 Autostart folder, 568 dialog boxes
from DOS, 131-132 components, 559-560 Components to Install, 80
hard drive partitions configuring, 570-574 Configure Filter, 1
creating, 125, 130, kfm (KDE file manager), Configure Mouse, 82
138-139 568-570 Edit New Partition, 79
deleting, 129-130 packages, 560-561 Kernel Configurator, 303
initializing, 140 panel, 562-566 Loadable Module
naming, 125-126 Qt toolkit, 560 Support, 301
partition tables, 126 Root menu, 567-568 digital-to-analog (D/A)
repartioning, 128 setting as default desktop, converters, 148
requirements, 127-128 561-562 dip command, 664-665
swap partitions, 140 taskbar, 566-567 dip (Dial-Up IP Protocol)
network configuration, 141 Templates folder, 568 driver, 664
preparation, 121-124 trash, 568 command mode, 665-666
requirements, 120-121 Destination field (routing command-line options,
rescue disks, 132-134 tables), 627 664-665
system configuration, 140 /dev directory, 441-442 diplogin, 670
system files, 134-135 /dev/sndstat, 153-154 dynamic IP (Internet
Debian Package development tools, 18 Protocol) addresses,
Management System, 174 device directories, 103 669-670
debug command (FTP), 687 device drivers. See drivers /etc/diphosts file, 670-671
debugging Apache, 766-767 device files, 441 static IP (Internet
DEC Alpha systems, 89-91 Device section (XFree86 Protocol) addresses,
decision structures, file), 521 667-668
350, 355 device-special files, 413 variables, 667
default accounts, 279 df command, 197, 411 diplogin, 670
default command (dip), 665 DHCP (Dynamic Host dir command (FTP),
defunct processes, 383 Configuration 687, 690
deiconified windows, 532 Protocol), 807 direct variable
-del option (route Diablo server, 829 assignment, 352
command), 626 diagnostics. See also
884
directives
directives disk partitions. See named.hosts files,
Apache config files, 762 partitions 803-805
#config, 770-771 disk space requirements, 95 named.rev files, 805-806
#echo, 770 diskless workstations, 615 spoofing, 795
#exec, 769 disks, floppy troubleshooting, 807-808
#flastmod, 770 booting from, 240-241 documentation
#fsize, 770 installing Linux from, backups, 260
#include, 769 63-64 How-To documents,
UserDir, 765 disks, hard. See hard drives 839-840, 844
directories, 418-420, display managers, 538-539 bibliography, 845-851
440-443. See also names of Display Properties copyrights, 859
specific directories command (KDE Root Distribution How-To, 17
Apache menu), 567 High Availability
accessing, 776-777 distributed-processing How-To, 857
indexes, 766 systems, 186-192 mini How-Tos, 851-857
archives bridges, 190 obtaining, 844
copying, 268-269 file servers, 189 submitting, 859
creating, 267-268 gateways, 190 translations, 844-845
changing, 690 hubs, 189 unmaintained
creating, 50, 424 NICs (network interface documents, 857
deleting, 50, 426-428 cards), 189 writing, 858
device, 103 repeaters, 189 XFree86 How-To, 504
home, 254, 408 routers, 190 Linux Documentation
inode numbers, 411-412 topologies, 190-191 Project, 839
LDAP (Lightweight workstations, 189 man pages, 48, 840
Directory Access Distribution How-To, 17 Open Source definition,
Protocol), 481-482 distribution limits, 746-747 872-874
client configuration, 484 DNS (Domain Name dollar sign ($), 341
database population, Service), 599, 794 FTP command, 687
483-484 domain name spaces, 795 vi command, 226
installing, 482-483 domains, 794 domains. See also DNS
navigating, 48-49 hosts, 794 (Domain Name Service)
pathnames, 408-409 name servers, 795 name spaces, 795
printing, 396-397 nodes, 794 names, 794
root, 408, 440 resolvers, 795 NIS (Network Information
shared, 493-494 /etc/host.conf files, Service), 480
UNIX, 418-419 796-797 DontZap flag
user, 765 /etc/resolv.conf files, (X server), 517
viewing contents of, 49 797-798 DontZoom flag
DIR_COLOR values, resource records, 800 (X server), 517
423-424 A, 802 DOS
disabling unused CNAME, 802 drive partitions, 104-107
accounts, 279 fields, 801 file commands, 52-53
disadvantages of Linux HINFO, 802 installing Linux from,
hardware problems, 21 MX, 802 131-132
overcoming, 23-24 name-server, 804 -down option (ifconfig
software problems, 22 NS, 802, 806 command), 623
system administration , 23 PTR, 802, 806 downloading files, 690-691
technical support, 20-21 SOA, 802 dpkg command, 174
disconnect command server configuration drivers, 103
(FTP), 687 named.boot files, 798-800 configuring, 322
Disk Druid, 78-79 named.ca files, 806-807 control characters, 323
885
email (electronic mail)
control keys, 323-324 Edit New Partition dialog canceling, 721-722
cooked mode, 323 box, 79 carbon copies, 732-733
dip (Dial-Up IP editing. See also editors client/server
Protocol), 664 commands, 347 messaging, 811
command mode, 665-666 user accounts, 290 commands, 734
command-line options, editors customizing environment,
664-665 Emacs, 204 734, 736
diplogin, 670 mail program, 722 deleting/undeleting, 729
dynamic IP addresses, vi, 204-205 elm mailer
669-670 adding text, 217-219 commands, 738-739
/etc/diphosts file, 670-671 buffers, 206 starting, 737-738
static IP addresses, changing/replacing text, environment variables, 735
667-668 222-223 EOT (end of transmission)
variables, 667 closing, 210-212 messages, 721
joystick drivers, 157-158 command mode, 206, 208 ESMTP (Extended Simple
raw mode, 323 command summary, Mail Transfer
sound card drivers 226-228 Protocol), 815
ALSA (Advanced Linux creating files, 208-209 forwarding, 731-732
Sound Architecture), cursor-positioning handshaking, 814
146 commands, 215-217 help, 727
comparison, 147-148 cutting/pasting text, historical overview, 810
OSS (Open Sound 223-225 LAN-based messaging,
System), 146 deleting text, 220-221 810-811
drives. See hard drives editing files, 209-210 mailing lists, 711-712,
-dstaddr option (ifconfig editing process, 206-207 733-734
command), 623 environment options, creating, 822-824
dumb terminals, 187 228-230 finding, 712
dump command, 263, input mode, 206, 208 Linux, 841
270-272 overwriting files, 215 LinuxPPC, 312
Dump Frequency field repeating commands, 225 mail reflectors, 712
(/etc/fstab file), 447 saving files, 213-214 majordomo scripts,
dust and system failure, 276 searching, 221 712-714
Dynamic Host searching/replacing subscribing to, 714-716
Configuration Protocol text, 225 MDAs (Mail Delivery
(DHCP), 807 starting, 207 Agents), 811
dynamic IP (Internet status line, 206 message formats, 815-816
Protocol) addresses, troubleshooting, 210, MTAs (Mail Transfer
669-670 221, 231 Agents), 811
undoing changes, MUAs (Mail User
212-213 Agents), 811
E vim, 204 Mutt client, 739-740
e command (vi), 226-227 electrical surges and system printing, 726
-e option failure, 276 quitting, 736-737
crontab command, 376
electronic mail. See email reading, 723-725
ps command, 380
elm replying to, 729-731
commands, 738-739 RFC (Request for
rlogin command, 696
starting, 737-738 Comments) standards,
ssh command, 699
telnet command, 685
Emacs, 204 812-814
eb command (vi), 228 email (electronic mail). See saving, 728-729
also mail program sending, 719-721, 726
echo command, 321, 665,
advantages, 718
#echo directive, 770
aliases, 733
886
email (electronic mail)
sendmail, 811, 816 escape characters, 340 Parallel IP interfaces, 625
aliases, 822 chat program, 673 reporting status of, 624
architecture, 817 xterm, 543 software loopback
configuring, 818-821 escaping to shell, 430 interfaces, 624-625
historical overview, 817 ESMTP (Extended Simple statistics, 634
rulesets, 821-822 Mail Transfer Protocol), ethers.byaddr map, 477
running as daemon, 815 ethers.byname map, 477
817-818 /etc directory (UNIX), ettiquette guidelines
SMTP (Simple Mail 418, 441 (Usenet), 830
Transfer Protocol), 599, /etc/conf.modules file, Excite search engine, 705
720, 814-815 151-152, 305 exclamation point (!), 687
system mailbox, 718 /etc/diphosts file, 670-671 #exec directive, 769
text editors, 722 /etc/exports file, 469-470 -exec option (find
undeliverable, 719 /etc/fstab file, 446-448, command), 433
writing, 721 471-472 exec processes, 331
emoticons, 748 /etc/host.conf file, DNS ExecCGI directive, 778
Emulate3Buttons option (Domain Name Service) executable files, 410
(XFree86 pointers), 519 resolvers, 796-797 Execute Command
end command /etc/hosts file, 620-621 command (KDE Root
(majordomo), 714 /etc/inittab file, menu), 567
end of transmission (EOT) 238-239, 540 executing commands, 332
messages, 721 /etc/lilo.conf file, 302 exit command (FTP), 688
end-of-line key, 323 /etc/networks file, 622 exiting. See closing
Enlightenment menu /etc/passwd file exporting
(GNOME), 585-586 user accounts, 250 NFS (Network File
Enlightenment window shadowed passwords, 289 System), 468-470
manager, 532, 536, 592 /etc/printcap file, 393, variables to shells, 362-363
entering commands, 45 397-400 expressions,
env command, 327 /etc/resolv.conf file, DNS Boolean, 705-706
environment variables, 325 (Domain Name Service) regular
HOME, 326, 328 resolvers, 797-798 special characters,
IFS, 333 /etc/shadow file, 290 341-342
LOGNAME, 326, 330 /etc/X11 files strings, 340
MAIL, 329 /etc/X11/prefdm, 540 ext2 file system, 449
mail program, 735 /etc/X11/qdm/Init/ extended partitions, 455
PATH, 326-329 Default, 540 Extended Simple Mail
PRINTER, 400 /etc/X11/qdm/qdm.conf, Transfer Protocol
PS1, 326, 329 540-541 (ESMTP), 815
PWD, 326 /etc/X11/qdm/Sessions/ extensions (file), 175, 706
SHELL, 326 Default, 540
TERM, 326, 329 /etc/X11/X, 540
TZ, 329 /etc/X11/xdm/Xsession,
F
EOT (end of transmission) 540-541
:f command (vi), 227
messages, 721 /etc/X11/xdm/
-f option
$errlvl variable, 667 Xsetup_0, 540
chat command, 671
error messages /etc/X11/xinit/Xclients,
file command, 410
Apache server, 541
gzip command, 176
766-767, 780 Ethernet ipchains command,
command feedback, 346 controller cards, 37, 194
646-648
unknown PCI device, 245 interfaces
ls command, 422
errorbells option (vi), 228 configuring, 625
mount command, 445
Esc command (vi), 228 initializing, 622
887
files
ps command, 380 types, changing, 457 /etc/host.conf file, DNS
ssh command, 699 verifying, 457-458 (Domain Name Service)
tar command, 265 unmounting, 448-449 resolvers, 796-797
F Virtual window manager. File Transfer Protocol. See /etc/hosts file, 620-621
See fvwm2 window FTP /etc/inittab file
manager File Transfer, Access, and 238-239, 540
FAQs (Frequently Asked Management (FTAM), 599 /etc/lilo.conf file, 302
Questions), 743 filename completion /etc/networks file, 622
fdisk command, 72-77, characters. See wildcards /etc/passwd file, 250, 289
108-109, 126, filenames, 408 /etc/printcap file, 393,
135-137, 451 files. See also directories, 397-400
command-line options, 411, 442 /etc/resolv.conf file, DNS
107, 452-453 Apache, 760-761 (Domain Name Service)
starting, 452-454 access.conf, 764-765 resolvers, 797-798
FIFO (first-in-first-out) httpd.conf, 762-764 /etc/shadow file, 290
buffers, 414 srm.conf, 764 /etc/X11 files, 540-541
file command, 410-411 archives executable, 410
file extensions 175, 706 copying, 268-269 filenames, 408
file manager (kfm), 568-570 creating, 267-268 finding, 266, 432-434
file servers, 189 backing up. See backups FTP (File Transfer
File System Device color highlighting, 333 Protocol) remote
(KDE), 570 compressing, 435-436 transfers
File System Specifier field configuration files, anonymous FTP,
(/etc/fstab file), 446 540-541 686-687, 693-695
file systems, 408, 440 copying, 51, 425 commmands, 687-689
building, 458-459 rdist command, 272-273 directories, changing, 690
checking for damage/ wget command, 273-274 downloads, 690-691
corruption, 450-451 Debian system files, /ftp/etc/group, 791
defined, 43, 66 134-135 /ftp/etc/passwd, 791
directories. See directories deleting, 51, 412, 426-428 remote directory listings,
ext2, 449 determining file type, 410 690
maintenance, 449 /dev/sndstat, 153-154 sample session, 692
Minix, 450 device files, 441 sessions, starting/ending,
mount points, 440 device-special files, 689-690
mounting, 444 413-414 transfer modes, 691
at boot time, 446-448 DNS (Domain Name transfer status, 691
interactively, 445 Service) resolvers, troubleshooting,
NFS (Network File 796-798 699-700
System) DNS (Domain Name uploads, 691
exporting, 468-470 Service) servers htaccess, 776-777
installing, 464-468 named.boot, 798-800 inode numbers, 411-412
mounting, 471-472 named.ca, 806-807 links
troubleshooting, 472-473 named.hosts, 803-805 ordinary, 412-413
partitions named.rev, 805-806 symbolic, 413
