Oracle DBA Code Examples

Oracle DBA Code Examples
Version Date: 30 July 2011 Editor: Ahmed Baraka
Page 1
Document Purpose This document is edited to be a quick reference of code examples used to achieve specific Oracle DBA tasks. No explanation on any subject is presented. The document is simply oriented based on the required task, the code to perform the task and any precautions or warnings when using the code. Also, it will be specified if the code is version specific. The document mainly demonstrates using SQL and PL/SQL code to achieve any task. It does not concentrate on using OEM to perform a task. However, if there will be a significant advantage, there could be just some hints on using OEM for some tasks.
Prerequisites The document assumes that the reader has already the knowledge of Oracle database administration.
How to Use the Document
1. Go to Contents section 2. Search the required task 3. Click on the required task link 4. Read the warnings and/or usage guideline, if any. 5. Make any modification in the code to match your case.
Oracle Database Versions The code presented in the document is to operate on Oracle database versions 10g and 11g. It will be stated, if the code is version specific.
Obtaining Latest Version of the Document Latest version can be obtained from my site or by emailing me at [email protected]
Usage Terms Anyone is authorized to copy this document to any means of storage and present it in any format to any individual or organization for non-commercial purpose free. No individual or organization may use this document for commercial purpose without a written permission from the editor. This document is for informational purposes only, and may contain typographical errors and technical inaccuracies. There is no warranty of any type for the code or information presented in this document. The editor is not responsible for any loses or damage resulted from using the information or executing the code in this document.
If any one wishes to correct a statement or a typing error or add a new piece of information,
please send the request to [email protected]
Page 2
Document Parts
Part 1 Part 2 Part 3 Part 4 Part 5 Part 6 Part 7 Part 8 Part 9 Part 10 Part 11 Part 12 Part 13
Oracle DBA Fundamentals _______________________ 36 Oracle Database Net Services ___________________ 238 Oracle Database Backup and Recovery ____________ 247 Oracle Database Security _______________________ 311 Oracle Database Performance Tuning _____________ 340 Oracle Automatic Storage Management (ASM) ______ 396 Oracle Real Application Cluster __________________ 410 Oracle RAC One Node __________________________ 529 Oracle Warehousing ___________________________ 551 Oracle Database Utilities _______________________ 556 Miscellaneous Oracle Database Topics _____________ 582 PL/SQL Samples ______________________________ 596 Appendixes__________________________________ 674
Page 3
Contents
Part 1
Oracle DBA Fundamentals _______________________ 36
DBA: Best Practices ____________________________________ 37

DBA: Best Practice Guidelines for Standalone and RAC Databases ____________37
Oracle Database Installation Consideration __________________ 38

Estimating Disk and Memoery Requirements ____________________________38 Optimal Flexible Architecture_________________________________________38 Oracle Products Installed with the 11.1 Release __________________________39
Installing Oracle 10g R2 on Enterprise Linux 4 _______________ 40

Installation Environment ____________________________________________40 Required Software _________________________________________________40 Used Hardware ___________________________________________________40 Installation Plan ___________________________________________________40
1. Preinstallation tasks __________________________________________________ 40 2. Oracle Database 10g Software Installation _________________________________ 44 3. Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software_____________ 44 4. Configure Listeners ___________________________________________________ 45 5. Create Database _____________________________________________________ 45 6. Postinstallation Tasks _________________________________________________ 45
Installing Oracle 11g R2 on Enterprise Linux 5 _______________ 47

1. Preinstallation tasks __________________________________________________ 47 2. Oracle Database 11g Software Installation _________________________________ 53 3. Apply Patchset_______________________________________________________ 53 4. Configure Listeners ___________________________________________________ 53 5. Create Database _____________________________________________________ 53 6. Postinstallation Tasks _________________________________________________ 53
Page 4
Installing Oracle 11g R2 on Enterprise Linux 5.5 with ASM ______ 54

1. Preinstallation tasks __________________________________________________ 55 2. Oracle Grid Infrastructure installation_____________________________________ 60 3. Oracle Grid Infrastructure Patching_______________________________________ 61 4. Oracle Database 11g R2 Software Installation ______________________________ 61 5. Oracle Database 11g R2 Software Patching ________________________________ 62 6. Install EM Agent in cluster nodes (if required) ______________________________ 62 7. ASM Diskgroups Creation ______________________________________________ 62 8. Database Creation ____________________________________________________ 62 9. Postinstallation tasks__________________________________________________ 62 10. General Useful Postinstallation Tasks in Linux _____________________________ 62
Managing Oracle Database Instance _______________________ 64

Product Release Number ____________________________________________64
Oracle Database Release Number Format ____________________________________ 64
Obtaining License Information________________________________________64 Managing the Instance Architecture ___________________________________64

Obtaining Information about the Instance Processes ___________________________ 64 Obtaining Information about the SGA _______________________________________ 65 Clearing the Buffer Cache ________________________________________________ 65
Database Administration Authentication ________________________________65

Using Operating System Authentication _____________________________________ 65 Using Password File Authentication _________________________________________ 66 Identifying Users SYSDBA or SYSOPER Users _________________________________ 66
Data Dictionary and Dynamic Performance Views _________________________66

Data Dictionary Creation _________________________________________________ 66
Startup and Shutdown______________________________________________66

Startup Levels _________________________________________________________ 66 Shutdown Levels _______________________________________________________ 66 Autostart of Database in Windows _________________________________________ 66 Automatically Starting Databases in Unix ____________________________________ 66
Page 5
Quiescing a Database ______________________________________________71 Suspending a Database _____________________________________________71 Dropping a Database _______________________________________________71 Initialization Files __________________________________________________71
Managing Initialization Files ______________________________________________ 71 Managing Parameters in SPFILE ___________________________________________ 72
Alert and Trace Files _______________________________________________72

Monitoring Alert and Trace Files ___________________________________________ 72
Managing Oracle Database Physical Structure ________________ 74

Managing Control Files______________________________________________74
Obtaining Control File information__________________________________________ 74 Creating Additional Copies, Renaming, and Relocating Control Files _______________ 74 Creating New Control Files _______________________________________________ 74 Backing Up Control Files _________________________________________________ 75 Manage the Size of Control Files ___________________________________________ 75 Multiplexing the Control File ______________________________________________ 76
Maintaining Online Redo Log Files _____________________________________76

Forcing Log Switches and Checkpoints ______________________________________ 76 Adding Online Redo Log File Groups ________________________________________ 76 Adding Online Redo Log File Members ______________________________________ 76 Dropping Online Redo Log File Groups ______________________________________ 76 Dropping Online Redo Log File Members _____________________________________ 76 Relocating and Renaming Redo Log Members_________________________________ 76 Verifying Blocks in Redo Log Files __________________________________________ 77 Clearing a Redo Log File _________________________________________________ 77 Viewing Redo Log Information_____________________________________________ 77
Managing Archived Redo Logs ________________________________________77

Obtaining Information about Archive Log ____________________________________ 77 Changing the Database Archiving Mode _____________________________________ 78 Specifying Archive Destinations and their Options _____________________________ 78 Specifying the Minimum Number of Successful Destinations _____________________ 79 Controlling Archiving to a Destination _______________________________________ 79 Controlling Trace Output Generated by the Archivelog Process ___________________ 79
Managing Tablespaces ______________________________________________79

Obtaining Tablespace Information__________________________________________ 79
Page 6
Creating a Locally Managed Tablespace _____________________________________ 83 Specifying Segment Space Management_____________________________________ 83 Adding Space to Tablespace ______________________________________________ 83 Specifying Nonstandard Block Sizes for Tablespaces ___________________________ 83 Using Bigfile Tablespace (BFT) ____________________________________________ 83 Using Temporary Tablespace______________________________________________ 84 Renaming a Tempfile ____________________________________________________ 84 Shrinking Temporary Tablespace __________________________________________ 84 Using Default Temporary Tablespace _______________________________________ 85 Using Temporary Tablespace Groups _______________________________________ 85 Suppressing Redo Generation for a Tablespace _______________________________ 85 Controlling Tablespaces Availability_________________________________________ 85 Using Read-Only Tablespaces _____________________________________________ 85 Renaming Tablespaces __________________________________________________ 85 Default Permanent Tabelspace ____________________________________________ 85 Dropping Tablespaces ___________________________________________________ 86 Managing the SYSAUX Tablespace _________________________________________ 86 Diagnosing and Repairing Locally Managed Tablespace Problems _________________ 86 Verifying the Integrity of Segments Created in ASSM Tablespaces. ________________ 86 Checking Consistency of Segment Extent Map with Tablespace File Bitmaps_________ 87 Verifying the Integrity of ASSM Tablespaces__________________________________ 88 Marking the Segment Corrupt or Valid ______________________________________ 89 Dropping a Corrupted Segment____________________________________________ 89 Dumping a Segment Header and Bitmap Blocks _______________________________ 90 Marking a DBA Range in Bitmap as Free or Used ______________________________ 90 Rebuilding the Appropriate Bitmap _________________________________________ 91 Rebuilding Quotas for Given Tablespace _____________________________________ 91 Migrating from a Dictionary-Managed to a Locally Managed Tablespace ____________ 91 Fixing the State of the Segments in A Tablespace _____________________________ 91 Scenario 1: Fixing Bitmap When Allocated Blocks are Marked Free (No Overlap) _____ 91 Scenario 2: Dropping a Corrupted Segment __________________________________ 92 Scenario 3: Fixing Bitmap Where Overlap is Reported __________________________ 92 Scenario 4: Correcting Media Corruption of Bitmap Blocks_______________________ 92 Transporting Tablespaces Between Databases ________________________________ 92
Managing Alert Thresholds __________________________________________94

Getting the Current Threshold Setting ______________________________________ 94 Setting Tablespace Alert Thresholds ________________________________________ 95 Restoring a Tablespace to Database Default Thresholds_________________________ 96
Page 7
Modifying Database Default Thresholds _____________________________________ 97 Viewing Alerts _________________________________________________________ 97
Managing Datafiles and Tempfiles _____________________________________98

Creating Datafiles ______________________________________________________ 98 Enabling and Disabling Automatic Extension for a Datafile _______________________ 98 Manually Resizing a Datafile ______________________________________________ 98 Bringing Datafiles Online or Taking Offline in ARCHIVELOG Mode _________________ 98 Taking Datafiles Offline in NOARCHIVELOG Mode ______________________________ 98 Renaming and Relocating Datafiles in a Single Tablespace_______________________ 99 Dropping Datafiles ______________________________________________________ 99 Copying a File on a Local File System _______________________________________ 99 Transferring a File to a Different Database __________________________________ 100 Dumping a Data Block __________________________________________________ 100
Managing Undo Tablespaces ________________________________________101

Obtaining Information on Undo ___________________________________________ 101 Enabling Automatic Undo Management _____________________________________ 102 Creating an Undo Tablespace ____________________________________________ 102 Setting Startup Undo Tablespace _________________________________________ 102 Tuning Undo Retention _________________________________________________ 102 Using Undo Advisor ____________________________________________________ 103 Setting the Undo Retention Period ________________________________________ 103 Enabling Retention Guarantee ____________________________________________ 103 Dropping an Undo Tablespace ____________________________________________ 103 To Drop a Corrupt UNDO Tablespace ______________________________________ 104
Using Oracle Managed Files (OMF) ___________________________________104
Managing Schema Objects ______________________________ 106

Chaching Small Tables in Memory ____________________________________106 Creating Virtual Columns ___________________________________________106 Creating Partitioned Tables _________________________________________106 Partition Maintenance Operations ____________________________________111 Setting Deferred Segment Creation___________________________________112 Creating Multiple Tables and Views in a Single Operation __________________112 Collecting Object Statistics _________________________________________113
Collecting Index Statistics _______________________________________________ 113 Collecting Table Statistics _______________________________________________ 113
Page 8
Collecting Schema Statistics _____________________________________________ 114
Validating Tables, Indexes, Clusters, and Materialized Views _______________115 Listing Chained and Migrated Rows of Tables and Clusters _________________115 Truncating Tables or Clusters _______________________________________116 Enabling and Disabling Triggers _____________________________________116 Managing Integrity Constraints ______________________________________116
Setting Constraint States and Deferability __________________________________ 116 Modifying, Renaming, or Dropping Existing Integrity Constraints ________________ 117 Reporting Constraint Exceptions __________________________________________ 117 Obtaining Information on Constraints ______________________________________ 118
Renaming Schema Objects _________________________________________118 Managing Object Dependencies ______________________________________118

Manually Recompiling Views Procedures and Packages ________________________ 118
Switching to a Different Schema _____________________________________118 Using DBMS_METADATA to Display Information About Schema Objects ______118 Specifying Storage Parameters at Object Creation _______________________118 Managing Resumable Space Allocation ________________________________119
Enabling Resumable Space Allocation ______________________________________ 119 Detecting Suspended Statements _________________________________________ 119 Obtaining Information about Suspended Statements __________________________ 121
Reclaiming Wasted Space __________________________________________121

Displaying Information About Space Usage for Schema Objects _________________ 121 Segment Advisor ______________________________________________________ 124 Shrinking Database Segments Online ______________________________________ 126 Deallocating Unused Space ______________________________________________ 126
Capacity Planning for Database Objects _______________________________127

Estimating the Space Use of a Table _______________________________________ 127 Obtaining Object Growth Trends __________________________________________ 128
Using the SQL Access Advisor _______________________________________129

Estimating the Space Use of a Table _______________________________________ 129
Managing Tables ______________________________________ 134

Obtaining Information about Tables __________________________________134 Creating Tables with some Options ___________________________________134
Page 9
DML Error Logging ________________________________________________135 Enabling Direct-Path INSERT ________________________________________135 Automatically Collecting Statistics on Tables____________________________136 Altering Tables ___________________________________________________136 Performing Online Redefinition with DBMS_REDEFINITION ________________137
Redefining a Table _____________________________________________________ 137 Redefining a Single Partition _____________________________________________ 139 Migrating BasicFile LOBs to SecureFiles ____________________________________ 140
Using Flashback Drop and Managing the Recycle Bin _____________________141 Managing Index-Organized Tables ___________________________________141 Managing External Tables __________________________________________142
Managing Indexes ____________________________________ 145

Using Indexes ___________________________________________________145 Using Bitmap Join Indexes (BJI) _____________________________________146 Partitioned Indexes _______________________________________________147
Managing Materialized Views ____________________________ 148

Obtaining Information about Materialized Views______________________________ 148 Monitoring the Progress of a Materialized View Refresh ________________________ 148 Materialized View Typical Refresh Errors____________________________________ 152 Using Materialized Views ________________________________________________ 152 Using Query Rewriting __________________________________________________ 155 ReWrite Hints_________________________________________________________ 157 Using EXPLAIN_MVIEW Procedure: Viewing Materialized View Capabilities _________ 157 Using DBMS_ADVISOR.TUNE_MVIEW ______________________________________ 157 Registering a User-defined Table as Materialized View _________________________ 158
Managing Clusters and Hash Clusters______________________ 159 Managing Views, Sequences, and Synonyms ________________ 161 Managing Transactions _________________________________ 162
Implementing Oracles Concurrency Control ____________________________162
Oracle Isolaction Levels _________________________________________________ 162 Oracle Lock Types _____________________________________________________ 162
Page 10
Identifying Blocking Sessions ____________________________________________ 163
Using Autonomous Transaction ______________________________________163 Managing Long Transactions with Workspace Manager____________________164
Repairing Corrupted Data _______________________________ 172

Options for Repairing Data Block Corruption ____________________________172 Detecting Corruptions Methods ______________________________________172 Using dbv (DBVerify) Utility_________________________________________172 Setting the Initialization Parameters for Detecting Corruption ______________172
Verifying Block Integrity in Real Time: DB_BLOCK_CHECKING __________________ 172 Verifying Block Integrity in Real Time: DB_BLOCK_CHECKSUM __________________ 173 Detecting lost write: DB_LOST_WRITE_PROTECT_____________________________ 173 Settubg the DB_ULTRA_SAFE Parameter (In Oracle 11g) ______________________ 173
Using ANALYZE Command __________________________________________173 Using EXP to Detect Corruption ______________________________________173 Using DBMS_REPAIR ______________________________________________174
DBMS_REPAIR Limitations and Restrictions _________________________________ 174 Evaluate the Costs and Benefits of Using DBMS_REPAIR _______________________ 174 Detect and Report Corruptions using DBMS_REPAIR __________________________ 175
Managing Automated Database Maintenance Tasks ___________ 177

Predefined Automated Maintenance Tasks _____________________________177 Predefined Maintenance Windows ____________________________________177 Obtaining Information about Predefined Maintenance Tasks________________177 Enabling and Disabling Maintenance Tasks _____________________________178 Configuring Maintenance Windows ___________________________________178
Managing Resources ___________________________________ 180

Obtaining Information on Database Resource Manager ___________________191 Monitoring Oracle Database Resource Manager _________________________193
Using Oracle Scheduler_________________________________ 195

Using Jobs ______________________________________________________195 Using Programs __________________________________________________202
Page 11
Using Schedules__________________________________________________203 Using Job Classes ________________________________________________204 Using Windows___________________________________________________205 Using Window Groups _____________________________________________206 Monitoring Job State with Email Notifications ___________________________207 Using File Watchers _______________________________________________209 Using Events Raised by the Scheduler_________________________________212 Using Events Raised by an Application (Events-Based Jobs)________________214 Using Chains ____________________________________________________217 Allocating Resources Among Jobs ____________________________________221 Administering Oracle Scheduler______________________________________221
Configuring Oracle Scheduler ____________________________________________ 221 Monitoring and Managing the Scheduler ____________________________________ 222 Enabling, Using and Disabling Remote External Jobs __________________________ 223 Import/Export and the Scheduler _________________________________________ 225 Scheduler Privileges____________________________________________________ 225 Scheduler Data Dictionary Views__________________________________________ 225 Using the UTL_FILE Package _____________________________________________ 226
Data Loading and Transforming Tools _____________________ 228 Using Database Links __________________________________ 229 Managing Diagnostic Data ______________________________ 230
Setting the Automatic Diagnostic Repository Directory ____________________230 Using adrci Tool __________________________________________________230
General usage of adrci__________________________________________________ 230 Using adrci to Package Incidents__________________________________________ 231
Managing Database Health Monitor ___________________________________231 Managing Data Recovery Advisor ____________________________________233 Using SQL Test Case Builder ________________________________________233
Patching Oracle Products _______________________________ 235

Using Oracle Opatch ______________________________________________235
Page 12
Part 2
Oracle Database Net Services ___________________ 238
Connectivity Naming Methods ___________________________ 239

The Local Naming Method __________________________________________239 The Easy Connect Naming Method ___________________________________239 The External Naming Method________________________________________239 The Directory Naming Method _______________________________________240 Database Resident Connection Pooling (DRCP) __________________________240
Oracle and Java Database Connectivity ____________________ 242

Establishing Database Connectivity ___________________________________242
Miscellaneous Connectivity Options _______________________ 244

Setting the Default Connect String ___________________________________244 Installing the Instant Client _________________________________________244 Setting Listener Options ___________________________________________244 Setting Access Controls ____________________________________________244 Changing Windows Hostname _______________________________________245
Part 3
Oracle Database Backup and Recovery ____________ 247
Backup Guidelines ____________________________________ 248

Causes of Unplanned Down Time ____________________________________248 Causes of Planned Down Time_______________________________________248 Oracles Solution to Down Time______________________________________249 Minimizing Unplanned Downtime Guidelines ____________________________249 SLA Sample _____________________________________________________250 Planning a Backup Strategy Guidelines ________________________________250
Examples of Backup Schedules for a Database_______________ 251 User-Managed Backups ________________________________ 252
Obtaining Database File Information __________________________________252
Making Whole Closed Database Backups ___________________________________ 252 Making a Whole Open Backup ____________________________________________ 252
Page 13
Making Tablespace Backups _____________________________________________ 253
Obtaining Backup Status Information _________________________________253 Checking Datafiles Taken as Backup _________________________________253 Handling Crash Before User-Manged Backup Ends _______________________253 Backing up Control File ____________________________________________254 Backing Up Initialization Files _______________________________________254
User-Managed Complete Recovery ________________________ 255

User-Managed Recovery in NOARCHIVELOG Mode _______________________255 User-Managed Recovery in NOARCHIVELOG Mode Without Redo Log File _____255 User-Managed Complete Recovery in ARCHIVELOG Mode _________________255 Re-Creating Lost Datafiles Without Backup _____________________________256
User-Managed Incomplete Recovery ______________________ 257

Common Situations Requiring Incomplete Recovery ______________________257 User-Managed Incomplete Recovery Steps _____________________________257 Recovering from Lost Control File by Re-Creating the Control File ___________257
Flash Recovery Area ___________________________________ 261

Obtaining Information on Flash Recovery Area __________________________261 Configuring Flash Recovery Area _____________________________________261 Backing Up the Flash Recovery Area __________________________________261 Moving the Flash Recovery Area _____________________________________262
Recovery Manager (RMAN)______________________________ 263

Using A Media Management Layer (MML) with RMAN _____________________263 Obtaining Information about and related to RMAN using Dictionary Views_____263 Starting RMAN ___________________________________________________265 Using rlwrap Utility with RMAN in Unix-Based Systems____________________266 Configuring the RMAN Environment __________________________________266 RMAN Channel Commands _________________________________________267 Duration in days of RMAN information in Control File _____________________268 Monitoring RMAN Jobs _____________________________________________268
Page 14
Using RMAN BACKUP Command _____________________________________268 Backing Up Control File and SPFile ___________________________________269 Backing Up Archived RedLogs _______________________________________269 Backup in NOARCHIVELOG Mode_____________________________________270 Encrypting RMAN Backups __________________________________________270 Using Compression in RMAN Backups _________________________________270 Using Multiplexed Backup Sets ______________________________________270 Using Parallelization of Backup Sets __________________________________270 Using Duplexed Backup Sets (Backupset Copies) ________________________271 Making Image Copies______________________________________________271 Validating Backup ________________________________________________272 Incremental Backup_______________________________________________272 Tags for Backups and Image Copies __________________________________272 Creating Archival Backups __________________________________________272 Monitoring RMAN Backups __________________________________________273 RMAN Complete Recovery _______________________________________273
Validating Backup Files _________________________________________________ 273 Previewing Backup Files Required by a Restore ______________________________ 274 Identifying Datafiles Requiring Recovery____________________________________ 274 Performing Complete Recovery ___________________________________________ 274 Restoring whole Database from RMAN Backups On a Different Node ______________ 275 Restoring whole Database from RMAN Backups from A 32 bit to 64 bit ____________ 278
RMAN Incomplete Recovery_________________________________________278 Simplified Recovery Through Resetlogs________________________________279 Recovering from Lost Control File using RMAN __________________________279 Block Media Recovery (BMR) ________________________________________280 Trial Recovery ___________________________________________________280 Handling Specific Errors During Recovery ______________________________280 Configuring Instance Crash Recovery Time (MTTR) ______________________282 Working with the Data Recovery Advisor in RMAN _______________________282 RMAN Maintenance _______________________________________________282
Cross Checking Backups and Copies _______________________________________ 282
Page 15
Deleting Backups and Copies ____________________________________________ 283 Changing the Availability of RMAN Backups and Copies ________________________ 283 Exempting a Backup or Copy from the Retention Policy ________________________ 283 The CATALOG Command ________________________________________________ 284 The CHANGE UNCATALOG Command ____________________________________ 284
RMAN Catalog ___________________________________________________284

Creating a Recovery Catalog _____________________________________________ 284 RMAN Catalog Reporting ________________________________________________ 285 Upgrading a Recovery Catalog ___________________________________________ 285 Importing Recovery Catalogs ____________________________________________ 285 Moving a Recovery Catalog ______________________________________________ 286 Dropping a Recovery Catalog ____________________________________________ 286
Virtual Private Catalogs ____________________________________________286 Using RMAN Scripts _______________________________________________287
Duplicating (Cloning) a Database _________________________ 289

Database Duplication (Cloning) Methods _______________________________289 Database Duplication Techniques ____________________________________289 Database Duplication Prerequisites ___________________________________289 Duplicating an Active Database using RMAN ____________________________289 Duplicating a Database without Recovery Catalog or Target Connection ______291 Manually Duplicating a Database_____________________________________293
Using Oracle Flashback Technology _______________________ 297 Flashback Options ____________________________________ 298 Preparing Your Database for Flashback ____________________ 299 Using Row Level Flashback Options _______________________ 300
Flashback Query _________________________________________________300 Flashback Versions Query __________________________________________300 Flashback Transaction Query ________________________________________301 Flashback Transaction (Backout) _____________________________________301
Using Table Level Flashback Options ______________________ 303

Page 16
Flashback Table __________________________________________________303 Flashback Drop __________________________________________________303 Flashback Data Archive ____________________________________________304
Using Flashback Database ______________________________ 308

When to use Flashback Database ____________________________________308 Flashback Database Considerations___________________________________308 Using Flashback Database __________________________________________308 Restore Points ___________________________________________________309
Part 4
Oracle Database Security _______________________ 311
Oracle Database Security Management ____________________ 312 Security Guidelines ____________________________________ 313 Managing Users ______________________________________ 315 Database Authentication _______________________________ 317
Managin Passwords _______________________________________________317 External (OS) Authentication ________________________________________317 Proxy Authentication ______________________________________________318 Logging In As a Different User_______________________________________318 Killing User Sessions from OS _______________________________________318
Controlling Database Access_____________________________ 320

System and Object Privileges _______________________________________320 Invoker Rights and Definer Rights____________________________________320 Roles __________________________________________________________320 Users, Roles, and Privileges Views____________________________________321 Fine-Grained Data Access (Virtual Private Database VPD) _________________322
Auditing Database ____________________________________ 325

Standard Auditing ________________________________________________325 Customizing Database Auditing with Triggers ___________________________326
Page 17
Auditing the Database Using System Trigger ___________________________328 Using Fine Grained Auditing_________________________________________331
Using Data Encryption _________________________________ 334

Oracle Transparent Data Encryption (TDE) _____________________________334 Tablespace Encryption _____________________________________________335
Fine-Grained Access Control for UTL_* Packages ____________ 337

Creating ACL ____________________________________________________337 Access Control Lists Maintenance ____________________________________338 Query Your Access Control List ______________________________________339
Part 5
Oracle Database Performance Tuning _____________ 340
Managing Performance Statistics _________________________ 341

Managing OS Statistics ____________________________________________341 Managing Database Statistics _______________________________________345
System and Session Statistics____________________________________________ 345 Time Model Statistics___________________________________________________ 346 Wait Events __________________________________________________________ 347 Active Session History (ASH)_____________________________________________ 349 Segment Statistics_____________________________________________________ 351
Handling Important Oracle Wait Events _______________________________351 List of Idle Waits Events ___________________________________________353 Using Performance Monitor Tool in Windows ____________________________355 Using OS Watcher for Windows ______________________________________355
Optimizing Performance in Windows Server ________________ 357

Optimizing Performance in Windows Server 2003________________________357
Tuning the Database Instance ___________________________ 358

Tuning the Shared Pool ____________________________________________358 Tuning the Buffer Cache ___________________________________________359 Tuning PGA _____________________________________________________361 Dumping the PGA ________________________________________________361
Page 18
Using Server Result Cache__________________________________________362 Obtaining Information about Object Locks _____________________________364 Handling a Hanging Database _______________________________________365 Accurately Measuring Process Size ___________________________________365
Managing Automatic Workload Repository (AWR) ____________ 367 Managing Automated Maintenance Tasks___________________ 369
Using Automatic Database Diagnostic Monitor (ADDM)____________________369 Using Automatic SQL Tuning Advisor _________________________________371
Implementing Automatic Memory Management______________ 374 Configuring DB_nK_CACHE_SIZE _________________________ 376 Managing Optimizer Operations __________________________ 377
Setting the Optimizer Mode _________________________________________377 Defining Access Paths and Joins for the Query Optimizer __________________377 Gathering Optimizer Statistics _______________________________________381
Gathering Object Statistics ______________________________________________ 381 Gathering System Statistics _____________________________________________ 382
Changing Statistics Preferences______________________________________382 Managing Pending and Published Statistics _____________________________383 Managing Extended Statistics _______________________________________384
MultiColumn Statistics __________________________________________________ 384 Expression Statistics ___________________________________________________ 385
A Simple Approach to Tuning SQL Statements _______________ 386 Using Application Tracing Tools __________________________ 387
Using the SQL Trace Facility and TKPROF ______________________________387 Using the Event 10046 to Trace SQL Code _____________________________387 Tracing End to End Application ______________________________________388 Enabling and Disabling Statistic Gathering for End to End Tracing ___________389
Page 19
Writing Efficient SQL __________________________________ 391 Improving SQL Processing Techniques_____________________ 394 Using SQL Tuning Advisor_______________________________ 395 Part 6 Oracle Automatic Storage Management (ASM) ______ 396
Managing Oracle ASM __________________________________ 397

Obtaining Information about ASM Instance_____________________________397 Creating an ASM Instance __________________________________________397 Managing Disk Groups and Disks in ASM_______________________________398
Fundementals of Managing Disk Groups and Disks____________________________ 398 Managing Disk Groups Attributes _________________________________________ 399
Monitoring Long-Running Operations _________________________________401 Migrating a Database to ASM________________________________________402 Moving a Tablespace to ASM ________________________________________402 Accessing an ASM instance from DB Console ___________________________403 Managing ASM Files _______________________________________________404 Using ASMCMD Utility _____________________________________________405 Using SYSASM Privilege and OSASM Group_____________________________407 Manually Upgrading Oracle AS from 10g to 11g _________________________408 Verifying Manually ASM Device ______________________________________408
Part 7
Oracle Real Application Cluster __________________ 410
Oracle RAC Possible Installation Configurations _____________ 411 Installing Oracle 10g R2 RAC on Enterprise Linux 4___________ 412
Installation Environment ___________________________________________412 Required Software ________________________________________________412 Used Hardware __________________________________________________412 Installation Plan __________________________________________________412
1. Preinstallation tasks _________________________________________________ 413 2. Oracle Clusterware installation _________________________________________ 420
Page 20
3. Oracle Database 10g Software Installation ________________________________ 424 4. Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software____________ 425 5. Install EM Agent in cluster nodes (if required) _____________________________ 426 6. Configure Listeners __________________________________________________ 426 7. Perform ASM installation ______________________________________________ 427 8. Perform cluster database creation ______________________________________ 427 9. Postinstallation tasks_________________________________________________ 429 10. Useful Postinstallation Tasks __________________________________________ 430
Installing Oracle 11g R2 RAC on Enterprise Linux 5___________ 431

Main Changes in Oracle 11g Release 2 RAC ____________________________431 Installation Environment ___________________________________________431 Required Software ________________________________________________431 Used Hardware __________________________________________________431 Installation Plan __________________________________________________432
1. Preinstallation tasks _________________________________________________ 433 2. Oracle Grid Infrastructure installation____________________________________ 440 3. Oracle Grid Infrastructure Patching______________________________________ 442 4. Oracle Database 11g R2 Software Installation _____________________________ 442 5. Oracle Database 11g R2 Software Patching _______________________________ 443 6. Install EM Agent in cluster nodes (if required) _____________________________ 443 7. ASM Diskgroups Creation _____________________________________________ 443 8. RAC Database Creation _______________________________________________ 444 9. Postinstallation tasks_________________________________________________ 445 10. General Useful Postinstallation Tasks in Linux ____________________________ 446
Installing Oracle 10g R2 RAC on Windows __________________ 447

Installation Methods ______________________________________________447 Installation Environment ___________________________________________447 Required Software ________________________________________________447 Used Virtual Hardware _____________________________________________447 Installation Plan __________________________________________________448
1. Preinstallation tasks _________________________________________________ 448 2. Oracle Clusterware installation _________________________________________ 450 3. Apply Patch Set 3 (10.2.0.4) on Clusterware software _______________________ 452 4. Oracle ASM 10g Software Installation____________________________________ 453
Page 21
5. Apply Patchset 3 (10.2.0.4) on ASM Software _____________________________ 454 6. Install EM Agent in cluster nodes (if required) _____________________________ 454 7. Configure Listeners __________________________________________________ 454 8. Create ASM Instance _________________________________________________ 455 9. Install Oracle RAC Database Home Software ______________________________ 456 10. Apply Patchset 3 (10.2.0.4) on Oracle RAC Software Home__________________ 457 11. Perform cluster database creation _____________________________________ 457 12. Useful Postinstallation Steps __________________________________________ 459
Cleaning Up Clusterware Installation on Windows____________ 460 Single Instance to RAC Conversion________________________ 462
The Tools to Convert a Single Instance DB to RAC _______________________462 Conversion Prerequisites for Oracle 10g R2 ____________________________462 Using rconfig Utitlity ______________________________________________462 Using DBCA _____________________________________________________463
Administering RAC Database ____________________________ 465 Administering Oracle Clusterware Components ______________ 466
Managing Cluserware Daemons and Processes __________________________466
Displaying Clusterware Processes _________________________________________ 466 Starting, Stopping, Enabling and Disabling crs Stack __________________________ 466
CSS Parameters __________________________________________________466 Administering Voting Disks in RAC ___________________________________466

Multiplexing Voting Disks________________________________________________ 466 Dynamically Adding and Removing Voting Disks after Installing RAC _____________ 467 Backing up Voting Disks ________________________________________________ 467 Recovering Voting Disks ________________________________________________ 467
Administering the Oracle Cluster Registry (OCR) ________________________467

Replacing the OCR _____________________________________________________ 468 Adding and Removing the OCR ___________________________________________ 468 Repairing the OCR _____________________________________________________ 468 Making Physical Backups of the OCR_______________________________________ 468 Recovering the OCR using the Physical Backups______________________________ 469 Making Logical Backups of the OCR (Exporting) ______________________________ 469 Making Logical Backups of the OCR (Importing)______________________________ 469
Page 22
Diagnosing OCR Problems with the OCRDUMP and OCRCHECK Utilities ____________ 470
Administering Storage _________________________________ 471

Datafile Access in Real Application Clusters _________________________________ 471 Redo Log File Storage in Real Application Clusters ____________________________ 471 Automatic Undo Management in Real Application Clusters ______________________ 471
Administering ASM Instances with SRVCTL in RAC _______________________472
Administering Cluster Databases _________________________ 473

Displaying Current Instance in SQL*Plus Prompt ________________________473 Starting and Stopping Instances and RAC Databases _____________________473
Starting Up and Shutting Down with SQL*Plus _______________________________ 473 Intermittent Windows Shutdown Issue in RAC Environments____________________ 473 Starting Up and Shutting Down with SRVCTL ________________________________ 473
Customizing How Oracle Clusterware Manages RAC Databases _____________474 Switching Between the Database Automatic and Manual Policies ____________474 Customizing Resource Parameters (like AUTO_START)____________________474 Handling Initialization Parameter Files in RAC___________________________475
Setting Server Parameter File Parameter Values for Real Application Clusters ______ 475 Parameters Used in RAC Databases _______________________________________ 475 Parameters that Must Have Identical Settings on All Instances __________________ 475 Parameters That Must Have Unique Settings on All Instances ___________________ 476 Parameters that Should Have Identical Settings on All Instances ________________ 476 ASM Instance Initialization Parameters and RAC _____________________________ 476
Dropping a RAC Database __________________________________________476
Workload Management in RAC ___________________________ 477

Types of Workload Distribution ______________________________________477 Connection Load Balancing _________________________________________477
Client-Side Load Balancing and Failover ____________________________________ 477 Server-Side Load Balancing______________________________________________ 477
Fast Application Notification (FAN) ___________________________________478 Using Fast Application Notification Callouts _____________________________478 Configuring the Server-Side ONS ____________________________________480 Administering Load Balancing Advisory ________________________________480
Page 23
Monitoring Load Balancing Advisory __________________________________481 Transparent Application Failover (TAF) ________________________________482
TAF Basic Configuration without FAN (From Client Side) _______________________ 482 TAF Basic Configuration with FAN (Server-Side)______________________________ 482 TAF Preconnect Configuration ____________________________________________ 482 Verifying TAF Configuration ______________________________________________ 483
Enabling Distributed Transaction Processing for Services __________________483
Administering Services _________________________________ 484

Service Attributes ________________________________________________484 Administering Services with DBCA____________________________________484 Administering Services with PL/SQL __________________________________484 Administering Services with SRVCTL __________________________________486 Controlling the Preferred and Available Instances ________________________486 Using Services with Client Applications ________________________________487 Services and the Scheduler _________________________________________487 Measuring Performance by Service Using the AWR _______________________488 Service Thresholds and Alerts _______________________________________490 Service Performance Views _________________________________________491 Restricted Session and Services _____________________________________491
Configuring Recovery Manager and Archiving _______________ 492

Backup Possible Distributions in RAC__________________________________492 RMAN Restore Scenarios for Real Application Clusters ____________________492
Cluster File System Restore Scheme_______________________________________ 492 Non-Cluster File System Restore Scheme ___________________________________ 492
RMAN and Oracle Net in Real Application Clusters _______________________492 Connecting to Specific Node ________________________________________492 Instance Recovery in Real Application Clusters __________________________493
Single Node Failure in Real Application Clusters ______________________________ 493 Multiple-Node Failures in Real Application Clusters____________________________ 493
Configuring the RMAN Snapshot Control File Location ____________________493 Configuring the RMAN Control File and SPFILE Autobackup Feature__________493
Page 24
Configuring Channels for RMAN in Real Application Clusters________________493

Configuring Channels to use Automatic Workload Balancing ____________________ 493 Configuring Channels to Use a Specific Instance _____________________________ 493
Node Affinity Awareness of Fast Connections ___________________________494 Archived Redo Log File Conventions in RAC ____________________________494 Archive Redo Log Configuration Scenarios _____________________________494
Automatic Storage Management and CFS Archiving Scheme ____________________ 494 Non-Cluster File System Local Archiving Scheme _____________________________ 494
Changing the Archiving Mode in Real Application Clusters _________________495 Deleting Archived Redo Logs after a Successful Backup ___________________495 Monitoring the Archiver Processes ____________________________________495 Log_Archive_Dest_1 Set To Default Even When DB_Recovery_File_Dest Is Set (Bug 6373164) __________________________________________________495 Media Recovery in Real Application Clusters ____________________________497 Parallel Recovery in Real Application Clusters ___________________________497 Using a Flash Recovery Area in RAC __________________________________497
Managing Backup and Recovery __________________________ 498 Administrative Options _________________________________ 499
Using Enterprise Manager Grid Control to Discover Nodes and Instances______499 Additional Information About SQL*Plus in RAC __________________________499
How SQL*Plus Commands Affect Instances _________________________________ 499 Displaying Running Instances ____________________________________________ 499 Displaying Connect Identifier_____________________________________________ 499
Quiescing RAC Databases __________________________________________500

Quiesced State and Cold Backups _________________________________________ 500
Transparent Data Encryption and Wallets in RAC ________________________500 Administering System and Network Interfaces with oifcfg _________________500
Defining Network Interfaces with oifcfg ____________________________________ 500 Syntax and Commands for the oifcfg Command-Line Tool ______________________ 500
Changing Public or Interconnect IP Subnet Configuration __________________501 Changing VIP Addresses ___________________________________________501
Adding Nodes and Instances on UNIX-Based Systems_________ 503

Page 25
Adding Nodes to a RAC Environment__________________________________503 Cloning Oracle Clusterware and RAC Software in Grid Environments _________503 Quick-Start Node and Instance Addition Procedures ______________________504
Adding an Oracle Clusterware Home to a New Node __________________________ 504 Adding an Oracle Home with RAC to a New Node _____________________________ 505
Detailed Node and Instance Addition Procedure _________________________505 Step 1: Connecting New Nodes to the Cluster___________________________505 Step 2: Extending Clusterware and Oracle Software to New Nodes __________505 Step 3: Preparing Storage on New Nodes ______________________________505 Step 4: Adding Nodes at the Oracle RAC Database Layer __________________506 Step 5: Adding Database Instances to New Nodes _______________________506
Deleting Nodes and Instances on UNIX-Based Systems________ 507

Option 1: Quick-Start Node and Instance Deletion Procedures______________507
Deleting an Oracle Home with RAC from an Existing Node ______________________ 507 Deleting an Oracle Clusterware Home from an Existing Node ___________________ 508
Option 2: Detailed Node and Instance Deletion Procedure _________________509 Step 1: Deleting DB Instances from Real Application Clusters Databases _____509
Using Enterprise Manager to Delete Database Instances from Existing Nodes_______ 509 Using DBCA in Interactive Mode to Delete Database Instances from Existing Nodes __ 509 Using DBCA in Silent Mode to Delete Instance from Existing Nodes_______________ 510
Step 2: Deleting Nodes from Real Application Clusters Databases ___________510 ASM Instance Clean-Up Procedures for Node Deletion ____________________511
Adding and Deleting Nodes and Instances on Windows-Based Systems ____________________________________________ 512
Cloning Oracle Clusterware and RAC Software in Grid Environments _________513 Quick-Start Node and Database Instance Addition and Deletion Procedures ___513
Adding an Oracle Clusterware Home to a New Node __________________________ 513 Adding an Oracle Home with RAC to a New Node _____________________________ 513 Deleting an Oracle Home with RAC from an Existing Node ______________________ 513 Deleting an Oracle Clusterware Home from an Existing Node ___________________ 513
Detailed Node and Database Instance Addition and Deletion Procedures ______513 Overview of Node Addition Procedures ________________________________513
Page 26
Step 1: Connecting New Nodes to the Cluster___________________________513

Making Physical Connections _____________________________________________ 513 Installing the Operating System __________________________________________ 513 Verifying the Installation with the Cluster Verification Utility ____________________ 513 Checking the Installation ________________________________________________ 513
Step 2: Extending Oracle Software to New Nodes at the Oracle Clusterware ___513 Step 3: Preparing Storage on New Nodes ______________________________513
Raw Device Storage Preparation for New Nodes ______________________________ 513
Step 4: Adding Nodes at the Oracle RAC Database Layer __________________513 Step 5: Adding Database Instances to New Nodes _______________________513
Using Enterprise Manager to Add Database Instances to New Nodes______________ 513 Using DBCA in Interactive Mode to Add Database Instances to New Nodes _________ 513 Using DBCA in Silent Mode to Add Database Instances to New Nodes _____________ 513 Connecting to iSQL*Plus after Adding a Node________________________________ 513
Adding Nodes that Already Have Clusterware and Oracle Software to a Cluster 514 Overview of Node Deletion Procedures ________________________________514 Step 1: Deleting Instances from Real Application Clusters Databases ________514
Using Enterprise Manager to Delete Database Instances from Existing Nodes_______ 514 Using DBCA in Interactive Mode to Delete Database Instances from Existing Nodes __ 514 Using DBCA in Silent Mode to Delete Instance from Existing Nodes_______________ 514
Step 2: Deleting Nodes from Real Application Clusters Databases ___________514 Step 3: ASM Instance Clean-Up Procedures for Node Deletion ______________514
Monitoring Performance ________________________________ 515

RAC Common Tuning Tips __________________________________________515 Instance Recovery and RAC_________________________________________515 Global Cache Wait Events __________________________________________515 Monitoring Performance in Enterprise Manager __________________________516
Using the Cluster Database Performance Page _______________________________ 516 Using the Cluster Database Instance Performance Page________________________ 516 Using the Cluster Performance Page _______________________________________ 517 Using the Cluster Interconnects Page ______________________________________ 517
Making Applications Highly Available Using Oracle Clusterware _ 518

Making an Application Highly Available Examples ________________________518
Page 27
Example1: Making an Application Highly Available ____________________________ 518 Example2: Making an Application Highly Available ____________________________ 520 Managing Automatic Oracle Clusterware Resource Operations for Action Scripts ____ 522 Displaying Clusterware Application and Application Resource Status Information ____ 522 Unregistering Applications and Application Resources _________________________ 523
RAC Troubleshooting __________________________________ 524

Diagnosing the Oracle Clusterware High Availability Components ___________524
Debugging Recommnedation_____________________________________________ 524 Clusterware Log Files and the Unified Log Directory Structure ___________________ 524 Dynamic Debugging____________________________________________________ 524 Component Level Debugging_____________________________________________ 524 Oracle Clusterware Shutdown and Startup __________________________________ 525 Enabling and Disabling Oracle Clusterware Daemons __________________________ 525 Diagnostics Collection Script _____________________________________________ 525 The Oracle Clusterware Alerts ____________________________________________ 525 Resource Debugging ___________________________________________________ 525 Checking the Health of the Clusterware ____________________________________ 525 Troubleshooting the Oracle Cluster Registry _________________________________ 525 Troubleshooting Hostname Changes and CSS________________________________ 525 Enabling Additional Tracing for Real Application Clusters High Availability__________ 526 Diagnosing Oracle Real Application Clusters Components ______________________ 526 Where to Find Files for Analyzing Errors ____________________________________ 526 Using Instance-Specific Alert Files in Real Application Clusters __________________ 526 Enabling Tracing for Java-Based Tools and Utilities in Real Application Clusters _____ 526 Resolving Pending Shutdown Issues _______________________________________ 526
Using the Cluster Verification Utility __________________________________526

Cluster Verify Locations _________________________________________________ 526 Cluster Verify Stages ___________________________________________________ 526 Cluster Verify Components ______________________________________________ 527 CVU Component Verification Examples _____________________________________ 527 Understanding CVU Commands, Help, Output, and Nodelist Shortcuts ____________ 528 Performing Various CVU Tests ____________________________________________ 528 Known Issues for the Cluster Verification Utility ______________________________ 528
Part 8
Oracle RAC One Node __________________________ 529
Installing Oracle 11g R2 RAC One Node on Linux 5 ___________ 530

Page 28
Installation Environment ___________________________________________530 Required Software ________________________________________________530 Used Hardware __________________________________________________530 Installation Plan __________________________________________________531
1. Preinstallation tasks _________________________________________________ 532 2. Oracle Grid Infrastructure installation____________________________________ 539 3. Oracle Grid Infrastructure Patching______________________________________ 542 4. Checking Oracle Grid Infrastructure Status _______________________________ 542 5. Oracle Database 11g R2 Software Installation _____________________________ 542 6. Oracle Database 11g R2 Software Patching _______________________________ 543 7. Install EM Agent in cluster nodes (if required) _____________________________ 544 8. ASM Diskgroups Creation _____________________________________________ 544 9. RAC Database Creation _______________________________________________ 545 10. Initialize the Database to RAC One Node (11.2.0.1 Only) ___________________ 547 11. Postinstallation tasks________________________________________________ 548 12. General Useful Postinstallation Tasks in Linux ____________________________ 548
Instance Relocation using Omotion (11.2.0.1)_______________ 549 Instance Relocation using Omotion (11.2.0.2)_______________ 550 Part 9 Oracle Warehousing ___________________________ 551
Oracle Warehouse Builder (OWB) ________________________ 552

Oracle Warehouse Builder Architecture ________________________________552 Starting and Stoping the Service_____________________________________552 Configuring the Repository and Workspaces ____________________________552 Steps of Using Warehouse Builder____________________________________553 Mapping Operators _______________________________________________554
Part 10
Oracle Database Utilities _______________________ 556
Using SQL*Plus_______________________________________ 557

Using SQL*Plus Command-Line Options _______________________________557 Starting SQL*Plus Session__________________________________________557 Controlling User Privileges in SQL*Plus ________________________________557
Page 29
Setting the SQL*Plus Environment with the SET Command ________________558 Setting SQL*Plus Preferances _______________________________________558 Logging SQL*Plus Errors ___________________________________________559 Key SQL*Plus "Working" Commands __________________________________559 Creating Command Files in SQL*Plus _________________________________559 Copying Tables with the COPY Command ______________________________559 Creating Web Pages Using SQL*Plus __________________________________560 Using SQL to Generate SQL _________________________________________560 Enabling AUTOTRACE for a User _____________________________________560 Using rlwrap Utility with SQL*Plus in Unix-Based Systems _________________561 Escaping Special Characters ________________________________________561
Using SQL*Loader Utility _______________________________ 563

Invoking SQL*Loader______________________________________________563 Using SQL*Loader Control File ______________________________________563 Loading Excel File into a Table using SQL*Loader ________________________569 Loading Large Fields into a Table ____________________________________569 Using Direct Load Options __________________________________________570
Data Pump Export and Import ___________________________ 571

Data Pump Components ___________________________________________571 Data Pump Export Interfaces________________________________________571 Export Modes Parameters __________________________________________571 Required Rrivileges _______________________________________________571 Invoking Export Data Pump Examples_________________________________571 Export Filtering Parameters _________________________________________572 Export Remapping Parameters ______________________________________572 Sampling Export Data _____________________________________________573 Export Encryption Parameters _______________________________________573 Export Estimating Parameters _______________________________________573 Export Network Link Parameter ______________________________________573 Export PARALELL Parameter ________________________________________574
Page 30
Import Modes Parameters __________________________________________574 File- and Directory-Related Parameters________________________________574 Using TABLE_EXISTS_ACTION Parameter ______________________________574 Import Filtering Parameters_________________________________________574 Import Remapping Parameters ______________________________________574 Ignoring Nondeferred Constraints ____________________________________575 Import Network Link Parameter _____________________________________575 Import Flashback Parameters _______________________________________575 Import PARALELL Parameter ________________________________________576 Monitoring a Data Pump Jobs _______________________________________576
LogMiner ____________________________________________ 577

Types of Supplemental Logging______________________________________577 Levels of Supplemental Logging _____________________________________577 Disabling Database-Level Supplemental Logging ________________________577 LogMiner Dictionary Options ________________________________________578 Redo Log File Options _____________________________________________578 OPTIONS possible values in DBMS_LOGMNR.START_LOGMNR: _____________578 Obtaining LogMiner Operational Information____________________________578 Examples of Using LogMiner ________________________________________579
Without Sepecifying the Redo Files ________________________________________ 579 Wit Sepecifying the Redo Files ___________________________________________ 580
Part 11
Miscellaneous Oracle Database Topics _____________ 582
Managing Oracle Database Control________________________ 583

Configuring and Using the Database Control ____________________________583 Implementing EM Database Control Auto Startup ________________________583 Dropping and Recreating the Management Repository ____________________584
Installing Oracle 10g R5 (10.2) Enterprise Manager Grid Control for Linux x86 ___________________________________________ 585
Installation Environment ___________________________________________585
Page 31
Required Software ________________________________________________585 Used Hardware __________________________________________________585 Installation Steps_________________________________________________585
Remote Diagnostic Agent (RDA)__________________________ 593

Using Remote Diagnostic Agent (RDA) ________________________________593
Connect Oracle to SQL Server____________________________ 594

Configuring and Using the Database Control ____________________________594
Part 12
PL/SQL Samples ______________________________ 596
PL/SQL Basics _______________________________________ 597

PL/SQL Data Types _______________________________________________597 Controlling Compile-Time Displayed Warnings __________________________598 Catching Returned Errors __________________________________________599 Hiding Code _____________________________________________________599 Controlling Program Flow___________________________________________599
Using Cursors ________________________________________ 602 Using Records ________________________________________ 606 Using Table Functions__________________________________ 608 Using Collections _____________________________________ 610
Using VARRAYS __________________________________________________610 Using Nested Tables ______________________________________________613 Using Associative Arrays ___________________________________________614 Using Collection API_______________________________________________616
Handling Errors_______________________________________ 621

Predefined Exceptions _____________________________________________621 Using User-Defined Exceptions ______________________________________621 Using RAISE_APPLICATION_ERROR __________________________________622
Page 32
Autonomous Transactions ______________________________ 624 Some Stored Subprobrams Concepts ______________________ 625
Serially Reusable Packages _________________________________________625 Stored Subprograms and Roles ______________________________________626 Invokers vs. Definers Rights _______________________________________626 Pinning an Programunit in the Shared Pool _____________________________627
Using Triggers _______________________________________ 628

Restrictions on Triggers _________________________________________________ 628
Using DML Triggers _______________________________________________628 Using Instead-of Triggers __________________________________________629 Using System Triggers_____________________________________________630 Handling Mutating Tables in Triggers _________________________________634 Dropping and Disabling Triggers _____________________________________635
Using Dynamic SQL____________________________________ 636

Working with Native Dynamic SQL ___________________________________636 Using DBMS_SQL _________________________________________________639
Calling Java from PL/SQL _______________________________ 648 Configuring Oracle Database to Use External Routines ________ 649 Using Large Objects (LOBs) _____________________________ 652
Creating LOB ____________________________________________________652 Using SQL with Internal LOBs _______________________________________652 Using LOBs in PL/SQL _____________________________________________653 Performance Considerations ________________________________________662
Using Returning Clause _________________________________________________ 662 Using CONTEXT Index __________________________________________________ 663
Migrating from LONGs to LOBs ______________________________________664
PL/SQL Performance Tuning Tips _________________________ 665

Use PL/SQL Profiler _______________________________________________665
Page 33
Use BULK COLLECT Clause _________________________________________666 Set PLSQL_OPTIMIZE_LEVEL and Subprogram Inlining ___________________670 Using Bind variables in Dynamic SQL _________________________________671 Use NOCOPY Keyword _____________________________________________672 Use Associative arrays _____________________________________________672 Use Server Result Cache ___________________________________________672
PL/SQL Miscellaneous Topics ____________________________ 673

Accessing V$ Views from PL/SQL_____________________________________673
Part 13
Appendixes__________________________________ 674
Program Units and Scripts Used in the Document ____________ 675

Return Parameter Value for Normal User ______________________________675 Applying Random Load on Database Sample 1 __________________________676
Setup _______________________________________________________________ 676 Using the Load Generator Scripts _________________________________________ 682
Applying Random Load on Database Sample 2 __________________________683

Setup _______________________________________________________________ 683 Using the Load Generator Scripts _________________________________________ 691
SQL Usage Samples ___________________________________ 693

Merge Command _________________________________________________693 Multitable Inserts _________________________________________________693 Parallel Insert ___________________________________________________694 Deleting Duplicate Rows in a Table ___________________________________694 Adding a Primary Key to a Table and then Filling it_______________________694
More DBA Scripts _____________________________________ 696

Compare Table Sizes in Two Databases _______________________________696
Using Linux for Oracle _________________________________ 697

Verifying the Kernel _______________________________________________697 Checking for a Tainted Kernel _______________________________________697
Page 34
Supported Hardware ______________________________________________697 Using Oracle Relink Utility __________________________________________697 Certified and Supported File Systems _________________________________698 Enterprise Linux Runlevels__________________________________________698 Using /etc/oratab File and dbstart Utility_______________________________698 Automating Jobs _________________________________________________699
Using cron ___________________________________________________________ 699 Using anacron ________________________________________________________ 700 Using at command_____________________________________________________ 700 Using batch command __________________________________________________ 701 Task Scheduler _______________________________________________________ 701 Configuring Linux Memory for Oracle ______________________________________ 701
Using Linux Performance Monitoring Tools _____________________________701

About Linux Tools _____________________________________________________ 702 Using Linux Tools______________________________________________________ 702
Checking Some General Guideline on Truning Oracle in Linux ______________703 Troubleshooting Oracle Database in Linux _____________________________704
Using OS Watcher (OSW) _______________________________________________ 704 Using OS Watcher Graphs(OSWg)_________________________________________ 706 Using the On-Board Monitor (LTOM) _______________________________________ 706 Using strace __________________________________________________________ 707
Page 35
Part 1
Oracle DBA Fundamentals
Page 36
DBA: Best Practices

DBA: Best Practice Guidelines for Standalone and RAC Databases
Use SPFILE Use temporary tablespaces Register all instances with remote listeners Use Database Resource Manager Use resumable space allocation Use Automatic Segment Space Management Use locally managed tablespaces Use Automatic Undo Management Enable block checking Enable Flashback Database Enable ARCHIVELOG mode and use a flash recovery area Use auto-tune checkpointing Log checkpoints to the alert log Multiplex production and standby redo logs Set CONTROL_FILE_RECORD_KEEP_TIME long enough Create two or more control files
Page 37
Oracle Database Installation Consideration

Estimating Disk and Memoery Requirements
o o o Make sure the storage vendor is listed in the Oracle Storage Compatibility Program (OSCP). Allocate swap space that is about two to three times your Oracle RAM allocation. On Estimating the RAM: OS RAM: 20 percent of total RAM for MS-Windows, 10% of RAM for UNIX Each Oracle connection consumes two megabytes of RAM + sort_area_size + hash_area_size If dynamic memory sizing is not used: Shared_pool_size: If all the SQL statements that sent to ORACLE are using bind variable adequately, then 300M is enough in most cases Large_pool_size: For dedicated Oracle server: 20-30M Java_pool_size=10M Data buffer: All the rest RAM should be allocated to Data buffer.
Optimal Flexible Architecture

The OFA is a set of recommendations from Oracle Corporation aimed at simplifying management of complex software and databases often running under multiple versions of software. OFA Guidelines o o Arrange at least four mount points that have names like /u01, /u02, /u03, and so on. If the Oracle software owner is "oracle", make /u01/app/oracle your ORACLE_BASE # mkdir -p /u01/app/oracle # chown -R oracle:oinstall /u01/app/oracle # chmod -R 775 /u01/app/oracle Make Oracle_HOME $ORACLE_BASE/product/version/{db|client|companion}_[n]. Examples: /u01/app/oracle/product/11.1.0/db_1 /u01/app/oracle/product/11.1.0/client_1 /u01/app/oracle/product/10.1.0.2.0/db_1 Install Flash Recovery Area in $ORACLE_BASE/flash_recovery_area in a mount point different from db files. Administrative Files are to be created in ORACLE_BASE/admin/dbname Store Oracle DB files in the mounted points. # mkdir /mount_point/oradata # chown oracle:oinstall /mount_point/oradata # chmod 775 /mount_point/oradata /u01/oradata/prod1/control01.ctl /u05/oradata/prod1/control02.ctl /u02/oradata/prod1/redo01.log /u04/oradata/prod1/redo02.log data file has this format: /u20/oradata/prod/system01.dbf
Page 38
o o o
Oracle Products Installed with the 11.1 Release

The following products are installed by default with database server 11g: o o o o o Oracle Application Express Oracle Warehouse Builder Oracle Configuration Manager: Tool that collects and uploads configuration information to the Oracle configuration repository Oracle SQL Developer Oracle Database Vault: Tool that enables you to secure business data
Page 39
Installing Oracle 10g R2 on Enterprise Linux 4

Installation Environment
Emulation software: VMWare Server 2 on for Windows OS: Oracle Linux Enterprise 4.5 for x86: kernel 2.6.9
Required Software
Oracle Database 10g Release 2 for Linux x86 32-bit
Used Hardware
In the VMWare: create one virtual machine (oradb1) with the following specs: o 2 GB RAM o an ethernet card o one local hardisk with 20 GB
Installation Plan
1. Preinstallation tasks: o o o 2. 3. 4. 5. 6. Hardware requirements Software requirements Environment configuration
Oracle Database 10g Software Installation Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software Configure Listeners Create Database Postinstallation tasks
1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Give the ethernet card IP 192.168.4.21 and the hostname oradb1.mydomain.com. Define a gateway. If it does not exist, make it same as the host IP address. o Insall the following packages: Desktop Environments o GNOME Desktop Environment
Page 40
Desktop o X Window System o Gnome
Applications o Graphical Internet (optional)
Servers o Do not select anything in this group.
Development o Development Tools
System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
Miscellaneous o Do not select anything in this group.
Complete the installation Install further packages: # to know distribution and version of Linux cat /etc/issue # to know kernel version (and its errata level) uname -r # from CD 3 rpm -Uvh libaio* rpm -Uvh openmotif21-2.1.30-11.RHEL4.6.i386.rpm rpm -Uvh openmotif-2.2.3-10.1.el4.i386.rpm # those packages downloaded from http://rpm.pbone.net rpm -e compat-libstdc++-296-2.96-132.7.2 rpm -Uvh compat-libstdc++-7.3-2.96.128.i386.rpm rpm -Uvh compat-libstdc++-devel-7.3-2.96.128.i386.rpm rpm -Uvh compat-gcc-7.3-2.96.128.i386.rpm rpm -Uvh compat-gcc-c++-7.3-2.96.128.i386.rpm # confirm the required packages are installed: rpm -qa|grep gccrpm -qa|grep glibcrpm -qa|grep compat-dbrpm -qa|grep compat-gccrpm -qa|grep compat-gcc-c++rpm -qa|grep compat-libstdc++rpm -qa|grep compat-libstdc++-develrpm -qa|grep control-center-2.8.0 rpm -qa|grep openmotif21rpm -qa|grep setarch# SELINUX must be disabled
Page 41
cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Check the hardware requirements # Hardware Requirements # At least 2 GB of physical memory grep MemTotal /proc/meminfo # swap space: twice the amount of physical memory grep SwapTotal /proc/meminfo # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 400 MB disk space in /tmp df -k /tmp # 4 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: df -h /dev/shm/ # to adjust the shared memory file system size: umount tmpfs mount -t tmpfs shmfs -o size=1200m /dev/shm Create the required network configuration: # Network names Resolution # configure /etc/hosts if no domain server is used cat /etc/hosts 127.0.0.1 localhost.localdomain oradb1.mydomain.com oradb1 Create and configure the required OS users and groups # inventory group groupadd -g 501 oinstall groupadd -g 502 dba # oracle software owner user /usr/sbin/useradd -u 200 -g oinstall -G dba oracle passwd oracle # make sure nobody user exists (if not there, create it useradd nobody) id nobody # The oracle User Environment # in /home/oracle/.bash_profile # export DISPLAY if required export ORACLE_BASE=/u01/app/oracle if [ $USER = "oracle" ]; then
Page 42
localhost
if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi export EDITOR=vi export ORACLE_HOME=$ORACLE_BASE/product/10.2.0/db_1 export ORACLE_PATH=$ORACLE_BASE/common/oracle/sql:.:$ORACLE_HOME/rdbms/admin export ORACLE_SID=ora10g export NLS_LANG=AMERICAN_AMERICA.WE8ISO8859P1 export NLS_DATE_FORMAT="mm/dd/yyyy hh24:mi:ss" export PATH=.:${PATH}:$HOME/bin:$ORACLE_HOME/bin:$ORA_CRS_HOME/bin export PATH=${PATH}:/usr/bin:/bin:/usr/bin/X11:/usr/local/bin export PATH=${PATH}:$ORACLE_BASE/common/oracle/bin export ORACLE_TERM=xterm export TNS_ADMIN=$ORACLE_HOME/network/admin export ORA_NLS10=$ORACLE_HOME/nls/data export LD_LIBRARY_PATH=$ORACLE_HOME/lib export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:$ORACLE_HOME/oracm/lib export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/lib:/usr/lib:/usr/local/lib export CLASSPATH=$ORACLE_HOME/JRE export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/jlib export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/rdbms/jlib export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/network/jlib export THREADS_FLAG=native export TEMP=/tmp export TMPDIR=/tmp # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf oracle soft nproc 2047 oracle hard nproc 16384 oracle soft nofile 1024 oracle hard nofile 65536 oracle soft memlock 3145728 oracle hard memlock 3145728 Configure kernel parameters and shell limits
Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Configuring kernel parameters and shell limits # they can be tuned for a production db # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf kernel.sem = 250 32000 100 128 # maximum size may be given to SGA (max 4GB) # kernel.shmmax = 536870912 (512 M) # following is 1 GB kernel.shmmax =1073741824 net.ipv4.ip_local_port_range = 1024 65000 net.core.rmem_default = 4194304 net.core.rmem_max = 4194304
Page 43
net.core.wmem_default = 262144 net.core.wmem_max = 262144 # to take immediate effect /sbin/sysctl -p Configure hangcheck-timer kernel module: # check hangcheck-timer Module Configuration # with this module, if the kernel hangs, the machine will reboot # verify the module is loaded /sbin/lsmod | grep -i hang # if not loaded, load it vi /etc/modprobe.conf options hangcheck-timer hangcheck_tick=30 hangcheck_margin=180 # execute and add in the file vi /etc/rc.local /sbin/modprobe hangcheck-timer Create the required directories for the Oracle database software: # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab # in the example above, /u01 should be owned by the root user # and writable by group oinstall cd / chown -R oracle:oinstall /u01 chmod -R 775 /u01/oracle mkdir -p /u01/app/oracle/product/10.2.0/db_1 mkdir /u01/stage chown -R oracle:oinstall /u01/app/oracle chmod -R 775 /u01/oracle chown oracle:oinstall /u01/stage As oracle, copy software installation into /u01/stage10g/db Disable screensavers on host & guest machines. o In Oracle Linux: Applications-> Preferences-> Screen Saver-> Mode: Disable Screen Saver o Do the same after logging off and logging on again as oracle user. Restart the machine.
2. Oracle Database 10g Software Installation -- start OUI su - oracle cd /u01/stage10g/db ./runInstaller Follow the steps. Install database software only. 3. Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software # extract 10g Release 2 (10.2.0.4) Patch Set 3 for Linux x86 to
Page 44
/u01/stage10g/patch10.2.0.4/Disk1 mkdir /u01/stage10g/patch10.2.0.4 cd /u01/stage10g/patch10.2.0.4/Disk1 ./runInstaller Select path of ORACLE_HOME ->Next 4. Configure Listeners cd /u01/app/oracle/product/10.2.0/db_1/bin ./netca & Add a new listener -- optionally, use net manager to manually register the database: ./netmgr 5. Create Database cd /u01/app/oracle/product/10.2.0/db_1/bin ./dbca & follow the steps to create a customized database of SID "ora10g" and Global nmae "ora10g.oradb1". Do not use specific templates (non-Custom). The DBCA may pop up the following message in the end of db creation: "Error securing Database control, Database Control has been brought up in nonsecure mode. To secure the Database Control execute the following commands: ... " # check the Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep 6. Postinstallation Tasks Verify OEM: # verify that OEM is working http://oradb1:1158/em # restart the dbconsole if required emctl status dbconsole emctl stop dbconsole emctl start dbconsole Verify iSQL*Plus: # in the browser http://oradb1:5560/isqlplus # if not running: isqlplusctl start Consider implementing automatic database startup. See: Automatically Starting Databases Consider implementing automatic EM Database Control startup. See: Implementing EM Database Control Auto Startup Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
For easy Oracle Home access: Oracle DBA Code Examples
Page 45
echo "alias db='cd $ORACLE_HOME'" >> /home/oracle/.bashrc
Page 46
Installing Oracle 11g R2 on Enterprise Linux 5

Note: for any installation, you should check the Release Notes documenation before taking any practical step.
Emulation software: VMWare Workstation 7 for Windows OS: Red Hat Enterprise Linux 5.2 for x86
Required Software
Oracle Database 11g Release 2 for Linux x86 32-bit
Used Hardware
In the VMWare: create one virtual machine with the following specs: o 2.0 GB RAM o One ethernet cards: can be configured as bridged or host-only in VMware. o One local hardisk with 32 GB on SCSI 0:0. It will be used for software installation. o One local hardisk with 20 GB on SCSI 1:0. It will be used for Oracle Database data files. o One local hardisk with 20 GB on SCSI 1:1. It will be used for Oracle Database flash recovery. o CPU Count: 2 (optional)
Installation Plan
1. Preinstallation tasks: o o o 2. 3. 4. 5. 6. Hardware requirements Software requirements Environment configuration
Oracle Database 11g Software Installation Apply Patchset Configure Listeners Create Database Postinstallation tasks
1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Oracle DBA Code Examples
Page 47
Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Give the ethernet card IP 192.168.4.100 and the hostname srv100.mydomain.com. Define a gateway. If it does not exist, make it same as the host IP address. o Insall the following packages: Desktop Environments o GNOME Desktop Environment Applications o Graphical Internet (optional) o Editors (optional) Development o Development Libraries o Development Tools Servers o Do not select anything in this group. Base System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
X Window System
Complete the installation RHEL 5 Bug: After the Installation compelets, RHEL 5.2 and below will hang on booting when it reaches to "starting udev" line. To solve this problem, shutdown the Vmware machine and change the CPU count and Core Count to only one. Implement the changes below, then shutdown the machine, set CPU count back to 2 and startup the machine. put the kernel command line parameters at the end of the "kernel" line: vi /boot/grub/grub.conf add divider=10 clocksource=acpi_pm For example: kernel /vmlinuz-2.6.18 .. clock=acpi_pm divider=10
For Vmware machines, install VMWare tools. Install further packages: # to know distribution and version of Linux (Red Hat Ent. 5.2 used) cat /etc/issue # to know kernel version (and its errata level) (2.6.18-92 or newer) uname -r # to list missed packages: rpm -q --qf '%{NAME}-%{VERSION}-%{RELEASE} (%{ARCH})\n' binutils \ compat-libstdc++-33 \ elfutils-libelf \ elfutils-libelf-devel \ gcc \
Page 48
gcc-c++ \ glibc \ glibc-common \ glibc-devel \ glibc-headers \ ksh \ libaio \ libaio-devel \ libgcc \ libstdc++ \ libstdc++-devel \ make \ sysstat \ unixODBC \ unixODBC-devel # for missed packages, install them: rpm -Uvh libaio-devel-0.3.106-3.2.i386.rpm rpm -Uvh unixODBC*
# SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Check the hardware requirements # Hardware Requirements (in cluster nodes) # At least 1.0 GB of physical memory grep MemTotal /proc/meminfo # swap space: same as the amount of physical memory grep SwapTotal /proc/meminfo # to display swap and memory in one command: free # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 1 GB disk space in /tmp df -h /tmp # 4 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: df -h /dev/shm/
Page 49
# to adjust the shared memory file system size: umount tmpfs mount -t tmpfs shmfs -o size=1200m /dev/shm Create the required network configuration: ping srv100 ping srv100.mydomain.com # Network names Resolution # configure /etc/hosts if no domain server is used cat /etc/hosts 127.0.0.1 srv100.mydomain.com srv100 localhost.localdomain localhost Create and configure the required OS users and groups # all group and user ids on all the nodes must have identical id # Grid Infrastructure (GI) and the Oracle RDBMS home will # be installed using different users: # oracle inventory group /usr/sbin/groupadd -g 501 oinstall /usr/sbin/groupadd -g 502 dba /usr/sbin/useradd -u 502 -g oinstall -G dba oracle # set passwords passwd oracle # make sure nobody user exists (if not there, create it useradd nobody) id nobody
# define the env variables for oracle user vi /home/oracle/.bash_profile # Oracle evn vars export EDITOR=vi export TMP=/tmp export TMPDIR=$TMP export ORACLE_HOSTNAME=srv100.mydomain.com export ORACLE_BASE=/u01/app/oracle export ORACLE_HOME=$ORACLE_BASE/product/11.2.0/db_1 export ORACLE_SID=ora11gr2 export ORACLE_TERM=xterm export PATH=/usr/sbin:$PATH export PATH=$ORACLE_HOME/bin:$PATH export LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib export CLASSPATH=$ORACLE_HOME/JRE:$ORACLE_HOME/jlib:$ORACLE_HOME/rdbms/jlib
# shell startup file vi /etc/profile if [ $USER = "oracle" ] || [ $USER = "grid" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi
Page 50
# for C shell vi /etc/csh.login if ( $USER = "oracle" || $USER = "grid" ) then limit maxproc 16384 limit descriptors 65536 endif Configure kernel parameters and shell limits
Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Configuring kernel parameters and shell limits # they can be tuned for a production db # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf # kernel.shmmax not stated in 11g R2 (max: 4GB) (169706.1) kernel.shmmni = 4096 kernel.sem = 250 32000 100 128 fs.aio-max-nr = 1048576 fs.file-max = 6815744 net.ipv4.ip_local_port_range = 9000 65500 net.core.rmem_default = 262144 net.core.rmem_max = 4194304 net.core.wmem_default = 262144 net.core.wmem_max = 1048576 # to take immediate effect /sbin/sysctl -p # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf oracle soft nproc 2047 oracle hard nproc 16384 oracle soft nofile 1024 oracle hard nofile 65536 vi /etc/pam.d/login session required pam_limits.so Create the required directories for the Oracle database software: # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab # Oracle Inventory Directory # as a root mkdir -p /u01/app/oraInventory chown -R oracle:oinstall /u01/app/oraInventory chmod -R 775 /u01/app/oraInventory # Oracle Base Directory mkdir -p /u01/app/oracle
Page 51
#needed to ensure that dbca is able to run after the rdbms installation mkdir -p /u01/app/oracle/cfgtoollogs chown -R oracle:oinstall /u01/app/oracle chmod -R 775 /u01/app/oracle # Oracle mkdir -p chown -R chmod -R RDBMS Home Directory /u01/app/oracle/product/11.2.0/db_1 oracle:oinstall /u01/app/oracle/product/11.2.0/db_1 775 /u01/app/oracle/product/11.2.0/db_1
# staging area to hold software installation mkdir -p /u01/stage11g/db chown -R oracle:oinstall /u01/stage11g/db chmod -R 775 /u01/stage11g/db Partition and mount the disks: #define the disks to partition fdisk -l | grep '^Disk' # as a root, for the disks /dev/sdb and /dev/sdc #partition the disks: fdisk /dev/sdb # answers: "n", "p", "1", "Return", "Return", "p" and "w" Note: if the following message appears after the "w" command: WARNING: Re-reading the partition table failed with error 16: Device or resource busy, then you can avoid restarting the machine by the following command: partprobe # to make sure partions are created ls -lX /dev/sd* # format the paritions mkfs.ext3 /dev/sdb1 mkfs.ext3 /dev/sdc1 # Mount the new disk mkdir /u02 mount /dev/sdb1 /u02 mkdir /u03 mount /dev/sdc1 /u03 df -H # Update /etc/fstab vi /etc/fstab /dev/sdb1 /dev/sdc1 # create mkdir -p chown -R chmod -R # create mkdir -p chown -R chmod -R
/u02 /u03
ext3 ext3
defaults defaults
1 2 1 2
folder for the db data /u02/oradata/ oracle:oinstall /u02/oradata/ 775 /u02/oradata/ folder for the flash area /u03/oraflash/ oracle:oinstall /u03/oraflash/ 775 /u03/oraflash/
As oracle, copy software installation into /u01/stage11g/db Oracle DBA Code Examples
Page 52
2. Oracle Database 11g Software Installation -- start OUI su - oracle cd /u01/stage11g/db ./runInstaller Follow the steps. Install database software only. 3. Apply Patchset
4. Configure Listeners cd /u01/app/oracle/product/10.2.0/db_1/bin ./netca & Add a new listener Add Naming mtehods: Local Naming, EZConnect 5. Create Database cd /u01/app/oracle/product/10.2.0/db_1/bin ./dbca & follow the steps to create a database with sid: ora11gr2 The DBCA may pop up the following message in the end of db creation: "Error securing Database control, Database Control has been brought up in nonsecure mode. To secure the Database Control execute the following commands: ... " 6. Postinstallation Tasks Verify OEM: # verify that OEM is working https://srv100.mydomain.com:1158/em # restart the dbconsole if required emctl status dbconsole emctl stop dbconsole emctl start dbconsole # check Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep Consider implementing automatic database startup. See: Automatically Starting Databases Consider implementing automatic EM Database Control startup. See: Implementing EM Database Control Auto Startup Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
For easy Oracle Home access: echo "alias db='cd $ORACLE_HOME'" >> /home/oracle/.bashrc
Page 53
Installing Oracle 11g R2 on Enterprise Linux 5.5 with ASM

Note: for any installation, you should check the Release Notes documenation before taking any practical step.
Emulation software: VMWare Workstation 7 for Windows OS: Oracle Enterprise Linux 5.5 for x86 64-bit
Required Software
Oracle Database 11g Release 2 for Linux x86 64-bit Oracle Database 11g Release 2 Grid Infrastructure (11.2.0.1.0) for Linux x86 64-bit
Used Hardware
In the VMWare: create one virtual machine with the following specs: o 2.5 GB RAM o One ethernet card configured as bridged or host-only in VMware o CPU Count: 2 o Disk1: 34 GB on SCSI 0:0 used to install the OS and software o Disk2: of 12 GB. It will be used for +Data. Set it on controller SCSI 1:1 o Disk3: of 6 GB. It will be used for +Flash. Set it on controller SCSI 1:2
Installation Plan
1. Preinstallation tasks 2. 3. 4. 5. 6. 7. 8. 9. Hardware requirements Software requirements Environment configuration
Oracle Grid Infrastructure installation Oracle Grid Infrastructure Patching Oracle Database 11g R2 Software Installation Oracle Database 11g R2 Software Patching Install EM Agent in cluster nodes (if required) ASM Diskgroups Creation Database Creation Complete postinstallation tasks Oracle DBA Code Examples
Page 54
10. Useful postinstallation tasks
1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Configure the swap area in the local hardisk to have 6 GB disk space. o Give the ethernet card IP 192.168.4.105 the hostname srv07 o Insall the following packages: Desktop Environments o GNOME Desktop Environment Applications o Graphical Internet (optional) o Editors (optional) Development o Development Libraries o Development Tools Servers o Do not select anything in this group. Base System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
X Window System
Complete the installation Install further packages: # to know distribution and version of Linux cat /etc/issue # to know kernel version (and its errata level) (2.6.18-194.el5) uname -r # to list missed packages: rpm -q --qf '%{NAME}-%{VERSION}-%{RELEASE} (%{ARCH})\n' binutils \ compat-libstdc++-33 \ elfutils-libelf \ elfutils-libelf-devel \ gcc \ gcc-c++ \ glibc \ glibc-common \ glibc-devel \
Page 55
glibc-headers \ ksh \ libaio \ libaio-devel \ libgcc \ libstdc++ \ libstdc++-devel \ make \ sysstat \ unixODBC \ unixODBC-devel # for missed packages, install them: rpm -Uvh elfutils-libelf-0.137-3.el5.i386.rpm rpm -ivh elfutils-libelf-devel-static-0.137-3.el5.i386.rpm elfutils-libelfdevel-0.137-3.el5.i386.rpm rpm -Uvh kernel-headers-2.6.18-194.el5.x86_64.rpm rpm -Uvh glibc-headers-2.5-49.x86_64.rpm rpm -Uvh glibc-devel-2.5-49.i386.rpm rpm -Uvh glibc-devel-2.5-49.x86_64.rpm rpm -Uvh libgomp-4.4.0-6.el5.x86_64.rpm rpm -Uvh gcc-4.1.2-48.el5.x86_64.rpm rpm -Uvh gcc-c++-4.1.2-48.el5.x86_64.rpm rpm -Uvh libstdc++-devel-4.1.2-48.el5.x86_64.rpm rpm -Uvh gcc-c++-4.1.2-48.el5.x86_64.rpm rpm -Uvh libaio-devel-0.3.106-5.i386.rpm rpm -Uvh libaio-devel-0.3.106-5.x86_64.rpm rpm -Uvh unixODBC-2.2.11-7.1.i386.rpm rpm -Uvh unixODBC-2.2.11-7.1.x86_64.rpm rpm -Uvh unixODBC-devel-2.2.11-7.1.i386.rpm rpm -Uvh unixODBC-devel-2.2.11-7.1.x86_64.rpm
# Oracle ASM Libaray and drivers can be downloaded from here # to know the kernel verion: uname -rm # In this case we need: # library and tools rpm -Uvh oracleasm-support-2.1.3-1.el5.x86_64.rpm rpm -Uvh oracleasm-2.6.18-194.el5-2.0.5-1.el5.x86_64.rpm rpm -Uvh oracleasmlib-2.0.4-1.el5.x86_64.rpm # SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Check the hardware requirements # Hardware Requirements (in cluster nodes) # At least 1.5 GB of physical memory but practically 1.5 is not fine grep MemTotal /proc/meminfo # swap space: same as the amount of physical memory grep SwapTotal /proc/meminfo # to display swap and memory in one command: free Oracle DBA Code Examples
Page 56
# if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 1 GB disk space in /tmp df -h /tmp # 8 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: df -h /dev/shm/ # to adjust the shared memory file system size: umount tmpfs mount -t tmpfs shmfs -o size=1200m /dev/shm Create the required network configuration: ping srv07 # Network names Resolution cat /etc/hosts 127.0.0.1 srv07 localhost.localdomain localhost Create and configure the required OS users and groups Note: When I tried using differenct OS users and groups for Grid Infrastrucutre and ASM, I got error later in DBCA. # Grid Infrastructure (GI) and the Oracle RDBMS home will # be installed using different users: /usr/sbin/groupadd -g 501 oinstall /usr/sbin/groupadd -g 502 dba /usr/sbin/groupadd -g 504 asmadmin /usr/sbin/useradd -u 502 -g oinstall -G dba,asmadmin oracle # set passwords passwd oracle # make sure nobody user exists (if not there, create it useradd nobody) id nobody # define the env variables for oracle user vi /home/oracle/.bash_profile # Oracle Settings export EDITOR=vi TMP=/tmp; export TMP TMPDIR=$TMP; export TMPDIR ORACLE_HOSTNAME=srv07; export ORACLE_HOSTNAME ORACLE_BASE=/u01/app/oracle; export ORACLE_BASE
Page 57
ORACLE_HOME=$ORACLE_BASE/product/11.2.0/db_1; export ORACLE_HOME ORACLE_SID=oradb; export ORACLE_SID ORACLE_TERM=xterm; export ORACLE_TERM PATH=/usr/sbin:$PATH; export PATH PATH=$ORACLE_HOME/bin:$PATH; export PATH LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib; export LD_LIBRARY_PATH CLASSPATH=$ORACLE_HOME/jdk/jre/lib/:$ORACLE_HOME/jlib:$ORACLE_HOME/rdbms/jlib; export CLASSPATH # shell startup file vi /etc/profile if [ $USER = "oracle" ] || [ $USER = "grid" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi # for C shell vi /etc/csh.login if ( $USER = "oracle" || $USER = "grid" ) then limit maxproc 16384 limit descriptors 65536 endif Configure kernel parameters and shell limits Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Kernel Parameters # to tune thme, refer to metalink document 169706.1 # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf kernel.shmmni = 4096 kernel.sem = 250 32000 100 128 fs.aio-max-nr = 1048576 fs.file-max = 6815744 net.ipv4.ip_local_port_range = 9000 65500 net.core.rmem_default = 262144 net.core.rmem_max = 4194304 net.core.wmem_default = 262144 net.core.wmem_max = 1048576 # to take immediate effect /sbin/sysctl -p # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf oracle oracle oracle oracle
Page 58
soft hard soft hard
nproc 2047 nproc 16384 nofile 1024 nofile 65536 Oracle DBA Code Examples
vi /etc/pam.d/login session required pam_limits.so Create partitions in the disks:
Note: On a real life storage, you would create a single whole-disk partition with exactly 1 MB offset on each LUN to be used as ASM Disk. In fdisk: u (to change units from cylinder to sectors), n, p, 1, 2048, w. # as a root, for the disks /dev/sdb .. /dev/sdd # confirm they are seen: ls /dev/sd* #partition the disks: fdisk /dev/sdb # answers: "n", "p", "1", "Return", "Return", "p" and "w" Note: if the following message appears after the "w" command: WARNING: Re-reading the partition table failed with error 16: Device or resource busy, then you can avoid restarting the machine by the following command: partprobe # to make sure partions are created ls -lX /dev/sd* Configure ASM drivers: # as root (to be done in all nodes) oracleasm configure -i Default user to own the driver interface []: oracle Default group to own the driver interface []: asmadmin Start Oracle ASM library driver on boot (y/n) [n]: y Fix permissions of Oracle ASM disks on boot (y/n) [y]: y Writing Oracle ASM library driver configuration: done # Load the kernel module using the following command: /usr/sbin/oracleasm init # If you have any problems, make sure you have the correct # version of the driver (may require Internet connection): /usr/sbin/oracleasm update-driver # mark the shared disks: (one node) /usr/sbin/oracleasm createdisk DISK1 /dev/sdb1 /usr/sbin/oracleasm createdisk DISK2 /dev/sdc1 # check the disks are marked and seen: /usr/sbin/oracleasm listdisks #If you need to unmark a disk that was used in a createdisk command: /usr/sbin/oracleasm deletedisk DISK1 /usr/sbin/oracleasm deletedisk DISK2 Create the required directories for the Oracle software: # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab
Page 59
# Oracle Inventory Directory # as a root mkdir -p /u01/app/oraInventory chown -R oracle:oinstall /u01/app/oraInventory chmod -R 775 /u01/app/oraInventory # Grid Infrastructure Home Directory mkdir -p /u01/11.2.0/grid chown -R oracle:oinstall /u01/11.2.0/grid chmod -R 775 /u01/11.2.0/grid # Oracle mkdir -p chown -R chmod -R Base Directory /u01/app/oracle oracle:oinstall /u01/app/oracle 775 /u01/app/oracle
#needed to ensure that dbca is able to run after the rdbms installation mkdir /u01/app/oracle/cfgtoollogs chown -R oracle:oinstall /u01/app/oracle chmod -R 775 /u01/app/oracle # Oracle mkdir -p chown -R chmod -R RDBMS Home Directory /u01/app/oracle/product/11.2.0/db_1 oracle:oinstall /u01/app/oracle/product/11.2.0/db_1 775 /u01/app/oracle/product/11.2.0/db_1
2. Oracle Grid Infrastructure installation # software staging folder mkdir -p /u01/app/stage/ora11gr2gridinfra chown -R oracle:oinstall /u01/app/stage/ora11gr2gridinfra chmod -R 775 /u01/app/stage/ora11gr2gridinfra mkdir -p /u01/app/stage/ora11gr2db chown -R oracle:oinstall /u01/app/stage/ora11gr2db chmod -R 775 /u01/app/stage/ora11gr2db # as oracle: copy Grid software files to /u01/app/stage/ora11gr2db # as root: chmod -R 775 /u01/app/stage/ora11gr2gridinfra # lunch OUI from the clusterware ( as oracle) cd /u01/app/stage/ora11gr2gridinfra ./runInstaller Installation Option >Select radio button 'Install and Configure Grid Infrastructure for a Standalone Server' >Next Product Language >Accept 'English' as language' >Next Creat ASM Disk Group >Disk Group Name: Data (12GB disk: DISK1) >Redundancy: external >Next
Page 60
NOTE: If you see an empty screen for you candidate disks it is likely that ASMLib has not been properly configured or installed. Try reconfigure them. If you are sure that ASMLib has been properly configured click on 'Change Discovery Path' and provide the correct destination. ASM Password >Specify and conform the password you want to use >Next Privileged OS Groups >Assign the correct OS groups for OS authentication (mostly default is OK) >Next Installation Location >ORACLE_BASE: /u01/app/oracle Software location: /u01/11.2.0/grid >Next Create Inventory >Specify the locations: /u01/app/oraInventory >Next Perform Prerequisite Checks >Check that status of all checks is Succeeded >Next Summary >Finish Execute Configuration Scripts >Run the scripts as instructed in the screen >OK >Next Message: The installation of the Grid Infrastructure was successfull. >Close 3. Oracle Grid Infrastructure Patching Apply patch set, if there is any. 4. Oracle Database 11g R2 Software Installation # as oracle: copy DB software files into /u01/app/stage/ora11gr2db # as root: chmod -R 775 /u01/app/stage/ora11gr2db -- start OUI su - oracle cd /u01/app/stage/ora11gr2db ./runInstaller Follow the steps. Install database software only. When executing root.sh, select "y" for ovewriting questions.
Page 61
5. Oracle Database 11g R2 Software Patching
6. Install EM Agent in cluster nodes (if required)
7. ASM Diskgroups Creation # as grid user: start the ASM Configuration Assistant (ASMCA) su - oracle cd /u01/11.2.0/grid/bin ./asmca >Disk Groups tab >Create button >Disk Group Name: FLASH >Redundancy: External >DISK2 >OK >Exit >Yes 8. Database Creation # as oracle cd /u01/app/oracle/product/11.2.0/db_1/bin ./dbca follow the steps to create a database with sid: oradb The DBCA may pop up the following message in the end of db creation: "Error securing Database control, Database Control has been brought up in nonsecure mode. To secure the Database Control execute the following commands: ... " # check Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep 9. Postinstallation tasks # backup the root.sh script cp /u01/app/oracle/product/11.2.0/db_1/root.sh ~/root.sh.bak cont>> 10. General Useful Postinstallation Tasks in Linux Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
/* Make crs_stat -t more readable */ /* copy the following script into ~/scripts/crstat.sh */ #!/usr/bin/ksh # # Sample 10g CRS resource status query script #
Page 62
# Description: # - Returns formatted version of crs_stat -t, in tabular # format, with the complete rsc names and filtering keywords # - The argument, $RSC_KEY, is optional and if passed to the script, will # limit the output to HA resources whose names match $RSC_KEY. # Requirements: # - $ORA_CRS_HOME should be set in your environment # suggested scrip name: crstat.sh RSC_KEY=$1 QSTAT=-u AWK=/usr/bin/awk
# if not available use /usr/bin/awk
# Table header:echo "" $AWK \ 'BEGIN {printf "%-45s %-10s %-18s\n", "HA Resource", "Target", "State"; printf "%-45s %-10s %-18s\n", "-----------", "------", "-----";}' # Table body: $ORA_CRS_HOME/bin/crs_stat $QSTAT | $AWK \ 'BEGIN { FS="="; state = 0; } $1~/NAME/ && $2~/'$RSC_KEY'/ {appname = $2; state=1}; state == 0 {next;} $1~/TARGET/ && state == 1 {apptarget = $2; state=2;} $1~/STATE/ && state == 2 {appstate = $2; state=3;} state == 3 {printf "%-45s %-10s %-18s\n", appname, apptarget, appstate; state=0;}'
# then add the following in the .bashrc of oracle user # if the file was saved in ~/scripts/crstat.sh alias crstat='~/scripts/crstat.sh' /* Easy Acces to crs and db homes */ # it is common to access bin directories in clusterware and db homes # add the following to .bashrc of oracle user alias db='cd /u01/app/oracle/product/11.2.0/db_1/bin' # add the following to .bashrc of grid user alias crs='cd /u01/app/oracle/crs/bin'
Page 63
Managing Oracle Database Instance

Product Release Number
Oracle Database Release Number Format SELECT * FROM PRODUCT_COMPONENT_VERSION; SELECT * FROM V$VERSION 1.2.3.4.5 1 2 3 4 5 Major database release number Maintenance release number application server release number component specific release number platform specific release number
Obtaining License Information

-- Number of users and CPU/Processors select L.SESSIONS_MAX, L.SESSIONS_WARNING, L.SESSIONS_CURRENT, L.SESSIONS_HIGHWATER, L.USERS_MAX, L.CPU_COUNT_CURRENT, L.CPU_SOCKET_COUNT_CURRENT, L.CPU_COUNT_HIGHWATER, L.CPU_CORE_COUNT_CURRENT, L.CPU_CORE_COUNT_HIGHWATER, L.CPU_SOCKET_COUNT_HIGHWATER from v$license l; -- Database Edition select banner from v$version where BANNER like '%Edition%'; -- Oracle Partitioning installed select decode(count(*), 0, 'No', 'Yes') from dba_part_tables where owner not in ('SYSMAN', 'SH', 'SYS', 'SYSTEM') and rownum = 1; -- Oracle Spatial installed: select decode(count(*), 0, 'No', 'Yes') from all_sdo_geom_metadata where rownum = 1; -- Oracle RAC installed: select decode(count(*), 0, 'No', 'Yes') from v$active_instances where rownum <= 2;
Managing the Instance Architecture

Obtaining Information about the Instance Processes ps ef | grep ora select name, description from V$BGPROCESS order by 1;
Page 64
Obtaining Information about the SGA select value from v$parameter where upper(name)='MEMORY_TARGET'; select value/1024/1024 mb from v$parameter where upper(name) = 'SGA_MAX_SIZE' -- size taken by a memory component select pool, round(sum(BYTES)/1024/1024) MB from V$SGASTAT group by pool select name , value/1024/1024 MB from v$parameter where upper(name) in ( 'DB_CACHE_SIZE','DB_RECYCLE_CACHE_SIZE','DB_KEEP_CACHE_SIZE', ' DB_2k_CACHE_SIZE', 'DB_8k_CACHE_SIZE', 'DB_16k_CACHE_SIZE') -- 10g SELECT COMPONENT , CURRENT_SIZE/1024/1024 MB FROM V$SGA_DYNAMIC_COMPONENTS WHERE CURRENT_SIZE <>0 -- Oracle 11g SELECT COMPONENT, ROUND(CURRENT_SIZE/1024/1024) CURRENT_SIZE , ROUND(MIN_SIZE/1024/1024) MIN, ROUND(MAX_SIZE/1024/1024) MAX FROM V$MEMORY_DYNAMIC_COMPONENTS; -- To know how Oracle has modified the memory area sizes by time select START_TIME, END_TIME, STATUS, COMPONENT, OPER_TYPE, OPER_MODE, PARAMETER, INITIAL_SIZE/1024/1024 INITIAL_SIZE_MB, TARGET_SIZE/1024/1024 TARGET_SIZE_MB, FINAL_SIZE/1024/1024 FINAL_SIZE_MB from V$MEMORY_RESIZE_OPS order by START_TIME, END_TIME Clearing the Buffer Cache ALTER SYSTEM FLUSH SHARED_POOL; ALTER SYSTEM FLUSH BUFFER_CACHE ;
Database Administration Authentication

Using Operating System Authentication Following are the steps to create an Oracle administrator using OS authenticaion. Creating normal user who is capable of using OS authentication is a bit different. See External (OS) Authentication. 1. 2. Create an operating system account for the user. Add the account to the OSDBA or OSOPER operating system defined groups. Unix User Groups: dba, oper Windows: ORA_DBA, ORA_OPER 3. 4. login to OS as the created user In SQL*Plus CONNECT / AS SYSDBA CONNECT / AS SYSOPER
Page 65
Using Password File Authentication 5. 6. 7. 8. 9. Create the password file orapwd FILE=filename PASSWORD=password ENTRIES=max_users alter system set REMOTE_LOGIN_PASSWORDFILE = EXCLUSIVE scope=SPFILE; CONN / AS SYSDBA GRANT SYSDBA TO GRANTEDUSER
10. CONN GRANTEDUSER/USERPASSWORD AS SYSDBA Identifying Users SYSDBA or SYSOPER Users SELECT * FROM V$PWFILE_USERS
Data Dictionary and Dynamic Performance Views

Data Dictionary Creation sql.bsq creates base tables catalog.sql creates data dictionary views catproc.sql runs scripts required for server-side PL/SQL they run as SYSDBA they are located in \rdbms\admin
Startup and Shutdown

Startup Levels STARTUP [FORCE] [RESTRICT] [PFILE=filename] [OPEN [RECOVER] [database] |MOUNT |NOMOUNT] ALTER DATABASE { MOUNT | OPEN } ALTER DATABASE OPEN [READ WRITE| READ ONLY] ALTER SYSTEM ENABLE | DISABLE RESTRICTED SESSION; Shutdown Levels SHUTDOWN [NORMAL | TRANSACTIONAL | IMMEDIATE | ABORT ] Autostart of Database in Windows As a service or name OracleServiceSID. You can manage the service using ORADIM utility. You can control the shutdown type of the database when you shutdown its corresponding service by editing the registry entry ORA_<SID>_SHUTDOWNTYPE in Oracle home entry.
Automatically Starting Databases in Unix For Oracle 10.1, refer to Note 222813.1 /* Example 1 */ # compatible with Note 281912.1 (and Note 760051.1) #(1) login as root #(2) Set it to Y in /etc/oratab mydb:/u01/app/oracle/product/10.2.0/db_1:Y
Page 66
#(3) Create dbora script in the directory /etc/init.d # fix ORA_HOME and ORA_OWNER values as required vi /etc/init.d/dbora #!/bin/bash # # description: Start/Stop the Databases.. # # chkconfig: 2345 99 10 # # processname: oracle # config: /etc/oratab # pidfile: /var/run/oracle.pid # Source function library. . /etc/init.d/functions RETVAL=0 ORA_OWNER="oracle" ORA_HOME="/u01/app/oracle/product/11.2.0/db_1" prog="oracle" start() { echo -n $"Starting $prog: " su - $ORA_OWNER -c "$ORA_HOME/bin/dbstart" su - $ORA_OWNER -c "$ORA_HOME/bin/lsnrctl start" RETVAL=$? echo [ $RETVAL -eq 0 ] && touch /var/lock/subsys/dbora return $RETVAL } stop() { echo -n $"Stopping $prog: " su - $ORA_OWNER -c "$ORA_HOME/bin/dbshut" su - $ORA_OWNER -c "$ORA_HOME/bin/lsnrctl stop" RETVAL=$? echo [ $RETVAL -eq 0 ] && rm -r /var/lock/subsys/dbora return $RETVAL } restart() { stop start }
Page 67
case "$1" in start) start ;; stop) stop ;; restart) restart ;; *) echo $"Usage: $0 {start|stop|restart}" exit 1 esac exit $?
#(4) as root: chgrp dba /etc/init.d/dbora chmod 750 /etc/init.d/dbora cd /sbin chkconfig --add dbora chkconfig --list | grep ora # to test, restart or: /etc/init.d/dbora start
/* Example 2 */ -- have been tested on an Oracle 10g Db on Linux 4 #(1) login as root #(2) Set it to Y in /etc/oratab mydb:/u01/app/oracle/product/10.2.0/db_1:Y #(3) Create dbora script in the directory /etc/init.d # fix ORACLE_HOME value as required #!/bin/sh # # /etc/rc.d/init.d/dbora # Description: Starts and stops the Oracle database and listeners # case "$1" in start) echo -n "Starting Oracle Databases: " echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Starting Oracle Databases as part of system up." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle su - oracle -c dbstart >> /var/log/oracle Oracle DBA Code Examples
Page 68
echo "Done." echo -n "Starting Oracle Listeners: " su - oracle -c "lsnrctl start" >> /var/log/oracle echo "Done." echo "" echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Finished." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle touch /var/lock/subsys/oracle ;; stop) echo -n "Shutting Down Oracle Listeners: " echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Shutting Down Oracle Databases as part of system down." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle su - oracle -c "lsnrctl stop" >> /var/log/oracle echo "Done." rm -f /var/lock/subsys/oracle echo -n "Shutting Down Oracle Databases: " su - oracle -c dbshut >> /var/log/oracle echo "Done." echo "" echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Finished." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle ;; restart) echo -n "Restarting Oracle Databases: " echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Restarting Oracle Databases as part of system up." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle su - oracle -c dbstop >> /var/log/oracle su - oracle -c dbstart >> /var/log/oracle echo "Done." echo -n "Restarting Oracle Listeners: " su - oracle -c "lsnrctl stop" >> /var/log/oracle su - oracle -c "lsnrctl start" >> /var/log/oracle echo "Done." echo "" echo "------------------------------------" >> /var/log/oracle date +"! %T %a %D : Finished." >> /var/log/oracle echo "------------------------------------" >> /var/log/oracle touch /var/lock/subsys/oracle ;; *) echo "Usage: oracle {start|stop|restart}" exit esac # (4) set file group and permissions chgrp dba dbora chmod 750 dbora # (5) ln -s ln -s ln -s ln -s
Page 69
Create the symbolic links: In Linux, /etc/init.d/dbora /etc/rc.d/rc3.d/K01dbora /etc/init.d/dbora /etc/rc.d/rc3.d/S99dbora /etc/init.d/dbora /etc/rc.d/rc5.d/K01dbora /etc/init.d/dbora /etc/rc.d/rc5.d/S99dbora Oracle DBA Code Examples
/* Example 3 */ #(1) login as root #(2) Set it to Y in /etc/oratab mydb:/u01/app/oracle/product/10.2.0/db_1:Y #(3) Create dbora script in the directory /etc/init.d # fix ORACLE_HOME value as required vi /etc/init.d/dbora #!/bin/bash # # oracle Init file for starting and stopping # Oracle Database. Script is valid for 10g and 11g versions. # # chkconfig: 35 80 30 # description: Oracle Database startup script # Source function library. . /etc/rc.d/init.d/functions ORACLE_OWNER="oracle" ORACLE_HOME="/u01/app/oracle/product/10.2.0/db_1" case "$1" in start) echo -n $"Starting Oracle DB:" su - $ORACLE_OWNER -c "$ORACLE_HOME/bin/dbstart $ORACLE_HOME" echo "OK" echo -n "Starting Oracle Listeners:" su - $ORACLE_OWNER -c "$ORACLE_HOME/bin/lsnrctl start" echo "OK" ;; stop) echo -n $"Stopping Oracle DB:" su - $ORACLE_OWNER -c "$ORACLE_HOME/bin/dbshut $ORACLE_HOME" echo "OK" echo -n "Stopping Oracle Listeners:" su - $ORACLE_OWNER -c "$ORACLE_HOME/bin/lsnrctl stop" echo "OK" ;; *) echo $"Usage: $0 {start|stop}" esac # (4) set file group and permissions chgrp dba /etc/init.d/dbora chmod 750 /etc/init.d/dbora # (5) ln -s ln -s ln -s ln -s Create the symbolic links: In Linux, /etc/init.d/dbora /etc/rc.d/rc3.d/K01dbora /etc/init.d/dbora /etc/rc.d/rc3.d/S99dbora /etc/init.d/dbora /etc/rc.d/rc5.d/K01dbora /etc/init.d/dbora /etc/rc.d/rc5.d/S99dbora
Page 70
# (6) configuring execution for specific runlevels chkconfig --add dbora --level 0356
Quiescing a Database
Users will remain logged in, and they can continue to execute their requests that are in progress, while the database is in the quiesced state. The database, however, will block all new transactional requests except those made by the users SYS and SYSTEM (not DBA grantees). ALTER SYSTEM QUIESCE RESTRICTED; ALTER SYSTEM UNQUIESCE; SELECT instance_name, status, shutdown_pending, active_state FROM v$instance;
Suspending a Database
All reads from and writes to the datafiles and control files are suspended. ALTER SYSTEM SUSPEND; ALTER SYSTEM RESUME; SELECT instance_name, status, shutdown_pending, active_state FROM v$instance;
Dropping a Database
Datafiles, redo log files, and control files are removed automatically. Alert log and parameter file are not deleted. CONNECT sys/sys_passwd AS SYSDBA SHUTDOW IMMEDIATE STARTUP RESTRICT MOUNT SELECT name FROM v$database; DROP DATABASE;
Initialization Files
Managing Initialization Files During startup, in $ORACLE_HOME/dbs (for UNIX/Linux) Oracle will look for the correct initialization file to use, in the following order: o o o spfile$ORACLE_SID.ora spfile.ora init$ORACLE_SID.ora
# currently used SPFiles (if null, pfile was used) select * from v$parameter where name = 'spfile' # create pfile CREATE SPFILE='/u01/oracle/dbs/test_spfile.ora' FROM PFILE='/u01/oracle/dbs/test_init.ora' # use pecific spfile during startup STARTUP PFILE = $ORACLE_HOME/dbs/initDBA1.ora
Page 71
# PFILE can indicate to use SPFILE SPFILE = /database/startup/spfileDBA1.ora
In Oracle 11g # write current values of instance parameters CREATE PFILE FROM MEMORY; CREATE SPFILE FROM MEMORY; Managing Parameters in SPFILE -- display current value of a parameter select * from v$parameter where name = 'spfile' -- parameter values set in SPFILE select NAME, ISSPECIFIED from V$SPPARAMETER where name like '%dest%'; -- in SQL Plus show parameter target -- set parameter value alter system set parameter=value scope=SPFILE|MEMORY|BOTH -- delete a parameter from SPFILe ALTER SYSTEM RESET undo_suppress_errors SCOPE=BOTH SID='node01'; -- Only sessions started after the statement is executed are affected -- this option is a must for the parameters whose ISSYS_MODIFIABLE column -- in V$PARAMETER is DEFERRED. You cannot use it, if the column value is FALSE ALTER SYSTEM SET parameter_name DEFERRED; -- undocumented parameters SELECT a.ksppinm parameter, a.ksppdesc description, b.ksppstvl session_value, c.ksppstvl instance_value FROM x$ksppi a, x$ksppcv b, x$ksppsv c WHERE a.indx = b.indx AND a.indx = c.indx AND SUBSTR (a.ksppinm,1,1) = '_' ORDER BY a.ksppinm;
Alert and Trace Files

Monitoring Alert and Trace Files # to obtain the Location of Trace Files column NAME format a35 column VALUE format a35 select name, value from v$parameter
Page 72
where name in ('background_dump_dest','user_dump_dest') # to set maximum size of trace files (excluding the alert file) # in number of block unless you sepcify K or M select * from v$parameter where upper(name )= 'MAX_DUMP_FILE_SIZE' alter session set MAX_DUMP_FILE_SIZE='100M'
Page 73
Managing Oracle Database Physical Structure

Managing Control Files
Obtaining Control File information select value from v$parameter where upper(name)= 'CONTROL_FILES' select * from v$controlfile; select * from v$controlfile_record_section Creating Additional Copies, Renaming, and Relocating Control Files
1. Shut down the database. 2. Copy an existing control file to a new location, using operating system commands. 3. Edit the CONTROL_FILES parameter
CONTROL_FILES="D:\ORACLE\ORADATA\ORA11G2\CONTROL01.CTL","D:\ORACLE\ORADATA\ORA1 1G2\CONTROL02.CTL";
4. Restart the database.
Note: if you are using SPFILE, STARTUP NOMOUNT then use ALTER SYSTEM SET .. SCOPE=SPFILE command. Creating New Control Files
1. Make a list of all datafiles and redo log files of the database. 2. Shut down the database. 3. Back up all datafiles and redo log files of the database. 4. STARTUP NOMOUNT 5. Use the CREATE CONTROLFILE statement:
CREATE CONTROLFILE SET DATABASE prod LOGFILE GROUP 1 ('/u01/oracle/prod/redo01_01.log', '/u01/oracle/prod/redo01_02.log'), GROUP 2 ('/u01/oracle/prod/redo02_01.log', '/u01/oracle/prod/redo02_02.log'), GROUP 3 ('/u01/oracle/prod/redo03_01.log', '/u01/oracle/prod/redo03_02.log') RESETLOGS | NORESETLOGS DATAFILE '/u01/oracle/prod/system01.dbf' SIZE 3M, '/u01/oracle/prod/rbs01.dbs' SIZE 5M, '/u01/oracle/prod/users01.dbs' SIZE 5M, '/u01/oracle/prod/temp01.dbs' SIZE 5M MAXLOGFILES 50 MAXLOGMEMBERS 3 MAXLOGHISTORY 400 MAXDATAFILES 200 MAXINSTANCES 6 ARCHIVELOG; Specify the RESETLOGS clause if you have lost any redo log groups in addition to control files. In this case, you will need to recover from the loss of the redo logs (step 8). You must specify the RESETLOGS clause if you have renamed the database. Otherwise, select the NORESETLOGS clause.
Page 74
Caution: The CREATE CONTROLFILE statement can potentially damage specified datafiles and redo log files. Omitting a filename can cause loss of the data in that file, or loss of access to the entire database. 6. Store a backup of the new control file on an offline storage device. 7. Edit the CONTROL_FILES initialization parameter 8. If you are renaming the database, edit the DB_NAME parameter in your instance parameter file. 9. Recover the database if necessary. If the new control file was created using the NORESETLOGS clause, you can recover the database with complete, closed database recovery. If the new control file was created using the RESETLOGS clause, you must specify USING BACKUP CONTROL FILE in your RECOVER command. 10. If you did not perform recovery, open the database normally. ALTER DATABASE OPEN; If you specified RESETLOGS when creating the control file: ALTER DATABASE OPEN RESETLOGS; Handling Errors During CREATE CONTROLFILE If Oracle Database sends you an error (usually error ORA-01173, ORA-01176, ORA-01177, ORA-01215, or ORA-01216) when you attempt to mount and open the database after creating a new control file, the most likely cause is that you omitted a file from the CREATE CONTROLFILE statement or included one that should not have been listed. Checking for Missing Files after Creating Control Files Check the alert log to see if the database has detected inconsistencies between the data dictionary and the control file. If a datafile exists in the data dictionary but not in the new control file, the database creates a placeholder entry in the control file under the name MISSINGnnnn, where nnnn is the file number in decimal. MISSINGnnnn is flagged in the control file as being offline and requiring media recovery. o If the actual datafile corresponding to MISSINGnnnn is read-only or offline normal, then you can make the datafile accessible by renaming MISSINGnnnn to the name of the actual datafile. o If MISSINGnnnn corresponds to a datafile that was not read-only or offline normal, you must drop the tablespace containing the datafile. Backing Up Control Files -- copy of the control file: ALTER DATABASE BACKUP CONTROLFILE TO '/oracle/backup/control.bkp'; -- commands to re-create the control file: ALTER DATABASE BACKUP CONTROLFILE TO TRACE; Show parameter user_dump_dest
Manage the Size of Control Files It is affected by MAXDATAFILES, MAXLOGFILES, MAXLOGMEMBERS, MAXLOGHISTORY, and MAXINSTANCES parameters in the CREATE DATABASE statement. Also it is affected by CONTROL_FILE_RECORD_KEEP_TIME
Page 75
Multiplexing the Control File 1. Alter the SPFILE: ALTER SYSTEM SET control_files ='$HOME/ORADATA/u01/ctrl01.ctl','$HOME/ORADATA/u02/ctrl02.ctl' SCOPE=SPFILE; 2. Shut down the database 3. Create additional control files: cp $HOME/ORADATA/u01/ctrl01.ctl $HOME/ORADATA/u02/ctrl02.ctl 4. Start the database: startup
Maintaining Online Redo Log Files

Forcing Log Switches and Checkpoints # Forcing a log switch ALTER SYSTEM SWITCH LOGFILE; # forcing checkpoints in seconds FAST_START_MTTR_TARGET = 600 ALTER SYSTEM CHECKPOINT;
Adding Online Redo Log File Groups # GROUP n is optional ALTER DATABASE ADD LOGFILE GROUP 3 ('/ORADATA/u01/log3a.rdo', '/ORADATA/u02/log3b.rdo') SIZE 1M; Adding Online Redo Log File Members ALTER DATABASE ADD LOGFILE MEMBER '/ORADATA/u04/log1c.rdo' TO GROUP 1, '/ORADATA/u04/log2c.rdo' TO GROUP 2, '/ORADATA/u04/log3c.rdo' TO GROUP 3; Dropping Online Redo Log File Groups SELECT GROUP#, ARCHIVED, STATUS FROM V$LOG; ALTER SYSTEM SWITCH LOGFILE; ALTER DATABASE DROP LOGFILE GROUP 3; Dropping Online Redo Log File Members ALTER DATABASE DROP LOGFILE MEMBER '$HOME/ORADATA/u04/log3c.rdo'; Relocating and Renaming Redo Log Members SHUTDOWN Copy the redo log files to the new location. STARTUP MOUNT
Page 76
ALTER DATABASE RENAME FILE '/diska/logs/log1a.rdo', '/diska/logs/log2a.rdo' TO '/diskc/logs/log1c.rdo', '/diskc/logs/log2c.rdo'; ALTER DATABASE OPEN; Verifying Blocks in Redo Log Files # it defaults to TURE alter system set DB_BLOCK_CHECKSUM=true ; Clearing a Redo Log File # if DB stops becuase log file is corrupted ALTER DATABASE CLEAR LOGFILE GROUP 2; # the redo log file will be available even if not archived ALTER DATABASE CLEAR UNARCHIVED LOGFILE GROUP 2; Viewing Redo Log Information SELECT * FROM V$LOG; # STATUS: INVALID (inaccessible), STALE (incomplete), DELETED, Blank (in use) SELECT * FROM V$LOGFILE; SELECT * FROM V$LOG G, V$LOGFILE M where G.GROUP#=M.GROUP# order by M.GROUP#
Managing Archived Redo Logs

Obtaining Information about Archive Log -- in SQL*Plus ARCHIVE LOG LIST -- to know archivelog mode of the database select LOG_MODE from V$DATABASE -- historical archived log information from the control file select RECID, NAME, DEST_ID, THREAD#, SEQUENCE#, RESETLOGS_CHANGE#, RESETLOGS_TIME, RESETLOGS_ID, FIRST_CHANGE#, FIRST_TIME, NEXT_CHANGE#, NEXT_TIME, BLOCKS * BLOCK_SIZE/1024/1024 MB , CREATOR, ARCHIVED, DELETED, STATUS, COMPLETION_TIME,
Page 77
END_OF_REDO, BACKUP_COUNT, ARCHIVAL_THREAD#, IS_RECOVERY_DEST_FILE, COMPRESSED, FAL, BACKED_BY_VSS from V$ARCHIVED_LOG; -- information about archive log destinations and their status select * from V$ARCHIVE_DEST -- information about ARCn processes select * from V$ARCHIVE_PROCESSES -- information about any backup made on archived log files select * from V$BACKUP_REDOLOG -- online groups and which one to be archived select * from V$LOG -- log history information select * from V$LOG_HISTORY Changing the Database Archiving Mode select log_mode from v$database ; CONN / AS SYSDBA SHUTDOWN Back up the database see Specifying Archive Destinations ( next section ) STARTUP MOUNT ALTER DATABASE ARCHIVELOG; -- or NOARCHIVELOG ALTER DATABASE OPEN; SHUTDOWN IMMEDIATE Back up the database Specifying Archive Destinations and their Options -- to local destinations LOG_ARCHIVE_DEST_1 = 'LOCATION=/disk1/archive' LOG_ARCHIVE_DEST_1 = 'LOCATION=USE_DB_RECOVERY_FILE_DEST' -- default is OPTIONAL, REOPEN in seconds (default 300) -- if REOPEN is omitted, ARCn will never open a destination after a failure LOG_ARCHIVE_DEST_2 = 'LOCATION=/disk2/archive MANDATORY REOPEN=600' LOG_ARCHIVE_DEST_3 = 'LOCATION=/disk3/archive OPTIONAL' -- to a standby db LOG_ARCHIVE_DEST_4 = 'SERVICE = standby1' -- control file format
Page 78
LOG_ARCHIVE_FORMAT=t%t_s%s_r%r.arc Specifying the Minimum Number of Successful Destinations alter system set LOG_ARCHIVE_MIN_SUCCEED_DEST=1 Controlling Archiving to a Destination alter system set LOG_ARCHIVE_DEST_STATE_2 = DEFER alter system set LOG_ARCHIVE_DEST_STATE_2 = ENABLE Controlling Trace Output Generated by the Archivelog Process LOG_ARCHIVE_TRACE takes combination of: 0 Disable archivelog tracing. This is the default. 1 Track archival of redo log file. 2 Track archival status for each archivelog destination. 4 Track archival operational phase. 8 Track archivelog destination activity. 16 Track detailed archivelog destination activity. 32 Track archivelog destination parameter modifications. 64 Track ARCn process state activity. 128 Track FAL (fetch archived log) server related activities. 256 Supported in a future release. 512 Tracks asynchronous LGWR activity. 1024 RFS physical client tracking. 2048 ARCn/RFS heartbeat tracking. 4096 Track real-time apply -- LOG_ARCHIVE_TRACE defaults to 0 select value from v$parameter where upper(name)='LOG_ARCHIVE_TRACE'; -- database must be mounted but not open. ALTER SYSTEM SET LOG_ARCHIVE_TRACE=12;
Managing Tablespaces
Obtaining Tablespace Information -- tablespace size usage report (for large db (>100GB), it takes long time) BREAK ON REPORT COMPUTE SUM OF tbsp_size ON REPORT compute SUM OF used ON REPORT compute SUM OF free ON REPORT COL tbspname FORMAT a20 HEADING 'Tablespace Name' COL tbsp_size FORMAT 999,999 HEADING 'Size|(MB)' COL used FORMAT 999,999 HEADING 'Used|(MB)' COL free FORMAT 999,999 HEADING 'Free|(MB)' COL pct_used FORMAT 999 HEADING'% Used' SELECT df.tablespace_name tbspname, sum(df.bytes)/1024/1024 tbsp_size, nvl(sum(e.used_bytes)/1024/1024,0) used, nvl(sum(f.free_bytes)/1024/1024,0) free, nvl((sum(e.used_bytes)*100)/sum(df.bytes),0) pct_used FROM DBA_DATA_FILES df,
Page 79
(SELECT file_id, SUM(nvl(bytes,0)) used_bytes FROM dba_extents GROUP BY file_id) e, (SELECT MAX(bytes) free_bytes, file_id FROM dba_free_space GROUP BY file_id) f WHERE e.file_id(+) = df.file_id AND df.file_id = f.file_id(+) GROUP BY df.tablespace_name ORDER BY 5 DESC -- for a specific tablespace: SELECT df.tablespace_name tbspname, sum(df.bytes)/1024/1024 tbsp_size, nvl(sum(e.used_bytes)/1024/1024,0) used, nvl(sum(f.free_bytes)/1024/1024,0) free, nvl((sum(e.used_bytes)*100)/sum(df.bytes),0) pct_used FROM DBA_DATA_FILES df, (SELECT file_id, SUM(nvl(bytes,0)) used_bytes FROM dba_extents WHERE TABLESPACE_NAME='REC_DATA' GROUP BY file_id) e, (SELECT MAX(bytes) free_bytes, file_id FROM dba_free_space WHERE TABLESPACE_NAME='REC_DATA' GROUP BY file_id) f WHERE e.file_id(+) = df.file_id AND df.file_id = f.file_id(+) AND TABLESPACE_NAME='REC_DATA' GROUP BY df.tablespace_name; -- ************************************************************* -- tablespace sizes (without getting used and free space) (quick response): SELECT DF.TABLESPACE_NAME TBSPNAME, ROUND(SUM(DF.BYTES)/1024/1024/1024,2) GB, COUNT(FILE_NAME) DATAFILES FROM DBA_DATA_FILES DF GROUP BY ROLLUP(DF.TABLESPACE_NAME ) ORDER BY 1
-- free space in temp tablespaces: select sum(free_blocks) from gv$sort_segment where tablespace_name = 'USER_TEMP' ;
-- tablespace info from control file select TS#,NAME,INCLUDED_IN_DATABASE_BACKUP, BIGFILE,FLASHBACK_ON,ENCRYPT_IN_BACKUP from V$TABLESPACE;
-- descriptions of tablespaces select TABLESPACE_NAME,BLOCK_SIZE,INITIAL_EXTENT, NEXT_EXTENT NEXT_EXTENT_SIZE,MIN_EXTENTS,MAX_EXTENTS, Oracle DBA Code Examples
Page 80
PCT_INCREASE,MIN_EXTLEN,STATUS, CONTENTS,LOGGING,FORCE_LOGGING, EXTENT_MANAGEMENT,ALLOCATION_TYPE,PLUGGED_IN, SEGMENT_SPACE_MANAGEMENT,DEF_TAB_COMPRESSION,RETENTION from DBA_TABLESPACES
-- segments contained in tablespaces select OWNER,SEGMENT_NAME,PARTITION_NAME,SEGMENT_TYPE,S.TABLESPACE_NAME, HEADER_FILE,HEADER_BLOCK,S.BYTES/1024/1024 SEGMENT_SIZE_MB, D.BLOCKS SEGMENT_BLOCKS,EXTENTS,S.INITIAL_EXTENT,S.NEXT_EXTENT NEXT_EXTENT_SIZE, S.MIN_EXTENTS,S.MAX_EXTENTS,S.PCT_INCREASE, FREELISTS,FREELIST_GROUPS,D.RELATIVE_FNO, D.FILE_NAME from DBA_SEGMENTS S, DBA_TABLESPACES T, DBA_DATA_FILES D where S.TABLESPACE_NAME = T.TABLESPACE_NAME AND T.TABLESPACE_NAME = D.TABLESPACE_NAME AND S.RELATIVE_FNO = D.RELATIVE_FNO AND T.TABLESPACE_NAME NOT IN ('SYSAUX','SYSTEM') order BY S.TABLESPACE_NAME -- extents contained by tablespaces SELECT EXTENT_ID, E.BLOCK_ID,E.BYTES/1024 EXTENT_SIZE_KB, E.OWNER,E.SEGMENT_NAME,E.PARTITION_NAME, E.SEGMENT_TYPE,D.FILE_NAME, S.TABLESPACE_NAME,HEADER_FILE, HEADER_BLOCK SEG_HEADER_BLOCK,S.BYTES/1024/1024 SEGMENT_SIZE_MB,D.BLOCKS SEGMENT_BLOCKS, EXTENTS SEG_EXTENTS FROM DBA_EXTENTS E, DBA_SEGMENTS S, DBA_DATA_FILES D WHERE E.OWNER=S.OWNER AND E.SEGMENT_NAME = S.SEGMENT_NAME AND NVL(E.PARTITION_NAME,'0') = NVL(S.PARTITION_NAME,'0') AND E.SEGMENT_TYPE = S.SEGMENT_TYPE AND E.FILE_ID = D.FILE_ID AND S.TABLESPACE_NAME NOT IN ('SYSAUX','SYSTEM') ORDER BY E.SEGMENT_NAME, E.OWNER, E.PARTITION_NAME,E.EXTENT_ID
-- free extents within tablespaces SELECT F.TABLESPACE_NAME,F.FILE_ID,F.BLOCK_ID, F.BYTES/1024/1024 FREE_MB, D.FILE_NAME FROM DBA_FREE_SPACE F, DBA_DATA_FILES D WHERE F.FILE_ID = D.FILE_ID UNION SELECT F.TABLESPACE_NAME, TO_NUMBER('') AS FILE_ID, TO_NUMBER('') AS BLOCK_ID, SUM(F.BYTES/1024/1024) FREE_MB, TO_CHAR('') AS FILE_NAME FROM DBA_FREE_SPACE F GROUP BY F.TABLESPACE_NAME, TO_NUMBER('') , TO_NUMBER('') ,TO_CHAR('') ORDER BY TABLESPACE_NAME
-- data files (from control file) SELECT FILE#,T.NAME TABLESPACE_NAME,D.NAME FILENAME, CREATION_CHANGE#,CREATION_TIME, RFILE#,STATUS, ENABLED,CHECKPOINT_CHANGE#,CHECKPOINT_TIME, UNRECOVERABLE_CHANGE#,UNRECOVERABLE_TIME,LAST_CHANGE#, LAST_TIME,OFFLINE_CHANGE#,ONLINE_CHANGE#, ONLINE_TIME,BYTES/1024/1024 FILESIZE_MB,BLOCKS, CREATE_BYTES,BLOCK_SIZE, PLUGGED_IN,BLOCK1_OFFSET,AUX_NAME FROM V$DATAFILE D, V$TABLESPACE T WHERE D.TS# = T.TS# ORDER BY TABLESPACE_NAME, D.RFILE#
-- data files
Page 81
select FILE_NAME,FILE_ID,T.TABLESPACE_NAME, F.STATUS FILE_STATUS, RELATIVE_FNO, AUTOEXTENSIBLE, ROUND(BYTES/1024/1024,2) FILESIZE_MB, ROUND(MAXBYTES/1024/1024,2) MAXSIZE_MB, ROUND((INCREMENT_BY*T.BLOCK_SIZE)/1024/1024,2) AUTOEXTENSION_SIZE_MB, ROUND(USER_BYTES/1024/1024,2) AVAILABLE_FOR_DATA_MB -- remaining size used from DBA_DATA_FILES F, DBA_TABLESPACES T -- for storing metadata where F.TABLESPACE_NAME = T.TABLESPACE_NAME order BY TABLESPACE_NAME, F.RELATIVE_FNO
-- tempfiles included in tablespaces select M.NAME TEMPFILE,FILE#,T.NAME TABLESPACE_NAME, CREATION_TIME,M.TS#,RFILE#, STATUS,ENABLED,BYTES, BLOCKS,CREATE_BYTES,BLOCK_SIZE from V$TEMPFILE M, V$TABLESPACE T where M.TS# = T.TS# order by T.NAME select FILE_NAME,FILE_ID,T.TABLESPACE_NAME, F.STATUS FILE_STATUS, RELATIVE_FNO, AUTOEXTENSIBLE, ROUND(BYTES/1024/1024,2) FILESIZE_MB, ROUND(MAXBYTES/1024/1024,2) MAXSIZE_MB, ROUND((INCREMENT_BY*T.BLOCK_SIZE)/1024/1024,2) AUTOEXTENSION_SIZE_MB from DBA_TEMP_FILES F, DBA_TABLESPACES where F.TABLESPACE_NAME = T.TABLESPACE_NAME order BY TABLESPACE_NAME, F.RELATIVE_FNO
-- extents in all locally managed temporary tablespaces SELECT E.TABLESPACE_NAME,E.FILE_ID, BLOCK_ID BEGIN_BLOCK#, ROUND(E.BYTES/1024,2) EXTENT_SIZE_KB, E.BLOCKS,OWNER FROM V$TEMP_EXTENT_MAP E
-- tablespace groups select GROUP_NAME, TABLESPACE_NAME from DBA_TABLESPACE_GROUPS order BY TABLESPACE_NAME -- user qoutas SELECT USERNAME,TABLESPACE_NAME, BYTES/1024 SIZE_KB, DECODE(MAX_BYTES,-1,-1,MAX_BYTES/1024/1024) MAX_MB FROM DBA_TS_QUOTAS ORDER BY USERNAME
-- diskspace usage by USER select owner, round(sum(bytes)/1024/1024,2) space_in_mb from dba_segments group by owner WHERE OWNER NOT IN ('SYSTEM', 'SYSMAN', 'SYS','WKSYS','WK_TEST','WMSYS','XDB','OUTLN','PERFSTAT','OLAPSYS','ORDSYS','M DSYS','EXFSYS','DMSYS','CTXSYS','REPTEST','SCOTT','RMAN')
Page 82
order by round(sum(bytes)/1024/1024,2) desc;
-- user temporary usage -- SEGTYPE: SORT,HASH,DATA,INDEX,LOB_DATA,LOB_INDEX SELECT USERNAME,SESSION_NUM SESSION_SN, SQLADDR,SQLHASH,SQL_ID, TABLESPACE, SEGTYPE,SEGFILE# INIT_EXTENT_FILE#,SEGBLK# INIT_EXTENT_BLK#,EXTENTS,BLOCKS,SEGRFNO# FROM V$TEMPSEG_USAGE Creating a Locally Managed Tablespace CREATE EXTENT CREATE EXTENT TABLESPACE MANAGEMENT TABLESPACE MANAGEMENT lmtbsb DATAFILE '/u02/oracle/data/lmtbsb01.dbf' SIZE 50M AUTOALLOCATE; lmtbsb DATAFILE '/u02/oracle/data/lmtbsb01.dbf' SIZE 50M LOCAL UNIFORM SIZE 128K; -- default 1MB
-- creating a tablespace in an ASM diskgroup CREATE TABLESPACE sample DATAFILE '+dgroup1'; CREATE TABLESPACE satbs DATAFILE '+DATA' size 50m; Specifying Segment Space Management CREATE TABLESPACE lmtbsb DATAFILE '/u02/oracle/data/lmtbsb01.dbf' SIZE 50M EXTENT MANAGEMENT LOCAL SEGMENT SPACE MANAGEMENT AUTO; -- though it's the default -- the other option is MANUAL Adding Space to Tablespace ALTER TABLESPACE test01 ADD DATAFILE '..' ALTER DATABASE DATAFILE '' RESIZE 500m; SIZE 1000M;
Specifying Nonstandard Block Sizes for Tablespaces -- must be 2KB, 4KB, 8KB, 16KB, or 32KB. -- requirement set DB_CACHE_SIZE, DB_nK_CACHE_SIZE select value/1024 KB from v$parameter where name='db_block_size'; CREATE TABLESPACE lmtbsb DATAFILE '/u02/oracle/data/lmtbsb01.dbf' SIZE 50M EXTENT MANAGEMENT LOCAL UNIFORM SIZE 128K BLOCKSIZE 8K; Using Bigfile Tablespace (BFT) Oracle recommends that you change the extent allocation policy from AUTOALLOCATE, which is the default, to UNIFORM and set a very high extent size. select TABLESPACE_NAME, BIGFILE from DBA_TABLESPACES order by BIGFILE; -- use BigFile tablespaces only with ASM or RAID volume manager. -- it can take up to 4G blocks. with 8K blocks = 32 terabyte datafile CREATE BIGFILE TABLESPACE bigtbs -- the other option SMALLFILE DATAFILE '/u02/oracle/data/bigtbs01.dbf' SIZE 50G -- T is acceptable EXTENT MANAGEMENT LOCAL UNIFORM SIZE 65536K; -- default 1m ALTER TABLESPACE bigtbs RESIZE 80G;
Page 83
ALTER TABLESPACE bigtbs AUTOEXTEND ON NEXT 20G; -- with this sitting, by default all tbs created later will be BFT CREATE DATABASE SET DEFAULT BIGFILE tablespace -- the other option SMALLFILE ALTER TABLESPACE SET DEFAULT BIGFILE TABLESPACE; SELECT property_value FROM database_properties WHERE property_name='DEFAULT_TBS_TYPE';
Using Temporary Tablespace Oracle recommends creating temporary tablespaces with multiples-of-64KB extent sizes. For large data warehousing, make it 1MB. select * from V$TEMPFILE; select * from DBA_TEMP_FILES; CREATE TEMPORARY TABLESPACE lmtemp TEMPFILE '/u02/oracle/data/lmtemp01.dbf' SIZE 20M REUSE EXTENT MANAGEMENT LOCAL UNIFORM SIZE 16M; -- default is 1M (AUTOALLOCATE not allowed) CREATE BIGFILE TABLESPACE bigtbs TEMPFILE '/u02/oracle/data/bigtbs01.dbf' SIZE 50G ALTER TABLESPACE lmtemp ADD TEMPFILE '/u02/oracle/data/lmtemp02.dbf' SIZE 18M REUSE; -- doesn't apply on default temporary tablespace ALTER TABLESPACE lmtemp TEMPFILE OFFLINE; ALTER TABLESPACE lmtemp TEMPFILE ONLINE; ALTER DATABASE TEMPFILE '/u02/oracle/data/lmtemp02.dbf' OFFLINE; ALTER DATABASE TEMPFILE '/u02/oracle/data/lmtemp02.dbf' ONLINE; ALTER DATABASE TEMPFILE '/u02/oracle/data/lmtemp02.dbf' RESIZE 18M; -- the tablespace isn't dropped ALTER DATABASE TEMPFILE '/u02/oracle/data/lmtemp02.dbf' DROP INCLUDING DATAFILES; Renaming a Tempfile ALTER DATABASE TEMPFILE 'C:\ORACLE\ORADATA\ORCL\TEMP02.DBF' OFFLINE; ren C:\ORACLE\ORADATA\ORCL\TEMP02.DBF TEMP03.DBF ALTER DATABASE RENAME FILE 'C:\ORACLE\ORADATA\ORCL\TEMP02.DBF' TO 'C:\ORACLE\ORADATA\ORCL\TEMP03.DBF'; ALTER DATABASE TEMPFILE 'C:\ORACLE\ORADATA\ORCL\TEMP03.DBF' ONLINE; Shrinking Temporary Tablespace ALTER TABLESPACE temp SHRINK SPACE KEEP 1000M; ALTER TABLESPACE temp SHRINK SPACE TEMPFILE tempfile '/u01/app/oracle/oradata/prod1/temp02.dbf' KEEP 100m; SELECT file#, name, bytes/1024/1024 mb FROM v$tempfile;
Page 84
Using Default Temporary Tablespace ALTER DATABASE DEFAULT TEMPORARY TABLESPACE temptbs02; -- can be temp tbs grp SELECT PROPERTY_NAME, PROPERTY_VALUE FROM database_properties WHERE property_name='DEFAULT_TEMP_TABLESPACE'; Using Temporary Tablespace Groups CREATE TEMPORARY TABLESPACE lmtemp2 TEMPFILE '/u02/oracle/data/lmtemp201.dbf' SIZE 50M TABLESPACE GROUP group1; ALTER TABLESPACE lmtemp TABLESPACE GROUP group2; -- remove it from a group ALTER TABLESPACE lmtemp3 TABLESPACE GROUP ''; -- Assigning a Tablespace Group as the Default Temporary Tablespace ALTER DATABASE mydb DEFAULT TEMPORARY TABLESPACE group2; select GROUP_NAME, TABLESPACE_NAME from DBA_TABLESPACE_GROUPS order BY TABLESPACE_NAME Suppressing Redo Generation for a Tablespace CREATE TABLESPACE .. NOLOGGING; Controlling Tablespaces Availability -- NORMAL, TEMPORARY, IMMEDIATE (not possible in NOARCHIVELOG) ALTER TABLESPACE users OFFLINE NORMAL; -- media recovery required if it was offline using TEMPORARY or IMMEDIATE ALTER TABLESPACE users ONLINE; Using Read-Only Tablespaces -- backup the tablespace after making it read only -- it waits for all transactions started before ALTER TABLESPACE flights READ ONLY; -- to list blocking transactions: SELECT SQL_TEXT, SADDR FROM V$SQLAREA,V$SESSION WHERE V$SQLAREA.ADDRESS = V$SESSION.SQL_ADDRESS AND SQL_TEXT LIKE 'alter tablespace%'; -- all transactions on top of SADDR retuned by previous statement -- are blocking transactions SELECT T.SES_ADDR, T.START_SCNB, S.USERNAME, S.MACHINE FROM V$TRANSACTION T, V$SESSION S WHERE T.SES_ADDR = S.SADDR ORDER BY START_SCNB; -- back to read write ALTER TABLESPACE flights READ WRITE; Renaming Tablespaces ALTER TABLESPACE users RENAME TO usersts; Default Permanent Tabelspace SELECT property_value FROM database_properties WHERE
Page 85
property_name='DEFAULT_PERMANENT_TABLESPACE'; ALTER DATABASE DEFAULT TABLESPACE users; Dropping Tablespaces DROP DROP DROP DROP TABLESPACE TABLESPACE TABLESPACE TABLESPACE users INCLUDING CONTENTS; test01 CASCADE CONSTRAINTS; users INCLUDING CONTENTS CASCADE CONSTRAINTS; users INCLUDING CONTENTS AND DATAFILES;
Managing the SYSAUX Tablespace A typical system with an average of 30 concurrent active sessions may require approximately 200 to 300 MB of space for its AWR data. -- to monitor is occupants -- to know which procedure to use to move an occupant SELECT OCCUPANT_NAME , OCCUPANT_DESC , SCHEMA_NAME ,MOVE_PROCEDURE ,MOVE_PROCEDURE_DESC ,SPACE_USAGE_KBYTES FROM V$SYSAUX_OCCUPANTS Diagnosing and Repairing Locally Managed Tablespace Problems Verifying the Integrity of Segments Created in ASSM Tablespaces. Use DBMS_SPACE_ADMIN.ASSM_SEGMENT_VERIFY. If ASSM is disabled, use SEGMENT_VERIFY. After execution, check sid_ora_process_ID.trc in USER_DUMP_DEST. The parameter verify_option takes one of the following constants: o o o SEGMENT_VERIFY_DEEP SEGMENT_VERIFY_BASIC (9) (10) default then the attrib parameter is considered.
SEGMENT_VERIFY_SPECIFIC (11)
attrib takes one of the following constants: o o o o o o o HWM_CHECK (12) checks whether high water mark information is accurate. BMB_CHECK (13) checks whether space bitmap blocks have correct backpointers to the segment header. SEG_DICT_CHECK (14) checks whether dictionary information for segment is accurate. EXTENT_TS_BITMAP_CHECK (15) checks whether extent maps are consistent with file level bitmaps. DB_BACKPOINTER_CHECK (16) checks whether datablocks have correct backpointers to the space metadata blocks. EXTENT_SEGMENT_BITMAP_CHECK (17) checks whether extent map in the segment match with the bitnaps in the segment. BITMAPS_CHECK (18) checks whether space bitmap blocks are accurate. varchar2(100) := 'EMPLOYEES'; varchar2(100) :='HR'; varchar2(100) :='TABLE'; varchar2(100); Oracle DBA Code Examples
declare v_segname v_segowner v_segtype v_tbs
Page 86
begin select into from where tablespace_name v_tbs dba_segments segment_name=v_segname and owner=v_segowner and segment_type=v_segtype;
DBMS_SPACE_ADMIN.ASSM_SEGMENT_VERIFY ( segment_owner =>v_segowner, segment_name =>v_segname, segment_type =>v_segtype, partition_name =>'', verify_option => DBMS_SPACE_ADMIN.SEGMENT_VERIFY_DEEP ); end; / -- Check sid_ora_process_ID.trc in USER_DUMP_DEST select value from v$parameter where upper(name)='USER_DUMP_DEST'
declare v_segname v_segowner v_segtype v_tbs begin select into from where
varchar2(100) := 'EMPLOYEES'; varchar2(100) :='HR'; varchar2(100) :='TABLE'; varchar2(100);
tablespace_name v_tbs dba_segments segment_name=v_segname and owner=v_segowner and segment_type=v_segtype;
DBMS_SPACE_ADMIN.ASSM_SEGMENT_VERIFY ( segment_owner =>v_segowner, segment_name =>v_segname, segment_type =>v_segtype, partition_name =>'', verify_option => DBMS_SPACE_ADMIN.SEGMENT_VERIFY_SPECIFIC, attrib => DBMS_SPACE_ADMIN.BMB_CHECK ); end; / -- Check sid_ora_process_ID.trc in USER_DUMP_DEST select value from v$parameter where upper(name)='USER_DUMP_DEST' Checking Consistency of Segment Extent Map with Tablespace File Bitmaps Use ASSM_SEGMENT_VERIFY to segment residing in a tablespace with automatic segment space management enabled and SEGMENT_VERIFY when it is disabled. After execution, check sid_ora_process_ID.trc in USER_DUMP_DEST. conn sys as sysdba declare v_segname varchar2(100) := 'EMPLOYEES'; v_segowner varchar2(100) :='HR'; Oracle DBA Code Examples
Page 87
v_tbs varchar2(100); v_fno number ; v_rfno number; v_hdr number; begin -- retreive tablespace name, absolute file number select tablespace_name, header_file, header_block into v_tbs, v_fno, v_hdr from dba_segments where segment_name=v_segname and owner=v_segowner; select relative_fno into v_rfno from dba_data_files
where tablespace_name = v_tbs and file_id=v_fno;
DBMS_SPACE_ADMIN.SEGMENT_VERIFY( tablespace_name =>v_tbs, header_relative_file =>v_rfno, header_block =>v_hdr, verify_option =>dbms_space_admin.SEGMENT_VERIFY_EXTENTS_GLOBAL -default SEGMENT_VERIFY_EXTENTS ); end; / -- Check sid_ora_process_ID.trc in USER_DUMP_DEST select value from v$parameter where upper(name)='USER_DUMP_DEST' Verifying the Integrity of ASSM Tablespaces Use DBMS_SPACE_ADMIN.ASSM_TABLESPACE_VERIFY If ASSM is disabled, use TABLESPACE_VERIFY After execution, check sid_ora_process_ID.trc in USER_DUMP_DEST. The parameter ts_option takes one of the following constants: o TS_VERIFY_BITMAPS (19) (Default) The bitmaps are verified against the extents. This will detect bits that are marked used or free wrongly and will also detect multiple allocation of extents. The file metadata will be validated against file$ and control file. TS_VERIFY_DEEP (20) verifies the file bitmaps as well perform checks on all the segments. TS_SEGMENTS (21) This option is used to invoke SEGMENT_VERIFY on all the segments in the tablespace.
o o
segment_option: when the TABLESPACE_VERIFY_SEGMENTS or TABLESPACE_VERIFY_DEEP is selected, the SEGMENT_OPTION can be specified optionally. When TS_VERIFY_SEGMENTS is specified, segment_option can be one of the following: o o SEGMENT_VERIFY_BASIC (9) SEGMENT_VERIFY_DEEP (10)
The value of segment_option is NULL when TS_VERIFY_DEEP or TS_VERIFY_BASIC is specified. After execution, check sid_ora_process_ID.trc in USER_DUMP_DEST conn sys as sysdba select name from v$tablespace order by name; begin
Page 88
DBMS_SPACE_ADMIN.ASSM_TABLESPACE_VERIFY( tablespace_name =>'EXAMPLE', ts_option => DBMS_SPACE_ADMIN.TS_VERIFY_DEEP, segment_option =>NULL); end; / -- Check sid_ora_process_ID.trc in USER_DUMP_DEST select value from v$parameter where upper(name)='USER_DUMP_DEST' Marking the Segment Corrupt or Valid corrupt_option takes one of the following: o o SEGMENT_MARK_CORRUPT (default) SEGMENT_MARK_VALID
conn sys as sysdba declare v_segname varchar2(100) := 'EMPLOYEES'; v_segowner varchar2(100) :='HR'; v_tbs varchar2(100); v_fno number ; v_rfno number; v_hdr number; begin -- retreive tablespace name, absolute file number select tablespace_name, header_file, header_block into v_tbs, v_fno, v_hdr from dba_segments where segment_name=v_segname and owner=v_segowner; select into from where relative_fno v_rfno dba_data_files tablespace_name = v_tbs and file_id=v_fno;
DBMS_SPACE_ADMIN.SEGMENT_CORRUPT ( tablespace_name =>v_tbs, header_relative_file =>v_rfno, header_block =>v_hdr, corrupt_option =>DBMS_SPACE_ADMIN.SEGMENT_MARK_CORRUPT); end; Dropping a Corrupted Segment Use SEGMENT_DROP_CORRUPT to drop a segment currently marked corrupt (without reclaiming space). The space for the segment is not released, and it must be fixed by using the TABLESPACE_FIX_BITMAPS Procedure or the TABLESPACE_REBUILD_BITMAPS Procedure. If the segment state is valid, the procedure returns ORA-03211 error. conn sys as sysdba declare v_segname varchar2(100) := 'EMP'; v_segowner varchar2(100) :='HR'; v_tbs varchar2(100); v_fno number ; Oracle DBA Code Examples
Page 89
v_rfno number; v_hdr number; begin -- retreive tablespace name, absolute file number select tablespace_name, header_file, header_block into v_tbs, v_fno, v_hdr from dba_segments where segment_name=v_segname and owner=v_segowner; select into from where relative_fno v_rfno dba_data_files tablespace_name = v_tbs and file_id=v_fno;
DBMS_SPACE_ADMIN.SEGMENT_DROP_CORRUPT( tablespace_name =>v_tbs, header_relative_file =>v_rfno, header_block => v_hdr); end; Dumping a Segment Header and Bitmap Blocks Use DBMS_SPACE_ADMIN.SEGMENT_DUMP The dump file named sid_ora_process_id.trc is generated in the location specified in the USER_DUMP_DEST conn sys as sysdba declare v_segname varchar2(100) := 'NAMES'; v_segowner varchar2(100) :='HR'; v_tbs varchar2(100); v_fno number; v_rfno number; v_hdr number; begin -- retreive tablespace name, absolute file number select tablespace_name, header_file, header_block into v_tbs, v_fno, v_hdr from dba_segments where segment_name=v_segname and owner=v_segowner; select into from where relative_fno v_rfno dba_data_files tablespace_name = v_tbs and file_id=v_fno;
DBMS_SPACE_ADMIN.SEGMENT_DUMP( tablespace_name =>v_tbs, header_relative_file =>v_rfno, header_block =>v_hdr, dump_option => DBMS_SPACE_ADMIN.SEGMENT_DUMP_EXTENT_MAP); end; Marking a DBA Range in Bitmap as Free or Used The procedure TABLESPACE_FIX_BITMAPS marks the appropriate DBA range (extent) as free or used in bitmap. The BEGIN and END blocks should be in extent boundary and should be extent multiple. fix_option takes one of the following
Page 90
o o
TABLESPACE_EXTENT_MAKE_FREE TABLESPACE_EXTENT_MAKE_USED
conn sys as sysdba DBMS_SPACE_ADMIN.TABLESPACE_FIX_BITMAPS ( tablespace_name =>'EXAMPLE', dbarange_relative_file =>4, dbarange_begin_block =>33, dbarange_end_block =>83, fix_option =>DBMS_SPACE_ADMIN.TABLESPACE_EXTENT_MAKE_FREE); Rebuilding the Appropriate Bitmap conn sys begin DBMS_SPACE_ADMIN.TABLESPACE_REBUILD_BITMAPS ( tablespace_name =>'EXAMPLE', bitmap_relative_file =>NULL, -- all files bitmap_block =>NULL); -- Block number of bitmap block to rebuild -- NULL = all blocks end; Rebuilding Quotas for Given Tablespace conn sys exec DBMS_SPACE_ADMIN.TABLESPACE_REBUILD_QUOTAS('USERS'); Migrating from a Dictionary-Managed to a Locally Managed Tablespace -- This operation is done online, but space management operations are blocked -- ASSM won't be active on migrated objects conn sys EXEC DBMS_SPACE_ADMIN.TABLESPACE_MIGRATE_TO_LOCAL ('USERS'); -- another way requiring table lock (better) ALTER TABLE emp MOVE TABLESPACE tbsp_new; ALTER INDEX emp_pk_idx REBUILD TABLESPACE tbsp_idx_new;
Fixing the State of the Segments in A Tablespace Use TABLESPACE_FIX_SEGMENT_STATES to fix the state of the segments in a tablespace in which migration was aborted. conn sys EXECUTE DBMS_SPACE_ADMIN.TABLESPACE_FIX_SEGMENT_STATES('TS1') Scenario 1: Fixing Bitmap When Allocated Blocks are Marked Free (No Overlap) The TABLESPACE_VERIFY procedure discovers that a segment has allocated blocks that are marked free in the bitmap, but no overlap between segments is reported. In this scenario, perform the following tasks: 1. Call the SEGMENT_DUMP procedure to dump the ranges that the administrator allocated to the segment. 2. For each range, call the TABLESPACE_FIX_BITMAPS procedure with the TABLESPACE_EXTENT_MAKE_USED option to mark the space as used.
Page 91
3. Call TABLESPACE_REBUILD_QUOTAS to fix up quotas. Scenario 2: Dropping a Corrupted Segment You cannot drop a segment because the bitmap has segment blocks marked "free". The system has automatically marked the segment corrupted. In this scenario, perform the following tasks: 1. Call the SEGMENT_VERIFY procedure with the SEGMENT_VERIFY_EXTENTS_GLOBAL option. If no overlaps are reported, then proceed with steps 2 through 5. 2. Call the SEGMENT_DUMP procedure to dump the DBA ranges allocated to the segment. 3. For each range, call TABLESPACE_FIX_BITMAPS with the TABLESPACE_EXTENT_MAKE_FREE option to mark the space as free. 4. Call SEGMENT_DROP_CORRUPT to drop the SEG$ entry. 5. Call TABLESPACE_REBUILD_QUOTAS to fix up quotas. Scenario 3: Fixing Bitmap Where Overlap is Reported The TABLESPACE_VERIFY procedure reports some overlapping. Some of the real data must be sacrificed based on previous internal errors. After choosing the object to be sacrificed, in this case say, table t1, perform the following tasks: 1. Make a list of all objects that t1 overlaps. 2. Drop table t1. If necessary, follow up by calling the SEGMENT_DROP_CORRUPT procedure. 3. Call the SEGMENT_VERIFY procedure on all objects that t1 overlapped. If necessary, call the TABLESPACE_FIX_BITMAPS procedure to mark appropriate bitmap blocks as used. 4. Rerun the TABLESPACE_VERIFY procedure to verify the problem is resolved. Scenario 4: Correcting Media Corruption of Bitmap Blocks A set of bitmap blocks has media corruption. In this scenario, perform the following tasks: 1. Call the TABLESPACE_REBUILD_BITMAPS procedure, either on all bitmap blocks, or on a single block if only one is corrupt. 2. Call the TABLESPACE_REBUILD_QUOTAS procedure to rebuild quotas. 3. Call the TABLESPACE_VERIFY procedure to verify that the bitmaps are consistent.
Transporting Tablespaces Between Databases Limitations on Transportable Tablespace Use The source and target database must use the same character set and national character set. Objects with underlying objects (such as materialized views) or contained objects (such as partitioned tables) are not transportable unless all of the underlying or contained objects are in the tablespace set. You cannot transport the SYSTEM tablespace or objects owned by the user SYS. This means that you cannot use TTS for PL/SQL, triggers, or views. These would have to be moved with export. You cannot transport a table with a materialized view unless the mview is in the transport set you create. You cannot transport a partition of a table without transporting the entire table.
Page 92
1. Check endian format of both platforms. For cross-platform transport, check the endian format of both platforms by querying the V$TRANSPORTABLE_PLATFORM view. You can find out your own platform name: select platform_name from v$database 2. Pick a self-contained set of tablespaces. The following statement can be used to determine whether tablespaces sales_1 and sales_2 are self-contained, with referential integrity constraints taken into consideration: DBMS_TTS.TRANSPORT_SET_CHECK( TS_LIST =>'sales_1,sales_2', INCL_CONSTRAINTS =>TRUE, FULL_CHECK =>TRUE) Note: You must have been granted the EXECUTE_CATALOG_ROLE role (initially signed to SYS) to execute this procedure. You can see all violations by selecting from the TRANSPORT_SET_VIOLATIONS view. If the set of tablespaces is self-contained, this view is empty. 3. Generate a transportable tablespace set. 3.1.Make all tablespaces in the set you are copying read-only. 3.2.Export the metadata describing the objects in the tablespace(s) EXPDP system/password DUMPFILE=expdat.dmp DIRECTORY=dpump_dir TRANSPORT_TABLESPACES = sales_1,sales_2 TRANSPORT_FULL_CHECK=Y 3.3.If you want to convert the tablespaces in the source database, use the RMAN RMAN TARGET / CONVERT TABLESPACE sales_1,sales_2 TO PLATFORM 'Microsoft Windows NT' FORMAT '/temp/%U' 4. Transport the tablespace set. Transport both the datafiles and the export file of the tablespaces to a place accessible to the target database. 5. Convert tablespace set, if required, in the destination database. Use RMAN as follows: RMAN> CONVERT DATAFILE '/hq/finance/work/tru/tbs_31.f', '/hq/finance/work/tru/tbs_32.f', '/hq/finance/work/tru/tbs_41.f' TO PLATFORM="Solaris[tm] OE (32-bit)" FROM PLATFORM="HP TRu64 UNIX" DBFILE_NAME_CONVERT= "/hq/finance/work/tru/", "/hq/finance/dbs/tru" PARALLELISM=5 Note: The source and destination platforms are optional. Note: By default, Oracle places the converted files in the Flash Recovery Area, without changing the datafile names. Note: If you have CLOB data on a small-endian system in an Oracle database version before 10g and with a varying-width character set and you are transporting to a database in a bigendian system, the CLOB data must be converted in the destination database. RMAN does not handle the conversion during the CONVERT phase. However, Oracle database automatically handles the conversion while accessing the CLOB data. If you want to eliminate this run-time conversion cost from this automatic conversion, you can issue the CREATE TABLE AS SELECT command before accessing the data. 6. Plug in the tablespace.
Page 93
IMPDP system/password DUMPFILE=expdat.dmp DIRECTORY=dpump_dir TRANSPORT_DATAFILES= /salesdb/sales_101.dbf, /salesdb/sales_201.dbf REMAP_SCHEMA=(dcranney:smith) REMAP_SCHEMA=(jfee:williams) If required, put the tablespace into READ WRITE mode. Using Transportable Tablespaces: Scenarios Transporting and Attaching Partitions for Data Warehousing 1. In a staging database, you create a new tablespace and make it contain the table you want to transport. It should have the same columns as the destination partitioned table. 2. Create an index on the same columns as the local index in the partitioned table. 3. Transport the tablespace to the data warehouse. 4. In the data warehouse, add a partition to the table. ALTER TABLE sales ADD PARTITION jul98 VALUES LESS THAN (1998, 8, 1) 5. Attach the transported table to the partitioned table by exchanging it with the new partition: ALTER TABLE sales EXCHANGE PARTITION jul98 WITH TABLE jul_sales INCLUDING INDEXES WITHOUT VALIDATION Publishing Structured Data on CDs A data provider can load a tablespace with data to be published, generate the transportable set, and copy the transportable set to a CD. When customers receive this CD, they can plug it into an existing database without having to copy the datafiles from the CD to disk storage. Note: In this case, it is highly recommended to set the READ_ONLY_OPEN_DELAYED initialization parameter to TRUE. Moving Databases Across Platforms Using Transportable Tablespaces You can use the transportable tablespace feature to migrate a database to a different platform. However, you cannot transport the SYSTEM tablespace. Therefore, objects such as sequences, PL/SQL packages, and other objects that depend on the SYSTEM tablespace are not transported. You must either create these objects manually on the destination database, or use Data Pump to transport the objects that are not moved by transportable tablespace.
Managing Alert Thresholds

Getting the Current Threshold Setting List of supported metrics can be found in the documentation "PL/SQL Packages and Types Reference": 10g, 11g or link or from the query below: SELECT METRIC_ID, METRIC_NAME, METRIC_UNIT, GROUP_ID, GROUP_NAME FROM V$METRICNAME ORDER BY METRIC_NAME -- current threshold settings select * from DBA_THRESHOLDS; set serveroutput on DECLARE V_WOPERATOR BINARY_INTEGER; V_WVALUE VARCHAR2(50);
Page 94
V_COPERATOR BINARY_INTEGER; V_CVALUE VARCHAR2(50); V_OBS_PERIOD BINARY_INTEGER; V_CON_PERIOD BINARY_INTEGER; FUNCTION GET_OPERATOR_NAME( P_OPER IN BINARY_INTEGER) RETURN VARCHAR2 IS BEGIN IF P_OPER =0 THEN RETURN 'GT'; ELSIF P_OPER =1 THEN RETURN 'EQ'; ELSIF P_OPER =2 THEN RETURN 'LT'; ELSIF P_OPER =3 THEN RETURN 'LE'; ELSIF P_OPER =4 THEN RETURN 'GE'; ELSIF P_OPER =5 THEN RETURN 'OPERATOR_CONTAINS'; ELSIF P_OPER =6 THEN RETURN 'NE'; ELSIF P_OPER =7 THEN RETURN 'OPERATOR_DO_NOT_CHECK'; END IF; END GET_OPERATOR_NAME; BEGIN DBMS_SERVER_ALERT.GET_THRESHOLD( metrics_id =>DBMS_SERVER_ALERT.TABLESPACE_PCT_FULL, warning_operator =>V_WOPERATOR , warning_value =>V_WVALUE, critical_operator =>V_COPERATOR, critical_value =>V_CVALUE, observation_period =>V_OBS_PERIOD, consecutive_occurrences =>V_CON_PERIOD, instance_name =>NULL, object_type => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE, object_name =>NULL); DBMS_OUTPUT.PUT_LINE('Warning Op.: '|| GET_OPERATOR_NAME(V_WOPERATOR)); DBMS_OUTPUT.PUT_LINE('Warning Val: '||V_WVALUE); DBMS_OUTPUT.PUT_LINE('Critical Op.: '|| GET_OPERATOR_NAME(V_COPERATOR)); DBMS_OUTPUT.PUT_LINE('Critical Val: '||V_CVALUE); EXCEPTION WHEN OTHERS THEN IF SQLCODE='-13799' THEN DBMS_OUTPUT.PUT_LINE('No threshold was found with the specified threshold key.(ORA-13799)'); ELSE RAISE; END IF; END; / Setting Tablespace Alert Thresholds warning_operator takes one of the following (not all applicable for all metrics): o o OPERATOR_EQ GE GT LE LT NE OPERATOR_CONTAINS Oracle DBA Code Examples
Page 95
OPERATOR_DO_NOT_CHECK (disables the alert for the specified metric)
-- set the free-space-remaining thresholds in the USERS tablespace to 10 MB (warning) -- and 2 MB (critical), and disable the percent-full thresholds. BEGIN DBMS_SERVER_ALERT.SET_THRESHOLD( metrics_id => DBMS_SERVER_ALERT.TABLESPACE_BYT_FREE, -- Tablespace FREE space in KB warning_operator => DBMS_SERVER_ALERT.OPERATOR_LT, -- GT is not applicable here warning_value => '10240', critical_operator => DBMS_SERVER_ALERT.OPERATOR_LT, critical_value => '2048', observation_period => 1, -- computation period (1-60 min) consecutive_occurrences => 1, -- violation times before alert instance_name => NULL, -- NULL= 'database_wide'. Passed value is not checked object_type => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE, object_name => 'USERS'); -- if NULL -> All Tablespaces DBMS_SERVER_ALERT.SET_THRESHOLD( metrics_id => DBMS_SERVER_ALERT.TABLESPACE_PCT_FULL, -- tablespace USAGE by % warning_operator => DBMS_SERVER_ALERT.OPERATOR_DO_NOT_CHECK, warning_value => '0', critical_operator => DBMS_SERVER_ALERT.OPERATOR_DO_NOT_CHECK, critical_value => '0', observation_period => 1, consecutive_occurrences => 1, instance_name => NULL, object_type => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE, object_name => 'USERS'); END; / SELECT * FROM dba_thresholds where object_name = 'USERS' and object_type='TABLESPACE'; Restoring a Tablespace to Database Default Thresholds You can restore the metric threshold values to revert to the database defaults by setting them to NULL in the DBMS_SERVER_ALERT.SET_THRESHOLD. BEGIN DBMS_SERVER_ALERT.SET_THRESHOLD( metrics_id => DBMS_SERVER_ALERT.TABLESPACE_PCT_FULL, -- tablespace USAGE by % warning_operator => NULL, -- do not use '' warning_value => NULL, critical_operator => NULL, critical_value => NULL, observation_period => 1, consecutive_occurrences => 1, instance_name => NULL, object_type => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE, object_name => 'USERS'); END; / Oracle DBA Code Examples
Page 96
SELECT * FROM dba_thresholds where object_name = 'USERS' and object_type='TABLESPACE'; Modifying Database Default Thresholds set the object_name to NULL BEGIN DBMS_SERVER_ALERT.SET_THRESHOLD( metrics_id => DBMS_SERVER_ALERT.TABLESPACE_PCT_FULL, -- redo with TABLESPACE_BYT_FREE warning_operator => DBMS_SERVER_ALERT.OPERATOR_GT, warning_value => '80', critical_operator => DBMS_SERVER_ALERT.OPERATOR_GT, critical_value => '92', observation_period => 1, consecutive_occurrences => 1, instance_name => NULL, object_type => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE, object_name => NULL); END; / SELECT * FROM dba_thresholds where object_name = 'USERS' and object_type='TABLESPACE'; Viewing Alerts -- outstanding alerts (to be cleared) SELECT SEQUENCE_ID,OWNER, OBJECT_NAME,OBJECT_TYPE,SUBOBJECT_NAME, REASON_ID, REASON,TIME_SUGGESTED, SUGGESTED_ACTION,ADVISOR_NAME,METRIC_VALUE, MESSAGE_TYPE,MESSAGE_GROUP,MESSAGE_LEVEL, HOSTING_CLIENT_ID,MODULE_ID,PROCESS_ID, HOST_ID,HOST_NW_ADDR,INSTANCE_NAME, INSTANCE_NUMBER,USER_ID,EXECUTION_CONTEXT_ID,CREATION_TIME FROM DBA_OUTSTANDING_ALERTS -- history of alerts that have been cleared SELECT SEQUENCE_ID,OWNER, OBJECT_NAME,SUBOBJECT_NAME,OBJECT_TYPE, REASON_ID, REASON,TIME_SUGGESTED, SUGGESTED_ACTION,ADVISOR_NAME,METRIC_VALUE, MESSAGE_TYPE,MESSAGE_GROUP,MESSAGE_LEVEL, HOSTING_CLIENT_ID,MODULE_ID,PROCESS_ID, HOST_ID,HOST_NW_ADDR,INSTANCE_NAME,INSTANCE_NUMBER, USER_ID,EXECUTION_CONTEXT_ID,CREATION_TIME FROM DBA_ALERT_HISTORY ORDER BY SEQUENCE_ID -- list of all metrics SELECT METRIC_ID, METRIC_NAME, METRIC_UNIT,GROUP_ID, GROUP_NAME FROM V$METRICNAME
Page 97
ORDER BY METRIC_NAME -- system-level metric values in memory SELECT BEGIN_TIME,END_TIME,INTSIZE_CSEC, GROUP_ID,ENTITY_ID,ENTITY_SEQUENCE, METRIC_ID,METRIC_NAME,VALUE,METRIC_UNIT FROM V$METRIC - also V$METRIC_HISTORY ORDER BY BEGIN_TIME, VALUE DESC -- alert types select INST_ID,REASON_ID,OBJECT_TYPE,TYPE,GROUP_NAME,SCOPE,INTERNAL_METRIC_CATEGORY, INTERNAL_METRIC_NAME from GV$ALERT_TYPES order by OBJECT_TYPE,TYPE
Managing Datafiles and Tempfiles