codes/types, 453-454 DOS files, 52-53 listing, 421-424
creating, 451-455 /etc directory fils log files, 283
extended, 455 /etc/conf.modules file, magic files, 410
partition table, 454-456 151-152, 305 moving, 51, 426
primary, 455 /etc/diphosts file, 670-671 name matching, 334-337
sizes, 457 /etc/exports file, 469-470 NIS (Network Information
swap paritions, 456, /etc/fstab file, 446-448, Service), 477-480
459-460 471-472 ordinary, 410-411
888
files
organizing, 424-425 filters, 398 kernel interface table, 634
password find command, 266, 432-434 routing tables, 627
fields, 250 Find utility (KDE), 563 flames, 743, 748, 753
login shell, 321 finding #flastmod directive, 770
patch files, 297 files, 266, 432-434 floppy disks
pathnames, 408-409 mailing lists, 712 booting from, 240-241
permissions, 177 packages, 168-169 installing Linux from,
changing, 415-416 finger command, 378-379 63-64
granting, 416-417 FIPS (first nondestructive flush command (dip), 665
relative permissions, interactive partition FM synthesis, 148
417-418 splitting) program, 129 focus, 530
subfields, 415 fire protection, 276 folders, 570. See also
viewing, 414 firewalls directories
recovering, 436-438 attackers, 655 follow-up articles (Usenet),
renaming, 426 defined, 636 751
restoring, 272 host, 639 fonts, 572
searching, 430 host separation, 644 for loops, 360
security, 281-282 kernels, 639-640 -force option (rpm
smb.conf monitoring, 654 command), 169
code listing, 487-491 network security Foreign Address field
[global] section, 491-492 policy, 655 (Active Internet
[homes] section, 492-493 networking options, Connections), 631
[printers] section, 493 640-643 forgotten passwords, 252
testing, 494 packet filter, 636-637 fork processes, 331
swap files creating, 645-654 forking, 766
activating/deactivating, forwarding, 636 form command (FTP), 688
461 ipchains utility, 646-651 forward slash (/), 49
creating, 460 masqerading, 636 forwarding
defined, 459 planning, 645-646 email, 731-732
time/date stamps, 434 proxy firewalls, compared, packet filter firewalls, 636
vi 638-639 ports, 659
creating, 208-209 passwords, 644 FreeBSD, 26
editing, 209-210 physical configurations, Frequently Asked Questions
overwriting, 215 639-644 (FAQs), 743
saving, 213-214 policies, 651-652 front-end processing,
viewing contents of, 428 implementing, 653-654 186-187
cat command, 429 testing, 654 fsck command, 450
less command, 52, what not to filter, 652 #fsize directive, 770
429-430 what to filter and FTAM (File Transfer,
more command, 52, 429 location, 652 Access, and Management),
od command, 430-432 port forwarding, 644-645 599
XF86Config, 516 proxy, 636-638 FTP (File Transfer
Device section, 521 packet filter firewalls, Protocol), 599, , 706, 782
Files section, 517 compared, 638-639 anonymous FTP, 686-687,
Keyboard section, standard, 636 693-695, 706, 782-783
517-518 transparent, 636 group files, 791
Monitor section, 519-521 software, 643-644 password files, 791
Pointer section, 518-519 first-in-first-out (FIFO) server permissions, 790
Screen section, 521-522 buffers, 414 sessions, 707, 783-789
ServerFlags section, 517 Flags field support for, 783
Files section (XFree86 Active UNIX Domain archie, 708-709
file), 517 Sockets, 632 commands, 687
889
GNOME (GNU Network Object Model Environment)
!, 687 reget, 688 menu, 553-554
$, 687 rename, 689 preparation, 553
?, 689 reset, 689 start-up programs, 554-555
account, 687 restart, 689
append, 687 rhelp, 689
ascii, 687 rmdir, 689
G
bell, 687 rstatus, 688
G command (vi), 227
binary, 687 runique, 689
G verification failure code
bye, 687 send, 689
(RPM), 173
case, 687 site, 689
-Gamma option (XFree86
cd, 687, 784, 787 size, 689
monitor), 520
cdup, 687 status, 689
Gateway field (routing
chmod, 687 struct, 689
tables), 627
close, 687 sunique, 689
gateways, 190, 616
cr, 687 system, 689
gdm display manager, 539
debug, 687 tenex, 689
General Public License
delete, 687 tick, 689
(GNU), 28-29, 862
diconnect, 687 trace, 689
applying to programs,
dir, 687 type, 689
868-869
exit, 688 umask, 689
preamble, 863-864
form, 688 user, 689
terms/conditions, 864-868
ftp, 783 verbose, 689
Genmask field (routing
get, 688, 690, 788 dir command, 690
tables), 627
glob, 688 directories
geometries (windows), 534
hash, 688 changing, 690
get command, 788
help, 688 remote directory
dip (Dial-up Protocol)
idle, 688 listings, 690
driver, 666
image, 688 downloads, 690-691
FTP (File Transfer
lcd, 688 file extensions, 706-707
Protocol), 688, 690
ls, 688, 690, 784 quitting, 690
majordomo, 714
macdef, 688 sessions
GIMP (GNU Image
mdelete, 688 closing, 690
Manipulation Program),
mdir, 688 example, 692
548-549
mget, 688, 691 starting, 689
glob command (FTP), 688
mkdir, 688 sites, 708, 840-841
[global] section (smb.conf
mls, 688 transfer modes, 691
file), 491-492
mode, 688 transfer status, 691
GNOME (GNU Network
modtime, 688 troubleshooting, 699-700
Object Model
mput, 688, 691 uploads, 691
Environment), 578-579
newer, 688 wu-ftpd
advantages/disadvantages,
nlist, 688 command-line options,
592-593
nmap, 688 789
configuring, 587
ntrans, 688 configuring, 789-790
applets, 589
open, 688 installing, 782
background, 590
passive, 688 ftp command, 783 desktop, 592
prompt, 688 /ftp/etc/group file, 791 launchers, 589
proxy, 688 /ftp/etc/passwd file, 791 Main Menu options, 588
put, 688, 691 full backups, 260 panel, 589
pwd, 688, 784 function keys, 67-68 screensavers, 591
quit, 688, 789 fvwm2 window manager, sounds, 591
quote, 688 532-535, 552
recv, 688
890
GNOME (GNU Network Object Model Environment)
subpanels, 589 graphics. See also GUIs kfm (KDE file manager),
window behavior, (graphical user interfaces) 568-570
590-591 GIMP (GNU Image packages, 560-561
Enlightenment menu, Manipulation Program), panel, 562-566
585-586 548-549 Qt toolkit, 560
GNOME Manifesto, Graphics mini Root menu, 567-568
579-580 How-To, 857 setting as default desktop,
installing, 580-581 greater than sign (>), 338 561-562
Midnight Commander File Group directive, 763 taskbar, 566-567
Manager, 586-587 group.byid map (NIS), 477 Templates folder, 568
panel group.byname map trash, 568
clock, 584 (NIS), 478 X Window System,
Control Center, 583 groups 504-506. See also
help system, 583 command groups, 342 XFree86
hiding, 584 group acounts, 253 client/server relationship,
Main Menu button, adding, 254 506-507
582-583 deleting, 254 display managers,
Netscape button, 583 security, 280 538-539
pager, 583-584 group files, 791 GIMP (GNU Image
task list, 584 grpck command, 292 Manipulation
Terminal button, 583 Gtk+, themes, 592 Program), 548-549
Root menu, 585 guest accounts, 279-280 historical overview, 505
selecting as default GUIs (graphical user input capabilities, 508
desktop, 581 interfaces), 504 navigating, 530-531
tear-off menus, 587 GNOME (GNU Network network
themes, 591-592 Object Model transparency, 507
GNU (GNU’s Not UNIX) Environment), 578-579 networking, 525, 527
General Public License, advantages/disadvantages, output capabilities, 507
28-29, 862 592-593 Seyon, 549-551
applying to programs, configuring, 587-592 themes, 537-538
868-869 Enlightenment menu, troubleshooting, 551-552
preamble, 863-864 585-586 user interface capabilities,
terms/conditions, 864-868 GNOME Manifesto, 507-508
GIMP (GNU Image 579-580 window managers,
Manipulation Program), installing, 580-581 531-537, 552-555
548-549 Midnight Commander xcalc, 545-548
Network Object Model File Manager, 586-587 xlock, 551
Environment. See panel, 582-584 xterm, 541-544
GNOME Root menu, 585 xv, 544-545
gopher, 710-711 selecting as default .gz file extension, 175
GOSIP (Government Open desktop, 581 gzip command,
Systems Interconnection tear-off menus, 587 175-177, 435
Profile), 599 themes, 591-592
goto command (dip), 666 KDE (K Desktop
granting file permissions, Environment), 558-559
H
416-417 advantages/disadvantages,
h command (vi), 226
graphical user interfaces. 574-575
-h option
See GUIs Autostart folder, 568
gzip command, 176
components, 559-560
ipchains command, 647
configuring, 570-574
ps command, 381
rpm command, 169
891
hubs
wget command, 274 monitors, 83 X Window System, 505
who command, 377 mouse devices XFree96, 505
hackers, 15, 20, 276 xterm, 543-544 home directories, 254,
halt command, 54, 245 xv program, 545 408, 443
handshaking, 814 support for, 21, 508-510 HOME environment
hard drives, 103-104 Harmony Project, 560 variable, 326, 328
installing Linux from, hash command (FTP), 688 $HOME/.Xclients file, 541
40, 65 headers (Usenet $HOME/.Xmodmap
Linux requirements, 34-35 articles), 750 file, 541
partitions, 126 help $HOME/.Xresources
adding, 109-111 GNOME (GNU Network file, 541
codes/types, 453-454 Object Model $HOME/.xsession file, 541
creating, 44, 72-79, 125, Environment), 583 [homes] section (smb.conf
130, 138-139, 451-455 Help Browser (KDE), 563 file), 492-493
deleting, 129-130 help command HorizSync option (XFree86
DOS drives, 104-107 dip (Dial-up monitor), 520
extended, 455 Protocol), 666 host name resolution. See
initializing, 140 FTP (File Transfer name resolution
inode tables, 411-412 Protocol), 688 hosts, 794
Linux installation, 96 majordomo, 714 access control, 775-776
n command, 455 Help on desktop command firewalls, 639, 644
naming, 125-126 (KDE Root menu), 567 messaging, 810
partition tables, 126, mail program, 727 hosts.byaddr map (NIS), 478
454-456 man pages, 48 hosts.byname map
primary, 455 hiding (NIS), 478
repartitioning, 128 GNOME (GNU Network HotBot search engine, 705
requirements, 127-128 Object Model How-To documents,
sizes, 457 Environment) panel, 584 839-840, 844
swap partitions, 111, 140, KDE (K Desktop bibliography, 845-851
456, 459-460 Environment) panel, 566 copyrights, 859
types, changing, 457 hierarchy of Usenet Distribution How-To, 17
verifying, 457-458 newsgroups, 827-828 High Availability
hard mounts, 472 High Availability How-To, 857
hardware How-To, 857 mini How-Tos, 851-857
backup media, 260, HINFO resource obtaining, 844
263-265 records, 802 submitting, 859
device conflicts, 161 history command, 347 translations, 844-845
drives. See hard drives history list (commands), unmaintained
Linux requirements, 32-33 45-46, 347 documents, 857
bus, 33 history of writing, 858
CD-ROMs, 37 email, 810 XFree86, 504
CPU, 33 Internet, 794 HP emulation (xcalc),
disk drives, 34-35 Linux, 14, 25-28 547-548
memory, 34 Minix, 27 .htaccess files, 776-777
monitors, 35-36 UNIX development, HTML (Hypertext Markup
mouse devices, 38 25-26 Language), 704
network access, 37 sendmail, 817 -http-user option (wget
printers, 38-39 TCP/IP (Transmission command), 274
swap space, 35 Control Protocol/ httpd.conf file, 762-764
tape drives, 38 Internet Protocol), 598 hubs, 189
LISA program Usenet, 744-745
autodetection, 101
892
-hw option (ifconfig command)
-hw option (ifconfig software loopback interface NFS (Network File
command), 623 configuration, 624-625 System)
hybrid topology, 191 status reports, 624 components, 464-465
hyperlinks, 702 IFS environment variable, init script, 465-468
Hypertext Markup 333 pmap_dump
Language (HTML), 704 ignore variable, 735 command, 465
hyphen (-), 332, 381 image command (FTP), 688 OpenLinux
images. See graphics disk space
/include directories, 419 requirements, 95
I #include directive, 769 fdisk program, 107-109
i command (vi), 226 incremental backups, 260 hard drive partitions, 96
-i option index command hard drive preparation,
cpio command, 268
(majordomo), 714 103-104
dip command, 665
indexes (Apache), 766 hardware detection,
ipchains command,
inetd daemon, 498 101-103
646-647
Inference Find! search install floppies, 96-99
ls command, 422
engine, 705 LILO (Linux
netstat command, 630
info command Loader), 114
ssh command, 699
(majordomo), 714 LISA, 100-101
wget command, 274
Infoseek search engine, 705 methods, 94
I/O (input/output) init command, 234-237, 666 modules disk, 99
processes, 332
init processes, 331 packages, 111-112
redirecting, 337-338
initializing partitions, 104-107,
network interfaces, 622 109-111
X Window System
partitions, 140 preparations, 95-96
input capabilities, 508
output capabilities, 507
inittab files, 238-239 prior configurations, 100
IBM PROFS, 810 INN news program, 744, reboot procedure, 115
ICMP, 648-649 828 system configuration,
iconified windows, 532 inode numbers, 411-412 112-114
idle command (FTP), 688 input devices system requirements, 94
joysticks, 157-158 troubleshooting, 115-118
IDs (identifiers)
mouse devices taper, 270
PIDs (process IDs), 331
configuring, 82 wu-ftpd, 782
SGIDs (Set Group
Linux requirements, 38 XFree86, 508-514
IDs), 282
selecting text, 46 hardware support,
SUIDs (Set User IDs), 282
xterm, 543-544 508-510
if command (dip), 666
xv program, 545 RPMs, 510-512
if field (/etc/printcap
file), 399 input mode (vi), 206, 208 installation (Linux), 62-63
if structures, 357-358 input/output. See I/O boot/supplemental disks,
Iface field insmod command, 303-304 44, 66-67
kernel interface table, 634
installation floppies, 96-99 components, 45, 80-82
routing tables, 627
installation (GNOME), Debian distribution, 120
ifconfig commmand, 197 580-581 base system, 141-143
kernel from CD-ROM, 131
command-line options,
command sequence, 298 from DOS, 131-132
622-623
modules, 303 fdisk command, 135-137
network interface
preparation, 296 hard drive partitions,
configuration, 625
LDAP (Lightweight 125-130, 138-140
Parallel IP interface
Directory Access
configuration, 625
Protocol), 482-483
LILO (Linux Loader), 114,
242
893
Internet
network kernel upgrades, 178-179 EOT (end of transmission)
configuration, 141 non-Linux software messages, 721
preparation, 121-124 file permissions, 177 ESMTP (Extended
requirements, 120-121 gzip command, 175-177 Simple Mail Transfer
rescue disks, 132-134 make command, 177 Protocol), 815
system configuration, 140 package formats, 175 forwarding, 731-732
system files, 134-135 troubleshooting, 178 handshaking, 814
DEC Alpha systems, 89-91 uninstalling, 178 help, 727
boot disks, 90-91 RPM (Red Hat Package historical overview, 810
supported hardware, 90 Manager), 168-170 LAN-based messaging,
function keys, 67-68 finding packages, 168 810-811
hardware requirements, querying packages, mailing lists, 711-716
32-33 172-173 MDAs (Mail Delivery
bus, 33 uninstalling packages, Agents), 811
CD-ROMs, 37 170-171 message formats, 815-816
CPU, 33 updating packages, MTAs (Mail Transfer
disk drives, 34-35 171-172 Agents), 811
memory, 34 verifying packages, 173 MUAs (Mail User
monitors, 35-36 Samba, 487 Agents), 811
mouse devices, 38 system administrator Mutt client, 739-740
network access, 37 duties, 167 printing, 726
printers, 38-39 terminology, 164 quitting, 736-737
swap space, 35 upgrades, 165 reading, 723-725
tape drives, 38 interactive processes, 366 replying to, 729-731
keyboard selection, 69 interfaces RFC (Request for
LILO (Linux Loader), bindings, 579 Comments) standards,
87-88 network. See network 812-814
media, 39 interfaces saving, 728-729
methods Internet, 598-599. See also sending, 719-721, 726
CD-ROM, 40, 63-64 protocols; Web sendmail, 811, 816-822
floppy disk, 63-64 connections SMTP (Simple Mail
hard drive, 40, 65 dip (Dial-Up IP Protocol) Transfer Protocol), 599,
network, 40, 65 driver, 664-671 720, 814-815
partitions, 44, 72 PPP (Point-to-Point system mailbox, 718
Disk Druid, 78-79 Protocol), 664, 671-679 text editors, 722
fdisk command, 72-73, SLIP (Serial Line undeliverable, 719
75-77 Internet Protocol), writing, 721
swap partitions, 77-78 664, 670 history of, 794
PCMCIA cards, 69 email structure, 702
preparation, 39, 65-66 advantages, 718 Usenet newsgroups
SCSI devices/drivers, 71 aliases, 733 accessing, 711
testing, 89 canceling, 721-722 articles, 743-744,
troubleshooting, 55-59, carbon copies, 732-733 748-753
88-89 client/server bandwidth, 743
update packages, 42-43 messaging, 811 client programs, 829
upgrades, 71 commands, 734 compared to BBSes, 742
Welcome screen, 68 customizing environment, culture, 748-749
installation (software), 734, 736 ettiquette guidelines, 830
164-167 deleting/undeleting, 729 FAQs (Frequently Asked
Debian Package elm mailer, 737-739 Questions), 743
Management System, 174 environment flames, 743, 748
dump, 271 variables, 735 hierarchy, 745, 827-828
894
Internet
history of, 744-745, ipchains program, 646-651 KDE (K Desktop
826-827 arguments, 647-649 Environment), 558-559
kill files, 753 built-in chains, 649-650 advantages/disadvantages,
moderated, 749 commands, 646 574-575
moderators, 830 ICMP types/subtypes, Autostart folder, 568
net.personalities, 743 648-649 components, 559-560
net.police, 743 IP masquerading, 659 configuring
netiquette, 747, 752-753 packet filter firewalls, 637 background, 572
news distributions, creating, 646-654 bookmarks, 571
746-747 planning, 645-646 colors, 572
news feeds, 742 port forwarding, 644 desktop organization,
news.announce. TOS masks, 649 573-574
newusers, 827 user-defined chains, 650 fonts, 572
newsgroup ipmasq program, 644 screensaver, 573
hierarchies, 746 ipmasqadm program Templates folder,
newsreaders, 742-743, firewall network 570-571
753-755, 829-832 options, 642 themes, 573
server programs, 828-829 forwarding ports, 659 kfm (KDE file manager),
signal-to-noise ratio, 743 ipportfw wrapper, 645 568-570
smileys, 744 ipportfw program, 644 packages, 560-561
spam, 751 -irq option (ifconfig panel
spawning, 827 command), 623 Application Starter, 562
subscribing to, 749-750 IRQ/DMA timeouts, 161 clock, 565
terminology, 743-744 ispell command, 369 Control Center, 563
newbies, 743 iterative structures, 350, Find utility, 563
InterNIC Web site, 605 355, 360-361 Help Browser, 563
interoperability, 19 hiding, 566
interrupt key, 324 Home Directory, 563
intro command
J Lock button, 564
(majordomo), 714 J command (vi), 226 Logout, 564
IP (Internet Protocol) -j option virtual desktops, 564-565
addresses, 599-604 ipchains program, 647
Window List, 563
ALL address, 604 PPP (Point-to-Point
od command, 431
classes, 604-605 Protocol) configuration,
ps command, 381
ipchains program, 659 677-679
joysticks, 157-158
IPv6, 599 Qt toolkit, 560
jtest program, 157
loopback address, 604, 621 Root menu, 567-568
masquerading, 655-657 setting as default desktop,
kernel components K 561-562
required, 657-658 taskbar, 566-567
kernel modules, 658
k command (vi), 226 Templates folder, 568
ports, forwarding, 659
K Destop Environment. See
trash, 568
setup, 658-660
KDE kdm display manager, 539
troubleshooting, 660-661
-k option kernel, 296
rcp command, 698
obtaining, 605 backups, 297
rlogin command, 696
SLIP (Serial Line Internet compiling, 301
rsh command, 697
Protocol) interfaces configuration
ssh command, 699
dynamic IP addresses, interactive text-based
669-670
K-Menu, 562 program, 298
static IP addresses, menu-based program, 300
667-668 options, 299
patch files, 297
895
Linux installation
source code files, 297 KOM (KDE Object applying to programs,
X Window system-based Model), 559 868-869
program, 300-301 Korn shells preamble, 863-864
defaults, 302 ksh, 829 terms/conditions, 864-868
firewalls, 639-640 pdksh, 320 Lightweight Directory
installation kwm window manager, Access Protocol. See
command sequence, 298 532, 536 LDAP
modules, 303 LILO (Linux Loader), 95,
preparing for, 296 239-243
IP (Internet Protocol)
L advantages/disadvantages,
masquerading 241
l command (vi), 226
requirements, 657-658 configuration, 242-243
-l option
kerneld daemon, 303-305 installing, 87-88, 114, 242
chat command, 671
loading, 151-152 uninstalling, 114, 245
crontab command, 376
modularized, 303
fdisk command, 452
links, 702
modules, 302 ordinary, 412-413
file command, 410
adding, 304 symbolic, 413, 778-779
fsck command, 450
commmands, 303
gzip command, 176
-links option (find
deleting, 304
ipchains program, 648
command), 433
installing, 303
kill command, 387
Linux Documentation
IP masquerading, 658
ls command, 281, 414, 422
Project, 839
viewing, 303
ps command, 381
Linux How-Tos. See
saving, 301
rlogin command, 696
How-To documents
troubleshooting, 160
rsh command, 697
Linux installation, 62-67
upgrading, 178-179 boot/supplemental disks,
ssh command, 699
version, determining, 296 44, 66-67
telnet command, 685
Kernel Configurator dialog L verification failure code components, 45, 80-82
box, 303 (RPM), 173 Debian distribution
Kernel Module Manager, LANs (Local Area base system, 141-143
101-102 Networks), 193, 810-811 fdisk command, 135-137
kerneld daemon, 303-305 last reboot command, 246 from CD-ROM, 131
Keyboard section (XFree86 lcd command (FTP), 688 from DOS, 131-132
file), 517-518 LDAP (Lightweight hard drive partitions,
keys, xcalc Directory Access 125-130, 138-140
HP emulation, 547-548 network configuration,
Protocol), 481-482
T1 emulation, 546-547 141
client configuration, 484
keyword Web searches, 705 database population,
preparation, 121-124
kfm (KDE file manager), 483-484
requirements, 120-121
568-570 installing, 482-483
rescue disks, 132-134
kill command, 367, 386 Leafnode server, 829 system configuration, 140
nontermination signals, system files, 134-135
legal issues. See copyrights;
389-390 DEC Alpha systems, 89-91
licenses
normal termination, function keys, 67-68
Legatto systems, 262
387-388 hardware requirements,
less command, 52, 429-430
terminating all 32-33
lf field (/etc/printcap file),
processes, 389 bus, 33
399
unconditional termination, CD-ROMs, 37
/lib directories (UNIX),
388-389 CPU, 33
418, 441
kill files, 753 licenses, GNU General disk drives, 34-35
kill key, 325 Public License, 28-29, 862 memory, 34
killall command, 387 monitors, 35-36
killing. See stopping mouse devices, 38
896
Linux installation
network access, 37 Linuxconf, 255 Manager), 102
printers, 38-39 LinuxPPC, 310-312 Loadable Module
swap space, 35 LISA (Linux Install and Support dialog box, 301
tape drives, 38 System Administration) Loader. See LILO
keyboard selection, 69 program, 100-101 loading
LILO (Linux Loader), listing files, 421-424 joystick modules, 157-158
87-88 listings kernel modules, 151-152
media, 39 access.conf file, 778 Local Address field (Active
methods adduser command, 251 Internet Connections),
CD-ROM, 40, 63-64 case structure, 355 631
floppy disk, 63-64 command-line parsing, 356 Local Area Networks
hard drive, 40, 65 dev/sndstat output (LANs), 193, 810-811
network, 40, 65 OSS sound driver, 154 local distribution
OpenLinux. See OSS/Lite sound (Usenet), 746
OpenLinux driver, 153 local loopback
partitions, 44, 72 dip (Dial-Up UP Protocol) addresses, 621
Disk Druid, 78-79 scripts $local variable, 667
fdisk command, 72-73, dynamic IP addresses, 669 localhosts, 621
75-77 static IP addresses, $locip variable, 667
swap partitions, 77-78 667-668 Lock button (KDE), 564
PCMCIA cards, 69 /etc/conf.modules file, Lock Screen command
preparation, 39, 65-66 151-152, 305 (KDE Root menu), 568
SCSI devices/drivers, 71 /etc/inittab file, 238-239 locking X Window System
testing, 89 /etc/lilo.conf file, default display, 551
troubleshooting, 55-59, kernel boot options, 302 LogFormat directive,
88-89 for loop, 360 773-774
update packages, 42-43 FTP (File Transfer logging,
upgrades, 71 Protocol) Apache, 772
Welcome screen, 68 anonymous FTP, CLF (Common Logfile
Linux Installation and Getting 693-695 Format), 774
Started Guide , 28 sample session, 692 mod_log_config module,
Linux Journal, 840 named.ca file, 807 773
Linux Loader. See LILO named.hosts file, 803 NCSA compatibility,
Linux, overview of, 14-16 named.rev file, 805 772-773
advantages, 17, 20 nfs init script, 465-468 virtual hosts, 774-775
educational uses, 19 partition table, 73, bootup entries, 234
portability, 18 108, 136 system file logs, 283
professional/corporate rc.local shell script, logging out, 47-48
advantages, 18-19 237-238 logical file system. See file
applications, 18 rc3.d directory, 236 system
disadvantages rdist configuration file, 273 logins, 47-48, 318-322
hardware problems, 21 safrm script, 330 root, 184
overcoming, 23-24 smb.conf file, 487 security, 278
software problems, 22 test command, 359 accounts without
system administration, 23 test command shell passwords, 279
technical support, 20-21 script, 358 command accounts, 280
distributions, 16-17 while loop, 361 default accounts, 279
history, 14 lists command group accounts, 280
Minix, 27 (majordomo), 714 guest accounts, 279-280
UNIX development, ln command, 413 unused accounts, 279
25-27 Load Kernel Modules shell environment, 325
ownership/rights, 29 functions (Kernel Module
897
managers
HOME variable, 326, -ma option (touch 733-734, 841
328 command), 434 creating, 822-824
LOGNAME variable, macdef command finding, 712
326, 330 (FTP), 688 LinuxPPC, 312
MAIL variable, 329 Macintosh PowerPC mail reflectors, 712
PATH variable, 326-329 platforms majordomo scripts,
PS1 variable, 326, 329 AppleTalk, 312-314 712-714
PWD variable, 326 LinuxPPC, 310-312 subscribing to, 714-716
SHELL variable, 326 MkLinux, 308-309 Main Menu (GNOME),