Creating Datafiles CREATE TABLESPACE CREATE TEMPORARY TABLESPACE ge 8-9 ALTER TABLESPACE ... ADD DATAFILE ALTER TABLESPACE ... ADD TEMPFILE ge 8-9 CREATE DATABASE ALTER DATABASE ... CREATE DATAFILE Enabling and Disabling Automatic Extension for a Datafile ALTER TABLESPACE users ADD DATAFILE '/u02/oracle/rbdb1/users03.dbf' SIZE 10M AUTOEXTEND ON NEXT 512K MAXSIZE 250M; Manually Resizing a Datafile -- if there is space in the datafile ALTER DATABASE DATAFILE '/u02/oracle/rbdb1/stuff01.dbf' RESIZE 100M; Bringing Datafiles Online or Taking Offline in ARCHIVELOG Mode ALTER DATABASE DATAFILE '/u02/oracle/rbdb1/stuff01.dbf' ONLINE; ALTER DATABASE DATAFILE '/u02/oracle/rbdb1/stuff01.dbf' OFFLINE; Bringing Datafiles Online or Taking Offline in ARCHIVELOG Mode ALTER DATABASE DATAFILE ... {ONLINE|OFFLINE} -- all datafiles will be affected in the following code -- this is different from ALTER TABLESPACE...ONLINE|OFFLINE which controls tablespace availability ALTER TABLESPACE ... DATAFILE {ONLINE|OFFLINE} ALTER TABLESPACE ... TEMPFILE {ONLINE|OFFLINE} Taking Datafiles Offline in NOARCHIVELOG Mode Use it when you want to drop the datafile.
Page 98
-- datafile cannot be brought ONLINE again ALTER DATABASE DATAFILE ... OFFLINE FOR DROP; -- the datafile MUST then be dropped -- for dictionary managed tablespace ALTER TABLESPACE ... DROP DATAFILE DROP TABLESPACE ... INCLUDING CONTENTS AND DATAFILES Renaming and Relocating Datafiles in a Single Tablespace ALTER TABLESPACE users OFFLINE NORMAL; Copy the datafiles to their new locations and rename them using the operating system. ALTER TABLESPACE users RENAME DATAFILE '/u02/oracle/rbdb1/user1.dbf', '/u02/oracle/rbdb1/user2.dbf' TO '/u02/oracle/rbdb1/users01.dbf', '/u02/oracle/rbdb1/users02.dbf'; -- for system, default temporary, or undo -- ALTER TABLESPACE cannot be used because you cannot take them OFFLINE mount the database ALTER DATABASE RENAME FILE '/u02/oracle/rbdb1/sort01.dbf', '/u02/oracle/rbdb1/user3.dbf' TO '/u02/oracle/rbdb1/temp01.dbf', '/u02/oracle/rbdb1/users03.dbf; Back up the database. Dropping Datafiles The following restrictions apply: o o o o o The database must be open. The datafile must be empty, otherwise use drop the tablespace. You cannot drop datafiles in a read-only tablespace. You cannot drop datafiles in the SYSTEM tablespace. If a datafile in a locally managed tablespace is offline, it cannot be dropped.
ALTER TABLESPACE example DROP DATAFILE ... ALTER TABLESPACE lmtemp DROP TEMPFILE ALTER DATABASE TEMPFILE .. DROP INCLUDING DATAFILES Copying a File on a Local File System The copied file must meet the following requirements: o o The size must be a multiple of 512 bytes. The size must be less than or equal to two terabytes.
Be aware not to coy a file that is being used by a process. If you are copying a database datafile, make it READ ONLY before you start to copy. CREATE DIRECTORY SOURCE_DIR AS '/usr/admin/source'; CREATE DIRECTORY DEST_DIR AS '/usr/admin/destination'; GRANT READ ON DIRECTORY source_dir TO strmadmin; GRANT WRITE ON DIRECTORY dest_dir TO strmadmin; CONNECT strmadmin/strmadminpw BEGIN DBMS_FILE_TRANSFER.COPY_FILE( source_directory_object => 'SOURCE_DIR', source_file_name => 'db1.dat',
Page 99
destination_directory_object => 'DEST_DIR', destination_file_name => 'db1_copy.dat'); END; Transferring a File to a Different Database In order to transfer a file the other way around, you must replace the PUT_FILE procedure with the GET_FILE procedure. If you are copying a database datafile, make it READ ONLY before you start to copy. You can monitor copying progress using V$SESSION_LONGOPS view.
CREATE DATABASE LINK ODB CONNECT TO system IDENTIFIED BY system_passwd USING 'prod1'; BEGIN DBMS_FILE_TRANSFER.PUT_FILE( SOURCE_DIRECTORY_OBJECT => 'SOURCE_DIR', SOURCE_FILE_NAME => 'mydata1.dbf', DESTINATION_DIRECTORY_OBJECT => 'DEST_DIR', DESTINATION_FILE_NAME => 'mydata2.dbf' DESTINATION_DATABASE => 'ODB.ACME.COM');
END;
/ BEGIN DBMS_FILE_TRANSFER.GET_FILE( SOURCE_DIRECTORY_OBJECT => 'SOURCE_DIR', SOURCE_FILE_NAME => 'TEST01.DBF', SOURCE_DATABASE => 'ODB.ACME.COM', DESTINATION_DIRECTORY_OBJECT => 'DEST_DIR', DESTINATION_FILE_NAME => 'TEST01.DBF'); END; /
Dumping a Data Block

ALTER SYSTEM DUMP DATAFILE 3 BLOCK 3281; -- To dump a number of consecutive blocks:: ALTER SYSTEM DUMP DATAFILE 5 BLOCK MIN 42 BLOCK MAX 50; select value from v$parameter where name='user_dump_dest'; /u01/app/oracle/admin/pasu/udump/pasu_ora_29673.trc
/* to dump index blocks */ -- get object id of the index: SELECT object_id FROM dba_objects WHERE object_name = 'MYINDEX'; -- do a treedump of the index: ALTER SESSION SET EVENTS 'immediate trace name treedump level 106315'; index height distinct index blocks in the lower level | RBA block at position zero | | | | | | branch: 01c3588a 29579402 (0: nrow: 222, level: 1)
number of entries lead block number (starts from -1) | non-deleted entries | | |
Page 100
leaf: leaf: leaf: leaf:
01c3588b 01c3588c 01c3588d 01c3588e
29579403 29579404 29579405 29579406
(-1: nrow: 485 rrow: 485) (0: nrow: 479 rrow: 479) (1: nrow: 479 rrow: 479) (2: nrow: 479 rrow: 479)
-- define the RBA of the block to dump then -- get its file# and block#: SELECT DBMS_UTILITY.DATA_BLOCK_ADDRESS_FILE(223456765), DBMS_UTILITY.DATA_BLOCK_ADDRESS_BLOCK(223456765) FROM dual;
-- dump the header block ALTER SYSTEM DUMP DATAFILE 7 BLOCK 328745;
-- also, you can find the root block ( it is the block following the header bloc): SELECT header_file, header_block+1 FROM dba_segments WHERE segment_name='MYINDEX';
Managing Undo Tablespaces

Obtaining Information on Undo
SHOW PARAMETER UNDO NAME TYPE VALUE ----------------- ------- ------------undo_management string AUTO undo_retention integer 900 undo_tablespace string UNDOTBS01 -- undo usage in 10-min-intervals for last 7 days select TO_CHAR(BEGIN_TIME, 'MM/DD/YYYY HH24:MI:SS') BEGIN_TIME, TO_CHAR(END_TIME, 'MM/DD/YYYY HH24:MI:SS') END_TIME, UNDOBLKS UNDO_BLOCKS, MAXQUERYLEN MAX_QUERY_LENGTH_INSEC, MAXQUERYID LONGEST_QUERY_ID, TXNCOUNT TOTAL_TRANSACTIONS, MAXCONCURRENCY from V$UNDOSTAT order by BEGIN_TIME DESC -- stats history of V$UNDOSTAT select TO_CHAR(BEGIN_TIME, 'MM/DD/YYYY HH24:MI:SS') BEGIN_TIME, TO_CHAR(END_TIME, 'MM/DD/YYYY HH24:MI:SS') END_TIME, INSTANCE_NUMBER,SNAP_ID, UNDOBLKS, TXNCOUNT TOTAL_TRANSACTIONS, MAXQUERYLEN MAX_QUERY_LENGTH_INSEC, MAXQUERYSQLID, MAXCONCURRENCY, SSOLDERRCNT ORA01555_CNT, NOSPACEERRCNT NOSPACE_CNT, ACTIVEBLKS,UNEXPIREDBLKS from DBA_HIST_UNDOSTAT
-- undo segments select SEGMENT_NAME,OWNER,STATUS from DBA_ROLLBACK_SEGS;
Page 101
-- undo extents select e.SEGMENT_NAME, e.TABLESPACE_NAME, e.EXTENT_ID, e.FILE_ID, e.BLOCK_ID, e.BYTES/1024 SIZE_KB, e.RELATIVE_FNO, e.STATUS -- ACTIVE EXPIRED UNEXPIRED from DBA_UNDO_EXTENTS e order by e.STATUS ASC -- undo sizes by STATUS select e.TABLESPACE_NAME, e.STATUS, to_char(sum(e.BYTES/1024),'999,999,999,999') SIZE_KB from DBA_UNDO_EXTENTS e group by e.TABLESPACE_NAME, e.STATUS order by e.STATUS
-- undo sizes consumed by active transactions by username SELECT s.username, sum(t.used_ublk) used_undo_blocks from v$session s, v$transaction t where s.saddr = t.ses_addr and t.status='ACTIVE' group by s.username order by s.username
Enabling Automatic Undo Management

select value from v$parameter where upper(name)='UNDO_MANAGEMENT'; alter system set UNDO_MANAGEMENT=AUTO scope=spfile ;
Creating an Undo Tablespace

CREATE UNDO TABLESPACE undotbs2 DATAFILE '/u01/oracle/rbdb1/undo0201.dbf' SIZE 200M REUSE AUTOEXTEND ON;
Setting Startup Undo Tablespace

select value from v$parameter where upper(name)='UNDO_TABLESPACE'; -- if there are multiple undo tablespaces -- after all active transactions have committed, the undo tablespace automatically -- goes from the PENDING OFFLINE mode to the OFFLINE mode. alter system set UNDO_TABLESPACE='UNDOTBS1'; -- the following switches out current undo tablespace to the next available one alter system set UNDO_TABLESPACE = '';
Tuning Undo Retention If the undo tablespace is configured with the AUTOEXTEND option, undo retention tuning is slightly different. In this case, the database tunes the undo retention period to be slightly longer than the longest-running query on the system at that time. For a fixed size undo tablespace, the database tunes the undo retention period based on 85% of the tablespace size, or on the warning alert threshold percentage for space used, whichever is lower. If you want a fixed undo size, use the Undo Advisor to specify the proper undo size for your requirement. You should estimate:
Page 102
o o
The length of your expected longest running query The longest interval that you will require for flashback operations
-- determine the current retention period select to_char(begin_time, 'DD-MON-RR HH24:MI') begin_time, to_char(end_time, 'DD-MON-RR HH24:MI') end_time, tuned_undoretention from v$undostat order by end_time; -- maximum query time SELECT round(MAX(maxquerylen)/60) Minutes FROM v$undostat;
Using Undo Advisor Use OEM or PL/SQL

-- retreive available AWR snaps select SNAP_ID, STARTUP_TIME, BEGIN_INTERVAL_TIME, END_INTERVAL_TIME from DBA_HIST_SNAPSHOT order by SNAP_ID; set serveroutput on DECLARE tid NUMBER; -- task ID tname VARCHAR2(30); -- task name oid NUMBER; BEGIN DBMS_ADVISOR.CREATE_TASK('Undo Advisor', tid, tname, 'Undo Advisor Task'); DBMS_ADVISOR.CREATE_OBJECT(tname, 'UNDO_TBS', null, null, null, null, oid); DBMS_ADVISOR.SET_TASK_PARAMETER(tname, 'TARGET_OBJECTS', oid); DBMS_ADVISOR.SET_TASK_PARAMETER(tname, 'START_SNAPSHOT', 52); DBMS_ADVISOR.SET_TASK_PARAMETER(tname, 'END_SNAPSHOT', 56); DBMS_ADVISOR.SET_TASK_PARAMETER(tname, 'INSTANCE', 1); DBMS_ADVISOR.execute_task(tname); DBMS_OUTPUT.PUT_LINE(tname); end; / SELECT DBA_ADVISOR_RECOMMENDATIONS.GET_TASK_REPORT('TASKNAME') FROM DUAL;
Setting the Undo Retention Period You must set UNDO_RETENTION parameter when: o o o The undo tablespace has the AUTOEXTEND option enabled You want to set undo retention for LOBs You want retention guarantee
select value from v$parameter where upper(name)='UNDO_RETENTION'; alter system set UNDO_RETENTION = 2400;
Enabling Retention Guarantee Enabling retention guarantee can cause multiple DML operations to fail. Use with caution.
select RETENTION from DBA_TABLESPACES where TABLESPACE_NAME='UNDOTBS1'; create undo tablespace undotbs01 .. RETENTION GUARANTEE; alter tablespace undotbs1 RETENTION GUARANTEE; alter tablespace undotbs1 RETENTION NOGUARANTEE;
Dropping an Undo Tablespace

DROP TABLESPACE undotbs_01;
Page 103
To Drop a Corrupt UNDO Tablespace An Undo tablespace containing a corrupted undo rollback segment may lead to one or more of the following: ORA-00376: file xx cannot be read at this time SIMON process being hang waiting for the event "wait for a undo record" forever!
Identify the bad rollback segment(s)(Oracle would report it anyway): -- looking for NEEDS RECOVERY or PARTIALLY AVAILABLE select segment_name, status from dba_rollback_segs where tablespace_name='undotbs_corrupt' and status not in ('OFFLINE', 'ONLINE');
-- Let's say it's _SYSSMU9$
-- (optionally) create a new undo tablespace -and replace it with the current one CREATE UNDO TABLESPACE undotbs2 DATAFILE '/u01/oracle/rbdb1/undo0201.dbf' SIZE 200M REUSE AUTOEXTEND ON; alter system set UNDO_TABLESPACE='UNDOTBS2'; -- Next, create pfile and in it: undo_management = MANUAL _offline_rollback_segments=_SYSSMU9$
-- shutdown the db -- open it using the pfile -- drop the offending rollback segment -- (and the undo tablespace altogether if the other one was created) -- Note: datafiles might not actually be deleted, check them drop tablespace .. including contents and datafiles
-- bounce the db (make it start using the spfile) -- this means it will go back to using the automatic undo
Using Oracle Managed Files (OMF)

OMF datafiles have to be created in one directory
set DB_CREATE_FILE_DEST, DB_CREATE_ONLINE_LOG_DEST_n, and DB_RECOVERY_FILE_DEST -- when OMF enabled CREATE TABLESPACE finance01; ALTER TABLESPACE finance01 ADD DATAFILE 500M; DROP TABLESPACE finance01; -- dbf auto deleted -- create db with OMF db_name=mydb DB_CREATE_FILE_DEST = '/u01/app/oracle/oradata' DB_RECOVERY_FILE_DEST_SIZE = 100M DB_RECOVERY_FILE_DEST = '/u04/app/oracle/oradata' LOG_ARCHIVE_DEST_1 = 'LOCATION = USE_DB_RECOVERY_FILE_DEST' SQL> connect sys/sys_passwd as sysdba Connected to an idle instance. Page 104
SQL> STARTUP NOMOUNT PFILE='initmydb.ora'; SQL> CREATE DATABASE mydb;
Page 105
Managing Schema Objects

Chaching Small Tables in Memory
ALTER TABLE hr.countries CACHE;
Creating Virtual Columns

CREATE TABLE EMPLOYEES ( empno number PRIMARY KEY, sal NUMBER (8,2) NOT NULL, annual_sal AS (sal*12), CONSTRAINT MaxAnSal CHECK (annual_sal BETWEEN 0 AND 2000000) ); SELECT TABLE_NAME, COLUMN_NAME, DATA_DEFAULT FROM DBA_TAB_COLUMNS WHERE TABLE_NAME='EMPLOYEES' AND COLUMN_NAME='ANNUAL_SAL'; ALTER TABLE employees ADD (income AS (salary*commission_pct));
Creating Partitioned Tables

/* range partitions */ CREATE TABLE student_history (student_id NUMBER(10), degree VARCHAR2(3), graduation_date DATE, final_gpa NUMBER) PARTITION BY RANGE (graduation_date) (PARTITION p_1997 VALUES LESS THAN (TO_DATE('01-JUN-1997','DD-MON-YYYY')) TABLESPACE ts1, PARTITION p_1998 VALUES LESS THAN (TO_DATE('01-JUN-1998','DD-MON-YYYY')) TABLESPACE ts2, PARTITION p_1999 VALUES LESS THAN (TO_DATE('01-JUN-1999','DD-MON-YYYY')) TABLESPACE ts3, PARTITION p_other VALUES LESS THAN (maxvalue) TABLESPACE ts4); CREATE TABLE sales_data (ticket_no NUMBER, sale_year INT NOT NULL, sale_month INT NOT NULL, sale_day INT NOT NULL) PARTITION BY RANGE (sale_year, sale_month, sale_day) (PARTITION sales_q1 VALUES LESS THAN (2008, 04, 01) TABLESPACE PARTITION sales_q2 VALUES LESS THAN (2008, 07, 01) TABLESPACE PARTITION sales_q3 VALUES LESS THAN (2008, 10, 01) TABLESPACE PARTITION sales_q4 VALUES LESS THAN (2009, 01, 01) TABLESPACE
ts1, ts2, ts3, ts4);
----------------------------------------------------------------/* Interval Partitioning */ -- You cant use a partitioning key that includes more than one column -- system generated partions have names SYS_Pn CREATE TABLE interval_sales ( prod_id NUMBER(6), cust_id NUMBER, time_id DATE, ..)
Page 106
PARTITION BY RANGE (time_id) INTERVAL(NUMTOYMINTERVAL(1, 'MONTH')) STORE IN (ts5, ts6, ts7) -- optional ( PARTITION ts1 VALUES LESS THAN (TO_DATE('1-1-2006', PARTITION ts2 VALUES LESS THAN (TO_DATE('1-1-2007', PARTITION ts3 VALUES LESS THAN (TO_DATE('1-7-2008', PARTITION ts4 VALUES LESS THAN (TO_DATE('1-1-2009',
'DD-MM-YYYY')), 'DD-MM-YYYY')), 'DD-MM-YYYY')), 'DD-MM-YYYY')) );
SELECT TABLE_NAME, PARTITION_NAME, PARTITION_POSITION, HIGH_VALUE FROM Remote DBA_TAB_PARTITIONS WHERE TABLE_NAME='POS_DATA' ORDER BY PARTITION_NAME; create table res ( res_id number not null, res_date date, hotel_id number(3), guest_id number ) partition by range (res_id) interval (100) store in (users) ( partition p1 values less than (101) );
-- selecting from the generated partition -- classic method select * from interval_sales partition for (SYS_P81); -- the other method (expanded partition access syntax) select * from interval_sales partition for (to_date('15-AUG-2009','dd-mon-yyyy')); alter table res truncate partition for (901);
-- a range partion can be converted to interval partition alter table pos_data_range set INTERVAL(NUMTOYMINTERVAL(1, 'MONTH')); -- convert an interval partition to range partition alter table pos_data_range set INTERVAL(); -- interval can be modified alter table pos_data set INTERVAL(NUMTOYMINTERVAL(3, 'MONTH')); -- round robin tablespaces can be modified alter table pos_data set STORE IN(tablespace1, tablespace2, tablespace3); ----------------------------------------------------------------/* Hash Partitioning */ -- used when range distribution is not predictable and for high cardinality columns -- Updates that would cause a record to move across partition boundaries are not allowed CREATE TABLE sales_data (ticket_no NUMBER, sale_year INT NOT NULL, sale_month INT NOT NULL, sale_day INT NOT NULL ) PARTITION BY HASH (ticket_no) PARTITIONS 4 STORE IN (ts1,ts2,ts3,ts4); ----------------------------------------------------------------/* List Partitioning */ CREATE TABLE sales_data (ticket_no NUMBER, sale_year INT NOT NULL, Page 107
sale_month INT NOT NULL, sale_day INT NOT NULL, destination_city CHAR(3), start_city CHAR(3)) PARTITION BY LIST (start_city) (PARTITION northeast_sales values ('NYC','BOS','PEN') TABLESPACE ts1, PARTITION southwest_sales values ('DFW','ORL','HOU') TABLESPACE ts2, PARTITION pacificwest_sales values('SAN','LOS','WAS') TABLESPACE ts3, PARTITION southeast_sales values ('MIA','CHA','ATL') TABLESPACE ts4); ----------------------------------------------------------------/* Reference Partitioning */ -- put child table data into parent table partitions -- You can use all partitioning strategies with reference partitioning, except interval partitioning CREATE TABLE orders ( order_id NUMBER(12), order_date DATE, order_mode VARCHAR2(8), customer_id NUMBER(6), order_status NUMBER(2), order_total NUMBER(8,2), sales_rep_id NUMBER(6), promotion_id NUMBER(6), CONSTRAINT orders_pk PRIMARY KEY(order_id)) PARTITION BY RANGE(order_date) ( PARTITION Q1_2005 VALUES LESS THAN (TO_DATE('01-APR-2005','DD-MON-YYYY')), PARTITION Q2_2005 VALUES LESS THAN (TO_DATE('01-JUL-2005','DD-MON-YYYY')), PARTITION Q3_2005 VALUES LESS THAN (TO_DATE('01-OCT-2005','DD-MON-YYYY')), PARTITION Q4_2005 VALUES LESS THAN (TO_DATE('01-JAN-2006','DD-MON-YYYY')) ); CREATE TABLE order_items ( order_id NUMBER(12) NOT NULL, line_item_id NUMBER(3) NOT NULL, product_id NUMBER(6) NOT NULL, unit_price NUMBER(8,2), quantity NUMBER(8), CONSTRAINT order_items_fk FOREIGN KEY(order_id) REFERENCES orders(order_id) ) PARTITION BY REFERENCE(order_items_fk); ----------------------------------------------------------------/* Virtual Column-Based Partitioning */ -- ENABLE ROW MOVEMENT clause ensures row migration among partitions when virtual column value changes CREATE TABLE sales ( prod_id NUMBER(6) NOT NULL, cust_id NUMBER NOT NULL, time_id DATE NOT NULL, channel_id CHAR(1) NOT NULL, promo_id NUMBER(6) NOT NULL, quantity_sold NUMBER(3) NOT NULL, amount_sold NUMBER(10,2) NOT NULL, total_amount AS (quantity_sold * amount_sold) ) PARTITION BY RANGE (time_id) INTERVAL (NUMTOYMINTERVAL(1,'MONTH')) SUBPARTITION BY RANGE(total_amount) SUBPARTITION TEMPLATE ( SUBPARTITION p_small VALUES LESS THAN (1000),
Page 108
SUBPARTITION p_medium VALUES LESS THAN (5000), SUBPARTITION p_large VALUES LESS THAN (10000), SUBPARTITION p_extreme VALUES LESS THAN (MAXVALUE) ) (PARTITION sales_before_2007 VALUES LESS THAN (TO_DATE('01-JAN-2007','dd-MON-yyyy'))) ENABLE ROW MOVEMENT PARALLEL; ----------------------------------------------------------------/* Virtual Column-Based Partitioning */ -- ENABLE ROW MOVEMENT clause ensures row migration among partitions when virtual column value changes CREATE TABLE sales ( prod_id NUMBER(6) NOT NULL, cust_id NUMBER NOT NULL, time_id DATE NOT NULL, channel_id CHAR(1) NOT NULL, promo_id NUMBER(6) NOT NULL, quantity_sold NUMBER(4) NOT NULL, amount_sold NUMBER(4) NOT NULL, total_amounts AS (quantity_sold * amount_sold) ) PARTITION BY RANGE (time_id) INTERVAL (NUMTOYMINTERVAL(1,'MONTH')) SUBPARTITION BY RANGE(total_amounts) SUBPARTITION TEMPLATE ( SUBPARTITION p_small VALUES LESS THAN (1000), SUBPARTITION p_medium VALUES LESS THAN (5000), SUBPARTITION p_large VALUES LESS THAN (10000), SUBPARTITION p_extreme VALUES LESS THAN (MAXVALUE) ) (PARTITION sales_before_2007 VALUES LESS THAN (TO_DATE('01-JAN-2007','dd-MON-yyyy'))) ENABLE ROW MOVEMENT PARALLEL; ----------------------------------------------------------------/* System Partitioning */ -- the application decided where to store the data CREATE TABLE docs ( ID NUMBER, Name VARCHAR2(255), Desc VARCHAR2(1000)) PARTITION BY SYSTEM ( PARTITION docs_p1 TABLESPACE ts1, PARTITION docs_p2 TABLESPACE ts2, PARTITION docs_p3 TABLESPACE ts3, PARTITION docs_p4 TABLESPACE ts4 ); -- PARTITION must be stated INSERT INTO docs PARTITION (ts1) VALUES (1, 'Oracle 11g New Features', 'New features in Oracle 11g Database.'); -- with DELETE command, PARTITION can be stated DELETE FROM docs PARTITION (ts2) WHERE doc_id=1002; DELETE FROM docs PARTITION (ts2); -- PARTITION can be used in queries to target specific partitions SELECT COUNT(*) FROM docs PARTITION (ts1) ----------------------------------------------------------------/* Range-Hash Partitioning */ CREATE TABLE scout_gear (equipno NUMBER,equipname VARCHAR(32),price NUMBER)
Page 109
PARTITION BY RANGE (equipno) SUBPARTITION BY HASH(equipname) SUBPARTITIONS 8 STORE IN (ts1, ts2, ts3, ts4) (PARTITION p1 VALUES LESS THAN (1000), PARTITION p2 VALUES LESS THAN (2000), PARTITION p3 VALUES LESS THAN (3000), PARTITION p4 VALUES LESS THAN (MAXVALUE)); ----------------------------------------------------------------/* Range-List Partitioning */ CREATE TABLE regional_sales (ticket_no NUMBER, sale_year INT NOT NULL, sale_month INT NOT NULL, sale_day DATE, destination_city CHAR(3), start_city CHAR(3)) PARTITION BY RANGE(sale_day) SUBPARTITION BY LIST (start_city) (PARTITION q1_2004 VALUES LESS THAN (TO_DATE('1-APR-2004','DD-MON-YYYY')) TABLESPACE ts1 (SUBPARTITION q12004_northeast_sales VALUES ('NYC','BOS','PEN'), SUBPARTITION q12004_southwest_sales VALUES ('DFW','ORL','HOU'), SUBPARTITION q12004_pacificwest_sales VALUES ('SAN','LOS','WAS'), SUBPARTITION q12004_southeast_sales VALUES ('MIA','CHA','ATL') ), PARTITION q2_2004 VALUES LESS THAN (TO_DATE('1-JUL-2004','DD-MON-YYYY')) TABLESPACE ts2 (SUBPARTITION q22004_northeast_sales VALUES ('NYC','BOS','PEN'), SUBPARTITION q22004_southwest_sales VALUES ('DFW','ORL','HOU'), SUBPARTITION q22004_pacificwest_sales VALUES ('SAN','LOS','WAS'), SUBPARTITION q22004_southeast_sales VALUES ('MIA','CHA','ATL') ), PARTITION q3_2004 VALUES LESS THAN (TO_DATE('1-OCT-2004','DD-MON-YYYY')) TABLESPACE ts3 (SUBPARTITION q32004_northeast_sales VALUES ('NYC','BOS','PEN'), SUBPARTITION q32004_southwest_sales VALUES ('DFW','ORL','HOU'), SUBPARTITION q32004_pacificwest_sales VALUES ('SAN','LOS','WAS'), SUBPARTITION q32004_southeast_sales VALUES ('MIA','CHA','ATL') ), PARTITION q4_2004 VALUES LESS THAN (TO_DATE('1-JAN-2005','DD-MON-YYYY')) TABLESPACE ts4 (SUBPARTITION q42004_northeast_sales VALUES ('NYC','BOS','PEN'), SUBPARTITION q42004_southwest_sales VALUES ('DFW','ORL','HOU'), SUBPARTITION q42004_pacificwest_sales VALUES ('SAN','LOS','WAS'), SUBPARTITION q42004_southeast_sales VALUES ('MIA','CHA','ATL') ) ); ----------------------------------------------------------------/* Interval-Range Partitioned Tables or Range-Range */ CREATE TABLE sales ( prod_id NUMBER(6), cust_id NUMBER, time_id DATE, channel_id CHAR(1), promo_id NUMBER(6), quantity_sold NUMBER(3), amount_sold NUMBER(10,2) ) PARTITION BY RANGE (time_id) INTERVAL (NUMTODSINTERVAL(1,'DAY')) SUBPARTITION BY RANGE(amount_sold) Page 110
SUBPARTITION TEMPLATE ( SUBPARTITION p_low VALUES LESS THAN (1000), SUBPARTITION p_medium VALUES LESS THAN (4000), SUBPARTITION p_high VALUES LESS THAN (8000), SUBPARTITION p_ultimate VALUES LESS THAN (maxvalue)) ( PARTITION before_2000 VALUES LESS THAN (TO_DATE('01-JAN-2000','dd-MON-yyyy'))); ----------------------------------------------------------------/* Interval-List Partitioned Tables */ CREATE TABLE sales ( prod_id NUMBER(6) , cust_id NUMBER , time_id DATE , channel_id CHAR(1) , promo_id NUMBER(6) , quantity_sold NUMBER(3) , amount_sold NUMBER(10,2) ) PARTITION BY RANGE (time_id) INTERVAL (NUMTODSINTERVAL(1,'DAY')) SUBPARTITION BY LIST (channel_id) SUBPARTITION TEMPLATE (SUBPARTITION p_catalog VALUES ('C') , SUBPARTITION p_internet VALUES ('I') , SUBPARTITION p_partners VALUES ('P') , SUBPARTITION p_direct_sales VALUES ('S') , SUBPARTITION p_tele_sales VALUES ('T')) ( PARTITION before_2000 VALUES LESS THAN (TO_DATE('01-JAN-2000','dd-MON-yyyy')));
Partition Maintenance Operations

Range and list partitions cant be coalesced. Hash partitions cant be dropped, split, or merged. Only list partitions allow the modification of partitions by adding and dropping the partition values.
-- Adding Partitions ALTER TABLE ticket_sales ADD PARTITION sales_quarter5 VALUES LESS THAN (TO_DATE('1-APR-2005','DD-MON-YYYY')) TABLESPACE ts5; -- Splitting a Partition ALTER TABLE ticket_sales SPLIT PARTITION ticket_sales01 AT (2000) INTO (PARTITION ticket_sales01A, ticket_sales01B); -- Merging Partitions ALTER TABLE ticket_sales MERGE PARTITIONS ticket_sales01, ticket_sales02 INTO PARTITION ticket_sales02; -- Renaming Partitions ALTER TABLE RENAME PARTITION fight_sales01 TO quarterly_sales01; -- Exchanging Partitions -- convert a regular nonpartitioned table into a partition of a partitioned table -- it only renames the source and destination ALTER TABLE ticket_sales EXCHANGE PARTITION ticket_sales02 WITH ticket_sales03; -- Dropping Partitions
Page 111
-- use UPDATE GLOBAL INDEXES keyword if data exists in the partition ALTER TABLE ticket_sales DROP PARTITION ticket_sales01; -- Coalescing Partitions -- applied on hash-partitioned and list-partitioned -- to reduce number of partitions ALTER TABLE ticket_sales COALESCE PARTITION;
Setting Deferred Segment Creation

When the table (heap or partitioned) is created, no segment is assigned until a row is inserted.
(Oracle 11g) -- db level: show parameter DEFERRED_SEGMENT_CREATION alter system set deferred_segment_creation=true scope=both ;
-- table level create table test ( .. ) SEGMENT CREATION DEFERRED partition by .. ; create table test ( .. ) SEGMENT CREATION IMMEDIATE ; -- (11.2.0.2) If you want to create the segments for objects where -- SEGMENT CREATION DEFERRED is set without waiting for -- inserting any rows: -- all the objects in a schema: conn / as sysdba begin DBMS_SPACE_ADMIN.MATERIALIZE_DEFERRED_SEGMENTS ( schema_name =>'SA'); end; -- specific table: begin DBMS_SPACE_ADMIN.MATERIALIZE_DEFERRED_SEGMENTS ( schema_name =>'SA', table_name=>'EMP'); end; -- specific partition begin DBMS_SPACE_ADMIN.MATERIALIZE_DEFERRED_SEGMENTS ( schema_name =>'SA', table_name=>'EMP', partition_name=>'PAR01'); end;
Creating Multiple Tables and Views in a Single Operation

If one included statement fails, non of the other statements will be executed.
CREATE SCHEMA AUTHORIZATION scott CREATE TABLE dept ( deptno NUMBER(3,0) PRIMARY KEY, dname VARCHAR2(15), loc VARCHAR2(25)) CREATE TABLE emp ( empno NUMBER(5,0) PRIMARY KEY, ename VARCHAR2(15) NOT NULL, job VARCHAR2(10), mgr NUMBER(5,0),
Page 112
hiredate DATE DEFAULT (sysdate), sal NUMBER(7,2), comm NUMBER(7,2), deptno NUMBER(3,0) NOT NULL CONSTRAINT dept_fkey REFERENCES dept) CREATE VIEW sales_staff AS SELECT empno, ename, sal, comm FROM emp WHERE deptno = 30 WITH CHECK OPTION CONSTRAINT sales_staff_cnst GRANT SELECT ON sales_staff TO human_resources;
Collecting Object Statistics

Collecting Index Statistics granularity parameter takes: o o o o o 'DEFAULT' - gathers global and partition-level statistics 'SUBPARTITION' - gathers subpartition-level statistics 'PARTITION' - gathers partition-level statistics 'GLOBAL' - gathers global statistics 'ALL' - gathers all (subpartition, partition, and global)
begin DBMS_STATS.GATHER_INDEX_STATS ( ownname =>'HR', indname =>'EMP_DEPARTMENT_IX', partname =>'', -- DEFAULT NULL estimate_percent => DBMS_STATS.AUTO_SAMPLE_SIZE, -- 0.1 to 100 DEFAULT to_estimate_percent_type (GET_PARAM('ESTIMATE_PERCENT')) stattab => null, -- DEFAULT NULL statid => null, -- DEFAULT NULL statown => null, -- DEFAULT NULL degree => null, -- parallelism NUMBER DEFAULT to_degree_type(get_param('DEGREE')) granularity => 'ALL', -- DEFAULT GET_PARAM('GRANULARITY') no_invalidate => false, -- BOOLEAN DEFAULT to_no_invalidate_type (GET_PARAM('NO_INVALIDATE')), force => false -- BOOLEAN DEFAULT FALSE ); end; / Collecting Table Statistics method_opt Accepts: o o o FOR ALL [INDEXED | HIDDEN] COLUMNS [size_clause] FOR COLUMNS [size clause] column|attribute [size_clause] [,column|attribute [size_clause]...] size_clause is defined as size_clause := SIZE {integer | REPEAT | AUTO | SKEWONLY}
- integer : Number of histogram buckets. Must be in the range [1,254]. - REPEAT : Collects histograms only on the columns that already have histograms. - AUTO : Oracle determines the columns to collect histograms based on data distribution and the workload of the columns.
Page 113
- SKEWONLY : Oracle determines the columns to collect histograms based on the data distribution of the columns. The default is FOR ALL COLUMNS SIZE AUTO granularity takes: 'ALL', 'AUTO', 'DEFAULT', 'GLOBAL', 'GLOBAL AND PARTITION', 'PARTITION', 'SUBPARTITION' begin DBMS_STATS.GATHER_TABLE_STATS ( ownname =>'HR', tabname =>'NAMES', partname =>'', estimate_percent => DBMS_STATS.AUTO_SAMPLE_SIZE, -- NUMBER DEFAULT to_estimate_percent_type(get_param('ESTIMATE_PERCENT')) block_sample => false, -- DEFAULT FALSE method_opt =>'FOR ALL COLUMNS SIZE AUTO' , -- DEFAULT get_param('METHOD_OPT') degree =>1, -- DEFAULT to_degree_type(get_param('DEGREE')) granularity =>'ALL' , -- DEFAULT GET_PARAM('GRANULARITY') cascade =>true, -- to_cascade_type(get_param('CASCADE')) stattab =>'', statid =>'', statown =>'', no_invalidate => false, -- DEFAULT to_no_invalidate_type (get_param('NO_INVALIDATE')) force => false ); end; / Collecting Schema Statistics method_opt Accepts: o o o FOR ALL [INDEXED | HIDDEN] COLUMNS [size_clause] FOR COLUMNS [size clause] column|attribute [size_clause] [,column|attribute [size_clause]...] size_clause is defined as size_clause := SIZE {integer | REPEAT | AUTO | SKEWONLY}
- integer : Number of histogram buckets. Must be in the range [1,254]. - REPEAT : Collects histograms only on the columns that already have histograms. - AUTO : Oracle determines the columns to collect histograms based on data distribution and the workload of the columns. - SKEWONLY : Oracle determines the columns to collect histograms based on the data distribution of the columns. The default is FOR ALL COLUMNS SIZE AUTO granularity takes: 'ALL', 'AUTO', 'DEFAULT', 'GLOBAL', 'GLOBAL AND PARTITION', 'PARTITION', 'SUBPARTITION' options takes : GATHER, GATHER AUTO, GATHER STALE, GATHER EMPTY. It also takes LIST STALE, LIST EMPTY, LIST AUTO (you should use an overloaded procedure different from the example above).
Note: GET_PARAM('..')is used till version 10g. In 11g, DBMS_STATS.GET_PREFS should be used. begin DBMS_STATS.GATHER_SCHEMA_STATS ( ownname =>'HR',
Page 114
estimate_percent =>DBMS_STATS.AUTO_SAMPLE_SIZE, --0.1-100 DEFAULT to_estimate_percent_type(get_param('ESTIMATE_PERCENT')) block_sample =>FALSE, method_opt =>'FOR ALL COLUMNS SIZE AUTO', -- DEFAULT get_param('METHOD_OPT') degree =>NULL, -- to_degree_type(get_param('DEGREE')), granularity =>'ALL', -- DEFAULT GET_PARAM('GRANULARITY'), cascade =>TRUE, -- DEFAULT to_cascade_type(get_param('CASCADE')), stattab =>'', -- DEFAULT NULL, statid =>'', -- DEFAULT NULL, options =>'GATHER', -- DEFAULT 'GATHER', statown =>'', -- DEFAULT NULL, no_invalidate =>false,-- DEFAULT to_no_invalidate_type (get_param('NO_INVALIDATE') force => FALSE); end; /
begin DBMS_STATS.GATHER_SCHEMA_STATS ( ownname =>'HR', estimate_percent =>NULL, method_opt =>'FOR ALL COLUMNS SIZE AUTO', degree => DBMS_STATS.DEFAULT_DEGREE, granularity =>'ALL', cascade =>TRUE, options =>'GATHER AUTO'); end; /
Validating Tables, Indexes, Clusters, and Materialized Views

Validating an object means verifying the integrity of the structure of it. If a corruption is reported, re-create the object. ANALYZE ANALYZE ANALYZE ANALYZE TABLE TABLE TABLE INDEX emp VALIDATE STRUCTURE; emp VALIDATE STRUCTURE CASCADE; emp VALIDATE STRUCTURE CASCADE ONLINE; loc_country_ix VALIDATE STRUCTURE;
Listing Chained and Migrated Rows of Tables and Clusters

-- set up -- following script creates CHAINED_ROWS table @ $ORACLE_HOME$\RDBMS\ADMIN\utlchain.sql -- handling chained and migrated rows -- 1 ANALYZE TABLE order_hist LIST CHAINED ROWS; ANALYZE CLUSTER emp_dept LIST CHAINED ROWS INTO CHAINED_ROWS; -- 2 SELECT * FROM CHAINED_ROWS WHERE TABLE_NAME = 'ORDER_HIST'; -- 3 if you have too many rows -- intermediate table to hold the chained rows
Page 115
CREATE TABLE int_order_hist AS SELECT * FROM order_hist WHERE ROWID IN (SELECT HEAD_ROWID FROM CHAINED_ROWS WHERE TABLE_NAME = 'ORDER_HIST'); -- 4 DELETE FROM order_hist WHERE ROWID IN (SELECT HEAD_ROWID FROM CHAINED_ROWS WHERE TABLE_NAME = 'ORDER_HIST'); -- 5 INSERT INTO order_hist SELECT * FROM int_order_hist; -- 6 DROP TABLE int_order_history; -- 7 DELETE FROM CHAINED_ROWS WHERE TABLE_NAME = 'ORDER_HIST'; -- 8 Use the ANALYZE statement again, and query the output table. Any rows that appear in the output table are chained. You can eliminate chained rows only by increasing your data block size.
Truncating Tables or Clusters

TRUNCATE TABLE emp_dept DROP STORAGE; -- default, leave space of MINIEXTENTS TRUNCATE TABLE emp_dept REUSE STORAGE; TRUNCATE TABLE emp DROP ALL STORAGE; -- all segments will be dropped
Enabling and Disabling Triggers

select table_name, trigger_name from user_triggers where status='DISABLED'; ALTER TRIGGER reorder ENABLE; ALTER TRIGGER reorder DISABLE; ALTER TABLE inventory ENABLE ALL TRIGGERS; ALTER TABLE inventory DISABLE ALL TRIGGERS;
Managing Integrity Constraints

Setting Constraint States and Deferability Constraint states: ENABLE, VALIDATE : inserts are checked, existing data in the table is checked ENABLE, NOVALIDATE (fast): inserts are checked, existing data in the table is NOT checked DISABLE, VALIDATE : DML on constrained columns disallowed DISABLE, NOVALIDATE : inserts are NOT checked, existing data in the table is NOT checked Constraint Checking (Deferability) Nondeferred: enforced at the end of every DML statement. Deferred : enforced when a transaction is committed. see next section for altering a constraint Oracle DBA Code Examples
Page 116
CONSTRAINT ... [[NOT] DEFERRABLE INITIALLY [DEFERRED|IMMEDIATE]] [ENABLE|DISABLE] [VALIDATE|NOVALIDATE] ALTER SESSION SET CONSTRAINT[S] = {IMMEDIATE|DEFERRED|DEFAULT} -- takes effect in TRANSACTION level -- the trans is rolled back, if COMMIT fails -- you can use SET CONSTRAINTS ALL IMMEDIATE to see whether COMMIT will fail before it -- really commits SET CONSTRAINT | CONSTRAINTS {constraint |ALL } {IMMEDIATE|DEFERRED}
SELECT CONSTRAINT_NAME, CONSTRAINT_TYPE,STATUS,DEFERRABLE,DEFERRED,VALIDATED FROM USER_CONSTRAINTS WHERE CONSTRAINT_NAME=''; ALTER TABLE emp ADD PRIMARY KEY (empno) DISABLE; ALTER TABLE countries ADD (UNIQUE(country_name) ENABLE NOVALIDATE); ALTER TABLE myTable ADD CONSTRAINT uq UNIQUE (id) DEFERRABLE INITIALLY DEFERRED; -- takes effect in TRANSACTION level -- not deferrable constraints are NOT affected set constraint job_fk deferred; set constraint job_fk immediate; set constraints all deferred; Modifying, Renaming, or Dropping Existing Integrity Constraints ALTER TABLE dept DISABLE CONSTRAINT dname_ukey; ALTER TABLE dept DISABLE PRIMARY KEY KEEP INDEX, DISABLE UNIQUE (dname, loc) KEEP INDEX; -- all FKs are also disabled ALTER TABLE dept DISABLE PRIMARY KEY CASCADE; ALTER TABLE dept ENABLE NOVALIDATE CONSTRAINT dname_ukey; ALTER TABLE dept ENABLE NOVALIDATE PRIMARY KEY, ENABLE NOVALIDATE UNIQUE (dname, loc); ALTER TABLE dept MODIFY CONSTRAINT dname_key VALIDATE; ALTER TABLE dept MODIFY PRIMARY KEY ENABLE NOVALIDATE; ALTER TABLE dept RENAME CONSTRAINT dname_ukey TO dname_unikey; ALTER TABLE dept DROP UNIQUE (dname, loc); ALTER TABLE emp DROP PRIMARY KEY KEEP INDEX, DROP CONSTRAINT dept_fkey; Reporting Constraint Exceptions The target is to put all the rows that voilate sepcific integrity contraints in the EXCEPTIONS table. 1. create the EXCEPTIONS table ORACLE_HOME\RDBMS\ADMIN\utlexcpt.sql 2. ALTER TABLE dept ENABLE PRIMARY KEY EXCEPTIONS INTO EXCEPTIONS; 3. SELECT * FROM EXCEPTIONS; 4. join the master table with the EXCEPTIONS SELECT deptno, dname, loc FROM dept, EXCEPTIONS WHERE EXCEPTIONS.constraint = 'SYS_C00610' AND dept.rowid = EXCEPTIONS.row_id; 5. fix the incorrect data 6. DELETE FROM EXCEPTIONS WHERE constraint = 'SYS_C00610';
Page 117
Obtaining Information on Constraints select * from USER_CONSTRAINTS; select * from USER_CONS_COLUMNS;
Renaming Schema Objects

rename mytable to mytable2
Managing Object Dependencies

Manually Recompiling Views Procedures and Packages ALTER ALTER ALTER ALTER VIEW emp_dept COMPILE; PROCEDURE update_salary COMPILE; PACKAGE acct_mgmt COMPILE BODY; PACKAGE acct_mgmt COMPILE PACKAGE;
Switching to a Different Schema

CONNECT scott/tiger ALTER SESSION SET CURRENT_SCHEMA = joe; -- the following is equivalent to SELECT * FROM joe.emp SELECT * FROM emp;
Using DBMS_METADATA to Display Information About Schema Objects

SET LONG 9999 -- storage clauses are not to be returned EXECUTE DBMS_METADATA.SET_TRANSFORM_PARAM(DBMS_METADATA.SESSION_TRANSFORM,'STORAGE',f alse); -- fetch the DDL of all tables (relational and objects) SELECT DBMS_METADATA.GET_DDL('TABLE',u.table_name) FROM USER_ALL_TABLES u WHERE u.nested='NO' AND (u.iot_type is null or u.iot_type='IOT'); -- return the parameter value to its default EXECUTE DBMS_METADATA.SET_TRANSFORM_PARAM(DBMS_METADATA.SESSION_TRANSFORM,'DEFAULT');
Specifying Storage Parameters at Object Creation

Objects include: Tables, Clusters, Indexes, LOBs, Varrays, Nested Tables, MViews CREATE TABLE MYTABLE(COL1 NUMBER, COL2 VARCHAR2(100)) STORAGE (INITIAL 10M BUFFER_POOL DEFAULT); -- MINEXTENTS SELECT EXTENT_ID, BYTES/1024 KB FROM DBA_EXTENTS WHERE SEGMENT_TYPE='TABLE' AND SEGMENT_NAME='MYTABLE';
Page 118
CREATE TABLE ... ( .., CONSTRAINT UQ)CON UNIQUE (PK_id) USING INDEX TABLESPACE MYTBS STORAGE (INITIAL 8K ) ); CREATE MATERIALIZED VIEW .. TABLESPACE MYTBS STORAGE (INITIAL 50K) USING INDEX STORAGE (INITIAL 25K) ...
Managing Resumable Space Allocation

Enabling Resumable Space Allocation -- the user should have this priv grant RESUMABLE to hr; select value from v$parameter where upper(name)='RESUMABLE_TIMEOUT'; -- to -- 0 alter -- or alter enable resumable space allocation system wide to disable system set RESUMABLE_TIMEOUT = 3600; in session level session set RESUMABLE_TIMEOUT = 3600;
-- beside the RESUMABLE_TIMEOUT parameter ALTER SESSION ENABLE RESUMABLE; ALTER SESSION ENABLE RESUMABLE TIMEOUT 3600; -- default 7200 -- 4340 is SID. NULL for current session EXEC DBMS_RESUMABLE.set_session_timeout(4340,18000); ALTER SESSION DISABLE RESUMABLE; -- naming resumable statements -- default is 'User username(userid), Session sessionid, Instance instanceid' -- name is used in the DBA_RESUMABLE and USER_RESUMABLE views ALTER SESSION ENABLE RESUMABLE TIMEOUT 3600 NAME 'insert into table'; Detecting Suspended Statements The AFTER SUSPEND System Event and Trigger is executed after a SQL statement has been suspended. -- source: Oracle Documentation CREATE OR REPLACE TRIGGER resumable_default AFTER SUSPEND ON DATABASE DECLARE /* declare transaction in this trigger is autonomous */ /* this is not required because transactions within a trigger are always autonomous */ PRAGMA AUTONOMOUS_TRANSACTION; cur_sid NUMBER; cur_inst NUMBER; errno NUMBER;
Page 119
err_type VARCHAR2; object_owner VARCHAR2; object_type VARCHAR2; table_space_name VARCHAR2; object_name VARCHAR2; sub_object_name VARCHAR2; error_txt VARCHAR2; msg_body VARCHAR2; ret_value BOOLEAN; mail_conn UTL_SMTP.CONNECTION; BEGIN -- Get session ID SELECT DISTINCT(SID) INTO cur_SID FROM V$MYSTAT; -- Get instance number cur_inst := userenv('instance'); -- Get space error information ret_value := DBMS_RESUMABLE.SPACE_ERROR_INFO(err_type,object_type,object_owner,table_space _name,object_name, sub_object_name); /* ------*/
If the error is related to undo segments, log error, send email to DBA, and abort the statement. Otherwise, set timeout to 8 hours. sys.rbs_error is a table which is to be created by a DBA manually and defined as (sql_text VARCHAR2(1000), error_msg VARCHAR2(4000), suspend_time DATE)
IF OBJECT_TYPE = 'UNDO SEGMENT' THEN /* LOG ERROR */ INSERT INTO sys.rbs_error ( SELECT SQL_TEXT, ERROR_MSG, SUSPEND_TIME FROM DBA_RESUMABLE WHERE SESSION_ID = cur_sid AND INSTANCE_ID = cur_inst); SELECT ERROR_MSG INTO error_txt FROM DBMS_RESUMABLE WHERE SESSION_ID = cur_sid and INSTANCE_ID = cur_inst; -- Send email to receipient via UTL_SMTP package msg_body:='Subject: Space Error Occurred Space limit reached for undo segment ' || object_name || on ' || TO_CHAR(SYSDATE, 'Month dd, YYYY, HH:MIam') || '. Error message was ' || error_txt; mail_conn := UTL_SMTP.OPEN_CONNECTION('localhost', 25); UTL_SMTP.HELO(mail_conn, 'localhost'); UTL_SMTP.MAIL(mail_conn, 'sender@localhost'); UTL_SMTP.RCPT(mail_conn, 'recipient@localhost'); UTL_SMTP.DATA(mail_conn, msg_body); UTL_SMTP.QUIT(mail_conn); -- Abort the statement DBMS_RESUMABLE.ABORT(cur_sid); ELSE -- Set timeout to 8 hours DBMS_RESUMABLE.SET_TIMEOUT(28800);
Page 120
END IF; /* commit autonomous transaction */ COMMIT; END; / Obtaining Information about Suspended Statements
SELECT USER_ID,SESSION_ID,INSTANCE_ID, COORD_INSTANCE_ID,COORD_SESSION_ID,STATUS, TIMEOUT,START_TIME,SUSPEND_TIME, RESUME_TIME,NAME,SQL_TEXT, ERROR_NUMBER,ERROR_PARAMETER1,ERROR_PARAMETER2, ERROR_PARAMETER3,ERROR_PARAMETER4,ERROR_PARAMETER5, ERROR_MSG FROM DBA_RESUMABLE; SELECT SID,SEQ#,EVENT, P1TEXT,P1,P1RAW, P2TEXT,P2,P2RAW, P3TEXT,P3,P3RAW, WAIT_CLASS_ID,WAIT_CLASS#,WAIT_CLASS, WAIT_TIME,SECONDS_IN_WAIT,STATE FROM V$SESSION_WAIT WHERE EVENT LIKE '%statement suspended%' exec DBMS_RESUMABLE.ABORT(159); select DBMS_RESUMABLE.GET_SESSION_TIMEOUT(159) from dual; exec DBMS_RESUMABLE.SET_SESSION_TIMEOUT(sessionID=>159,timeout=>8000);
Reclaiming Wasted Space

Displaying Information About Space Usage for Schema Objects /* Using PL SQL */ SET SERVEROUTPUT ON declare -- IN vars v_segment_owner VARCHAR2(30) :='HR'; v_segment_name VARCHAR2(30) :='NAMES'; v_segment_type VARCHAR2(30) :='TABLE'; -- TABLE, TABLE PARTITION, TABLE SUBPARTITION, INDEX, INDEX PARTITION, INDEX SUBPARTITION, CLUSTER, LOB, LOB PARTITION, LOB SUBPARTITION v_partition_name VARCHAR2(30) :=NULL; -- OUT vars v_total_blocks NUMBER; v_total_bytes NUMBER; v_unused_blocks NUMBER; v_unused_bytes NUMBER; v_last_used_extent_file_id NUMBER; -- file id of last Extend containing data v_filename varchar2(300); v_last_used_extent_block_id NUMBER; v_last_used_block NUMBER; begin DBMS_SPACE.UNUSED_SPACE ( segment_owner => v_segment_owner, segment_name => v_segment_name , segment_type => v_segment_type,
Page 121
total_blocks => v_total_blocks , total_bytes => v_total_bytes , unused_blocks => v_unused_blocks, unused_bytes => v_unused_bytes , last_used_extent_file_id => v_last_used_extent_file_id , last_used_extent_block_id =>v_last_used_extent_block_id , last_used_block => v_last_used_block , partition_name => v_partition_name ); if v_last_used_extent_file_id is not null then select file_name into v_filename from dba_data_files where file_id = v_last_used_extent_file_id ; end if; DBMS_OUTPUT.PUT_LINE('Total Blocks : '||v_total_blocks); DBMS_OUTPUT.PUT_LINE('Total MBytes : '||to_char(v_total_bytes/1024/1024)); DBMS_OUTPUT.PUT_LINE('Unused Blocks: '||v_unused_blocks); DBMS_OUTPUT.PUT_LINE('Unused MBytes: '||to_char(v_unused_bytes/1024/1024)); DBMS_OUTPUT.PUT_LINE('File of Last Extent Containing Data: '||'('||v_last_used_extent_file_id ||') '||v_filename ); DBMS_OUTPUT.PUT_LINE('Starting Block ID of the Last Extent Containing Data: '||v_last_used_block ); DBMS_OUTPUT.PUT_LINE('Last Block Within This Extent Which Contains Data: '||v_last_used_block ); end; / -- space usage of data blocks under the segment High Water Mark with -- auto segment space management Tablespaces -- following code in Oracle 10g set serveroutput on declare -- IN vars v_segment_owner VARCHAR2(100):='HR'; v_segment_name VARCHAR2(100) :='NAMES2'; v_segment_type VARCHAR2(100) :='TABLE'; v_partition_name VARCHAR2(100) :=null; -- OUT vars v_unformatted_blocks NUMBER; v_unformatted_bytes NUMBER; v_fs1_blocks NUMBER; v_fs1_bytes NUMBER; v_fs2_blocks NUMBER; v_fs2_bytes NUMBER; v_fs3_blocks NUMBER; v_fs3_bytes NUMBER; v_fs4_blocks NUMBER; v_fs4_bytes NUMBER; v_full_blocks NUMBER; v_full_bytes NUMBER; v_segment_size_blocks NUMBER; v_segment_size_bytes NUMBER; v_used_blocks NUMBER; v_used_bytes NUMBER; v_expired_blocks NUMBER; v_expired_bytes NUMBER;
Page 122
v_unexpired_blocks NUMBER; v_unexpired_bytes NUMBER; begin DBMS_SPACE.SPACE_USAGE( segment_owner => v_segment_owner , segment_name => v_segment_name , segment_type => v_segment_type , unformatted_blocks => v_unformatted_blocks , unformatted_bytes => v_unformatted_bytes , fs1_blocks => v_fs1_blocks , fs1_bytes => v_fs1_bytes , fs2_blocks => v_fs2_blocks , fs2_bytes => v_fs2_bytes , fs3_blocks => v_fs3_blocks , fs3_bytes => v_fs3_bytes , fs4_blocks => v_fs4_blocks , fs4_bytes => v_fs4_bytes , full_blocks => v_full_blocks , full_bytes => v_full_bytes , partition_name => v_partition_name ); DBMS_OUTPUT.PUT_LINE('Free Space in MB:'); DBMS_OUTPUT.PUT_LINE('Unformatted Mbytes: '||v_unformatted_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE('Free Space in MB:'); DBMS_OUTPUT.PUT_LINE(' 0 to 25% free space: '|| v_fs1_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE(' 25 to 50% free space: '|| v_fs2_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE(' 50 to 75% free space: '|| v_fs3_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE(' 75 to 100% free space: '|| v_fs4_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE(' Full Mbytes : '|| v_full_bytes/1024/1024); DBMS_OUTPUT.PUT_LINE('Free Space in Blocks:'); DBMS_OUTPUT.PUT_LINE('Unformatted Blocks: ' || v_unformatted_blocks); DBMS_OUTPUT.PUT_LINE(' 0 to 25% free space: '|| v_full_blocks); DBMS_OUTPUT.PUT_LINE(' 25 to 50% free space: '|| v_fs2_blocks); DBMS_OUTPUT.PUT_LINE(' 50 to 75% free space: '|| v_fs3_blocks); DBMS_OUTPUT.PUT_LINE(' 75 to 100% free space: '|| v_fs4_blocks); DBMS_OUTPUT.PUT_LINE(' Full blocks : '|| v_full_blocks); exception when others then if sqlcode = '-942' then DBMS_OUTPUT.PUT_LINE('Object Does not Exist.'); else raise; end if; end; /
/* Using Views */ -- segment info select OWNER,SEGMENT_NAME,PARTITION_NAME,SEGMENT_TYPE,S.TABLESPACE_NAME, HEADER_FILE,HEADER_BLOCK,S.BYTES/1024/1024 SEGMENT_SIZE_MB, D.BLOCKS SEGMENT_BLOCKS,EXTENTS,S.INITIAL_EXTENT, S.NEXT_EXTENT NEXT_EXTENT_SIZE, S.MIN_EXTENTS,S.MAX_EXTENTS,S.PCT_INCREASE, FREELISTS,FREELIST_GROUPS,D.RELATIVE_FNO, D.FILE_NAME from DBA_SEGMENTS S, DBA_TABLESPACES T, DBA_DATA_FILES D where S.TABLESPACE_NAME = T.TABLESPACE_NAME AND T.TABLESPACE_NAME =
Page 123
D.TABLESPACE_NAME AND S.RELATIVE_FNO = D.RELATIVE_FNO AND T.TABLESPACE_NAME NOT IN ('SYSAUX','SYSTEM') order BY S.TABLESPACE_NAME; -- extent info SELECT EXTENT_ID, E.BLOCK_ID,E.BYTES/1024 EXTENT_SIZE_KB, E.OWNER,E.SEGMENT_NAME,E.PARTITION_NAME, E.SEGMENT_TYPE,D.FILE_NAME, S.TABLESPACE_NAME,HEADER_FILE, HEADER_BLOCK SEG_HEADER_BLOCK,S.BYTES/1024/1024 SEGMENT_SIZE_MB,D.BLOCKS SEGMENT_BLOCKS, EXTENTS SEG_EXTENTS FROM DBA_EXTENTS E, DBA_SEGMENTS S, DBA_DATA_FILES D WHERE E.OWNER=S.OWNER AND E.SEGMENT_NAME = S.SEGMENT_NAME AND NVL(E.PARTITION_NAME,'0') = NVL(S.PARTITION_NAME,'0') AND E.SEGMENT_TYPE = S.SEGMENT_TYPE AND E.FILE_ID = D.FILE_ID AND S.TABLESPACE_NAME NOT IN ('SYSAUX','SYSTEM') ORDER BY E.SEGMENT_NAME, E.OWNER, E.PARTITION_NAME,E.EXTENT_ID; -- Segment that cannot allocate additional extents SELECT a.SEGMENT_NAME, a.SEGMENT_TYPE, a.TABLESPACE_NAME, a.OWNER FROM DBA_SEGMENTS a WHERE a.NEXT_EXTENT >= (SELECT MAX(b.BYTES) FROM DBA_FREE_SPACE b WHERE b.TABLESPACE_NAME = a.TABLESPACE_NAME) OR a.EXTENTS = a.MAX_EXTENTS OR a.EXTENTS = '8192' ; -- data_block_size -- free extents within tablespaces SELECT F.TABLESPACE_NAME,F.FILE_ID,F.BLOCK_ID, F.BYTES/1024/1024 FREE_MB, D.FILE_NAME FROM DBA_FREE_SPACE F, DBA_DATA_FILES D WHERE F.FILE_ID = D.FILE_ID UNION SELECT F.TABLESPACE_NAME, TO_NUMBER('') AS FILE_ID, TO_NUMBER('') AS BLOCK_ID, SUM(F.BYTES/1024/1024) FREE_MB, TO_CHAR('') AS FILE_NAME FROM DBA_FREE_SPACE F GROUP BY F.TABLESPACE_NAME, TO_NUMBER('') , TO_NUMBER('') ,TO_CHAR('') ORDER BY TABLESPACE_NAME; Segment Advisor The Automatic Segment Advisor is started by a Scheduler job that is configured to run during the default maintenance window(s). Input for DBMS_ADVISOR.CREATE_OBJECT ATTR4 is unused. Specify NULL OBJECT_TYPE ATTR1 ATTR2 ATTR3 TABLESPACE tbs name TABLE schema name table name INDEX schema name index name TABLE PARTITION schema name table name table partition name INDEX PARTITION schema name index name index partition name TABLE SUBPART. schema name table name table subpartition name INDEX SUBPART. schema name index name index subpartition name LOB schema name segment name LOB PARTITION schema name segment name lob partition name LOB SUBPART. schema name segment name lob subpartition name -- object priv EXECUTE ON DBMS_ADVISOR or the system priv ADVISOR
Page 124
-- run Segment Advisor on hr.employees variable id number; declare v_name varchar2(100); v_descr varchar2(500); obj_id number; begin v_name:='Manual_Employees'; v_descr:='Segment Advisor Example'; dbms_advisor.create_task ( advisor_name => 'Segment Advisor', task_id => :id, task_name => v_name, task_desc => v_descr); -- identify the target object dbms_advisor.create_object ( task_name => v_name, object_type => 'TABLE', attr1 => 'HR', attr2 => 'EMPLOYEES', attr3 => NULL, attr4 => NULL, attr5 => NULL, object_id => obj_id); dbms_advisor.set_task_parameter( task_name => v_name, parameter => 'recommend_all', value => 'TRUE'); dbms_advisor.execute_task(v_name); end; / -- viewing segment results -- asa_recommendations ( all_runs in varchar2 DEFAULT 'TRUE', -show_manual in varchar2 DEFAULT 'TRUE', -show_findings in varchar2 DEFAULT 'FALSE' ) select tablespace_name, segment_name, segment_type, partition_name, recommendations, c1 from table(dbms_space.asa_recommendations('TRUE', 'TRUE', 'TRUE')); select * from table(dbms_space.asa_recommendations('TRUE', 'TRUE', 'TRUE')); -- alternatively use DBA_ADVISOR_* SELECT OWNER,REC_ID,TASK_ID, TASK_NAME,FINDING_ID,TYPE, RANK,PARENT_REC_IDS,BENEFIT_TYPE, BENEFIT,ANNOTATION_STATUS,FLAGS FROM DBA_ADVISOR_RECOMMENDATIONS WHERE TASK_ID = :ID SELECT OWNER,TASK_ID,TASK_NAME, FINDING_ID,TYPE,PARENT, OBJECT_ID,IMPACT_TYPE,IMPACT,MESSAGE,MORE_INFO FROM DBA_ADVISOR_FINDINGS WHERE TASK_ID = :ID
Page 125
-- suggested actions to perform SELECT OWNER,TASK_ID,TASK_NAME, REC_ID,ACTION_ID,OBJECT_ID, COMMAND,COMMAND_ID,FLAGS, ATTR1,ATTR2,ATTR3, ATTR4,ATTR5,ATTR6, NUM_ATTR1,NUM_ATTR2,NUM_ATTR3, NUM_ATTR4,NUM_ATTR5,MESSAGE FROM DBA_ADVISOR_ACTIONS WHERE TASK_ID = :ID -- analyzed objects SELECT OWNER,OBJECT_ID,TYPE,TYPE_ID, TASK_ID,TASK_NAME,ATTR1,ATTR2, ATTR3,ATTR4,ATTR5,OTHER FROM DBA_ADVISOR_OBJECTS WHERE TASK_ID = :ID /* information about the Advisor Runs */ SELECT AUTO_TASKID,SNAPID,SEGMENTS_SELECTED,SEGMENTS_PROCESSED, TABLESPACE_SELECTED,TABLESPACE_PROCESSED,RECOMMENDATIONS_COUNT, START_TIME, END_TIME FROM DBA_AUTO_SEGADV_SUMMARY; /* control information that show which object analyzed by the Advisor */ SELECT * FROM DBA_AUTO_SEGADV_CTL Shrinking Database Segments Online -- required ALTER TABLE names ENABLE ROW MOVEMENT -- it acquires table lock
-- shrink table (defrage and then move the HWM) ALTER TABLE names SHRINK SPACE; -- it acquires table lock -- defrage and don't move HWM (later you can SHRINK SPACE) ALTER TABLE names SHRINK SPACE COMPACT; -- no table lock -- LOBs and the Indexes are alos shrinked ALTER TABLE names SHRINK SPACE CASCADE; -- shrink LOB only ALTER TABLE names MODIFY LOB (perf_review) (SHRINK SPACE); -- shrink partition ALTER TABLE names MODIFY PARTITION cust_P1 SHRINK SPACE; -- IOT ALTER TABLE names SHRINK SPACE CASCADE; -- IOT Overflow Segment ALTER TABLE names OVERFLOW SHRINK SPACE; Deallocating Unused Space deallocates unused space at the end of the segment; no data is moved Oracle DBA Code Examples
Page 126
ALTER TABLE mytable DEALLOCATE UNUSED KEEP integer; ALTER INDEX myindex DEALLOCATE UNUSED KEEP integer; ALTER CLUSTER cluster DEALLOCATE UNUSED KEEP integer;
Capacity Planning for Database Objects

Estimating the Space Use of a Table DBMS_SPACE.CREATE_TABLE_COST can be used on dictionary-managed or locally managed tablespaces. -- based on average row size SET SERVEROUTPUT ON declare v_table_name varchar2(30) :='NAMES'; v_owner varchar2(30) :='HR'; v_tablespace_name varchar2(100) := 'USERS'; v_avg_row_size number ; -- anticipated average row size v_row_count number := 1000000; v_pct_free number := 10; v_used_bytes number; -- User data space including PCTFREE v_alloc_bytes number; -- The anticipated size of the table taking into account the tablespace extent characteristics x char(1); begin -- check the tablespace exists begin SELECT 'x' into x from dba_tablespaces where tablespace_name=v_tablespace_name ; exception when no_data_found then raise_application_error(-20001,'Tablespace: ' || v_tablespace_name || ' does not exist'); end; -- retreive average row size begin SELECT AVG_ROW_LEN into V_avg_row_size FROM DBA_TABLES where table_name=v_table_name and owner=v_owner; exception when no_data_found then raise_application_error(-20002,'Table : ' || V_OWNER ||'.'||v_table_name || ' does not exist'); end; DBMS_SPACE.CREATE_TABLE_COST ( tablespace_name => v_tablespace_name, avg_row_size => v_avg_row_size, row_count => v_row_count, pct_free => v_pct_free, used_bytes => v_used_bytes, alloc_bytes => v_alloc_bytes); dbms_output.put_line('User Data Space (MB): '|| TO_CHAR(v_used_bytes/1024/1024 ,'999,999,999,999,999.99')); dbms_output.put_line('Data Space in the Tablespace (MB): '||
Page 127
TO_CHAR(v_alloc_bytes/1024/1024,'999,999,999,999,999.99')); end; / -- based on column info declare ub NUMBER; ab NUMBER; cl sys.create_table_cost_columns; begin cl := sys.create_table_cost_columns( sys.create_table_cost_colinfo('NUMBER',10), sys.create_table_cost_colinfo('VARCHAR2',30), sys.create_table_cost_colinfo('VARCHAR2',30), sys.create_table_cost_colinfo('DATE',NULL)); DBMS_SPACE.CREATE_TABLE_COST('SYSTEM',cl,100000,0,ub,ab); DBMS_OUTPUT.PUT_LINE('Used Bytes: ' || TO_CHAR(ub)); DBMS_OUTPUT.PUT_LINE('Alloc Bytes: ' || TO_CHAR(ab)); end; / Obtaining Object Growth Trends User DBMS_SPACE.OBJECT_GROWTH_TREND o o o o o start_time Statistics generated after this time will be used in generating the growth trend end_time Statistics generated until this time will be used in generating the growth trend interval The interval at which to sample skip_interpolated Whether interpolation of missing values should be skipped single_data_point_flag Whether in the absence of statistics the segment should be sampled timepoint The time at which the statistic was recorded space_usage space_alloc quality The space used by data The size of the segment including overhead and unused space
Returned columns: o o o o
The quality of result: "GOOD", "INTERPOLATED", "PROJECTION" GOOD: accurate figure PROJECTED: figures projected from the data collected by the AWR INTERPOLATED: no data was available for calculations
COL TIMEPOINT FORMAT A30 SELECT * FROM table( DBMS_SPACE.OBJECT_GROWTH_TREND ( object_owner =>'HR', object_name =>'NAMES', object_type =>'TABLE', partition_name =>NULL, start_time =>NULL, end_time =>NULL, interval =>to_dsinterval('0 00:10:00') skip_interpolated => 'FALSE',
Page 128
, -- Day to Second interval
timeout_seconds =>NULL, single_datapoint_flag =>'TRUE') )
Using the SQL Access Advisor

Estimating the Space Use of a Table The SQL Access Advisor sources: o o o A hypothetical database workload An actual workload you provide SQL cache
-- Method 1: to invoke SQL Access Advisor task linked to a workload declare l_taskname VARCHAR2(30) := 'sql_access_test_task'; l_task_desc VARCHAR2(128) := 'Test SQL Access'; l_wkld_name VARCHAR2(30) := 'test_work_load'; l_saved_rows NUMBER := 0; l_failed_rows NUMBER := 0; l_num_found NUMBER; BEGIN -- create an SQL Access Advisor task. select COUNT(*) into l_num_found from DBA_ADVISOR_TASKS where TASK_NAME = l_taskname ; IF l_num_found = 0 THEN DBMS_ADVISOR.CREATE_TASK ( ADVISOR_NAME => DBMS_ADVISOR.SQLACCESS_ADVISOR, TASK_NAME => l_taskname, TASK_DESC => l_task_desc); END IF; -- reset the task ( remove all recommendations, and intermediate data from the task) DBMS_ADVISOR.RESET_TASK(TASK_NAME => l_taskname); -- create a workload. SELECT COUNT(*) INTO l_num_found FROM USER_ADVISOR_SQLW_SUM WHERE WORKLOAD_NAME = l_wkld_name; IF l_num_found = 0 THEN DBMS_ADVISOR.CREATE_SQLWKLD(WORKLOAD_NAME => l_wkld_name); END IF; -- link the workload to the task SELECT count(*) INTO l_num_found FROM USER_ADVISOR_SQLA_WK_MAP WHERE TASK_NAME = l_taskname AND WORKLOAD_NAME = l_wkld_name; IF l_num_found = 0 THEN
Page 129
DBMS_ADVISOR.ADD_SQLWKLD_REF( TASK_NAME => l_taskname, WORKLOAD_NAME => l_wkld_name); END IF; -- Set workload parameters. DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, DBMS_ADVISOR.ADVISOR_UNLIMITED); DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, 'PRIORITY,OPTIMIZER_COST'); DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_SQLWKLD_PARAMETER(l_wkld_name, DBMS_ADVISOR.ADVISOR_UNUSED);
'ACTION_LIST', 'MODULE_LIST', 'SQL_LIMIT', 'ORDER_LIST', 'USERNAME_LIST', 'VALID_TABLE_LIST',
-- unmark the required option /* -- import the current contents of the server's SQL cache DBMS_ADVISOR.IMPORT_SQLWKLD_SQLCACHE(l_wkld_name, 'REPLACE', 2, l_saved_rows, l_failed_rows); -- load a SQL workload from an existing SQL Tuning Set DBMS_ADVISOR.IMPORT_SQLWKLD_STS (l_wkld_name, 'SOURCE_STS_Name', 'REPLACE',2, l_saved_rows, l_failed_rows); */ -- Set task parameters. DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, '_MARK_IMPLEMENTATION', 'FALSE'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'EXECUTION_TYPE', 'INDEX_ONLY'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'MODE', 'COMPREHENSIVE'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'STORAGE_CHANGE', DBMS_ADVISOR.ADVISOR_UNLIMITED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'DML_VOLATILITY', 'TRUE'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'ORDER_LIST', 'PRIORITY,OPTIMIZER_COST'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'WORKLOAD_SCOPE', 'PARTIAL'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'DEF_INDEX_TABLESPACE', DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'DEF_INDEX_OWNER', DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'DEF_MVIEW_TABLESPACE', DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'DEF_MVIEW_OWNER', DBMS_ADVISOR.ADVISOR_UNUSED); -- Execute the task: control will not return till the execution finishes DBMS_ADVISOR.execute_task(task_name => l_taskname); END; /
-- Method 2: to invoke SQL Access Advisor linked to a specific STS declare l_taskname VARCHAR2(30) := 'sql_access_test_task2'; l_task_desc VARCHAR2(128) := 'Test SQL Access for a STS';
Page 130
l_wkld_name VARCHAR2(30) := 'test_work_load'; l_sts_name VARCHAR2(30) := 'test_sts'; l_saved_rows NUMBER := 0; l_failed_rows NUMBER := 0; l_num_found NUMBER; BEGIN -- create an SQL Access Advisor task, if it doesnt exist select COUNT(*) into l_num_found from DBA_ADVISOR_TASKS where TASK_NAME = l_taskname ; IF l_num_found = 0 THEN DBMS_ADVISOR.CREATE_TASK ( ADVISOR_NAME => DBMS_ADVISOR.SQLACCESS_ADVISOR, TASK_NAME => l_taskname, TASK_DESC => l_task_desc); END IF; -- reset the task ( remove all recommendations, and intermediate data from the task) DBMS_ADVISOR.RESET_TASK(TASK_NAME => l_taskname); -- check if STS already exists select count(*) into l_num_found from DBA_SQLSET where upper(NAME) = upper(l_sts_name) ; IF l_num_found <> 0 THEN DBMS_SQLTUNE.DROP_SQLSET ( sqlset_name => l_sts_name); END IF; -- create STS DBMS_SQLTUNE.CREATE_SQLSET(SQLSET_NAME => l_sts_name, DESCRIPTION =>'To test Access Advisor'); /* unmark the required option -- (Option 1) Load l_sts_name from an AWR baseline. -- The data has been filtered to select only the top 30 SQL statements ordered by elapsed time. declare baseline_cur DBMS_SQLTUNE.SQLSET_CURSOR; begin -- a ref cursor is opened to select from the specified baseline OPEN baseline_cur FOR SELECT VALUE(p) FROM TABLE (DBMS_SQLTUNE.SELECT_WORKLOAD_REPOSITORY( 'peak baseline',NULL, NULL, 'elapsed_time', NULL, NULL, NULL,30 )) p; -- Next the statements and their statistics are loaded into the STS DBMS_SQLTUNE.LOAD_SQLSET( SQLSET_NAME=>l_sts_name, POPULATE_CURSOR=>baseline_cur); end; -- (Option 2) Load l_sts_name with SQL statements that are not owned by SYS and -- their elapsed time is greater than 20,000 seconds. declare sql_cur DBMS_SQLTUNE.SQLSET_CURSOR;
Page 131
begin -- a ref cursor is opened to select the required SQL statments OPEN sql_cur FOR SELECT VALUE(p) FROM TABLE (DBMS_SQLTUNE.SELECT_CURSOR_CACHE('parsing_schema_name <> ''SYS'' and elapsed_time > 2000000',NULL, NULL, NULL, NULL,1, NULL, 'ALL')) p; -- the statements are loaded into the STS DBMS_SQLTUNE.LOAD_SQLSET( SQLSET_NAME=>l_sts_name, POPULATE_CURSOR=>sql_cur); end; -- (Option 3) Copy the contents of a SQL workload object to a SQL Tuning Set -- check the example above for creating a workload DBMS_ADVISOR.COPY_SQLWKLD_TO_STS ('My_WorkLoad', l_sts_name, 'REPLACE'); */ -- link the STS to the task DBMS_ADVISOR.ADD_STS_REF (l_taskname, null, l_sts_name); -- Set task parameters. DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'FALSE'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'INDEX_ONLY'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.ADVISOR_UNLIMITED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, 'PRIORITY,OPTIMIZER_COST'); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.ADVISOR_UNUSED); DBMS_ADVISOR.SET_TASK_PARAMETER(l_taskname, DBMS_ADVISOR.ADVISOR_UNUSED);
'_MARK_IMPLEMENTATION', 'EXECUTION_TYPE', 'MODE', 'COMPREHENSIVE'); 'STORAGE_CHANGE', 'DML_VOLATILITY', 'TRUE'); 'ORDER_LIST', 'WORKLOAD_SCOPE', 'PARTIAL'); 'DEF_INDEX_TABLESPACE', 'DEF_INDEX_OWNER', 'DEF_MVIEW_TABLESPACE', 'DEF_MVIEW_OWNER',
-- Execute the task: control will not return till the execution finishes DBMS_ADVISOR.EXECUTE_TASK(task_name => l_taskname); END; /
-- Method 3: Quick Tune for a single given statement begin -- a task and a workload will be created then the task will be executed DBMS_ADVISOR.QUICK_TUNE( ADVISOR_NAME => DBMS_ADVISOR.SQLACCESS_ADVISOR, TASK_NAME => 'names_quick_tune', ATTR1 => 'SELECT id FROM hr.names n WHERE id = 100'); end; / Following are examples of how to manage the task and obtain information about the advisors output repot:
Page 132
-- while the task is being executed, you can monitor it using the following query: select TASK_NAME, STATUS, PCT_COMPLETION_TIME, ERROR_MESSAGE from DBA_ADVISOR_LOG where TASK_NAME ='sql_access_test_task'; -- if you need to terminate the executing task (may be time consuming) exec DBMS_ADVISOR.CANCEL_TASK(TASK_NAME =>'sql_access_test_task'); -- Display the resulting script ( method 1) SET LONG 100000 SET PAGESIZE 50000 SELECT DBMS_ADVISOR.GET_TASK_SCRIPT('sql_access_test_task') AS script FROM dual; SET PAGESIZE 24 -- Display the resulting script ( method 2) CREATE DIRECTORY ADVISOR_RESULTS AS 'C:\TEMP\'; exec DBMS_ADVISOR.CREATE_FILE(DBMS_ADVISOR.GET_TASK_SCRIPT('sql_access_test_task'), 'ADVISOR_RESULTS','advscript.sql'); -- alternatively, use the following queries -- benefit is the total improvement in execution cost of all the queries select REC_ID, RANK, BENEFIT, TYPE "Recommendation Type" from DBA_ADVISOR_RECOMMENDATIONS where TASK_NAME = 'sql_access_test_task' order by RANK; -- display processed statements in the workload select SQL_ID, REC_ID, PRECOST, POSTCOST, (PRECOST-POSTCOST)*100/PRECOST AS PERCENT_BENEFIT from USER_ADVISOR_SQLA_WK_STMTS where TASK_NAME = 'sql_access_test_task' AND workload_name = 'test_work_load'; -- see the actions for each recommendations select REC_ID, ACTION_ID, SUBSTR(COMMAND,1,30) AS COMMAND from USER_ADVISOR_ACTIONS where TASK_NAME = 'sql_access_test_task' ORDER BY rec_id, action_id; -- to delete a given task exec DBMS_ADVISOR.DELETE_TASK('sql_access_test_task');
Page 133
Managing Tables
Obtaining Information about Tables
-- 11g SELECT OWNER,TABLE_NAME,TABLESPACE_NAME,CLUSTER_NAME,IOT_NAME,STATUS,PCT_FREE, PCT_USED,INI_TRANS,MAX_TRANS,INITIAL_EXTENT, NEXT_EXTENT,MIN_EXTENTS,MAX_EXTENTS,PCT_INCREASE,FREELISTS,FREELIST_GROUPS,LOGGING, BACKED_UP,NUM_ROWS,BLOCKS,EMPTY_BLOCKS,AVG_SPACE,CHAIN_CNT, AVG_ROW_LEN,AVG_SPACE_FREELIST_BLOCKS,NUM_FREELIST_BLOCKS,DEGREE, INSTANCES,CACHE,TABLE_LOCK,SAMPLE_SIZE,LAST_ANALYZED,PARTITIONED,IOT_TYPE, TEMPORARY,SECONDARY,NESTED,BUFFER_POOL,ROW_MOVEMENT,GLOBAL_STATS,USER_STATS, DURATION,SKIP_CORRUPT,MONITORING,CLUSTER_OWNER,DEPENDENCIES,COMPRESSION, COMPRESS_FOR,DROPPED,READ_ONLY FROM DBA_TABLES WHERE TABLE_NAME='EMP' AND OWNER='HR'; -- columns SELECT OWNER,TABLE_NAME,COLUMN_NAME,DATA_TYPE,DATA_TYPE_MOD,DATA_TYPE_OWNER,DATA_LENGTH,DATA_P RECISION,DATA_SCALE,NULLABLE,COLUMN_ID,DEFAULT_LENGTH,DATA_DEFAULT,NUM_DISTINCT,LOW_VAL UE,HIGH_VALUE,DENSITY,NUM_NULLS,NUM_BUCKETS,LAST_ANALYZED,SAMPLE_SIZE,CHARACTER_SET_NAM E,CHAR_COL_DECL_LENGTH,GLOBAL_STATS,USER_STATS,AVG_COL_LEN,CHAR_LENGTH,CHAR_USED,V80_FM T_IMAGE,DATA_UPGRADED,HISTOGRAM FROM DBA_TAB_COLUMNS WHERE TABLE_NAME='EMP' AND OWNER='HR'; -- table gathered statistics SELECT OWNER,TABLE_NAME,PARTITION_NAME,PARTITION_POSITION,SUBPARTITION_NAME,SUBPARTITION_POSIT ION,OBJECT_TYPE,NUM_ROWS,BLOCKS,EMPTY_BLOCKS,AVG_SPACE,CHAIN_CNT,AVG_ROW_LEN,AVG_SPACE_ FREELIST_BLOCKS,NUM_FREELIST_BLOCKS,AVG_CACHED_BLOCKS,AVG_CACHE_HIT_RATIO,SAMPLE_SIZE,L AST_ANALYZED,GLOBAL_STATS,USER_STATS,STATTYPE_LOCKED,STALE_STATS FROM DBA_TAB_STATISTICS WHERE TABLE_NAME='EMP' AND OWNER='HR'; -- comments select * from DBA_TAB_COMMENTS; select * from DBA_COL_COMMENTS; -- object and relational tables SELECT OWNER,TABLE_NAME,TABLESPACE_NAME,CLUSTER_NAME,IOT_NAME,STATUS,PCT_FREE,PCT_USED,INI_TRA NS,MAX_TRANS,INITIAL_EXTENT,NEXT_EXTENT,MIN_EXTENTS,MAX_EXTENTS,PCT_INCREASE,FREELISTS, FREELIST_GROUPS,LOGGING,BACKED_UP,NUM_ROWS,BLOCKS,EMPTY_BLOCKS,AVG_SPACE,CHAIN_CNT,AVG_ ROW_LEN,AVG_SPACE_FREELIST_BLOCKS,NUM_FREELIST_BLOCKS,DEGREE,INSTANCES,CACHE,TABLE_LOCK ,SAMPLE_SIZE,LAST_ANALYZED,PARTITIONED,IOT_TYPE,OBJECT_ID_TYPE,TABLE_TYPE_OWNER,TABLE_T YPE,TEMPORARY,SECONDARY,NESTED,BUFFER_POOL,ROW_MOVEMENT,GLOBAL_STATS,USER_STATS,DURATIO N,SKIP_CORRUPT,MONITORING,CLUSTER_OWNER,DEPENDENCIES,COMPRESSION,COMPRESS_FOR,DROPPED FROM DBA_ALL_TABLES WHERE TABLE_NAME='EMP' AND OWNER='HR';
Creating Tables with some Options