TERM variable, Netatalk, 312-314 582-583
326, 329 SheepShaver, 312 maintaining file systems,
TZ variable, 329 Yellow Dog Linux, 449
telnet, 684 309-310 major timeouts, 472
terminal environment magazines, 840 majordomo, 712-714,
control keys, 324-325 magic files, 410 822-824
device drivers, 322-323 mail clients (MUAs), 814 make command, 177, 298
user accounts, 250, 253 Mail Delivery Agents man command, 48, 840
LOGNAME environment (MDAs), 811 man pages, 48, 840
variable, 326, 330 MAIL environment managers
Logout (KDE), 564 variable, 329 display managers
Logout command (KDE mail program, 719 configuring, 539
Root menu), 568 aliases, 733 gdm, 539
loopback address, 604 commands, 734 kdm, 539
loops customizing, 734-736 xdm, 538
for, 360 email messages Midnight Commander File
while, 361 canceling, 721-722 Manager, 586-587
lp field (/etc/printcap file), carbon copies, 732-733 RPM (Red Hat Package
399 deleting/undeleting, 729 Manager), 167-168
lpc command, 395-396 forwarding, 731-732 finding packages,
lpd daemon, 394-395 printing, 726 168-169
lpq command, 395 program/command installing packages,
lpr command, 395, 726 results, 723 169-170
lprm command, 395 reading, 723-726
ls command, 49, 281, 332, replying to, 729-731 querying packages,
414, 421-424, 688-690, saving, 728-729 172-173
784 sending, 720-721, 726 uninstalling packages,
lsmod command, 302-303 writing, 721 170-171
Lycos search engine, 705 environment variables, 735 updating packages,
help, 727 171-172
mailing lists, 733-734 verifying packages,
M quitting, 736-737 173
text editors, 722 window managers
M command (vi), 226
-m option mail reflectors, 712 AfterStep, 532, 535
dip command, 665
Mail Transfer Agents Backbox, 532
fdisk command, 452
(MTAs), 811 Blackbox, 536
file command, 410
Mail User Agents configuring, 536-537
ipchains command, 647
(MUAs), 811 Enlightenment, 532, 536
ps command, 381
mail. See email fvmn2, 532, 552-555
tar command, 265
mail.aliases map (NIS), 478 fvmn95, 532
M verification failure code mail.byaddr map (NIS), 478 fvwm2, 533-534
(RPM), 173 mailboxes, 718 fvwm95, 535
mailing lists, 711-712, kwm, 532, 536
898
managers
Twm, 531-533 Midnight Commander File monitoring
Window Maker, Manager (GNOME), firewalls, 654
532, 535 586-587 networks, 630-631
mapping host names to IP MIME (Multipurpose active network
addresses. See name Internet Mail Extensions) connections, 631-633
resolution types, 571 interface statistics, 634
maps (NIS), 477-480 mini How-To documents, kernel routing table, 633
masquerading 851-857 processes, 379-383
IP (Internet Protocol) Minix, 25, 27, 450 systems, 196-198
addresses, 655-657 minor timeouts, 472 users
ipchains program, 659 misc newsgroups, 745 finger command, 378-379
kernel components mixer, 149 ps commmand, 284
required, 657-658 mkdir command, 50, who command, 376-378
kernel modules, 658 424, 688 monitors
ports, forwarding, 659 mkfs command, 458-459 configuring, 83
setup, 658-660 MkLinux, 308-309 Linux requirements, 35-36
troubleshooting, 660-661 mkswap command, 459 more command, 52, 429
packet filter firewalls, 636 mlabel command, 53 mount command, 445, 471
matching filenames, 334 mls command, 688 Mount Options field
* (asterisk), 335 mmd command, 53 (/etc/fstab file), 447
[ ] expression, 336-337 mode command, 666, 688 Mount Point field
? (question mark), 336 Modeline option (XFree86 (/etc/fstab file), 446
mattrib command, 53 monitor), 520 mount points, 440
MAUs (Multistation Access modem command (dip), 666 mounting
Units), 191 $modem variable, 667 CD-ROMs, 117
mcd command, 53 modems, 187, 285 /etc/exports file
mcopy command, 53 moderated newsgroups, 749 options, 470
MDAs (Mail Delivery moderators, 830 file systems, 444
Agents), 811 modes at boot time, 446-448
mdel command, 53 device drivers, 323 interactively, 445
mdelete command, 688 vi, 206-208 NFS (Network File
mdir command, 53, 688 modprobe command, 303 System)
memory modtime command, 688 /etc/fstab file, 471-472
buffers, 54 modules hard mounts, 472
Linux requirements, 34 kernel, 302 interactively, 472
XFree86 requirements, 508 adding, 304 soft mounts, 472
menus (X Window commands, 303 mouse devices
System), 531 deleting, 304 configuring, 82
messages, email. See email installing, 303 Linux requirements, 38
messaging programs. See IP (Internet Protocol) selecting text, 46
elm; mail program masquerading, 658 xterm, 543-544
Met field (kernel interface viewing, 303 xv program, 545
table), 634 PAM (Pluggable moving
metoo variable, 735 Authentication Modules), files, 51, 426
Metric field (routing 287-288 vi cursor, 215-217
tables), 627 configuration files, 288 mput command (FTP), 688,
-metric option (ifconfig control flags, 288-289 691
command), 623 modules disk, 99 mrd command, 53
mformat command, 53 mod_log_config module mren command, 53
mget command, 688, 691 (Apache), 773
MIDI (musical instrument Monitor section (XFree86
digital interface), 149 file), 519-521
899
Netatalk
MS-DOS continuing after logout, named.hosts files, 803-805
drive partitions, 104-107 383-384 named.rev files, 805-806
file commands, 52-53 daemons, 366 name-server records, 804
installing Linux from, defunct, 383 named daemon
131-132 interactive, 366 named.boot file, 798-800
MTAs (Mail Transfer monitoring, 379-383 named.ca file, 806-807
Agents), 811 multiple, 368-370 named.hosts file, 803-805
MTU field (kernel interface parent, 369 named.rev file, 805-806
table), 634 sending signals to, named pipes, 414
-mtu option (ifconfig 389-390 named.boot files, 798-800
command), 623 stopping, 386-389 named.ca files, 806-807
$mtu variable, 667 zombie, 383 named.hosts files, 803-805
mtype command, 53 queues, 366 named.rev files, 805-806
MUAs (Mail User scheduled tasks names
Agents), 811 at command, 370-372 files, 408, 426
multimedia batch command, 372-373 network nodes, 606-608
applications, 158-159 cron command, 373 partitions, 125-126
CDs, 155 crontab command, pathnames, 408-409
playing, 156 373-376 Naughton, Patrick J., 551
saving to MP3 format, priority, 384-386 navigating
156 time-sharing, 366 directories, 48-49
signatures, 156 user monitoring GNOME (GNU Network
troubleshooting, 162 finger command, 378-379 Object Model
joysticks, 157-158 who command, 376-378 Environment), 582
MIDI (musical instrument multiuser systems. See Enlightenment menu,
digital interface), 149 networks 585-586
online resources, 159-160 multiuser/multitasking Midnight Commander
sound cards commands, 367 File Manager, 586-587
A/D (analog-to-digital) music. See sound panel, 582-584
converters, 148 musical instrument digital Root menu, 585
configuring, 149-152 interface (MIDI), 149 tear-off menus, 587
D/A (digital-to-analog) Mutt email client, 739-740 KDE (K Desktop
converters, 148 mv command, 51, 426 Environment)
drivers, 146-148 mx field (/etc/printcap Autostart folder, 568
mixer, 149 file), 400 kfm (KDE file manager),
testing, 153-155 MX resource records, 802 568-570
troubleshooting, 160 panel, 562-566
sound sampling, 148 Root menu, 567-568
multiport adapters, 188
N taskbar, 566-567
Multipurpose Internet Mail n command (vi), 227 Templates folder, 568
Extensions (MIME) -n option trash, 568
types, 571 fdisk command, 452
X Window System
Multistation Access Units gzip command, 176
focus, 530
(MAUs), 191 ipchains command,
menus, 531
multitasking, 185 647-648
virtual terminals, 531
commands, 367
ls command, 422
nc distribution (Usenet),
preemptive, 14
mount command, 445
746
processes
netstat command, 630
NCP (Network Control
background processes,
name servers, 795 Protocol) protocol, 817
368-369
named.boot files, 798-800
net.personalities, 743
batch, 366
named.ca files, 806-807
net.police, 743
child, 369 Netatalk, 312-314
900
netgroup map (NIS)
netgroup map (NIS), 478 administering. See installing Linux from,
netgroup.byhost map administration 40, 65
(NIS), 478 bridges, 611 LANs (Local Area
netgroup.byuser map centralized-processing Networks), 193
(NIS), 478 systems, 186-188 monitoring, 196-198
netid.byname map client/server model, 192 naming, 606-608
(NIS), 478 configuring NIS (Network Information
netiquette, 743, 747, bandwidth, 614 Service)
752-753 bridges, 617 clients, 476
-netmask option (ifconfig connection media, 614 domains, 480
command), 623 dial-up connections, 615 files, 477-480
netmasks.byaddr map diskless workstations, 615 maps, 477-480
(NIS), 479 physical location, 614 security, 481
netnews. See Usenet remote connections, 616 servers, 476
netstat command, 201 routers, 616 nodes, 599
active network switches, 617 private address blocks, 658
connections, listing, TCP/IP, 84 RFCs (Requests for
631-633 connections, 612-614 Comment), 605-606
command-line options, distributed-processing routing
630-631 systems, 188-192 RIP (Routing Information
interface statistics, 634 bridges, 190 Protocol), 610-611
kernel routing table, file servers, 189 segmentation, 611
printing, 633 gateways, 190 security
nettiquette, 830 hubs, 189 breaches, 286
network administrators. See NICs (network interface dial-up passwords, 292
system administrators cards), 189 enforcing, 285-286
network command repeaters, 189 modems, 285
(dip), 666 routers, 190 shadow passwords,
Network File System. See workstations, 189 289-292
NFS firewall options, 640-643 terminal lockouts, 285
Network Information GNOME (GNU Network subnet masks, 608-610
Center (NIC), 599 Object Model subnetting, 608-610
Network Information Environment), 578-579 TCP/IP. See TCP/IP
Service. See NIS advantages/disadvantages, networks
network interface cards 592-593 topologies
(NICs), 189 configuring, 587-592 bus, 191
network interfaces Enlightenment menu, hybrid, 191
configuring, 625 585-586 ring, 191
initializing, 622 GNOME Manifesto, star, 190
Parallel IP interfaces, 625 579-580 troubleshooting, 200-201,
reporting status of, 624 installing, 580-581 617-618
software loopback Midnight Commander X Window System case
interfaces, 624-625 File Manager, 586-587 study, 525, 527
statistics, 634 panel, 582-584 networks.byaddr map (NIS),
Network layer (OSI Root menu, 585 479
model), 601 selecting as default networks.byname map
Network News Transport desktop, 581 (NIS), 479
Protocol (NNTP), 744 tear-off menus, 587 New command (KDE Root
network transparency, 507 themes, 591-592 menu), 567
networks. See also Internet newbies, 743
access, 37 newer command (FTP), 688
news distributions, 746-747
901
OpenLinux installation
news feeds, 742 clients, 476 selecting as default
news newsgroups, 745 domains, 480 desktop, 581
news. See Usenet files, 477-480 tear-off menus, 587
news.announce.newusers maps, 477-480 themes, 591-592
newsgroup, 827 security, 481 obtaining IP addresses, 605
newsgroups, 826, 837-839. servers, 476 octal dump (od) command,
See also Usenet nlist command (FTP), 688 430-432
client programs, 829 nmap command (FTP), 688 online resources. See also
ettiquette guidelines, 830 nmbd daemon, 486 Web sites
hierarchy, 827-828 NN newsreader, 829 FTP sites, 840-841
history of, 826-827 NNTP (Network News How-To documents,
moderators, 830 Transport Protocol), 744 839-840
news.announce. nodes, 599, 606-608, 794 bibliography, 845-851
newusers, 827 noheader variable, 735 copyrights, 859
newsreaders, 742-743, 750 nohup command, 344, 367, Distribution How-To, 17
kill files, 753 383-384 High Availability How-
NN, 829 NoTrapSignals flag (X To, 857
pine, 831-832 server), 517 mini How-Tos, 851-857
rn, 753-755 NS resource records, obtaining, 844
TIN, 829 802, 806 submitting, 859
trn, 754 nslookup commmand, 198 translations, 844-845
server programs, 828-829 ntrans command (FTP), 688 unmaintained
spawning, 827 nu option (vi), 228 documents, 857
newsreaders, 742-743, 750 number comparisons, 359 writing, 858
kill files, 753 number option (vi), 228 XFree86 How-To, 504
NN, 829 Linux Documentation
pine, 831-832 Project, 839
rn, 753-755
O magazines, 840
TIN, 829 mailing lists, 312, 841
o command (vi), 226
trn, 754 newgroups, 837-839
-o option
NFS (Network File System), cpio command, 268
open command (FTP), 688
464, 599 ipchains program, 648
Open Shortest Path First
exporting, 468-470
mount command, 445
(OSFP) protocol, 611
installing
netstat command, 630
Open Sound System (OSS),
components, 464-465
ssh command, 699
146
init script, 465-468
wget command, 274
Open Source, 872-874
pmap_dump
objects, GNOME (GNU Open System
command, 465
Network Object Model Interconnection. See OSI
mounting
Environment), 578-579 model
/etc/fstab file, 471-472
advantages/disadvantages,
open systems, 18-19
hard mounts, 472
592-593
opening anonymous FTP
interactively, 472
configuring, 587-592
sessions, 783
soft mounts, 472
Enlightenment menu,
OpenLDPA, 482-483
troubleshooting, 472-473
585-586
OpenLinux Base, 25
NIC (Network Information GNOME Manifesto,
OpenLinux installation, 94,
Center), 599 579-580
100
nice command, 367, installing, 580-581
disk space requirements, 95
384-385 Midnight Commander File
fdisk program, 107-109
NICs (network interface Manager, 586-587
hard drive partitions, 96
cards), 189 panel, 582-584
hard drive preparation,
NIS (Network Information Root menu, 585
103-104
Service)
902
OpenLinux installation
hardware detection,
101-103
P task list, 584
Terminal button, 583
install floppies, 96-99 KDE (K desktop
p command (vi), 227
LILO (Linux Loader), 114 environment), 562
-p option
LISA, 100-101 Application Starter, 562
fdisk command, 452
methods, 94 clock, 565
ipchains command, 647
modules disk, 99 Control Center, 563
rcp command, 698
packages, 111-112 Find utility, 563
ssh command, 699
partitions, 104-107, Help Browser, 563
Package Manager. See RPM
109-111 hiding, 566
packages, 42-43, 560-561
preparations, 95-96 Home Directory, 563
finding, 168-169
prior configurations, 100 Lock button, 564
formats, 175
reboot procedure, 115 Logout, 564
installing, 80-82
system configuration, virtual desktops, 564-565
installing. See software
112-114 Window List, 563
installation
system requirements, 94
querying, 172-173
PAP (Password
troubleshooting, 115-118
updating, 171-172
Authentication
OpenParts, 559 verifying, 173
Protocol), 676
operating systems. See packet filter firewalls, Parallel IP (PLIP)
names of specific operating 636-637 interfaces, 625
systems creating, 645-654
parent processes, 331,
Options ExecCGI directive, firewall policies, 651-652
369, 766
777 implementing, 653-654
parity command (dip), 666
ordinary files, 410-411 testing, 654
parsing command line, 332
ordinary links, 412-413 what not to filter, 652
partitions, 126
organizing what to filter and
adding, 109-111
files, 424-425 codes/types, 453-454
location, 652
home directories, 254 creating, 44, 125, 130,
forwarding, 636
OSI (Open Source ip chains utility, 646-651
138-139
Interconnect model) masquerading, 636
Disk Druid, 78-79
model, 600-602 planning, 645-646
fdisk command, 72-73,
(Open System 75-77, 451-455
proxy firewalls, compared,
Interconnection), 599 swap partitions, 77-78
638-639
Application layer, 602 deleting, 129-130
pager (GNOME), 583-584
Data Link layer, 601 DOS drives, 104-107
PAM (Pluggable
Network layer, 601 extended, 455
Authentication Modules),
Physical layer, 601 initializing, 140
287-288
Presentation layer, 601 inode tables, 411-412
configuration files, 288
Session layer, 601 Linux installation, 96
control flags, 288-289
Transport layer, 601 n command, 455
panel
OSPF (Open Shortest Path GNOME (GNU Network
naming, 125-126
First), 611 Object Model
partition table, 126
OSS (Open Sound Environment), 582
verifying, 455-456
System), 146 clock, 584
viewing, 454-455
output. See I/O (input/ Control Center, 583
primary, 455
output), 507 help system, 583
repartitioning, 128
overwriting files, vi, 215 hiding, 584
requirements, 127-128
owner field (resource Main Menu button,
sizes, 457
records), 801 582-583
swap partitions, 111,
ownership of Linux, 29 Netscape button, 583
140, 456
ownership of Linux. See pager, 583-584
creating, 459-460
copyrights, 28 defined, 459
903
primary partitions
types, changing, 457 file permissions, 414 POP (Post Office
verifying, 457-458 changing, 415-416 Protocol), 718
Pass Number field granting, 416-417 populating databases,
(/etc/fstab file), 447 relative permissions, 483-484
passive command 417-418 port command (dip), 666
(FTP), 688 subfields, 415 -port option (telnet
passwd command, 252, 291 viewing, 414 command), 685
passwd.byname map FTP (File Transfer $port variable, 667
(NIS), 479 Protocol) servers, 790 portability of Linux, 18
passwd.byuid map printing, 394 Portable Operating System
(NIS), 479 viewing, 281 Interface (POSIX), 19
Password Authentication physical configuration of ports
Protocol (PAP), 676 firewalls, 639-644 Apache, 762
password command host, 639 forwarding,
(dip), 666 host separation, 644 defined, 644
password file kernels, 639-640 firewalls, 644-645
fields, 250 networking options, IP masquerading, 659
FTP (File Transfer 640-643 POSIX (Portable Operating
Protocol), 791 passwords, 644 System Interface), 19
login shell, 321 software, 643-644 posting newsgroup
Password PAM (Pluggable Physical layer (OSI articles, 743
Authentication Modules) model), 601 follow-up articles, 751
module, 287 physical network location, new articles, 751-752
passwords 614 pound sign (#), 351
choosing, 278 physical security, 276-277 PowerPC platforms
deleting, 252 PIDs (process IDs), 331 AppleTalk, 312-314
dial-up, 292 pine, 831-832 LinuxPPC, 310-312
firewalls, 644 ping command, 200 MkLinux, 308-309
forgotten, 252 pipe character (|), 369-370 Netatalk, 312-314
root, 87, 277 pipes, 337, 414 SheepShaver, 312
shadow, 289 planning Yellow Dog Linux,
command-line tools, networks, 615-616 309-310
290-292 packet filter firewalls, PPP (Point-to-Point
/etc/password files, 289 645-646 Protocol), 671
/etc/shadow files, 290 platforms. See names of accounts, 675-676
Shadow Suite, 289 specific platforms chat program
user accounts, 250, 252 playing CDs, 156 command-line options,
paste operations, PLIP (Parallel IP) 671-672
223-225, 348 interfaces, 625 scripts, 672-673
patch command, 297 Pluggable Authentication configuring with KDE (K
patch files, 297 Modules. See PAM desktop environment),
PATH environment pmap_dump command, 465 677-679
variable, 326-329 Pointer section (XFree86 pppd daemon, 673-675
Path field (Active UNIX file), 518-519 requirements, 664
Domain Sockets), 633 -pointopoint option security, 676-677
pathnames, 408-409 (ifconfig command), 623 pppd daemon, 673-675
pdksh shell, 320 policies (firewall), 651-652 preemptive multitasking, 14
-percent option (rpm implementing, 653-654 Presentation layer (OSI
command), 169 testing, 654 model), 601
period (.), 227, 341, 409 what not to filter, 652 primary partitions, 455
Perl, 644 what to filter and
permissions, 177, 281 location, 652
904
print command (dip)
print command (dip), 666 nice command, 384-385 zombie, 383
-print option (find renice comamnd, 385-386 processing models
command), 433 private network address centralized-processing
printcap entries, 401 blocks, 658 systems, 186-188
printenv command, 325 Probe-Only mode distributed-processing
printer daemons, 393 (XFree86), 523 systems, 188-192
PRINTER variable, 400 problems. See bridges, 190
[printers] section (smb.conf troubleshooting file servers, 189
file), 493 proc directory, 442 gateways, 190
printers processes, 325 hubs, 189
configuring, 401-402 background, 368-369 NICs (network interface
field values, 404 background processing, cards), 189
filters, 404-405 343-344 repeaters, 189
local/remote, 403 at command, 344-345 routers, 190
printtool menu batch command, 345 topologies, 190-191
commands, 403 cron daemon, 344 workstations, 189
defined, 393 crontab command, PROFS email, 810
Linux requirements, 38-39 345-346 program installation,
selecting, 392 nohup command, 344 164-167, 175. See also
troubleshooting, 403, batch, 366 programs
405 child, 369 Debian Package
printing, 392-393, 397 connecting with pipes, 337 Management System, 174
directories, 396-397 continuing after logout, kernel upgrades, 178-179
email, 726 383-384 non-Linux software
/etc/printcap file, 393, daemons, 366 file permissions, 177
397-400 defunct, 383 gzip command, 175-177
filters, 398 exec, 331 make command, 177
permissions, 394 forks, 331 package formats, 175
printcap entries, 401 init, 331 troubleshooting, 178
printer daemons, 393 interactive, 366 uninstalling, 178
PRINTER variable, 400 monitoring, 379-383 RPM (Red Hat Package
printers multiple, 368-370 Manager)
configuring, 401-405 parent, 331, 369 finding packages, 168
defined, 393 PIDs (process IDs), 331 querying packages,
Linux requirements, scheduling, 370 172-173
38-39 at command, 370-372 uninstalling packages,
selecting, 392 batch command, 372-373 170-171
troubleshooting, 403, 405 cron command, 373 updating packages,
programs/commands crontab command, 171-172
lpc, 395-396 373-376 verifying packages, 173
lpd daemon, 394-395 priority, 384-386 system administrator
lpq, 395 sending signals to, 389-390 duties, 167
lpr command, 395 standard I/O (input/ terminology, 164
lprm, 395 output), 332 upgrades, 165, 198
spooling, 392 stopping (kill programs, 24-25. See also
troubleshooting, command), 386 commands
403, 405 normal termination, A News, 744
printtool menu 387-388 B News, 744
commands, 403 terminating all browsers, 703
priority (scheduled processes, 389 C News, 744
processes) unconditional courtney (Perl), 644
termination, 388-389 development tools, 18
905
protocols
FIPS (first nondestructive protocol suites. See TCP/ uploads, 691
interactive partition IP (Transmission Control wu-ftpd, 782, 789-790
splitting), 129 Protocol/Internet gopher, 710-711
GIMP (GNU Image Protocol) IP (Internet Protocol) . See
Manipulation Program), protocols, 702 IP addresses
548-549 AppleTalk, 312-314 LDAP (Lightweight
INN news, 744 archie, 708-709 Directory Access
LILO (Linux Loader) CHAP (Challenge Protocol), 481-482
installing, 114 Handshake client configuration, 484
Linux install, 95 Authentication database population,
uninstalling, 114 Protocol), 676 483-484
LISA, 100-101 DHCP (Dynamic Host installing, 482-483
mail. See mail program Configuration Protocol), Netatalk, 312-314
multimedia applications, 807 NNTP (Network News
158-159 DNS (Domain Name Transport Protocol), 744
Mutt, 739-740 Service) OSPF (Open Shortest Path
OpenLinux Base, 25 domain name spaces, 795 First), 611
Qt toolkit, 560 domains, 794 PAP (Password
rescue utilities, 240-241 hosts, 794 Authentication Protocol),
Seyon, 549-551 name servers, 795 676
smbstatus, 486 nodes, 794 POP (Post Office
SuperProbe, 515-516 resolvers, 795-798 Protocol), 718
switchdesk, 581 resource records, 800-802 PPP (Point-to-Point
testparm, 486 server configuration, Protocol)
timidity, 148 798-800, 803-807 accounts, 675-676
WordPerfect for Linux, 25 spoofing, 795 chat program, 671-673
xcalc calculator, 545 troubleshooting, 807-808 configuring with KDE (K
HP emulation, 547-548 ESMTP (Extended Simple desktop environment),
T1 emulation, 546-547 Mail Transfer 677-679
xlock, 551 Protocol), 815 pppd daemon, 673-675
xterm FTAM (File Transfer, requirements, 664
cursors, 543 Access, and security, 676-677
emulations, 542-543 Management), 599 RIP, (Routing Information
escape sequences, 543 FTP (File Transfer Protocol), 599, 610-611
mouse usage, 543-544 Protocol), 599, 706 SLIP (Serial Line Internet
xv anonymous FTP, Protocol). See also dip
buttons, 544-545 686-687, 693-695, driver
mouse usage, 545 706-707, 782-791 accounts, 670
prompt command (FTP), commands, 687-689 requirements, 664
688 directories, changing, 690 SMB. See Samba
prompts, 45, 320 downloads, 690-691 SMTP (Simple Mail
protecting resources, 776 file extensions, 706-707 Transfer Protocol),
Proto field remote directory 599, 720, 814-815
Active Internet listings, 690 SNMP (Simple Network
Connections, 631 sample session, 692 Management
Active UNIX Domain sessions, starting/ending, Protocol), 599
Sockets, 632 689-690 TCP (Transmission
Protocol option (XFree86 sites, 708 Control Protocol), 600
keyboard), 518 transfer modes, 691 Telnet, 600, 709-710
transfer status, 691 UDP (User Datagram
troubleshooting, 699-700 Protocol), 600
906
protocols
WAIS (Wide Area command), 172 CNAME, 802
Information Servers), 714 Qt toolkit, 560 fields, 801
XNS (Xerox Network querying packages, 172-173 HINFO, 802
Systems), 610 question mark (?) MX, 802
protocols.byname map FTP command, 689 name-server, 804
(NIS), 479 wildcard, 336 NS, 802, 806
protocols.bynumber map queues, 366 PTR, 802, 806
(NIS), 479 quit command, 666, SOA, 802
proxy command (FTP), 688 688, 789 recovery. See also backups
proxy firewalls, 636-638 quitting. See closing email messages, 729
packet filter firewalls, quotation marks ("), 340 files, 436-438
compared, 638-639 quote command (FTP), 688 recv command (FTP), 688
standard, 636 .qz file extension, 707 Recv-Q field (Active
transparent, 636 Internet
ps command, 197, 284, 331, Connections), 631
367, 379-383
R Red Hat Package Manager.