-- specific tablspace create table mytab ... tablspace mytbs -- Parallelizing Table Creation CREATE TABLE mytb PARALLEL AS SELECT .. ALTER SESSION FORCE PARALLEL DDL; -- created tables will be parallelized -- nologging Page 134
CREATE TABLE mytb NOLOGGING AS SELECT .. -- compression CREATE TABLE mytb ( ... ) COMPRESS FOR ALL OPERATIONS; -- OK in OLTP CREATE TABLE sales_history ( ... ) COMPRESS FOR DIRECT_LOAD OPERATIONS; -- OK in OLAP CREATE TABLE sales_history ( ... ) COMPRESS; -- same as above CREATE TABLE sales (saleskey number, quarter .. region varchar2(10)) COMPRESS PARTITION BY LIST (region) (PARTITION northwest VALUES ('NORTHWEST'), PARTITION southwest VALUES ('SOUTHWEST'), PARTITION northeast VALUES ('NORTHEAST') NOCOMPRESS, PARTITION southeast VALUES ('SOUTHEAST'), PARTITION western VALUES ('WESTERN')) -- encrypting columns (REQUIRES OEPENED WALLET) CREATE TABLE mytb ( .. ssn NUMBER(9) ENCRYPT, ..) -- temporary table -- transaction level CREATE GLOBAL TEMPORARY TABLE mytb (..) ON COMMIT DELETE ROWS; -- session level CREATE GLOBAL TEMPORARY TABLE mytb (..) ON COMMIT PRESERVE ROWS;
DML Error Logging

DML error logging works with INSERT, UPDATE, MERGE, and DELETE statements. The following conditions cause the statement to fail and roll back without invoking the error logging capability: o o o o Violated deferred constraints Out-of-space errors Any direct-path INSERT operation (INSERT or MERGE) that raises a unique constraint or index violation Any UPDATE or MERGE that raises a unique constraint or index violation
-- 1. Create an error logging table. -- LONG, CLOB, BLOB, BFILE, and ADT datatypes are not supported in the columns EXECUTE DBMS_ERRLOG.CREATE_ERROR_LOG(dml_table_name=>'EMP', err_log_table_name=>'ERR_EMP'); EXECUTE DBMS_ERRLOG.CREATE_ERROR_LOG(dml_table_name=>'EMP', err_log_table_name=>'ERR_EMP', err_log_table_owner=>'HR', err_log_table_space=>'USERS'); -- 2 include an error logging clause INSERT INTO emp VALUES(1001, 'abcd', 75, NULL) LOG ERRORS INTO ERR_EMP REJECT LIMIT UNLIMITED; INSERT INTO hr.emp SELECT .. FROM .. LOG ERRORS INTO err_emp ('daily_load') REJECT LIMIT 25 -- 3 Query the error logging table and take corrective action
Enabling Direct-Path INSERT

To activate direct-path INSERT in serial mode, you must specify the APPEND hint in each INSERT statement, either immediately after the INSERT keyword, or immediately after the SELECT keyword in the subquery of the INSERT statement. Oracle DBA Code Examples
Page 135
insert /*+ append */ into customer (select xxx); When you are inserting in parallel DML mode, direct-path INSERT is the default. In order to run in parallel DML mode, the following requirements must be met: o o You must have Oracle Enterprise Edition installed. You must enable parallel DML in your session. To do this, run the following statement: ALTER SESSION { ENABLE | FORCE } PARALLEL DML; o You must specify the parallel attribute for the target table, either at create time or subsequently, or you must specify the PARALLEL hint for each insert operation. insert /*+ parallel */ into names3 (select * from names) If nologging is specified, the insert will finish faster. select logging from dba_tables where owner='HR' and table_name='EMPLOYEES'; select logging from dba_tablespaces where tablespace_name = 'USERS'; Index maintenance is done by end of command. Consider drop and then re-creating the indexes after the statement. Exclusive table lock is acquired.
Automatically Collecting Statistics on Tables

In Oracle Database 11g, statistics are collected automatically for tables regardless of the the MONITORING and NOMONITORING keywords when creating them.
Altering Tables
If you enable compression for all operations on a table, you can drop table columns. If you enable compression for direct-path inserts only, you cannot drop columns. -- move the table to a new segment -- indexes will be UNUSABLE and must be rebuilt -- statistics must be regathered ALTER TABLE hr.emp MOVE STORAGE ( INITIAL 20K NEXT 40K MINEXTENTS 2 MAXEXTENTS 20 PCTINCREASE 0 ); alter table emp MOVE TABLESPACE example ; alter table emp move storage ( NEXT 1024K PCTINCREASE 0 ) COMPRESS; -- allocating and deallocating extents alter table hr.emp allocate extent; alter table hr.emp deallocate extent; -- allocating and deallocating extents alter table hr.emp allocate extent; alter table hr.emp deallocate unused;
Page 136
-- Marking Columns Unused ALTER TABLE hr.emp SET UNUSED (hiredate, mgr); SELECT * FROM DBA_UNUSED_COL_TABS; ALTER TABLE hr.emp DROP UNUSED COLUMNS statement; -- less undo space is consumed when checkpoint is done periodically ALTER TABLE hr.admin_emp DROP UNUSED COLUMNS CHECKPOINT 250; -- read only and read write ALTER TABLE employees READ ONLY; ALTER TABLE employees READ WRITE; select TABLE_NAME, READ_ONLY from user_tables where table_name='EMPLOYEES';
Performing Online Redefinition with DBMS_REDEFINITION

Redefining a Table /* This example illustrates online redefinition of the previously created table hr.admin_emp, which at this point only contains columns: empno, ename, job,deptno. The table is redefined as follows: -New columns mgr, hiredate, sal, and bonus are added. -The new column bonus is initialized to 0 -The column deptno has its value increased by 10. -The redefined table is partitioned by range on empno. */ --1 Verify that the table is a candidate for online redefinition. BEGIN DBMS_REDEFINITION.CAN_REDEF_TABLE( UNAME => 'HR', TNAME => 'ADMIN_EMP', OPTIONS_FLAG => dbms_redefinition.cons_use_pk); -- OR DBMS_REDEFINITION.CONS_USE_ROWID END; / -- 2. Create an interim table hr.int_admin_emp. CREATE TABLE hr.int_admin_emp (empno NUMBER(5) PRIMARY KEY, ename VARCHAR2(15) NOT NULL, job VARCHAR2(10), mgr NUMBER(5), hiredate DATE DEFAULT (sysdate), sal NUMBER(7,2), deptno NUMBER(3) NOT NULL, bonus NUMBER (7,2) DEFAULT(1000)) PARTITION BY RANGE(empno) (PARTITION emp1000 VALUES LESS THAN (1000) TABLESPACE admin_tbs, PARTITION emp2000 VALUES LESS THAN (2000) TABLESPACE admin_tbs2); -- (Optional) If you are redefining a large table alter session force parallel dml parallel 2; alter session force parallel query parallel 2; --3. Start the redefinition process. BEGIN
Page 137
DBMS_REDEFINITION.START_REDEF_TABLE( UNAME => 'HR', ORIG_TABLE => 'ADMIN_EMP', INT_TABLE => 'int_admin_emp', col_mapping=>'empno empno, ename ename, job job, deptno+10 deptno, 0 bonus', -- NULL if same columns in both tables options_flag => dbms_redefinition.cons_use_pk); END; / -- 4 Copy dependent objects. DECLARE N NUMBER; BEGIN DBMS_REDEFINITION.COPY_TABLE_DEPENDENTS( uname =>'HR', orig_table =>'admin_emp', int_table => 'int_admin_emp', copy_indexes =>dbms_redefinition.cons_orig_params, -- 0 don't copy copy_triggers =>TRUE, copy_constraints =>TRUE, copy_privileges =>TRUE, ignore_errors => TRUE, -- default FALSE num_errors => n, copy_statistics => FALSE, -- default FALSE copy_mvlog => FALSE); -- default FALSE -- bcuz ignore_errors=true if n =0 then dbms_output.put_line('Success.'); else dbms_output.put_line('Number of Errors: ' || n); end if; END; / /* Note that the ignore_errors argument is set to TRUE for this call. The reason is that the interim table was created with a primary key constraint, and when COPY_TABLE_DEPENDENTS attempts to copy the primary key constraint and index from the original table, errors occurs. You can ignore these errors, but you must run the query shown in the next step to see if there are other errors. */ -- 5. Query the DBA_REDEFINITION_ERRORS view to check for errors. select object_name, base_table_name, ddl_txt from DBA_REDEFINITION_ERRORS; -- 6. Optionally, synchronize the interim table hr.int_admin_emp BEGIN DBMS_REDEFINITION.SYNC_INTERIM_TABLE('hr', 'admin_emp', 'int_admin_emp'); END; / --7. Complete the redefinition. BEGIN DBMS_REDEFINITION.FINISH_REDEF_TABLE('hr', 'admin_emp', 'int_admin_emp'); END; / /* The table hr.admin_emp is locked in the exclusive mode only for a small
Page 138
window toward the end of this step. After this call the table hr.admin_emp is redefined such that it has all the attributes of the hr.int_admin_emp table. */ -- 8 Wait for any long-running queries against the interim table to complete, and then drop the interim table. Redefining a Single Partition / * We want to move the oldest partition of a range-partitioned sales table to a tablespace named TBS_LOW_FREQ. The table containing the partition to be redefined is defined as follows: CREATE TABLE salestable (s_productid NUMBER, s_saledate DATE, s_custid NUMBER, s_totalprice NUMBER) TABLESPACE users PARTITION BY RANGE(s_saledate) (PARTITION sal03q1 VALUES LESS THAN (TO_DATE('01-APR-2003', 'DD-MON-YYYY')), PARTITION sal03q2 VALUES LESS THAN (TO_DATE('01-JUL-2003', 'DD-MON-YYYY')), PARTITION sal03q3 VALUES LESS THAN (TO_DATE('01-OCT-2003', 'DD-MON-YYYY')), PARTITION sal03q4 VALUES LESS THAN (TO_DATE('01-JAN-2004', 'DD-MON-YYYY'))); The table has a local partitioned index that is defined as follows: CREATE INDEX sales_index ON salestable (s_saledate, s_productid, s_custid) LOCAL; */ -- 1. Ensure that salestable is a candidate for redefinition. BEGIN DBMS_REDEFINITION.CAN_REDEF_TABLE( uname => 'STEVE', tname => 'SALESTABLE', options_flag => DBMS_REDEFINITION.CONS_USE_ROWID, part_name => 'sal03q1'); END; / -- 2 Create the interim table in the TBS_LOW_FREQ tablespace. Because this is a redefinition of a range partition, the interim table is non-partitioned. CREATE TABLE int_salestable (s_productid NUMBER, s_saledate DATE, s_custid NUMBER, s_totalprice NUMBER) TABLESPACE tbs_low_freq; -- 3. Start the redefinition process using rowid. BEGIN DBMS_REDEFINITION.START_REDEF_TABLE( uname => 'STEVE', orig_table => 'salestable', int_table => 'int_salestable', col_mapping => NULL, options_flag => DBMS_REDEFINITION.CONS_USE_ROWID, part_name => 'sal03q1'); END; /
Page 139
-- 4. Manually create any local indexes on the interim table. CREATE INDEX int_sales_index ON int_salestable (s_saledate, s_productid, s_custid) TABLESPACE tbs_low_freq; -- 5. Optionally synchronize the interim table. BEGIN DBMS_REDEFINITION.SYNC_INTERIM_TABLE( uname => 'STEVE', orig_table => 'salestable', int_table => 'int_salestable', part_name => 'sal03q1'); END; / -- 6. Complete the redefinition. BEGIN DBMS_REDEFINITION.FINISH_REDEF_TABLE( uname => 'STEVE', orig_table => 'salestable', int_table => 'int_salestable', part_name => 'sal03q1'); END; / -- 7. Wait for any long-running queries against the interim table to complete, and then drop the interim table. -- The following query shows that the oldest partition has been moved to the new tablespace: select partition_name, tablespace_name from user_tab_partitions where table_name = 'SALESTABLE'; Migrating BasicFile LOBs to SecureFiles -- Grant privileges required for online redefinition. GRANT EXECUTE ON DBMS_REDEFINITION TO pm; GRANT ALTER ANY TABLE TO pm; GRANT DROP ANY TABLE TO pm; GRANT LOCK ANY TABLE TO pm; GRANT CREATE ANY TABLE TO pm; GRANT SELECT ANY TABLE TO pm; -- Privileges required to perform cloning of dependent objects. GRANT CREATE ANY TRIGGER TO pm; GRANT CREATE ANY INDEX TO pm; CONNECT pm CREATE TABLE cust(c_id NUMBER PRIMARY KEY, c_zip NUMBER, c_name VARCHAR(30) DEFAULT NULL, c_lob CLOB); INSERT INTO cust VALUES(1, 94065, 'hhh', 'ttt'); -- Creating Interim Table -- no need to specify constraints because they are copied over from the original table CREATE TABLE cust_int(c_id NUMBER NOT NULL, c_zip NUMBER, c_name VARCHAR(30) DEFAULT NULL, c_lob CLOB)
Page 140
LOB(c) STORE AS SECUREFILE (...); declare col_mapping VARCHAR2(1000); begin -- map all the columns in the interim table to the original table col_mapping :='c_id c_id , '||'c_zip c_zip , '||'c_name c_name, '||'c_lob c_lob'; DBMS_REDEFINITION.START_REDEF_TABLE('pm', 'cust', 'cust_int', col_mapping); end; declare error_count pls_integer := 0; begin DBMS_REDEFINITION.COPY_TABLE_DEPENDENTS('pm', 'cust', 'cust_int', 1, TRUE,TRUE,TRUE,FALSE, error_count); DBMS_OUTPUT.PUT_LINE('errors := ' || TO_CHAR(error_count)); end; exec DBMS_REDEFINITION.FINISH_REDEF_TABLE('pm', 'cust', 'cust_int'); -- Drop the interim table DROP TABLE cust_int; DESC cust; -- to prove that the primary key on the c_id column is preserved after migration. INSERT INTO cust VALUES(1, 94065, 'hhh', 'ttt'); SELECT * FROM cust;
Using Flashback Drop and Managing the Recycle Bin See Flashback Drop.
Managing Index-Organized Tables

-- Creating an Index-Organized Table -- if the length of a row exceeds 20% of the index -- block size, then the column that exceeded that threshold and -- all columns after it are moved to the overflow segment. -- Specify MAPPING TABLE to create a mapping table; it's needed to create a bitmap index on the IOT CREATE TABLE admin_docindex( token char(20), doc_id NUMBER, token_frequency NUMBER, token_offsets VARCHAR2(2000), CONSTRAINT pk_admin_docindex PRIMARY KEY (token, doc_id)) ORGANIZATION INDEX TABLESPACE admin_tbs PCTTHRESHOLD 20 OVERFLOW TABLESPACE admin_tbs2; -- Creating Index-Organized Tables that Contain Object Types CREATE OR REPLACE TYPE admin_typ AS OBJECT (col1 NUMBER, col2 VARCHAR2(6)); CREATE TABLE admin_iot (c1 NUMBER primary key, c2 admin_typ) ORGANIZATION INDEX;
Page 141
CREATE TABLE admin_iot2 OF admin_typ (col1 PRIMARY KEY) ORGANIZATION INDEX; -- oken_offsets column value is always stored in the overflow area TABLE admin_docindex2( token CHAR(20), doc_id NUMBER, token_frequency NUMBER, token_offsets VARCHAR2(2000), CONSTRAINT pk_admin_docindex2 PRIMARY KEY (token, doc_id)) ORGANIZATION INDEX TABLESPACE admin_tbs PCTTHRESHOLD 20 -- default 50 INCLUDING token_frequency OVERFLOW TABLESPACE admin_tbs2; -- Parallelizing Index-Organized Table Creation CREATE TABLE admin_iot3(i PRIMARY KEY, j, k, l) ORGANIZATION INDEX PARALLEL AS SELECT * FROM hr.jobs; -- Altering Index-Organized Tables ALTER TABLE admin_docindex INITRANS 4 OVERFLOW INITRANS 6; ALTER TABLE admin_docindex PCTTHRESHOLD 15 INCLUDING doc_id; -- if there's no existing overflow segment ALTER TABLE admin_iot3 ADD OVERFLOW TABLESPACE admin_tbs2; -- Moving (Rebuilding) Index-Organized Tables ALTER TABLE admin_docindex MOVE; -- just rebuild ALTER TABLE admin_docindex MOVE ONLINE; ALTER TABLE admin_iot_lob MOVE LOB (admin_lob) STORE AS (TABLESPACE admin_tbs3); -- IOT is rebuilt along with its overflow segment ALTER TABLE admin_docindex MOVE TABLESPACE admin_tbs2 OVERFLOW TABLESPACE admin_tbs3; -- Creating Secondary Indexes on Index-Organized Tables CREATE INDEX Doc_id_index on Docindex(Doc_id, Token);
Managing External Tables

/* obtain info about external tables */ select OWNER,TABLE_NAME,TYPE_OWNER, TYPE_NAME,DEFAULT_DIRECTORY_OWNER,DEFAULT_DIRECTORY_NAME, REJECT_LIMIT,ACCESS_TYPE,ACCESS_PARAMETERS, PROPERTY from DBA_EXTERNAL_TABLES; select OWNER, TABLE_NAME, LOCATION, DIRECTORY_OWNER, DIRECTORY_NAME from DBA_EXTERNAL_LOCATIONS;
/* Creating an External Table and Loading Data */ CONNECT / AS SYSDBA; -- Set up directories and grant access to hr CREATE OR REPLACE DIRECTORY admin_dat_dir AS 'c:\temp\data'; CREATE OR REPLACE DIRECTORY admin_log_dir AS 'c:\temp\log'; CREATE OR REPLACE DIRECTORY admin_bad_dir AS 'c:\temp\bad'; GRANT READ ON DIRECTORY admin_dat_dir TO hr; GRANT WRITE ON DIRECTORY admin_log_dir TO hr;
Page 142
GRANT WRITE ON DIRECTORY admin_bad_dir TO hr; -- hr connects. Provide the user password (hr) when prompted. CONNECT hr -- create the external table CREATE TABLE admin_ext_employees (employee_id NUMBER(4), first_name VARCHAR2(20), last_name VARCHAR2(25), job_id VARCHAR2(10), manager_id NUMBER(4), hire_date DATE, salary NUMBER(8,2), commission_pct NUMBER(2,2), department_id NUMBER(4), email VARCHAR2(25)) ORGANIZATION EXTERNAL ( TYPE ORACLE_LOADER -- or ORACLE_DATAPUMP (for unload) DEFAULT DIRECTORY admin_dat_dir ACCESS PARAMETERS ( records delimited by newline badfile admin_bad_dir:'empxt%a_%p.bad' logfile admin_log_dir:'empxt%a_%p.log' fields terminated by ',' missing field values are null ( employee_id, first_name, last_name, job_id, manager_id, hire_date char date_format date mask "dd-mon-yyyy", salary, commission_pct, department_id, email ) ) LOCATION ('empxt1.dat', 'empxt2.dat') ) PARALLEL -- useful to huge data REJECT LIMIT UNLIMITED; -- more options -- in ACCESS PARAMETERS you can add LOAD WHEN (job != MANAGER) -- using SQL*Loader to generated script for creating external tables -- (1) prepare control file such as: LOAD DATA INFILE * INTO TABLE test_emp FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"' (employee_id,first_name,last_name,hire_date,salary,manager_id) BEGINDATA 12345,"sam","alapati",sysdate,50000,99999 23456,"mark","potts",sysdate,50000,99999 -- (2) issue SQL*Loader using EXTERNAL_TABLE =GENERATE_ONLY NOT_USED EXECUTE sqlldr USERID=system/sammyy1 CONTROL=test.ctl \ EXTERNAL_TABLE=GENERATE_ONLY -- (3) check the generated log file
/* Unloading into External Tables */ -- (using datapump drivers)
Page 143
CREATE TABLE test_xt ORGANIZATION EXTERNAL( TYPE ORACLE_DATAPUMP DEFAULT DIRECTORY ext_data_dir ACCESS PARAMETERS (COMPRESSION ENABLED) LOCATION ('test_xt.dmp')) AS SELECT * FROM scott.dept; -- encrypting loaded data CREATE TABLE TEST ORGANIZATION EXTERNAL (TYPE ORACLE_DATAPUMP DEFAULT DIRECTORY test_dir1 ACCESS PARAMETERS (ENCRYPTION ENABLED) LOCATION ('test.dmp')); -- unload into multiple files CREATE TABLE customers ORGANIZATION EXTERNAL (TYPE ORACLE_DATAPUMP DEFAULT DIRECTORY ext_data_dir ACCESS PARAMETERS (NOBADFILE NOLOGFILE) LOCATION ('customers1.exp', 'customers2.exp', 'customers3.exp', 'customers4.exp')) PARALLEL 4 REJECT LIMIT UNLIMITED AS SELECT c.*, co.country_name, co.country_subregion, co.country_region FROM customers c, countries co where co.country_id=c.country_id;
-- enable parallel for loading (good if lots of data to load) ALTER SESSION ENABLE PARALLEL DML; -- load the data in hr employees table INSERT INTO emp (employee_id, first_name, last_name, job_id, manager_id, hire_date, salary, commission_pct, department_id, email) SELECT * FROM admin_ext_employees;
/* Altering External Tables */ ALTER TABLE admin_ext_employees REJECT LIMIT 100; -- access driver processes only the columns in the select list (default) ALTER TABLE admin_ext_employees PROJECT COLUMN REFERNCED; -- the access driver processes all of the columns -- This setting always provides a consistent set of returned rows ALTER TABLE admin_ext_employees PROJECT COLUMN ALL; -- default directory ALTER TABLE admin_ext_employees DEFAULT DIRECTORY admin_dat2_dir; -- access parameter ALTER TABLE admin_ext_employees ACCESS PARAMETERS (FIELDS TERMINATED BY ';'); -- location ALTER TABLE admin_ext_employees LOCATION ('empxt3.txt', 'empxt4.txt');
Page 144
Managing Indexes
Using Indexes
When creating an extremely large index, consider allocating a larger temporary tablespace specially for the index creation. If you want the optimizer to consider the invisible indexes in its operation, you can set the new initialization parameter OPTIMIZER_USE_INVISIBLE_INDEXES to TRUE (the default is FALSE). You can set the parameter in the system and session levels. -- obtain info about indexes select * from DBA_INDEXES; select * from DBA_IND_COLUMNS; -- for funcation-based indexes select * from DBA_IND_EXPRESSIONS; -- optimizer stats select * from DBA_IND_STATISTICS; -- obtained from ANALYZE INDEX...VALIDATE STRUCTURE select * from INDEX_STATS; -- create index CREATE INDEX emp_ename ON emp(ename) TABLESPACE users STORAGE (INITIAL 20K NEXT 20k PCTINCREASE 75); CREATE INDEX emp_ename ON emp(ename) COMPUTE STATISTICS; CREATE INDEX emp_name ON emp (mgr, emp1, emp2, emp3) ONLINE; CREATE UNIQUE INDEX dept_unique_index ON dept (dname) TABLESPACE indx; CREATE TABLE emp ( empno NUMBER(5) PRIMARY KEY, age INTEGER) ENABLE PRIMARY KEY USING INDEX TABLESPACE users; CREATE TABLE a ( a1 INT PRIMARY KEY USING INDEX (create index ai on a (a1))); CREATE TABLE b( b1 INT, b2 INT, CONSTRAINT bu1 UNIQUE (b1, b2) USING INDEX (create unique index bi on b(b1, b2)), CONSTRAINT bu2 UNIQUE (b2, b1) USING INDEX bi); CREATE TABLE c(c1 INT, c2 INT); CREATE INDEX ci ON c (c1, c2); ALTER TABLE c ADD CONSTRAINT cpk PRIMARY KEY (c1) USING INDEX ci; -- bitmap index Note: do not create it on a table that might be updated by more than one session, otherwise you risk seeing ORA-02049 error now and then! CREATE BITMAP INDEX gender_idx ON employee(gender) TABLESPACE emp_index_05; -- Reverse-Key Indexes -- good when you do sequential insertion of values into the index CREATE INDEX reverse_idx ON employee(emp_id) REVERSE; -- function based CREATE INDEX area_index ON rivers (area(geo)); -- compress CREATE INDEX emp_ename ON emp(ename) TABLESPACE users COMPRESS 1; ALTER INDEX emp_ename REBUILD NOCOMPRESS;
Page 145
-- Invisible Index (11g) CREATE INDEX name_indx ON employees(emp_name) INVISIBLE; SELECT /*+ index (EMP_NAME NAME_INDEX) */ ... ALTER INDEX name_indx VISIBLE; ALTER INDEX name_indx INVISIBLE; SELECT INDEX_NAME, VISIBILITY FROM DBA_INDEXES WHERE INDEX_NAME='NAME_INDX'; -- Bitmap Join Index (see next section). -- rebuilding an index ALTER INDEX emp_name REBUILD; ALTER INDEX emp_name REBUILD ONLINE; -- Monitoring Index Usage -- as the index owner -- Note: generally, do not drop an index build on a FK column ALTER INDEX index MONITORING USAGE; ALTER INDEX index NOMONITORING USAGE; Select INDEX_NAME,TABLE_NAME,MONITORING,USED,START_MONITORING,END_MONITORING From V$OBJECT_USAGE -- as the index owner WHERE INDEX_NAME='EMP_LNAME_IND'; -- Monitoring Space Use of Indexes -- 1. Analyzing statistics exec DBMS_STATS.GATHER_INDEX_STATS( OWNNAME=>'HR', INDNAME=>'EMP_LNAME_IND') -- 2. Validating the index ANALYZE INDEX EMP_LNAME_IND VALIDATE STRUCTURE; -- 3. Checking PCT_USED SELECT PCT_USED FROM INDEX_STATS WHERE NAME = 'EMP_LNAME_IND'; -- 4. Dropping and rebuilding (or coalescing) the index ALTER INDEX emp_name REBUILD; -- estimating index space -- gather index and its underlying table statistics SET SERVEROUTPUT ON declare l_index_ddl VARCHAR2(1000); l_used_bytes NUMBER; l_allocated_bytes NUMBER; BEGIN DBMS_SPACE.create_index_cost ( DDL => 'create index persons_idx on persons(person_id)', USED_BYTES => l_used_bytes, ALLOC_BYTES => l_allocated_bytes); DBMS_OUTPUT.PUT_LINE ('used = ' || to_char(l_used_bytes/1024) || ' KB ' || ' allocated = ' || to_char(l_allocated_bytes/1024) || ' KB'); END; /
Using Bitmap Join Indexes (BJI)

A bitmap join index is a bitmap index for the join of two or more tables. It is designed when joining fact tables with dimension tables as in warehouse databases (start schema model). CREATE BITMAP INDEX NAMES_CTR_bji ON NAMES2(COUNTRIES.COUNTRY_NAME)
Page 146
FROM NAMES2, COUNTRIES WHERE NAMES2.NAT_ID=COUNTRIES.ID NOLOGGING COMPUTE STATISTICS; SELECT /*+ INDEX_COMBINE(N NAMES_CTR_bji)*/ C.COUNTRY_NAME FROM NAMES2 N, COUNTRIES C WHERE N.NAT_ID = C.ID;
Partitioned Indexes
-- Global Indexes -- can be nonpartitioned or partitioned CREATE INDEX ticketsales_idx ON ticket_sales(month) GLOBAL PARTITION BY range(month) (PARTITION ticketsales1_idx VALUES LESS PARTITION ticketsales1_idx VALUES LESS PARTITION ticketsales2_idx VALUES LESS PARTITION ticketsales3_idx VALUES LESS CREATE INDEX hgidx ON tab (c1,c2,c3) GLOBAL PARTITION BY HASH (c1,c2) (PARTITION p1 TABLESPACE ts1, PARTITION p2 TABLESPACE ts2, PARTITION p3 TABLESPACE ts3, PARTITION p4 TABLESPACE ts4); -- Local Indexes CREATE INDEX ticket_no_idx ON ticket_sales(ticket__no) LOCAL TABLESPACE localidx_01; indexes
THAN THAN THAN THAN
(3) (6) (9) (MAXVALUE));
Page 147
Managing Materialized Views

Obtaining Information about Materialized Views -- all info about the mviews SELECT OWNER, MVIEW_NAME, CONTAINER_NAME, QUERY, QUERY_LEN, UPDATABLE, UPDATE_LOG, MASTER_ROLLBACK_SEG, MASTER_LINK, REWRITE_ENABLED, REWRITE_CAPABILITY, REFRESH_MODE, REFRESH_METHOD, BUILD_MODE, FAST_REFRESHABLE, LAST_REFRESH_TYPE, LAST_REFRESH_DATE, STALENESS, AFTER_FAST_REFRESH, UNKNOWN_PREBUILT, UNKNOWN_PLSQL_FUNC, UNKNOWN_EXTERNAL_TABLE, UNKNOWN_CONSIDER_FRESH, UNKNOWN_IMPORT, UNKNOWN_TRUSTED_FD, COMPILE_STATE, USE_NO_INDEX, STALE_SINCE FROM DBA_MVIEWS ORDER BY OWNER, MVIEW_NAME
-- list mview last refreshed on year or longer ago -- Note: they cause the mview logs linked to them to get so large by time SELECT OWNER, MVIEW_NAME, TO_CHAR(LAST_REFRESH_DATE,'DD-MM-YY') LAST_REFRESHED_DATE FROM DBA_MVIEWS WHERE LAST_REFRESH_DATE < SYSDATE-365 ORDER BY LAST_REFRESH_DATE ASC
-- determine the master table (and last refresh time) select OWNER, NAME, MASTER_OWNER, MASTER, LAST_REFRESH from DBA_SNAPSHOT_REFRESH_TIMES
-- FAST_REFRESHABLE mviews which couldn't make FAST refresh -- in their last refreshes SELECT OWNER, MVIEW_NAME, REFRESH_METHOD, FAST_REFRESHABLE, LAST_REFRESH_TYPE, LAST_REFRESH_DATE, STALENESS, AFTER_FAST_REFRESH, COMPILE_STATE, STALE_SINCE FROM DBA_MVIEWS WHERE FAST_REFRESHABLE<>'NO' and LAST_REFRESH_TYPE<>'FAST' ORDER BY OWNER, MVIEW_NAME
-- all MATERIALIZED VIEW LOGS SELECT LOG_OWNER, MASTER, LOG_TABLE, LOG_TRIGGER, ROWIDS, PRIMARY_KEY, OBJECT_ID, FILTER_COLUMNS, SEQUENCE, INCLUDE_NEW_VALUES FROM DBA_MVIEW_LOGS ORDER BY LOG_OWNER, MASTER Monitoring the Progress of a Materialized View Refresh Reference: Document ID 258021.1 /* Determine if a Specific MVIEW is Being Refreshed */
Page 148
select o.owner, o.object_name mview, username, s.sid from v$lock l, dba_objects o, v$session s where o.object_id=l.id1 and l.type='JI' and l.lmode=6 and s.sid=l.sid and o.object_type='TABLE';
/* Determine if a Refresh Group is Being Refreshed */ -- Given the name of the refresh group and its owner, the -- following query can be used to identify if a refresh -- is being executed by a job queue process: select s.sid, s.username from dba_jobs_running jr, v$session s, dba_jobs j where jr.sid=s.sid and j.job=jr.job and upper(j.what) like '%REFRESH%<name of the refresh group>%'; -- to determine if a refresh has been executed manually ( -- not through a background job ) select username, sid, rowner, rname from ( select username, s.sid, rc.rowner, rc.rname, count(*) from v$lock l, dba_objects o, v$session s, dba_refresh_children rc where o.object_id=l.id1 and l.type='JI' and l.lmode=6 and s.sid=l.sid and o.object_type='TABLE' and o.object_name=rc.name and o.owner=rc.owner and rc.type='SNAPSHOT' group by username, s.sid, rc.rowner, rc.rname having count(*)=( select count(*) from dba_refresh_children where rowner= rc.rowner and rname=rc.rname and type='SNAPSHOT') );
/* Determine which MVIEW in a Refresh Group is Being Refreshed */ select SID, SERIAL#, CURRMVOWNER, CURRMVNAME from v$mvrefresh
/* Determine the Current Phase of a Refresh */ -- if in Propagation phase (push and purge) -- locks are released once push and purge finished -- in Push phase: select l.sid, decode( count(*), 0, 'No propagation in progress', 'Propagation in progress' ) State from v$lock l, dbms_lock_allocated la where l.type='UL' and l.lmode=4 and l.id1=la.lockid and
Page 149
la.name='ORA$DEF$EXE$PushCommonLock' group by l.sid; -- identifying the target site of the propagation: select l.sid, 'Currently propagating to ' || substr(la.name, 13) from v$lock l, dbms_lock_allocated la where l.type='UL' and l.lmode=6 and l.id1=la.lockid and la.name like 'ORA$DEF$EXE$%'; -- in Purge phase: -- if refresh is invoked with purge_option>0 and parallelism>0 select l.sid, decode( count(*), 0, 'No purge in progress', 'Purge is in progress' ) State from v$lock l, dbms_lock_allocated la where l.type='UL' and l.lmode=6 and l.id1=la.lockid and la.name='ORA$DEF$EXE$PurgeCommonLock' group by l.sid;
-- Refresh Subphases ( SETUP, INSTATNIATION and WRAPUP) -- to get: - the type of the ongoing refresh - the phase of the ongoing refresh - the number of DMLs performed by the refresh select CURRMVOWNER_KNSTMVR || '.' || CURRMVNAME_KNSTMVR "MVIEW BEING REFRESHED", decode( REFTYPE_KNSTMVR, 1, 'FAST', 2, 'COMPLETE', 'UNKNOWN' ) REFTYPE, decode(GROUPSTATE_KNSTMVR, 1, 'SETUP', 2, 'INSTANTIATE', 3, 'WRAPUP', 'UNKNOWN' ) STATE, TOTAL_INSERTS_KNSTMVR INSERTS, TOTAL_UPDATES_KNSTMVR UPDATES, TOTAL_DELETES_KNSTMVR DELETES from X$KNSTMVR X WHERE type_knst=6 and exists (select 1 from v$session s where s.sid=x.sid_knst and s.serial#=x.serial_knst);
/* Steps to Determine Whether a Refresh is Hanging, or Moving Slowly */ -- if the refresh in Propagation: -- check for blocking sessions: -- to identify if a row being pushed is locked by a users session: select s.username, w.holding_session holder, h.username holder_name, s.row_wait_row# rowno, o.object_name, o.owner from dba_waiters w, dba_objects o, v$session s, v$session h
Page 150
where w.waiting_session = s.sid and w.holding_session = h.sid and s.row_wait_obj# = o.object_id; -- Check the wait events -- if refresh is being executed from DBMS_JOB: select sw.sid, sw.event, sw.p1, sw.p2 from v$session_wait sw where sw.sid in ( select qs.sid from dba_jobs_running jr, dba_jobs j, v$px_session qs where jr.job = j.job and jr.sid = qs.qcsid and upper(j.what) like '%..%') and sw.wait_time = 0; -- if refresh is being executed by a users session: select sw.sid, sw.event, sw.p1, sw.p2 from v$session_wait sw, v$session s where sw.sid = s.sid and sw.sid in (select qs.sid from v$px_session qs) and sw.wait_time = 0
-- Check if a large error is being queued: -- to identify if a large error is being queued at the remote master site: select count(*) from v$sqlarea a, v$session s where s.sql_address = a.address and s.sql_hash_value = a.hash_value and a.sql_text like '%DEF$_AQERROR%' and s.username = 'REPADMIN';
-- if refresh in WRAPUP phase: -- get the wait event : -- in normal cases, it's read scattered or sequential: select event, p1, p2, p3 from v$session_wait where sid=<SID of the session at master site>; -- the object that is being accessed by: select owner, segment_name from dba_extents where file_id=<P1> and <P2> between block_id and block_id+blocks-1; -- if the accessed object is MView log, check the following section -which diagnoses the MView Log
-- if refresh in INSTANTIATION phase: -- the rows are pulled from the master site -- check the wait: select event, p1, p2, p3 from v$session_wait where sid=<SID of the session at master site>;
Page 151
-- if the event is enqueue: check the blocking session select s.username, w.holding_session holder, h.username holder_name, s.row_wait_row# rowno, o.object_name, o.owner from dba_waiters w, dba_objects o, v$session s, v$session h where w.waiting_session = s.sid and w.holding_session = h.sid and s.row_wait_obj# = o.object_id; -- if event the session if the wait submit a SR is at is in SQL*Net message from dblink: master site should be checked for the type of the wait "SQL*Net message from dblink" in both sites, Oracle Support.
Materialized View Typical Refresh Errors ORA-12004: "REFRESH FAST cannot be used for ... This error indicates a problem with the log at master. See Note:179469.1 for further information. ORA-12034: "snapshot log on "%s"."%s" younger than last refresh" This error also indicates a problem with the log at master. See Note:204127.1 for further information. ORA-23402: refresh was aborted because of conflicts caused by deferred txns This error is caused by outstanding conflicts logged in the DefError table at the master. This can be workaround by setting refresh_after_errors to true. See Note:1031119.6 and Note:39232.1 for the details. ORA-23385: replication parallel push heap_size argument not valid This error is caused if the heap_size value is set to NULL. Query rgroup$ to obtain the current value of heap_size and use dbms_refresh.change to set it to not null value. IE. 0. See Note:49558.1 for the error definition. Using Materialized Views Good reference is Note ID 179466.1 If USER1 wants to refresh MView created by USER2 and MView Master Table is owned by USER3, then the following privileges are required: grant SELECT ANY TABLE to USER2; grant CREATE TABLE, CREATE ANY MATERIALIZED VIEW to USER2; grant FLASHBACK on USER3.OAS_DOCLINE to USER1; grant FLASHBACK on USER3.OAS_DOCLINE to USER2; grant FLASHBACK on USER3.MLOG$_OAS_DOCLINE to USER1; grant FLASHBACK on USER3..MLOG$_OAS_DOCLINE to USER2; For COMPLETE refresh, to avoid logging, you can consider setting ATOMIC_REFRESH=false. It makes the refresh faster. When creating materialized view, you should keep in mind the following restrictions: The defining query of the materialized view cannot contain any non-repeatable expressions (ROWNUM, SYSDATE, non-repeatable PL/SQL functions, and so on).
Page 152
The query cannot contain any references to RAW or LONG RAW datatypes or object REFs Restrictions on Fast Refresh The defining query of the materialized view is restricted as follows: The materialized view must not contain references to non-repeating expressions like SYSDATE and ROWNUM. The materialized view must not contain references to RAW or LONG RAW data types. It cannot contain a SELECT list subquery. It cannot contain analytical functions (for example, RANK) in the SELECT clause. It cannot contain a MODEL clause. It cannot contain a HAVING clause with a subquery. It cannot contain nested queries that have ANY, ALL, or NOT EXISTS. It cannot contain a [START WITH ] CONNECT BY clause. It cannot contain multiple detail tables at different sites. ON COMMIT materialized views cannot have remote detail tables. Nested materialized views must have a join or aggregate. If you received the following error when trying to refresh a materialized view, trying compiling it: ORA-00942: table or view does not exist ORA-06512: at "SYS.DBMS_SNAPSHOT", alter materialized view compile -- required privs GRANT CREATE MATERIALIZED VIEW TO hr; GRANT QUERY REWRITE TO hr; -- mv log -- you must specify the ROWID, SEQUENCE and the INCLUDING NEW VALUES -- For aggregate materialized views, it must also contain every column -in the table referenced in the materialized view CREATE MATERIALIZED VIEW LOG ON employees WITH SEQUENCE, ROWID (employee_id,department_id,salary) INCLUDING NEW VALUES; CREATE MATERIALIZED VIEW LOG ON departments WITH SEQUENCE, ROWID (department_id, department_name) INCLUDING NEW VALUES; ALTER MATERIALIZED VIEW LOG ON sales ADD ROWID (prod_id, cust_id, time_id, amount_sold) INCLUDING NEW VALUES; DROP MATERIALIZED VIEW LOG ON sales; -- create mv -- the options BUILD IMMEDIATE | DEFERRED REFRESH NEVER, FAST, FORCE, COMPLETE - on commit or on demand ENABLE QUERY REWRITE -- is a must for rewriting queries. DISABLE WITH PRIMARY KEY START WITH ROUND(SYSDATE + 1) + 11/24 NEXT NEXT_DAY(TRUNC(SYSDATE), 'MONDAY') + 15/24 recommended to gather stats on it then EXECUTE DBMS_STATS.GATHER_TABLE_STATS ( 'HR','DEPT01_MV', estimate_percent
Page 153
=> 20, block_sample
=> TRUE, cascade
=> TRUE);
-- (1) Materialized Views with Aggregates -- vm log must also contain every column -in the table referenced in the materialized view -- there must be a COUNT(*) and a COUNT(column) on any aggregated columns CREATE MATERIALIZED VIEW dept01_mv (department_id, department_name, salary_count, average_salary) TABLESPACE ts1 BUILD IMMEDIATE REFRESH FAST ENABLE QUERY REWRITE AS select d.department_id, d.department_name, count(e.salary), -- it is a must for mv with aggregate or count(*), avg(e.salary) from departments d, employees e where d.department_id = e.department_id group by d.department_id, d.department_name; CREATE MATERIALIZED VIEW product_sales_mv TABLESPACE ts1 BUILD IMMEDIATE -- DEFERRED REFRESH FAST -- FAST, FORCE, COMPLETE - on commit or on demand ENABLE QUERY REWRITE AS SELECT p.prod_name, SUM(s.amount_sold) AS dollar_sales, COUNT(*) AS cnt, COUNT(s.amount_sold) AS cnt_amt FROM sales s, products p WHERE s.prod_id = p.prod_id GROUP BY p.prod_name;
-- (2) Materialized Views Containing Only Joins -- contain only joins and no aggregates -- ROWID column must be present in each materialized view log and mv SELECT CREATE MATERIALIZED VIEW LOG ON sales WITH ROWID; CREATE MATERIALIZED VIEW LOG ON times WITH ROWID; CREATE MATERIALIZED VIEW LOG ON customers WITH ROWID; CREATE MATERIALIZED VIEW detail_sales_mv PARALLEL BUILD IMMEDIATE REFRESH FAST AS SELECT s.rowid "sales_rid", t.rowid "times_rid", c.rowid "customers_rid", c.cust_id, c.cust_last_name, s.amount_sold, s.quantity_sold, s.time_id FROM sales s, times t, customers c WHERE s.cust_id = c.cust_id(+) AND s.time_id = t.time_id(+);
-- (3) Nested Materialized Views -- a materialized view whose definition is based on another materialized view CREATE MATERIALIZED VIEW LOG ON sales WITH ROWID; CREATE MATERIALIZED VIEW LOG ON customers WITH ROWID; CREATE MATERIALIZED VIEW LOG ON times WITH ROWID; /*create materialized view join_sales_cust_time as fast refreshable at COMMIT time */ CREATE MATERIALIZED VIEW join_sales_cust_time REFRESH FAST ON COMMIT AS
Page 154
SELECT c.cust_id, c.cust_last_name, s.amount_sold, t.time_id, t.day_number_in_week, s.rowid srid, t.rowid trid, c.rowid crid FROM sales s, customers c, times t WHERE s.time_id = t.time_id AND s.cust_id = c.cust_id; /* create materialized view log on join_sales_cust_time and include the required columns */ CREATE MATERIALIZED VIEW LOG ON join_sales_cust_time WITH ROWID (cust_last_name, day_number_in_week, amount_sold) INCLUDING NEW VALUES; /* create the single-table aggregate materialized view sum_sales_cust_time on join_sales_cust_time as fast refreshable at COMMIT time */ CREATE MATERIALIZED VIEW sum_sales_cust_time REFRESH FAST ON COMMIT AS SELECT COUNT(*) cnt_all, SUM(amount_sold) sum_sales, COUNT(amount_sold) cnt_sales, cust_last_name, day_number_in_week FROM join_sales_cust_time GROUP BY cust_last_name, day_number_in_week; -- Refreshing a Nested Materialized View DBMS_MVIEW.REFRESH('SALES_MV,COST_MV', nested => TRUE); /* commenting on mvs */ COMMENT ON MATERIALIZED VIEW sales_mv IS 'sales materialized view'; SELECT MVIEW_NAME, COMMENTS FROM USER_MVIEW_COMMENTS WHERE MVIEW_NAME = 'SALES_MV'; Using Query Rewriting o o o A query is rewritten only when a certain number of conditions are met: Query rewrite must be enabled for the session. A materialized view must be enabled for query rewrite. The rewrite integrity level should allow the use of the materialized view. Use Column Alias Lists in the CREATE MATERIALIZED VIEW command to let the query rewrite work /* Required Privs */ grant query rewrite to hr; -- if the mv uses a table from other schema grant query rewrite on employees to scott; or grant global query rewrite to hr; /* Ensuring that Query Rewrite Takes Effect */ -- (1) query rewrite enabled for the mv select REWRITE_ENABLED from user_mviews where mview_name='DEPT01_MV'; ALTER MATERIALIZED VIEW dept01_mv ENABLE QUERY REWRITE ; -- (2) QUERY_REWRITE_ENABLED = TRUE | FORCE ALTER SESSION SET QUERY_REWRITE_ENABLED=true; select get_par('query_rewrite_enabled') from dual; -- for user with dba priv select value from v$parameter where upper(name)='QUERY_REWRITE_ENABLED'; alter system set QUERY_REWRITE_ENABLED=TRUE; -- (3) OPTIMIZER_MODE = ALL_ROWS, FIRST_ROWS, or FIRST_ROWS_n, (where n = 1, 10, 100, 1000) select get_par('OPTIMIZER_MODE') from dual; -- (4) OPTIMIZER_FEATURES_ENABLE must be 10.0.0 or higher select value from v$parameter where upper(name)='OPTIMIZER_FEATURES_ENABLE'; Oracle DBA Code Examples
Page 155
/* Column Alias */ -- failed example CREATE MATERIALIZED VIEW sales_mv ENABLE QUERY REWRITE AS SELECT s.time_id sales_tid, c.time_id costs_tid FROM sales s, products p, costs c WHERE s.prod_id = p.prod_id AND c.prod_id = p.prod_id AND p.prod_name IN (SELECT prod_name FROM products); -- the right way CREATE MATERIALIZED VIEW sales_mv (sales_tid, costs_tid) ENABLE QUERY REWRITE AS SELECT s.time_id, c.time_id FROM sales s, products p, costs c WHERE s.prod_id = p.prod_id AND c.prod_id = p.prod_id AND p.prod_name IN (SELECT prod_name FROM products); /* Rewrite Integrity */ -- QUERY_REWRITE_INTEGRITY: ENFORCED (returned data is 100% correct), TRUSTED, STALE_TOLERATED /* Verifying that Query Rewrite has Occurred */ EXPLAIN PLAN FOR select d.department_id, d.department_name, count(e.salary), avg(e.salary) from departments d, employees e where d.department_id = e.department_id group by d.department_id, d.department_name; -- if rewriting occurred, you should see operation: MAT_VIEW REWRITE ACCESS SELECT OPERATION, OBJECT_NAME FROM PLAN_TABLE; /* Using the EXPLAIN_REWRITE Procedure with Query Rewrite */ -- target: to know why rewriting didn't occur -- 1) create EXPLAIN_REWRITE table @/oracle/oradb10g/rdbms/admin/utlxrw.sql -- 2) execute declare v varchar2(4000); begin v := 'select d.department_id, d.department_name, count(e.salary), avg(e.salary) from departments d, employees e where d.department_id = e.department_id group by d.department_id, d.department_name'; DBMS_MVIEW.EXPLAIN_REWRITE( query => v, mv =>'', -- you can specify the comma-separated mv(s) or NULL to consider all mvs statement_id => 'ID6'); -- client-supplied unique identifier to distinguish output messages end; / SELECT message, original_cost, rewritten_cost FROM rewrite_table ORDER BY sequence; -- alternatively to see neat output use SYS.XRW (see Wareshouse documentation)
Page 156
ReWrite Hints /* REWRITE and NOREWRITE Hints */ SELECT /*+ NOREWRITE */ ... SELECT /*+ REWRITE (sum_sales_pscat_week_mv) */ /* The Rewrite_or_Error Hint */ -- use the mv or error SELECT /*+ REWRITE_OR_ERROR */ prod_id, SUM(quantity_sold) AS sum_sales_qty FROM sales_data GROUP BY prod_id; /* NO_MULTIMV_REWRITE hint prevents the query from being rewritten with more than one materialized NO_BASETABLE_MULTIMV_REWRITE hint prevents the query from being rewritten with a combination of materialized views and the base tables */ Using EXPLAIN_MVIEW Procedure: Viewing Materialized View Capabilities Target: to determine if a materialized view is fast refreshable and what types of query rewrite you can perform with a particular materialized view. DBMS_MVIEW.EXPLAIN_MVIEW ( mv -- mv, a SELECT or a CREATE MATERIALIZED VIEW statement statement_id -- any id --(1) create MV_CAPABILITIES_TABLE @/oracle/oradb10g/rdbms/admin/utlxmv.sql --(2) execute declare v varchar2(4000); begin v := 'DEPT01_MV'; DBMS_MVIEW.EXPLAIN_MVIEW( V,'ID01'); Commit; end; / SELECT CAPABILITY_NAME, POSSIBLE, RELATED_TEXT,RELATED_NUM, MSGNO, MSGTXT FROM MV_CAPABILITIES_TABLE WHERE STATEMENT_ID='ID01' ORDER BY SEQ / SELECT MSGTXT FROM MV_CAPABILITIES_TABLE WHERE STATEMENT_ID='ID01' ORDER BY SEQ / Using DBMS_ADVISOR.TUNE_MVIEW The TUNE_MVIEW procedure shows how to decompose a materialized view into two or more materialized views or to restate the materialized view in a way that is more advantageous for fast refresh and query rewrite. TUNE_MVIEW analyzes and processes the input statement and generates two sets of output results: one for the materialized view implementation and the other for undoing the CREATE materialized view operations. Oracle DBA Code Examples
...
Page 157
ADVISOR privilege required to execute the procedure. declare task_cust_mv VARCHAR2(30); create_mv_ddl VARCHAR2(4000); begin task_cust_mv := 'cust_mv'; create_mv_ddl := 'CREATE MATERIALIZED VIEW REPTEST.MP_OAS_BALANCE_MV BUILD IMMEDIATE REFRESH FORCE ON DEMAND WITH PRIMARY KEY as SELECT cmpcode || LPAD (TO_CHAR (yr), 4, 0) "BAL_KEY", CMPCODE, YR, PERIOD, BALCODE, CURCODE, REPBASIS, CURFLAG, ACCODE, DEBIT_VALUE, CREDIT_VALUE, 1 UNMARKER, ROWID RID FROM OAS_BALANCE WHERE CMPCODE NOT IN (''XYZ'', ''KH'', ''GUI'')'; DBMS_ADVISOR.TUNE_MVIEW(task_cust_mv, create_mv_ddl); end; / -- Access IMPLEMENTATION Output SELECT * FROM USER_TUNE_MVIEW WHERE TASK_NAME= 'cust_mv' AND SCRIPT_TYPE='IMPLEMENTATION'; -- Save IMPLEMENTATION Output in a Script File CREATE DIRECTORY TUNE_RESULTS AS '/myscript' GRANT READ, WRITE ON DIRECTORY TUNE_RESULTS TO PUBLIC; EXECUTE DBMS_ADVISOR.CREATE_FILE(DBMS_ADVISOR.GET_TASK_SCRIPT('cust_mv'), 'TUNE_RESULTS', 'mv_create.sql');
Registering a User-defined Table as Materialized View -- table and mv have the same name CREATE TABLE sum_sales_tab PCTFREE 0 TABLESPACE demo AS SELECT s.prod_id, SUM(amount_sold) AS dollar_sales, SUM(quantity_sold) AS unit_sales FROM sales s GROUP BY s.prod_id; CREATE MATERIALIZED VIEW sum_sales_tab_mv ON PREBUILT TABLE WITHOUT REDUCED PRECISION ENABLE QUERY REWRITE AS SELECT s.prod_id, SUM(amount_sold) AS dollar_sales, SUM(quantity_sold) AS unit_sales FROM sales s GROUP BY s.prod_id;
Page 158
Managing Clusters and Hash Clusters

Hashing is useful when you have the following conditions: Most queries are equality queries on the cluster key: SELECT ... WHERE cluster_key = ...; The tables in the hash cluster are primarily static in size so that you can determine the number of rows and amount of space required for the tables in the cluster.
-- obtaining info on clusters select * from DBA_CLUSTERS; -- map table columns to cluster columns select * from DBA_CLU_COLUMNS; -- create cluster CREATE CLUSTER emp_dept (deptno NUMBER(3)) SIZE 600 TABLESPACE users STORAGE (INITIAL 200K NEXT 300K MINEXTENTS 2 MAXEXTENTS 20 PCTINCREASE 33); CREATE TABLE emp ( empno NUMBER(5) PRIMARY KEY, ename VARCHAR2(15) NOT NULL, . . . deptno NUMBER(3) REFERENCES dept) CLUSTER emp_dept (deptno); CREATE TABLE dept ( deptno NUMBER(3) PRIMARY KEY, . . . ) CLUSTER emp_dept (deptno); -- create cluster index CREATE INDEX emp_dept_index ON CLUSTER emp_dept TABLESPACE users STORAGE (INITIAL 50K NEXT 50K MINEXTENTS 2 MAXEXTENTS 10 PCTINCREASE 33); -- dropping a cluster -- no tables DROP CLUSTER emp_dept; -- there are tables DROP CLUSTER emp_dept INCLUDING TABLES; DROP CLUSTER emp_dept INCLUDING TABLES CASCADE CONSTRAINTS; -- drop a table in the cluster drop table dept; /* Managing Hash Clusters */ -- obtain info selet * from DBA_CLUSTERS ;
Page 159
selet * from DBA_CLU_COLUMNS; selet * from DBA_CLUSTER_HASH_EXPRESSIONS; -- Creating Hash Clusters CREATE CLUSTER trial_cluster (trialno NUMBER(5,0)) TABLESPACE users STORAGE (INITIAL 250K NEXT 50K MINEXTENTS 1 MAXEXTENTS 3 PCTINCREASE 0) HASH IS trialno HASHKEYS 150; CREATE TABLE trial ( trialno NUMBER(5,0) PRIMARY KEY, ...) CLUSTER trial_cluster (trialno); -- Creating a Sorted Hash Cluster CREATE CLUSTER call_detail_cluster ( telephone_number NUMBER, call_timestamp NUMBER SORT, call_duration NUMBER SORT ) HASHKEYS 10000 HASH IS telephone_number SIZE 256; CREATE TABLE call_detail ( telephone_number NUMBER, call_timestamp NUMBER SORT, call_duration NUMBER SORT, other_info VARCHAR2(30) ) CLUSTER call_detail_cluster ( telephone_number, call_timestamp, call_duration ); -- following select gets advantage from the sorted hash cluster SELECT * WHERE telephone_number = 6505551212; -- Creating Single-Table Hash Clusters CREATE CLUSTER peanut (variety NUMBER) SIZE 512 SINGLE TABLE HASHKEYS 500; -- Dropping Hash Clusters DROP CLUSTER emp_dept;
Page 160
Managing Views, Sequences, and Synonyms

-- create views CREATE VIEW sales_staff AS SELECT empno, ename, deptno FROM emp WHERE deptno = 10 WITH CHECK OPTION CONSTRAINT sales_staff_cnst; -- Creating Views with Errors CREATE FORCE VIEW AS ...; -- creating sequence CREATE SEQUENCE emp_sequence INCREMENT BY 1 START WITH 1 NOMAXVALUE NOCYCLE CACHE 10; -- alter a sequence ALTER SEQUENCE emp_sequence INCREMENT BY 10 MAXVALUE 10000 CYCLE CACHE 20; -- Creating Synonyms CREATE PUBLIC SYNONYM public_emp FOR jward.emp;
Page 161
Managing Transactions
Implementing Oracles Concurrency Control
Oracle Isolaction Levels o o o o There are four possible isolation levels: Serializable: doesnt allow any concurrent. Repeatable read Read uncommitted Read committed (Default): Oracle only guarantees statement-level isolation here (changes between reads are seen), not transaction-level isolation.
command SET TRANSACTION READ ONLY
Its effect When set, the transaction that follows operates on essentially a snapshot of the database at the time the command was issued. This is especially useful when multiple select statements are executed over the course of a transaction, and data must be consistent. Set the transactgion back to read write if we try to modify a record that already has a DML Row Exclusive lock on that record, the attempt to update will wait until the locks are released. if we try to modify a record that already has a DML Row Exclusive lock on that record, the attempt to update will fail
SET TRANSACTION READ WRITE SET TRANSACTION ISOLATION LEVEL READ COMMITTED SET TRANSACTION ISOLATION LEVEL SERIALIZABLE
-- must be the first command in an transaction -- to enable transaction-level isolation ALTER SESSION SET ISOLATION LEVEL SERIALIZABLE; SET TRANSACTION ISOLATION LEVEL SERIALIZABLE; -- alternatively ( repeatable reads ) SET TRANSACTION READ ONLY Oracle Lock Types DML Locks: Row-Level Lock: Exclusive Table-Level Lock: Row exclusive
DDL Locks: DDL is not allowed if DML Lock was there and "ORA-00054: resource bus" will be returned unless DDL_LOCK_TIMEOUT is defined in seconds (11g). Explicit Table Locking: DDL requires Table exclusive DML lock. You can specify that a DDL command wait for a specific length of time before it fails: LOCK TABLE ... IN lockmode MODE [NOWAIT | WAIT integer] lockmode: [ROW] SHARE, [[SHARE] ROW] EXCLUSIVE, SHARE UPDATE integer in seconds
Latches: are internal mechanisms that protect shared data structures in the SGA. Data dictionary locks: whenever the dictionary objects are being modified. Oracle DBA Code Examples
Page 162
Distributed locks: used in a distributed database system or in RAC. Internal locks: are used by Oracle to protect access to structures such as datafiles, tablespaces, and rollback segments. -- current locks in the DB select SID, DECODE(TO_CHAR(BLOCK),'0','Not-Blocking','1','Blocking') IS_BLOCKING, DECODE(TYPE,'TM','DML enqueue','TX','Transaction enqueue','UL','User supplied',TYPE) LOCK_TYPE, DECODE(TO_CHAR(LMODE),'0','None','1','Null','2','Row-S (SS)','3','Row-X (SX)','4','Share (S)','5','S/Row-X (SSX)','6','Exclusive (X)') HELD_LMODE, DECODE(TO_CHAR(REQUEST),'0','None','1','Null','2','Row-S (SS)','3','Row-X (SX)','4','Share (S)','5','S/Row-X (SSX)','6','Exclusive (X)') REQUEST_LMODE from v$lock; -- Oracle 11g: to allow DDL wait for lock instead or returning error ALTER SESSION SET ddl_lock_timeout = 30; -- explicit table locking (to acquire an exclusive lock=no updates) -- to release the lock: ROLLBACK, COMMIT lock table emp in EXCLUSIVE mode nowait ;
Identifying Blocking Sessions -- oracle supplied script printing blocking sessions in tree-like view @$ORACLE_HOME/rdbms/admin/utllockt.sql select SID, DECODE(TYPE,'TM','DML enqueue','TX','Transaction enqueue','UL','User supplied',TYPE) LOCK_TYPE, DECODE(TO_CHAR(LMODE),'0','None','1','Null','2','Row-S (SS)','3','Row-X (SX)','4','Share (S)','5','S/Row-X (SSX)','6','Exclusive (X)') HELD_LMODE, DECODE(TO_CHAR(REQUEST),'0','None','1','Null','2','Row-S (SS)','3','Row-X (SX)','4','Share (S)','5','S/Row-X (SSX)','6','Exclusive (X)') REQUEST_LMODE from v$lock where BLOCK=1; SELECT sid, username, blocking_session blocking_sid FROM V$SESSION WHERE blocking_session_status='VALID';
Using Autonomous Transaction

The autonomous transactions give you the ability to commit or roll back the subprograms changes independent of the main program. -- write errors log CREATE OR REPLACE PROCEDURE error_log (error__msg in varchar2, procedure_name IN VARCHAR2) IS PRAGMA AUTONOMOUS_TRANSACTION; BEGIN INSERT INTO log_table (error_msg, procedure_name) VALUES (error_msg,procedure_name)); COMMIT; EXCEPTION WHEN OTHERS THEN ROLLBACK; END error_log; /
Page 163
-- using DDL statements in triggers -- performing an audit of database queries -- performing an audit of and failed (unauthorized) database activity CREATE OR REPLACE TRIGGER aud_bef_trig BEFORE INSERT ON emp FOR EACH ROW DECLARE PRAGMA AUTONOMOUS_TRANSACTION BEGIN INSERT INTO audit_employee VALUES ( :new.username, 'before insert', sysdate); COMMIT; END; /
Managing Long Transactions with Workspace Manager

o o Workspace Manager is by default installed when you create a database using DBCA but not manually. To check, make sure WMSYS exists in DBA_USERS. With Workspace Manager, you can maintain multiple versions of data, which you can keep or discard as necessary. Its benifits: To enable simultaneous read and write access to data during long transactions. To create multiple data scenarios for what-if analyses. The original table will be renamed to tableName_LT, new table tableName_AUX will be created. Users be default are placed in LIVE workspace. Refresh: update the data in the refreshed workspace with the changes made in its parent workspace. Merge: update the parent workspace with the changes made to the merged workspace. Reference: Application Developers Guide - Workspace Manager -- should return LIVE select dbms_wm.getworkspace from dual; /* go back in time within a workspace */ -- the example is using the LIVE workspace -- privs begin dbms_wm.grantSystemPriv ( 'ACCESS_ANY_WORKSPACE, 'MERGE_ANY_WORKSPACE, 'CREATE_ANY_WORKSPACE, 'REMOVE_ANY_WORKSPACE, 'ROLLBACK_ANY_WORKSPACE' 'USER1' , 'YES'); end; /
' ' ' ' ,
|| || || ||
conn user1
Page 164
create table dept ( DEPTNO NUMBER(3) PRIMARY KEY, DNAME VARCHAR2(14) ) / create table EMP ( EMPNO NUMBER(4) PRIMARY KEY, -- PK is mandatory ENAME VARCHAR2(10), DEPTNO NUMBER(3) references dept(deptno)) /
-- set version enabled tables -- VIEW_WO_OVERWRITE = without Overwrite begin dbms_wm.enableVersioning('EMP','VIEW_WO_OVERWRITE'); dbms_wm.enableVersioning('DEPT','VIEW_WO_OVERWRITE'); end; / select TABLE_NAME, STATE, HISTORY from USER_WM_VERSIONED_TABLES order by TABLE_NAME ;
-- testing data -- Note: rows cannot be inserted if the Workspace isn't -- in its latest version by DBMS_WM.GotoSavePoint(); insert into dept values (1,'d1'); commit; insert into emp values (1,'e1',1); insert into emp values (2,'e2',1); commit; select * from emp, dept where emp.deptno=dept.deptno; -- save current time var dt_1 varchar2(21) begin select to_char(sysdate, 'dd.mm.yyyy hh24:mi:ss') into :dt_1 from dual; end; / -- insert extra rows insert into dept values(2,'d2'); commit; insert into emp values(3,'e3',2); insert into emp values(4,'e4',2); commit; select * from emp, dept where emp.deptno=dept.deptno; -- return to dt_1 time begin dbms_wm.gotoDate(to_date(:dt_1, 'dd.mm.yyyy hh24:mi:ss')); end;
Page 165
/ commit; select * from emp, dept where emp.deptno=dept.deptno; -- disable versioning begin dbms_wm.disableVersioning('DEPT'); dbms_wm.disableVersioning('EMP'); end; / /* Creating Workspaces and Merging/Refreshing Data */ -- privs begin dbms_wm.grantSystemPriv ( 'ACCESS_ANY_WORKSPACE, ' || 'MERGE_ANY_WORKSPACE, ' || 'CREATE_ANY_WORKSPACE, ' || 'REMOVE_ANY_WORKSPACE, ' || 'ROLLBACK_ANY_WORKSPACE' , 'USER1' , 'YES'); end; /
conn user1 create table dept ( DEPTNO NUMBER(3) PRIMARY KEY, DNAME VARCHAR2(14) ) / create table EMP ( EMPNO NUMBER(4) PRIMARY KEY, -- PK is mandatory ENAME VARCHAR2(10), DEPTNO NUMBER(3) references dept(deptno)) / insert into dept values (1,'d1'); commit; insert into emp values (1,'e1',1); insert into emp values (2,'e2',1); commit; -- set version enabled tables -- VIEW_WO_OVERWRITE = without Overwrite -- this option make WAREHOUSES_HIST view contain -- complete history information about data changes begin -- both table must be version enabled because they are linked dbms_wm.enableVersioning('DEPT,EMP','VIEW_WO_OVERWRITE'); end; / -- create workspaces Oracle DBA Code Examples
Page 166
begin dbms_wm.createworkspace ('WS1'); dbms_wm.createworkspace ('WS2'); end; / select workspace, parent_workspace from user_workspaces; -- move logged on user to WS1 begin dbms_wm.gotoworkspace('WS1'); end; / select * from emp, dept where emp.deptno=dept.deptno; -- insert extra rows insert into dept values(2,'d2'); commit; insert into emp values(3,'e3',2); insert into emp values(4,'e4',2); commit; select * from emp, dept where emp.deptno=dept.deptno; -- go back to Live workspace begin dbms_wm.gotoworkspace('LIVE'); end; / select * from emp, dept where emp.deptno=dept.deptno; -- change some data within the LIVE workspaces insert into dept values(3,'d3'); commit; insert into emp values(5,'e5',3); insert into emp values(6,'e6',3); commit;
select * from emp, dept where emp.deptno=dept.deptno; -- move to Workspace WS2 begin dbms_wm.gotoworkspace('WS2'); end; / -- you won't see changes made in the LIVE workspace nor the changes -- made in the WS1 workspace. select * from emp, dept where emp.deptno=dept.deptno; insert into dept values(4,'d3'); commit; insert into emp values(7,'e7',4); insert into emp values(8,'e8',4);
Page 167
commit; -- REFRESH WS1 begin dbms_wm.refreshworkspace('WS1'); end; / -- move to WS1 begin dbms_wm.gotoworkspace('WS1'); end; / select * from emp, dept where emp.deptno=dept.deptno; -- MERGE WS2 begin dbms_wm.mergeworkspace('WS2'); end; / -- verify merge succeeded begin dbms_wm.gotoworkspace('LIVE'); end; / select * from emp, dept where emp.deptno=dept.deptno;
-- cleaning up begin dbms_wm.disableversioning ('dept, emp', true); end; / begin dbms_wm.gotoworkspace ('LIVE'); dbms_wm.removeworkspace('WS1'); dbms_wm.removeworkspace('WS2'); end; / /* Resolving Conflicts */ -- privs begin dbms_wm.grantSystemPriv ( 'ACCESS_ANY_WORKSPACE, 'MERGE_ANY_WORKSPACE, 'CREATE_ANY_WORKSPACE, 'REMOVE_ANY_WORKSPACE, 'ROLLBACK_ANY_WORKSPACE' 'USER1' , 'YES'); end; /
' ' ' ' ,
|| || || ||
Page 168
conn user1 create table dept ( DEPTNO NUMBER(3) PRIMARY KEY, DNAME VARCHAR2(14) ) / create table EMP ( EMPNO NUMBER(4) PRIMARY KEY, -- PK is mandatory ENAME VARCHAR2(10), DEPTNO NUMBER(3) references dept(deptno)) / insert into insert into commit; insert into insert into insert into insert into commit; dept values (1,'d1'); dept values (2,'d2'); emp emp emp emp values values values values (1,'e1',1); (2,'e2',1); (3,'e3',2); (4,'e4',2);
-- set version enabled tables begin dbms_wm.enableVersioning('DEPT,EMP','VIEW_WO_OVERWRITE'); end; /
-- create workspace and goto it begin dbms_wm.createworkspace('WS1'); end; / begin dbms_wm.gotoworkspace('WS1'); end; / -- make an update update emp set ename='in WS1' where empno=1; commit; -- go back to LIVE workspace begin dbms_wm.gotoworkspace('LIVE'); end; / -- make an update update emp set ename='in LIVE' where empno=1; commit; -- to see the conflicts in xxx_CONF, you should leave LIVE workspace begin dbms_wm.gotoworkspace('WS1');
Page 169
end; / column wm_workspace format a16 select * from EMP_CONF;
/* To resolve the conflict */ -- (1) call dbms_wm.beginresolve begin dbms_wm.BeginResolve('WS1'); end; / -- (2) call dbms_wm.resolveconflicts begin -- Keep update from Parent (LIVE) dbms_wm.resolveconflicts( workspace => 'WS1', -- workspace to check for conflicts with others table_name => 'EMP', where_clause => 'empno=1', keep => 'PARENT'); end; / -- OR begin dbms_wm.resolveconflicts( workspace => 'WS1', table_name => 'EMP', where_clause => 'empno=1', keep => 'CHILD'); end; / commit; -- (3) call dbms_wm.commitresolve. begin dbms_wm.CommitResolve('WS1'); end; /
-- let's see if we still have a conflict: select * from emp_conf; -- check the data select * from emp ; -- let's see what data in LIVE begin dbms_wm.gotoworkspace('LIVE'); end; / select * from emp ; -- if data modified in WS1 was chosen, merge it with LIVE
Page 170
begin dbms_wm.mergeworkspace('WS1'); end; / -- Cleaning up begin dbms_wm.gotoworkspace('LIVE'); dbms_wm.removeworkspace('WS1'); dbms_wm.disableversioning('emp,dept'); end; /
Page 171
Repairing Corrupted Data

Options for Repairing Data Block Corruption
Drop and re-create an object after the corruption is detected. If data block corruption is limited to a subset of rows, then another option is to rebuild the table by selecting all data except for the corrupt rows. Use the DBMS_REPAIR package to detect and repair corrupt blocks in tables and indexes. Use Flashback Versions Query to query values of a row over a period of time. Then arrange undo plan. Block Media Recovery BMR
Detecting Corruptions Methods

DB_VERIFY utility ANALYZE TABLE .. VALIDATE STRUCTURE [CASCADE] Setting the initialization parameters: DB_BLOCK_CHECKING and DB_BLOCK_CHECKSUM exp utility DBMS_REPAIR Data Recovery Advisor automatically runs after any corruption is detected and can alos proactively invoked. See Using Data Recovery Advisor in RMAN.
Using dbv (DBVerify) Utility

The utility checks only for logical corruption below the HWM. dbv file=D:\ORACLE\ORADATA\ORA10G\USERS01.DBF dbv file=example01.dbf blocksize=8192 .. examine "Total Pages Marked Corrupt" blocksize=4096
Setting the Initialization Parameters for Detecting Corruption

Verifying Block Integrity in Real Time: DB_BLOCK_CHECKING Block checking typically causes 1% to 10% overhead, depending on workload. You should set DB_BLOCK_CHECKING to FULL if the performance overhead is acceptable. Possible Values: o o o o OFF: No block checking (except for SYSTEM) (default) LOW: Basic block header checks are performed after block contents change in memory MEDIUM: All LOW checks, as well as block checking for all non-index-organized table blocks, are performed. FULL: All LOW and MEDIUM checks, as well as checks on index blocks, are performed.
alter system set DB_BLOCK_CHECKING=FULL;

Page 172
alter session set DB_BLOCK_CHECKING=LOW; Verifying Block Integrity in Real Time: DB_BLOCK_CHECKSUM Can prevent corruption caused by underlying I/O systems FULL setting causes 4-5% overhead DB_BLOCK_CHECKSUM = { OFF | FALSE | TYPICAL | TRUE |FULL } select value from v$parameter where upper(name)='DB_BLOCK_CHECKSUM'; Detecting lost write: DB_LOST_WRITE_PROTECT A data block lost write occurs when an I/O subsystem acknowledges the completion of the block write, while in fact the write did not occur in the persistent storage. DB_LOST_WRITE_PROTECT = { NONE | TYPICAL | FULL } Settubg the DB_ULTRA_SAFE Parameter (In Oracle 11g)
This parameter is used to set the effective values of the parameters: DB_BLOCK_CHECKING, DB_LOST_WRITE_PROTECT, DB_BLOCK_CHECKSUM. This parameter takes one of the following values: off this value means any values you set for any of the three parameters will not be overridden. The effective value of the parameters will be as follows:
Parameter Active Value
data only
DB_BLOCK_CHECKING DB_LOST_WRITE_PROTECT DB_BLOCK_CHECKSUM. data and index
medium typical full
The effective value of the parameters will be as follows:

Parameter Active Value
DB_BLOCK_CHECKING DB_LOST_WRITE_PROTECT DB_BLOCK_CHECKSUM.
full typical full
Using ANALYZE Command

Validates the structure of a table or table partitions, and index or index partitions. Does not mark blocks as soft corrupt; only reports them When corruption detected, an error message returned. ANALYZE TABLE table_name VALIDATE STRUCTURE CASCADE;
Using EXP to Detect Corruption

exp utility reports an errors when it encounters corruption. $ exp hr/hr tables=departments About to export specified tables via Conventional Path ... . . . EXP-00056: ORACLE error 1578 encountered ORA-01578: ORACLE data block corrupted (file # 5, block # 51)
Page 173
ORA-01110: data file 5: '/u01/app/oracle/oradata/..'
Using DBMS_REPAIR
DBMS_REPAIR Limitations and Restrictions o o o o o DBMS_REPAIR procedures have the following limitations: Tables with LOB datatypes, nested tables, and varrays are supported, but the out of line columns are ignored. Clusters are supported in the SKIP_CORRUPT_BLOCKS and REBUILD_FREELISTS procedures, but not in the CHECK_OBJECT procedure. Index-organized tables and LOB indexes are not supported. The DUMP_ORPHAN_KEYS procedure does not operate on bitmap indexes or functionbased indexes. The DUMP_ORPHAN_KEYS procedure processes keys that are no more than 3,950bytes long.
Evaluate the Costs and Benefits of Using DBMS_REPAIR Before using DBMS_REPAIR you must weigh the benefits of its use in relation to the liabilities. You should also examine other options available for addressing corrupt objects. Begin by answering the following questions: What is the extent of the corruption? To determine if there are corruptions and repair actions, execute the CHECK_OBJECT procedure and query the repair table. What other options are available for addressing block corruptions? Consider the following: If the data is available from another source, then drop, re-create, and repopulate the object. Issue the CREATE TABLE...AS SELECT statement from the corrupt table to create a new one. Ignore the corruption by excluding corrupt rows from SELECT statements. Perform media recovery. What logical corruptions or side effects are introduced when you use DBMS_REPAIR to make an object usable? Can these be addressed? What is the effort required to do so?
It is possible that you do not have access to rows in blocks marked corrupt. However, a block can be marked corrupt even if there are rows that you can validly access. It is also possible that referential integrity constraints are broken when blocks are marked corrupt. If this occurs, then disable and reenable the constraint; any inconsistencies are reported. After fixing all problems, you should be able to reenable the constraint. Logical corruption can occur when there are triggers defined on the table. For example, if rows are reinserted, should insert triggers be fired or not? You can address these issues only if you understand triggers and their use in your installation. If indexes and tables are not synchronized, then execute the DUMP_ORPHAN_KEYS procedure to obtain information from the keys that might be useful in rebuilding corrupted data. Then issue the ALTER INDEX...REBUILD ONLINE statement to synchronize the table with its indexes. If repair involves loss of data, can this data be retrieved? You can retrieve data from the index when a data block is marked corrupt. The DUMP_ORPHAN_KEYS procedure can help you retrieve this information.
Page 174
Detect and Report Corruptions using DBMS_REPAIR -- Reports corruptions and identifies the associated repair directives -- 1. build the repair table BEGIN DBMS_REPAIR.ADMIN_TABLES ( table_name => 'REPAIR_TABLE', table_type => DBMS_REPAIR.REPAIR_TABLE, action => DBMS_REPAIR.CREATE_ACTION, tablespace => 'USERS'); END; / SELECT OBJECT_NAME, BLOCK_ID, CORRUPT_TYPE, MARKED_CORRUPT, CORRUPT_DESCRIPTION, REPAIR_DESCRIPTION FROM REPAIR_TABLE; -- 2. report corruptions SET SERVEROUTPUT ON DECLARE num_corrupt INT; BEGIN num_corrupt := 0; DBMS_REPAIR.CHECK_OBJECT ( schema_name => 'HR', object_name => 'DEPARTMENTS', repair_table_name => 'REPAIR_TABLE', corrupt_count => num_corrupt); END; / -- 3. make the object usable: ignoring corruptions during table and index scans -- mark corrupted blocks SET SERVEROUTPUT ON DECLARE num_fix INT; BEGIN num_fix := 0; DBMS_REPAIR.FIX_CORRUPT_BLOCKS ( schema_name => 'HR', object_name => 'DEPARTMENTS', object_type => DBMS_REPAIR.TABLE_OBJECT, repair_table_name => 'REPAIR_TABLE', fix_count => num_fix); END; / SELECT OBJECT_NAME, BLOCK_ID, MARKED_CORRUPT FROM REPAIR_TABLE; -- skip blocks marked as corrupted BEGIN DBMS_REPAIR.SKIP_CORRUPT_BLOCKS ( schema_name => 'HR', object_name => 'DEPARTMENTS', object_type => DBMS_REPAIR.TABLE_OBJECT, flags => DBMS_REPAIR.SKIP_FLAG); -- or NOSKIP_FLAG END; /
Page 175
select SKIP_CORRUPT from DBA_TABLES where table_name='DEPARTMENTS' and woner='HR'; -- 4. reports on index entries that point to rows in corrupt data blocks -- may be useful for rebuilding lost rows BEGIN DBMS_REPAIR.ADMIN_TABLES ( table_name => 'ORPHAN_KEY_TABLE', table_type => DBMS_REPAIR.ORPHAN_TABLE, action => DBMS_REPAIR.CREATE_ACTION, tablespace => 'USERS'); END; / SET SERVEROUTPUT ON DECLARE num_orphans INT; BEGIN num_orphans := 0; DBMS_REPAIR.DUMP_ORPHAN_KEYS( schema_name => 'SCOTT', object_name => 'PK_DEPT', object_type => DBMS_REPAIR.INDEX_OBJECT, repair_table_name => 'REPAIR_TABLE', orphan_table_name => 'ORPHAN_KEY_TABLE', key_count => num_orphans); DBMS_OUTPUT.PUT_LINE('Orphan Key Count: ' ||TO_CHAR(num_orphans)); END; /
Page 176
Managing Automated Database Maintenance Tasks

Predefined Automated Maintenance Tasks
Automatic Optimizer Statistics Collection Automatic Segment Advisor Automatic SQL Tuning Advisor
Predefined Maintenance Windows

Window Name MONDAY_WINDOW TUESDAY_WINDOW WEDNESDAY_WINDOW THURSDAY_WINDOW FRIDAY_WINDOW SATURDAY_WINDOW SUNDAY_WINDOW Description Starts at 10 p.m. on Monday and ends at 2 a.m. Starts at 10 p.m. on Tuesday and ends at 2 a.m. Starts at 10 p.m. on Wednesday and ends at 2 a.m. Starts at 10 p.m. on Thursday and ends at 2 a.m. Starts at 10 p.m. on Friday and ends at 2 a.m. Starts at 6 a.m. on Saturday and is 20 hours long. Starts at 6 a.m. on Sunday and is 20 hours long.
Obtaining Information about Predefined Maintenance Tasks