PS1 environment variable, :r command (vi), 227 See RPM
326, 329 R command (vi), 226 redirecting I/O (input/
PTR records, 806 -r option output), 337-338
PTR resource records, 802 crontab command, 376
redirection symbols, 338
Public Domain Korn Shell fsck command, 450
redraw option (vi), 228
(pdksh), 320 ipchains command, 646
Ref field (routing
publickey.byname map ls command, 422
tables), 627
(NIS), 479 mount command, 445
RefCnt field (Active UNIX
publicly writable spaces, netstat command, 630
Domain Sockets), 632
779-780 ps command, 381
RefererIgnore directive, 772
put command (FTP), rcp command, 698
RefererLog directive, 772
688, 691 ssh command, 699
Refresh Desktop command
pwck command, 291 wget command, 274
(KDE Root menu), 567
pwd command, 409, raw mode (device reget command (FTP), 688
688, 784 drivers), 323 regular expressions
PWD environment rawrite command, 66-67, 99 special characters, 341-342
variable, 326 Raymond, Eric S., 504 strings, 340
rc.local files, 237-238 relative pathnames, 409
relative permissions,
Q rc.sysinit file, 238
417-418
rc3.d directories, 236
:q command (vi), 227 rcp command, 697-698 remote connections, 616
-q option rdist command, 272-273 remote file transfers (FTP)
anonymous FTP, 686-687,
fdisk command, 452 read command, 352
693-695
gzip command, 176 reading
commands, 687-689
ls command, 422 email, 723
directories, changing, 690
rpm command, 172 current messages,
downloads, 690-691
ssh command, 699 724-725
remote directory listings,
wget command, 274 from files, 725-726
690
-qa option (rpm next message, 725
sample session, 692
command), 172 newsgroup articles, 750
sessions, starting/ending,
-qf option (rpm reboot command, 54, 245
689-690
command), 172 rebooting Linux, 115
transfer modes, 691
-ql option (rpm rec newsgroups, 745
transfer status, 691
command), 172 records (resource), 800
-qp option (rpm A, 802
907
RX-DRP field (kernel interface table)
troubleshooting, 699-700 NS, 802, 806 routes
uploads, 691 PTR, 802, 806 adding, 628
remote monitor SOA, 802 deleting, 630
(RMON), 599 restart command (FTP), 689 printing, 633
remote terminal sessions restarting. See also starting routing tables, 627
rcp, 697-698 kerneld daemon, 304-305 Routing Information
rlogin, 695-696 Linux, 54 Protocol (RIP), 599
rsh, 696-697 restore command, 272 rp field (/etc/printcap
ssh, 698-699 restoring files, 272 file), 399
telnet reverse resolution, 795 rpc.bynumber map
command-line options, RFCs (Requests for (NIS), 479
684-685 Comments), 605-606, rpc.mountd daemon, 468
example, 685-686 812-814 rpc.nfsd daemon, 468
$remote variable, 667 rhelp command (FTP), 689 RPCs (Remote Procedure
removing. See deleting ring topology, 191 Calls), 599
rename command RIP (Routing Information RPM (Red Hat Package
(FTP), 689 Protocol), 599, 610-611 Manager), 167-168
renaming files, 426 rlogin command, 695-696 finding packages, 168-169
renice command, 367, rm command, 51, 426-428 installing packages,
385-386 rm field (/etc/printcap 169-170
repartitioning hard drives, file), 399 querying packages,
22, 128 rmdir command, 50, 689 172-173
repeaters, 189 rmmod command, 303-304 uninstalling packages,
repeating commands, 225 RMON (remote 170-171
-replacefiles option (rpm monitor), 599 updating packages,
command), 169 $rmtip variable, 667 171-172
replacing text, 222-223 rn newsreader, 753-755 verifying packages, 173
replying to root accounts, 284 XFree86, 510
email, 729-731 root directories, 408, dependency errors, 524
newsgroup articles, 440, 764 optional RPMs, 511-512
750-751 root logins, 184 recommended RPMs,
report option (vi), 228 Root menu 510-511
Requests for Comments GNOME (GNU Network X server RPMs, 511
(RFCs), 605-606, 812-814 Object Model rpm command, 168
rescue disks, creating, Environment), 585 installation options, 169
66-67, 132-134 KDE (K desktop query options, 172
rescue utilities, 240-241 environment), 567-568 rsh command, 696-697
reset command, 666, 689 root passwords, 87, 277 rstatus command (FTP),
resolvers, 795 route program 688
/etc/host.conf files, command-line options, 626 rule specifications, 646
796-797 routes rulesets
/etc/resolv.conf files, adding, 628 sendmail, 821-822
797-798 deleting, 630 standard rulesets, 822
resolving names. See name routing tables, 627 syntax, 821
resolution routers, 190, 616 run levels, 235-239
resource records, 800 routing, 610, 626 runique command
A, 802 examples, 628-630 (FTP), 689
CNAME, 802 network segmentation, 611 running swat
fields, 801 policies, 626 inetd, 498
HINFO, 802 RIP (Routing Information from Web, 498-499
MX, 802 Protocol), 610-611 RX-DRP field (kernel
name-server, 804 route program, 626-630 interface table), 634
908
RX-ERR field (kernel interface table)
RX-ERR field (kernel processes, 370 search engines, 704-705
interface table), 634 at command, 370-372 telnet, 709-710
RX-OK field (kernel batch command, 372-373 WAIS (Wide Area
interface table), 634 cron command, 373 Information
RX-OVR field (kernel crontab command, Servers), 714
interface table), 634 373-376 secret keys, 676
priority, 384-386 Secure Sockets Layer
regular basis, 373 (SSL), 768
S sci newsgroups, 745 security
-s option screen, clearing, 52 Apache, 777
fsck command, 450
Screen section (XFree86 CGI scripts, 777-778
gzip command, 176
file), 521-522 publicly writable spaces,
ipchains program, 647
screen-capture programs, 779-780
ln command, 413
xv, 544-545 server-side includes, 778
ls command, 422
screensavers symbolic links, 778-779
GNOME (GNU Network breaches, 286
od command, 431
Object Model enforcing, 285-286
ps command, 381
Environment), 591 file permissions, 177, 414
S verification failure code
KDE (K Desktop changing, 415-416
(RPM), 173
Environment), 573 granting, 416-417
safrm shell script, 330
Samba, 486-487 script aliases, 764 relative permissions,
installing, 487
scripts 417-418
chat, 672-673 subfields, 415
server, 494-495
init command, 237 viewing, 414
shared directories, 493-494
majordomo, 712-714 file systems
smb.conf file
printcap entries, 401 permissions, 281
code listing, 487-491
shell scripts, 320, 348-349 SGIDs (Set Group
[global] section, 491-492
boot process, 236-238 IDs), 282
[homes] section, 492-493
comments, 351 SUIDs (Set User
[printers] section, 493
control structures, IDs), 282
testing, 494
355-361 modems, 285
smbclient program,
exit status, 356 networks, policy, 655
495-496
running commands, 345 NIS (Network Information
swat program
safrm, 330 Service), 481
command buttons, 497
special characters, 354 PAM (Pluggable
running from inetd, 498
testing, 358-360 Authentication Modules),
running from Web,
variables, 351-353 287-289
498-499
writing, 350-351 passwords
SampleRate option
(XFree86 pointers), 519 SCSI (small computer choosing, 278
sampling, 148 systems interface) devices/ dial-up passwords, 292
saving drivers, 34, 71 root, 87, 277
email, 728-729
search engines, 704-705 shadow passwords,
kernel versions, 301
Search.Com, search engine, 289-292
vi files, 213-214
705 physical, 276-277
sbin directory, 441 searching PPP (Point-to-Point
files, 221, 430 Protocol), 676-677
sbin/route program. See
Web social engineering,
route program
archie, 708-709 282-283
scheduling
Boolean expressions, SSL (Secure Sockets
background processes, 373
705-706 Layer), 768
backups, 260, 262-263
gopher, 710-711 system file logs, 283
keywords, 705 terminal lockouts, 285
909
shells
tips/guidelines, 292-294 #exec directive, 769 sessions, FTP (File Transfer
user accounts, 278 #flastmod directive, 770 Protocol)
accounts without #fsize directive, 770 closing, 690
passwords, 279 #include directive, 769 starting, 689
command accounts, 280 ServerFlags section set command, 229, 325, 734
default accounts, 279 (XFree86 file), 517 Set Group IDs (SGIDs), 282
group accounts, 280 ServerNumLock option set search command
guest accounts, 279-280 (XFree86 keyboard), 518 (archie), 709
passwd command, 291 servers, 702 Set User IDs (SUIDs), 282
root accounts, 284 Apache. See Apache Seyon, 549-551
unused accounts, 279 defined, 506 sf field (/etc/printcap
user monitoring, 284 file servers, 189 file), 400
segmenting networks, 611 FTP (File Transfer SGIDs (Set Group IDs), 282
selecting, text, 46 Protocol) sh field (/etc/printcap
semicolon (;), 226, 342 group files, 791 file), 400
send command, 666, 689 password files, 791 sh shell, 320, 325-330
Send-Q field (Active permissions, 790 shadow passwords, 289
Internet name servers, 795 command-line tools,
Connections), 631 NIS (Network Information 290-292
sending Service), 476 /etc/password files, 289
email, 719-721 Samba, 494-495 /etc/shadow files, 290
carbon copies, 732-733 Usenet, 828-829 Shadow Suite, 289
while reading WAIS (Wide Area grpck command, 292
messages, 726 Information Servers), 714 pwck command, 291
signals to processes, services, starting at boot, useradd command, 290
389-390 85-87 userdel command, 291
sendmail, 811 services.byname map usermod command, 290
aliases, 822 (NIS), 480 shared directories, 493-494
architecture, 817 session environment shared-file messaging, 810
configuring, 818-821 shell environment, 325 SheepShaver, 312
ESMTP (Extended Simple HOME variable, SHELL environment
Mail Transfer 326, 328 variable, 326
Protocol), 815 LOGNAME variable, shell scripts, 320, 348-349
handshaking, 814 326, 330 boot process, 236-238
historical overview, 817 MAIL variable, 329 control structures
mailing lists, 822-824 PATH variable, 326-329 case, 355-356
RFCs (Requests for PS1 variable, 326, 329 if, 357-358
Comments), 812-814 PWD variable, 326 iterative, 360-361
rulesets SHELL variable, 326 exit status, 356
standard rulesets, 822 TERM variable, running commands, 345
syntax, 821 326, 329 safrm, 330
running as daemon, TZ variable, 329 special characters, 354
817-818 terminal environment testing, 358-360
SMTP (Simple Mail control keys, 324-325 variables, 351-353
Transfer Protocol), device drivers, 322-323 writing
814-815 Session layer (OSI comments, 351
sendmail.cf file, 819 model), 601 echo command, 350-351
server-side includes, Session Message Block shell variables, substituting,
768-769 protocol. See Samba 339
Apache security, 778 Session PAM (Pluggable shells, 319
#config directive, 770-771 Authentication Modules) ash, 240-241
#echo directive, 770 module, 288 bash, 321
910
shells
Bourne, 320, 325-330 signatures (Usenet articles), SNMP (Simple Network
C, 320 744, 752 Management
command aliases, 363 SIGTERM signal, 389 Protocol), 599
command-line parsing, 332 SIGUSR1 signal, 389 SOA resource records, 802
customizing, 361-363 Simple Mail Transfer soc newsgroups, 745
defined, 319 Protocol (SMTP), 599, social engineering, 282-283
determining usage, 321 720, 814-815 sockets, SSL (Secure
environment Simple Network Sockets Layer), 768
configuration, 325 Management Protocol soft mounts, 472
HOME variable, (SNMP), 599 software. See programs
326, 328 site command (FTP), 689 software installation. See
LOGNAME variable, sites. See Web sites program installation
326, 330 size command (FTP), 689 software loopback
MAIL variable, 329 -size option (find interfaces, 624-625
PATH variable, 326-329 command), 433 sound
PS1 variable, 326, 329 slash (/), 49, 341, 408 CDs, 155
PWD variable, 326 Slashdot Web site, 589 playing, 156
SHELL variable, 326 sleep command (dip), 666 saving to MP3
TERM variable, SLIP (Serial Line Internet format, 156
326, 329 Protocol), 664 signatures, 156
TZ variable, 329 accounts, 670 troubleshooting, 162
escaping to, 430 dip driver GNOME (GNU Network
exporting variables, command mode, 665-666 Object Model
362-363 command-line options, Environment), 591
pdksh, 320 664-665 MIDI (musical instrument
prompts, 320 diplogin, 670 digital interface), 149
subshells, 342-343 dynamic IP addresses, sampling, 148
T, 320 669-670 sound cards, 146
troubleshooting, 364 /etc/diphosts file, 670-671 A/D (analog-to-digital)
show search command static IP addresses, converters, 148
(archie), 708 667-668 configuring, 149-152
showmatch option (vi), 228 variables, 667 D/A (digital-to-analog)
showmode option (vi), requirements, 664 converters, 148
228-230 sm option (vi), 228 drivers, 146-148
shutdown command, 54-55, smart terminals, 187 mixer, 149
244 SMB protocol. See Samba testing, 153-155
shutting down Linux, 54-55, smb.conf file troubleshooting, 160
243 code listing, 487-491 synthesized, 148
halt command, 245 [global] section, 491-492 troubleshooting, 161-162
reboot command, 245 [homes] section, 492-493 sound cards, 146
shutdown command, [printers] section, 493 A/D (analog-to-digital)
54-55, 244 testing, 494 converters, 148
troubleshooting, 244-247 smbclient program, 486, configuring
sigfiles, 744, 752 495-496 automated configuration
SIGHUP signal, 389 smbd daemon, 486 programs, 150
SIGKILL signal, 389 smbstatus utility, 486 hardware information,
-signal option (kill smd option (vi), 228-230 149
command), 387 smileys, 744, 748 kernel modules, 151-152
signals SMTP (Simple Mail D/A (digital-to-analog)
sending to processes, Transfer Protocol), 599, converters, 148
389-390 720, 814-815
signal-to-noise ratio, 743 sndconfig program, 150
911
system prompt
drivers, 146-147 status of swais clients, 714
ALSA (Advanced Linux network interfaces, 624 swap files
Sound Architecture), processes, 379-383 activating/
146 status command (FTP), 689 deactivating, 461
comparison, 147-148 status line (vi), 206 creating, 460
OSS (Open Sound sticky bits, 415 defined, 459
System), 146 sticky windows, 534 swap partitions, 456
mixer, 149 stopbits command (dip), 666 configuring, 140
testing, 153-155 stopping creating, 77-78, 111,
troubleshooting, 160 processes (kill 459-460
spam, 751 command), 386 defined, 459
spawning newsgroups, 827 normal termination, Linux requirements, 35
special variables, 327-328 387-388 swapoff command, 461
speed command (dip), 666 terminating all processes, swapon command, 460
$speed variable, 667 389 swat program
spoofing (DNS), 795 unconditional command buttons, 497
spooling, 392 termination, 388-389 running
srm.conf file, 764 Samba, 495 inetd, 498
ssh command, 698-699 strings, 340 Web, 498-499
sshd daemon, 238 struct command (FTP), 689 switchdesk command, 561,
SSL (Secure Sockets structures 581
Layer), 768 case, 355-356 switches, 617
Stallman, Richard, 862 control, 350 symbolic links, 413, 778-779
standard I/O (input/ decision structures, 355 synthesized sound
output), 332 if, 357-358 FM synthesis, 148
standard proxy firewalls, 636 iterative, 355, 360-361 wavetable synthesis, 148
star topology, 190 stty command, 324
starting stty sane command, 325 sys admin (systems
Apache, 765-766 su command, 283 administrator)
elm, 737-738 submitting How-To system administrators, 23,
fdisk, 452-454 documents, 859 47, 184, 192. See also
FTP (File Transfer subnet masks, 608-610 administration
Protocol) sessions, 689 subnetting, 608-610 backups, 287
kerneld daemon, 304-305 subscribe command defined, 164
Linux, 54 (majordomo), 713 hardware/software issues,
processes subscribing to 193-194
background processes, mailing lists, 714-716 peripherals, 195-196
368-369 Usenet newsgroups, role and duties, 184-185
multiple processes, 749-750 software installation duties,
368-370 subshells, 342-343 167
Samba, 495 substituting software upgrades, 198
vi, 207 command output, 353-354 system monitoring,
startup services, command results, 339-340 196-198
configuring, 85-87 shell variables, 339 system setup, 194-195
State field SUIDs (Set User IDs), 282 training, 198-199
Active Internet sunique command (FTP), system clock, 85
Connections, 632 689 system command
Active UNIX Domain SuperProbe utility, 515-516 (FTP), 689
Sockets, 633 superusers. See system System Commander, 41, 66
static IP (Internet Protocol) administrators system files. See files
addresses, 667-668 supplemental disks, 44 system mailbox, 718
surge suppressors, 276 system prompt, 45
912
system requirements
system requirements, 32-33 tape backups, 260 tcsh shell, 829
bus, 33 tape drives, 38 tear-off menus
CD-ROMs, 37 taper command, 263, (GNOME), 587
CPU, 33 269-270 technical support, 20-21
disk drives, 34-35 tar command, 263-268 telecommunications,
disk space, 22 advantages/ 549-551
memory, 34 disadvantages, 265 telinit command, 235
monitors, 35-36 command-line options, Telnet, 600, 684-686,
mouse devices, 38 265-266 709-710
network access, 37 examples, 266-268 Templates folder
OpenLinux install, 94 .tar file extension, 707 (KDE), 568
printers, 38-39 task list (GNOME), 584 tenex command (FTP), 689
swap space, 35 taskbar (KDE), 566-567 term command (dip), 666
tape drives, 38 tasks, 373 TERM environment
TCP (Transmission Control variable, 326, 329
Protocol), 600 TERM variable, 207
T TCP/IP (Transmission Terminal button
t command (vi), 457 Control Protocol/Internet (GNOME), 583
-t option Protocol), 598, 603 terminals,
configuring, 84 emulators, 541-542
cpio command, 268
history, 598 cursors, 543
dip command, 664
network connections, emulations, 542-543
fdisk command, 452
612-614 escape sequences, 543
fsck command, 450
gzip command, 176
TCP/IP networks mouse usage, 543-544
ls command, 422
(Transmission Control environment,
mkfs command, 458
Protocol/Internet configuring, 322
mount command, 445
Protocol), 620 control keys, 324-325
configuration files, 620 device drivers, 322-323
netstat command, 631
/etc/hosts file, 620-621 lockouts, 285
od command, 431
/etc/networks file, 622 virtual, 318, 600
ps command, 381
ssh command, 699
interfaces terminating commands, 342
tar command, 266
configuring, 625 terminating. See stopping
T shell, 320 initializing, 622 test command, 358-360
T verification failure code Parallel IP interfaces, 625 testing
reporting status of, 624 firewall policies, 654
(RPM), 173
software loopback Linux installation, 89
T1 emulation, 546-547
interfaces, 624-625 shell scripts, 358-360
Tab window manager,
statistics, 634 smb.conf file, 494
531-533
monitoring, 630-631 sound cards, 153-155
tables
inode, 411-412
active network testparm utility, 486
partition tables, 126
connections, 631-633 text
interface statistics, 634 cutting/pasting, 348
verifying, 455-456
kernel routing table, 633 selecting, 46
viewing, 454-455
routing tables
name resolution, 620-621 text editors
routing Emacs, 204
fields, 627
examples, 628-630 mail program, 722
printing, 633
policies, 626 vi, 204-205
routes, adding/deleting,
route program, 626 adding text, 217-219
628, 630
routes, adding/deleting, buffers, 206
viewing, 627
628, 630 changing/replacing text,
talk newsgroups, 745
routing tables, 627 222-223
Tannebaum, Andrew, 25, 27
tcpdump program, 641 closing, 210-212
913
undoing actions
command mode, 206-208 top command, 196-197 shutdown process, 244-247
command summary, topologies, 190-191 software installation, 178
226-228 Torvalds, Linus, 14, 27-28 sound cards, 160
creating files, 208-209 TOS (Type of Service), 649 sound problems, 161-162
cursor-positioning touch command, 434 vi, 210, 221, 231
commands, 215-217 trace command (FTP), 689 window managers, 551-552
cutting/pasting text, traceroute command, XFree86, 524-525
223-225 200-201 ttl field (resource
deleting text, 220-221 -trailers option (ifconfig records), 801
editing files, 209-210 command), 623 tty command, 413
editing process, 206-207 training system tty devices, 413
environment options, administrators, 198-199 Twm window manager,
228-230 translation modes 531-533
input mode, 206-208 (Seyon), 551 TX-DRP field (kernel
overwriting files, 215 Transmission Control interface table), 634
repeating commands, 225 Protocol/Internet TX-ERR field (kernel
saving files, 213-214 Protocol. See TCP/IP interface table), 634
searching, 221 transparency (network), 507 TX-OK field (kernel
searching/replacing text, transparent proxy interface table), 634
225 firewalls, 636 TX-OVR field (kernel
starting, 207 Transport layer (OSI interface table), 634
status line, 206 model), 601 type command (FTP), 689
troubleshooting, 210, trash (KDE), 568 Type field
221, 231 trn newsreader, 754 Active UNIX Domain
undoing changes, Troll Tech, Qt toolkit, 560 Sockets, 632
212-213 troubleshooting /etc/fstab file, 447
vim, 204 Apache, 766-767 resource records, 801
themes at command, 390 Type of Service, 649
GNOME (GNU Network boot process, 244-247 TZ environment
Object Model CDs, 162 variable, 329
Environment), 591-592 crontab command, 390
KDE (K Desktop device conflicts, 161
Environment), 573 DNS (Domain Name
U
X Window System, Service), 807-808
-u option
537-538 error messages, unknown
fdisk command, 452
threads, 750 PCI device, 245
ls command, 423
throttling, 766 file recovery, 436-438
netstat command, 631
tick command (FTP), 689 FTP (File Transfer
ps command, 381
tilde (~), 227 Protocol), 699-700
who command, 377
time slices, 366 IP (Internet Protocol)
U verification failure code
time-sharing, 366 masquerading, 660-661
(RPM), 173
time/date stamp, 434 kernel modules, 160
UDP (User Datagram
timeout command (dip), 666 Linux installation, 55-59,
Protocol), 600
timidity program, 148 88-89
umask command (FTP), 689
TIN newsreader, 829 networks, 200-201,
umount command, 448
TkZip program, 435 617-618
Unclutter Windows
tmp directory, 442 NFS (Network File
command (KDE Root
/tmp directories System), 472-473
menu), 568
(UNIX), 419 OpenLinux installation,
undeliverable email, 719
toggle options (vi), 230 115-118
undoing actions, 212-213
tools. See commands; printing/printers, 403-405
programs shells, 364
914
uniform resource locators (URLs)
uniform resource locators FAQs (Frequently Asked /usr directory, 419,
(URLs), 703-704 Questions), 743 443-444, 540
uninstalling flames, 743, 748 utilities. See commands;
LILO (Linux Loader), hierarchies, 745, 827-828 programs
114, 245 history, 744-745, 826-827
software, 170-171, 178 kill files, 753
UNIX, 14 moderated, 749
V
directories, 418-419 moderators, 830
v option
history, 26-27 netiquette, 747, 752-753
chat command, 671
USL (UNIX System news distributions,
dip command, 665
Laboratories), 26 746-747
fdisk command, 452
unknown PCI device error news feeds, 742
fsck command, 450
message, 245 news.announce.
gzip command, 176
unmounting file systems, newusers, 827
mkfs command, 458
448-449 newsgroup hierarchies, 746
mount command, 445
unsubscribe command newsreaders, 742-743
netstat command, 631
(majordomo), 713 NN, 829
od command, 431
unused security pine, 831-832
ps command, 381
accounts, 279 rn, 753-755
ssh command, 699
-up option (ifconfig TIN, 829
tar command, 266
command), 623 trn, 754
wget command, 274
update packages, 42-43 server programs, 828-829
V7 UNIX, 826
upgrading signal-to-noise ratio, 743
var directory, 443
kernel, 178-179 smileys, 744
variables, 350
software, 165, spam, 751
assigning values to,
171-172, 198 spawning, 827
351-353
uploading files, 691 subscribing to, 749-750
dip driver, 667
uptime command, 247 terminology, 743-744
environment, 325
URLs (uniform resource users, 743
HOME, 326-328
locators), 94, 703-704 user command (FTP), 689 IFS, 333
us distribution (Usenet), User Datagram Protocol LOGNAME, 326, 330
746 (UDP), 600 MAIL, 329
Use field (routing tables), user directories, 765 PATH, 326-329
627 User field (Active Internet PS1, 326, 329
Usenet newsgroups, 742, Connections), 632 PWD, 326
826, 837-839 user IDs, 763 SHELL, 326
accessing, 711 user-defined chains, 650 TERM, 326, 329
articles, 743 useradd command, 290 TZ, 329
conventions, 748 userdel command, 291 exporting to shells,
emoticons, 748 UserDir directive, 765 362-363
flames, 753 usermod command, 290 mail program, 735
headers, 750 users PRINTER, 400
posting, 743, 751-752 adding, 290
substituting, 339
reading, 750 accounts. See accounts
TERM, 207
replying to, 750-751 authentication, 287-289
verbose command
sigfiles, 744, 752 monitoring
(FTP), 689
threads, 750 finger command, 378-379
verifying
bandwidth, 743 ps command, 284
backups, 261
client programs, 829 who command, 376-378
packages, 173
compared to BBSes, 742 USL (UNIX System partition table, 455-456
culture, 748-749 Laboratories), 26 partitions, 457-458
ettiquette guidelines, 830
915
window managers
versions of Linux, 16-17 virtual memory, 508 BRU, 262
VertRefresh option virtual terminals (VTs), 318, Caldera, 94
(XFree86 monitor), 520 531, 600 FTP sites, 708
vi, 204-205 vmstat commmand, 198 Getting GNOME, 580
buffers, 206 VTSysReq option (XFree86 GNOME Manifesto, 579
closing, 210-212 keyboard), 518 gopher sites, 711
command summary, InterNIC, 605
226-228 Legatto systems, 262
cursor-positioning
W Linux Installation and
commands, 215-217 Getting Guide, 28
:w command (vi), 227
editing process, 206-207 Linux resources, 836-837
-w option, 226, 367, 457
environment options, MkLinux, 309
fdisk command, 452
228-230 multimedia resources,
mount command, 445
files 159-160
netstat command, 631
creating, 208-209 Mutt, 740
od command, 431
editing, 209-210 Open Source definition,
ps command, 381
overwriting, 215 872
WABI (Windows
saving, 213-214 SheepShaver, 312
Applications Binary
modes, 206-208 Slashdot, 589
Interface), 22
repeating commands, 225 ssh utility, 238
WAIS (Wide Area
searching, 221 wu-ftpd, 782
Information Servers), 714
starting, 207 XFree86 How-To, 504
wait command (dip), 666
status line, 206 Yellow Dog Linux, 310
warn option (vi), 228
text
wavetable synthesis, 148 Welcome screen (Linux
adding, 217-219
Web, 702 installation), 68
changing/replacing,
browsers, 703
Welsh, Matt, 839
222-223
clients, 702
wget command, 273-274
cutting/pasting, 223-225
links, 702
which command, 714
deleting, 220-221
protocols/services
while loops, 361
searching/replacing, 225
archie, 708-709
who command, 367,
troubleshooting, 210,
FTP (File Transfer
376-378, 714
221, 231
Protocol), 706-708
whois commmand, 198
undoing changes, 212-213
gopher, 710-711
Wide Area Information
video cards, 509-510 telnet, 709-710
Servers (WAIS), 714
video chipsets, 36 WAIS (Wide Area
widgets, 560
viewing Information Servers),
wildcards, 334
active network * (asterisk), 335
714
connections, 631-633 ? (question mark), 336
searching
command history, 45-46 [ ] expression, 336-337
Boolean expressions,
directory contents, 49
705-706
Window List (KDE), 563
file contents, 428
keywords, 705
Window Maker window
cat command, 429
search engines, 704-705
manager, 532, 535
less command, 52,
sendmail configuration
window managers, 531
429-430 AfterStep, 532, 535
files, 819
more command, 52, 429 Backbox, 536
servers. See servers
od command, 430-432 Blackbox, 532
structure, 702
file permissions, 281, 414 configuring, 536-537
URLs (uniform resource
kernel modules, 303 Enlightenment, 532, 536
locators), 703-704
partition table, 454-455 fvwm2, 532-534, 552-555
Usenet newsgroups,
routing tables, 627 fvwm95, 532, 535
accessing, 711
vim, 204 Web sites kwm, 532, 536