-- currently running Scheduler jobs created for automated maintenance tasks select CLIENT_NAME,JOB_NAME,JOB_SCHEDULER_STATUS,TASK_NAME,TASK_TARGET_TYPE,TASK_TARG ET_NAME,TASK_PRIORITY,TASK_OPERATION From DBA_AUTOTASK_CLIENT_JOB ; -- automated maintenance task statistics over 7-day and 30-day periods select CLIENT_NAME,STATUS,CONSUMER_GROUP,CLIENT_TAG,PRIORITY_OVERRIDE,ATTRIBUTES,WIND OW_GROUP,SERVICE_NAME,RESOURCE_PERCENTAGE,USE_RESOURCE_ESTIMATES,MEAN_JOB_DURA TION,MEAN_JOB_CPU,MEAN_JOB_ATTEMPTS,MEAN_INCOMING_TASKS_7_DAYS,MEAN_INCOMING_T ASKS_30_DAYS,TOTAL_CPU_LAST_7_DAYS,TOTAL_CPU_LAST_30_DAYS,MAX_DURATION_LAST_7_ DAYS,MAX_DURATION_LAST_30_DAYS,WINDOW_DURATION_LAST_7_DAYS,WINDOW_DURATION_LAS T_30_DAYS From DBA_AUTOTASK_CLIENT -- history of automated maintenance task job runs select CLIENT_NAME,WINDOW_NAME,WINDOW_START_TIME,WINDOW_DURATION,JOB_NAME,JOB_STATUS, JOB_START_TIME,JOB_DURATION,JOB_ERROR,JOB_INFO from DBA_AUTOTASK_JOB_HISTORY order by WINDOW_START_TIME desc; -- Lists the windows that belong to MAINTENANCE_WINDOW_GROUP select WINDOW_NAME,WINDOW_NEXT_TIME,WINDOW_ACTIVE,AUTOTASK_STATUS,OPTIMIZER_STATS,SEG MENT_ADVISOR,SQL_TUNE_ADVISOR,HEALTH_MONITOR from DBA_AUTOTASK_WINDOW_CLIENTS;
Page 177
-- per-window history of job execution counts for each -- automated maintenance task select CLIENT_NAME,WINDOW_NAME,WINDOW_START_TIME,WINDOW_DURATION,JOBS_CREATED,JOBS_ST ARTED,JOBS_COMPLETED,WINDOW_END_TIME from DBA_AUTOTASK_CLIENT_HISTORY;
Enabling and Disabling Maintenance Tasks

select CLIENT_NAME,STATUS,CONSUMER_GROUP,CLIENT_TAG,PRIORITY_OVERRIDE, ATTRIBUTES,WINDOW_GROUP,SERVICE_NAME,RESOURCE_PERCENTAGE, USE_RESOURCE_ESTIMATES,MEAN_JOB_DURATION,MEAN_JOB_CPU,MEAN_JOB_ATTEMPTS, MEAN_INCOMING_TASKS_7_DAYS,MEAN_INCOMING_TASKS_30_DAYS,TOTAL_CPU_LAST_7_DAYS, TOTAL_CPU_LAST_30_DAYS,MAX_DURATION_LAST_7_DAYS,MAX_DURATION_LAST_30_DAYS, WINDOW_DURATION_LAST_7_DAYS,WINDOW_DURATION_LAST_30_DAYS from DBA_AUTOTASK_CLIENT; -- disable BEGIN DBMS_AUTO_TASK_ADMIN.DISABLE( client_name => 'sql tuning advisor', operation => NULL, window_name => NULL); -- null means all widows END; / -- enable BEGIN DBMS_AUTO_TASK_ADMIN.ENABLE( client_name => 'sql tuning advisor', operation => NULL, window_name => NULL); END; / -- enable or disable ALL automated maintenance tasks for ALL windows EXECUTE DBMS_AUTO_TASK_ADMIN.DISABLE; EXECUTE DBMS_AUTO_TASK_ADMIN.ENABLE; -- Enabling and Disabling Maintenance Tasks for Specific Maintenance Windows BEGIN DBMS_AUTO_TASK_ADMIN.DISABLE( client_name => 'sql tuning advisor', operation => NULL, window_name => 'MONDAY_WINDOW'); END; /
Configuring Maintenance Windows

-- Modifying a Maintenance Window -- see Using Windows SECTION BEGIN dbms_scheduler.disable( name => 'SATURDAY_WINDOW');
Page 178
dbms_scheduler.set_attribute( name => 'SATURDAY_WINDOW', attribute => 'DURATION', value => numtodsinterval(4, 'hour')); dbms_scheduler.enable( name => 'SATURDAY_WINDOW'); END; / -- Creating a New Maintenance Window BEGIN DBMS_SCHEDULER.CREATE_WINDOW( window_name => 'EARLY_MORNING_WINDOW', duration => numtodsinterval(1, 'hour'), resource_plan => 'DEFAULT_MAINTENANCE_PLAN', repeat_interval => 'FREQ=DAILY;BYHOUR=5;BYMINUTE=0;BYSECOND=0'); dbms_scheduler.add_window_group_member( group_name => 'MAINTENANCE_WINDOW_GROUP', window_list => 'EARLY_MORNING_WINDOW'); END; /
-- Removing a Maintenance Window -- window continues to exist but no longer runs automated maintenance tasks. -- Any other Scheduler jobs assigned to this window continue to run as usual. BEGIN DBMS_SCHEDULER.REMOVE_WINDOW_GROUP_MEMBER( group_name => 'MAINTENANCE_WINDOW_GROUP', window_list => 'EARLY_MORNING_WINDOW'); END; /
Page 179
Managing Resources
The steps: 1. Create a pending area. 2. Create a resource consumer group. 3. Create a resource plan. 4. Create a plan directive. 5. Validate the pending area. 6. Submit the pending area -- required privs exec DBMS_RESOURCE_MANAGER_PRIVS.GRANT_SYSTEM_PRIVILEGE (GRANTEE_NAME => 'scott', PRIVILEGE_NAME => 'ADMINISTER_RESOURCE_MANAGER'); -- creating demo users conn system create user u1 identified by u1 default tablespace users; create user u2 identified by u2 default tablespace users; create user u3 identified by u3 default tablespace users; grant resource, connect to u1; grant resource, connect to u2; grant resource, connect to u3; CREATE TABLE U1.EMP AS SELECT * FROM HR.EMPLOYEES; CREATE TABLE U2.EMP AS SELECT * FROM HR.EMPLOYEES; CREATE TABLE U3.EMP AS SELECT * FROM HR.EMPLOYEES;
-- Creating a Pending Area exec dbms_resource_manager.create_pending_area(); -- if you make mistakes exec dbms_resource_manager.clear_pending_area; -- Resource Consumer Groups -- pre-configured CGs: -DEFAULT_CONSUMER_GROUP,OTHER_GROUPS,SYS_GROUP,LOW_GROUP begin dbms_resource_manager.create_consumer_group ( CONSUMER_GROUP => 'PROGRAMMERS', COMMENT => 'IT programmers', CPU_MTH => 'ROUND-ROBIN' -- other possible value is RUN_TO_COMPLETION ); end; / begin dbms_resource_manager.create_consumer_group ( CONSUMER_GROUP => 'CLERKS', COMMENT => 'Group of Clerks', CPU_MTH => 'ROUND-ROBIN' );
Page 180
end; /
set linesize 100 column consumer_group format a15 column comments format a40 column cpu_method format a11 select DISTINCT consumer_group, cpu_method, comments from dba_rsrc_consumer_groups order by 1 / -- Consumer Group Mapping -- Assigning User Sessions to Consumer Groups begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => dbms_resource_manager.oracle_user, -- it is constant (no quots) VALUE => 'U1', CONSUMER_GROUP => 'PROGRAMMERS'); end; / begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => dbms_resource_manager.oracle_user, -- it is constant (no quots) VALUE => 'U2', CONSUMER_GROUP => 'CLERKS'); end; / -- possible attributes are: CLIENT_OS_USER CLIENT_PROGRAM CLIENT_MACHINE MODULE_NAME MODULE_NAME_ACTION ORACLE_USER SERVICE_NAME SERVICE_MODULE -- /* for demo only begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => CLIENT_OS_USER, VALUE => 'graciej', CONSUMER_GROUP => 'OLAP_GROUP'); end; / */
-- Establishing Mapping Priorities begin dbms_resource_manager.SET_CONSUMER_GROUP_MAPPING_PRI(
Page 181
EXPLICIT => 1, CLIENT_OS_USER => 5, CLIENT_MACHINE => 3, CLIENT_PROGRAM => 4, ORACLE_USER => 2, MODULE_NAME => 6, MODULE_NAME_ACTION => 7, SERVICE_NAME => 8, SERVICE_MODULE => 9, SERVICE_MODULE_ACTION => 10); end; / -- ERROR BEGIN DBMS_RESOURCE_MANAGER.SET_MAPPING_PRIORITY( EXPLICIT => 1, CLIENT_MACHINE => 2, MODULE_NAME => 3, ORACLE_USER => 4, SERVICE_NAME => 5, CLIENT_OS_USER => 6, CLIENT_PROGRAM => 7, MODULE_NAME_ACTION => 8, SERVICE_MODULE=>9, SERVICE_MODULE_ACTION=>10); END; /
-- Resource Plans -- Simple Resource Plan (single-level resource plans for allocating CPU) /* 10g (deprecated in 11g) */ BEGIN DBMS_RESOURCE_MANAGER.CREATE_SIMPLE_PLAN( SIMPLE_PLAN => 'JOBS_PLAN', CONSUMER_GROUP1 => 'PROGRAMMERS', GROUP1_CPU => 75, CONSUMER_GROUP2 => 'CLERKS', GROUP2_CPU => 25); end; / -- Note: the final output is still multi-level begin DBMS_RESOURCE_MANAGER.DELETE_PLAN ('JOBS_PLAN'); end; / /* 11g : to implement the following plan: CG Level1 Level2 Level 3 SYS_GROUP 100% MYGROUP1 80% MYGROUP2 20% OTHER_GROUPS 100% */ BEGIN DBMS_RESOURCE_MANAGER.CREATE_SIMPLE_PLAN(SIMPLE_PLAN => 'SIMPLE_PLAN1', CONSUMER_GROUP1 => 'MYGROUP1', GROUP1_PERCENT => 80, CONSUMER_GROUP2 => 'MYGROUP2', GROUP2_PERCENT => 20); END; -- ***************** /* Complex Plan: CG, Resource P., Res. Plan Dir. */ -- create resource plan (10g)
Page 182
begin dbms_resource_manager.create_plan( PLAN => 'NW_PLAN', COMMENT => 'New World Plan', CPU_MTH => 'EMPHASIS'); -- or RATIO end; / -- create resource plan 11g -- MGMT_MTH for specifying how much CPU each consumer group or subplan gets. -- 'EMPHASIS', the default method, is for single-level or multilevel plans -that use percentages. 'RATIO' is for single-level plans that use ratios BEGIN DBMS_RESOURCE_MANAGER.CREATE_PLAN( PLAN => 'NW_PLAN', COMMENT => 'New World Plan', MGMT_MTH => 'EMPHASIS', SUB_PLAN => FALSE -- If TRUE, the plan cannot be used as the top plan ); END; / select DISTINCT plan, num_plan_directives, cpu_method from dba_rsrc_plans;
-- Resource Plan Directives begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'New World Directive', GROUP_OR_SUBPLAN => 'PROGRAMMERS', MAX_IDLE_TIME => '120'); end; / -- other parameters are CPU_Pn where n from 1 to 8 (10g) MGMT_Pn where n from 1 to 8 (11g) ACTIVE_SESS_POOL_P1 QUEUEING_P1 PARALLEL_DEGREE_LIMIT_P1 SWITCH_GROUP -- CG to switch to, also accept 'CANCEL_SQL' 'KILL_SESSION' SWITCH_TIME -- time (in CPU seconds) for a call before SWITCH_GROUP is taken SWITCH_ESTIMATE -- TRUE or FALSE. MAX_EST_EXEC_TIME UNDO_POOL -- in KB MAX_IDLE_TIME MAX_IDLE_BLOCKER_TIME -- in seconds SWITCH_TIME_IN_CALL -- (10g) SWITCH_FOR_CALL -- (11g) SWITCH_IO_MEGABYTES -- (11g) max megabytes of I/O by a session SWITCH_IO_REQS -- (11g) max number of I/O requests SWITCH_FOR_CALL -- (11g) If TRUE, a session that was automatically switched -- to another consumer group (according to SWITCH_TIME, -- SWITCH_IO_MEGABYTES, or SWITCH_IO_REQS) is returned to -- its original consumer group when the top level call
Page 183
-- completes. -- examples in 11g BEGIN DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE ( PLAN => 'DAYTIME', GROUP_OR_SUBPLAN => 'REPORTING', COMMENT => 'Reporting group', MGMT_P1 => 15, PARALLEL_DEGREE_LIMIT_P1 => 8, ACTIVE_SESS_POOL_P1 => 4); DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE ( PLAN => 'DAYTIME', GROUP_OR_SUBPLAN => 'OTHER_GROUPS', COMMENT => 'This one is required', MGMT_P1 => 10); END; / -- Creating Multi-Level Plan Directives -- 10g -- 11g: replace CPU_Pn with MGMT_Pn begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'SYSTEM USERS', GROUP_OR_SUBPLAN => 'SYS_GROUP', CPU_P1=> 100); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'Clerks CPU quota', GROUP_OR_SUBPLAN => 'CLERKS', CPU_P2=> 30); end; / BEGIN dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'Progs CPU quota', GROUP_OR_SUBPLAN => 'PROGRAMMERS', CPU_P2=> 70); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'OTHER_GROUPS CPU ALLOCATION', GROUP_OR_SUBPLAN => 'OTHER_GROUPS', CPU_P3=> 100); end; /
Page 184
-- Creating Automatic Consumer Group Switching Directives begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'LIMIT CLERKS EXECUTION TIME', GROUP_OR_SUBPLAN => 'CLERKS', SWITCH_GROUP => 'CANCEL_SQL', SWITCH_TIME => 3600); -- one hour end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'NW_PLAN', COMMENT => 'SWITCH PROGRAMMERS TEMPORARILY', GROUP_OR_SUBPLAN => 'PROGRAMMERS', SWITCH_TIME_IN_CALL => 900, SWITCH_GROUP => 'LOW_GROUP', SWITCH_ESTIMATE => TRUE); end; / -- Updating Resource Plan Directives begin dbms_resource_manager.update_plan_directive( PLAN => 'NW_PLAN', GROUP_OR_SUBPLAN => 'PROGRAMMERS', NEW_SWITCH_ESTIMATE => FALSE); end; / -- Deleting Resource Plan Directives begin dbms_resource_manager.delete_plan_directive( PLAN => 'NW_PLAN', GROUP_OR_SUBPLAN => 'PROGRAMMERS'); end; /
-- validating Pending Area exec dbms_resource_manager.validate_pending_area; -- submitting Pending Area exec dbms_resource_manager.submit_pending_area;
-- Enabling the Database Resource Manager ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = NW_PLAN; -- Switching Resource Consumer Groups -- for a session begin dbms_resource_manager.switch_consumer_group_for_sess ( SESSION_ID => '145',
Page 185
SESSION_SERIAL=> '49', CONSUMER_GROUP => 'SYS_GROUP'); end; / -- for all sessions of a specific user begin dbms_resource_manager.switch_consumer_group_for_user ( USER => 'HR', CONSUMER_GROUP => 'LOW_GROUP'); end; /
-- switch C Groups by client -- require priv begin dbms_resource_manager_privs.grant_switch_consumer_group( GRANTEE_NAME => 'U1', -- a role name can be given as well CONSUMER_GROUP => 'PROGRAMMERS', GRANT_OPTION => FALSE); end; / -- Enabling Users or Applications to Manually Switch Consumer Groups DECLARE original_group varchar2(30); dummy varchar2(30); BEGIN DBMS_SESSION.SWITCH_CURRENT_CONSUMER_GROUP( 'MARKETING', original_group, FALSE); -- execute some SQL null; DBMS_SESSION.SWITCH_CURRENT_CONSUMER_GROUP( original_group, dummy, FALSE); END; / -- Granting and Revoking the Switch Privilege BEGIN -- GRANT_.. or REVOKE_.. DBMS_RESOURCE_MANAGER_PRIVS.GRANT_SWITCH_CONSUMER_GROUP( GRANTEE_NAME => 'SCOTT', CONSUMER_GROUP => 'OLTP', GRANT_OPTION => false); END; / -- Enabling Oracle Database Resource Manager and Switching Plans ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = 'mydb_plan'; -- the resource plan can be changed only by setting -- the RESOURCE_MANAGER_PLAN initialization parameter. -- It disallows changes by Window opening. -- This restriction can be lifted by rerunning the command without FORCE ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = 'FORCE:mydb_plan'; -- disable the Resource Manager ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = '';
Page 186
-- **************** /* Putting it all together Example implementation OFF_HOURS_PLAN SYS_GROUP NIGHT_GROUP REPORTS_GROUP OTHER_GROUPS
*/
10 5 2 1
DAY_PLAN LEVEL1 SYS_GROUP 100% LEVEL2 OLTP_PLAN 100% LEVEL1 OLTP_GROUP LEVEL2 OTHERS_GROUPS LEVEL3 OTHERS_GROUPS 100%
90% 100%
DREPORTS_GROUP
10%
exec dbms_resource_manager.create_pending_area(); begin dbms_resource_manager.create_consumer_group('OLTP_GROUP','Incoming orders'); end; / begin dbms_resource_manager.create_consumer_group('DREPORTS_GROUP','DAYTIME REPORTS'); end; / begin dbms_resource_manager.create_consumer_group('NIGHT_GROUP','BULK LOADS'); end; / begin dbms_resource_manager.create_consumer_group('REPORTS_GROUP','OFF HOURS REPORTS'); end; / -- Creating the Resource Plans begin dbms_resource_manager.create_plan( PLAN => 'DAY_PLAN', COMMENT => 'NORMAL WORKING HOURS '); end; / begin dbms_resource_manager.create_plan (PLAN => 'OLTP_PLAN', COMMENT => 'ORDER ENTRY SUB-PLAN'); end; /
Page 187
begin dbms_resource_manager.create_plan( PLAN => 'OFF_HOURS_PLAN', COMMENT => 'GOVERNS NON-WORKING HOURS', CPU_MTH => 'RATIO'); end; / -- PLAN DIRECTIVE begin dbms_resource_manager.create_plan_directive( PLAN => 'OFF_HOURS_PLAN', GROUP_OR_SUBPLAN => 'SYS_GROUP', COMMENT => 'CPU ALLOCATION FOR SYS_GROUP', CPU_P1 => 10); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'OFF_HOURS_PLAN', GROUP_OR_SUBPLAN => 'NIGHT_GROUP', COMMENT => 'CPU ALLOCATION FOR NIGHTLY JOBS', CPU_P1 => 5); end; /
begin dbms_resource_manager.create_plan_directive( PLAN => 'OFF_HOURS_PLAN', GROUP_OR_SUBPLAN => 'REPORTS_GROUP', COMMENT => 'CPU ALLOCATION FOR NIGHTLY REPORTS', CPU_P1 => 2); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'OFF_HOURS_PLAN', GROUP_OR_SUBPLAN => 'OTHER_GROUPS', COMMENT => 'CPU ALLOCATION FOR OTHER_GROUPS', CPU_P1 => 1); end; /
-- Creating the OLTP_PLAN Plan Directives begin dbms_resource_manager.create_plan_directive( PLAN => 'OLTP_PLAN', GROUP_OR_SUBPLAN => 'OLTP_GROUP', COMMENT => 'CPU ALLOCATION FOR OLTP USERS', CPU_P1 => 90); end; /
Page 188
begin dbms_resource_manager.create_plan_directive( PLAN => 'OLTP_PLAN', GROUP_OR_SUBPLAN => 'DREPORTS_GROUP', COMMENT => 'CPU ALLOCATION FOR DAYTIME REPORTING', CPU_P1 => 10); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'OLTP_PLAN', GROUP_OR_SUBPLAN => 'OTHER_GROUPS', COMMENT => 'CPU ALLOCATION FOR OTHER_GROUPS', CPU_P2 => 100); end; /
-- DAY_PLAN Plan Directives begin dbms_resource_manager.create_plan_directive( PLAN => 'DAY_PLAN', GROUP_OR_SUBPLAN => 'SYS_GROUP', COMMENT => 'CPU ALLOCATION FOR SYS_GROUP', CPU_P1 => 100); end; /
begin dbms_resource_manager.create_plan_directive( PLAN => 'DAY_PLAN', GROUP_OR_SUBPLAN => 'OLTP_PLAN', COMMENT => 'CPU ALLOCATION FOR OLTP_PLAN SUB-PLAN', CPU_P2 => 100); end; / begin dbms_resource_manager.create_plan_directive( PLAN => 'DAY_PLAN', GROUP_OR_SUBPLAN => 'OTHER_GROUPS', COMMENT => 'CPU ALLOCATION FOR OTHER_GROUPS', CPU_P3 => 100); end; / exec dbms_resource_manager.validate_pending_area; exec dbms_resource_manager.submit_pending_area;
-- Enabling the Resource Plans ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = 'DAY_PLAN'; ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = 'OFF_HOURS_PLAN'; ALTER SYSTEM SET RESOURCE_MANAGER_PLAN = '';
Page 189
-- testing exec dbms_resource_manager.create_pending_area(); begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => dbms_resource_manager.oracle_user, VALUE => 'U1', CONSUMER_GROUP => 'NIGHT_GROUP'); end; / begin dbms_resource_manager_privs.grant_switch_consumer_group( GRANTEE_NAME => 'U1', -- a role name can be given as well CONSUMER_GROUP => 'NIGHT_GROUP', GRANT_OPTION => FALSE); end; / begin dbms_resource_manager_privs.grant_switch_consumer_group( GRANTEE_NAME => 'U2', CONSUMER_GROUP => 'REPORTS_GROUP', GRANT_OPTION => FALSE); end; / begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => dbms_resource_manager.oracle_user, -- it is constant (no quots) VALUE => 'U2', CONSUMER_GROUP => 'REPORTS_GROUP'); end; / begin dbms_resource_manager_privs.grant_switch_consumer_group( GRANTEE_NAME => 'U3', CONSUMER_GROUP => 'OLTP_GROUP', GRANT_OPTION => FALSE); end; / begin dbms_resource_manager.set_consumer_group_mapping( ATTRIBUTE => dbms_resource_manager.oracle_user, VALUE => 'U3', CONSUMER_GROUP => 'OLTP_GROUP'); end; / exec dbms_resource_manager.submit_pending_area; SET SERVEROUTPUT ON
Page 190
DECLARE D DATE := SYSDATE; DUMMY NUMBER; BEGIN DBMS_OUTPUT.PUT_LINE('Start Time:'|| to_char(D,'hh24:mi:ss')); FOR I IN 1..3000000 LOOP DUMMY := SQRT(DBMS_RANDOM.VALUE(1,8100)); END LOOP; DBMS_OUTPUT.PUT_LINE('End Time:'|| to_char(sysdate,'hh24:mi:ss')); DBMS_OUTPUT.PUT_LINE('Elapsed :'|| round((sysdate-d)*24*60*60,2) || ' s' ); END; /
Obtaining Information on Database Resource Manager

-- resource consumer groups that exist in the database select CONSUMER_GROUP_ID,CONSUMER_GROUP,CPU_METHOD,MGMT_METHOD,INTERNAL_USE,COMMENTS ,CATEGORY,STATUS,MANDATORY from DBA_RSRC_CONSUMER_GROUPS; -- resource directives select PLAN,GROUP_OR_SUBPLAN,TYPE,CPU_P1,CPU_P2,CPU_P3,CPU_P4,CPU_P5,CPU_P6,CPU_P7,C PU_P8,MGMT_P1,MGMT_P2,MGMT_P3,MGMT_P4,MGMT_P5,MGMT_P6,MGMT_P7,MGMT_P8,ACTIVE_ SESS_POOL_P1,QUEUEING_P1,PARALLEL_DEGREE_LIMIT_P1,SWITCH_GROUP,SWITCH_FOR_CAL L,SWITCH_TIME,SWITCH_IO_MEGABYTES,SWITCH_IO_REQS,SWITCH_ESTIMATE,MAX_EST_EXEC _TIME,UNDO_POOL,MAX_IDLE_TIME,MAX_IDLE_BLOCKER_TIME,SWITCH_TIME_IN_CALL,COMME NTS,STATUS,MANDATORY from DBA_RSRC_PLAN_DIRECTIVES; -- the groups to which a user or role is allowed to belong or be switched select GRANTEE,GRANTED_GROUP,GRANT_OPTION,INITIAL_GROUP from DBA_RSRC_CONSUMER_GROUP_PRIVS -- Viewing Plan Information select PLAN_ID,PLAN,NUM_PLAN_DIRECTIVES,CPU_METHOD,MGMT_METHOD,ACTIVE_SESS_POOL_MTH, PARALLEL_DEGREE_LIMIT_MTH,QUEUEING_MTH,SUB_PLAN,COMMENTS,STATUS,MANDATORY from DBA_RSRC_PLANS -- Viewing Current Consumer Groups for Sessions SELECT SID,SERIAL#,USERNAME,RESOURCE_CONSUMER_GROUP FROM V$SESSION; -- display the current top level plan and all of its descendent subplans SELECT NAME, IS_TOP_PLAN FROM V$RSRC_PLAN; -- mapping pairs for all of the session attributes select ATTRIBUTE,VALUE,CONSUMER_GROUP,STATUS from DBA_RSRC_GROUP_MAPPINGS; -- current mapping priority of each attribute select ATTRIBUTE,PRIORITY,STATUS from DBA_RSRC_MAPPING_PRIORITY; -- historical information on resource plan activation: snapshots of V$RSRC_PLAN_HISTORY
Page 191
select SNAP_ID,DBID,INSTANCE_NUMBER,SEQUENCE#,START_TIME,END_TIME,PLAN_ID,PLAN_NAME, CPU_MANAGED from DBA_HIST_RSRC_PLAN; -- historical statistical information on consumer groups. snapshots of: V$RSRC_CONS_GROUP_HISTORY select SNAP_ID,DBID,INSTANCE_NUMBER,SEQUENCE#,CONSUMER_GROUP_ID,CONSUMER_GROUP_NAME, REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CONSUMED_CPU_TIME,YIELDS,ACTIVE_SESS_LIMIT_H IT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIME,SWITCHES_IN_IO_M EGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS,SWITCHES_OUT_IO_RE QUESTS,SQL_CANCELED,ACTIVE_SESS_KILLED,IDLE_SESS_KILLED,IDLE_BLKR_SESS_KILLED ,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAITS,SMALL_READ_MEGA BYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRITE_MEGABYTES,SMALL_ READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARGE_WRITE_REQUESTS from DBA_HIST_RSRC_CONSUMER_GROUP ; -- for each consumer group in the plan showing the cumulative statistics for the consumer group. -- sequence# as as in V$RSRC_PLAN_HISTORY select SEQUENCE#,ID,NAME,REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CONSUMED_CPU_TIME,YIELDS,A CTIVE_SESS_LIMIT_HIT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIM E,SWITCHES_IN_IO_MEGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS, SWITCHES_OUT_IO_REQUESTS,SQL_CANCELED,ACTIVE_SESS_KILLED,IDLE_SESS_KILLED,IDL E_BLKR_SESS_KILLED,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAI TS,SMALL_READ_MEGABYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRIT E_MEGABYTES,SMALL_READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARG E_WRITE_REQUESTS from V$RSRC_CONS_GROUP_HISTORY; select SEQUENCE#,ID,NAME,REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CONSUMED_CPU_TIME,YIELDS,A CTIVE_SESS_LIMIT_HIT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIM E,SWITCHES_IN_IO_MEGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS, SWITCHES_OUT_IO_REQUESTS,SQL_CANCELED,ACTIVE_SESS_KILLED,IDLE_SESS_KILLED,IDL E_BLKR_SESS_KILLED,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAI TS,SMALL_READ_MEGABYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRIT E_MEGABYTES,SMALL_READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARG E_WRITE_REQUESTS from V$RSRC_CONS_GROUP_HISTORY; -- resource usage and stats data on currently active resource consumer groups select ID,NAME,ACTIVE_SESSIONS,EXECUTION_WAITERS,REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CO NSUMED_CPU_TIME,YIELDS,QUEUE_LENGTH,CURRENT_UNDO_CONSUMPTION,ACTIVE_SESSION_L IMIT_HIT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIME,SWITCHES_I N_IO_MEGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS,SWITCHES_OUT _IO_REQUESTS,SQL_CANCELED,ACTIVE_SESSIONS_KILLED,IDLE_SESSIONS_KILLED,IDLE_BL KR_SESSIONS_KILLED,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAI TS,SMALL_READ_MEGABYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRIT E_MEGABYTES,SMALL_READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARG E_WRITE_REQUESTS,V$RSRC_CONSUMER_GROUP FROM v$rsrc_consumer_group; -- a history of resources consumed and cumulative CPU wait time (due to resource management) per consumer group for the past minute select BEGIN_TIME,END_TIME,INTSIZE_CSEC,SEQUENCE#,CONSUMER_GROUP_ID,CONSUMER_GROUP_N AME,CPU_CONSUMED_TIME,CPU_WAIT_TIME,IO_REQUESTS,IO_MEGABYTES
Page 192
from V$RSRCMGRMETRIC; -- when resource plans were enabled or disabled on the instance SELECT sequence# seq, name plan_name, to_char(start_time, 'DD-MON-YY HH24:MM') start_time, to_char(end_time, 'DD-MON-YY HH24:MM') end_time, window_name FROM v$rsrc_plan_history; -- how the session has been affected by the Resource Manager select SID,CURRENT_CONSUMER_GROUP_ID,ORIG_CONSUMER_GROUP_ID,MAPPING_ATTRIBUTE,MAPPED _CONSUMER_GROUP,STATE,ACTIVE,CURRENT_IDLE_TIME,CURRENT_CPU_WAIT_TIME,CPU_WAIT _TIME,CURRENT_CPU_WAITS,CPU_WAITS,CURRENT_CONSUMED_CPU_TIME,CONSUMED_CPU_TIME ,CURRENT_ACTIVE_TIME,ACTIVE_TIME,CURRENT_QUEUED_TIME,QUEUED_TIME,CURRENT_YIEL DS,YIELDS,CURRENT_UNDO_CONSUMPTION,MAX_UNDO_CONSUMPTION,SQL_CANCELED,QUEUE_TI ME_OUTS,ESTIMATED_EXECUTION_LIMIT_HIT,CURRENT_IO_SERVICE_TIME,IO_SERVICE_TIME ,CURRENT_IO_SERVICE_WAITS,IO_SERVICE_WAITS,CURRENT_SMALL_READ_MEGABYTES,SMALL _READ_MEGABYTES,CURRENT_LARGE_READ_MEGABYTES,LARGE_READ_MEGABYTES,CURRENT_SMA LL_WRITE_MEGABYTES,SMALL_WRITE_MEGABYTES,CURRENT_LARGE_WRITE_MEGABYTES,LARGE_ WRITE_MEGABYTES,CURRENT_SMALL_READ_REQUESTS,SMALL_READ_REQUESTS,CURRENT_SMALL _WRITE_REQUESTS,SMALL_WRITE_REQUESTS,CURRENT_LARGE_READ_REQUESTS,LARGE_READ_R EQUESTS,CURRENT_LARGE_WRITE_REQUESTS,LARGE_WRITE_REQUESTS from v$rsrc_session_info;
Monitoring Oracle Database Resource Manager

-- display the currently active resource plan and its subplans. SELECT name, is_top_plan FROM v$rsrc_plan; -- resource usage and stats data on currently active resource consumer groups select ID,NAME,ACTIVE_SESSIONS,EXECUTION_WAITERS,REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CO NSUMED_CPU_TIME,YIELDS,QUEUE_LENGTH,CURRENT_UNDO_CONSUMPTION,ACTIVE_SESSION_L IMIT_HIT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIME,SWITCHES_I N_IO_MEGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS,SWITCHES_OUT _IO_REQUESTS,SQL_CANCELED,ACTIVE_SESSIONS_KILLED,IDLE_SESSIONS_KILLED,IDLE_BL KR_SESSIONS_KILLED,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAI TS,SMALL_READ_MEGABYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRIT E_MEGABYTES,SMALL_READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARG E_WRITE_REQUESTS,V$RSRC_CONSUMER_GROUP FROM v$rsrc_consumer_group; -- how the session has been affected by the Resource Manager select SID,CURRENT_CONSUMER_GROUP_ID,ORIG_CONSUMER_GROUP_ID,MAPPING_ATTRIBUTE,MAPPED _CONSUMER_GROUP,STATE,ACTIVE,CURRENT_IDLE_TIME,CURRENT_CPU_WAIT_TIME,CPU_WAIT _TIME,CURRENT_CPU_WAITS,CPU_WAITS,CURRENT_CONSUMED_CPU_TIME,CONSUMED_CPU_TIME ,CURRENT_ACTIVE_TIME,ACTIVE_TIME,CURRENT_QUEUED_TIME,QUEUED_TIME,CURRENT_YIEL DS,YIELDS,CURRENT_UNDO_CONSUMPTION,MAX_UNDO_CONSUMPTION,SQL_CANCELED,QUEUE_TI ME_OUTS,ESTIMATED_EXECUTION_LIMIT_HIT,CURRENT_IO_SERVICE_TIME,IO_SERVICE_TIME ,CURRENT_IO_SERVICE_WAITS,IO_SERVICE_WAITS,CURRENT_SMALL_READ_MEGABYTES,SMALL _READ_MEGABYTES,CURRENT_LARGE_READ_MEGABYTES,LARGE_READ_MEGABYTES,CURRENT_SMA LL_WRITE_MEGABYTES,SMALL_WRITE_MEGABYTES,CURRENT_LARGE_WRITE_MEGABYTES,LARGE_ WRITE_MEGABYTES,CURRENT_SMALL_READ_REQUESTS,SMALL_READ_REQUESTS,CURRENT_SMALL _WRITE_REQUESTS,SMALL_WRITE_REQUESTS,CURRENT_LARGE_READ_REQUESTS,LARGE_READ_R EQUESTS,CURRENT_LARGE_WRITE_REQUESTS,LARGE_WRITE_REQUESTS from v$rsrc_session_info; SELECT se.sid sess_id, co.name consumer_group,
Page 193
se.state, se.consumed_cpu_time cpu_time, se.cpu_wait_time, se.queued_time FROM v$rsrc_session_info se, v$rsrc_consumer_group co WHERE se.current_consumer_group_id = co.id; -- when resource plans were enabled or disabled on the instance SELECT sequence# seq, name plan_name, to_char(start_time, 'DD-MON-YY HH24:MM') start_time, to_char(end_time, 'DD-MON-YY HH24:MM') end_time, window_name FROM v$rsrc_plan_history; -- how resources were shared among the consumer groups over time select sequence# seq, name, cpu_wait_time, cpu_waits, consumed_cpu_time from V$RSRC_CONS_GROUP_HISTORY; -- sequence# as as in V$RSRC_PLAN_HISTORY select SEQUENCE#,ID,NAME,REQUESTS,CPU_WAIT_TIME,CPU_WAITS,CONSUMED_CPU_TIME,YIELDS,A CTIVE_SESS_LIMIT_HIT,UNDO_LIMIT_HIT,SWITCHES_IN_CPU_TIME,SWITCHES_OUT_CPU_TIM E,SWITCHES_IN_IO_MEGABYTES,SWITCHES_OUT_IO_MEGABYTES,SWITCHES_IN_IO_REQUESTS, SWITCHES_OUT_IO_REQUESTS,SQL_CANCELED,ACTIVE_SESS_KILLED,IDLE_SESS_KILLED,IDL E_BLKR_SESS_KILLED,QUEUED_TIME,QUEUE_TIME_OUTS,IO_SERVICE_TIME,IO_SERVICE_WAI TS,SMALL_READ_MEGABYTES,SMALL_WRITE_MEGABYTES,LARGE_READ_MEGABYTES,LARGE_WRIT E_MEGABYTES,SMALL_READ_REQUESTS,SMALL_WRITE_REQUESTS,LARGE_READ_REQUESTS,LARG E_WRITE_REQUESTS from V$RSRC_CONS_GROUP_HISTORY;
Page 194
Using Oracle Scheduler

Using Jobs
For Batch Files in Windows: o o o The OracleJobscheduler service must exist in a running state The user who runs this service should have the Logon as batch job privilege .bat file cannot be run directly, but should be called as an argument of cmd.exe
-- creating a job begin dbms_scheduler.create_job ( job_name => 'Calc_Job', job_type => 'stored_procedure', -- stored_procedure, plsql_block, executable, chain job_action => 'POPULATE_DEPT_SALS', start_date => SYSDATE, -- or to_date('01-08-2006 17:46','dd-mm-yyyy hh24:mi'), repeat_interval => 'FREQ=MINUTELY; INTERVAL=2', comments => 'Annual Department Salaries'); end; / BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => 'update_sales', job_type => 'STORED_PROCEDURE', job_action => 'OPS.SALES_PKG.UPDATE_SALES_SUMMARY', start_date => '28-APR-03 07.00.00 PM Australia/Sydney', repeat_interval => 'FREQ=DAILY;INTERVAL=2', /* every other day */ end_date => '20-NOV-04 07.00.00 PM Australia/Sydney', job_class => 'batch_update_jobs', comments => 'My new job'); END; / -- for SYSTEM=TRUE: automatically created by Oracle jobs select OWNER,JOB_NAME,JOB_SUBNAME,JOB_STYLE,JOB_CREATOR,CLIENT_ID,GLOBAL_UID,PROGRAM _OWNER,PROGRAM_NAME,JOB_TYPE,JOB_ACTION,NUMBER_OF_ARGUMENTS,SCHEDULE_OWNER,SC HEDULE_NAME,SCHEDULE_TYPE,START_DATE,REPEAT_INTERVAL,EVENT_QUEUE_OWNER,EVENT_ QUEUE_NAME,EVENT_QUEUE_AGENT,EVENT_CONDITION,EVENT_RULE,END_DATE,JOB_CLASS,EN ABLED,AUTO_DROP,RESTARTABLE,STATE,JOB_PRIORITY,RUN_COUNT,MAX_RUNS,FAILURE_COU NT,MAX_FAILURES,RETRY_COUNT,LAST_START_DATE,LAST_RUN_DURATION,NEXT_RUN_DATE,S CHEDULE_LIMIT,MAX_RUN_DURATION,LOGGING_LEVEL,STOP_ON_WINDOW_CLOSE,INSTANCE_ST ICKINESS,RAISE_EVENTS,SYSTEM,JOB_WEIGHT,NLS_ENV,SOURCE,DESTINATION,CREDENTIAL _OWNER,CREDENTIAL_NAME,INSTANCE_ID,DEFERRED_DROP,COMMENTS,FLAGS from dba_SCHEDULER_JOBS;
-- Setting Repeat Intervals FREQ YEARLY, MONTHLY, WEEKLY, DAILY, HOURLY, MINUTELY, and SECONDLY. INTERVAL a number of maximum 999 BYMONTH 1-12 1,4,6 BYYEARDAY any positive or negative number BYMONTHDAY any positive or negative number (eg -1 last day of the month)
Page 195
BYDAY (MON, TUE, and so on) can be prefixed with a number -1FRI BYHOUR 0-23 BYMINUTE 0-59 BYSECOND 0-59 Every Monday FREQ=WEEKLY; BYDAY=MON; Every other Monday FREQ=WEEKLY; BYDAY=MON; INTERVAL=2; Last day of each month FREQ=MONTHLY; BYMONTHDAY=-1; Every January 7 FREQ=YEARLY; BYMONTH=JAN; BYMONTHDAY=7; Second Wednesday of FREQ=MONTHLY; BYDAY=2WED; each month Every hour FREQ=HOURLY; Every 4 hours FREQ=HOURLY; INTERVAL=4; Hourly on the first FREQ=HOURLY; BYMONTHDAY=1; day of each month 15th day of every other FREQ=MONTHLY; BYMONTHDAY=15; INTERVAL=2 month -- MORE EXAMPLES Daily at 4, 5, and 6PM: FREQ=DAILY; BYHOUR=16,17,18; 15th day of every other month: FREQ=MONTHLY; INTERVAL=2; BYMONTHDAY=15; 29th day of every month: FREQ=MONTHLY; BYMONTHDAY=29; Second Wednesday of each month: FREQ=MONTHLY; BYDAY=2WED; Last Friday of the year: FREQ=YEARLY; BYDAY=-1FRI; Every 50 hours: FREQ=HOURLY; INTERVAL=50; Last day of every other month: FREQ=MONTHLY; INTERVAL=2; BYMONTHDAY=-1; Hourly for the first three days of every month: FREQ=HOURLY; BYMONTHDAY=1,2,3; Last workday of every month (assuming that workdays are Monday through Friday): FREQ=MONTHLY; BYDAY=MON,TUE,WED,THU,FRI; BYSETPOS=-1 Last workday of every month, excluding company holidays. (This example references an existing named schedule called Company_Holidays): FREQ=MONTHLY; BYDAY=MON,TUE,WED,THU,FRI; EXCLUDE=Company_Holidays; BYSETPOS=-1 Noon every Friday and on company holidays: FREQ=YEARLY;BYDAY=FRI;BYHOUR=12;INCLUDE=Company_Holidays -- Testing Repeat Intervals set serveroutput on DECLARE start_date TIMESTAMP; return_date_after TIMESTAMP; next_run_date TIMESTAMP; BEGIN start_date := to_timestamp('01-01-2006 00:00:00','DD-MM-YYYY HH24:MI:SS'); DBMS_SCHEDULER.EVALUATE_CALENDAR_STRING( CALENDAR_STRING => 'FREQ=MONTHLY;INTERVAL=1', START_DATE => start_date, RETURN_DATE_AFTER => null, NEXT_RUN_DATE => next_run_date); DBMS_OUTPUT.PUT_LINE('next_run_date: ' || next_run_date); END; / alter session set nls_date_language=English / DECLARE
Page 196
start_date TIMESTAMP; return_date_after TIMESTAMP; next_run_date TIMESTAMP; BEGIN start_date := to_timestamp('10-10-2004 10:00:00','DD-MM-YYYY HH24:MI:SS'); return_date_after := start_date; FOR i IN 1..10 LOOP DBMS_SCHEDULER.EVALUATE_CALENDAR_STRING( 'FREQ=MONTHLY; INTERVAL=2; BYMONTHDAY=15', start_date, return_date_after, next_run_date); DBMS_OUTPUT.PUT_LINE('Next Run Date: ' || to_char(next_run_date,'Dy dd-MMRR HH24:MI')); return_date_after := next_run_date; END LOOP; END; /
-- Creating a Set of Regular Jobs DECLARE newjob sys.job; newjobarr sys.job_array; BEGIN -- Create an array of JOB object types newjobarr := sys.job_array(); -- Allocate sufficient space in the array newjobarr.extend(5); -- Add definitions for 5 jobs FOR i IN 1..5 LOOP -- Create a JOB object type newjob := sys.job(job_name => 'TESTJOB' || to_char(i), job_style => 'REGULAR', job_template => 'PROG1', repeat_interval => 'FREQ=MINUTELY;INTERVAL_15', start_date => systimestamp + interval '600' second, max_runs => 2, auto_drop => FALSE, enabled _> TRUE ); -- Add it to the array newjobarr(i) := newjob; END LOOP; -- Call CREATE_JOBS to create jobs in one transaction DBMS_SCHEDULER.CREATE_JOBS(newjobarr, 'TRANSACTIONAL'); END; /
-- Creating a Set of Lightweight Jobs DECLARE newjob sys.job; newjobarr sys.job_array; BEGIN newjobarr := sys.job_array(); newjobarr.extend(5); FOR i IN 1..5 LOOP newjob := sys.job(job_name => 'lwjob_' || to_char(i), job_style => 'LIGHTWEIGHT',
Page 197
job_template => 'PROG1', repeat_interval => 'FREQ=MINUTELY;INTERVAL=15', start_date => systimestamp + interval '10' second, enabled => TRUE); newjobarr(i) := newjob; end loop; DBMS_SCHEDULER.CREATE_JOBS(newjobarr, 'TRANSACTIONAL'); END; /
-- Creating Jobs Using a Named Program BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => 'my_new_job1', program_name => 'my_saved_program', repeat_interval => 'FREQ=DAILY;BYHOUR=12', comments => 'Daily at noon'); END; /
-- Creating Jobs Using a Named Schedule BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => 'my_new_job2', job_type => 'PLSQL_BLOCK', job_action => 'BEGIN SALES_PKG.UPDATE_SALES_SUMMARY; END;', schedule_name => 'my_saved_schedule'); END; / -- Creating Jobs Using a Named Program and Schedule BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => 'my_new_job3', program_name => 'my_saved_program1', schedule_name => 'my_saved_schedule1'); END; / -- Creating Remote External Jobs -- Note: check section " Enabling and Disabling Remote External Jobs" BEGIN DBMS_SCHEDULER.CREATE_JOB( job_name => 'CLEANLOGS', job_type => 'EXECUTABLE', job_action => '/home/logowner/cleanlogs', repeat_interval => 'FREQ=DAILY; BYHOUR=23', enabled => FALSE); DBMS_SCHEDULER.SET_ATTRIBUTE('CLEANLOGS', 'credential_name', 'LOGOWNER'); DBMS_SCHEDULER.SET_ATTRIBUTE('CLEANLOGS', 'destination', 'app455:12345'); DBMS_SCHEDULER.ENABLE('CLEANLOGS'); END; / -- in multiple hosts declare
Page 198
job_prefix varchar2(30) := 'remote_'; job_name varchar2(30); destinations dbms_utility.lname_array; begin destinations(1) := 'host1:1234'; destinations(2) := 'host2:1234'; destinations(3) := 'host3:1234'; destinations(4) := 'host4:1234'; for i in 1..destinations.LAST loop job_name := dbms_scheduler.generate_job_name(job_prefix); dbms_scheduler.create_job(job_name, job_type=>'executable', job_action=>'/u01/app/ext_backup', number_of_arguments=>0, enabled=>false); dbms_scheduler.set_attribute(job_name,'destination',destinations(i)); dbms_scheduler.set_attribute(job_name,'credential_name','remote_cred'); dbms_scheduler.enable(job_name); end loop; end; / -- run a shell script that uses SQL*Plus to submit the statements -- external authentication is assumed #!/bin/sh export ORACLE_HOME=/u01/app/oracle/product/11.1.0/db_1 export ORACLE_SID=orcl export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$ORACLE_HOME/lib # The following command assumes external authentication $ORACLE_HOME/bin/sqlplus / << EOF set serveroutput on; select * from dual; EXIT; EOF -- Copying Jobs -- the new job is created disabled begin DBMS_SCHEDULER.COPY_JOB ( old_job =>'MY_OLD_JOB', new_job =>'MY_NEW_JOB'); end;
-- altering jobs -- value paramter may be {VARCHAR2|TIMESTAMP WITH TIMEZONE| -- PLS_INTEGER|BOOLEAN|INTERVAL DAY TO SECOND} begin DBMS_SCHEDULER.SET_ATTRIBUTE ( name =>'FULL_BAKCUP', attribute =>'auto_drop', -- see possible values below value =>TRUE); end; / -- attribute possible values for JOBS auto_drop
Page 199
comments credential_name credential to use when running an external job database_role Oracle Data Guard 'PRIMARY' or 'LOGICAL STANDBY' destination host and port on which to run a remote external job end_date if no value, the job repeats forever event_spec follow_default_timezone instance_id instance_stickiness TRUE or FALSE job_action job_class job_priority the order in which jobs from that class are picked up by the job coordinator. 1(first)-5, 3 default job_type 'PLSQL_BLOCK', 'STORED_PROCEDURE','EXECUTABLE', 'CHAIN' job_weight in parallel technology, 1-100 logging_level DBMS_SCHEDULER.LOGGING_OFF, LOGGING_FAILED_RUNS, LOGGING_RUNS, LOGGING_FULL max_failures number of times a job can fail on consecutive max_run_duration (INTERVAL DAY TO SECOND) if job duration exceeds this value, the Scheduler raises an event of type JOB_OVER_MAX_DUR. It is then up to your event handler to decide whether or not to allow the job to continue max_runs the maximum number of consecutive scheduled runs. disabled then. number_of_arguments The number of arguments if the program is inlined parallel_instances TRUE or FALSE program_name raise_events job_started CONSTANT PLS_INTEGER := 1 job_succeeded CONSTANT PLS_INTEGER := 2 job_failed CONSTANT PLS_INTEGER :=4 job_broken CONSTANT PLS_INTEGER :=8 job_completed CONSTANT PLS_INTEGER :=16 job_stopped CONSTANT PLS_INTEGER :=32 job_sch_lim_reached CONSTANT PLS_INTEGER :=64 job_disabled CONSTANT PLS_INTEGER :=128 job_chain_stalled CONSTANT PLS_INTEGER :=256 job_all_events CONSTANT PLS_INTEGER := 511 job_run_completed CONSTANT PLS_INTEGER := job_succeeded + job_failed + job_stopped repeat_interval Either a PL/SQL function returning the next date and time on which to run, or calendaring syntax expression restartable schedule_limit time limit after which the job will not start ( from 1 minute to 99 days ) schedule_name start_date stop_on_window_close
-- Setting Multiple Job Attributes for a Set of Regular Jobs DECLARE newattr sys.jobattr; newattrarr sys.jobattr_array; j number; BEGIN
Page 200
-- Create new JOBATTR array newattrarr := sys.jobattr_array(); -- Allocate enough space in the array newattrarr.extend(20); j := 1; FOR i IN 1..5 LOOP -- Create and initialize a JOBATTR object type newattr := sys.jobattr(job_name => 'TESTJOB' || to_char(i), attr_name => 'MAX_FAILURES', attr_value => 5); -- Add it to the array. newattrarr(j) := newattr; j := j + 1; newattr := sys.jobattr(job_name => 'TESTJOB' || to_char(i), attr_name => 'COMMENTS', attr_value => 'Bogus comment'); newattrarr(j) := newattr; j := j + 1; newattr := sys.jobattr(job_name => 'TESTJOB' || to_char(i), attr_name => 'END_DATE', attr_value => systimestamp + interval '24' hour); newattrarr(j) := newattr; j := j + 1; newattr := sys.jobattr(job_name => 'TESTJOB' || to_char(i), attr_name => 'SCHEDULE_LIMIT', attr_value => interval '1' hour); newattrarr(j) := newattr; j := j + 1; END LOOP; -- Call SET_JOB_ATTRIBUTES to set all 20 set attributes in one transaction DBMS_SCHEDULER.SET_JOB_ATTRIBUTES(newattrarr, 'TRANSACTIONAL'); END; / -- stopping a job -- stop job job1 and all jobs in the job class dw_jobs. BEGIN DBMS_SCHEDULER.STOP_JOB('job1, sys.dw_jobs'); END; /
-- Dropping Jobs BEGIN DBMS_SCHEDULER.DROP_JOB ('job1, job3, sys.jobclass1, sys.jobclass2'); END; / -- Disabling Jobs BEGIN DBMS_SCHEDULER.DISABLE('job1, job2, job3, sys.jobclass1'); END; / -- Enabling Jobs BEGIN DBMS_SCHEDULER.ENABLE ('job1, job2, job3,sys.jobclass1');
Page 201
END; /
Using Programs
-- Creating Programs BEGIN DBMS_SCHEDULER.CREATE_PROGRAM ( program_name => 'my_program1', program_action => '/usr/local/bin/date', program_type => 'EXECUTABLE', comments => 'My comments here'); END; / BEGIN DBMS_SCHEDULER.CREATE_PROGRAM ( program_name => 'oe.my_program1', program_type => 'PLSQL_BLOCK', program_action => 'BEGIN DBMS_STATS.GATHER_TABLE_STATS(''oe'', ''sales''); END;', number_of_arguments => 0, enabled => TRUE, comments => 'My comments here'); END; / SELECT PROGRAM_NAME FROM DBA_SCHEDULER_PROGRAMS WHERE PROGRAM_NAME = 'MY_PROGRAM1'; -- Defining Program Arguments BEGIN DBMS_SCHEDULER.DEFINE_PROGRAM_ARGUMENT ( program_name => 'operations_reporting', argument_position => 2, argument_name => 'end_date', argument_type => 'VARCHAR2', default_value => '12-DEC-03'); END; / -- dropping a program argument BEGIN DBMS_SCHEDULER.DROP_PROGRAM_ARGUMENT ( program_name => 'operations_reporting', argument_position => 2); DBMS_SCHEDULER.DROP_PROGRAM_ARGUMENT ( program_name => 'operations_reporting', argument_name => 'end_date'); END; / -- sets the value (non-Null) of an argument of the associated PROGRAM OBJECT for a job -- by argument position BEGIN DBMS_SCHEDULER.SET_JOB_ARGUMENT_VALUE(
Page 202
job_name => 'ops_reports', argument_position => 2, argument_value => '12-DEC-03'); END; / -- by argument name BEGIN DBMS_SCHEDULER.SET_JOB_ARGUMENT_VALUE( job_name => 'ops_reports', argument_name => 'END_DATE', argument_value => '12-DEC-03'); END; / -- setting an argument to NULL value BEGIN DBMS_SCHEDULER.SET_ATTRIBUTE_NULL ( job_name => 'ops_reports', argument_name => 'END_DATE'); END; / -- Dropping Programs BEGIN DBMS_SCHEDULER.DROP_PROGRAM('program1, program2, program3'); END; / -- Disabling Programs BEGIN DBMS_SCHEDULER.DISABLE ( name => 'ops_reports', force =>FALSE, -- the default commit_semantics => 'STOP_ON_FIRST_ERROR'); -- TRANSACTIONAL or ABSORB_ERRORS END; / -- Enabling Programs BEGIN DBMS_SCHEDULER.ENABLE('program1, program2, program3'); END; /
Using Schedules
-- Creating Schedules BEGIN DBMS_SCHEDULER.CREATE_SCHEDULE ( schedule_name => 'my_stats_schedule', start_date => SYSTIMESTAMP, end_date => SYSTIMESTAMP + INTERVAL '30' day, repeat_interval => 'FREQ=HOURLY; INTERVAL=4', comments => 'Every 4 hours'); END; /
Page 203
-- altering schedules BEGIN dbms_scheduler.set_attribute( name => 'my_stats_schedule', attribute => 'REPEAT_INTERVAL', value => 'FREQ=HOURLY; INTERVAL=2'); END; / attribute: comments end_date event_spec repeat_interval source The host name that the database is running on start_date -- Dropping Schedules BEGIN DBMS_SCHEDULER.DROP_SCHEDULE ( schedule_name => 'my_stats_schedule', force => FALSE); -- DEFAULT END; /
Using Job Classes

Jobs classes provide a way to group jobs for resource allocation and prioritization, and a way to easily assign a set of attribute values to member jobs. -- Creating Job Classes BEGIN DBMS_SCHEDULER.CREATE_JOB_CLASS ( job_class_name => 'finance_jobs_class', resource_consumer_group => 'finance_group'); END; / BEGIN dbms_scheduler.set_attribute( name => ' finance_jobs_class', attribute => 'logging_level', value => DBMS_SCHEDULER.LOGGING_FAILED_RUNS); END; / ATTRIBUTE: comments logging_level DBMS_SCHEDULER.LOGGING_OFF, LOGGING_FAILED_RUNS, LOGGING_RUNS, LOGGING_FULL resource_consumer_group jobs run under this resource consumer group service -- Dropping Job Classes BEGIN DBMS_SCHEDULER.DROP_JOB_CLASS('jobclass1, jobclass2, jobclass3'); END; /
Page 204
Using Windows
-- Creating Windows begin dbms_scheduler.create_window ( window_name => 'WORK_HOURS_WINDOW', resource_plan => 'DAY_PLAN', start_date => SYSTIMESTAMP, -- also schedule_name repeat_interval => 'FREQ=DAILY; BYHOUR=8', duration => INTERVAL '10' HOUR, window_priority => 'HIGH'); -- or LOW end; / BEGIN DBMS_SCHEDULER.CREATE_WINDOW ( window_name => 'my_window1', resource_plan => 'my_res_plan1', start_date => '15-JUL-03 1.00.00AM US/Pacific', repeat_interval => 'FREQ=DAILY', end_date => '15-SEP-03 1.00.00AM US/Pacific', duration => interval '80' MINUTE, comments => 'This is my first window'); END; / BEGIN DBMS_SCHEDULER.CREATE_WINDOW ( window_name => 'my_window2', schedule_name => 'my_stats_schedule', resource_plan => 'my_resourceplan1', duration => interval '60' minute, comments => 'My window'); END; / SELECT WINDOW_NAME FROM DBA_SCHEDULER_WINDOWS WHERE WINDOW_NAME = 'MY_WINDOW1'; -- Altering Windows BEGIN dbms_scheduler.set_attribute( name => 'MYWINDOW', attribute => 'window_priority', value => 'LOW'); END; / ATTRIBUTE: comments duration end_date repeat_interval resource_plan schedule_name start_date window_priority -- openinng a window begin
Page 205
can be used
dbms_scheduler.open_window ( window_name => 'WORK_HOURS_WINDOW', duration => INTERVAL '20' MINUTE); end; / -- closing a window begin dbms_scheduler.close_window ('WORK_HOURS_WINDOW'); end; / -- dropping a window BEGIN DBMS_SCHEDULER.DROP_WINDOW ('window1, window2, window3, windowgroup1, windowgroup2'); END; / -- Disabling Windows BEGIN DBMS_SCHEDULER.DISABLE ('sys.window1, sys.window2, sys.window3, sys.windowgroup1, sys.windowgroup2'); END; / -- Enabling Windows BEGIN DBMS_SCHEDULER.ENABLE ('sys.window1, sys.window2, sys.window3'); END; / -- Window Logging column log_id format a3 column window_name format a20 column operation format a8 select trim(log_id) log_id, trunc(log_date) log_date, window_name, operation from dba_scheduler_window_log / column actual_duration format a20 select trim(log_id) log_id, trunc(log_date) log_date, window_name, actual_duration from dba_scheduler_window_details order by 2 desc / begin DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE('LOG_HISTORY','60'); end; /
Using Window Groups

If you create a window group, add windows to it, and then name this window
Page 206
group in a job's schedule_name attribute, the job runs during all the windows in the window group. -- Creating Window Groups BEGIN DBMS_SCHEDULER.CREATE_WINDOW_GROUP ( group_name => 'downtime', window_list => 'weeknights, weekends'); END; / -- Dropping Window Groups BEGIN DBMS_SCHEDULER.DROP_WINDOW_GROUP('windowgroup1, windowgroup2'); END; / -- Adding a Member to a Window Group BEGIN DBMS_SCHEDULER.ADD_WINDOW_GROUP_MEMBER ('window_group1', 'window1, window2'); END; / -- Dropping a Member from a Window Group BEGIN DBMS_SCHEDULER.REMOVE_WINDOW_GROUP_MEMBER('window_group1', 'window1,window2'); END; / -- Enabling a Window Group BEGIN DBMS_SCHEDULER.ENABLE('sys.windowgroup1', 'sys.windowgroup2, sys.windowgroup3'); END; / -- Disabling a Window Group: but not the members BEGIN DBMS_SCHEDULER.DISABLE('sys.windowgroup1,sys.windowgroup2'); END; /
Monitoring Job State with Email Notifications

You can configure a job to send e-mail notifications when it changes state.
/* Configuration */ CONN / AS SYSDBA BEGIN -- define the SMTP server DBMS_SCHEDULER.set_scheduler_attribute('email_server', 'smtp.mydomain.com:25'); -- optionally define default sender address, which -- is used if the sender parameter is not specified DBMS_SCHEDULER.set_scheduler_attribute('email_sender', '[email protected]'); END; /
Page 207
-- to enable/disable encryption is for the SMTP server connection -- only (11.2.0.2) -- possible values: NONE, SSL_TLS, STARTTLS exec DBMS_SCHEDULER.set_scheduler_attribute('email_server_encryption','SSL_TLS')
-- Authentication -- If the SMTP server requires authentication, then the Scheduler uses the -- user name and password stored in the specified credential object -- default NULL exec dbms_scheduler.create_credential('hrcredential','hr','hrpassword'); exec DBMS_SCHEDULER.set_scheduler_attribute('email_server_credential','hrcredential')
/* Using Email Notification */ -- You call ADD_JOB_EMAIL_NOTIFICATION once for each different set of notifications -- that you want to configure for a job. -- associate an email notification with the job -- using the default subject and body BEGIN DBMS_SCHEDULER.add_job_email_notification ( job_name => 'email_notification_job', recipients => '[email protected]', events => 'job_started, job_succeeded'); END; -- subject and body specified: BEGIN DBMS_SCHEDULER.ADD_JOB_EMAIL_NOTIFICATION ( job_name => 'email_notification_job', recipients => '[email protected], [email protected]', sender => '[email protected]', subject => 'Scheduler Job Notification-%job_owner%.%job_name%-%event_type%', body => '%event_type% occurred at %event_timestamp%. %error_message%', events => 'JOB_FAILED, JOB_BROKEN, JOB_DISABLED, JOB_SCH_LIM_REACHED'); END;
-- configures an additional e-mail notification for the same job -- for a different event BEGIN DBMS_SCHEDULER.ADD_JOB_EMAIL_NOTIFICATION ( job_name => 'email_notification_job', recipients => '[email protected]', events => 'JOB_OVER_MAX_DUR'); END;
-- The notification fires only if a job fails with "600" "700" error codes BEGIN DBMS_SCHEDULER.add_job_email_notification ( job_name => 'email_notification_job', recipients => '[email protected]', events => 'job_failed', filter_condition => ':event.error_code=600 or :event.error_code=700'); END;
/* Removing Email Notification */ Page 208
-- remove the notification from specified recipient/event BEGIN DBMS_SCHEDULER.remove_job_email_notification ( job_name => 'email_notification_job', recipients => '[email protected]', events => 'job_succeeded'); END; -- remove the notification from all recipients and events BEGIN DBMS_SCHEDULER.remove_job_email_notification ( job_name => 'email_notification_job'); END;
/* Obtain Info about Email Notifications */ SELECT job_name, recipient, event, filter_condition, subject, body FROM user_scheduler_notifications;
Using File Watchers

/* Obtain Info about FW */ SELECT file_watcher_name, destination, directory_path, file_name, credential_name FROM user_scheduler_file_watchers;
/* Configuration */ -- by default, a destination is checked every 10 mins. To change this: -- only sys can do it: CONN / AS SYSDBA set serveroutput on declare v varchar2(1000); begin DBMS_SCHEDULER.GET_ATTRIBUTE ( 'FILE_WATCHER_SCHEDULE','REPEAT_INTERVAL', v); DBMS_OUTPUT.PUT_LINE(V); end; / BEGIN DBMS_SCHEDULER.set_attribute( 'file_watcher_schedule', 'repeat_interval', 'freq=minutely; interval=5'); END; /
/* Creating File Watcher */ -- create OS credential: BEGIN DBMS_SCHEDULER.create_credential( credential_name => 'fw_credential', username => 'oracle', password => 'oracle'); END; /
Page 209
-- Grant EXECUTE on the credential to the schema that owns the -- event-based job that the file watcher will start: GRANT EXECUTE ON fw_credential to DSSUSER;
-- create file watcher: BEGIN DBMS_SCHEDULER.create_file_watcher( file_watcher_name => 'data_fw', directory_path => '/tmp/test', -- if '?' = ORACLE_HOME file_name => '*.dat', -- wildcard supported credential_name => 'fw_credential', destination => NULL, -- NULL=local server enabled => FALSE); END; /
-- Grant EXECUTE on the file watcher to any schema that owns an event-based job -- that references the file watcher. GRANT EXECUTE ON data_fw to DSSUSER;
-- create a program raised by the file watcher BEGIN DBMS_SCHEDULER.create_program( program_name => 'import_data_prog', program_type => 'stored_procedure', program_action => 'import_data_proc', number_of_arguments => 1, enabled => FALSE); END; /
-- define the metadata argument using the event_message attribute -- the metadata contains info about the file, such as its name: BEGIN DBMS_SCHEDULER.define_metadata_argument( program_name => 'import_data_prog', metadata_attribute => 'event_message', argument_position => 1); END; /
-- create the defined procedure: -- It must accept an argument of the SCHEDULER_FILEWATCHER_RESULT type CREATE TABLE received_files ( fileinfo VARCHAR2(4000), rdate date ); CREATE OR REPLACE PROCEDURE import_data_proc (p_sfwr SYS.SCHEDULER_FILEWATCHER_RESULT) AS v_message received_files.fileinfo%type; BEGIN v_message := p_sfwr.directory_path || '/' p_sfwr.file_size || ')'; INSERT INTO received_files VALUES (v_message, sysdate); COMMIT; END; /
||
p_sfwr.actual_file_name
||
'
('
||
Page 210
-- create the job: BEGIN DBMS_SCHEDULER.create_job( job_name => 'import_data_job', program_name => 'import_data_prog', event_condition => NULL, -- 'tab.user_data.file_size < 1024' queue_spec => 'data_fw', -- file watcher name auto_drop => FALSE, enabled => FALSE); END; /
-- By default, the arrival of new files will be ignored if the job is already running. -- If you need the job to fire for each new arrival, regardless of whether the job is already -- running or not, set the PARALLEL_INSTANCES attribute for the job to true. The job -- will then be run as a lightweight job: BEGIN DBMS_SCHEDULER.set_attribute('import_data_job','parallel_instances',TRUE); END; /
-- Enable all the objects: EXEC DBMS_SCHEDULER.enable('data_fw','import_data_prog','import_data_job');
-- to test: echo "This is a test" > /tmp/test/f1.dat echo "This is a test too" > /tmp/test/f2.dat echo "Yes another test" > /tmp/test/f3.dat
select * from received_files order by rdate desc;
/* Managing File Watchers */ -- enable/disable EXEC DBMS_SCHEDULER.enable('data_fw') ; EXEC DBMS_SCHEDULER.disable('data_fw') ;
-- change an attribute: begin DBMS_SCHEDULER.SET_ATTRIBUTE ( name => 'data_fw', attribute =>'directory_path', value =>'/home/oracle/receivedfiles' ) ; end; / begin DBMS_SCHEDULER.SET_ATTRIBUTE ( name => 'data_fw', attribute =>'file_name', value =>'*.txt' ) ; end; /
Page 211
begin DBMS_SCHEDULER.SET_ATTRIBUTE ( name => 'data_fw', attribute =>'credential_name', value =>'fw_credential2' ) ; end; /
-- to drop a file watchers: DBMS_SCHEDULER.DROP_FILE_WATCHER('data_fw');
Using Events Raised by the Scheduler

-- by default messages raised by the scheduler are -- deleted after 24 hours: event_expiry_time SET SERVEROUTPUT ON DECLARE V VARCHAR2(200); BEGIN DBMS_SCHEDULER.GET_SCHEDULER_ATTRIBUTE ( attribute =>'EVENT_EXPIRY_TIME', value =>V); DBMS_OUTPUT.PUT_LINE(nvl(V,'24 hours')); -- 24 hours if null END; / declare n number := 48*60*60; begin DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE( attribute =>'EVENT_EXPIRY_TIME', value =>n); -- in seconds (24 hours if NULL) end; / -- making sure the required privs are granted -- EXECUTE on both DBMS_AQ and DBMS_AQADM CONN SYS select grantee, privilege, table_name from DBA_TAB_PRIVS where table_name in ('DBMS_AQ','DBMS_AQADM') and grantee='USER1'; GRANT GRANT GRANT GRANT GRANT EXECUTE ON DBMS_AQ TO USER1; EXECUTE ON DBMS_AQADM TO USER1; SELECT ON DBA_AQ_AGENTS TO USER1; CREATE JOB TO USER1; CREATE EXTERNAL JOB TO USER1;
begin DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('ENQUEUE_ANY','USER1',FALSE); DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('DEQUEUE_ANY','USER1',FALSE); DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('MANAGE_ANY','USER1',FALSE); end; /

Page 212
-- create the job BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => '"USER1"."U1JOB"', job_type => 'EXECUTABLE', job_action => 'C:\windows\system32\cmd.exe', number_of_arguments => 3, start_date => systimestamp, repeat_interval => 'FREQ=SECONDLY;INTERVAL=10', comments => 'Testing Events raised by scheduler', auto_drop => FALSE, enabled => FALSE); DBMS_SCHEDULER.set_job_argument_value('"USER1"."U1JOB"',1,'/c'); DBMS_SCHEDULER.set_job_argument_value('"USER1"."U1JOB"',2, 'c:\temp\testme.bat'); DBMS_SCHEDULER.set_job_argument_value('"USER1"."U1JOB"',3, 'passed to bat'); END; /
-- alter job to raise events BEGIN DBMS_SCHEDULER.SET_ATTRIBUTE( name => '"USER1"."U1JOB"', attribute => 'raise_events', value => dbms_scheduler.job_started + dbms_scheduler.job_succeeded + dbms_scheduler.job_failed + dbms_scheduler.job_broken + dbms_scheduler.job_completed + dbms_scheduler.job_stopped + dbms_scheduler.job_sch_lim_reached + dbms_scheduler.job_disabled + dbms_scheduler.job_chain_stalled ); END; / -- enable Job BEGIN DBMS_SCHEDULER.ENABLE( '"USER1"."U1JOB"' ); END; /
-- Consuming Scheduler-Raised Events with your Application -- follow either Plan A or Plan B -- Plan A --1. conn sys or with user of MANAGE ANY QUEUE --2. Subscribe to the queue using a new or existing agent --3. Run the procedure DBMS_AQADM.ENABLE_DB_ACCESS(agent_name, db_username); -- Plan B /* DBMS_SCHEDULER.ADD_EVENT_QUEUE_SUBSCRIBER(subscriber_name); where subscriber_name is the name of the Oracle Streams Advanced Queuing (AQ) agent to be used to subscribe to the Scheduler event queue. (If it is NULL, an agent is created whose name is the user name of the calling user.) This call both creates a subscription to the Scheduler event queue and grants
Page 213
the user permission to dequeue using the designated agent. */ conn user1 exec DBMS_SCHEDULER.ADD_EVENT_QUEUE_SUBSCRIBER -- opposite: REMOVE_EVENT_QUEUE_SUBSCRIBER -- Events are dequeued from the scheduler event queue using the DBMS_AQ SET SERVEROUTPUT ON DECLARE l_dequeue_options DBMS_AQ.dequeue_options_t; l_message_properties DBMS_AQ.message_properties_t; l_message_handle RAW(16); l_queue_msg sys.scheduler$_event_info; BEGIN l_dequeue_options.consumer_name := 'USER1'; DBMS_AQ.dequeue(queue_name => 'SYS.SCHEDULER$_EVENT_QUEUE', dequeue_options => l_dequeue_options, message_properties => l_message_properties, payload => l_queue_msg, msgid => l_message_handle); COMMIT; DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line DBMS_OUTPUT.put_line END; / ('event_type : ('object_owner : ('object_name : ('event_timestamp: ('error_code : ('event_status : ('log_id : ('run_count : ('failure_count : ('retry_count : ' ' ' ' ' ' ' ' ' ' || || || || || || || || || || l_queue_msg.event_type); l_queue_msg.object_owner); l_queue_msg.object_name); l_queue_msg.event_timestamp); l_queue_msg.error_code); l_queue_msg.event_status); l_queue_msg.log_id); l_queue_msg.run_count); l_queue_msg.failure_count); l_queue_msg.retry_count);
-- remove the job and unsubscribes the user from the scheduler event queue EXEC DBMS_SCHEDULER.drop_job('"USER1"."U1JOB"') EXEC DBMS_SCHEDULER.remove_event_queue_subscriber
Using Events Raised by an Application (Events-Based Jobs)

-- by default messages raised by the scheduler are -- deleted after 24 hours: event_expiry_time declare n number := 48*60*60; begin DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE ( attribute =>'event_expiry_time', value =>n); -- in seconds (24 hours if NULL) end; / -- making sure the required privs are granted -- EXECUTE on both DBMS_AQ and DBMS_AQADM select grantee, privilege, table_name from dba_tab_privs where table_name in ('DBMS_AQ','DBMS_AQADM')
Page 214
and grantee='TEST_USER'; GRANT EXECUTE ON DBMS_AQ TO test_user; GRANT EXECUTE ON DBMS_AQADM TO test_user; GRANT SELECT ON DBA_AQ_AGENTS TO test_user; begin DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('ENQUEUE_ANY','test_user',FALSE); DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('DEQUEUE_ANY','test_user',FALSE); DBMS_AQADM.GRANT_SYSTEM_PRIVILEGE('MANAGE_ANY','test_user',FALSE); end; / -- CONNECT AS test_user -- create type for the message to receive connect marvin/panic create or replace type MY_MSGT as object ( msg varchar2(20) ) / begin DBMS_AQADM.CREATE_QUEUE_TABLE ( QUEUQ_TABLE=>'MY_QT', QUEUQ_PAYLOAD_TYPE => 'MY_MSGT', MULTIPLE_CONSUMERS => TRUE ); DBMS_AQADM.CREATE_QUEUE( QUEUE_NAME => 'MY_Q', QUEUE_TABLE => 'MY_QT' ); DBMS_AQADM.START_QUEUE(QUEUE_NAME=>'MY_Q'); end; / -- example of event_condition = 'tab.user_data.event_type = ''CARD_SWIPE'' and extract hour from tab.user_data.event_timestamp < 9' BEGIN dbms_scheduler.create_job ( job_name => '"TEST_USER"."BCKUP_01"', job_type => 'EXECUTABLE', job_action => '/home/oracle/bin/rman.sh', event_condition => 'tab.user_data.msg=''GO''', queue_spec => '"TEST_USER"."MY_Q"', -- agent, queue name start_date => systimestamp, job_class => NULL, comments => 'backup a database', auto_drop => FALSE, number_of_arguments => 1, enable => FALSE ); END; / -- Altering a Job to Raise Events BEGIN DBMS_SCHEDULER.SET_ATTRIBUTE('dw_reports', 'raise_events', DBMS_SCHEDULER.JOB_FAILED + DBMS_SCHEDULER.JOB_SCH_LIM_REACHED); END; / raise_events job_started CONSTANT PLS_INTEGER := 1
Page 215
job_succeeded CONSTANT PLS_INTEGER := 2 job_failed CONSTANT PLS_INTEGER :=4 job_broken CONSTANT PLS_INTEGER :=8 job_completed CONSTANT PLS_INTEGER :=16 job_stopped CONSTANT PLS_INTEGER :=32 job_sch_lim_reached CONSTANT PLS_INTEGER :=64 job_disabled CONSTANT PLS_INTEGER :=128 job_chain_stalled CONSTANT PLS_INTEGER :=256 job_all_events CONSTANT PLS_INTEGER := 511 job_run_completed := job_succeeded + job_failed + job_stopped -- if required, set arguments and attributes -- job argument passed to the job DBMS_SCHEDULER.SET_JOB_ARGUMENT_VALUE ( job_name => '"TEST_USER"."BCKUP_01"', argument_position => 1, argument_value => 'db_01' ); -- for remote job DBMS_SCHEDULER.SET_ATTRIBUTE ( name => '"TEST_USER"."BCKUP_01"', attribute => 'destination', value => 'pantzer:15021' ); DBMS_SCHEDULER.SET_ATTRIBUTE ( name => '"TEST_USER"."BCKUP_01"', attribute => 'credential_name', value => '"TEST_USER."JOBS_CRED2"' ); END; / -- enable Job BEGIN DBMS_SCHEDULER.ENABLE( '"TEST_USER"."BCKUP_01"' ); END; / DECLARE my_msgid RAW(16); props dbms_aq.message_properties_t; enqopts dbms_aq.enqueue_options_t; BEGIN sys.dbms_aq.enqueue('marvin.bckup_q', enqopts, props, marvin.bckup_msgt('GO'), my_msgid); end; / COMMIT; -- * alternatively scheduler can be used for the same BEGIN DBMS_SCHEDULER.CREATE_EVENT_SCHEDULE ( schedule_name => 'entry_events_schedule', start_date => SYSTIMESTAMP,
Page 216
event_condition => 'tab.user_data.event_type = ''CARD_SWIPE''', queue_spec => 'entry_events_q, entry_agent1'); END; / BEGIN DBMS_SCHEDULER.SET_ATTRIBUTE ('entry_events_schedule', 'event_spec', 'tab.user_data.event_type = ''BAD_BADGE''', 'entry_events_q, entry_agent1'); END; /
Using Chains
-- Creating Chains BEGIN DBMS_SCHEDULER.CREATE_CHAIN ( chain_name => 'my_chain1', rule_set_name => NULL, evaluation_interval => NULL, comments => 'My first chain'); END; / -- Defining Chain Steps -- a step point to program, another chain, event BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_STEP ( chain_name => 'my_chain1', step_name => 'my_step1', program_name => 'my_program1'); DBMS_SCHEDULER.DEFINE_CHAIN_STEP ( chain_name => 'my_chain1', step_name => 'my_step2', program_name => 'my_chain2'); END; / -- event-based (waits for an event to occur) BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_EVENT_STEP ( chain_name => 'my_chain1', step_name => 'my_step3', event_schedule_name => 'my_event_schedule'); END; / -- Adding Rules to a Chain -- Each rule has a condition and an action -- rule that starts the chain always evaluates to TRUE -- A chain job does not complete until one of the rules containing the END action -- evaluates to TRUE. If no such a step, the job enters the CHAIN_STALLED state. BEGIN Oracle DBA Code Examples
Page 217
DBMS_SCHEDULER.DEFINE_CHAIN_RULE ( chain_name => 'my_chain1', condition => 'TRUE', action => 'START step1', rule_name => 'my_rule1', comments => 'start the chain'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ( chain_name => 'my_chain1', condition => 'step1 completed', action => 'START step2', rule_name => 'my_rule2'); END; / /* :step_name.attribute. (step_name refers to a typed object.) Possible attributes are: completed, state, start_date, end_date, error_code, and duration. Possible values for the state attribute include: 'NOT_STARTED', 'SCHEDULED', 'RUNNING', 'PAUSED', 'STALLED', 'SUCCEEDED', 'FAILED', and 'STOPPED'. If a step is in the state 'SUCCEEDED', 'FAILED', or 'STOPPED', its completed attribute is set to 'TRUE', otherwise completed is 'FALSE'. */ -- examples of conditions step3 succeeded step3 failed -- Example: 2 BEGIN DBMS_SCHEDULER.CREATE_CHAIN ( chain_name => 'my_chain1', rule_set_name => NULL, evaluation_interval => NULL, comments => NULL); END; / --- define three steps for this chain. BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain1', 'step1', 'my_program1'); DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain1', 'step2', 'my_program2'); DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain1', 'step3', 'my_program3'); END; / --- define corresponding rules for the chain. BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_RULE('my_chain1', 'TRUE', 'START step1'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ( 'my_chain1', 'step1 COMPLETED', 'Start step2, step3'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ( 'my_chain1', 'step2 COMPLETED AND step3 COMPLETED', 'END'); END; / -- Example 3: BEGIN DBMS_SCHEDULER.CREATE_CHAIN (
Page 218
chain_name => 'my_chain2', rule_set_name => NULL, evaluation_interval => NULL, comments => NULL); END; / --- define three steps for this chain. BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain2', 'step1', 'my_program1'); DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain2', 'step2', 'my_program2'); DBMS_SCHEDULER.DEFINE_CHAIN_STEP('my_chain2', 'step3', 'my_program3'); END; / --- define corresponding rules for the chain. BEGIN DBMS_SCHEDULER.DEFINE_CHAIN_RULE ('my_chain2', 'TRUE', 'START step1'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ( 'my_chain2', 'step1 SUCCEEDED', 'Start step2'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ('my_chain2', 'step1 COMPLETED AND step1 NOT SUCCEEDED', 'Start step3'); DBMS_SCHEDULER.DEFINE_CHAIN_RULE ('my_chain2', 'step2 COMPLETED OR step3 COMPLETED', 'END'); END; / -- Enabling Chains BEGIN DBMS_SCHEDULER.ENABLE ('my_chain1'); END; / -- Creating Jobs for Chains -- either use the RUN_CHAIN procedure or create a job of type -- 'CHAIN'. job action refers to chain name -- a step job created for every step BEGIN DBMS_SCHEDULER.CREATE_JOB ( job_name => 'chain_job_1', job_type => 'CHAIN', job_action => 'my_chain1', repeat_interval => 'freq=daily;byhour=13;byminute=0;bysecond=0', enabled => TRUE); END; / -- Dropping Chains BEGIN DBMS_SCHEDULER.DROP_CHAIN ( chain_name => 'my_chain1', force => TRUE); END; / -- Running Chains BEGIN DBMS_SCHEDULER.RUN_CHAIN ( chain_name => 'my_chain1', job_name => 'quick_chain_job',
Page 219
start_steps => 'my_step1, my_step2'); END; / -- Dropping Rules from a Chain BEGIN DBMS_SCHEDULER.DROP_CHAIN_RULE ( chain_name => 'my_chain1', rule_name => 'my_rule1', force => TRUE); END; / -- Disabling Chains BEGIN DBMS_SCHEDULER.DISABLE ('my_chain1'); END; / -- Dropping Chain Steps BEGIN DBMS_SCHEDULER.DROP_CHAIN_STEP ( chain_name => 'my_chain2', step_name => 'my_step2', force => TRUE); END; / -- Altering Chain Steps BEGIN DBMS_SCHEDULE.ALTER_CHAIN ( chain_name => 'my_chain1', step_name => 'my_step3', attribute => 'SKIP', value => TRUE); END; / -- for a running step BEGIN DBMS_SCHEDULER.ALTER_RUNNING_CHAIN ( job_name => 'my_job1', step_name => 'my_step1', attribute => 'PAUSE', value => TRUE); END; / -- Handling Stalled Chains -- option1: alter the state of one of its steps with the -- check ALL_SCHEDULER_RUNNING_CHAINS and ALL_SCHEDULER_CHAIN_RULES BEGIN DBMS_SCHEDULER.ALTER_RUNNING_CHAIN ( job_name => 'my_job1', step_name => 'my_step1', attribute => 'SUCCEEDED', value => TRUE); END;
Page 220
/ -- option2: -- if one or more rules are incorrect, you can use the -- DEFINE_CHAIN_RULE procedure to replace them -- check ALL_SCHEDULER_CHAIN_RULES -- After adding or updating rules, you must run EVALUATE_RUNNING_CHAIN
Allocating Resources Among Jobs