virtual desktops, 564-565 themes, 537-538
916
window managers
troubleshooting, 551-552 X Window System, 504-506 troubleshooting, 524-525
Twm, 531-533 client/server relationship, XF86Config file,
Window Maker, 532, 535 506-507 516-517
windows. See also window configuring, 83 xlock, 551
managers; X Window display managers, 538-539 xterm, 541-542
System GIMP (GNU Image cursors, 543
Applications Binary Manipulation Program), emulations, 542-543
Interface (WABI), 22 548-549 escape sequences, 543
geometries, 534 historical overview, 505 mouse usage, 543-544
sticky, 534 input capabilities, 508 xv, 544-545
wm= option (vi), 229 navigating, 530-531 X11 routines, 559
word search option (vi), 229 network transparency, 507 X11R6. See X Window
WordPerfect for Linux, 25 networking, 525-527 System
workstations, 189 output capabilities, 507 xcalc, 545
world distribution Seyon, 549-551 HP emulation, 547-548
(Usenet), 746 themes, 537-538 T1 emulation, 546-547
World Wide Web. See Web user interface capabilities, Xconfigurator, 514-515
:wq command (vi), 227 507-508 xdm display manager, 538
wrapmargin option (vi), 229 window managers, 531 XENIX, 27
writing AfterStep, 532, 535 Xerox Network Systems
email, 721 Backbox, 536 (XNS), 610
How-To documents, 858 Blackbox, 532 xf86config, 515-516
shell scripts configuring, 536-537 xf86config command, 523
comments, 351 Enlightenment, 532, 536 XF86Config file, 516
echo command, 350-351 fvwm2, 532-534, Device section, 521
special characters, 354 552-555 Files section, 517
variables, 351-353 fvwm95, 532, 535 Keyboard section, 517-518
ws option (vi), 229 kwm, 532, 536 Monitor section, 519-521
wu-ftpd, 782 troubleshooting, 551-552 Pointer section, 518-519
command-line options, 789 Twm, 531-533 Screen section, 521-522
configuring, 789-790 Window Maker, ServerFlags section, 517
installing, 782 532, 535 XF86Setup, 515
WWW (World Wide Web). xcalc, 545-548 XFree86, 17, 504. See also
See Web XFree86, 504 X Window System
configuring, configuring, 514
514-516, 523 SuperProbe utility,
X Device file, 521 515-516
hardware support, Xconfigurator, 514-515
x command (vi), 226
508-510 xf86config command, 523
X Consortium, 505
historical overview, 505 XF86Setup, 515
-x option
How-To document, 504 hardware support, 508-510
fdisk command, 453
installing, 508, 512-514 historical overview, 505
ipchains command, 647
Keyboard file, 517-518 How-To document, 504
ls command, 423
Monitor file, 519-521 installing, 508, 512-514
netstat command, 631
Pointer file, 518-519 Probe-Only mode, 523
ps command, 381
Probe-Only mode, 523 resolutions, 524
rcp command, 698
resolutions, 524 RPMs
rlogin command, 696
RPMs (Red Hat Package dependency errors, 524
rsh command, 697
Managers), optional RPMs, 511-512
ssh command, 699
510-512, 524 recommended RPMs,
tar command, 265
Screen file, 521-522 510-511
startup, 540-541 X server RPMs, 511
917
zones
startup, 540-541
troubleshooting, 524-525
XF86Config file, 516
Device section, 521
Files section, 517
Keyboard section,
517-518
Monitor section, 519-521
Pointer section, 518-519
Screen section, 521-522
ServerFlags section, 517
xlock, 551
XNS (Xerox Network
Systems), 610
xor argument, 648
xterm, 541-542
cursors, 543
emulations, 542-543
escape sequences, 543
mouse usage, 543-544
xterm sessions, 531
xv, 544-545
Y
-y option
chat command, 671
ipchains program, 648
Yahoo! search engine, 704
Yellow Dog Linux, 309-310
ypservers map (NIS), 480
Z
.Z file extension, 175, 707
-Z option
ipchains command, 646
tar command, 265
-z option (file command),
410
zcat command, 435
.zip file extension, 707
zless command, 436
zombie processes, 383
zones, 795
You might also like
- Linux For Developers - Jumpstart Your Linux Programming Skills (PDFDrive) PDF100% (3)Linux For Developers - Jumpstart Your Linux Programming Skills (PDFDrive) PDF224 pages
- Forensic APFS File Recovery: Jonas Plum Andreas DewaldNo ratings yetForensic APFS File Recovery: Jonas Plum Andreas Dewald11 pages
- Linux System Administration For The 2020s The Modern Sysadmin Leaving100% (9)Linux System Administration For The 2020s The Modern Sysadmin Leaving314 pages
- Red Hat Enterprise Linux-6-Storage Administration Guide-En-USNo ratings yetRed Hat Enterprise Linux-6-Storage Administration Guide-En-US224 pages
- Linux The Complete Manual - 2nd Edition, 2021100% (1)Linux The Complete Manual - 2nd Edition, 2021148 pages
- RedhatLinux Certified Professional Step-by-Step-guide Aravikumar48@gmail PDF100% (1)RedhatLinux Certified Professional Step-by-Step-guide Aravikumar48@gmail PDF112 pages
- Pub - Live Linux Cds Building and Customizing Bootables PDF100% (1)Pub - Live Linux Cds Building and Customizing Bootables PDF451 pages
- Hitchcock K. Linux System Administration For The 2020s. The Modern Sysadmin 2022No ratings yetHitchcock K. Linux System Administration For The 2020s. The Modern Sysadmin 2022349 pages
- CentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 SystemsFrom EverandCentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 SystemsNo ratings yet
- Special Edition Using Microsoft Windows XP Professional Robert Cowart - The ebook with all chapters is available with just one clickNo ratings yetSpecial Edition Using Microsoft Windows XP Professional Robert Cowart - The ebook with all chapters is available with just one click56 pages
- Special Edition Using Microsoft Windows XP Professional Robert Cowart all chapter instant download100% (14)Special Edition Using Microsoft Windows XP Professional Robert Cowart all chapter instant download60 pages
- Arman Danesh, Gautam Das-Special Edition Using Linux System Administration (Special Edition Using) - Que (2000)No ratings yetArman Danesh, Gautam Das-Special Edition Using Linux System Administration (Special Edition Using) - Que (2000)856 pages
- Special Edition Using Windows XP Professional Robert Cowart download100% (1)Special Edition Using Windows XP Professional Robert Cowart download47 pages
- Linux For System Administrators and Devops100% (1)Linux For System Administrators and Devops99 pages
- Linux Networking Cookbook 1st Edition Carla Schroderpdf download100% (1)Linux Networking Cookbook 1st Edition Carla Schroderpdf download50 pages
- Linux For System Administrators and DevopsNo ratings yetLinux For System Administrators and Devops99 pages
- Lenovo Iomega Ix4 300d NTWK Stor 4tb 70b89000na Users Manual 329743No ratings yetLenovo Iomega Ix4 300d NTWK Stor 4tb 70b89000na Users Manual 329743147 pages
- (Ebook) Linux Network Architecture by Klaus Wehrle ISBN 9780131777200, 0131777203 download pdfNo ratings yet(Ebook) Linux Network Architecture by Klaus Wehrle ISBN 9780131777200, 0131777203 download pdf75 pages
- Lenovo Iomega Ix4-300d NTWK Stor 4tb 70B89000NANo ratings yetLenovo Iomega Ix4-300d NTWK Stor 4tb 70B89000NA147 pages
- Foundations of Python Network Programming 1st Edition John Goerzen - Quickly download the ebook to never miss any content100% (1)Foundations of Python Network Programming 1st Edition John Goerzen - Quickly download the ebook to never miss any content35 pages
- Complete Download Linux Network Architecture Klaus Wehrle PDF All Chapters100% (12)Complete Download Linux Network Architecture Klaus Wehrle PDF All Chapters60 pages
- Sams Teach Yourself UNIX System Administration in 24 Hours100% (1)Sams Teach Yourself UNIX System Administration in 24 Hours525 pages
- (Ebook) Linux Kernel Development by Robert Love ISBN 9780672325120, 0672325128 pdf download100% (1)(Ebook) Linux Kernel Development by Robert Love ISBN 9780672325120, 0672325128 pdf download47 pages
- (Ebooks PDF) Download Special Edition Using TCP IP Niit (Usa) Inc. Full Chapters100% (16)(Ebooks PDF) Download Special Edition Using TCP IP Niit (Usa) Inc. Full Chapters75 pages
- 21 - SSUG17DE - Dominic Mueller-Wicke - Backup of Spectrum Scale Using Spectrum ArchiveNo ratings yet21 - SSUG17DE - Dominic Mueller-Wicke - Backup of Spectrum Scale Using Spectrum Archive34 pages
- IBM DB2 UDB Version 8 Update ConsiderationsNo ratings yetIBM DB2 UDB Version 8 Update Considerations7 pages
- AWS Certified Solutions Architect Associate Practice Test 6No ratings yetAWS Certified Solutions Architect Associate Practice Test 6159 pages
- Asghar Ghori - RHCSA & RHCE Red Hat Enterprise Linux 7 Training and Exam Preparation Guide - ErrataNo ratings yetAsghar Ghori - RHCSA & RHCE Red Hat Enterprise Linux 7 Training and Exam Preparation Guide - Errata4 pages
- All Stages of Linux Booting Process Explained0% (1)All Stages of Linux Booting Process Explained12 pages
- A Win32-Based Technique For Finding and Hashing NTFS Alternate Data StreamsNo ratings yetA Win32-Based Technique For Finding and Hashing NTFS Alternate Data Streams14 pages
- Ems Fault Tolerant Configuration On Amazon v2.7No ratings yetEms Fault Tolerant Configuration On Amazon v2.723 pages
- CIS Debian Linux 8 Benchmark v2.0.1 PDFNo ratings yetCIS Debian Linux 8 Benchmark v2.0.1 PDF507 pages
- Lecture 08, File Handling, Shared PreferencesNo ratings yetLecture 08, File Handling, Shared Preferences40 pages
- Instructions To Install CentOS Linux 9 On VMWare-2No ratings yetInstructions To Install CentOS Linux 9 On VMWare-222 pages
- Linux For Developers - Jumpstart Your Linux Programming Skills (PDFDrive) PDFLinux For Developers - Jumpstart Your Linux Programming Skills (PDFDrive) PDF
- Forensic APFS File Recovery: Jonas Plum Andreas DewaldForensic APFS File Recovery: Jonas Plum Andreas Dewald
- Linux System Administration For The 2020s The Modern Sysadmin LeavingLinux System Administration For The 2020s The Modern Sysadmin Leaving
- Red Hat Enterprise Linux-6-Storage Administration Guide-En-USRed Hat Enterprise Linux-6-Storage Administration Guide-En-US
- RedhatLinux Certified Professional Step-by-Step-guide Aravikumar48@gmail PDFRedhatLinux Certified Professional Step-by-Step-guide Aravikumar48@gmail PDF
- Pub - Live Linux Cds Building and Customizing Bootables PDFPub - Live Linux Cds Building and Customizing Bootables PDF
- Hitchcock K. Linux System Administration For The 2020s. The Modern Sysadmin 2022Hitchcock K. Linux System Administration For The 2020s. The Modern Sysadmin 2022
- Unix / Linux FAQ: with Tips to Face InterviewsFrom EverandUnix / Linux FAQ: with Tips to Face Interviews
- CentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 SystemsFrom EverandCentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 Systems
- Special Edition Using Microsoft Windows XP Professional Robert Cowart - The ebook with all chapters is available with just one clickSpecial Edition Using Microsoft Windows XP Professional Robert Cowart - The ebook with all chapters is available with just one click
- Special Edition Using Microsoft Windows XP Professional Robert Cowart all chapter instant downloadSpecial Edition Using Microsoft Windows XP Professional Robert Cowart all chapter instant download
- Arman Danesh, Gautam Das-Special Edition Using Linux System Administration (Special Edition Using) - Que (2000)Arman Danesh, Gautam Das-Special Edition Using Linux System Administration (Special Edition Using) - Que (2000)
- Special Edition Using Windows XP Professional Robert Cowart downloadSpecial Edition Using Windows XP Professional Robert Cowart download
- Linux Networking Cookbook 1st Edition Carla Schroderpdf downloadLinux Networking Cookbook 1st Edition Carla Schroderpdf download
- Lenovo Iomega Ix4 300d NTWK Stor 4tb 70b89000na Users Manual 329743Lenovo Iomega Ix4 300d NTWK Stor 4tb 70b89000na Users Manual 329743
- (Ebook) Linux Network Architecture by Klaus Wehrle ISBN 9780131777200, 0131777203 download pdf(Ebook) Linux Network Architecture by Klaus Wehrle ISBN 9780131777200, 0131777203 download pdf
- Foundations of Python Network Programming 1st Edition John Goerzen - Quickly download the ebook to never miss any contentFoundations of Python Network Programming 1st Edition John Goerzen - Quickly download the ebook to never miss any content
- Complete Download Linux Network Architecture Klaus Wehrle PDF All ChaptersComplete Download Linux Network Architecture Klaus Wehrle PDF All Chapters
- Sams Teach Yourself UNIX System Administration in 24 HoursSams Teach Yourself UNIX System Administration in 24 Hours
- (Ebook) Linux Kernel Development by Robert Love ISBN 9780672325120, 0672325128 pdf download(Ebook) Linux Kernel Development by Robert Love ISBN 9780672325120, 0672325128 pdf download
- (Ebooks PDF) Download Special Edition Using TCP IP Niit (Usa) Inc. Full Chapters(Ebooks PDF) Download Special Edition Using TCP IP Niit (Usa) Inc. Full Chapters
- 21 - SSUG17DE - Dominic Mueller-Wicke - Backup of Spectrum Scale Using Spectrum Archive21 - SSUG17DE - Dominic Mueller-Wicke - Backup of Spectrum Scale Using Spectrum Archive
- AWS Certified Solutions Architect Associate Practice Test 6AWS Certified Solutions Architect Associate Practice Test 6
- Asghar Ghori - RHCSA & RHCE Red Hat Enterprise Linux 7 Training and Exam Preparation Guide - ErrataAsghar Ghori - RHCSA & RHCE Red Hat Enterprise Linux 7 Training and Exam Preparation Guide - Errata
- A Win32-Based Technique For Finding and Hashing NTFS Alternate Data StreamsA Win32-Based Technique For Finding and Hashing NTFS Alternate Data Streams
- Instructions To Install CentOS Linux 9 On VMWare-2Instructions To Install CentOS Linux 9 On VMWare-2