For jobs, resource allocation is specified by associating a job class with a consumer group, or by associating a job class with a database service name and mapping that database service to a consumer group. BEGIN DBMS_SCHEDULER.CREATE_JOB_CLASS ( job_class_name => 'finance_jobs', resource_consumer_group => 'finance_group'); END; /
Administering Oracle Scheduler

Configuring Oracle Scheduler -- Setting Scheduler Privileges GRANT SCHEDULER_ADMIN TO username; -- create jobs, schedules, or programs GRANT CREATE JOB TO scott; -- alter, or drop windows, job classes, or window groups GRANT MANAGE SCHEDULER TO adam; -- Setting Chain Privileges -- create a chain in his own schema, create rules, rule sets, -- and evaluation contexts in his own schema BEGIN DBMS_RULE_ADM.GRANT_SYSTEM_PRIVILEGE(DBMS_RULE_ADM.CREATE_RULE_OBJ, 'username'); DBMS_RULE_ADM.GRANT_SYSTEM_PRIVILEGE (DBMS_RULE_ADM.CREATE_RULE_SET_OBJ, 'username'); DBMS_RULE_ADM.GRANT_SYSTEM_PRIVILEGE (DBMS_RULE_ADM.CREATE_EVALUATION_CONTEXT_OBJ, 'username'); END; / -- Setting Scheduler Attributes -- default_timezone select dbms_scheduler.stime from dual; DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE('default_timezone','US/Eastern'); SELECT DISTINCT TZNAME FROM V$TIMEZONE_NAMES; -- log_history DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE('log_history','90'); -- max_job_slave_processes (default NULL) -- event_expiry_time SET SERVEROUTPUT ON DECLARE
Page 221
V VARCHAR2(200); BEGIN DBMS_SCHEDULER.GET_SCHEDULER_ATTRIBUTE ( attribute =>'EVENT_EXPIRY_TIME', value =>V); DBMS_OUTPUT.PUT_LINE(nvl(V,'24 hours')); -- 24 hours if null END; / declare n number := 48*60*60; begin DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE( attribute =>'EVENT_EXPIRY_TIME', value =>n); -- in seconds (24 hours if NULL) end; / Monitoring and Managing the Scheduler -- Viewing the Currently Active Window and Resource Plan SELECT WINDOW_NAME, RESOURCE_PLAN FROM DBA_SCHEDULER_WINDOWS WHERE ACTIVE='TRUE';
-- Finding Information About Currently Running Jobs SELECT JOB_NAME, STATE FROM DBA_SCHEDULER_JOBS WHERE JOB_NAME = 'MY_EMP_JOB1'; SELECT JOB_NAME , STATE, START_DATE, END_DATE, LAST_START_DATE, NEXT_RUN_DATE, LAST_RUN_DURATION, FAILURE_COUNT FROM USER_SCHEDULER_JOBS ORDER BY START_DATE DESC; -- progress of currently running jobs SELECT * FROM ALL_SCHEDULER_RUNNING_JOBS; -- job that is part of a running chain SELECT * FROM ALL_SCHEDULER_RUNNING_CHAINS WHERE JOB_NAME='MY_JOB1';
-- Monitoring and Managing Window and Job Logs SELECT JOB_NAME, OPERATION, OWNER FROM DBA_SCHEDULER_JOB_LOG;
-- Job Run History Details select log_id, job_name, status, to_char(log_date, 'DD-MON-YYYY HH24:MI') log_date from dba_scheduler_job_run_details where job_name = 'MY_JOB14';
-- Controlling Job Logging: at job or class levels logging_level attribute in the CREATE_JOB_CLASS: DBMS_SCHEDULER.LOGGING_OFF, DBMS_SCHEDULER.LOGGING_RUNS, DBMS_SCHEDULER.LOGGING_FULL
Page 222
DBMS_SCHEDULER.SET_ATTRIBUTE ( 'mytestjob', 'logging_level', DBMS_SCHEDULER.LOGGING_FULL);
-- Window Logs SELECT LOG_ID, TO_CHAR(LOG_DATE, 'MM/DD/YYYY'), WINDOW_NAME, OPERATION FROM DBA_SCHEDULER_WINDOW_LOG; SELECT LOG_ID, WINDOW_NAME, ACTUAL_START_DATE, ACTUAL_DURATION FROM DBA_SCHEDULER_WINDOW_DETAILS;
-- Purging Logs DBMS_SCHEDULER.SET_SCHEDULER_ATTRIBUTE('log_history','90'); -- days -- can be specified at job class create DBMS_SCHEDULER.SET_ATTRIBUTE('class2','log_history','30'); -- Purging Logs Manually DBMS_SCHEDULER.PURGE_LOG(); -- urges all entries from the jog log that are -- older than three days. The window log is not affected DBMS_SCHEDULER.PURGE_LOG(log_history => 3, which_log => 'JOB_LOG'); DBMS_SCHEDULER.PURGE_LOG(log_history => 10, job_name => 'job1, sys.class2');
-- Changing Job Priorities -- 1(high)-5 BEGIN DBMS_SCHEDULER.SET_ATTRIBUTE ( name => 'my_emp_job1', attribute => 'job_priority', value => 1); END; / SELECT JOB_NAME, JOB_PRIORITY FROM DBA_SCHEDULER_JOBS;
-- Monitoring Running Chains SELECT * FROM USER_SCHEDULER_RUNNING_CHAINS WHERE JOB_NAME = 'MY_CHAIN_JOB'; Enabling, Using and Disabling Remote External Jobs -- 1) Setting Up the Database conn sys -- verify that the XML DB option is installed: DESC RESOURCE_VIEW -- Enable HTTP connections to the database: -- port between 1 and 65536, and for UNIX and Linux greater than 1023 BEGIN DBMS_XDB.SETHTTPPORT(port); END; / -- run script SQL> @?/rdbms/admin/prvtrsch.plb -- Set a registration password BEGIN DBMS_SCHEDULER.SET_AGENT_REGISTRATION_PASS('mypassword'); END;
Page 223
/ --------------2) Installing, Configuring, and Starting the Scheduler Agent install Scheduler agent from Oracle Database Gateways in the Database Media Pack in its own HOME To install, configure, and start the Scheduler agent on a remote host: Log in to the remote host as oracle Run the Oracle Universal Installer -> Next -> select Oracle Scheduler Agent -> Next -> Home -> eneter hostname, port -> install -> run script_path/root.sh as root -> Exit cehck the agent configuration parameters in the file: schagent.conf
-- Register the Scheduler agent with a database that is to run remote external jobs on -- the agent's host computer. Use the following command: AGENT_HOME/bin/schagent -registerdatabase db_host db_http_port where: db_host is the host name or IP address of the host on which the database resides. db_http_port is the port number that the database listens on for HTTP connections. SELECT DBMS_XDB.GETHTTPPORT() FROM DUAL; -- zero means disabled Repeat the previous step for each database that is to run remote external jobs on the agent's host. Start the Scheduler agent with the following command: AGENT_HOME/bin/schagent -start -- 3) Stopping the Scheduler Agent On UNIX and Linux: AGENT_HOME/bin/schagent -stop On Windows, stop the service OracleSchedulerExecutionAgent -- 4) Disabling Remote External Jobs DROP USER REMOTE_SCHEDULER_AGENT CASCADE; Registration of new scheduler agents and execution of remote external jobs is disabled until you run prvtrsch.plb again.
/* to create a remote external job */ -- create a credential object exec dbms_scheduler.create_credential('hrcredential','hr','hrpassword'); -- Grant privileges grant execute on system.hrcrdential to someuser; -- create the remote external job: begin dbms_scheduler.create_job( job_name => 'remove_logs', job_type => 'executable',
Page 224
job_action => '/u01/app/oracle/logs/removelogs', repeat_interval => 'freq=daily; byhour=23', enabled => false); end; / -- set the CREDENTIAL_NAME attribute exec dbms_scheduler.set_attribute('remove_logs','credential_name','hrcredential'); -- set the DESTINATION attribute exec dbms_scheduler.set_attribute('remove_logs','destination', 'localhost.localdomain:1521'); --enable the external job exec dbms_scheduler.enable('remove_logs'); Import/Export and the Scheduler You must use the Data Pump utilities (impdp and expdp) to export Scheduler objects. After you import Scheduler credentials, you must reset the passwords using the SET_ATTRIBUTE procedure of the DBMS_SCHEDULER package.
Scheduler Privileges o o o o o o o o o System Privileges: CREATE JOB CREATE ANY JOB CREATE EXTERNAL JOB EXECUTE ANY PROGRAM EXECUTE ANY CLASS MANAGE SCHEDULER Object Privileges: on jobs, programs, chains, schedules and job classes. EXECUTE ALTER ALL
Scheduler Data Dictionary Views DBA_SCHEDULER_CHAINS DBA_SCHEDULER_CHAIN_RULES DBA_SCHEDULER_CHAIN_STEPS DBA_SCHEDULER_CREDENTIALS DBA_SCHEDULER_GLOBAL_ATTRIBUTE DBA_SCHEDULER_JOBS DBA_SCHEDULER_JOB_ARGS DBA_SCHEDULER_JOB_CLASSES DBA_SCHEDULER_JOB_LOG DBA_SCHEDULER_JOB_ROLES DBA_SCHEDULER_JOB_RUN_DETAILS DBA_SCHEDULER_PROGRAMS DBA_SCHEDULER_PROGRAM_ARGS DBA_SCHEDULER_REMOTE_DATABASES Oracle DBA Code Examples
Page 225
DBA_SCHEDULER_REMOTE_JOBSTATE DBA_SCHEDULER_RUNNING_CHAINS DBA_SCHEDULER_SCHEDULES DBA_SCHEDULER_WINDOWS DBA_SCHEDULER_WINDOW_DETAILS DBA_SCHEDULER_WINDOW_GROUPS DBA_SCHEDULER_WINDOW_LOG DBA_SCHEDULER_WINGROUP_MEMBERS
Using the UTL_FILE Package CREATE DIRECTORY MYDIR AS '/home/oracle/'; GRANT READ, WRITE ON DIRECTORY utl_dir to .. | public; DECLARE fHandle UTL_FILE.FILE_TYPE; v_username dba_users.username%TYPE; CURSOR users IS SELECT username FROM dba_users order by username; n number := 0; BEGIN -- options: r w a fHandle := UTL_FILE.FOPEN('MYDIR','utlfile.txt','w'); UTL_FILE.PUT_LINE(fHandle,'SN'||CHR(9) || 'User NAME'); UTL_FILE.FCLOSE(fHandle); /* re-Open the utlfile.txt for append, and get its file handle */ fHandle := UTL_FILE.FOPEN('MYDIR','utlfile.txt','a'); OPEN users; LOOP FETCH users INTO v_username; EXIT when users%NOTFOUND; n := n + 1; /* Write a line of text to the file utlfile.txt */ UTL_FILE.PUT_LINE(fHandle,n || chr(9) || v_username); /* Read a line from the file utltext.txt */ -- UTL_FILE.GET_LINE(fHandle,v_username||v_failed||v_life||v_lock); END LOOP; CLOSE users; UTL_FILE.FCLOSE(fHandle); EXCEPTION WHEN UTL_FILE.INVALID_PATH THEN RAISE_APPLICATION_ERROR(-20100,'Invalid Path'); WHEN UTL_FILE.INVALID_MODE THEN RAISE_APPLICATION_ERROR(-20101,'Invalid Mode'); WHEN UTL_FILE.INVALID_OPERATION then RAISE_APPLICATION_ERROR(-20102,'Invalid Operation'); WHEN UTL_FILE.INVALID_FILEHANDLE then RAISE_APPLICATION_ERROR(-20103,'Invalid Filehandle'); WHEN UTL_FILE.WRITE_ERROR then RAISE_APPLICATION_ERROR(-20104,'Write Error'); WHEN UTL_FILE.READ_ERROR then RAISE_APPLICATION_ERROR(-20105,'Read Error'); WHEN UTL_FILE.INTERNAL_ERROR then RAISE_APPLICATION_ERROR(-20106,'Internal Error'); WHEN OTHERS THEN UTL_FILE.FCLOSE(fHandle);
Page 226
END; /
Page 227
Data Loading and Transforming Tools

Oracles ETL (Extraction-Transformation-Loading) solution includes the following components: SQL*Loader: see the section Using SQL*Loader Utility in Oracle Database Utilities part. External tables: see Managing External Tables Multitable inserts Merging Data Table functions: Table functions produce a set of rows as output. Instead of defining the transform declaratively in SQL, you define it procedurally in PL/SQL. See Table Functions. Transportable tablespaces: see Transporting Tablespaces Between Databases Oracle Warehouse Builder (OWB): OWB offers you a wizard-driven facility to load data into the database through SQL*Loader, load data from an Oracle database or other databases such as Sybase, Informix, and Microsoft SQL Server via Oracle Transparent Gateways.
Page 228
Using Database Links

/* Info about DB Links */ select OWNER, DB_LINK, USERNAME, HOST, CREATED from DBA_DB_LINKS;
/* Privs */ grant CREATE DATABASE LINK to hr; grant CREATE PUBLIC DATABASE LINK to hr;
/* Private */ CONNECT system/system_passwd@mydb CREATE DATABASE LINK MONITOR CONNECT TO hr IDENTIFIED BY hr USING 'monitor'; Create database link orcl2.net connect to sa identified by a using '(DESCRIPTION=(ADDRESS = (PROTOCOL = TCP)(HOST =10.4.x.x) (PORT=1521)) (connect_data=(service_name=orcl)))'; /* Public */ CREATE PUBLIC DATABASE LINK MONITOR CONNECT TO hr IDENTIFIED BY hr USING 'monitor';
Page 229
Managing Diagnostic Data

Setting the Automatic Diagnostic Repository Directory
It replaces USER_DUMP_DEST, BACKGROUND_DUMP_DEST and CORE_DUMP_DEST parameters. Default value: $ORACLE_BASE/diag/rdbms/$INSTANCE_NAME/$ORACLE_SID If you havent set the ORACLE_BASE variable, the value of the DIAGNOSTIC_DEST parameter defaults to $ORACLE_HOME/log.
show parameter DIAGNOSTIC_DEST SELECT VALUE FROM V$PARAMETER WHERE NAME ='diagnostic_dest'; ALTER SYSTEM SET DIAGNOSTIC_DEST ='C:\ORACLE\diag'; SELECT NAME, VALUE FROM V$DIAG_INFO;
Using adrci Tool

General usage of adrci adrci help adrci>help adrci>help show incident -- running adrci in batch mode adrci exec 'command [; comamnd]. . . ' adrci script=file_name
adrci>show base adrci>show homes adrci>show homepath adrci>set homepath diag\rdbms\ora11g\ora11g
-- spooling adrci>spool /u01/myfiles/myadrci.txt adrci> ... adrci>spool off
-- view alert log set editor notepad.exe show alert
Page 230
show alert tail 30 show alert p "MESSAGE TEXT LIKE '%ORA-600%'*
-- List Trace Files show tracefile
-- View Incidents show incident show incident mode detail p "incident_id=112564" Using adrci to Package Incidents
With adrci tool, you can package all the diagnostic files related to specific problems into a ZIP file to submit it to Oracle support. To do so, you use special commands called IPS as shown in the following steps: 1. Create a logical package: use ips create package command to create an empty logical package as shown in the example below. The package will be given a serially generated number. adrci>ips create package 2. Add diagnostic data to the logical package: this is done by ips add incident command as shown below: adrci>ips add incident 112564 package 1 Actually, there are formats of the ips create package command which enables you to perform the steps 1 and 2 in one command. Following are those command formats: o ips create package problem o o o ips create package problem key ips create package incident ips create package time
3. Generate the physical package. The files related to the incident will be collected in a ZIP file. The following example shows the command to perform this task: adrci>ips generate package 1 in /u01/myfiles/incidents If you decide to add or change any diagnostic data later, you can do so by generating an incremental ZIP file. Modify the command as follows to achieve that: adrci>ips generate package 1 in /u01/myfiles/incidents incremental You will notice that the generated file has the phase INC in its name indicating that it is an incremental ZIP file. ips commands behavior is controlled by various configuration options. To display those configuration options, use the command ips show configuration.
Managing Database Health Monitor

To display list of the check that can be performed, issue the following query: SELECT NAME, DESCRIPTION, OFFLINE_CAPABLE FROM V$HM_CHECK;
Page 231
The OFFLINE_CAPABLE column defines whether you can perform the check when the database is offline or not. Running Health Checks Using the DBMS_HM A DBA can use DBMS_HM to manually invoke the database check. To retrieve the list of checks that can be run manually by users, issue the following query: SELECT NAME FROM V$HM_CHECK WHERE INTERNAL_CHECK = 'N'; Use the procedure RUN_CHECK to perform a database health check. Its first parameter CHECKNAME is mandatory and it takes one of the returned names by the query above. exec DBMS_HM.RUN_CHECK(CHECK_NAME=>'DB Structure Integrity Check', RUN_NAME=>'HM01'); Most health checks accept input parameters. You can view parameter names and descriptions with the V$HM_CHECK_PARAM view. Some parameters are mandatory while others are optional. The following query displays parameter information for all health checks: select C.NAME CHECK_NAME, P.NAME PARAMETER_NAME, P.TYPE, P.DEFAULT_VALUE, P.DESCRIPTION from V$HM_CHECK_PARAM P, V$HM_CHECK C where P.CHECK_ID = C.ID and C.INTERNAL_CHECK = 'N' order by C.NAME; Input parameters are passed to the INPUT_PARAMS argument of the RUN_CHECK procedure as name/value pairs separated by semicolons (;). The following example illustrates how to pass the transaction ID as a parameter to the Transaction Integrity Check: begin DBMS_HM.RUN_CHECK ( CHECK_NAME => ' Transaction Integrity Check', sensitive RUN_NAME => 'MY_RUN', INPUT_PARAMS => 'TXN_ID=7.33.2'); end;
-- passed value is case
Database Health checks executions are stored in ADR and can be viewed by either querying the V$HM_RUN: SELECT * FROM V$HM_RUN; Another option is to run the adrci command show hm_run: adrci>show hm_run You can view a report of a particular Health check by using the following adrci command: adrci>show report hm_run HM01 Alternatively, you can DBMS_HM package as shown in the following code example: declare v_rpt clob; begin v_rpt := DBMS_HM.GET_RUN_REPORT('HM01'); end; Findings, if any, detected by the checks can be obtained from V$HM_FINDING and recommendations from V$HM_RECOMMENDATION. Running Health Checks Using the Enterprise Manager After connecting as SYSDBA, under the Advisor Central page, you will see Checkers link which can be used to manually invoke any Health check.
Page 232
Managing Data Recovery Advisor

Data Recovery Advisor is an Oracle Database 11g tool that automatically diagnoses data failures, determines and presents appropriate repair options, and executes repairs at the user's request. Data Recovery Advisor can diagnose failures such as the following: Inaccessible components like datafiles and control files. Physical corruptions such as block checksum failures and invalid block header Field values Inconsistent datafiles (online and offline) I/O failures
The advisor however doe not recover from failures on standby databases or RAC environment. This advisor can be used through RMAN or the Enterprise Manager. Using Data Recovery Advisor with RMAN Following are the RMAN commands to use Data Recovery Advisor: 1. List failures by running the LIST FAILURE command. Following are variations of using the command: RMAN>LIST FAILURE; RMAN>LIST OPEN; RMAN>LIST CLOSED; 2. Optionally, execute LIST FAILURE ... DETAIL to list details of an individual failure. RMAN>LIST FAILURE 105 DETAIL; 3. If you suspect that failures exist that have not been automatically diagnosed by the database, then run VALIDATE DATABASE to check for corrupt blocks and missing files. If a failure is detected, then RMAN logs it into the ADR, where it can be accessed by the Data Recovery Advisor. 4. Determine repair options by running the ADVISE FAILURE command. RMAN>ADVISE FAILURE; 5. Choose a repair option. You can repair the failures manually or run the REPAIR FAILURE command to fix them automatically. By default, the REPAIR FAILURE command prompts the user to confirm the repair, but this can be prevented using the NOPROMPT keyword. Be aware that the previous command must be issued before using REPAIR FAILURE command. The following form of the command informs you how RMAN plans to repair the failure: RMAN>REPAIR FAILURE PREVIEW 6. You may wish to change the priority of a failure (to HIGH or LOW), if it does not represent a problem to you, or even manually close it. This can be done by the CHANGE FAILURE command: RMAN> CHANGE FAILURE 202 PRIORITY LOW;
Note
Data Recovery Advisor may detect or handle some logical corruptions. But in general, corruptions of this type require help from Oracle Support Services.
Using Data Recovery Advisor with the Enterprise Manager Access the Data Recovery Advisor in the Enterprise Manager by following the links: Availability> Manage> Perform Recovery> Perform Automated Repair
Using SQL Test Case Builder

The SQL Test Case Builder aims at capturing the information pertaining to a SQL-related problem, along with the exact environment under which the problem occurred, so that the problem can be reproduced Page 233
and tested on a separate Oracle database instance. Once the test case is ready, you can upload the problem to Oracle Support to enable support personnel to reproduce and troubleshoot the problem. The information gathered by SQL Test Case Builder includes the query being executed, table and index definitions (but not the actual data), PL/SQL functions, procedures, and packages, optimizer statistics, and initialization parameter settings. The output of the SQL Test Case Builder is a SQL script that contains the commands required to recreate all the necessary objects and the environment. Accessing SQL Test Case Builder Using DBMS_SQLDIAG The DBMS_SQLDIAG has a procedure named EXPORT_SQL_TESTCASE which is used to generate a SQL test case for a given SQL statement, SQL Id (taken from V$SQL) or an incident id. Following steps should be followed: 1. Create directory to hold the SQL test case files. CREATE DIRECTORY sql_tes_dir AS 'C:\Oracle\TestCase'; 2. Execute the proper form of EXPORT_SQL_TESTCASE procedure. Following is an example using a passed SQL statement. DECLARE V_SQL CLOB := 'SELECT * FROM HR.NAMES WHERE ID BETWEEN 100 AND 1000'; V_TESTCASE CLOB; BEGIN DBMS_SQLDIAG.EXPORT_SQL_TESTCASE ( DIRECTORY =>'SQL_TES_DIR', SQL_TEXT =>V_SQL, USER_NAME =>'HR', BIND_LIST =>NULL, EXPORTENVIRONMENT =>TRUE, EXPORTMETADATA =>TRUE, EXPORTDATA =>TRUE, SAMPLINGPERCENT =>100, CTRLOPTIONS =>NULL, TIMELIMIT =>0, TESTCASE_NAME =>'RETURN_NAMES', -- generated scripts prefix TESTCASE =>V_TESTCASE); END;
Accessing SQL Test Case Builder Using the Enterprise Manager From Enterprise Manager, the SQL Test Case Builder is accessible only when a SQL incident occurs. SQLrelated problem is referred to as a SQL incident. To access the SQL Test Case Builder, follow the links the Support Workbench page> Click on an incident ID> Investigate and Resolve section> Oracle Support> Generate Additional Dumps and Test Cases> click on the icon in the Go To Task.
Page 234
Patching Oracle Products

Using Oracle Opatch
References Oracle Universal Installer and OPatch User's Guide On metalink: OPatch Utility Guide - 10.2 [ID 554417.1] How to Download It On metalink, search Patch 6880880. At document writing, it is on: https://updates.oracle.com/ARULink/PatchDetails/process_form?patch_num=6880880 Environment Variables OPatch Uses ORACLE_HOME - Oracle home location. OPATCH_DEBUG - Log level that specifies the amount of logging OPatch should perform. OPATCH_PLATFORM_ID - Unique platform ID. PATH - Path information. Backup Recommendations It is highly recommended that you back up the ORACLE_HOME before any patch operation. You can use any method, such as zip, cp -r, tar, and cpio. Info about a Utility To find out more about the options available with each command, please use the following syntax: opatch util <Utility_Name> -help lsinventory It is used to check what is currently installed on the system. opatch lsinventory -detail to list all available Oracle homes: opatch lsinventory -all
NApply It is used to apply a set of patches under a directory. opatch napply <patch_location> -id 1,2,3 -skip_subset -skip_duplicate This will apply patches 1, 2, and 3 which are under < the patch_location> directory. OPatch will skip duplicate patches and subset patches (patches under <patch_location> that are subsets of patches installed in the ORACLE_HOME) NRollback It is used to rollback a set of patches installed in an ORACLE_HOME. You can invoke the command using "opatch nrollback" or "opatch util nrollback". opatch nrollback -id 1,2,3 UpdateRemoteNodes (RAC) It is used to propagate/remove files/directories to/from remote nodes using files under ORACLE_HOME/.patch_storage/<ID>/rac/*. The directories listed in copy_dirs.txt will be copied to remote nodes. The files listed in copy_files.txt wil be copied to remote nodes. The directories listed in remove_dirs.txt will be deleted from remote nodes.
Page 235
The files listed in remove_files.txt will be deleted from remote nodes. Cleanup It is used to clean up 'restore.sh, make.txt' files and 'rac, scratch, backup' directories in the ORACLE_HOME/.patch_storage directory. If -ps option is used, then it cleans the above specified areas only for that patch, else for all patches under ORACLE_HOME/.patch_storage. You will be still able to rollback patches after this cleanup. opatch util cleanup -ps 6121193_Jun_21_2008_04_19_82 CopyListedFiles (RAC) It is used to copy all files listed in ORACLE_HOME/.patch_storage/<ID>/rac/copy_files.txt to remote nodes. If -fp option is used, then one can specify the path of the file containing the list of files to be copied. The files mentioned in this file will be copied to the remote nodes. opatch util copylistedfiles -fp a -remote_nodes ceintcb-a5 CopyListedFilesTest (RAC) It is used to copy a single file to remote nodes. The usage remains the same as CopyListedFiles. opatch util copylistedfilestest -fp /home/oracle/a -remote_nodes ceintcb-a5 CopyListedDirs (RAC) It is used to recursively copy all directories listed in ORACLE_HOME/.patch_storage/<ID>/rac/copy_dirs.txt to remote nodes. If -dp option is used, then one can specify the path of the file containing the list of directories to be copied. The directories mentioned in this file will be copied to the remote nodes. CopyListedDirsTest (RAC) It is used to copy a single file to remote nodes. The usage remains the same as CopyListedDirs. RemoveListedFiles (RAC) It is used to remove files listed in ORACLE_HOME/.patch_storage/<ID>/rac/remove_files.txt on remote nodes. If -fr option is used, then one can specify the path of the file containing the list of files to be removed. The files mentioned in this file will be removed from the remote nodes. RemoveListedFilesTest (RAC) It is used to remove a single file from remote nodes. The usage remains the same as RemoveListedFiles. RemoveListedDirs (RAC) It is used to recursively remove directories listed in ORACLE_HOME/.patch_storage/<ID>/rac/remove_dirs.txt from remote nodes. If -dr option is used, then one can specify the path of the file containing the list of directories to be removed. The directories mentioned in this file will be removed from the remote nodes. RemoveListedDirsTest (RAC) It is used to remove a single directory from remote nodes. The usage remains the same as RemoveListedDirs. RunLocalMake It is used to invoke re-link on the local node. The make commands are stored in ORACLE_HOME/.patch_storage/<ID>/make.txt. You need to use the -ps option to specify the Patch ID with timestamp. A directory by this name will be present under ORACLE_HOME/.patch_storage. The make.txt file present under ORACLE_HOME/.patch_storage/Patch ID with timestamp/ will be used to perform the local make operation. This command cannot be run if you have already run Cleanup as it would have removed these make.txt files.
Page 236
opatch util runlocalmake -ps 6121250_Jun_21_2007_04_16_11 RunRemoteMake (RAC) It is used to invoke re-link on remote nodes. The make commands are stored in ORACLE_HOME/.patch_storage/<ID>/rac/makes_cmd.txt. The usage remains the same as RunLocalMake. RunAnyCommand (RAC) It is used to run any command on remote nodes. The command should be specified using the -cmd option. opatch util runanycommand -remote_nodes ceintcb-a5 -cmd ls Verify It is used to run the patch verification process to ensure that the patch was applied to the ORACLE_HOME. It uses the defined ORACLE_HOME and the given patch location via -ph, to run the check. opatch util verify -ph ~/6646853/6121183 LoadXML It is used to check the validity of an XML file. The -xmlInput option can be used to specify the path of the xml file. opatch util loadxml -xmlInput $ORACLE_HOME/inventory/ContentsXML/comps.xml
Page 237
Part 2
Oracle Database Net Services
Page 238
Connectivity Naming Methods

The Local Naming Method
The TNS_ADMIN environment variable tells Oracle where to locate tnsnames.ora and sqlnet.ora files. /* Starting Oracle Net Configuration Assistant */ export DISPLAY=172.16.14.15:0.0 netca
The Easy Connect Naming Method

The EZCONNECT keyword should be in the NAMES.DIRECTORY_PATH variable in the sqlnet.ora file You cant use any advanced features of Oracle networking such as connection pooling, external procedure calls, or Heterogeneous Services. sqlplus system/[email protected]:1521/mydb.myorg.org # the default port number is 1521 sqlplus system/[email protected]/mydb.myorg.org
The External Naming Method

The external naming method uses external naming services such as the Network Information Service (NIS), originally developed by Sun Microsystems, to resolve net service names. 1. Have your system administrator configure NIS if it isnt already in place. 2. Create a tnsnames.ora file as you would in the local naming method. 3. Convert the tnsnames.ora file to a tnsnames map, which youll need for the NIS server later on. You can derive the tnsnames map from the tnsnames.ora file by having your system administrator use the tns2nis command, as shown here: # tns2nis tnsnames.ora 4. Copy the tnsnames map file to the server on which the NIS is running. 5. Install the tnsnames map file on the NIS server using the makedbm NIS program, as shown here: # makedbm tnsnames /var/yp/'domainname'/tnsnames 6. Test the NIS installation of the tnsnames map by using the following command: # ypmatch net_service_name tnsnames You should get a confirmation back in the following form: description=(address=(protocol=tcp) (host=host_name)(port=port_number))) (connect_data=(service_name=service_name)))
Page 239
7. Edit the sqlnet.ora file as follows: NAMES_DIRECTORY_PATH=(nis, hostname, tnsnames) The nis method should be listed first inside the brackets so that Oracle Net will attempt to resolve the service name using NIS first. Apart from that, the order of the items in the brackets doesnt matter.
The Directory Naming Method

The directory naming method stores database connection information in a Lightweight Directory Access Protocol (LDAP)compliant directory server (like Oracle Internet Directory). The connect identifiers are stored under an Oracle context that contains entries for use with OID.
Database Resident Connection Pooling (DRCP)

DRCP (11g) is especially designed to help architectures such as PHP with the Apache server, that cant take advantage of middle-tier connection pooling because they used multiprocess single-threaded application servers. DRCP enables applications such as these to easily scale up to server connections in the tens of thousands. DRCP is controlled by the following configuration parameters:
INACTIVITY_TIMEOUT MAX_LIFETIME_SESSION MAX_USE_SESSION maximum idle time for a pooled server before it is terminated. time to live TTL duration for a pooled session. maximum number of times a connection can be taken and released to the pool. the maximum and minimum number of pooled servers in the connections pool. pool would increment by this number of pooled server when pooled server are unavailable at application request time. maximum time of inactivity by the client after getting a server from the pool. If the client does not issue a database call after grabbing a server from the pool, the client will be forced to relinquish control of the pooled server and will get an error. The freed up server may or may not be returned to the pool. turn on SESSION_CACHED_CURSORS for all connections in the pool. This is an existing initialization parameter
MAX_SIZE and
MIN_SZIE
INCRSIZE
MAX_THINK_TIME
SESSION_CACHED_CURSORS
/* Enabling and Disabling DRCP */ conn sys as sysdba -- the ramins open after DB restart exec dbms_connection_pool.start_pool(); select connection_pool, status, maxsize from dba_cpool_info; exec dbms_connection_pool.stop_pool(); -- specify using DRCP -- in EZCONNECT method (.Net 11g) myhost.comany.com:1521/mydb.company.com:POOLED -- tnsnames mydb = (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp) (HOST=myhost.company.com) (SERVER=POOLED))) Oracle DBA Code Examples
Page 240
/* Configuring DRCP */ begin DBMS_CONNECTION_POOL.ALTER_PARAM( PARAM_NAME =>'INACTIVITY_TIMEOUT', PARAM_VALUE=>'3600'); end; / -- restore parameter values to their defaults exec dbms_connection_pool.restore_defaults() /* Monitor DRCP */ SELECT STATUS,MINSIZE,MAXSIZE,INCRSIZE,SESSION_CACHED_CURSORS,INACTIVITY_TIMEOUT FROM DBA_CPOOL_INFO; SELECT NUM_OPEN_SERVERS, NUM_BUSY_SERVERS, NUM_REQUESTS, NUM_HITS NUM_MISSES, NUM_WAITS, NUM_PURGED, HISTORIC_MAX FROM V$CPOOL_STATS; -- class-level stats Select * From V$CPOOL_CC_STATS
Page 241
Oracle and Java Database Connectivity

Establishing Database Connectivity
o o o o JDBC driver types: JDBC thin driver JDBC OCI driver JDBC server-side thin driver JDBC server-side internal driver
-- load the drivers -- method 1 DriverManager.registerDriver ("new oracle.jdbc.OracleDriver()"); -- method 2 Class.forName("oracle.jdbc.driver.OracleDriver"); -- establish the connection connection conn=DriverManager.getConnection( "jdbc:oracle:thin:@prod1:1521:finprod", username, passwd); -- Creating the Statement Object statement stmt = conn.createStatement(); -- Handling Queries string first_name,last_name; number salary; resultSet rs = stmt.executeQuery("SELECT * FROM Employees"); while (rs.next()) { first_name = rs.getString("first_name"); last_name = rs.getString("last_name"); salary = rs.getNumber("salary"); system.out.println(first_name + last_name +" with salary of:" + salary); } -- Handling DDL and Nonquery DML Statements statement stmt = conn.createStatement(); stmt.executeUpdate("CREATE TABLE Emp" + "(last_name VARCHAR2(30), first_name VARCHAR2(20), salary number"); -- auto committed by default stmt.executeUpdate("INSERT INTO Emp " + "VALUES ('Lname', 'Fname', salary)"); -- transaction control conn.setAutoCommit(false); conn.commit(); conn.rollback(); -- Error Handling
Page 242
try { conn.setAutoCommit(false); stmt.executeUpdate(".."); conn.commit(); conn.setAutoCommit(true); } catch(SQLException ex) { system.err.println("SQLException: " + ex.getMessage()); conn.rollback(); conn.setAutoCommit(true); }
Page 243
Miscellaneous Connectivity Options

Setting the Default Connect String
-- in unix export TWO_TASK=mydb -- in windows: set LOCAL key to mydb value in the registry or: SET LOCAL=<mydb> -- later you don't have to specify the connect string sqlplus scott/tiger sqlplus [ [<option>] [<logon>] [<start>] ]
Installing the Instant Client

Oracles new Instant Client software allows you to run your applications without installing the standard Oracle Client or having an ORACLE_HOME. (1) Download and install Oracle Instant Client software. You must install the Basic client package and tehn you can also include any of the advanced optional packages. http://www.oracle.com/technology/software/tech/oci/instantclient/index.html (2) Unzip the downloaded package and copy it wherever you like. (3) In Linux, set LD_LIBRARY_PATH to instantclient directory. On Windows, set PATH to instantclient
Setting Listener Options

/* Queue Size */ -- concurrent connection requests can be made LISTENER= (DESCRIPTION= (ADDRESS=(PROTOCOL=tcp)(HOST=myserver)(PORT=1521)(QUEUESIZE=10))) /* Password */ -- ask for a password before executing any admin command LSNRCTL> set password LSNRCTL> change_password
Setting Access Controls

-- in the sqlnet.ora -- only the addresses in the list are allowed to make connections tcp.validnode_checking = yes tcp.invited_nodes = (server1.us.mycompany.com,172.14.16.152) -- addresses in the list are excluded
Page 244
tcp.excluded_nodes = (server1.us.mycompany.com,172.14.16.152)
Changing Windows Hostname

Step 1 Create Hosts Entry for Old Hostname Locate your hosts file, typically located at %WINDIR%\system32\drivers\etc\hosts and add an entry for the old (current) hostname.
# # HOSTS file # win2k3r2 172.16.10.10
Note the IP address this is the address of a Loopback Adapter installed on the guest machine. As outlined by the Oracle Installer, a Loopback Adapter is required on systems that do not have a static IP address (as do virtual machines using NAT, etc.) Step 2 Uninstall Enterprise Manager Console Because there are configuration settings stored with Enterprise Manager Console that reference the hostname, the same must be uninstalled.
emca -deconfig dbcontrol db -repos drop
Note, before executing this command, ensure that the Oracle instance is running it has to be in order for Enterprise Manager Configuration Assistance to drop the repository and de-configure the Console. Step 3 Stop All Oracle Services Once the uninstall of Enterprise Manage Console has completed, stop all Oracle Services on the guest machine. iSQL*Plus Service typically named Oracle<OracleHomeName>iSQL*Plus Oracle Listener Service typically named Oracle<OracleHomeName>TNSListener Oracle Database Instance Service typically named OracleServer<SID>
Step 4 Update listener.ora and tnsnames.ora Once all the Oracle services have stopped, update the listener.ora and tnsnames.ora files, located in %ORACLE_HOME%\network\admin to reflect the desired (new) hostname.
LISTENER = (DESCRIPTION_LIST = (DESCRIPTION = (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1)) (ADDRESS = (PROTOCOL = TCP)(HOST = win2k3r2)(PORT = 1521)) ) ) DEVBOX = (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = win2k3r2)(PORT = 1521)) (CONNECT_DATA =
Page 245
(SERVER = DEDICATED) (SERVICE_NAME = devbox) ) )

Step 5 Rename Host and Restart Now, rename the computer and restart the guest machine. Step 6 Ensure Oracle Instance is Running Once the guest machine has started up, log in and ensure the Oracle instance is running using the following command line (typically required, unless the instance, not the Windows Service, is configured to auto-start.)
oradim -startup sid devbox

Step 7 Reinstall Enterprise Manager Console After ensuring the Oracle instance is running, reinstall Enterprise Manager Console using the following command line:
emca -config dbcontrol db -repos create

Step 8 Validate Enterprise Manager Console Installation Lastly, after the successful installation of Enterprise Manager Console, validate the installation by navigating to the logon page typically http://<hostname>:1158/em/.
Page 246
Part 3 Recovery
Oracle Database Backup and
Page 247
Backup Guidelines
Causes of Unplanned Down Time
Software Failures o Operating system o Database o Middleware o Application o Network Hardware Failures o CPU o Memory o Power supply o Bus o Disk o Tape o Controllers o Network o Power Human Errors o Operator error o User error o DBA o System admin. o Sabotage Disasters o Fire o Flood o Earthquake o Power failure o Bombing
Causes of Planned Down Time

Routine Operations o Backups o Performance mgmt o Security mgmt o Batches Periodic Maintenance o Storage maintenance o Initialization parameters
Page 248
o Software patches o Schema management o Operating system o Middleware o Network New deployments o HW upgrade o OS upgrades o DB upgrades o MidW upgrades o App upgrades o Net upgrades
Oracles Solution to Down Time

Failure Category Unplanned Down Time Failure Sub-Category System Failures Solutions RAC Data Guard Streams Fast-start Fault Recovery RAC Data Guard Streams Data Failures RMAN backup/recovery ASM Flashback Hardware Assisted Resilient Data (HARD) Data Guard and Streams Planned Down Time System Changes Data Changes Rolling upgrades Dynamic provisioning Online redefinition
Minimizing Unplanned Downtime Guidelines

Use RAID data storage with mirroring ( 1+0 is a good choice). Maintain offsite storage of your backups with a reliable vendor. Make is part of your recovery testing program. Normally, for a production database, operating in Archivelog mode is a must. Multiplex the control files on separate disk drives managed by different disk controllers. Oracle strongly recommends that you multiplex the redo log file. After every major structural change, back up the control file. You can take backup of the control file every hour. If you backup to a tape, backup to two copies. The media might be defective. Make auxiliary files part of your backup: (SPFILE) or the init.ora, sqlnet.ora, tnsnames.ora, password and wallet files. Log your backup operations. Make every application has its own tablespace. Use Data Pump utility for supplemental protection. Make a plan to make sure that the backups are actually readable and valid. Oracle DBA Code Examples
Page 249
Make database recovery testing plan. Always keep a redundancy set online (use flash recovery area for this purpose) so you can recover faster. A redundancy set has: o o o o o o Last backup of all datafiles Last backup of the control file Multiplexed copies of the current redo log files Copies of the current control file The archived redo logs since the last backup Auxilary files: SPFILE or the init.ora, listener.ora, and tnsnames.ora, pwsd
SLA Sample
Standard Processing Services. The Provider shall furnish and allow access to the processing environments listed below: a. Mid-tier processing. (1) Applications to be processed: Financial Information Systems (FIS) to include: LIST OF FIS APPLICATIONS Other Departmental Applications (2) Hours of Availability. Interactive: Monday-Friday* 07:00-17:00* Saturday, Sunday and Holidays Not Applicable *Application will be a web-based 247365 system WITH the exception of the scheduled maintenance periods (see below) Batch: Not applicable Maintenance: Monthly, Fourth Weekend of Every Month (3) Standard Processing/Service Requirements. All of the systems/applications listed in paragraph (1) above are required to be operational 98% of the total time listed in paragraph (2) above. The Information Systems Department will provide a method for the Department of Finance to monitor operational percentages. (4) Processing of data will be limited to the functionality/processing that was being conducted at the time of handing over the operations to the Information Services Department. List Database failure possible reasons or scenarios and the time required to recover for each reason.
Planning a Backup Strategy Guidelines

If possible, have your required backup files on disks (faster than tapes). Automatically delete obsolete backups. If DML load makes it feasible, use incremental backups.
Page 250
Examples of Backup Schedules for a Database

/* Example 1 */ # you'll have: image copy of the database (L0), # incremental backup L1 # Archived redo logs between current time and L0 RUN { # apply L1 to the specified datafile images RECOVER COPY OF DATABASE WITH TAG 'incr_update'; # create L0 datafile images (first time) then L1 BACKUP INCREMENTAL LEVEL 1 FOR RECOVER OF COPY WITH TAG 'incr_update' DATABASE; }
/* Example 2 */ # you'll have image copy of db L0, 3 L1s, # archived log between current time and L0 # assumption: Size the flash recovery area so it holds three days worth of incremental backups. RUN { RECOVER COPY OF DATABASE TAG "whole_db_copy" UNTIL TIME 'SYSDATE-3'; BACKUP INCREMENTAL LEVEL 1 FOR RECOVER OF COPY WITH TAG "whole_db_copy" DATABASE; }
Page 251
User-Managed Backups
Obtaining Database File Information
SELECT name, status FROM v$datafile; SELECT t.name tablespace, f.name datafile FROM v$tablespace t, v$datafile f WHERE t.ts# = f.ts# ORDER BY t.name; SELECT name FROM v$controlfile; select member from V$LOGFILE Making Whole Closed Database Backups # 1) shutdown the db # 2) copy all db physical files: datafiles, control files, redo log files, parameter file, password file, wallet file (if there's a one). Ensure that the complete pathnames of the files are noted #!/bin/ksh ORACLE_SID=$1 export ORACLE_SID export ORAENV_ASK=NO BACKUP_DIR=/u02/app/oracle . oraenv sqlplus -s system/mypsswrd << EOF SET HEAD OFF FEED OFF ECHO OFF TRIMSPOOL ON LINESIZE 200 SPOOL /u01/app/oracle/dba/cold_backup.ksh SELECT 'cp ' ||file_name|| ' ${BACKUP_DIR}' from sys.dba_data_files; SELECT 'cp ' ||name || ' ${BACKUP_DIR}' from V$controlfile; SELECT 'cp ' ||member|| ' ${BACKUP_DIR}' from V$logfile; SPOOL OFF; EXIT; EOF # 3) open the db Making a Whole Open Backup #!/bin/ksh ORACLE_SID=$1 export ORACLE_SID export ORACLE_ASK=NO BACKUP_DIR=/u01/app/oracle/backup export BACKUP_DIR sqlplus -s "sys/sys_password as sysdba" << EOF set linesize 200 set head off set feed off SPOOL /u01/app/oracle/dba/hot_backup.ksh BEGIN dbms_output.put_line ('alter database begin backup;'); for f1 in (select file_name fn from sys.dba_data_files)
Page 252
loop dbms_output.put_line( 'host cp '||f1.fn|| ' $BACKUP_DIR'); end loop; dbms_output.put_line ('alter database end backup;'); dbms_output.put_line('alter database backup controlfile to '|| ' $BACKUP_DIR/control'|| ';'); dbms_output.put_line('alter system switch logfile;'); END; / SPOOL OFF; EXIT EOF Making Tablespace Backups # offline ALTER TABLESPACE users OFFLINE; copy datafiles ALTER TABLESPACE users ONLINE; # online ALTER TABLESPACE sysaux BEGIN BACKUP; copy datafiles ALTER TABLESPACE sysaux END BACKUP;
Obtaining Backup Status Information

SELECT * FROM v$backup;
Checking Datafiles Taken as Backup

The utility checks only for logical corruption below the HWM. dbv file=D:\ORACLE\ORADATA\ORA10G\USERS01.DBF dbv file=D:\ORACLE\ORADATA\ORA10G\USERS01.DBF blocksize=4096 blocksize=8192
Handling Crash Before User-Manged Backup Ends

#1) make sure that one or more datafiles left in online backup mode: SELECT * FROM v$backup; #2) ALTER DATABASE END BACKUP; #3) SELECT * FROM v$backup; #4) ALTER DATABASE OPEN;
Page 253
Backing up Control File

# create a text trace file: ALTER DATABASE BACKUP CONTROLFILE TO TRACE # binary copy of the control file ALTER DATABASE BACKUP CONTROLFILE to 'c:\temp\controlfile.ctl'
Backing Up Initialization Files

# create pfile CREATE SPFILE='/u01/oracle/dbs/test_spfile.ora' FROM PFILE='/u01/oracle/dbs/test_init.ora' # In Oracle 11g, write current values of instance parameters CREATE PFILE FROM MEMORY; CREATE SPFILE FROM MEMORY;
Page 254
User-Managed Complete Recovery

User-Managed Recovery in NOARCHIVELOG Mode
#(1 SHUTDOWN ABORT #(2 copy datafiles, control files and redo log files from backup location into the original destinations #(3 STARTUP
User-Managed Recovery in NOARCHIVELOG Mode Without Redo Log File

#(1 SHUTDOWN IMMEDIATE #(2 Restore the most recent whole database backup with operating system commands. #(3 mimic incomplete recovery: SQL>RECOVER DATABASE UNTIL CANCEL USING BACKUP CONTROLFILE; SQL>CANCEL #4) SQL> ALTER DATABASE OPEN RESETLOGS;
User-Managed Complete Recovery in ARCHIVELOG Mode

If the lost file is the SYSTEM, UNDO or SYSAUX, the databse will be mostly closed. # Checking steps: #Identifying datafiles that need recovery SELECT * FROM v$recover_file; SELECT file_id f#, file_name, tablespace_name tablespace, status FROM dba_data_files; # Locating Archived Log Files to Apply SELECT * FROM v$recovery_log; #1) if the database is closed, open mount STARTUP MOUNT #2) make sure the datafile is offline select file#, name , status from v$datafile; ALTER DATABASE datafile '/disk2/data/df2.dbf' offline; #3) restore the damaged or lost datafile using os commands if it should be resotred in a new location:
Page 255
ALTER DATABASE RENAME FILE '/ORADATA/u03/users01.dbf' TO '/ORADATA/u04/users01.dbf'; #4) if archive redo log file are stored in location different from log_archive_dest_n: #4.1) Specifying the location and name at the recover prompt: Specify log: {<RET>=suggested | filename | AUTO | CANCEL} #4.2)Using the ALTER SYSTEM ARCHIVE command: SQL> ALTER SYSTEM ARCHIVE LOG START TO <new location>; #4.3)Using the RECOVER FROM <LOCATION> command: SQL> RECOVER FROM '<new location>' DATABASE #4.4) SQL>SET LOGSOURCE /new_directory #4.5) ALTER DATABASE RECOVER FROM '/new_directory'; #5) recover the datafile(s) RECOVER [AUTOMATIC] DATABASE RECOVER [AUTOMATIC] TABLESPACE <NUMBER> | <NAME> RECOVER [AUTOMATIC] DATAFILE <'filename'> | <NAME> #6) if db is open, take the datafile online: ALTER DATABASE DATAFILE '/disk2/data/df2.dbf' ONLINE; ALTER TABLESPACE user_data ONLINE; #7) if required: ALTER DATABASE OPEN;
Re-Creating Lost Datafiles Without Backup

# if the database open: ALTER TABLESPACE table_data OFFLINE IMMEDIATE; # if the database is closed, open mount STARTUP MOUNT ALTER DATABASE DATAFILE 4 OFFLINE; # re-create the datafile with the same name: ALTER DATABASE CREATE DATAFILE '/ORADATA/u03/users01.dbf'; #or #re-create the datafile but with a new filename or location: ALTER DATABASE CREATE DATAFILE '/ORADATA/u03/users01.dbf' AS '/ORADATA/u04/users01.dbf'; RECOVER TABLESPACE table_data; RECOVER DATAFILE '/ORADATA/u03/users01.dbf '; ALTER TABLESPACE table_data ONLINE; ALTER DATABASE DATAFILE '/ORADATA/u03/users01.dbf' ONLINE;
Page 256
User-Managed Incomplete Recovery

Common Situations Requiring Incomplete Recovery
Missing archive Loss of redo logs User error Loss of control files: but you have a backup of an old binary copy
User-Managed Incomplete Recovery Steps

You must have the following to recover: A valid offline or online backup containing all datafiles. All archived redo logs, from the restored backup to before the time of failure. 1. Perform a full closed backup of the existing database. 2. Restore all datafiles. You may need to restore archived logs. If archive log changed to different destination: SET LOGSOURCE <LOCATION> STARTUP MOUNT 3. Place the database in mount mode and insure that the datafiles are online. SELECT file_id f#, file_name, tablespace_name tablespace, status FROM dba_data_files; 4. Recover the database. RECOVER [AUTOMATIC] DATABASE until .. until time 'YYYY-MM-DD:HH:MI:SS' until cancel until scn <integer> using backup control file recover database until time '2002-03-09:11:44:00'
5. Open the database by using the RESETLOGS option and verify the recovery. alter database open resetlogs; Note: If log files need to be re-created on another disk due to media failure, use the ALTER DATABASE DROP LOG GROUP and ALTER DATABASE ADD LOG GROUP commands to create the log files manually. 6. Perform a whole closed backup of the database.
Recovering from Lost Control File by Re-Creating the Control File

ALTER DATABASE BACKUP CONTROLFILE TO TRACE; the command generates file in <ORACLE_ADR>\rdbms\<GSID>\<SID>\trace\<SID>_ora_nnn.trc The exact file name will also be found in <ORACLE_ADR>\rdbms\<GSID>\<SID>\trace\alert_<SID>.log
Page 257
-- The following are current System-scope REDO Log Archival related -- parameters and can be included in the database initialization file. --- LOG_ARCHIVE_DEST='' -- LOG_ARCHIVE_DUPLEX_DEST='' --- LOG_ARCHIVE_FORMAT=ARC%S_%R.%T --- DB_UNIQUE_NAME="ora11g" --- LOG_ARCHIVE_CONFIG='SEND, RECEIVE, NODG_CONFIG' -- LOG_ARCHIVE_MAX_PROCESSES=4 -- STANDBY_FILE_MANAGEMENT=MANUAL -- STANDBY_ARCHIVE_DEST=%ORACLE_HOME%\RDBMS -- FAL_CLIENT='' -- FAL_SERVER='' --- LOG_ARCHIVE_DEST_10='LOCATION=USE_DB_RECOVERY_FILE_DEST' -- LOG_ARCHIVE_DEST_10='OPTIONAL REOPEN=300 NODELAY' -- LOG_ARCHIVE_DEST_10='ARCH NOAFFIRM NOEXPEDITE NOVERIFY SYNC' -- LOG_ARCHIVE_DEST_10='REGISTER NOALTERNATE NODEPENDENCY' -- LOG_ARCHIVE_DEST_10='NOMAX_FAILURE NOQUOTA_SIZE NOQUOTA_USED NODB_UNIQUE_NAME' -- LOG_ARCHIVE_DEST_10='VALID_FOR=(PRIMARY_ROLE,ONLINE_LOGFILES)' -- LOG_ARCHIVE_DEST_STATE_10=ENABLE --- LOG_ARCHIVE_DEST_1='LOCATION=C:\oracle\oracledb11g\RDBMS' -- LOG_ARCHIVE_DEST_1='MANDATORY NOREOPEN NODELAY' -- LOG_ARCHIVE_DEST_1='ARCH NOAFFIRM EXPEDITE NOVERIFY SYNC' -- LOG_ARCHIVE_DEST_1='NOREGISTER NOALTERNATE NODEPENDENCY' -- LOG_ARCHIVE_DEST_1='NOMAX_FAILURE NOQUOTA_SIZE NOQUOTA_USED NODB_UNIQUE_NAME' -- LOG_ARCHIVE_DEST_1='VALID_FOR=(PRIMARY_ROLE,ONLINE_LOGFILES)' -- LOG_ARCHIVE_DEST_STATE_1=ENABLE --- Below are two sets of SQL statements, each of which creates a new -- control file and uses it to open the database. The first set opens -- the database with the NORESETLOGS option and should be used only if -- the current versions of all online logs are available. The second -- set opens the database with the RESETLOGS option and should be used -- if online logs are unavailable. -- The appropriate set of statements can be copied from the trace into -- a script file, edited as necessary, and executed when there is a -- need to re-create the control file. --Set #1. NORESETLOGS case --- The following commands will create a new control file and use it -- to open the database. -- Data used by Recovery Manager will be lost. -- Additional logs may be required for media recovery of offline -- Use this only if the current versions of all online logs are -- available. -- After mounting the created controlfile, the following SQL -- statement will place the database in the appropriate -- protection mode: -- ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PERFORMANCE
Page 258
STARTUP NOMOUNT CREATE CONTROLFILE REUSE DATABASE "ORA11G" NORESETLOGS NOARCHIVELOG MAXLOGFILES 16 MAXLOGMEMBERS 3 MAXDATAFILES 100 MAXINSTANCES 8 MAXLOGHISTORY 292 LOGFILE GROUP 1 'C:\ORACLE\ORADATA\ORA11G\REDO01.LOG' SIZE 50M, GROUP 2 'C:\ORACLE\ORADATA\ORA11G\REDO02.LOG' SIZE 50M, GROUP 3 'C:\ORACLE\ORADATA\ORA11G\REDO03.LOG' SIZE 50M -- STANDBY LOGFILE DATAFILE 'C:\ORACLE\ORADATA\ORA11G\SYSTEM01.DBF', 'C:\ORACLE\ORADATA\ORA11G\SYSAUX01.DBF', 'C:\ORACLE\ORADATA\ORA11G\UNDOTBS01.DBF', 'C:\ORACLE\ORADATA\ORA11G\USERS01.DBF', 'C:\ORACLE\ORADATA\ORA11G\FDA_TBS.DBF' CHARACTER SET AR8MSWIN1256 ; -- Commands to re-create incarnation table -- Below log names MUST be changed to existing filenames on -- disk. Any one log file from each branch can be used to -- re-create incarnation records. -- ALTER DATABASE REGISTER LOGFILE 'C:\ORACLE\FLASH_RECOVERY_AREA\ORA11G\ARCHIVELOG\2010_03_19\O1_MF_1_1_%U_.ARC' ; -- Recovery is required if any of the datafiles are restored backups, -- or if the last shutdown was not normal or immediate. RECOVER DATABASE -- Database can now be opened normally. ALTER DATABASE OPEN; -- Commands to add tempfiles to temporary tablespaces. -- Online tempfiles have complete space information. -- Other tempfiles may require adjustment. ALTER TABLESPACE TEMP ADD TEMPFILE 'C:\ORACLE\ORADATA\ORA11G\TEMP01.DBF' SIZE 20971520 REUSE AUTOEXTEND ON NEXT 655360 MAXSIZE 32767M; -- End of tempfile additions. --Set #2. RESETLOGS case --- The following commands will create a new control file and use it -- to open the database. -- Data used by Recovery Manager will be lost. -- The contents of online logs will be lost and all backups will -- be invalidated. Use this only if online logs are damaged. -- After mounting the created controlfile, the following SQL -- statement will place the database in the appropriate -- protection mode: -- ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PERFORMANCE STARTUP NOMOUNT CREATE CONTROLFILE REUSE DATABASE "ORA11G" RESETLOGS NOARCHIVELOG MAXLOGFILES 16 MAXLOGMEMBERS 3 MAXDATAFILES 100 MAXINSTANCES 8 MAXLOGHISTORY 292
Page 259
LOGFILE GROUP 1 'C:\ORACLE\ORADATA\ORA11G\REDO01.LOG' SIZE 50M, GROUP 2 'C:\ORACLE\ORADATA\ORA11G\REDO02.LOG' SIZE 50M, GROUP 3 'C:\ORACLE\ORADATA\ORA11G\REDO03.LOG' SIZE 50M -- STANDBY LOGFILE DATAFILE 'C:\ORACLE\ORADATA\ORA11G\SYSTEM01.DBF', 'C:\ORACLE\ORADATA\ORA11G\SYSAUX01.DBF', 'C:\ORACLE\ORADATA\ORA11G\UNDOTBS01.DBF', 'C:\ORACLE\ORADATA\ORA11G\USERS01.DBF', 'C:\ORACLE\ORADATA\ORA11G\FDA_TBS.DBF' CHARACTER SET AR8MSWIN1256 ; -- Commands to re-create incarnation table -- Below log names MUST be changed to existing filenames on -- disk. Any one log file from each branch can be used to -- re-create incarnation records. -- ALTER DATABASE REGISTER LOGFILE 'C:\ORACLE\FLASH_RECOVERY_AREA\ORA11G\ARCHIVELOG\2010_03_19\O1_MF_1_1_%U_.ARC' ; -- Recovery is required if any of the datafiles are restored backups, -- or if the last shutdown was not normal or immediate. RECOVER DATABASE USING BACKUP CONTROLFILE -- Database can now be opened zeroing the online logs. ALTER DATABASE OPEN RESETLOGS; -- Commands to add tempfiles to temporary tablespaces. -- Online tempfiles have complete space information. -- Other tempfiles may require adjustment. ALTER TABLESPACE TEMP ADD TEMPFILE 'C:\ORACLE\ORADATA\ORA11G\TEMP01.DBF' SIZE 20971520 REUSE AUTOEXTEND ON NEXT 655360 MAXSIZE 32767M; -- End of tempfile additions. --
Page 260
Flash Recovery Area

Obtaining Information on Flash Recovery Area
-- FRA Configuration show parameter DB_RECOVERY_FILE_DEST_SIZE show parameter DB_RECOVERY_FILE_DEST -- related parameters show parameter DB_CREATE_FILE_DEST show parameter LOG_ARCHIVE_DEST_1 show parameter LOG_ARCHIVE_DEST_2 -- space usage select NAME, round(SPACE_LIMIT/1024/1024/1024,4) SPACE_LIMIT_GB, round(SPACE_USED/1024/1024/1024,4) SPACE_USED_GB, round(SPACE_RECLAIMABLE/1024/1024/1024,4) SPACE_RECLAIMABLE_GB, NUMBER_OF_FILES from V$RECOVERY_FILE_DEST; select FILE_TYPE, PERCENT_SPACE_USED, PERCENT_SPACE_RECLAIMABLE, NUMBER_OF_FILES from V$FLASH_RECOVERY_AREA_USAGE order by PERCENT_SPACE_USED desc
Configuring Flash Recovery Area

If configured, LOG_ARCHIVE_DEST and LOG_ARCHIVE_DUPLEX_DEST parameters cannot be used. -- configuring FRA ALTER SYSTEM SET DB_RECOVERY_FILE_DEST_SIZE = 4G SCOPE=BOTH; ALTER SYSTEM SET DB_RECOVERY_FILE_DEST = 'C:\ORACLE\RECOVERY_AREA' SCOPE=BOTH ALTER SYSTEM SET DB_RECOVERY_FILE_DEST = '+dskgrp1' -- related parameters DB_CREATE_FILE_DEST = /u02/test/oradata/dbfiles/ LOG_ARCHIVE_DEST_1 = 'LOCATION=/u03/test/arc_dest1' LOG_ARCHIVE_DEST_2 = 'LOCATION=USE_DB_RECOVERY_FILE_DEST' -- disable FRA ALTER SYSTEM SET DB_RECOVERY_FILE_DEST = '' -- 85% warning and 79% critical threshold are by default configured
Backing Up the Flash Recovery Area

In order to back up the flash recovery area itself using RMAN, you must set CONFIGURE BACKUP OPTIMIZATION to ON. Neither of the two commands, BACKUP RECOVERY AREA nor BACKUP RECOVERY FILES, will back up any permanent files or the flashback logs in the flash recovery area You can back up the flash recovery area only to a tape device using the following backup commands:
BACKUP RECOVERY AREA
Page 261
o This command backs up all flash recovery files in the current or previous flash recovery area destinations. o It backs up only those files that have never been backed up to tape before. o The files that the command will back up include full backups, incremental backups, control file autobackups, archive logs, and datafile copies. BACKUP RECOVERY FILES This command backs up all the files that the BACKUP RECOVERY AREA command does, but from all areas on your file system, not just from the flash recovery area. BACKUP RECOVERY FILE DESTINATION Use this command to move disk backups created in the flash recovery area to tape.
Moving the Flash Recovery Area

ALTER SYSTEM SET DB_RECOVERY_FILE_DEST='/u01/app/oracle/new_area' SCOPE=BOTH Eventually, Oracle will delete all the transient files from the previous flash recovery area location, when each of them becomes eligible for deletion. However, if you want to move your current permanent files, transient files, or flashback logs to the new flash recovery area, you can do so by using the standard file-moving procedures. If the database flashback logs are enabled and you moved the flash recovery area, the flashback option should be switched off and on.
Page 262
Recovery Manager (RMAN)

Using A Media Management Layer (MML) with RMAN
Oracle maintains the Oracle Backup Solutions Program (BSP). The partners are listed in http://otn.oracle.com/deploy/availability/htdocs/bsp.htm#MMV Some of the important one are Legato Systems (NetWorker) and VERITAS (NetBackup).
Obtaining Information about and related to RMAN using Dictionary Views

/* archives log files */ -- created, backed up, and cleared in the database select RECID, NAME, DEST_ID, THREAD#, SEQUENCE#, RESETLOGS_CHANGE#, RESETLOGS_TIME, RESETLOGS_ID, FIRST_CHANGE#, FIRST_TIME, NEXT_CHANGE#, NEXT_TIME, round(BLOCKS * BLOCK_SIZE/1024/1024,2) MB , CREATOR, ARCHIVED, DELETED, STATUS, COMPLETION_TIME, END_OF_REDO, BACKUP_COUNT, ARCHIVAL_THREAD#, IS_RECOVERY_DEST_FILE, COMPRESSED, FAL, BACKED_BY_VSS from V$ARCHIVED_LOG ;
/* Backup Sets */ select RECID, STAMP, SET_STAMP, SET_COUNT, BACKUP_TYPE, CONTROLFILE_INCLUDED, INCREMENTAL_LEVEL, PIECES, START_TIME, COMPLETION_TIME, ELAPSED_SECONDS,
Page 263
BLOCK_SIZE, INPUT_FILE_SCAN_ONLY, KEEP, KEEP_UNTIL, KEEP_OPTIONS, MULTI_SECTION from V$BACKUP_SET order by RECID;
/* Backup Pieces*/ SELECT RECID, STAMP, SET_STAMP, -- link to V$BACKUP_SET SET_COUNT, PIECE#, COPY#, DEVICE_TYPE, HANDLE, COMMENTS, MEDIA, MEDIA_POOL, CONCUR, TAG, STATUS, START_TIME, COMPLETION_TIME, ELAPSED_SECONDS, DELETED, BYTES, IS_RECOVERY_DEST_FILE, RMAN_STATUS_RECID, RMAN_STATUS_STAMP, COMPRESSED, BACKED_BY_VSS, ENCRYPTED, BACKED_BY_OSB, FROM V$BACKUP_PIECE ORDER BY SET_STAMP;
/* Files in Backup Sets */ -- control files and datafiles in backup sets from the control file select RECID, STAMP, SET_STAMP, SET_COUNT BACKUPSET_COUNT, B.FILE#, d.NAME DATAFILE_NAME, B.CREATION_CHANGE#, B.CREATION_TIME, RESETLOGS_CHANGE#, RESETLOGS_TIME, INCREMENTAL_LEVEL, INCREMENTAL_CHANGE#, B.CHECKPOINT_CHANGE#,
Page 264
B.CHECKPOINT_TIME, ABSOLUTE_FUZZY_CHANGE#, MARKED_CORRUPT, MEDIA_CORRUPT, LOGICALLY_CORRUPT, DATAFILE_BLOCKS, B.BLOCKS, B.BLOCK_SIZE, OLDEST_OFFLINE_RANGE, COMPLETION_TIME, CONTROLFILE_TYPE, USED_CHANGE_TRACKING, BLOCKS_READ, USED_OPTIMIZATION, B.FOREIGN_DBID, B.PLUGGED_READONLY, B.PLUGIN_CHANGE#, B.PLUGIN_RESETLOGS_CHANGE#, B.PLUGIN_RESETLOGS_TIME, SECTION_SIZE, UNDO_OPTIMIZED from V$BACKUP_DATAFILE b, V$DATAFILE d where B.FILE# (+)= D.FILE# ORDER BY b.CREATION_TIME DESC;
/* Archive Log files in the Backup Sets */ SELECT RECID, STAMP, SET_STAMP, SET_COUNT, THREAD#, SEQUENCE#, RESETLOGS_CHANGE#, RESETLOGS_TIME, FIRST_CHANGE#, FIRST_TIME, NEXT_CHANGE#, NEXT_TIME, round(BLOCKS * BLOCK_SIZE/1024/1024,2) MB , TERMINAL FROM V$BACKUP_REDOLOG ORDER BY RECID DESC; /* Corruption */ -- blocks have been found corrupt during a backup of a backup set select * from V$BACKUP_CORRUPTION; -- database blocks that were corrupted after the last backup select * from V$DATABASE_BLOCK_CORRUPTION
Starting RMAN
export ORACLE_SID=mydb rman target /
Page 265
rman target sys/psw cmdfile D:\..\dblevel0.ora log D:\..\dblevel0.log append rman target / @myrmanscript.ora rman target orcl catalog rman/rman@nick # log parameter doesnt display output to you. To work around: rman | tee /u01/app/oracle/rman.log # using dynamic script: passing parameters to command file # command file may contain: BACKUP DATABASE TAG &1 FORMAT '/u02/oracle/bck/&2%U.bck' # the script running rman may contain: export format=$2 export mytag=$3 rman @'/u01/app/oracle/scripts/my_backup.cmd' USING $format $mytag # you may run the script as: myscript.sh db01012010 HR01012010
Using rlwrap Utility with RMAN in Unix-Based Systems

/* Install */ # Option 1: for rpm version: # can be downloaded from: # http://rpm.pbone.net # http://ivan.kartik.sk rpm -ivh rlwrap*.rpm
# Option 2: for tar version # download rlwrap-0.30.tar.gz (search the net or from http://www.ahmedbaraka.com/download/oracle/rlwrap-0.30.tar.gz ) # unzip the file and install su gunzip rlwrap-0.30.tar.gz tar -xvf rlwrap-0.30.tar cd rlwrap-0.30 ./configure make make install # echo "alias rman2='rlwrap rman'" >> /home/oracle/.bashrc
Configuring the RMAN Environment

/* Handling Configuration in General */ # To list current configuration SHOW ALL # To set value for a configuration setting CONFIGURE CHANNEL DEVICE TYPE DISK FORMAT '/?/%U'; CONFIGURE CONTROL FILE AUTOBAKCUP ON # To reset the setting value CONFIGURE CONTROL FILE AUTOBAKCUP CLEAR
Page 266
/* Backup Retention Policy */ CONFIGURE RETENTION POLICY TO REDUNDANCY 2; CONFIGURE RETENTION POLICY TO RECOVERY WINDOW OF 14 DAYS;
/* Default Device Type */ CONFIGURE DEFAULT DEVICE TYPE TO sbt;
/* Encryption */ -- Transparent or Dual mode CONFIGURE ENCRYPTION FOR DATABASE ON
/* Compression */ CONFIGURE DEVICE TYPE DISK BACKUP TYPE TO COMPRESSED BACKUPSET
/* Defining Parallelism */ CONFIGURE DEVICE TYPE DISK PARALLELISM 4;
/* Configuring Default Channels */ -- number of channels to use depends on DISK PARALLELISM CONFIGURE CHANNEL 1 DEVICE TYPE DISK FORMAT '/u01/oracle/oradata/backup/%U'; CONFIGURE CHANNEL 2 DEVICE TYPE DISK FORMAT '/u02/oracle/oradata/backup/%U';
/* Backup Optimization */ -- datafile of a version identical to its backup won't be backuped up CONFIGURE BACKUP OPTIMIZATION ON;
/* Control File Auto-Backup */ CONFIGURE CONTROLFILE AUTOBACKUP ON; -- Control File Backup Location and Format CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '/u01/oracle/oradta/backup/cf_%F';
/* Archivelog Deletion Policy */ -- all destinations are affected including flash recovery area CONFIGURE ARCHIVELOG DELETION POLICY TO BACKED UP 2 TIMES TO SBT;
RMAN Channel Commands

# Manual Channel Allocation RUN { ALLOCATE CHANNEL c1 TYPE disk FORMAT = '/db01/BACKUP/usr0520.bak'; BACKUP DATAFILE '/db01/ORADATA/users01.dbf';} # Automatic Channel Allocation CONFIGURE DEFAULT DEVICE TO DISK | SBT ;
Page 267
CONFIGUE DEVICE TYPE DISK PARALLELISM n ; # Automatic Channel Options CONFIGURE CHANNEL DEVICE TYPE DISK FORMAT = '/BACKUP/RMAN/%U' ; CONFIGURE CHANNEL DEVICE TYPE DISK MAXPIECESIZE 2G ;
Duration in days of RMAN information in Control File

--in days select value from v$parameter where upper(name)='CONTROL_FILE_RECORD_KEEP_TIME'; alter system set CONTROL_FILE_RECORD_KEEP_TIME=30 ;
Monitoring RMAN Jobs

SELECT SID, RECID, STAMP, PARENT_RECID, PARENT_STAMP, SESSION_RECID, SESSION_STAMP, ROW_LEVEL, ROW_TYPE, COMMAND_ID, OPERATION, STATUS, MBYTES_PROCESSED, START_TIME, END_TIME from V$RMAN_STATUS ORDER BY RECID DESC; # to view message generated by the RMAN operation select SID, RECID, STAMP, SESSION_RECID, SESSION_STAMP, OUTPUT from V$RMAN_OUTPUT;
Using RMAN BACKUP Command

BACKUP command options FULL INCREMENTAL LEVEL n INCLUDE CURRENT CONTROLFILE FILESPERSET n # maximum number of input files in each backup set SKIP OFFLINE | READONLY | INACCESSIBLE MAXSETSIZE n K|M|G DELETE INPUT # applicable on archived logs, datafile copies or backup sets. FORMAT : %c copy number %p backup piece number %s backup set number %d database name %n database name padded with 8 characters %t backup set time stamp %u compressed version of %s and %t %U (default) equivalent t%u_%p_%c /* Backup Piece Size */ ALLOCATE CHANNEL MAXPIECESIZE = integer CONFIGURE CHANNEL MAXPIECESIZE = integer
Page 268
/* Backup examples */ -- make a whole database backup BACKUP DATABASE FORMAT '/tmp/%U' TAG='weekly_bak' SQL 'ALTER SYSTEM ARCHIVE LOG CURRENT'; -- Backing Up Tablespaces BACKUP TABLESPACE system, users, tools; -- Backing up Datafiles and Datafile Copies BACKUP DATAFILE 1,2,3,4, DATAFILECOPY '/tmp/system01.dbf' -- Backing Up Backup Sets (from disk to tape or from disk tdisk) BACKUP DEVICE TYPE sbt BACKUPSET ALL;
Backing Up Control File and SPFile

/* Control File Autobackup */ CONFIGURE CONTROLFILE AUTOBACKUP enable ; /* Control File Backup Format */ SET CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE disk TO 'controlfile_%F'; CONFIGURE CONTROLFILE AUTOBACKUP FORMAT '..' ; /* Backing Up the Control File Manually */ BACKUP CURRENT CONTROLFILE TAG='mondaypmbackup' ; /* Including the Control File in a Backup Set */ BACKUP TABLESPACE users INCLUDE CURRENT CONTROLFILE; /* Backing Up the Server Parameter File */ -- Automatically backed up when CONFIGURE CONTROLFILE AUTOBACKUP = ON -- Explicitly BACKUP SPFILE
Backing Up Archived RedLogs

/* Using BACKUP ARCHIVELOG command */ BACKUP ARCHIVELOG ALL ; BACKUP ARCHIVELOG NOT BACKED UP 2 TIMES; -- To delete backed up copy of the archived log file BACKUP ARCHIVELOG ALL DELETE INPUT ; -- To delete logs from all enabled archiving destinations. BACKUP ARCHIVELOG ALL DELETE ALL INPUT ; -- To specify a range of archived redlogs by time BACKUP ARCHIVELOG UNTIL TIME 'SYSDATE-7'; BACKUP ARCHIVELOG FROM TIME 'SYSDATE-30' UNTIL TIME 'SYSDATE-7'; -- To specify a range of archived redlogs by SCN BACKUP ARCHIVELOG UNTIL SCN = 320 ; BACKUP ARCHIVELOG SCN BETWEEN 205 AND 320 ;
Page 269
-- To specify a range of archived redlogs by log sequence number BACKUP ARCHIVELOG UNTIL SEQUENCE = 501 BACKUP ARCHIVELOG FROM SEQUENCE integer /* Using BACKUP ... PLUS ARCHIVELOG: */ BACKUP DATABASE PLUS ARCHIVELOG;
Backup in NOARCHIVELOG Mode

1. 2. 3. 4. Shut down cleanly Mount the database. Allocate multiple channels, if not using automatic. Run the BACKUP command.
Encrypting RMAN Backups

Three possible encryption modes for your backups: o Transparent mode: It requires Oracle Wallet. It is the default encryption mode. CONFIGURE ENCRYPTION FOR DATABASE ON Password mode: It requires you to provide a password. It is best suited for backups restored at remote locations. SET ENCRYPTION ON IDENTIFIED BY password ONLY o Dual mode: It can use either Oracle Wallets or passwords. After making sure the wallet is open: SET ENCRYPTION ON IDENTIFIED BY password o If there is no wallet or the wallet is closed: SET DECRYPTION IDENTIFIED BY password1 {, password2,, passwordn}
Using Compression in RMAN Backups

Only Backupsets can be compressed: RMAN> CONFIGURE DEVICE TYPE DISK BACKUP TYPE TO COMPRESSED BACKUPSET RMAN> BACKUP AS COMPRESSED BACKUPSET DATABASE
Using Multiplexed Backup Sets

RMAN can read from two datafiles simultaneously, and then combine the blocks from these datafiles into a single backup piece. Multiplexing can be controlled by the following: o o The FILESPERSET parameter on the BACKUP command The MAXOPENFILES parameter of the ALLOCATE CHANNEL and CONFIGURE CHANNEL commands (default is 8)
Using Parallelization of Backup Sets

Parallelization of backup sets is achieved by: o
Page 270
Configuring PARALLELISM to greater than 1 Oracle DBA Code Examples
o o
allocating multiple channels Specifying many files in the BACKUP command
BACKUP (DATAFILE 1,2,3 FILESPERSET = 1 CHANNEL ORA_DISK_1) (DATAFILECOPY '/tmp/system01.dbf', '/tmp/tools01.dbf' FILESPERSET = 2 CHANNEL ORA_DISK_2); RUN { ALLOCATE CHANNEL c1 DEVICE TYPE sbt PARMS="ENV= (BACKUP_SERVER=tape_server1)"; ALLOCATE CHANNEL c2 DEVICE TYPE sbt PARMS="ENV= (BACKUP_SERVER=tape_server2)"; BACKUP (DATAFILE 1,2,3 CHANNEL c1) (DATAFILECOPY '/tmp/system01.dbf', '/tmp/tools01.dbf' FILESPERSET = 2 CHANNEL c2) }
Using Duplexed Backup Sets (Backupset Copies)

CONFIGURE . . . BACKUP COPIES option (maximum 4) SET BACKUP COPIES in a RUN block. The COPIES option in the BACKUP command. -- BACKUP COPIES BACKUP DEVICE TYPE DISK COPIES 2 DATAFILE 1 FORMAT '/disk1/df1_%U', '/disk2/df1_%U'; BACKUP COPIES 2 DATAFILE 1, DATAFILE 2 FORMAT '/BACKUP1/%U','/BACKUP2/%U'; BACKUP COPIES 3 INCREMENTAL LEVEL = 0 DATABASE; -- CONFIGURE BACKUP COPIES CONFIGURE CHANNEL DEVICE TYPE DISK FORMAT '/save1/%U', '/save2/%U'; CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE sbt T2; CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE sbt T2;
Making Image Copies

BACKUP AS COPY.. BACKUP AS COPY DATABASE; BACKUP AS COPY TABLESPACE SYSAUX; BACKUP AS COPY DATAFILE 2; BACKUP AS COPY COPY OF DATABASE; BACKUP AS BACKUPSET COPY OF TABLESPACE SYSAUX; BACKUP AS COPY COPY OF DATAFILE 2; -- to check
Page 271
list backupset of tablespace tbs1; In 9i, (DEPRICATED IN 10G ) COPY DATAFILE { 'filename'| integer } | DATAFILECOPY {'filename' |TAG='tag_name'} | ARCHIVELOG 'filename' | CURRENT CONTROLFILE | CONTROLFILECOPY {'filename'|TAG='tag_name'} TO AUXNAME | 'filename' COPY DATAFILE '/ORADATA/users_01_db01.dbf' to '/BACKUP/users01.dbf' tag=DF3 ;
Validating Backup
When you run BACKUP VALIDATE, RMAN checks datafiles for physical and logical block corruption but it does not actually produce any backup sets or image copies. # only database level BACKUP VALIDATE DATABASE ARCHIVELOG ALL; # db, tablespace, datafile, block levels VALIDATE BACKUPSET 1;
Incremental Backup
Level 0 or Level 1. Differential: since last 1 or 0, Cumulative: since last 0 (faster recovery). BACKUP INCREMENTAL LEVEL N [CUMULATIVE] backup of: DATAFILE, DATAFILECOPY, TABLESPACE, or DATABASE. BACKUP INCREMENTAL LEVEL 0 ...
Tags for Backups and Image Copies

BACKUP .. TAG='tag_name'
Creating Archival Backups

The BACKUP ... KEEP command can be used to create a backup that is both all-inclusive (every file needed including archived redo logs) and exempt from the backup retention policy. In Oracle 11g, some modifications made on the RMAN BACKUP ... KEEP command. In the new version of the command, the KEEP, NOKEEP, FOREVER, and UNTIL TIME options are retained. However, the LOGS and NOLOGS options are not there any longer. Instead, you have a new option, RESTORE POINT. The RESTORE POINT option lets RMAN automatically create a normal restore point.
Note: You cant use the KEEP clause for backup files in the flash recovery area. Also, you cannot use the CHANGE ... KEEP command for backup files stored in the flash recovery area. RUN
Page 272
{ ALLOCATE CHANNEL c1 DEVICE TYPE sbt PARMS 'ENV=(OB_MEDIA_FAMILY=archival_backup)'; -- with forever option (recovery catalog is required) BACKUP DATABASE TAG BAKQ108 KEEP FOREVER RESTORE POINT FY08Q1; -- backup will be kept for 365 days (long-term) BACKUP DATABASE TAG BAKQ108 KEEP UNTIL TIME 'SYSDATE+365' RESTORE POINT FY08Q1; -- After one day, the backup becomes obsolete, -- regardless the configured retention policy BACKUP DATABASE FORMAT '/u01/oraclebck/%U.bck' TAG TESTDB KEEP UNTIL 'SYSDATE+1' RESTORE POINT TESTDB08; }
If you want to change the status of a regular backup to an archival backup, use the CHANGE command as follows:
CHANGE BACKUP TAG 'weekly_bkp' KEEP FOREVER; -- make it follow back the retention policy CHANGE BACKUP TAG 'weekly_bkp' NOKEEP;
Monitoring RMAN Backups

To correlate a process with a channel during a backup: 1. In each session, set the COMMAND ID to a different value RUN { ALLOCATE CHANNEL c1 TYPE sbt; SET COMMAND ID TO 'sess1'; BACKUP DATABASE; } 2. Query the joined V$SESSION and V$PROCESS views
SELECT SID, SPID, CLIENT_INFO FROM V$PROCESS p, V$SESSION s WHERE p.ADDR = s.PADDR AND s.CLIENT_INFO LIKE '%id=sess%';
The CLIENT_INFO column displays in the following format: id=command_id,rman channel=channel_id 3. Query the V$SESSION_LONGOPS view to get the status of the backup or copy
SELECT s.SID, S.SERIAL#, CLIENT_INFO, round(L.ELAPSED_SECONDS/60,2) ELAPSED_TIME_MIN ,L.SOFAR, L.TIME_REMAINING/60 TIME_REMAINING_MIN FROM V$PROCESS p, V$SESSION s, V$SESSION_LONGOPS l WHERE p.ADDR = s.PADDR and L.SID = s.SID and L.SERIAL# = s.SERIAL# AND s.CLIENT_INFO LIKE '%rman%' AND NVL(L.TIME_REMAINING,0)<>0;
RMAN
Complete Recovery
Validating Backup Files You can use VALIDATE Validate backup sets before you use them from a recovery. You can run RESTORE ... VALIDATE to test whether RMAN can restore a specific file or set of files from a backup. RMAN chooses which backups to use.
Page 273
LIST COPY; LIST BACKUP; LIST BACKUP RECOVERABLE; VALIDATE BACKUPSET 1; -- no restore is actually done -- lack of errors: validation found no problem RESTORE DATABASE VALIDATE; RESTORE ARCHIVELOG ALL VALIDATE; Previewing Backup Files Required by a Restore The RESTORE . . . PREVIEW command provides a detailed report of all backups that are necessary for that RESTORE command to succeed. RESTORE RESTORE RESTORE RESTORE DATABASE PREVIEW; DATABASE PREVIEW SUMMARY; TABLESPACE users PREVIEW; DATAFILE 3 PREVIEW;
Identifying Datafiles Requiring Recovery SELECT r.FILE# AS df#, d.NAME AS df_name, t.NAME AS tbsp_name, d.STATUS, r.ERROR, r.CHANGE#, r.TIME FROM V$RECOVER_FILE r, V$DATAFILE d, V$TABLESPACE t WHERE t.TS# = d.TS# AND d.FILE# = r.FILE#; Performing Complete Recovery -- Recover a Database in ARCHIVELOG Mode STARTUP MOUNT; RESTORE DATABASE; RECOVER DATABASE;
-- Restore Datafiles to a New Location run{ set newname for datafile 1 to '/newdir/system01.dbf'; restore database; switch datafile all; # record in control file recover database;} -- alternatively: run { sql "alter tablespace tbs1 offline immediate" ; set newname for datafile 6 to 'C:\APP\ADMINISTRATOR\ORADATA\ORA11G\tbs1.dbf'; restore tablespace tbs1; switch datafile 6; recover tablespace tbs1; } sql "alter tablespace tbs1 online" ;
-- Recover a Tablespace run{ sql "alter tablespace users offline immediate"; restore tablespace users; recover tablespace users; sql "alter tablespace users online"; }
Page 274
-- Relocate a Tablespace RUN{ SQL "alter tablespace users offline immediate"; SET NEWNAME FOR DATAFILE '/ORADATA/u03/users01.dbf' TO '/ORADATA/u04/users01.dbf'; RESTORE TABLESPACE users; SWITCH datafile 3; # Update the control file and recovery catalog RECOVER TABLESPACE users; #Recover the tablespace SQL "alter tablespace tbs1 online";} Restoring whole Database from RMAN Backups On a Different Node The directory structures can be the same or different The procedure applies for database in ARCHIVELOG mode or NOARCHIVELOG mode. Reference: Document ID 419137.1 The following example assume restoring a database from node1 to node2. /* Move the following files to the NODE 2 */ + The database backup pieces to location '/node2/database/backup' + Controlfile backup piece to the location '/node2/database/backup' + The parameter file i.e init.ora file to the default location: Unix: $ORACLE_HOME/dbs Windows: $ORACLE_HOME/database
/* Edit the PFILE on NODE 2 to change the environment specific parameters and create thier directories */ user_dump_dest= background_dump_dest= control_files= <--specify the new location of controlfiles db_recovery_file_dest=
/* set the Oracle environment variables and start the database in nomount mode */ Unix: export ORACLE_HOME=/u01/oracle/product/ora10g export ORACLE_SID=ora10g export PATH=$ORACLE_HOME/bin:$PATH Windows: set ORACLE_HOME=D:\oracle\product\10.1.0\db_1 set ORACLE_SID=oradb set PATH=D:\oracle\product\10.1.0\db_1\bin;%PATH%
/* Create a password file: */ Unix: $orapwd file=$ORACLE_HOME/dbs/orapw$ORACLE_SID.ora password=*** entries=5 Windows: orapwd file=%ORACLE_HOME%\database\PWD%ORACLE_SID%.ora password=*** entries=5
/* In Windows: create the SID service: */
Page 275
oradim -NEW -SID oradb -STARTMODE auto
/* invoke Rman on the NODE 2 */ -- in Windows, it's already started so shutdown first rman target / Recovery Manager: Release 10.2.0.1.0 - Production on Tue Feb 13 00:36:55 2007 Copyright (c) 1982, 2005, Oracle. All rights reserved. connected to target database (not started) RMAN> startup nomount pfile=X:\..\PFILE.ORA Oracle instance started Total System Global Area 205520896 bytes Fixed Size 1218508 bytes Variable Size 75499572 bytes Database Buffers 121634816 bytes Redo Buffers 7168000 bytes
/* Restore the controlfile from the backup piece. */ -- controlfiles will be resotred to locations indicated in pfile RMAN> restore controlfile from 'D:\temp\backup\ORADB\AUTOBACKUP\2011_07_06\O1_MF_S_755814486_718GDZ6L_.BKP'; Starting restore at 06/07/11 allocated channel: ORA_DISK_1 channel ORA_DISK_1: sid=380 devtype=DISK channel ORA_DISK_1: restoring controlfile channel ORA_DISK_1: restore complete output filename=D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\CONTROL01.CTL output filename=D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\CONTROL02.CTL output filename=D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\CONTROL03.CTL Finished restore at 06/07/11
/* Mount the database */ RMAN> alter database mount;
/* Now catalog the backup pieces that were shipped from NODE 1 */ RMAN> catalog backuppiece '/node2/database/backup/o1_mf_annnn_TAG20070213T002925_2x21m6ty_.bkp'; RMAN> catalog backuppiece '/node2/database/backup/o1_mf_annnn_TAG20070213T002825_2x21kbds_.bkp'; RMAN> catalog backuppiece '/node2/database/backup/o1_mf_nnndf_TAG20070213T002827_2x21kd12_.bkp'; RMAN>catalog backuppiece 'D:\temp\backup\ORADB\BACKUPSET\2011_07_06\O1_MF_NNNDF_TAG20110706T203113_718G CL0P_.BKP';
/* In case of ARCHIVELOGMODE: Get to know the last sequence available in the archivelog backup using the following command.*/ -- This will help us in recovering the database till that archivelog. RMAN > list backup of archivelog all; Let us assume the last sequence of last archivelog in the backup is 50.
Page 276
/* Rename the Redologfiles,so that they can be created in new locations the database is opened in resetlogs */ SQL>conn sys as sysdba -- get registered files from SELECT * FROM V$LOGFILE SQL> alter database rename file '/node1/database/prod/redo01.log' to '/node2/database/prod/redo01.log'; .... .... SQL>alter database rename file alter database rename file 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\REDO01.LOG' to 'D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\REDO01.LOG'; .... ....
/* Now restore the datafiles to new locations and recover. */ Since we are recovering the database here till the archivelog sequence 50 the sequence number in the SET UNTIL SEQUENCE clause should be 50 Note: If we are restoring the Rman backups from tapes, then we should ensure the same media manager variables that were used during backups are maintained during restore too. The following document articles gives the information of various media manager Environment Variables: NOTE.312737.1 RMAN and Specific Media Managers Environment Variables. Note: in NOARCHIVEMOE, omit recover database below RMAN> run { set until sequence 51; set newname for datafile 1 to '/node2/database/prod/sys01.dbf'; set newname for datafile 2 to '/node2/database/prod/undotbs01.dbf'; set newname for datafile 3 to '/node2/database/prod/sysaux01.dbf'; set newname for datafile 4 to '/node2/database/prod/users01.dbf'; set newname for datafile 5 to '/node2/database/prod/1.dbf'; set newname for datafile 6 to '/node2/database/prod/sysaux02.dbf'; set newname for datafile 7 to '/node2/database/prod/undotbs02.dbf'; restore database; switch datafile all; recover database; alter database open resetlogs; }
Note: in NOARCHIVEMOE, omit recover database below RMAN> run { set newname for datafile 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SYSTEM01.DBF' to 'D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SYSTEM01.DBF'; set newname for datafile 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\UNDOTBS01.DBF' to 'D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\UNDOTBS01.DBF'; set newname for datafile 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SYSAUX01.DBF' to 'D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SYSAUX01.DBF'; set newname for datafile 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\USERS01.DBF' to D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\USERS01.DBF'; set newname for datafile 'E:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SATBS01.DBF' to 'D:\ORACLE\PRODUCT\10.1.0\DB_1\ORADB\SATBS01.DBF'; restore database; switch datafile all; alter database open resetlogs; Oracle DBA Code Examples
Page 277
/* Addd tempfile to temporary tablespaces */ -- list of temporary tablespaces that require datafiles are in alert file ALTER TABLESPACE temp ADD TEMPFILE ' MAXSIZE 8G;
..'
SIZE 500M AUTOEXTEND ON NEXT 100M
/* Some recommended post-restore */ -- user created jobs will automatically run if their scheduled date is due -- you may consider stopping them as job owner: -- to list of the jobs: select JOB, BROKEN, SCHEMA_USER, NEXT_DATE, WHAT from dba_jobs; begin dbms_job.broken( 387 ,true); commit; end; / Restoring whole Database from RMAN Backups from A 32 bit to 64 bit Reference: Document ID 467676.1 -- follow all the steps in the pervious section titled "Restoring whole Database from RMAN Backups On a Different Node" till you reach to the step "Now restore the datafiles to new locations and recover." -- implement the steps in that step but change opening statement as follows: ALTER DATABASE OPEN RESETLOGS MIGRATE; -- all PL/SQL objects should be re-compiled -- after opening the database with resetlogs: -- invalidate and all pl/sql modules SQL> @ ?/rdbms/admin/utlirp.sql SQL> shutdown immediate; SQL> startup; -- recompiles invalid objects in the database SQL> @ ?/rdbms/admin/utlrp.sql SQL> shutdown immediate; SQL> startup
RMAN Incomplete Recovery

1. Mount the database. 2. The following steps should be followed: RUN { # multiple channels for parallelization ALLOCATE CHANNEL c1 TYPE DISK; ALLOCATE CHANNEL c2 TYPE DISK; # recover until time, SCN or sequence SET UNTIL TIME = '2001-12-09:11:44:00'; SET UNTIL TIME "to_date('09-05-2004 00:00:20', 'dd-mm-yyyy hh24:mi:ss')" ; SET UNTIL SEQUENCE 120 THREAD 1;#120 not included
Page 278
# ALL datafiles must be restored RESTORE DATABASE; RECOVER DATABASE; ALTER DATABASE OPEN RESETLOGS; } 3. If using a recovery catalog, register the new incarnation of the database using the command: RESET DATABASE 4. Perform a whole database backup. Note: Insure that NLS_LANG and NLS_DATE_FORMAT environment variables are set appropriately. Note: check the alert.log for any errors during recovery. Note: If you need to restore archived redo log files to a new location use the SE SET ARCHIVELOG DESTINATION TO <location> command.
Simplified Recovery Through Resetlogs

Target: to recovery through current and ancestor database incarnations. You must reset the current incarnation of the database back to the incarnation to which your target SCN belongs. You must use the control file from the older incarnation that contains the target SCN. Find out the incarnation key for the incarnation that was current at the time you want to recover your database to. RMAN>LIST INCARNATION Start the database in the following way: RMAN> STARTUP FORCE NOMOUNT; Reset the current incarnation: RMAN> RESET DATABASE TO INCARNATION 2; Restore the old control file from a backup: RMAN> RESTORE CONTROLFILE FROM AUTOBACKUP; RMAN> ALTER DATABASE MOUNT; Restore and recover the database: RMAN> RESTORE DATABASE; RMAN> RECOVER DATABASE UNTIL SCN 1000; Open the database: RMAN> ALTER DATABASE OPEN RESETLOGS;
Recovering from Lost Control File using RMAN

If database isn't already shutdown: RMAN>shutdown abort RMAN>startup nomount RMAN>RESTORE CONTROLFILE FROM AUTOBACKUP; ALTER DATABASE MOUNT; RECOVER DATABASE;
Page 279
ALTER DATABASE OPEN RESETLOGS;
Block Media Recovery (BMR)

RMANs BLOCKRECOVER command recovers blocks marked as corrupt in the V$BACKUP_CORRUPTION and V$COPY_CORRUPTION views. BMR helps on the following goals: o o Faster recovery time. Increased database availability.
BMR can repair only physically corrupted blocks but not the logically corrupted blocks. -- typical scenario: ORA_11578: ORACLE data block corrupted (file# 9, block# 21) ORA=01110: data file 9: /u01/app/oracle/oradata/remorse/users_01.dbf' -- the command will search flashback logs then backup files -- specific block RECOVER DATAFILE 9 BLOCK 21; -- all blocks in V$DATABASE_BLOCK_CORRUPTION RECOVER CORRUPTION LIST; -- to query which blocks recovered: SELECT * FROM V$DATABASE_BLOCK_CORRUPTION;
Trial Recovery
Target: to estimate size of corruption in a recovery process. It lets you know whether there is corruption and, if there is, the extent of the corruption. -- You can use the TEST option for any RECOVER command RECOVER DATABASE TEST RECOVER DATABASE USING BACKUP CONTROLFILE UNTIL CANCEL TEST RECOVER TABLESPACE users TEST RECOVER DATABASE UNTIL CANCEL TEST -- You can limit the number of data blocks trial recovery can corrupt in memory: RECOVER DATABASE TEST ALLOW 10 CORRUPTION; SQL> RECOVER DATABASE UNTIL CANCEL TEST; ORA-10574: Test recovery did not corrupt any data block ORA-10573: Test recovery tested redo from change 9948095 to 9948095 ORA-10570: Test recovery complete
Handling Specific Errors During Recovery

ORA-01194 Error: ----------------- this error may raise during startup a cloned database -- resolution: provide the online redo log file to recover SQL> startup
Page 280
ORACLE instance started. .. Database mounted. ORA-01589: must use RESETLOGS or NORESETLOGS option for database open SQL> alter database open noresetlogs; alter database open noresetlogs * ERROR at line 1: ORA-01588: must use RESETLOGS option for database open SQL> alter database open resetlogs; alter database open resetlogs * ERROR at line 1: ORA-01194: file 1 needs more recovery to be consistent ORA-01110: data file 1: 'C:\ORACLE\ORADATA\MANAGER\SYSTEM01.DBF' SQL> RECOVER DATABASE UNTIL CANCEL USING BACKUP CONTROLFILE; ORA-00279: change 405719 generated at 06/30/2008 15:51:04 needed for thread 1 ORA-00289: suggestion : C:\ORACLE\RDBMS\ARC00019.001 ORA-00280: change 405719 for thread 1 is in sequence #19 Specify log: {<RET>=suggested | filename | AUTO | CANCEL} C:\ORACLE\ORADATA\MANAGER\REDO03.LOG Log applied. Media recovery complete. SQL> alter database open resetlogs; Database altered.
ORA-01152 Error: ----------------- resolution: provide the online redo log file to recover ORA-00289: suggestion : /u01/app/oracle/admin/finance/arch/finance/_0000012976.arc ORA-00280: change 962725326 for thread 1 is in sequence #12976 ORA-00278: logfile'/u01/app/oracle/admin/finance/arch/finance/_0000012975.arc' no longer needed for this recovery Specify log: {<RET>=suggested | filename | AUTO | CANCEL} ORA-01547: warning: RECOVER succeeded but OPEN RESETLOGS would get error below ORA-01152: file 1 was not restored from a sufficiently old backup ORA-01110: data file 1: '/pase16/oradata/finance/system_01.dbf'ORA-01112: media recovery not started Specify log: {<RET>=suggested | filename | AUTO | CANCEL} /pase04/oradata/finance/redo01a.rdo ORA-00279: change 962746677 generated at 07/30/2008 04:33:52 needed for thread 1 ORA-00289: suggestion : /u01/app/oracle/admin/finance/arch/finance/_0000012978.arc ORA-00280: change 962746677 for thread 1 is in sequence #12978 ORA-00278: log file '/pase04/oradata/finance/redo01a.rdo' no longer needed for this recovery Specify log: {<RET>=suggested | filename | AUTO | CANCEL} /pase04/oradata/finance/redo02a.rdo Log applied. Media recovery complete.
Page 281
ORA-00376 Error: ---------------reason: might be datafile or tablespace being offline resolution: bringing the tablespace or datafile online ORA-00376: file 10 cannot be read at this time ORA-01110: data file 10: '/u01/app/oracle/remorse/data_01.dbf'
Configuring Instance Crash Recovery Time (MTTR)

SELECT VALUE FROM V$PARAMETER WHERE UPPER(NAME)='FAST_START_MTTR_TARGET'; -- in seconds ALTER DATABASE SET FAST_START_MTTR_TARGET=60; SELECT recovery_estimated_ios, estimated_mttr, target_mttr FROM v$instance_recovery;
Working with the Data Recovery Advisor in RMAN

list list LIST LIST list list list failure; failure critical; FAILURE HIGH; FAILURE LOW; failure open; failure closed; failure exclude failure 12345;
advise failure; repair failure preview;
repair failure; The V$IR_REPAIR view shows the results of the REPAIR FAILURE command: select REPAIR_ID,ADVISE_ID,SUMMARY,RANK from V$IR_REPAIR;
RMAN Maintenance
Cross Checking Backups and Copies Use CROSSCHECK command to ensure that data about backup sets and image copies in the recovery catalog or control file is synchronized with corresponding files on disk or in the media management catalog. CROSSCHECK BACKUPSET OF DATABASE; CROSSCHECK BACKUP OF TABLESPACE users DEVICE TYPE sbt COMPLETED BEFORE 'SYSDATE-31'; Page 11 Oracle Database Administration Fundamentals II (Note Sheets) CROSSCHECK BACKUP OF ARCHIVELOG ALL SPFILE;
Page 282
CROSSCHECK BACKUPSET 1338, 1339, 1340; CROSSCHECK BACKUPPIECE TAG = 'nightly_backup'; CROSSCHECK CONTROLFILECOPY '/tmp/control01.ctl'; CROSSCHECK DATAFILECOPY 113, 114, 115; Note: If the backup or copy is no longer available, then RMAN marks it as EXPIRED. You can determine which files are marked EXPIRED by issuing a LIST EXPIRED command. Deleting Backups and Copies /* Deleting Specified Backups and Copies */ DELETE BACKUPPIECE 101; DELETE CONTROLFILECOPY '/tmp/control01.ctl'; DELETE NOPROMPT ARCHIVELOG UNTIL SEQUENCE = 300; DELETE BACKUP OF TABLESPACE users DEVICE TYPE sbt; DELETE COPY OF CONTROLFILE LIKE '/tmp/%'; DELETE NOPROMPT BACKUP OF SPFILE COMPLETED BEFORE 'SYSDATE-7'; DELETE NOPROMPT ARCHIVELOG ALL BACKED UP 3 TIMES TO sbt; /* Deleting Expired Backups and Copies */ -- files are marked as EXPIRED when not found by CROSSCHECK DELETE EXPIRED BACKUP; -- REPORT OBSOLETE DELETE EXPIRED COPY; DELETE NOPROMPT EXPIRED BACKUP OF TABLESPACE users DEVICE TYPE sbt COMPLETED BEFORE 'SYSDATE-31'; /* Deleting Backups and Copies Rendered Obsolete by the Retention Policy */ DELETE OBSOLETE; /* Deleting Backups and Copies Defined as Obsolete */ DELETE OBSOLETE REDUNDANCY = 3; DELETE OBSOLETE RECOVERY WINDOW OF 7 DAYS; /* Forcing the Deletion of Backups and Copies */ DELETE FORCE NOPROMPT BACKUPSET TAG 'weekly_bkup'; Changing the Availability of RMAN Backups and Copies If a file is marked UNAVAILABLE, RMAN will not use the file when a RESTORE or RECOVER command is issued. Use the command CHANGE ... UNAVAILABLE CHANGE CHANGE CHANGE CHANGE DATAFILECOPY '/DB01/BACKUP/users01.dbf' UNAVAILABLE; BACKUP OF CONTROLFILE UNAVAILABLE; BACKUP OF CONTROLFILE AVAILABLE; COPY OF ARCHIVELOG SEQUENCE BETWEEN 230 AND 240 UNAVAILABLE;
Exempting a Backup or Copy from the Retention Policy Note: Specify KEEP ... LOGS to save archived logs for a possible incomplete recovery and KEEP ... NOLOGS not to save archived logs for a possible incomplete recovery. Note: The KEEP FOREVER clause requires the use of a recovery catalog. Note: Use CHANGE ... NOKEEP to make the file conform to the retention policy. Use the command CHANGE ... KEEP FOREVER|UNTIL CHANGE BACKUPSET 123 KEEP FOREVER NOLOGS; CHANGE DATAFILECOPY '/DB01/BACKUP/users01.dbf' KEEP UNTIL 'SYSDATE+60';
Page 283
The CATALOG Command Use CATALOG command to make RMAN aware of the existence of objects that are not recorded in the repository. You need to make RMAN aware of the existence of archived redo log files that are not recorded in the repository, if you manually have restored your control file from a backup. CATALOG DATAFILECOPY '/DB01/BACKUP/users01.dbf'; CATALOG CONTROLFILECOPY /DB01/BACKUP/db.ctl; CATALOG ARCHIVELOG '/ORADATA/ARCHIVE1/arch_12.arc', '/ORADATA/ARCHIVE1/arch_13.arc'; # start searching for all uncataloged files in the directory CATALOG START WITH '/disk1/backups/'; # Catalog all files and the contents of directories which # begin with the pattern "/backup/MYSID/arch". CATALOG START WITH '/backup/MYSID/arch'; # Catalog all files in the current recovery area. CATALOG RECOVERY AREA NOPROMPT; # Catalog all files in the current recovery area. # This is an exact synonym of the previous command. CATALOG DB_RECOVERY_FILE_DEST NOPROMPT; The CHANGE UNCATALOG Command Run the CHANGE ... UNCATALOG command to perform the following actions on RMAN repository records: o Delete a specific backup or copy record from the recovery catalog o Update a backup or copy record in the target control file repository to status DELETED RMAN does not touch the specified physical files: it only alters the repository records for these files. You can use this command when you have deleted a backup or copy through a means other than RMAN. The CHANGE ... UNCATALOG is not supported for backupsets. Use Run the CHANGE ... UNAVAILABLE or CHANGE ... DELETE instead. CHANGE ARCHIVELOG UNCATALOG; CHANGE DATAFILECOPY '/DB01/BACKUP/users01.dbf' UNCATALOG;
RMAN Catalog
Creating a Recovery Catalog
1. Create tablespace 2. Create catalog owner
CREATE USER RMAN IDENTIFIED BY rman TEMPORARY TABLESPACE temp DEFAULT TABLESPACE rman_tbs QUOTA UNLIMITED ON rman_tbs;
3. Grant privileges GRANT connect, resource, recovery_catalog_owner to rman; 4. Create catalog rman catalog rman_db1/rman_db1@catdb RMAN> create catalog tablespace rman_ts; Page 284
5. Connect to target database as SYSDBA RMAN TARGET SYS/SYS@TEST2 CATALOG RMAN/RMAN@TEST1 6. Register target database ( make sure ORACLE_SID is properly set ): REGISTER DATABASE;
RMAN Catalog Reporting

Use REPORT and LIST commands. Alternatively you can use SQL commands to query the RMAN data dictionary views. Examples of those view are the following: RC_DATABASE RC_DATAFILE RC_STORED_SCRIPT RC_STORED_SCRIPT_LINE RC_TABLESPACE
/* report command */ REPORT SCHEMA REPORT OBSOLETE; REPORT NEED BACKUP; # datafile containing undergone a nologging data (must be backed up) REPORT UNRECOVERABLE /* list command */ LIST BACKUP; LIST BACKUP RECOVERABLE; LIST COPY; LIST ARCHIVELOG ALL; LIST SCRIPT NAMES; LIST GLOBAL SCRIPT NAMES; LIST INCARNATION; Upgrading a Recovery Catalog
If your RMAN client is from the Oracle 11.1 release, but the recovery catalog schema is from an older version, you must upgrade the recovery catalog.
-- to know current version of ur catalog SELECT * FROM rcver; -- upgrade steps 1) If the recovery catalog owner that you created is from a release before 10.1, execute the following GRANT command (assuming that rman is the catalog owner): SQL> GRANT CREATE TYPE TO rman; 2) Start RMAN and connect to the recovery catalog database. RMAN> connect catalog rman/rman; 3) Execute the UPGRADE CATALOG command. RMAN> UPGRADE CATALOG; 4) Confirm the command by rerunning it. RMAN> UPGRADE CATALOG; You can now use the recovery catalog with the RMAN client from the Oracle Database 11g release. Importing Recovery Catalogs
To merge two recovery catalogs, one from the 10.2 release and the other from 11g, into a single 11g release catalog schema.
Page 285
1. Connect to the destination recovery catalog. $ rman RMAN> connect catalog rman/rman@rman11 2. Issue the IMPORT CATALOG command RMAN> import catalog rman1/rman1@rman10; To specify which database to register: RMAN> import catalog rman10/rman10@tenner dbid = 123456, 1234557; RMAN> import catalog rman10/rman10@tenner db_name = testdb, mydb; -- by default, the databases are unregistered from the source catalog: RMAN> import catalog rman10/rman10@tenner NO UNREGISTER Moving a Recovery Catalog -- 1) -- 2) RMAN> RMAN> create empty recovery catalog connect and import: connect catalog rman/rman@target_db import catalog rman10/rman10@source_db;
Dropping a Recovery Catalog connect catalog rman/rman@target_db DROP CATALOG;
Virtual Private Catalogs

In Oracle Database 11g, you can restrict access to the recovery catalog by granting access to only a subset of the metadata in the recovery catalog. The subset that a user has read/write access to is termed as virtual private catalog, or just virtual catalog. The central or source recovery catalog is now called the base recovery catalog. Following are the steps to create a new private catalog for the database user SCOTT: # grant the role RECOVERY_CATALOG_OWNER to the user SQL>GRANT RECOVERY_CATALOG_OWNER TO scott; # in RMAN session, connect as the base catalog owner RMAN>CONNECT CATALOG rman/rman@mydb RMAN>GRANT CATALOG FOR DATABASE db1, db2 TO SCOTT; # connect as the granted user (virtual catalog onwer) and create the virtual catalog RMAN>CONNECT CATALOG scott/lion@mydb RMAN>CREATE VIRTUAL CATALOG; # make sure only granted dbs are seen RMAN>LIST INCARNATION; If the catalog is to be used for releases pre-Oracle 11g clients, in the SQL*Plus log on as the virtual private catalog owner and run the following procedure, where "rman" is the name of the base catalog owner: SQL> CONN scott/lion@mydb SQL> EXEC rman.DBMS_RCVCAT.CREATE_VIRTUAL_CATALOG;
Note
A virtual private catalog owner can create a local stored script, but has only read-only access to global scripts. The CATALOG FOR DATABASE privileges include the privilege to register and unregister those databases for which the catalog for database privilege was granted. The set of views and synonyms that makes up the virtual private catalog is stored in the schema of the virtual catalog owner.
Page 286
Managing Virtual Private Catalogs The base recovery catalog owner can optionally grant a virtual catalog owner the right to register new target databases in the recovery catalog by specifying the REGISTER database clause with the GRANT command. Following is an example: RMAN> grant register database to scott; The virtual catalog owner must have the SYSDBA and SYSOPER privileges on the traget database, to perform most of the RMAN operations on it. Following are examples of removing the privileges from a virtual catalog owner: # To remove recovery catalog access to a database from a user: RMAN>CONNECT CATALOG RMAN/RMAN@MYDB; RMAN>REVOKE CATALOG FOR DATABASE db1 FROM scott; # To revoke the ability to register new databases from a virtual private catalog owner: RMAN>REVOKE REGISTER DATABASE FROM scott; # To revoke both the catalog and register privileges from a user: RMAN>REVOKE ALL PRIVILEGES FROM scott; Dropping a Virtual Private Catalog Virtual private catalog owners can drop the private recovery catalog they own by issuing the DROP CATALOG command. Following is an example: # Log in as the virtual catalog owner: RMAN>CONNECT CATALOG scott/<password>@mydb; # Issue the drop catalog command RMAN>DROP CATALOG;
Caution
When the DROP CATALOG command is issued by the virtual catalog owner, all the metadata pertaining to it is deleted from the base recovery catalog.
Using RMAN Scripts

# create script CREATE SCRIPT nightly_backup { ALLOCATE CHANNEL c1 TYPE DISK; BACKUP DATABASE FORMAT '/u01/app/oracle/%u'; SQL 'ALTER DATABASE BACKUP CONTROLFILE TO TRACE'; } # run it RUN {EXECUTE SCRIPT nightly_backup;} # create global script: accessible by any registered db CREATE GLOBAL SCRIPT global_full_backup {.. # creating and using dynamic scripts: CREATE SCRIPT myscript { ... BACKUP TAG &1 .. RUN { EXECUTE SCRIPT myscript USING arch_bkp FY09Q4;} # print a script on the screen ( or query RC_STORED_SCRIPT_LINE )
Page 287
PRINT GLOBAL SCRIPT global_full_backup; # listing script names LIST SCRIPT NAMES; # update a script REPLACE SCRIPT full_backup { .. # converting stored script to text file PRINT script nightly_backup to file 'test.txt'; # delete a script DELETE SCRIPT 'my-script';
Page 288
Duplicating (Cloning) a Database

A duplicate database is a copy of your target database with a new, unique database identifier (DBID).
Database Duplication (Cloning) Methods

Using RMAN DUPLICATE command Using the OEM Database Control Manually with SQL*Plus
Database Duplication Techniques

Following are the techniques for database duplication: o o o Duplicating an Active Database: From an active database, connected to the target and auxiliary instances (no backup is required). Duplicating a Database with a Target Connection: From backup, connected to the target and auxiliary instances. Duplicating a Database with Recovery Catalog Without Target Connection: From backup, connected to the auxiliary instance, not connected to the target, but with recovery catalog connection. Duplicating a Database without Recovery Catalog or Target Connection: From backup, connected to the auxiliary instance, not connected to the target and the recovery catalog.
Database Duplication Prerequisites

Both the target and destination databases must be on an identical operating system platform. The password file must exist for the source database and both the target and destination databases must have the same SYS password. The target database must be open or in mount state. If the target database is open, it must be in ARCHIVELOG mode.
Duplicating an Active Database using RMAN

-Create a password file in the destination server: orapwd FILE=D:\oracle\product\11.1.0\db_1\database\PWDora11g2.ora PASSWORD=ora11g ENTRIES=10 ignorecase=n -Establish Oracle Net connectivity to the auxiliary instance in both the source and destination servers: ORA11G2 = (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = moep10)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED)
Page 289
(SERVICE_NAME = ora11g2) ) )
-Add the auxiliary database service to the listener configuration file in the source server: SID_LIST_LISTENER = (SID_LIST = (SID_DESC = (GLOBAL_DBNAME = ora11g) (ORACLE_HOME = C:\app\Administrator\product\11.1.0\db_1) (SID_NAME = ora11g) ) (SID_DESC = (GLOBAL_DBNAME = ora11g2) (ORACLE_HOME = C:\app\Administrator\product\11.1.0\db_1) (SID_NAME = ora11g2) ) )
- create a text-based initialization parameter file for the auxiliary instance that contains only one parameter: DB_NAME in ORACLEHOME\database\initora11g2.ora DB_NAME=ora11g2
- create Oracle SID service (in Windows) oradim -new -sid ora11g2
- connect to the auxilary instance: sqlplus /nolog conn sys/ora11g@ora11g2 as sysoper STARTUP NOMOUNT pfile=D:\oracle\product\11.1.0\db_1\database\InitOra11g2.ora
-- Start and Configure RMAN Before Duplication # source database RMAN>CONNECT TARGET SYS@ora11g # duplicate database instance RMAN>CONNECT AUXILIARY SYS@ora11g2
-- You may want to increase the parallelism setting of your source database disk channels CONFIGURE DEVICE TYPE DISK PARALLELISM 1 ;
-- check parameters containing folder info: col value format a35 col name format a30 select NAME, VALUE FROM v$parameter where upper(value) like '%D:\%';
Page 290
-- Run the DUPLICATE command -- duplicating a database with a different directory structure -- ( CREATE REQUIRED DIRS) DUPLICATE TARGET DATABASE TO ora11g2 FROM ACTIVE DATABASE DB_FILE_NAME_CONVERT 'D:\oracle\oradata\ora11g','D:\oracle\oradata\ora11g2' SPFILE PARAMETER_VALUE_CONVERT 'D:\oracle\oradata\ora11g', 'D:\oracle\oradata\ora11g2' SET MEMORY_TARGET '350M' SET LOG_FILE_NAME_CONVERT 'D:\oracle\oradata\ora11g', 'D:\oracle\oradata\ora11g2' SET audit_file_dest 'D:\oracle\admin\ora11g2\adump' SET CONTROL_FILES="D:\ORACLE\ORADATA\ORA11G2\CONTROL01.CTL","D:\ORACLE\ORADATA\ORA 11G2\CONTROL02.CTL";
Duplicating a Database without Recovery Catalog or Target Connection

Oracle 11g R2 This is backup-based duplication - Make Backups and Archived Logs Accessible to the Duplicate Instance: - if backup files location cannot be shared: Create a new directory in the source host that has the same name as the directory on the destination host that will contain the backups: in source mkdir C:\oracle\stagingbackups in source RUN { ALLOCATE CHANNEL c1 DEVICE TYPE DISK FORMAT 'C:\oracle\stagingbackups\%U'; BACKUP AS COPY DATABASE; BACKUP AS COPY CURRENT CONTROLFILE FORMAT 'C:\oracle\stagingbackups\control01.ctl'; } OR RUN { ALLOCATE CHANNEL c1 DEVICE TYPE DISK FORMAT 'C:\oracle\stagingbackups\%U'; BACKUP COPY OF DATABASE ; BACKUP AS COPY CURRENT CONTROLFILE FORMAT 'C:\oracle\stagingbackups\control01.ctl'; } - Manually transfer the backups in the new directory on the source host to the identically named directory on the destination host. - in source: -- check parameters containing folder info: col value format a35 col name format a30 select NAME, VALUE FROM v$parameter where upper(value) like '%C:\%'; OR select NAME, VALUE FROM v$parameter where name in Oracle DBA Code Examples
Page 291
('control_files','db_recovery_file_dest','audit_file_dest','diagnostic_dest');
- in destination: -Create a password file in the destination server: orapwd FILE=C:\oracle\oracledb11g\database\PWDora11g2.ora PASSWORD=ora11g ENTRIES=10 ignorecase=n - create a text-based initialization parameter file for the auxiliary instance that contains only one parameter: DB_NAME in ORACLEHOME\database\initora11g2.ora DB_NAME=ora11g2
- create Oracle SID service (in Windows) oradim -new -sid ora11g2
-configure Net connectivity to the auxiliary instance in the destination server: ORA11G2 = (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = pc02)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = ora11g2.pc02) ) )
- configure the listner, if no listener already exists - manually register the database in the listner.ora: SID_LIST_LISTENER = (SID_LIST = (SID_DESC = (GLOBAL_DBNAME = ora11g2.pc02) (ORACLE_HOME = C:\oracle\oracledb11g) (SID_NAME = ora11g2) ) ) LISTENER = (DESCRIPTION_LIST = (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = PC02)(PORT = 1521)) ) (DESCRIPTION = (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1521)) ) )
- connect to the auxilary instance: set ORACLE_SID=ORA11G2 sqlplus /nolog

Page 292
conn sys/ora11g@ora11g2 as sysoper STARTUP NOMOUNT pfile=C:\oracle\oracledb11g\database\initora11g2.ora
-- Start and Configure RMAN Before Duplication # duplicate database instance set ORACLE_SID=ORA11G2 rman CONNECT AUXILIARY SYS@ora11g2 -- You may want to increase the parallelism setting of your source database disk channels CONFIGURE DEVICE TYPE DISK PARALLELISM 1 ; -- Run the DUPLICATE command -- duplicating a database with a different directory structure -- ( CREATE REQUIRED DIRS) DUPLICATE TARGET DATABASE TO ora11g2 BACKUP LOCATION 'C:\oracle\stagingbackups' DB_FILE_NAME_CONVERT 'C:\oracle\oradata\ora11g','C:\oracle\oradata\ora11g2' SPFILE PARAMETER_VALUE_CONVERT 'C:\oracle\oradata\ora11g', 'C:\oracle\oradata\ora11g' SET MEMORY_TARGET '620M' SET LOG_FILE_NAME_CONVERT 'C:\oracle\oradata\ora11g', 'C:\oracle\oradata\ora11g2' SET audit_file_dest 'C:\ORACLE\ADMIN\ORA11G2\ADUMP' SET CONTROL_FILES="C:\ORACLE\ORADATA\ORA11G2\CONTROL01.CTL","C:\ORACLE\ORADATA\ORA 11G2\CONTROL02.CTL";
Manually Duplicating a Database

-- it is assumed that the duplicate database will have different name (ora11g2) from prod (ora11g) 1. Copy the prod database files to the target location. - in prod SHUTDOWN IMMEDIATE COPY THE DATABASE DATA FILES (EXCLUDE THE TEMFILES) + redo log file + spfiles CREATE PFILE='C:\TEMP\MYPFILE.TXT' FROM MEMORY; Edit the generated PFILE to modify directory-based parameters like: control_files audit_file_dest db_recovery_file_dest diagnostic_dest Change DB name: db_domain db_name
2. Prepare a text file for the creation of a control file for the new database as follows: - in prod SQL> ALTER DATABASE BACKUP CONTROLFILE TO TRACE;
Page 293
in ORACLE_BASE\diag\rdbms\ora11g\ora11g\trace\ folder Modify as required the parameters below, include them in the parameter file and create their folders: Note: if you define diagnostic_dest, you don't have to create its subdirectories -- The following are current System-scope REDO Log Archival related -- parameters and can be included in the database initialization file. --- LOG_ARCHIVE_DEST='' -- LOG_ARCHIVE_DUPLEX_DEST='' --- LOG_ARCHIVE_FORMAT=ARC%S_%R.%T --- DB_UNIQUE_NAME="ora11g2" --- LOG_ARCHIVE_CONFIG='SEND, RECEIVE, NODG_CONFIG' -- LOG_ARCHIVE_MAX_PROCESSES=4 -- STANDBY_FILE_MANAGEMENT=MANUAL -- STANDBY_ARCHIVE_DEST=C:\oracle\oracledb11g\RDBMS -- FAL_CLIENT='' -- FAL_SERVER='' --- LOG_ARCHIVE_DEST_10='LOCATION=USE_DB_RECOVERY_FILE_DEST' -- LOG_ARCHIVE_DEST_10='OPTIONAL REOPEN=300 NODELAY' -- LOG_ARCHIVE_DEST_10='ARCH NOAFFIRM NOEXPEDITE NOVERIFY SYNC' -- LOG_ARCHIVE_DEST_10='REGISTER NOALTERNATE NODEPENDENCY' -- LOG_ARCHIVE_DEST_10='NOMAX_FAILURE NOQUOTA_SIZE NOQUOTA_USED NODB_UNIQUE_NAME' -- LOG_ARCHIVE_DEST_10='VALID_FOR=(PRIMARY_ROLE,ONLINE_LOGFILES)' -- LOG_ARCHIVE_DEST_STATE_10=ENABLE ---- the current versions of all online logs are available -- Note: the generated code will use "REUSE DATABASE .. NORESETLOGS"; change it to "SET DATABASE.. RESETLOGS " if -- you want to use a differnt db name CREATE CONTROLFILE SET DATABASE "ORA11G2" RESETLOGS ARCHIVELOG MAXLOGFILES 16 MAXLOGMEMBERS 3 MAXDATAFILES 100 MAXINSTANCES 8 MAXLOGHISTORY 292 LOGFILE GROUP 1 'C:\ORACLE\ORADATA\ORA11G2\REDO01.LOG' SIZE 50M, GROUP 2 'C:\ORACLE\ORADATA\ORA11G2\REDO02.LOG' SIZE 50M, GROUP 3 'C:\ORACLE\ORADATA\ORA11G2\REDO03.LOG' SIZE 50M -- STANDBY LOGFILE DATAFILE 'C:\ORACLE\ORADATA\ORA11G2\SYSTEM01.DBF', 'C:\ORACLE\ORADATA\ORA11G2\SYSAUX01.DBF', 'C:\ORACLE\ORADATA\ORA11G2\UNDOTBS01.DBF', 'C:\ORACLE\ORADATA\ORA11G2\USERS01.DBF', 'C:\ORACLE\ORADATA\ORA11G2\FDA_TBS.DBF' CHARACTER SET AR8MSWIN1256 ;
Page 294
3. in destination: Create a password file in the destination server: orapwd FILE=C:\oracle\oracledb11g\database\PWDora11g2.ora PASSWORD=ora11g ENTRIES=10 ignorecase=n
4. Create Oracle SID service (in Windows) oradim -new -sid ora11g2
5. Configure the listner, if no listener already exists
6. Connect to the instance: set ORACLE_SID=ORA11G2 sqlplus /nolog conn / as sysdba # use pfile from step 1 STARTUP NOMOUNT pfile=C:\oracle\oracledb11g\database\initora11g2.ora sample of its contetns follows: audit_file_dest='C:\ORACLE\ADMIN\ORA11G2\ADUMP' audit_trail='DB' compatible='11.1.0.0.0' control_files='C:\ORACLE\ORADATA\ORA11G2\CONTROL01.CTL' control_files='C:\ORACLE\ORADATA\ORA11G2\CONTROL02.CTL' control_files='C:\ORACLE\ORADATA\ORA11G2\CONTROL03.CTL' db_block_size=8192 db_domain='pc02' db_name='ora11g2' db_recovery_file_dest='C:\oracle\flash_recovery_area' db_recovery_file_dest_size=4048M diagnostic_dest='C:\ORACLE' dispatchers='(PROTOCOL=TCP) (SERVICE=ora11gXDB)' log_buffer=5654016 # log buffer update memory_target=620M open_cursors=400 optimizer_dynamic_sampling=2 optimizer_mode='ALL_ROWS' plsql_warnings='DISABLE:ALL' # PL/SQL warnings at init.ora processes=400 query_rewrite_enabled='TRUE' remote_login_passwordfile='EXCLUSIVE' result_cache_max_size=2112K sessions=390 skip_unusable_indexes=TRUE undo_tablespace='UNDOTBS1' LOG_ARCHIVE_FORMAT=ARC%S_%R.%T DB_UNIQUE_NAME="ora11g2" LOG_ARCHIVE_DEST_10='LOCATION=USE_DB_RECOVERY_FILE_DEST' 7. Create the control file be executing the command in step 2 8. Execute: ALTER DATABASE OPEN RESETLOGS;
Page 295
8. Execute: ALTER TABLESPACE TEMP ADD TEMPFILE 'C:\ORACLE\ORADATA\ORA11G2\TEMP01.DBF' size 250m autoextend on next 10m maxsize 4g; 9. Execute: ALTER DATABASE RENAME GLOBAL_NAME TO ora11g2.pc02; ALTER SYSTEM REGISTER;
Page 296
Using Oracle Flashback Technology
Page 297
Flashback Options
Row Level: o o o o Flashback Query Flashback Versions Query Flashback Transaction Query Flashback Transaction Backout
Table level: o o o Flashback Table Flashback Drop Flashback Data Archive (FDA)
Database level: o Flashback Database
Page 298
Preparing Your Database for Flashback

-- must be AUTO show parameter UNDO_MANAGEMENT -- set undo retension target select value from v$parameter where upper(name)='UNDO_RETENTION'; alter system set UNDO_RETENTION = 2400; -- get undo tbs name show parameter UNDO_TABLESPACE -- consider guaranteeing undo retention: ALTER TABLESPACE UNDOTBS1 RETENTION GUARANTEE ;
Page 299
Using Row Level Flashback Options

Flashback Query
-- if the user isn't the owner GRANT FLASHBACK ON employees TO scott; GRANT FLASHBACK ANY TABLE TO scott; -- what was the salary at that time SELECT employee_id, department_id, salary FROM employees AS OF TIMESTAMP TO_TIMESTAMP ('2008-11-07 11:30:00', 'YYYY-MM-DD HH:MI:SS') WHERE employee_id=101; select employee_id , salary from employees AS OF SCN 794532059 where employee_id=101
/* using DBMS_FLASHBACK */ -- set the specified point in time in the past select ... -- the output will be as of current time EXECUTE DBMS_FLASHBACK.ENABLE_AT_TIME (TO_TIMESTAMP '11-DEC-2008 10:00:00', 'DD-MON-YYYY hh24:MI:SS'); select ... -- the output will be as of the set time EXECUTE DBMS_FLASHBACK.DISABLE ();
-- to get current SCN SELECT current_scn from V$DATABASE; select DBMS_FLASHBACK.GET_SYSTEM_CHANGE_NUMBER from dual;
Flashback Versions Query

The VERSIONS clause cannot span DDL commands. Syntax: SELECT [pseudocolumns] . . . /* provide details about the row history */ FROM . . . /* table name goes here */ VERSIONS BETWEEN {SCN|TIMESTAMP {expr|MINVALUE} AND {expr|MAXVALUE}} [AS OF{SCN|TIMESTAMP expr}] WHERE [pseudocolumns . . . ] . . . Pseudo-Columns: VERSIONS_STARTSCN and VERSIONS_STARTTIME: VERSIONS_ENDSCN and VERSIONS_ENDTIME: if null, current or deleted VERSIONS_OPERATION: I,D,U VERSIONS_XID: trans ID
Note: The actual time might be up to three seconds earlier or later than the time you specify with a time stamp.
Page 300
Flashback Versions Query restrictions: o You cant apply the VERSIONS clause across DDL operations. o The query will ignore purely physical row changes as happen, for example, during a segment shrink operation.
SELECT versions_xid AS XID, versions_startscn AS START_SCN, versions_endscn AS END_SCN, versions_operation AS OPERATION, last_name, salary FROM EMPLOYEES VERSIONS BETWEEN SCN MINVALUE AND MAXVALUE WHERE employee_id = 101; SET LINESIZE 100 COL START_TIME FORMAT A21 COL END_TIME FORMAT A21 SELECT versions_xid AS XID, VERSIONS_STARTTIME AS START_TIME, VERSIONS_ENDTIME AS END_TIME, VERSIONS_OPERATION AS OPERATION, last_name, salary FROM EMPLOYEES VERSIONS BETWEEN TIMESTAMP MINVALUE AND MAXVALUE WHERE employee_id = 101;
Flashback Transaction Query

It is actually a SELECT from FLASHBACK_TRANSACTION_QUERY /* requirements */ --(1) highly recommended ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; --(2) required privs grant SELECT ANY TRANSACTION to scott; /* using it */ SELECT to_char(start_timestamp,'dd-Mon hh24:mi'), operation, undo_sql FROM flashback_transaction_query WHERE start_timestamp >= TO_TIMESTAMP ('2009-11-09 05:00:00', 'YYYY-MM-DD HH:MI:SS') AND table_owner='HR' AND table_name='EMPLOYEES' order by start_timestamp desc; /* using flashback transaction with versions query */ SELECT xid, START_SCN , commit_scn COMMIT, OPERATION, logon_user, undo_sql FROM flashback_transaction_query WHERE xid = HEXTORAW('0600230050010000'); -- obtained from version query
Flashback Transaction (Backout)

Target: to back-out a committed transaction and all dependent transactions while the database is still online. Oracle DBA Code Examples
Page 301
OEM provides an easier to manage interface. DBMS_FLASHBACK.TRANSACTION_BACKOUT parameters: NUMBEROFXIDS Number of transactions to be backed out XIDS OPTIONS TIMEHINT SCNHINT array of transaction identifiers nocascade (default), cascade, nocascade_force, noconflict_only time at the start of the transaction you are backing out SCN at the beginning of the transaction you are backing out
-- Setting Up for Flashback Transaction Backout ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS;
-- required privs grant EXECUTE on DBMS_FLASHBACK to hr; grant SELECT ANY TRANSACTION to hr;
-- get the transaction ID -- use v$ views or the flashback transaction query or log miner -- following example from sql in cache: select A.SQL_TEXT, B.XID from V$OPEN_CURSOR A, V$TRANSACTION B, V$SESSION C where A.SID=C.SID AND C.TADDR=B.ADDR and A.SQL_TEXT LIKE '%delete%';
-- backout the trans declare V_XID SYS.XID_ARRAY; begin V_XID := SYS.XID_ARRAY('03001800BC0D0000'); DBMS_FLASHBACK.TRANSACTION_BACKOUT( NUMTXNS => 1, XIDS=>V_XID, OPTIONS=>DBMS_FLASHBACK.CASCADE); -- children trans also backed out end; / -- to make it permanent: you must now COMMIT or ROLLBACK -- Information about transaction Backouts done can be obtained from: DBA_FLASHBACK_TXN_STATE: any transaction shown in this view is backed out. DBA_FLASHBACK_TXN_REPORT: compensating status of all transactions in the database
Page 302
Using Table Level Flashback Options

Flashback Table
ROW MOVEMENT must be enabled in advance and before flashback recovery time. Generates undo and redo data Syntax: FLASHBACK TABLE [schema.]table [,[schema.]table] ... TO {SCN|TIMESTAMP} expr By default, Oracle disables all relevant triggers and reenables them upon completing the table recovery. Several restrictions apply: o o You cant flash back a table to a time preceding any DDL operation involving a change in table structure. If the flashback operation involves multiple tables, all of the tables must be flashed back or none.
-- required privs grant FLASHBACK ANY TABLE to hr; grant FLASHBACK on hr.employees to hr; -- SELECT, INSERT, DELETE, and ALTER privileges on the table to be flashed back are required -- required on the table -- also enable it for the tables that reference it ALTER TABLE emp ENABLE ROW MOVEMENT; -- TAKE current SCN before flashback to go back to initial stat of table if -the flashback doesnt do good SELECT current_scn from V$DATABASE; select DBMS_FLASHBACK.GET_SYSTEM_CHANGE_NUMBER from dual FLASHBACK TABLE emp TO SCN 5759290864; FLASHBACK TABLE emp TO TIMESTAMP TO_TIMESTAMP ('2008-01-30 07:00:00', 'YYYYMM-DD HH24:MI:SS'); FLASHBACK TABLE emp TO TIMESTAMP TO_TIMESTAMP ('2009-04-05 10:00:00', 'YYYYMM-DD HH24:MI:SS') ENABLE TRIGGERS;
Flashback Drop
When table is dropped, its dependents' names will also renamed by system generated names. If a table is undropped, the cryptic system-generated names of the dependents remain. -- Necessary Privileges **** be the owner or have the drop privileges (DROP TABLE or DROP ANY TABLE) on a table. You must have the SELECT privilege and the FLASHBACK privilege on an object in order to query that object in the Recycle Bin.
Page 303
-- list dropped tables **** -- db level SELECT owner, original_name, object_name,ts_name, droptime, CAN_UNDROP FROM dba_recyclebin; -- user level SELECT * FROM RECYCLEBIN; SHOW RECYCLEBIN
-- Enabling and Disabling the Recycle Bin *** -- session level ALTER SESSION SET recyclebin = OFF; -- db level -- it affects the parameter file ALTER SYSTEM SET recyclebin = OFF;
-- retreive data from a dropped table SELECT * FROM "BIN$xTMPjHZ6SG+1xnDIaR9E+g==$0";
-- restore table from drop FLASHBACK TABLE table_name TO BEFORE DROP; FLASHBACK TABLE "BIN$xTMPjHZ6SG+1xnDIaR9E+g==$0" TO BEFORE DROP; FLASHBACK TABLE "BIN$xTMPjHZ6SG+1xnDIaR9E+g==$0" TO BEFORE DROP RENAME TO NEW_PERSONS; -- you can rename any dependent SELECT INDEX_NAME FROM USER_INDEXES WHERE TABLE_NAME = 'JOB_HISTORY'; ALTER INDEX "BIN$DBo9UChtZSbgQFeMiAdCcQ==$0" RENAME TO JHIST_JOB_IX; -- purge a table(s) DROP TABLE table_name PURGE; PURGE TABLE int_admin_emp; PURGE TABLE BIN$jsleilx392mk2=293$0; PURGE TABLESPACE users; PURGE TABLESPACE users USER scott; PURGE RECYCLEBIN; -- purge all objects in the db PURGE DBA_RECYCLEBIN;
Flashback Data Archive

FDA is part of the "Oracle Total Recall" option in Oracle database 11g.
Page 304
Disabling flashback archiving for a table or dropping its flashback archive object will result in all the historical data for that table being lost. It also requires SYSDBA or FLASHBACK ARCHIVE ADMINISTER privilege. For a table with Flashback Archiving enabled, you cannot issue the following DDL commands: ALTER TABLE (except adding a column), DROP TABLE, RENAME TABLE and TRUNCATE TABLE. /* to obtain info about FDA */ -- information on all flashback archives contained in the database SELECT * FROM DBA_FLASHBACK_ARCHIVE; -- information on all tablespaces containing flashback archives SELECT * FROM DBA_FLASHBACK_ARCHIVE_TS; -- which flashback archive a given table is assigned to. SELECT TABLE_NAME, OWNER_NAME, FLASHBACK_ARCHIVE_NAME FROM DBA_FLASHBACK_ARCHIVE_TABLES; -- history tables names created by Oracle: select * from dba_FLASHBACK_ARCHIVE_TABLES; /* Setting up the Data Flashback Archive */ --1) for system level FDA, FLASHBACK ARCHIVE ADMINISTER priv is required SELECT * FROM DBA_SYS_PRIVS WHERE PRIVILEGE LIKE '%FLASHBACK ARC%' -- for specific table: GRANT FLASHBACK ARCHIVE ON hr_hist TO scott;
--2) Create FDA object -- The statement in the example above may return ORA-55603: Invalid Flashback Archive command error, if you try to create a flashback archive in a non-empty tablespace. I figured out a workaround which is to put the tablespace name between double quotations. CREATE FLASHBACK ARCHIVE hr_hist -- DEFAULT keyword may be used TABLESPACE fda_archives -- mandatory (and it must be with ASSM) QUOTA 5G -- optional in M,G,T,P RETENTION 24 MONTH; -- mandatory (in YEAR, MONTH, DAY)
--3) Enable Flashback Data Archiving for existing or new tables -- Create the table, using the default archive location. CREATE TABLE my_table(..) FLASHBACK ARCHIVE; -- Modify a table to use the default archive location -- Note: if there is not default flashback archive, an error will be raised ALTER TABLE my_table FLASHBACK ARCHIVE; -- Create a table to use a non-default archivelocation CREATE TABLE my_table (..) FLASHBACK ARCHIVE hr_arc;
Page 305
-- Modify a table to use a non-default archive location. ALTER TABLE my_table FLASHBACK ARCHIVE hr_arc; -- Modify a table to stop (disable) archiving. ALTER TABLE my_table NO FLASHBACK ARCHIVE;
/* Altering Flashback Archives */ -- make myflash the default flashback archive (as SYSDBA) ALTER FLASHBACK ARCHIVE myflash SET DEFAULT; -- add space to the flashback archive ALTER FLASHBACK ARCHIVE myflash ADD TABLESPACE mytbs; -- remove the tablespace from use by the flashback archive -- (assign it to another tablespace first) ALTER FLASHBACK ARCHIVE myflash REMOVE TABLESPACE mytbs; -- change the quota for the archive ALTER FLASHBACK ARCHIVE myflash MODIFY TABLESPACE mytbs QUOTA 10G; -- undefined quota (make the space unlimited) ALTER FLASHBACK ARCHIVE myflash MODIFY TABLESPACE mytbs; -- change the archive retention time ALTER FLASHBACK ARCHIVE myflash MODIFY RETENTION 2 YEAR; -- purge all archived data ALTER FLASHBACK ARCHÌVE myflash PURGE ALL; -- purge data older than 2 days ALTER FLASHBACK ARCHIVE MYFLASH PURGE BEFORE TIMESTAMP( SYSTIMESTAMP INTERVAL '2' DAY); -- dropping FDA: DROP FLASHBACK ARCHIVE myflash; /* Using Oracle Flashback Data Archives */ SELECT LAST_NAME, SALARY FROM HR.EMPLOYEES AS OF TIMESTAMP TO_TIMESTAMP ('2008-01-01 00:00:00','YYYY-MM-DD HH24:MI:SS'); SELECT LAST_NAME, SALARY FROM HR.EMPLOYEES AS OF TIMESTAMP (SYSTIMESTAMP - INTERVAL '6' MONTH); SELECT LAST_NAME, SALARY FROM HR.EMPLOYEES VERSIONS BETWEEN TIMESTAMP TO_TIMESTAMP ('2008-01-01 00:00:00','YYYY-MM-DD HH24:MI:SS') -- or MINVALUE AND TO_TIMESTAMP ('2008-01-01 15:00:00','YYYY-MM-DD HH24:MI:SS') -- or MAXVALUE
Page 306
WHERE EMPLOYEE_ID = 200; FLASHBACK TABLE employees TO TIMESTAMP (SYSTIMESTAMP INTERVAL '6' MONTH);
Page 307
Using Flashback Database

When to use Flashback Database
To retrieve a dropped schema When a user error affects the entire database When you truncate a table in error When a batch job performs only partial changes
Flashback Database Considerations

You cannot use Flashback Database when: The control file has been restored or re-created. A tablespace has been dropped. A data file has been reduced in size, for example by a shrink.
Using Flashback Database

It is essential that you monitor the flash recovery areas size to ensure that you have sufficient space so as not to risk losing any of the Flashback Database logs. /* Configuring Flashback Database */ -- Check that your database is in the archivelog mode: ARCHIVE LOG LIST -- in minutes ( 1440= 1 day) ALTER SYSTEM SET DB_FLASHBACK_RETENTION_TARGET=1440; -- in mount level ALTER DATABASE FLASHBACK ON; ALTER DATABASE OPEN;
/* Exempting a tablespace form db flashback */ ALTER TABLESPACE users FLASHBACK OFF; -- to switch it on again: ALTER TABLESPACE users FLASHBACK ON;
/* Disabling Flashback Database */ -- in mount level -- all Flashback Database logs will be deleted ALTER DATABASE FLASHBACK OFF;
Page 308
/* Obtain info about Flashback Database */ -- to find out how far you can flash back your database SELECT oldest_flashback_scn, to_char(oldest_flashback_time,'dd-Mon hh24:mi:ss') oldest_flashback_time FROM v$flashback_database_log; -- to estimate the space required by flashback db: SELECT estimated_flashback_size/1024/1024 estimated_flashback_size, retention_target, flashback_size/1024/1024 flashback_size FROM v$flashback_database_log; -- last 24-hour stats data (each row is one-hour ) SELECT begin_time, end_time, flashback_data, db_data, redo_data, estimated_flashback_size AS EST_FB_SZE FROM V$FLASHBACK_DATABASE_STAT order by begin_time desc;
/* Flashback db in action */ -- flashback db to a past time STARTUP MOUNT; FLASHBACK DATABASE TO SCN 5964663; FLASHBACK DATABASE TO SEQUENCE 12345; FLASHBACK DATABASE TO TIMESTAMP(SYSDATE -1/24); -- You can flash back to just before the last RESETLOGS operation by : FLASHBACK DATABASE TO SCN 5964663 TO BEFORE RESETLOGS; -- you can check the data before open resetlogs ALTER DATABASE OPEN READ ONLY; -- open resetlogs ALTER DATABASE OPEN RESETLOGS; -- to undo the results of the entire Flashback operation RECOVER DATABASE; /* Assissted commands */ -- current scn SELECT current_scn FROM V$DATABASE;
Restore Points
Guaranteed restore points use a separate logging mechanism from the Flashback logging used for a Flashback Database operation. However, if you use a guaranteed restore point and Flashback Database is enabled, Oracle wont delete any Flashback logs and thus FRA will eventually become full. Turning off Flashback Database if youre using guaranteed restore points. Oracle DBA Code Examples
Page 309
With guaranteed restore points, you can only flashback to exactly the restore point time. -- obtain info about rp: SELECT name, scn, storage_size, time, guarantee_flashback_database FROM v$restore_point;
-- creating ordinary restore point CREATE RESTORE POINT rp_test; DROP RESTORE POINT rp_test;
-- creating a guaranteed restore point CREATE RESTORE POINT test_guarantee GUARANTEE FLASHBACK DATABASE;
-- to use later the rp: FLASHBACK DATABASE TO RESTORE POINT test_guarantee;
-- to know if flashback db status: if only guaranteed rp is enabled, it'll return RESTORE POINT ONLY SELECT flashback_on FROM v$database;
Page 310
Part 4
Oracle Database Security
Page 311
Oracle Database Security Management

The main aspects of Oracle database security management: o o o o o Authorization: controlling access to data Authentication: restricting access to legitimate users Auditing: ensuring accountability on the part of the users Encryption: safeguarding key data in the database Enterprise Security: managing the security of the entire organizational information structure
Page 312
Security Guidelines
Apply Security Patches. http://technet.oracle.com/deploy/security/alerts.htm Enable "Automatic Secure Configuration" when creating a new database. Lock default users. Exceptions: SYS, SYSTEM, DBSNMP, SYSMAN, and MGMT_VIEW. Strict password aging and expiration policies. Do not use hard coded passwords in your script. If possible, avoid Operating System Authentication. Enable SYSDBA operations by setting AUDIT_SYS_OPERATIONS to TRUE. The audit records are written to the operating system's audit trail. Avoid granting ANY privileges and privileges with the ADMIN option. Whenver possible, use roles rather than granting privileges directly to users. Dont grant any unnecessary roles or privileges to PUBLIC. select count(*) from DBA_TAB_PRIVS where GRANTEE='PUBLIC'; Set the UMASK variable to 022 to the Oracle software OS user owner. Remove the SETUID on all Oracle files. Unless needed, remove mentions to EXTPROC in both the listener.ora file on the server and the tnsnames.ora file on the client. Then remove all EXTPROC executables from your $ORACLE_HOME/bin directory ( usually extproc and xtproc0). If needed, refere to Document ID 175429.1 in metalink. Set a password to the listener. Secure the linstener.ora file. ADMIN_RESTRICTIONS=ON You may use Server-side Access Controls in the sqlnet.ora file as follows: -- only the addresses in the list are allowed to make connections tcp.validnode_checking = yes tcp.invited_nodes = (server1.us.mycompany.com,172.14.16.152) -- addresses in the list are excluded tcp.excluded_nodes = (server1.us.mycompany.com,172.14.16.152) Disable remote OS authentication by setting the following parameter in the init.ora : REMOTE_OS_AUTHENT=FALSE Consider the proper setting of the security-related parameters: # password case sensitivity (11g) select value from v$parameter where upper(name)='SEC_CASE_SENSITIVE_LOGON' alter system set SEC_CASE_SENSITIVE_LOGON = TRUE; # maximum number of authentication attempts that can be made by a # client on a connection to the server process. Default 10 (11g) select value from v$parameter where lower(name)='sec_max_failed_login_attempts' alter system set sec_max_failed_login_attempts=10 scope=spfile; For application users: Oracle DBA Code Examples
Page 313
o o o o
Grant privileges to users through roles rather than direct access. Consider using application enabled roles. The users can make data changes only through procedures not using direct DML statements. Consider preventing normal users from using SQL*Plus:
-- disable INSERT INTO SYSTEM.PRODUCT_USER_PROFILE(PRODUCT,userid,attribute,char_value) VALUES('SQL*Plus','TESTER','ROLES','TEST123'); -- enable DELETE FROM product_user_profile WHERE userid='TESTER' AND char_value = 'TEST123';
Page 314
Managing Users
Recommendation: create default tablespace to every user. For managing resources, it might be better using Resource Manager. User Profile parameters: o Resource limit paramters: CONNECT_TIME: (in minutes) IDLE_TIME CPU_PER_CALL: CPU time used per each call CPU_PER_SESSION SESSIONS_PER_USER LOGICAL_READS_PER_SESSION LOGICAL_READS_PER_CALL PRIVATE_SGA: applicable only to shared server architecture systems COMPOSITE_LIMIT: o Password management parameters (see the following section): FAILED_LOGIN_ATTEMPTS PASSWORD_LIFE_TIME PASSWORD_GRACE_TIME PASSWORD_LOCK_TIME PASSWORD_REUSE_TIME PASSWORD_REUSE_MAX PASSWORD_VERIFY_FUNCTION -- creating user CREATE USER user1 IDENTIFIED BY urs1754 TEMPORARY TABLESPACE TEMPTBS01 DEFAULT TABLESPACE user1ts QUOTA 500M ON user1ts PROFILE 'SALES_RPOF'; GRANT CREATE SESSION TO salapati; ALTER USER user1 QUOTA 100M ON user1ts; GRANT UNLIMITED TABLESPACE TO user1; SELECT tablespace_name, username, bytes FROM DBA_TS_QUOTAS WHERE username='USER1'; -- dropping user DROP USER user1; DROP USER user1 cascade; REVOKE CREATE SESSION FROM user1;
Page 315
-- Creating and Using User Profiles CREATE PROFILE SALES_PROF LIMIT connect_time 120 failed_login_attempts 3 idle_time 60 sessions_per_user 2; ALTER PROFILE test LIMIT sessions_per_user 4 failed_login_attempts 4; -- to make Oracle enforce the resource limits in -- the profile, if they are used ALTER SYSTEM SET resource_limit=true; ALTER USER salapati PROFILE SALES_PROF; SELECT profile FROM dba_users WHERE username = 'USER1'; SELECT DISTINCT resource_name, limit FROM dba_profiles WHERE profile='DEFAULT';
-- Password Management Function -- utlpwdmg.sql script creates verify_function_11g ALTER PROFILE DEFAULT LIMIT PASSWORD_LIFE_TIME 180 PASSWORD_GRACE_TIME 7 PASSWORD_REUSE_TIME UNLIMITED PASSWORD_REUSE_MAX UNLIMITED FAILED_LOGIN_ATTEMPTS 10 PASSWORD_LOCK_TIME 1 PASSWORD_VERIFY_FUNCTION verify_function_11G; -- Dropping a User Profile DROP PROFILE test CASCADE;
Page 316
Database Authentication
Managin Passwords
select value from v$parameter where name='sec_case_sensitive_logon'; -- dynamic alter system set sec_case_sensitive_logon=false ; SELECT username, password, password_versions FROM dba_users order by 1; /* making a password expired */ ALTER USER hr IDENTIFIED BY hr PASSWORD EXPIRE; ALTER PROFILE test_profile LIMIT PASSWORD_LIFE_TIME 30; -- in days (refere to profile section) ALTER USER hr PROFILE test_profile; /* password file */ select value from v$parameter where upper(name)='REMOTE_LOGIN_PASSWORDFILE'; SELECT * FROM v$pwfile_users; orapwd FILE=testpwd PASSWORD=remorse1 ENTRIES=20
External (OS) Authentication

Following are the steps to setting up OS Authentication on UNIX/Linux and Windows platforms. # to enable external authentication over the net REMOTE_OS_AUTHENT=TRUE
# create the OS user useradd ahmedb passwd ahmedb # set the parameter OS_AUTHENT_PREFIX SHOW PARAMETER os_authent_prefix alter system set os_authent_prefix='ops$' scope=spfile; alter system set OS_AUTHENT_PREFIX = '' scope=spfile; # in DB: create the DB user with the prefix # in Unix CREATE USER ops$ahmedb IDENTIFIED EXTERNALLY; GRANT CONNECT TO ops$ahmedb; # in Windows
Page 317
CREATE USER "OPS$MYDOMAIN.COM\AHMEDB" IDENTIFIED EXTERNALLY; GRANT CONNECT TO "OPS$MYDOMAIN.COM\AHMEDB"; # in Windows: # set the following in the file "%ORACLE_HOME%\network\admin\sqlnet.ora": SQLNET.AUTHENTICATION_SERVICES= (NTS) # test: su - ahmedb export ORACLE_HOME=/u01/app/oracle/product/11.2.0/db_1 export PATH=$PATH:$ORACLE_HOME/bin export ORACLE_SID=ora11gr2 sqlplus /
Proxy Authentication
-- to authorize connections by a database user logging on from -- a middle-tier node, using password authentication. ALTER USER user1 GRANT CONNECT THROUGH appserv AUTHENTICATED USING PASSWORD; -ALTER USER user1 GRANT CONNECT THROUGH appserv;
Logging In As a Different User

If you want to log on as a different user and you do not know his password. -- 1) obtain his encrypted password SELECT 'alter user tester identified by values '||password||';' FROM user$ WHERE username='HR'; -- 2) set a new password to him ALTER USER hr IDENTIFIED BY newpassword; -- 3) log on using the new password and execute your commands -- 4) reset the user password back to the original one (note the qout) ALTER USER tester IDENTIFIED BY VALUES '1825ACAA229030F1';
Killing User Sessions from OS

-- on Unix -- 1. obtain process# SELECT process, sid, serial# FROM v$session WHERE username='&user'; -- 2. kill the process kill -9 345678
Page 318
-- on Windows -- 1. obain thread # SELECT sid, spid as thread, osuser, s.program FROM v$process p, v$session s WHERE p.addr = s.paddr; -- kill the process using orakill utiliy (orakill DBsid thread#) orakill MyDB 6200
Page 319
Controlling Database Access

System and Object Privileges
/* system privs */ GRANT CREATE SESSION TO hr; GRANT CREATE SESSION TO hr WITH ADMIN OPTION; GRANT ANY OBJECT to hr; -- only SELECT ANY DICTIONARY is NOT included GRANT ALL PRIVILEGES TO ourdba; REVOKE DELETE ANY TABLE FROM ourdba; -- dictionary views become accessible grant SELECT_CATALOG_ROLE to ourdev;
/* object privs */ GRANT DELETE ON bonuses TO hr WITH GRANT OPTION; GRANT UPDATE (product_id) ON sales01 TO hr; GRANT SELECT, UPDATE ON emp_view TO PUBLIC; GRANT SELECT ON oe.customers_seq TO hr; GRANT ALL ON EMPLOYEES TO hr; GRANT EXECUTE ON employee_pkg TO hr; GRANT QUERY REWRITE TO hr; GRANT READ ON DIRECTORY bfile_dir TO hr; -- incorrect REVOKE UPDATE (hostname) ON ods_process FROM hr; -- correct REVOKE UPDATE ON ods_process FROM hr;
Invoker Rights and Definer Rights

CREATE OR REPLACE PROCEDURE delete_emp (p_emp_id number) AUTHID CURRENT_USER IS BEGIN DELETE FROM emp WHERE emp_id = p_emp_id; COMMIT; END; /
Roles
Predefined roles: Oracle DBA Code Examples
Page 320
o o o o o o
CONNECT RESOURCE: CREATE CLUSTER, CREATE INDEXTYPE, CREATE OPERATOR, CREATE PROCEDURE, CREATE SEQUENCE, CREATE TABLE, CREATE TRIGGER, CREATE TYPE DBA EXP_FULL_DATABASE IMP_FULL_DATABASE RECOVERY_CATALOG_OWNER
-- create role CREATE ROLE new_dba; /* Role Authorization */ -- (1) Database authorization CREATE ROLE clerk IDENTIFIED BY password; -- (2) Database authorization with a PL/SQL package -- the role is enabled by a hr.admin package: CREATE ROLE admin_role IDENTIFIED USING hr.admin; -- (3) Externally CREATE ROLE accts_rec IDENTIFIED EXTERNALLY; -- (4) Globally: enabled by an enterprise directory service CREATE ROLE supervisor IDENTIFIED GLOBALLY; /* Disabling and Enabling a Role */ -- disable INSERT INTO SYSTEM.PRODUCT_USER_PROFILE(PRODUCT,userid,attribute,char_value) VALUES('SQL*Plus','TESTER','ROLES','TEST123'); -- enable DELETE FROM product_user_profile WHERE userid='TESTER' AND char_value = 'TEST123'; -- Dropping a Role DROP ROLE admin_user;
Users, Roles, and Privileges Views

DBA_USERS DBA_ROLES DBA_COL_PRIVS Shows DBA_ROLE_PRIVS DBA_SYS_PRIVS Shows DBA_TAB_PRIVS Shows ROLE_ROLE_PRIVS ROLE_SYS_PRIVS ROLE_TAB_PRIVS SESSION_PRIVS Shows SESSION_ROLES Shows Provides information about users Shows all the roles in the database column-level object grants Shows users and their roles users who have been granted system privileges users and their privileges on tables Shows roles granted to roles Shows system privileges granted to roles Shows table privileges granted to roles privileges currently enabled for the current session roles currently enabled for the current session
-- list of everything granted to a user: SELECT 'GRANT ' || PRIVILEGE||' to "'||grantee||'";' FROM dba_sys_privs WHERE
Page 321
grantee='FINANCE' union all SELECT 'GRANT ' || GRANTED_ROLE||' to "'||grantee||'";' FROM dba_role_privs WHERE grantee='FINANCE' union all SELECT 'GRANT ' || PRIVILEGE||' to "'||grantee||'";' FROM dba_tab_privs WHERE grantee='FINANCE';
Fine-Grained Data Access (Virtual Private Database VPD)

Fine-grained security within the database is implemented by DBMS_RLS Fine-grained access control policy: attached to DB object. It can be applied to SELECT, INSERT, UPDATE, INDEX, and DELETE statements. SYS user is not affected by FGAC. Common Predefined Attributes in the USERENV instance entryID current_user session_user db_name host os_user terminal ip_address IP external_name Instance ID Auditing entry identifier Name of the user who started the session Database username by which the current user is authenticated Name of the machine on which the database is running Operating system account name Client terminal through which the database is being accessed address of the client machine External name of the database user Name of the database
/* Using Application Context */ -- using pre-defined context SELECT sys_context ('USERENV', 'OS_USER') FROM DUAL; SELECT first_name,last_name,employee_id FROM employees WHERE UPPER(last_name)=sys_context('USERENV', 'SESSION_USER'); -- using user-defined application context: var=val pair per session CONNECT system/system_passwd; GRANT CREATE ANY CONTEXT TO hr; CONNECT hr/hr; CREATE CONTEXT employee_info USING hr.context; CREATE OR REPLACE PACKAGE hr_context AS PROCEDURE select_emp_no ; END; / CREATE OR REPLACE PACKAGE BODY hr_context as PROCEDURE select_emp_no IS empnum number; BEGIN SELECT employee_id INTO empnum FROM employees WHERE UPPER(last_name) = sys_context('USERENV', 'SESSION_USER');
Page 322
DBMS_SESSION.SET_CONTEXT('EMPLOYEE_INFO', 'EMP_NUM', EMPNUM); END select_emp_no; END; / -- set application context: upon login trigger CREATE OR REPLACE TRIGGER hr.security_context AFTER LOGON ON DATABASE BEGIN hr_context.select_emp_no; END; / /* using fine-grained access control */ -- here are the steps --(1) define the policy function: which will generate the predicates -- the function must adhere to the following syntax: -- FUNCTION policy_function (object_schema IN VARCHAR2, object_name VARCHAR2) -RETURN VARCHAR2 returning max of 2000 Bytes CREATE OR REPLACE PACKAGE hr_security AS FUNCTION empnum_sec (A1 VARCHAR2, A2 VARCHAR2) RETURN varchar2; END; / CREATE OR REPLACE PACKAGE BODY hr_security AS FUNCTION empnum_sec (A1 VARCHAR2, A2 VARCHAR2) RETURN varchar2 IS d_predicate varchar2 (2000); BEGIN d_predicate:= 'employee_id = SYS_CONTEXT("EMPLOYEE_INFO","EMP_NUM")'; RETURN d_predicate; END empnum_sec; END hr_security; / -- to make all db user being able to use it GRANT EXECUTE ON hr_security TO public; --(2) create security policy RLS=row-level security -- POLICY_TYPE parameter in DBMS_RLS.ADD_POLICY takes: -- DBMS_RLS.DYNAMIC (default) -- DBMS_RLS.STATIC -- DBMS_RLS.SHARED_STATIC -- DBMS_RLS.CONTEXT_SENSITIVE -- DBMS_RLS.SHARED_CONTEXT_SENSITIVE BEGIN DBMS_RLS.ADD_POLICY (OBJECT_SCHEMA=>'HR', OBJECT_NAME=>'EMPLOYEES', POLICY_NAME=>'MANAGER_POLICY',
Page 323
FUNCTION_SCHEMA=>'HR', POLICY_FUNCTION=>'hr_security.empnum_sec', -- func generates the predicates STATEMENT_TYPES='SELECT' ); END; / SELECT object_name, policy_name, sel, ins, upd, del, enable FROM all_policies; -- Column-Level VPD -- You can apply column-level VPD to a table or a view BEGIN DBMS_RLS.ADD_POLICY (OBJECT_SCHEMA=>'HR', OBJECT_NAME=>'EMPLOYEES', POLICY_NAME=>'MANAGER_POLICY', FUNCTION_SCHEMA=>'HR', POLICY_FUNCTION=>'hr_security.empnum_sec, STATEMENT_TYPE=INSERT,UPDATE, SEC_RELEVANT_COLS=>'salary,commission') END; / /* Exempting from Access Policy */ GRANT EXEMPT ACCESS POLICY to hr;
Page 324
Auditing Database
Standard Auditing
Set AUDIT_TRAIL to: NONE (default), OS, DB (SYS.AUD$), DB_EXTENDED (SYS.AUD$ + the columns SQLBIND and SQLTEXT CLOB), XML (in OS), 'XML, EXTENDED'. If you set AUDIT_TRAIL to DB, change the tablespace of SYS.AUD$ from SYSTEM. Audited info: o o o o o o Operating system login Database username Terminal and session identifiers Operation performed or attempted Date and time stamp SQL text that triggered the auditing
select value from v$parameter where name='audit_trail'; -- if not defined: $ORACLE_HOME/rdbms/audit/ select value from v$parameter where name='audit_file_dest'; alter system set audit_trail=db_extended scope=spfile; ALTER SYSTEM SET audit_trail = xml,extended SCOPE=SPFILE ; -- AUDIT focused by -- DB/user -- success/failure and -- grouped by session or access audit session ; AUDIT SESSION BY hr; AUDIT select table BY hr BY SESSION; AUDIT DELETE ANY TABLE BY hr WHENEVER NOT SUCCESSFUL; AUDIT UPDATE ANY TABLE; AUDIT SELECT,INSERT,UPDATE,DELETE ON employees BY ACCESS WHENEVER SUCCESSFUL; AUDIT ALL PRIVILEGES; /* turn NOAUDIT NOAUDIT NOAUDIT NOAUDIT audit off */ select table BY hr; ALL; /* turns off all statement auditing */ ALL PRIVILEGES; /* turns off all privilege auditing */ ALL ON DEFAULT; /* turns off all object auditing */
/* obtain info on audit */ select * from DBA_STMT_AUDIT_OPTS; select OS_USERNAME, USERNAME, USERHOST, ACTION_NAME, to_char(LOGOFF_TIME,'dd-mm-yy hh24:mi:ss') LOGOFF, SESSION_CPU from DBA_AUDIT_SESSION; select USERNAME, USERHOST, ACTION_NAME, OBJ_NAME from DBA_AUDIT_OBJECT
Page 325
ORDER BY USERNAME;
/* flushing db audit */ CONN / AS SYSDBA DELETE FROM SYS.AUD$;
Customizing Database Auditing with Triggers

/* The following procedure generate code required to write an auditing trigger Usage: execute the procedure and pass the table name. Then select from VW order by I to view the code */ CREATE TABLE VW ( I NUMBER, X VARCHAR2(4000)); SEQUENCE SEQ_I ; CREATE PROCEDURE CREATE_AUDIT2 ( P_TABLE VARCHAR2) IS V_NCOL_LIST VARCHAR2(4000) ; V_OCOL_LIST VARCHAR2(4000) ; V_NCOL_LIST2 VARCHAR2(4000) ; V_OCOL_LIST2 VARCHAR2(4000) ; V_OUTPUT VARCHAR2(4000); PROCEDURE INSERT_VW( V VARCHAR2) IS BEGIN INSERT INTO VW VALUES(SEQ_I.NEXTVAL, V ); END; BEGIN -- ** generate the target Audit Table -- create sequence INSERT_VW('CREATE SEQUENCE SEQ_'|| P_TABLE || '_AUDIT ;'); INSERT_VW('CREATE TABLE '|| P_TABLE || '_AUDIT ('); INSERT_VW('ID NUMBER CONSTRAINT '|| P_TABLE || '_AUDIT_ID PRIMARY KEY,'); -- N Columns FOR R IN ( SELECT TABLE_NAME,COLUMN_NAME,DATA_TYPE,DATA_LENGTH,DATA_PRECISION,DATA_SCALE FROM USER_TAB_COLUMNS WHERE TABLE_NAME = UPPER(P_TABLE) ) LOOP IF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NULL THEN INSERT_VW( 'N' || R.COLUMN_NAME || ' NUMBER ,'); V_NCOL_LIST := V_NCOL_LIST || 'N' || R.COLUMN_NAME || ','; V_NCOL_LIST2 := V_NCOL_LIST2 || ' :NEW.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NOT NULL AND R.DATA_SCALE = 0 THEN INSERT_VW( 'N' || R.COLUMN_NAME || ' NUMBER('|| R.DATA_PRECISION ||'),'); V_NCOL_LIST := V_NCOL_LIST || 'N' || R.COLUMN_NAME || ','; V_NCOL_LIST2 := V_NCOL_LIST2 || ' :NEW.' || R.COLUMN_NAME || ',';
Page 326
ELSIF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NOT NULL AND R.DATA_SCALE <> 0 THEN INSERT_VW( 'N' || R.COLUMN_NAME || ' NUMBER('|| R.DATA_PRECISION ||','|| R.DATA_SCALE ||'),'); V_NCOL_LIST := V_NCOL_LIST || 'N' || R.COLUMN_NAME || ','; V_NCOL_LIST2 := V_NCOL_LIST2 || ' :NEW.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'VARCHAR2' THEN INSERT_VW( 'N' || R.COLUMN_NAME || ' VARCHAR2('|| R.DATA_LENGTH ||'),'); V_NCOL_LIST := V_NCOL_LIST || 'N' || R.COLUMN_NAME || ','; V_NCOL_LIST2 := V_NCOL_LIST2 || ' :NEW.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'DATE' THEN INSERT_VW( 'N' || R.COLUMN_NAME || ' DATE,'); V_NCOL_LIST := V_NCOL_LIST || 'N' || R.COLUMN_NAME || ','; V_NCOL_LIST2 := V_NCOL_LIST2 || ' :NEW.' || R.COLUMN_NAME || ','; END IF; END LOOP; -- O columns FOR R IN ( SELECT TABLE_NAME,COLUMN_NAME,DATA_TYPE,DATA_LENGTH,DATA_PRECISION,DATA_SCALE FROM USER_TAB_COLUMNS WHERE TABLE_NAME = UPPER(P_TABLE) ) LOOP IF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NULL THEN INSERT_VW( 'O' || R.COLUMN_NAME || ' NUMBER ,'); V_OCOL_LIST := V_OCOL_LIST || 'O' || R.COLUMN_NAME || ','; V_OCOL_LIST2 := V_OCOL_LIST2 || ' :OLD.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NOT NULL AND R.DATA_SCALE = 0 THEN INSERT_VW( 'O' || R.COLUMN_NAME || ' NUMBER('|| R.DATA_PRECISION ||'),'); V_OCOL_LIST := V_OCOL_LIST || 'O' || R.COLUMN_NAME || ','; V_OCOL_LIST2 := V_OCOL_LIST2 || ' :OLD.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'NUMBER' AND R.DATA_PRECISION IS NOT NULL AND R.DATA_SCALE <> 0 THEN INSERT_VW( 'O' || R.COLUMN_NAME || ' NUMBER('|| R.DATA_PRECISION ||','|| R.DATA_SCALE ||'),'); V_OCOL_LIST := V_OCOL_LIST || 'O' || R.COLUMN_NAME || ','; V_OCOL_LIST2 := V_OCOL_LIST2 || ' :OLD.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'VARCHAR2' THEN INSERT_VW( 'O' || R.COLUMN_NAME || ' VARCHAR2('|| R.DATA_LENGTH ||'),'); V_OCOL_LIST := V_OCOL_LIST || 'O' || R.COLUMN_NAME || ','; V_OCOL_LIST2 := V_OCOL_LIST2 || ' :OLD.' || R.COLUMN_NAME || ','; ELSIF R.DATA_TYPE = 'DATE' THEN INSERT_VW( 'O' || R.COLUMN_NAME || ' DATE,'); V_OCOL_LIST := V_OCOL_LIST || 'O' || R.COLUMN_NAME || ','; V_OCOL_LIST2 := V_OCOL_LIST2 || ' :OLD.' || R.COLUMN_NAME || ','; END IF; END LOOP; INSERT_VW( 'ACTION VARCHAR2(6),'); INSERT_VW( 'ACTION_TIME DATE,'); INSERT_VW( 'HOST_NAME VARCHAR2(300),'); INSERT_VW( 'OS_USERNAME VARCHAR2(250));'); -- generate the Auditing Trigger
Page 327
INSERT_VW('CREATE OR REPLACE TRIGGER TRG_' || P_TABLE || '_AUDIT'); INSERT_VW('AFTER INSERT OR UPDATE OR DELETE ON '|| P_TABLE); INSERT_VW('FOR EACH ROW'); INSERT_VW('BEGIN'); INSERT_VW('IF INSERTING THEN'); INSERT_VW('INSERT INTO ' || P_TABLE || '_AUDIT(ID,'); INSERT_VW(V_NCOL_LIST); INSERT_VW('ACTION,ACTION_TIME,HOST_NAME,OS_USERNAME)'); INSERT_VW('VALUES'); INSERT_VW('(SEQ_'|| P_TABLE || '_AUDIT.NEXTVAL,'); INSERT_VW( V_NCOL_LIST2); INSERT_VW( '''INSERT'',sysdate, SYS_CONTEXT(''USERENV'',''HOST''),SYS_CONTEXT(''USERENV'',''OS_USER''));'); INSERT_VW('END IF;'); INSERT_VW('IF DELETING THEN'); INSERT_VW('INSERT INTO ' || P_TABLE || '_AUDIT(ID,'); INSERT_VW(V_OCOL_LIST); INSERT_VW('ACTION,ACTION_TIME,HOST_NAME,OS_USERNAME)'); INSERT_VW('VALUES'); INSERT_VW('(SEQ_'|| P_TABLE || '_AUDIT.NEXTVAL,'); INSERT_VW( V_OCOL_LIST2); INSERT_VW( '''DELETE'', SYSDATE,SYS_CONTEXT(''USERENV'',''HOST''),SYS_CONTEXT(''USERENV'',''OS_USER'') );'); INSERT_VW('END IF;'); INSERT_VW('IF UPDATING THEN'); INSERT_VW('INSERT INTO ' || P_TABLE || '_AUDIT(ID,'); INSERT_VW(V_NCOL_LIST ); INSERT_VW( V_OCOL_LIST); INSERT_VW('ACTION,ACTION_TIME,HOST_NAME,OS_USERNAME)'); INSERT_VW('VALUES'); INSERT_VW('(SEQ_'|| P_TABLE || '_AUDIT.NEXTVAL,'); INSERT_VW( V_NCOL_LIST2 ); INSERT_VW( V_OCOL_LIST2 ); INSERT_VW( '''UPDATE'', SYSDATE,SYS_CONTEXT(''USERENV'',''HOST''),SYS_CONTEXT(''USERENV'',''OS_USER'') );'); INSERT_VW('END IF;'); INSERT_VW('END;'); COMMIT; END CREATE_AUDIT2;
Auditing the Database Using System Trigger

System-Level trigger types: o o o o
Page 328
Database startup Logon and Logoff DDL Server error Oracle DBA Code Examples
/* Required privs */ grant ADMINISTER DATABASE TRIGGER to user1; /* Obtain Info about DB triggers */ -- list the db triggers SELECT a.obj#, a.sys_evts, b.name FROM trigger$ a,obj$ b WHERE a.sys_evts > 0 AND a.obj#=b.obj# AND baseobject = 0; /* Examples */ /* If db cannot start because of an error in the AFTER STARTUP trigger */ set linesize 150 col NAME format a30 col VALUE format a20 col DESCRIPTION format a60 SELECT x.ksppinm NAME, y.ksppstvl VALUE, ksppdesc DESCRIPTION FROM x$ksppi x, x$ksppcv y WHERE x.inst_id = userenv('Instance') AND y.inst_id = userenv('Instance') AND x.indx = y.indx AND x.ksppinm = '_system_trig_enabled'; -- enable or disable db triggers ALTER SYSTEM SET "_system_trig_enabled" = TRUE SCOPE=BOTH; -- Log On Log Off trigger CREATE OR REPLACE TRIGGER logon_audit AFTER LOGON ON DATABASE BEGIN INSERT INTO connection_audit (login_date, user_name) VALUES (SYSDATE, USER); END logon_audit; / CREATE OR REPLACE TRIGGER logoff_audit_trig AFTER LOGOFF ON DATABASE BEGIN INSERT INTO logon_audit VALUES (user, sys_context('userenv', 'sessionid'), null, sysdate, sys_context('userenv', 'host')); END; / -- trigger to trap unsuccessful logons /* other errors that could be trapped include: ORA-01004 - default username feature not supported ORA-01005 - null password given ORA-01035 - Oracle only available to users with restricted session priv ORA-01045 - create session privilege not granted */ CREATE OR REPLACE TRIGGER logon_failures AFTER SERVERERROR ON DATABASE BEGIN
Page 329
IF (IS_SERVERERROR(1017)) THEN INSERT INTO connection_audit (login_date, user_name) VALUES (SYSDATE, 'ORA-1017'); END IF; END logon_failures; / -- DDL triggers /* BEFORE / AFTER ALTER BEFORE / AFTER ANALYZE BEFORE / AFTER ASSOCIATE STATISTICS BEFORE / AFTER AUDIT BEFORE / AFTER COMMENT BEFORE / AFTER CREATE BEFORE / AFTER DDL BEFORE / AFTER DISASSOCIATE STATISTICS BEFORE / AFTER DROP BEFORE / AFTER GRANT BEFORE / AFTER NOAUDIT BEFORE / AFTER RENAME BEFORE / AFTER REVOKE BEFORE / AFTER TRUNCATE AFTER SUSPEND */ CREATE OR REPLACE TRIGGER ddl_log_trig AFTER DDL ON DATABASE BEGIN INSERT INTO ddl_log (username, change_date, object_type, object_owner, database, event_name ) VALUES (ORA_LOGIN_USER, sysdate, ora_dict_obj_type, ora_dict_obj_owner, ora_database_name, ora_sysevent) END; / -- Disable granting privileges to PUBLIC CREATE OR REPLACE TRIGGER ddl_trig BEFORE GRANT ON DATABASE DECLARE g_list DBMS_STANDARD.ORA_NAME_LIST_T; n PLS_INTEGER; BEGIN n := ORA_GRANTEE(g_list); FOR i IN 1..n LOOP IF g_list(i) = 'PUBLIC' THEN RAISE_APPLICATION_ERROR(-20997,'Public Grants Not Allowed'); END IF; END LOOP; END; / -- System Errors CREATE TABLE servererror_log ( error_datetime TIMESTAMP, error_user VARCHAR2(30), db_name VARCHAR2(9), error_stack VARCHAR2(2000), captured_sql VARCHAR2(1000)); CREATE OR REPLACE TRIGGER log_server_errors AFTER SERVERERROR ON DATABASE
Page 330
DECLARE captured_sql VARCHAR2(1000); BEGIN SELECT q.sql_text INTO captured_sql FROM gv$sql q, gv$sql_cursor c, gv$session s WHERE s.audsid = audsid AND s.prev_sql_addr = q.address AND q.address = c.parent_handle; INSERT INTO servererror_log (error_datetime, error_user, db_name, error_stack, captured_sql) VALUES (systimestamp, sys.login_user, sys.database_name, dbms_utility.format_error_stack, captured_sql); END log_server_errors; /
Using Fine Grained Auditing

------ Show all currently active FGA Policies in the database ----COL object_schema FORMAT A10 HEADING 'Object|Schema' COL object_name FORMAT A20 HEADING 'Object Name' WRAP COL policy_name FORMAT A16 HEADING 'Policy Name' WRAP COL policy_text FORMAT A24 HEADING 'Policy Text' WRAP COL policy_column FORMAT A16 HEADING 'Policy Column' WRAP COL enabled FORMAT A05 HEADING 'On?' COL siud_options FORMAT A04 HEADING 'SIUD|Set' SELECT policy_name ,policy_text ,policy_column ,enabled ,object_schema ,object_name ,DECODE(sel,'YES','Y','N') || DECODE(ins,'YES','Y','N')|| DECODE(upd,'YES','Y','N') || DECODE(del,'YES','Y','N') siud_options FROM dba_audit_policies; -- required priv grant execute on DBMS_FGA to hr; ------ Show Fine-Grained Auditing results so far -- the view is based on SYS.FGA_LOG$ ----TTITLE 'Current Fine-Grained Auditing (FGA) Results' COL audit_date FORMAT A10 HEADING 'Audit|Date' COL policy_name FORMAT A16 HEADING 'Policy Name' WRAP COL object_schema FORMAT A10 HEADING 'Object|Schema' COL object_name FORMAT A20 HEADING 'Object Name' WRAP COL db_user FORMAT A10 HEADING 'DBUser' COL sql_text FORMAT A36 HEADING 'SQL Text' WRAP SELECT TO_CHAR(timestamp,'dd/mm/yyyy hh24:mi:ss') audit_date ,db_user ,object_schema ,object_name
Page 331
,policy_name ,sql_text FROM dba_fga_audit_trail ORDER BY timestamp;
------ Use the new combined audit trail view (DBA_COMMON_AUDIT_TRAIL) -- to see results of both Standard (i.e. AUDIT) and Fine-Grained -- Auditing (i.e. via DBMS_FGA) ----COL audtype FORMAT A03 HEADING 'Aud|Typ' COL db_user FORMAT A10 HEADING 'DBUser' COL object_schema FORMAT A06 HEADING 'Object|Schema' COL object_name FORMAT A20 HEADING 'Object Name' WRAP COL policy_name FORMAT A16 HEADING 'Policy Name' WRAP COL audit_date FORMAT A10 HEADING 'Audit|Date' COL sql_text FORMAT A32 HEADING 'SQL Text' WRAP SELECT DECODE(audit_type, 'Fine Grained Audit', 'FGA' ,'Standard Audit', 'STD' ,'UNK') audtype ,db_user ,object_schema ,object_name ,policy_name ,TO_CHAR(extended_timestamp,'mm/dd/yyyy hh24:mi:ss') audit_date ,sql_text FROM dba_common_audit_trail WHERE db_user NOT IN ('SYS','SYSTEM','DBSNMP','SYSMAN') ORDER BY extended_timestamp, db_user, object_schema, object_name; ----BEGIN DBMS_FGA.ADD_POLICY( object_schema => 'AP' -- if null, logon user schema ,object_name => 'VENDORS' ,policy_name => 'VENDORS_LO' ,audit_condition => 'ACTIVE_IND <> ''Y''' -- if NULL=TRUE ,audit_column => 'ACTIVE_IND,CREDIT_CARD,CREDIT_LIMIT' ,handler_schema => NULL ,handler_module => NULL -- the procedure will fire on audit ,enable => TRUE -- default is TRUE ,statement_types => 'SELECT' ,audit_trail => DBMS_FGA.DB_EXTENDED ,audit_column_opts => DBMS_FGA.ANY_COLUMNS -- or DBMS_FGA.ALL_COLUMNS ); END; / BEGIN DBMS_FGA.ADD_POLICY( object_schema => 'AP' ,object_name => 'RV_INVOICE_DETAILS' ,policy_name => 'RV_INVOICE_LO' ,audit_condition => NULL ,audit_column => 'VENDOR_NAME,INVOICE_ID,EXTENDED_AMT,VENDOR_CREDIT_LIMIT' ,handler_schema => NULL ,handler_module => NULL ,enable => FALSE ,statement_types => 'SELECT'
Page 332
,audit_trail => DBMS_FGA.DB_EXTENDED ,audit_column_opts => DBMS_FGA.ALL_COLUMNS ); END; /
/* || Listing 1.5: FGA Policy Maintenance */ BEGIN ------ Disabling an enabled, existing FGA policY ----DBMS_FGA.DISABLE_POLICY( object_schema => 'AP' ,object_name => 'INVOICES' ,policy_name => 'INVOICES_HI' ); ------ Dropping an enabled, existing FGA policY ----DBMS_FGA.DROP_POLICY( ,policy_name => 'RV_INVOICE_LOW' ,object_name => ' RV_INVOICE_DETAILS' ,policy_name => 'RV_INVOICE_LO' ); END; /
Page 333
Using Data Encryption

Oracle Transparent Data Encryption (TDE)
Caution: Wallet file must be included in your backup. Column length changes on disk. Actual lengths not reported by DUMP or VSIZE. The Wallet must be opened after instance restart. /* Setting up TDE */ -- 1. Create the Wallet file: -- add the following to the sqlnet.ora ENCRYPTION_WALLET_LOCATION = (SOURCE= (METHOD=file) (METHOD_DATA= (DIRECTORY=C:\oracle\OraDb10g\admin\ora10g\wallet))) -- 2. Set the master key: this is done only once: ALTER SYSTEM SET ENCRYPTION KEY IDENTIFIED BY <password>; -- 3. Create tables that contain encrypted columns -- possible algorithms are AES128, (AES192), AES256, or 3DES168 -- the salt increases the protection but prevents indexing on the column. CREATE TABLE emp ( first_name VARCHAR2(128), ... empID NUMBER ENCRYPT NO SALT, salary NUMBER(6) ENCRYPT USING '3DES168', comm NUMBER(6) ENCRYPT); ALTER TABLE EMP MODIFY ( SAL ENCRYPT NO SALT ); DESC EMP /* Existing Tables and TDE */ --Add encrypted columns: ALTER TABLE emp ADD (ssn VARCHAR2(11) ENCRYPT); Encrypt unencrypted columns: ALTER TABLE emp MODIFY (first_name ENCRYPT); Disable column encryption: ALTER TABLE emp MODIFY (first_name DECRYPT); --Add or remove salt: ALTER TABLE emp MODIFY (first_name ENCRYPT [NO] SALT); --Change keys and the encryption algorithm: ALTER TABLE emp REKEY USING '3DES168'; -- To Test TDE SELECT DBMS_ROWID.ROWID_TO_ABSOLUTE_FNO (ROWID,USER,'EMP'), DBMS_ROWID.ROWID_BLOCK_NUMBER (ROWID)
Page 334
FROM EMP; File Nu Block Number ------- -----------4 63 ALTER SESSION SET EVENTS '10389 trace name context forever, level 1'; ALTER SYSTEM DUMP DATAFILE 4 BLOCK 63; /* Opening and Closing the Wallet */ -- The Wallet must be opened after instance restart. ALTER SYSTEM SET ENCRYPTION WALLET OPEN IDENTIFIED BY password> ALTER SYSTEM SET ENCRYPTION WALLET CLOSE -- To verify that a wallet is open, you can query the V$ENCRYPTION_WALLET: select WRL_PARAMETER, STATUS from V$ENCRYPTION_WALLET
Tablespace Encryption
In Oracle Database 11g, you can encrypt an entire tablespace. Encrypted Tablespace Limitations You cannot encrypt an existing tablespace. exp and imp utilities are not supported with objects in the encrypted tablespaces. Whereas expdp and impdp utilities are supported. You cannot re-create the tablespace encryption key. The NO SALT option is not supported. Temporary and undo tablespaces cannot be encrypted. You cannot transport an encrypted tablespace to a database that already has an Oracle wallet for TDE. In this case, use Oracle Data Pump to export the objects in the tablespace using the expdp with ENCRYPTION_MODE=password and then import them to the destination database. BFILES and external tables are not encrypted. Logically, encrypted tablespace is less efficient than normal un-encrypted tablespace.
Caution Losing the master key or the wallet file will lead to losing the data in the encrypted tablespace. Encrypting a Tablespace Create and open a wallet file, as explained in the previous section. The tablespace creation statement for an encrypted tablespace has the following syntax: CREATE TABLESPACE <tbsp_name> ... [ENCRYPTION [USING <ALGORITHM>]] -- specify encryption algorithm DEFAULT STORAGE(ENCRYPT) -- encrypt objects in the tablespace CREATE TABLESPACE tbsp1 DATAFILE '/u01/app/oracle/test/tbsp1_01.dbf' SIZE 500m ENCRYPTION
Page 335
DEFAULT STORAGE (ENCRYPT); CREATE TABLESPACE mytbsp2 DATAFILE '/u01/app/oracle/test/mytbsp2_01.dbf' size 500m ENCRYPTION USING '3DES168' DEFAULT STORAGE (ENCRYPT); The ALGORITHM clause accepts one of the following values: o o o o AES192 AES128 AES256 Advanced Encryption Standard (the default). Advanced Encryption Standard 128-bit encryption Advanced Encryption Standard 256-bit encryption 3DES168 Triple Data Encryption Standard 168-bit encryption
To know whether an existing tablespace is encrypted or not, issue the following query: select vt.NAME, vet.ENCRYPTIONALG, vet.ENCRYPTEDTS from V$ENCRYPTED_TABLESPACES vet, V$TABLESPACE vt where vet.TS#=vt.TS# SELECT tablespace_name, encrypted 2 FROM dba_tablespaces;
Page 336
Fine-Grained Access Control for UTL_* Packages

Oracle Database 11g provides a mechanism to refine the level of access to the network access packages UTL_TCP, UTL_SMTP, UTL_MAIL, UTL_HTTP, and UTL_INADDR.
Creating ACL
You can use the DBMS_NETWORK_ACL_ADMIN package to facilitate management of the UTL_* network access packages as in the following steps: 1) Create an Access Control List (ACL): All ACL definitions are stored in XML DB in the form of XML documents. The ACL XML files reside in the /sys/acls directory of the XML DB repository. Following is an example of using the CREATE_ACL procedure to create an XML file called dba.xml: begin DBMS_NETWORK_ACL_ADMIN.CREATE_ACL ( ACL => 'dba.xml', -- case sensitive DESCRIPTION=> 'Network Access Control for the DBAs', PRINCIPAL => 'SCOTT', -- user or role the privilege is granted or denied (upper case) IS_GRANT => TRUE, -- privilege is granted or denied PRIVILEGE => 'connect', -- or 'resolve' (case sensitive) START_DATE => null, -- when the access control entity ACE will be valid END_DATE => null); -- ACE expiration date (TIMESTAMP WITH TIMEZONE format) end; Regarding the PRIVILEGE parameter, the database user needs the connect privilege to an external network host computer if he or she is connecting using the UTL_TCP, UTL_HTTP, UTL_SMTP, and UTL_MAIL utility packages. To resolve a host name that was given as a host IP address, or the IP address that was given as a host name, with the UTL_INADDR package, grant the database user the resolve privilege. You can then query the RESOURCE_VIEW view to find the dba.xml ACL in the /sys/acls directory: select ANY_PATH from RESOURCE_VIEW where ANY_PATH LIKE '/sys/acls/dba%' Too may entries in the ACL may lead to significant XML DB performance drop because ACL are checked for each access to Oracle XML DB repository. As general rule of thumb, ACL check operations perform best when the number of ACEs in the ACL is at 16 entries or less. 2) Add Access Control Entries: Once you create the initial ACL, you can continue to add more privileges to the XML file. The following example will add the user RAMI to the dba.xml file and grant him network access: begin DBMS_NETWORK_ACL_ADMIN.ADD_PRIVILEGE ( ACL => 'dba.xml', PRINCIPAL => 'RAMI', IS_GRANT => TRUE,
Page 337
PRIVILEGE => 'connect', START_DATE => null, -- if the time interval is defined, END_DATE => null); -- the ACE will expire after the specified date range end; / COMMIT; In ACL, the security entries are evaluating in order precedence. If you have two contradicting entries in the list, the first one in the order will take effect. You can control the order number of an added entry as follows: begin DBMS_NETWORK_ACL_ADMIN.ADD_PRIVILEGE ( POSITION => 1, -- on the top ACL => 'dba.xml', PRINCIPAL => 'SAMI', IS_GRANT => FALSE, PRIVILEGE => 'connect', START_DATE => null, END_DATE => null); end; 3) Assign Hosts: The ASSIGN_ACL procedure is used to authorize access to one or more network hosts as follows: begin DBMS_NETWORK_ACL_ADMIN.ASSIGN_ACL ( ACL => 'dba.xml', HOST => 'dbaexpert.com', LOWER_PORT => 80, UPPER_PORT => 443); end; COMMIT; The lower port and the upper port define the lower and the upper boundaries of the allowable port range. They should be set for connect privileges not resolve privileges. 4) Validate that the ACL permissions worked accordingly. Following is an example to test the code in the previous step. select UTL_HTTP.REQUEST('http://www.ahmedbaraka.com') from dual; If the sufficient ACL privileges or ACL assignments are not provided, you will receive the ORA24247 error.
Access Control Lists Maintenance

Use DELETE_PRIVILEGE to remove an access control entry from the XML file. exec DBMS_NETWORK_ACL_ADMIN.DELETE_PRIVILEGE( ACL=>'dba.xml', PRINCIPAL=> 'RAMI'); Use the DROP_ACL procedure to remove the XML file from the /sys/acls directory as follows: exec DBMS_NETWORK_ACL_ADMIN.DROP_ACL ( ACL=>'dba.xml' );
Page 338
Query Your Access Control List

To display list of the ACLs created in the database, use the following query: select HOST, LOWER_PORT, UPPER_PORT, ACL from DBA_NETWORK_ACLS You can query the DBA_NETWORK_ACL_PRIVILEGES view to query network privileges granted or denied for the access control list as follows: select PRINCIPAL, PRIVILEGE, IS_GRANT from DBA_NETWORK_ACL_PRIVILEGES where ACL like '%dba.xml' Logged on users can use the following query to see their access entries in the dba.xml file: select HOST, LOWER_PORT, UPPER_PORT, STATUS privilege from USER_NETWORK_ACL_PRIVILEGES where HOST in (select * from table(DBMS_NETWORK_ACL_UTILITY.DOMAINS('dbaexpert.com'))) and PRIVILEGE = 'connect' order by DBMS_NETWORK_ACL_UTILITY.DOMAIN_LEVEL(host) desc, LOWER_PORT;
Page 339
Part 5 Tuning
Oracle Database Performance
Page 340
Managing Performance Statistics

Managing OS Statistics
CPU Statistics: o doesn't exceed 95% in total. o check DB share on CPU Virtual Memory Statistics: o validate that memory usage does not increase after the system has reached a steady state after startup. Disk I/O Statistics: o current response time should be between 5 to 20 ms for a single block IO o the length of the disk queues shouldn't exceed two. Network Statistics: o Look at the network round-trip ping time and the number of collisions. Investigate it, if the network is causing large delays in response time.
Reducing Disk Contention: o o o o o o Increase the number of disks in the storage system Separate the database and the redo log files For a large table, use partitions to reduce I/O Stripe the data either manually or by using a RAID disk-striping system Invest in cutting-edge technology, such as file caching, to avoid I/O bottlenecks Consider using ASM
/* from the database */ -- to gather system stats see Gathering System Statistics -- all stats select STAT_NAME, VALUE, OSSTAT_ID, COMMENTS, CUMULATIVE from V$OSSTAT; /* CPU */ -- one-hour history of the Host CPU Utilization select BEGIN_TIME, END_TIME, GROUP_ID, METRIC_ID, METRIC_NAME, VALUE, METRIC_UNIT from V$SYSMETRIC_HISTORY where METRIC_NAME LIKE '%Host CPU%' -- top session CPU-comsumers SELECT n.username, s.sid, s.value FROM v$sesstat s,v$statname t, v$session n WHERE s.statistic# = t.statistic# AND n.sid = s.sid AND t.name='CPU used by this session'
Page 341
AND s.value <> 0 ORDER BY s.value desc; -- Decomposition of Total CPU Usage -- if the parsing or recursive CPU usage PCT is high, then tuning is required -- Recursive CPU Usage is for data dict lookups and executing PL/SQL programs SELECT name, value, round(value/(select sum(value) from v$sysstat WHERE NAME IN ('CPU used by this session', 'recursive cpu usage','parse time cpu'))*100,2) PCT FROM V$SYSSTAT WHERE NAME IN ('CPU used by this session', 'recursive cpu usage','parse time cpu') order by value DESC
/* io tuning */ -- IO related waits select EVENT, TOTAL_WAITS_FG, TOTAL_TIMEOUTS_FG, TIME_WAITED_FG, AVERAGE_WAIT_FG, WAIT_CLASS, TOTAL_WAITS, TOTAL_TIMEOUTS, TIME_WAITED, AVERAGE_WAIT from V$SYSTEM_EVENT WHERE WAIT_CLASS = 'User I/O' order by WAIT_CLASS; -- io stats select FILE_NO, FILETYPE_NAME, SMALL_READ_MEGABYTES "Single-block MegaBytes Reads", SMALL_WRITE_MEGABYTES "Single-block MegaBytes Writes", SMALL_READ_REQS "Single-block Read Requests", SMALL_WRITE_REQS "Single-block Write Requests", SMALL_READ_SERVICETIME "Total S-Block Read Time", SMALL_WRITE_SERVICETIME "Total S-Block Write Time", -- decode(SMALL_READ_REQS,0,0,SMALL_READ_SERVICETIME/SMALL_READ_REQS) "Per SBlock Read Response T", SMALL_SYNC_READ_REQS, SMALL_SYNC_READ_LATENCY "S-Block Sync Read Latency (ms)", LARGE_READ_MEGABYTES "Multi-block MegaBytes Reads", LARGE_WRITE_MEGABYTES "Multi-block MegaBytes Writes", LARGE_READ_REQS "Multi-block Read Requests", LARGE_WRITE_REQS "Multi-block Write Requests", LARGE_READ_SERVICETIME "Total M-Block Read Time", LARGE_WRITE_SERVICETIME "Total M-Block Write Time", ASYNCH_IO, RETRIES_ON_ERROR from V$IOSTAT_FILE order by FILE_NO /
Page 342
-- Datafiles IO since instance startup SELECT d.name, f.phyrds reads, f.phywrts wrts, (f.readtim / decode(f.phyrds,0,-1,f.phyrds))/10 ReadRespoinseTime_ms, (f.writetim / decode(f.phywrts,0,-1,phywrts))/10 WriteRespoinseTime_ms, SINGLEBLKRDTIM/10 Single_Block_ReadTime_ms FROM v$datafile d, v$filestat f WHERE d.file# = f.file# ORDER BY d.name; -- Datafiles IO History SELECT f.snap_id, f.filename, f.phyrds reads, f.phywrts wrts, (f.readtim / decode(f.phyrds,0,-1,f.phyrds))/10 ReadRespoinseTime_ms, (f.writetim / decode(f.phywrts,0,-1,phywrts))/10 WriteRespoinseTime_ms, SINGLEBLKRDTIM/10 Single_Block_ReadTime_ms, wait_count, time waittime FROM DBA_HIST_FILESTATXS f ORDER BY f.snap_id desc, filename
/* on Windows */ --Performance Monitor tool. For Windows Server 2003, can be downloaded from: http://www.microsoft.com/downloads/details.aspx?familyid=09115420-8c9d-46b9-a9a59bffcd237da2&displaylang=en -- see using Using Performance Tool in Windows -- see using OS Watcher -- see also Optimizing Windows Server
/* on Unix */ /* CPU */ # process per second sar -c 2 10 -- for all CPUs without interval, since last reboot sar -c P 0 -- for first CPU # CPU% utilization sar -u 4 5 # from vmstat output # us: user CPU Time % sy: system CPU Time % id: idle CPU Time % wa: Waiting for IO CPU Time % # alternatively vmstat -s | grep "cpu ticks" # CPU% utilization
Page 343
iostat c iostat 4 5 c -- 5 times every 4 seconds
/* Virtual Memory Statistics */ -- Obtain info about memory cat /proc/meminfo -- Page ins and page outs vmstat -s | grep "pages paged" sar -B 2 50 -- every 2 seconds for 50 times -- Swap ins and swap outs # from vmstat output: swpd (so (swapped out) or si) should ideally be 0 # alternatively: vmstat s -S M | grep "swap" sar r sar -W -- Active and inactive pages: you shouldn't have too few inactive memory pages: vmstat -S M # alternatively: vmstat -s -S M | grep "memory" sar r sar R
/* Disk I/O Statistics */ # disk usage df h # from vmstat output: bi (blocks in) bo (blocks out) # alternatively: #vmstat -s -S M | grep "pages" # reports disk statistics vmstat -d | grep d # tps: transfer per second iostat d iostat 4 5 d -- 5 times every 4 seconds iostat -d k -- display output in kilobytes instead of blocks iostat -d x -- extended report # rtps wtps read/write requests per second issued to the physical disk # bread/s bwrtn/s data read/write from the drive in blocks per second sar b 5 10 # bloc device usage # number of sectors (512 byte) read/written per second # if avque is greater larger than 1, disk contention is there sar -d
/* Network Statistics */
Page 344
# ping and check the latency ping ... # sar -n DEV | EDEV | SOCK | FULL # DEV= network devices rxpck/s packets received per second txpck/s packets transmitted per second rxbyt/s bytes received per second txbyt/s bytes transmitted per second # EDEV = failures from the network devices rxerr/s bad packets received per second. txerr/s errors that happened per second while transmitting packets coll/s collisions that happened per second while transmitting packets rxdrop/s received packets dropped per second because of a lack of space in linux buffers txdrop/s transmitted packets dropped per second because of a lack of space in linux buffers txcarr/s carrier-errors that happened per second while transmitting packets rxfram/s frame alignment errors that happened per second on received packets. rxfifo/s FIFO overrun errors that happened per second on received packets. txfifo/s FIFO overrun errors that happened per second on transmitted packets. # SOCK = sockets totsck Total number of used sockets. tcpsck Number of TCP sockets currently in use. udpsck Number of UDP sockets currently in use. rawsck Number of RAW sockets currently in use. ip-frag Number of IP fragments currently in use. # display network interfaces netstat i # summary stats on each protocol netstat s | less
Managing Database Statistics

System and Session Statistics (and their executed statements) Time Model Statistics Wait Events Active Session History
System and Session Statistics /* System stats */ select NAME, decode(CLASS, '1','User','2','Redo','4','Enqueue','8','Cache','16','OS','32','RAC','64','SQL','1 28','Debug', CLASS) STAT_CLASS, VALUE from V$SYSSTAT order by NAME;
/* Session stats */ select Oracle DBA Code Examples
Page 345
T.SID,S.USERNAME, S.MACHINE, S.MODULE, S.ACTION, N.NAME, decode(N.CLASS, '1','User','2','Redo','4','Enqueue','8','Cache','16','OS','32','RAC','64','SQL','1 28','Debug', N.CLASS) STAT_CLASS, VALUE from V$SESSTAT T, V$STATNAME N, V$SESSION S WHERE T.STATISTIC#=N.STATISTIC# and T.SID=S.SID and S.USERNAME NOT IN ('SYSTEM','SYS','DBSNMP','SYSMAN') order by S.SID , N.NAME;
/* SQL Stats */ high buffer gets = using the wrong index, the wrong driving table in a join, or a similar SQL-related error buffer gets and disk reads are at identical levels = a missing index
-- top io consumers SELECT executions, buffer_gets, disk_reads, rows_processed,SORTS, FROM V$SQL WHERE buffer_gets > 100000 OR disk_reads > 100000 ORDER BY buffer_gets + 100*disk_reads DESC;
sql_text
-- top CPU consumers SELECT executions, ROUND(elapsed_time/1000000, 2) elapsed_seconds, ROUND(cpu_time/1000000, 2) cpu_secs , sql_text from (select * from v$sql order by elapsed_time desc) WHERE rownum <6 Time Model Statistics DB Time is an indicator instance workload. DB Time = CPU time + non-idle Wait time (of all the sessions accumulatively) -- how long since the instance started select STARTUP_TIME , ROUND((SYSDATE-STARTUP_TIME)*24,2) HOURS from v$instance; -- system-wide time-based stat SELECT STAT_NAME "Stat. Name", round(VALUE/1000000) "Value (s)" , round(VALUE/1000000/60) "Value (min)" FROM V$SYS_TIME_MODEL;
-- SESSION-wide time-based stats SELECT E.SID, S.USERNAME, S.MACHINE, S.MODULE, S.ACTION, STAT_NAME "Stat. Name", round(VALUE/1000000) "Value (s)" , round(VALUE/1000000/60) "Value (min)" FROM V$SESS_TIME_MODEL E, V$SESSION S WHERE E.SID = S.SID and S.USERNAME NOT IN ('SYSTEM','SYS','DBSNMP','SYSMAN')
Page 346
order by E.SID; Wait Events The wait events are only the symptoms of problems, most likely within the application code. After defining the troubled waiting event, you can get further info by tracing the suspected session. -- TIMED_STATISTICS must be true (default) show parameter TIMED_STATISTICS /* Wait time for the whole instance */ -- System metrics captured in recent 60-sec or 15-sec -- If 'Database Wait Time Ratio' is higher than 'Database CPU Time Ratio', -consider looking for bottlenecks select GROUP_ID, METRIC_NAME, VALUE, METRIC_UNIT, INTSIZE_CSEC/100 Interval_Duration, TO_CHAR(BEGIN_TIME,'HH24:MI:SS') BEGIN_TIME, TO_CHAR(END_TIME,'HH24:MI:SS') END_TIME from V$SYSMETRIC -- where METRIC_NAME IN ('Database Wait Time Ratio','Database CPU Time Ratio') order BY END_TIME DESC -- in the recent hour SELECT GROUP_ID, METRIC_NAME, VALUE, METRIC_UNIT, ROUND(INTSIZE_CSEC/100) Interval_Duration, TO_CHAR(BEGIN_TIME,'HH24:MI:SS') BEGIN_TIME, TO_CHAR(END_TIME,'HH24:MI:SS') END_TIME FROM V$SYSMETRIC_HISTORY WHERE METRIC_NAME IN ('Database Wait Time Ratio','Database CPU Time Ratio') ORDER BY END_TIME DESC
/* WAIT EVENTS */ -- waite events in the instance: -- Top Wait Classes By Instance Total -- AWR reports could also assist you select WAIT_CLASS, TIME_WAITED, round(TIME_WAITED/TOT_WAIT*100,2) TIME_WAITED_PCT, TIME_WAITED_FG, round(TIME_WAITED_FG/TOT_WAIT*100,2) TIME_WAITED_FG_PCT from V$SYSTEM_WAIT_CLASS, (select sum(TIME_WAITED) TOT_WAIT from V$SYSTEM_WAIT_CLASS where WAIT_CLASS <>'Idle'), (select sum(TIME_WAITED_FG) TOT_WAIT_FG from V$SYSTEM_WAIT_CLASS where WAIT_CLASS <>'Idle') where WAIT_CLASS <>'Idle' order by TIME_WAITED_FG_PCT DESC
Page 347
-- Wait Classes by Instance Wide in the Last Hour select TO_CHAR(BEGIN_TIME,'HH24:MI') BEGIN_TIME, TO_CHAR(END_TIME,'HH24:MI') END_TIME, INTSIZE_CSEC/100 Interval_sec, WAIT_CLASS#, (SELECT DISTINCT WAIT_CLASS FROM V$EVENT_NAME X WHERE X.WAIT_CLASS#=W.WAIT_CLASS#) WAIT_CLASS_NAME, TIME_WAITED, DBTIME_IN_WAIT "Pct of DB Time spent", WAIT_COUNT from V$WAITCLASSMETRIC_HISTORY W where WAIT_CLASS#<>6 -- Idle order by BEGIN_TIME DESC, TIME_WAITED DESC;
-- Wait Events by Instance Total -- typically waits by foreground processes are what we care about select EVENT, TIME_WAITED_FG, ROUND(TIME_WAITED_FG/TOT_WAIT_FG*100,2) TIME_WAITED_PCT, TOTAL_WAITS_FG, TOTAL_TIMEOUTS_FG, AVERAGE_WAIT_FG, WAIT_CLASS, TOTAL_WAITS, TOTAL_TIMEOUTS, TIME_WAITED, AVERAGE_WAIT from V$SYSTEM_EVENT, (SELECT SUM(TIME_WAITED_FG) TOT_WAIT_FG FROM V$SYSTEM_EVENT where WAIT_CLASS <> 'Idle') where WAIT_CLASS <> 'Idle' order by TIME_WAITED_FG DESC; -- if buffer busy was on the top waits, get more info about -- block contention statistics select CLASS, TIME, COUNT from V$WAITSTAT order by TIME desc
-- Wait Events by Sessions -- detailed information on V$SESSION ( or V$SESSION_WAIT ) -- cols in V$SESSION_WAIT already contained in V$SESSION select E.SID, S.USERNAME, S.MACHINE, S.MODULE, S.ACTION, S.STATUS, E.WAIT_CLASS, E.EVENT, S.STATE, WAIT_TIME, TIME_WAITED, SECONDS_IN_WAIT, AVERAGE_WAIT, P1TEXT, P1, P2TEXT, P2,
Page 348
P3TEXT, P3, TOTAL_WAITS, MAX_WAIT, TOTAL_TIMEOUTS, ROW_WAIT_OBJ# WAITED_OBJECT, TO_CHAR(S.LOGON_TIME,'HH24:MI') LOGON_TIME, S.BLOCKING_SESSION from V$SESSION_EVENT E, V$SESSION S where E.SID = S.SID and S.USERNAME NOT IN ('SYSTEM','SYS','DBSNMP','SYSMAN') and E.WAIT_CLASS <> 'Idle' order by E.TIME_WAITED desc /
-- if the problem in the latch, further details can be obtained: SELECT ADDR Latch_Object_Address, LATCH#, LEVEL#, NAME, HASH, GETS, MISSES, case misses when 0 then 0 else misses/(misses+GETS+SLEEPS) end MISSES_RATIO, SLEEPS, IMMEDIATE_GETS, IMMEDIATE_MISSES, SPIN_GETS, WAIT_TIME WAIT_TIME_US FROM V$LATCH order by MISSES desc Active Session History (ASH) Used when you want to analyze for a previous period not taken yet by AWR (let's say last 10 minutes). V$ACTIVE_SESSION_HISTORY is flushed into DBA_HIST_ACTIVE_SESS_HISTORY when AWR is taken (every hour by default). -- ASH list SELECT SAMPLE_ID, SAMPLE_TIME, SESSION_ID, SESSION_SERIAL#, SESSION_TYPE, FLAGS, USER_ID, SQL_ID, SQL_CHILD_NUMBER, SQL_OPCODE, FORCE_MATCHING_SIGNATURE, TOP_LEVEL_SQL_ID, TOP_LEVEL_SQL_OPCODE, SQL_PLAN_HASH_VALUE, SQL_PLAN_LINE_ID, SQL_PLAN_OPERATION, SQL_PLAN_OPTIONS, SQL_EXEC_ID, SQL_EXEC_START, PLSQL_ENTRY_OBJECT_ID, PLSQL_ENTRY_SUBPROGRAM_ID, PLSQL_OBJECT_ID, PLSQL_SUBPROGRAM_ID, QC_INSTANCE_ID, QC_SESSION_ID, QC_SESSION_SERIAL#, EVENT, EVENT_ID, EVENT#, SEQ#, P1TEXT, P1, P2TEXT, P2, P3TEXT, P3, WAIT_CLASS, WAIT_CLASS_ID, WAIT_TIME, SESSION_STATE, TIME_WAITED, BLOCKING_SESSION_STATUS, BLOCKING_SESSION, BLOCKING_SESSION_SERIAL#, CURRENT_OBJ#, CURRENT_FILE#, CURRENT_BLOCK#, CURRENT_ROW#, CONSUMER_GROUP_ID, XID, REMOTE_INSTANCE#, IN_CONNECTION_MGMT, IN_PARSE, IN_HARD_PARSE, IN_SQL_EXECUTION, IN_PLSQL_EXECUTION, IN_PLSQL_RPC, IN_PLSQL_COMPILATION, IN_JAVA_EXECUTION, IN_BIND, IN_CURSOR_CLOSE, SERVICE_HASH, PROGRAM, MODULE, ACTION, CLIENT_ID FROM V$ACTIVE_SESSION_HISTORY Where SESSION_TYPE <> 'BACKGROUND' and USER_ID not in ( select u.user_id from dba_users u where username IN
Page 349
('SYS','SYSTEM','DBSNMP','SYSMAN') ) order by Sample_Time desc
-- ASH report -- info about the SQL that ran during the time you specify $ORACLE_HOME/rdbms/admin/ashrpt.sql The ashrpti.sql lets you specify the db instance
-- Top Waited-for Objects SELECT (SELECT o.object_name FROM dba_objects o WHERE o.object_id = current_obj#) (SELECT o.object_type FROM dba_objects o WHERE o.object_id = current_obj#) a.event, SUM (a.wait_time + a.time_waited) FROM v$active_session_history a WHERE ( a.wait_time + a.time_waited ) > 0 AND a.current_obj# IS NOT NULL AND a.sample_time BETWEEN SYSDATE GROUP BY a.event, current_obj# ORDER BY total_wait_time DESC;
object_name,
object_type, total_wait_time
15 / 1440 AND SYSDATE
-- Top Waits SELECT a.event, SUM(a.wait_time + a.time_waited) total_wait_time FROM v$active_session_history a WHERE a.sample_time between sysdate - 30/2880 and sysdate GROUP BY a.event ORDER BY total_wait_time DESC;
-- Top Waiting Users SELECT s.sid, s.username, SUM(a.wait_time + a.time_waited) total_wait_time FROM v$active_session_history a, v$session s WHERE a.sample_time between sysdate - 30/2880 and sysdate AND a.session_id=s.sid GROUP BY s.sid, s.username ORDER BY total_wait_time DESC;
-- Top Waiting SQL Statements SELECT a.user_id,d.username,s.sql_text, SUM(a.wait_time + a.time_waited) total_wait_time FROM v$active_session_history a, v$sqlarea s, dba_users d WHERE a.sample_time between sysdate - 30/2880 and sysdate
Page 350
AND a.sql_id = s.sql_id AND a.user_id = d.user_id GROUP BY a.user_id,s.sql_text, d.username; -- Top Resource Consuming SQL SELECT hash_value, executions, ROUND (elapsed_time/1000000, 2) total_time, ROUND (cpu_time/1000000, 2) cpu_seconds FROM (SELECT * FROM V$SQL ORDER BY elapsed_time desc); -- ASH history SELECT * FROM DBA_HIST_ACTIVE_SESS_HISTORY Where SESSION_TYPE <> 2 -- 'BACKGROUND' (SESSION_TYPE in V$ is varchar2) order by SNAP_ID desc Segment Statistics You drill down into segment stats from the instance, sessions or ASM wait stats. select OWNER, OBJECT_TYPE, OBJECT_NAME, TABLESPACE_NAME, STATISTIC_NAME, VALUE from V$SEGMENT_STATISTICS -- WHERE statistic_name='buffer busy waits' order by value desc;
Handling Important Oracle Wait Events

Buffer Busy Waits Check type of block is causing the wait. Use locally managed tablespaces with ASSM. Consider using global hash-partitioned indexes. Tune SQL statements as necessary to fix these waits. Check that the sar -d utility might indicate high request queues and service times. Check your core dump directory is not too big. db file scattered read Db File Sequential Read Reduce the demand for physical I/Os: Raising the buffer cache component, add missing indexes on key tables, Optimize SQL statements. Increase the capacity of the system to handle more I/Os. Indicate index usage. Increase PGA_AGGREGATE_TARGET. If the objects arent too large, you can use the DEFAULT and KEEP buffer pools to retain them in memory. When you have file# and Block# from wait details, you can get the object name: select segment_name from dba_extents where file_id = <file#> and <block#> between block_id and block_id + blocks - 1 and rownum = 1 Direct Path Read and Direct Path Write It occurs while performing a direct read or write into the PGA, bypassing the SGA buffer cache. Enable PGA auto tuning. Increase the number of disks.
Page 351
Free Buffer Waits
The number of dirty buffers in cache too high for the db writer. Increase Db Buffer Cache Increase DB_WRITER_PROCESSES (one per CPU) Check the IO
Enqueue Waits
Sessions are waiting for locks held by other sessions. Query V$ENQUEUE_STAT select * from V$ENQUEUE_STAT where CUM_WAIT_TIME<>0 OR FAILED_REQ#<>0 May caused by infrequent commits and dictionary managed tbs.
Latch Free
SELECT a.name "Latch Name", a.gets "Gets (Wait)", a.misses "Misses (Wait)", (1 - (misses / gets)) * 100 "Latch Hit Ratio %" FROM V$LATCH a WHERE a.gets != 0 UNION SELECT a.name "Latch Name", a.gets "Gets (Wait)", a.misses "Misses (Wait)", 100 "Latch Hit Ratio" FROM V$LATCH a WHERE a.gets = 0 ORDER BY 4; shared pool latch (and the library cache latches) reasons: An undersized shared pool Failure to use bind variables Using dissimilar SQL statements and failing to reuse statements Users frequently logging off and logging back into the application Failure to keep cursors open after each execution Using a shared pool size thats too large cache buffers LRU chain and cache buffer chain may be caused by table full scan or unselective indexes.
Log Buffer Space
Check I/O to the redo log disk Consider increasing LOG_BUFFER (so that Log File Sync wait event doesn't increase as well) For a materialized view with COMPLETE refresh, consider setting ATOMIC_REFRESH=false in DBMS_MVIEW.REFRESH to avoid logging
Log File Switch
Check the archive destination isn't full. Increase sizes of redo log files. Increase the number of archiver (ARCn) processes. Small values lead to "redo log space requests" in V$SYSSTAT
Log File Sync library cache pin
Too-frequent commits Redo Log I/O bottleneck It occurs when you are compiling or parsing a PL/SQL object or a view and Oracle wants to pin an object in memory in the library cache and
Page 352
ensure no other processes can update the object at the same time. This is my script for tracking down who is blocking who in the event of a library cache pin event:
select decode(lob.kglobtyp, 0, 'NEXT OBJECT', 1, 'INDEX', 2, 'TABLE', 3, 'CLUSTER', 4, 'VIEW', 5, 'SYNONYM', 6, 'SEQUENCE', 7, 'PROCEDURE', 8, 'FUNCTION', 9, 'PACKAGE', 11, 'PACKAGE BODY', 12, 'TRIGGER', 13, 'TYPE', 14, 'TYPE BODY', 19, 'TABLE PARTITION', 20, 'INDEX PARTITION', 21, 'LOB', 22, 'LIBRARY', 23, 'DIRECTORY', 24, 'QUEUE', 28, 'JAVA SOURCE', 29, 'JAVA CLASS', 30, 'JAVA RESOURCE', 32, 'INDEXTYPE', 33, 'OPERATOR', 34, 'TABLE SUBPARTITION', 35, 'INDEX SUBPARTITION', 40, 'LOB PARTITION', 41, 'LOB SUBPARTITION', 42, 'MATERIALIZED VIEW', 43, 'DIMENSION', 44, 'CONTEXT', 46, 'RULE SET', 47, 'RESOURCE PLAN', 48, 'CONSUMER GROUP', 51, 'SUBSCRIPTION', 52, 'LOCATION', 55, 'XML SCHEMA', 56, 'JAVA DATA', 57, 'SECURITY PROFILE', 59, 'RULE', 62, 'EVALUATION CONTEXT', 'UNDEFINED') object_type, lob.KGLNAOBJ object_name, pn.KGLPNMOD lock_mode_held, pn.KGLPNREQ lock_mode_requested, ses.sid, ses.serial#, ses.username FROM x$kglpn pn, v$session ses, x$kglob lob, v$session_wait vsw WHERE pn.KGLPNUSE = ses.saddr and pn.KGLPNHDL = lob.KGLHDADR and lob.kglhdadr = vsw.p1raw and vsw.event = 'library cache pin' order by lock_mode_held desc
direct path read temp
This happens in: The sorts are too large to fit in memory and some of the sort data is written out directly to disk. Parallel slaves are used for scanning data. This is the biggest wait for large data warehouse sites. Consider setting sort_area_size or pga_aggreagate_target
enq: JI - contention
The session is waiting for a materialized view to finish its undergoing refresh. For example, you are trying to refresh two materialized views reading from same master table(s) in the same time. Another example, you are trying to compile a materialized view when another session is refreshing it.
List of Idle Waits Events

On Oracle 10g, the idle waits events are: AQ Proxy Cleanup Wait ASM background timer DIAG idle wait EMON idle wait
Page 353
KSV master wait LNS ASYNC archive log LNS ASYNC dest activation LNS ASYNC end of log LogMiner: client waiting for transaction LogMiner: slave waiting for activate message LogMiner: wakeup event for builder LogMiner: wakeup event for preparer LogMiner: wakeup event for reader Null event PX Deq Credit: need buffer PX Deq Credit: send blkd PX Deq: Execute Reply PX Deq: Execution Msg PX Deq: Par Recov Execute PX Deq: Signal ACK PX Deq: Table Q Normal PX Deq: Table Q Sample PX Deque wait PX Idle Wait Queue Monitor Shutdown Wait Queue Monitor Slave Wait Queue Monitor Wait SQL*Net message from client SQL*Net message to client SQL*Net more data from client STREAMS apply coord waiting for slave message STREAMS apply slave idle wait STREAMS apply slave waiting for coord message STREAMS capture process filter callback wait for ruleset STREAMS fetch slave waiting for txns STREAMS waiting for subscribers to catch up Streams AQ: RAC qmn coordinator idle wait Streams AQ: deallocate messages from Streams Pool Streams AQ: delete acknowledged messages Streams AQ: qmn coordinator idle wait Streams AQ: qmn slave idle wait Streams AQ: waiting for messages in the queue Streams AQ: waiting for time management or cleanup tasks Streams fetch slave: waiting for txns class slave wait client message dispatcher timer gcs for action gcs remote message ges remote message i/o slave wait jobq slave wait knlqdeq lock manager wait for remote message master wait null event parallel query dequeue pipe get pmon timer queue messages rdbms ipc message
Page 354
rdbms ipc reply reliable message slave wait smon timer virtual circuit status wait for activate message wait for unread message on broadcast channel wakeup event for builder wakeup event for preparer wakeup event for reader wakeup time manager
Using Performance Monitor Tool in Windows

Instant monitor: click on the '+' icon to start up new counters. Click on 'Process' under 'Performance Object'. Select counters, 'Private Bytes' (what Task Manager shows), Thread Count, 'Handle Count', and 'Virtual Bytes' in the 'Select counters from list' window. Also select the process, 'Oracle' in the 'Select instance from list' window. Click on the 'Add' button. Scheduled monitor: click Counter Logs node, right click on the right hand side pane, select New Log Settings, type the name, select the objects to monitor, set the schedule and the generated log file.
Using OS Watcher for Windows

OS Watcher for Unix and Linux, see Using OS Watcher for Linux. OS Watcher for Windows (OSWFW) is several batch files that run the Windows utility logman and schtasks. If OSWFW is installed on RAC, when it is first run, it will read the Cluster Registry and copy itself to the other nodes in a directory that is the same name as the directory it was first unzipped in. Its reference is Note 433472.1 /* Removing OSWFW */ oswatcher remove
/* Initially configure OSWFW */ OSWatcher {ARG1} {ARG2} {ARG3} ARG1 = Snapshot interval in seconds ARG2 = Runtime Interval - hours OSWatcher will run ARG3 = RAC - detect Real Application Cluster OSWatcher 30 48
/* OSWatcher stop */ OSWatcher stop {node name}
/* Getting the Status of OSWFW */ OSWatcher status
Page 355
/* Querying details of a specific counter or task */ OSWatcher query {node name} {counter / OSWCleanup / OSWPrivNet} OSWatcher query all oswatcher query OSWThread
/* Extracting the Names of the Counters in a Capture File */ relog {trace_file_name} -q # to sort the output: relog {trace_file_name} -q | sort /+1
Page 356
Optimizing Performance in Windows Server

Optimizing Performance in Windows Server 2003
/* disable un-required services */ Usually, the following services are not required in a database server and can be disabled: Alerter ClipBook Computer Browser DHCP Client DHCP Server Fax Service File Replication INfrared Monitor Internet Connection Sharing Messenger NetMeeting Remote Desktop Sharing Network DDE Network DDE DSDM NWLink NetBIOS NWLink IPX/SPX Print Spooler TCP/IP NetBIOS Helper Service: unless member of a domain Telephony Telnet Uninterruptible Power Supply
/* Optimize Performance for Background Services */ Oracle database runs as a background service. To increase performance for background services, follow these steps: 1. Click Start, click Control Panel, and then click System. 2. Click the Advanced tab, and then click Settings under Performance. 3. Click the Advanced tab, click Background services, and then click OK
/* Optimize Data Throughput for Network Applications */ 1. In Windows Explorer, right-click My Network Places, and then click Properties. 2. Right-click the Local Area Connection that you want to optimize, then click Properties. 3. In the This connection uses the following items list, click (but do not clear its check box) File and Printer Sharing for Microsoft Networks, and then click Properties. 4. Click Maximum data throughput for network applications, click OK, then Close.
Page 357
Tuning the Database Instance

Tuning the Shared Pool
/* Data Dictionary Hit Ratio */ -- around 90 (depending on the application, low value in some cases acceptable) SELECT (sum(gets - getmisses - fixed)) / SUM(gets) "data dictionary hit ratio" from v$rowcache;
/* Pinning Objects in the Shared Pool (LC)*/ SELECT type, COUNT(*) OBJECTS, SUM(DECODE(KEPT,'YES',1,0)) KEPT, SUM(loads) - count(*) reloads FROM V$DB_OBJECT_CACHE GROUP BY type ORDER BY objects DESC; select * FROM V$DB_OBJECT_CACHE order by loads desc; -- pin objects with high loads: -- un-retained after db restart -- you can make script to load then when db starts EXECUTE SYS.DBMS_SHARED_POOL.KEEP(NEW_EMP.PKG, PACKAGE); EXECUTE SYS.DBMS_SHARED_POOL.UNKEEP(NEW_EMP.PKG,PACKAGE);
/* Measuring Library Cache Efficiency */ -- lib cache hit ratio SELECT SUM(pinhits)/sum(pins) Library_cache_hit_ratio FROM V$LIBRARYCACHE;
-- statement reloads SELECT namespace, pins, pinhits, reloads FROM V$LIBRARYCACHE ORDER BY namespace;
-- lib cache memory usage select LC_NAMESPACE, LC_INUSE_MEMORY_OBJECTS, LC_INUSE_MEMORY_SIZE, LC_FREEABLE_MEMORY_OBJECTS, LC_FREEABLE_MEMORY_SIZE from V$LIBRARY_CACHE_MEMORY order by LC_INUSE_MEMORY_OBJECTS desc, LC_INUSE_MEMORY_SIZE desc, LC_FREEABLE_MEMORY_OBJECTS desc, LC_FREEABLE_MEMORY_SIZE desc;
-- to estimate the optimal shared pool size -- you can just set the MEMORY_TARGET select SHARED_POOL_SIZE_FOR_ESTIMATE, SHARED_POOL_SIZE_FACTOR, ESTD_LC_SIZE, ESTD_LC_MEMORY_OBJECTS, ESTD_LC_TIME_SAVED, ESTD_LC_TIME_SAVED_FACTOR, ESTD_LC_LOAD_TIME, ESTD_LC_LOAD_TIME_FACTOR, ESTD_LC_MEMORY_OBJECT_HITS from V$SHARED_POOL_ADVICE
Page 358
order by SHARED_POOL_SIZE_FOR_ESTIMATE desc;
-- if literal values rather than bind values are used -- by the applications (high hard parse): CURSOR_SHARING=FORCE (recommended) or SIMILAR (not EXACT)
/* Setting CURSOR_SPACE_FOR_TIME */ If CURSOR_SPACE_FOR_TIME=TRUE, you ensure that the cursors for the application cannot be deallocated while the application cursors are still open. It will then eliminate the Oracles overhead to check whether the cursor is flushed from the library cache. It will result in increase in the shared pool memory.
/* Setting SESSION_CACHED_CURSORS */ It ensures that for any cursor for which more than three parse requests are made, the parse requests are automatically cached in the session cursor cache. It avoids high soft parse. Good to use in Forms-based apps. ALTER SESSION SET SESSION_CACHED_CURSORS = value; If the value of session cursor cache hits is low compared to the total parse count for a session, then the SESSION_CACHED_CURSORS parameter value should be increased (also good to make it larger than OPEN_CURSORS): select NAME, VALUE from v$sysstat where name = 'session cursor cache hits' select SID, NAME, VALUE from v$sesstat, V$STATNAME where V$SESSTAT.STATISTIC# = V$STATNAME.STATISTIC# and NAME LIKE 'parse count (total)' order by VALUE desc alter system set SESSION_CACHED_CURSORS=300 scope=spfile ;
Tuning the Buffer Cache

/* Sizing the Buffer Cache */ -- use MEMORY_TARGET -- size assigned to it select current_size/1024/1024 from V$SGA_DYNAMIC_COMPONENTS where component = 'DEFAULT buffer cache'
-- different areas in the buffer cache select count(*) blocks, State from ( select decode (state, 0, 'Free', 1, decode (lrba_seq,0, 'Available','Being Used'), 3, 'Being Used',
Page 359
state) State from x$bh ) group by rollup(state) order by count(*) desc;
-- to get the buffer cache hit ratio SELECT NAME, PHYSICAL_READS, DB_BLOCK_GETS, CONSISTENT_GETS, round( 1 - (PHYSICAL_READS/(DB_BLOCK_GETS + CONSISTENT_GETS)),4)*100 "HitRatio" FROM V$BUFFER_POOL_STATISTICS;
/* Using Multiple Pools for the Buffer Cache */ -- x$bh: contains a record (the buffer header) for each block in the buffer select count(*), State from ( select decode (state, 0, 'Free', 1, decode (lrba_seq,0, 'Available','Being Used'), 3, 'Being Used', state) State from x$bh ) group by state
/* Sizing the Buffer Cache */ -- use MEMORY_TARGET -- to get the buffer cache hit ratio SELECT NAME, PHYSICAL_READS, DB_BLOCK_GETS, CONSISTENT_GETS, round( 1 - (PHYSICAL_READS/(DB_BLOCK_GETS + CONSISTENT_GETS)),4)*100 "HitRatio" FROM V$BUFFER_POOL_STATISTICS;
/* Using Multiple Pools for the Buffer Cache */ -- To determine objects that are candidates for the recycle buffer pool -- x$bh: contains a record (the buffer header) for each block in the buffer select obj object, o.object_name, count(*) buffers, round((count(*)/totsize) * 100,2) percent_cache FROM x$bh, (select count(*) totsize FROM x$bh ), DBA_OBJECTS o WHERE ( tch=1 -- touch count OR (tch = 0 and lru_flag <10)) and obj=o.object_id(+) GROUP BY obj, o.object_name, totsize having round((count(*)/totsize) * 100,2) > 1 order by percent_cache desc; -- objects candidate for keep pool -- objects of at least 25 buffers and have an average touch count of more than 5 SELECT obj object, o.object_name, count(*) buffers, AVG(tch) average_touch_count FROM x$bh, dba_objects o WHERE obj=o.object_id(+) and lru_flag = 8
Page 360
GROUP BY obj, o.object_name HAVING avg(tch) > 5 AND count(*) > 25 order by avg(tch) desc
-- define the areas DB_KEEP_CACHE_SIZE=16MB DB_RECYCLE_CACHE_SIZE=16MB
-- change the table's pool ALTER TABLE test1 STORAGE (buffer_pool keep); ALTER TABLE test2 STORAGE (buffer_pool recycle);
Tuning PGA
-- PGA stats SELECT NAME, VALUE, UNIT FROM V$PGASTAT -- PGA used by each session SELECT s.sid,a.username, round(s.value/1024,2) KB FROM V$SESSTAT S, V$STATNAME N, V$SESSION A WHERE n.STATISTIC# = s.STATISTIC# and name = 'session pga memory' AND s.sid=a.sid ORDER BY s.value desc; -- PGA used by sessions select PID, SERIAL#, CATEGORY, round(ALLOCATED/1024,2) ALLOCATED_KB1, USED, round(MAX_ALLOCATED/1024,2) MAX_ALLOCATED_KB from V$PROCESS_MEMORY -- PGA used by processes SELECT program, pga_used_mem, pga_alloc_mem, pga_freeable_mem, pga_max_mem from V$PROCESS;
Dumping the PGA

SQL>conn / as sysdba SQL> alter system set max_dump_file_size=unlimited; SQL> connect sys as sysdba SQL> oradebug setospid <OSpid from query above for unix systems> If on a MSwindows platform you will need to attach using the setorapid instead of setospid SQL> oradebug setorapid <Orapid from query above>
Page 361
SQL> oradebug SQL> oradebug SQL> oradebug SQL> oradebug SQL> oradebug SQL>exit
unlimit dump errorstack 10 dump heapdump 536870917 tracefile_name (shows the path and filename information) close_trace (closes the trace file)
Using Server Result Cache

For SQL Result Cache, if an underlying table is updated, the result cache will be invalidate and the statement will be re-executed.
Caution When a session reads from a PL/SQL function result cache, the function body is not executed. This means, if the function includes any IO or auditing code, this code will not actually be executed.
SQL Result Cache Restrictions: Queries against data dictionary objects and temporary tables are not supported. Queries that use the following SQL functions: CURRENT_DATE, CURRENT_TIMESTAMP, LOCAL_TIMESTAMP, USERENV/SYS_CONTEXT, SYS_GUID, SYSDATE and SYS_TIMESTAMP are not supported. Queries with bind variables can reuse a cached result only for identical variable values. Results of the queries retrieving non current version of data are not cached in the result cache. Results of the flashback queries are not cached.
Restrictions on PL/SQL Function Result Cache include: The function cannot be defined in a module using invokers rights. The function cannot be used in an anonymous block. The function cannot have any OUT or IN OUT parameters. The function cannot have IN parameters that are BLOB, CLOB, NCLOB, REF CURSOR, collections, objects, or records. The function cannot return a BLOB, CLOB, NCLOB, REF CURSOR, OBJECTS, or records. It can return a collection as long as the collection does not contain one of these types. /* Configuring Result Cache */ RESULT_CACHE_MAX_SIZE maximum amount of SGA memory (in bytes) that can be used by the Result Cache (taken from shared pool). If the value of this parameter is 0, then the feature is disabled. show parameter RESULT_CACHE_MAX_SIZE ALTER SYSTEM SET RESULT_CACHE_MAX_SIZE =8M; The parameter RESULT_CACHE_MAX_RESULT specifies the percentage of RESULT_CACHE_MAX_SIZE that any single result can use. Its default value is five. ALTER SYSTEM SET RESULT_CACHE_MAX_RESULT =25;
Page 362
The parameter RESULT_CACHE_REMOTE_EXPIRATION specifies the number of minutes that a result using a remote object is allowed to remain valid. Setting this parameter to 0 (the default) implies that results using remote objects should not be cached.
/* Controlling Result Cache Behavior */ The RESULT_CACHE_MODE initialization parameter: MANUAL The ResultCache operator is added, only if you use the RESULT_CACHE hint in the SQL query. FORCE The ResultCache operator is added to the root of all SELECT statements, if that is possible. However, if the statement contains a NO_RESULT_CACHE hint, then the hint takes precedence over the parameter setting. ALTER SYSTEM SET RESULT_CACHE_MODE =FORCE; ALTER SESSION SET RESULT_CACHE_MODE =FORCE;
SELECT /*+ result_cache */ AVG(SALARY), E.DEPARTMENT_ID FROM HR.EMPLOYEES E, HR.DEPARTMENTS D WHERE E.DEPARTMENT_ID = D.DEPARTMENT_ID GROUP BY E.DEPARTMENT_ID;
/* PL/SQL Function Result Cache */ CREATE OR REPLACE FUNCTION get_name (id NUMBER) RETURN VARCHAR2 RESULT_CACHE RELIES_ON(emp) IS ... -- Package specification CREATE OR REPLACE PACKAGE department_pks IS -- Function declaration FUNCTION get_dept_info (dept_id NUMBER) RETURN dept_info_record RESULT_CACHE; END department_pks; CREATE OR REPLACE PACKAGE BODY department_pks AS -- Function definition FUNCTION get_dept_info (dept_id NUMBER) RETURN dept_info_record RESULT_CACHE RELIES_ON (EMPLOYEES) IS BEGIN ...
/* Bypass Result Cache */ begin DBMS_RESULT_CACHE.BYPASS(TRUE); DBMS_RESULT_CACHE.FLUSH; end; exec DBMS_RESULT_CACHE.BYPASS(FALSE);
/* Monitoring Result Cache */ -- Memory taken for Result Cache SELECT NAME, ROUND(BYTES/1024,2) KB FROM V$SGAstat WHERE upper(name) like '%RESULT%';
Page 363
-- objects in the result cache select ID, TYPE, STATUS, BUCKET_NO, HASH, NAME, NAMESPACE, CREATOR_UID, DEPEND_COUNT, BLOCK_COUNT, SCN, COLUMN_COUNT, PIN_COUNT, SCAN_COUNT, ROW_COUNT, ROW_SIZE_MAX, ROW_SIZE_MIN, ROW_SIZE_AVG, BUILD_TIME, LRU_NUMBER, OBJECT_NO, INVALIDATIONS, SPACE_OVERHEAD, SPACE_UNUSED, CACHE_ID, CACHE_KEY, to_char(CREATION_TIMESTAMP,'HH12:MI AM') CREATE_TIME from V$RESULT_CACHE_OBJECTS order by type desc; -- result cache stats select ID, NAME, VALUE from V$RESULT_CACHE_STATISTICS; -- using DBMS_RESULT_CACHE -- check the status of the Result Cache -- Note: this is the reliable method to know whether result cache is enabled or not SQL>select DBMS_RESULT_CACHE.STATUS from dual; -- display report on result cache memory SQL>set serveroutput on SQL>exec DBMS_RESULT_CACHE.MEMORY_REPORT -- turn bypass mode on and off SQL>exec DBMS_RESULT_CACHE.BYPASS (TRUE); -- to flush the result cache SQL>exec DBMS_RESULT_CACHE.FLUSH
Obtaining Information about Object Locks

-- DML locks by transactions on objects SELECT l.object_id, l.session_id, l.oracle_username, DECODE(l.locked_mode, 1, 'No Lock', 2, 'Row Share', 3, 'Row Exclusive', 4, 'Shared Table', 5, 'Shared Row Exclusive', 6, 'Exclusive') locked_mode, (select o.object_name from DBA_OBJECTS o WHERE o.object_id=l.object_id) object_name FROM V$LOCKED_OBJECT l order by ORACLE_USERNAME; SELECT dbl.lock_type, dbl.mode_held, dbl.blocking_others, dbo.object_name object_locked, dbo.object_type FROM dba_locks dbl, v$session v, dba_objects dbo WHERE v.username not in ('SYS','SYSTEM','DBSNMP','SYSMAN','OUTLN','TSMSYS','WMSYS','EXFSYS','CTXSYS','XDB' ,'ORDSYS','MDSYS','OLAPSYS','WKSYS','WK_TEST','IX') AND dbl.session_id = v.sid AND dbo.object_id = dbl.lock_id1;
Page 364
-- blocking session select ADDR, KADDR, l.SID, s.USERNAME, S.MACHINE, S.TERMINAL, S.PROGRAM, decode(l.TYPE, 'TM','DML enqueue','TX','Transaction enqueue','UL','User supplied') Block_Type, ID1, ID2, LMODE, REQUEST, CTIME, BLOCK from v$lock l, v$session s where s.sid = l.sid and l.block=1;
-- blocking and blocked sessions SELECT SUBSTR(TO_CHAR(w.session_id),1,5) WSID, p1.spid WPID, SUBSTR(s1.username,1,12) "WAITING User", SUBSTR(s1.osuser,1,8) "OS User", SUBSTR(s1.program,1,20) "WAITING Program", s1.client_info "WAITING Client", SUBSTR(TO_CHAR(h.session_id),1,5) HSID, p2.spid HPID, SUBSTR(s2.username,1,12) "HOLDING User", SUBSTR(s2.osuser,1,8) "OS User", SUBSTR(s2.program,1,20) "HOLDING Program", s2.client_info "HOLDING Client", o.object_name "HOLDING Object" FROM gv$process p1, gv$process p2, gv$session s1, gv$session s2, dba_locks w, dba_locks h, dba_objects o WHERE -- w.last_convert > 120 AND (Objects locked for 2 mins) h.mode_held != 'None' AND h.mode_held != 'Null' AND w.mode_requested != 'None' AND s1.row_wait_obj# = o.object_id AND w.lock_type(+) = h.lock_type AND w.lock_id1(+) = h.lock_id1 AND w.lock_id2 (+) = h.lock_id2 AND w.session_id = s1.sid (+) AND h.session_id = s2.sid (+) AND s1.paddr = p1.addr (+) AND s2.paddr = p2.addr (+) ORDER BY w.last_convert desc;
Handling a Hanging Database

Archiver Process Stuck. In the alert log file, you'll see: ORA-00257: archiver error. Connect internal only, until freed. o o Redirect archiving to a different directory. Clear the archive log destination by removing some archive logs.
Sys Audit is full. Check and delete $ORACLE_HOME/rdbms/audit Make sure the OS isn't swapping.
Accurately Measuring Process Size

Some tools like ps and top give you a misleading idea as to the process size, because they include the common shared TEXT sizes in individual processes.
Page 365
-- memory occupied by every session SELECT sid, n.name|| ' ('||s.statistic#||')', to_char(round(value/1024),'999,999') || ' KB' KB FROM v$sesstat s, v$statname n WHERE s.statistic# = n.statistic# AND n.name like '%ga memory%' ORDER BY SID, value DESC;
-- total memory allocated to the PGA and UGA memory SELECT 'Total PGA', round(SUM(value)/1024/1024,2) MB FROM V$SESSTAT s, V$STATNAME n WHERE s.statistic# = n.statistic# AND n.name in ('session pga memory') group by n.name union SELECT 'Total UGA', round(SUM(value)/1024/1024,2) MB FROM V$SESSTAT s, V$STATNAME n WHERE s.statistic# = n.statistic# AND n.name in ('session uga memory') group by n.name;
-- another way to know PGA used size select round(value/1024/1024,2) MB from V$PGASTAT where name='total PGA inuse';
Page 366
Managing Automatic Workload Repository (AWR)

The AWR and ADDM are Oracle products that need special licensing through the purchase of the Diagnostic Pack. -- To manually creating a snapshot: dbms_workload_repository.create_snapshot()
-- to list snapshots taken by AWR select SNAP_ID, STARTUP_TIME, BEGIN_INTERVAL_TIME, END_INTERVAL_TIME, FLUSH_ELAPSED, SNAP_LEVEL, ERROR_COUNT from DBA_HIST_SNAPSHOT order by STARTUP_TIME DESC, SNAP_ID DESC
-- To drop a range of snapshots: dbms_workload_repository.drop_snapshot_range (low_snap_id => 40,high_snap_id => 60, dbid => 2210828132) -- To modify a AWR setting: begin DBMS_WORKLOAD_REPOSITORY.MODIFY_SNAPSHOT_SETTINGS( retention => 43200, -- in mins interval => 30, -- between snaps in mins dbid => 3310949047 -- If NULL, the local dbid will be used ); end; / -- To get AWR setting select DBID, SNAP_INTERVAL, RETENTION from DBA_HIST_WR_CONTROL;
/* Creating and Deleting AWR Snapshot Baselines */ begin dbms_workload_repository.create_baseline (start_snap_id => 125, end_snap_id => 185, baseline_name => 'peak_time baseline', dbid => 2210828132 ); end; / -- to list created baselines: select DBID, BASELINE_ID, BASELINE_NAME, START_SNAP_ID, START_SNAP_TIME, END_SNAP_ID, END_SNAP_TIME from DBA_HIST_BASELINE order by START_SNAP_ID; -- To drop a snapshot baseline: -- if cascade is true, the snaps will also be deleted exec dbms_workload_repository.drop_baseline (baseline_name => 'peak_time baseline', cascade => FALSE)
Page 367
/* Creating AWR Reports */ <ORACLE_HOME>\RDBMS\ADMIN\awrrpt.sql -- allows you to select a single instance: <ORACLE_HOME>\RDBMS\ADMIN\awrrpti.sql
Page 368
Managing Automated Maintenance Tasks

Automatic Optimizer Statistics Collection Automatic Segment Advisor Automatic SQL Tuning Advisor /* Obtain Info about AMTs (11g) */ select * from DBA_AUTOTASK_TASK
/* Monitoring AMT */ SELECT client_name, status, attributes, window_group,service_name FROM dba_autotask_client;
/* Enabling and Disableing a Maintenance Task */ begin DBMS_AUTO_TASK_ADMIN.ENABLE (client_name => 'sql tuning advisor', operation => 'automatic sql tuning task', window_name => 'monday_window'); end; / begin dbms_auto_task_admin.disable (client_name => 'sql tuning advisor', operation => 'automatic sql tuning task', window_name => 'monday_window'); end; /
Using Automatic Database Diagnostic Monitor (ADDM)

The AWR and ADDM are Oracle products that need special licensing through the purchase of the Diagnostic Pack. /* Obtain info about ADDM */ -- list of all advisors tasks select * from DBA_ADVISOR_TASKS; -- findings (all advisors) select * from DBA_ADVISOR_FINDINGS where task_name='ADDM 01012009'; -- findings only add ADDM select * from DBA_ADDM_FINDINGS where task_name='ADDM 01012009'; -- recommendations select * from DBA_ADVISOR_RECOMMENDATIONS where task_name='ADDM 01012009';
Page 369
/* Setting Up ADDM */ -- CONTROL_MANAGEMENT_PACK_ACCESS should be DIAGNOSTIC or DIAGNOSTIC+TUNING -- STATISTICS_LEVEL should be TYPICAL or ALL (not BASIC) select name, value from v$parameter where upper(name) in ('CONTROL_MANAGEMENT_PACK_ACCESS','STATISTICS_LEVEL'); alter system set control_management_pack_access='DIAGNOSTIC+TUNING'
/* Determining Optimal I/O Performance */ -- Oracle assumes the value of the parameter (not intialization parameter) -- DBIO_EXPECTED is 10 milliseconds. SELECT PARAMETER_VALUE FROM DBA_ADVISOR_DEF_PARAMETERS WHERE ADVISOR_NAME='ADDM' AND PARAMETER_NAME='DBIO_EXPECTED' If your hardware is significantly different, you can set the parameter value one time for all subsequent ADDM executions: DBMS_ADVISOR.SET_DEFAULT_TASK_PARAMETER('ADDM','DBIO_EXPECTED', 8000);
/* Retreiving AWR snapshots */ SELECT SNAP_ID , INSTANCE_NUMBER , TO_CHAR(BEGIN_INTERVAL_TIME,'DD-MON-RR HH24:MI') BEGIN_TIME, TO_CHAR(END_INTERVAL_TIME,'DD-MON-RR HH24:MI') END_TIME FROM DBA_HIST_SNAPSHOT order by SNAP_ID desc;
/* Running ADDM in Database Mode */ DECLARE -- task name shouldn't exists in db TNAME VARCHAR2(50) :='ADDM DB 01012009'; BEGIN DBMS_ADDM.ANALYZE_DB ( task_name =>TNAME, -- it is IN OUT para begin_snapshot =>43, end_snapshot =>44); END; / -- also the report will then be generated @C:\oracle\oracledb11g\RDBMS\ADMIN\addmrpt.sql
/* Running ADDM in Instance Mode */ -- used in RAC DECLARE TNAME VARCHAR2(50) :='ADDM INST 01012009'; BEGIN DBMS_ADDM.ANALYZE_INST (
Page 370
task_name =>TNAME, -- it is IN OUT para begin_snapshot =>43, end_snapshot =>44, instance_number=>1); END; /
/* Running ADDM in Partial Mode */ -- subset of all database instances (in RAC) DECLARE -- task name shouldn't exists in db TNAME VARCHAR2(50) :='ADDM PART 01012009'; BEGIN DBMS_ADDM.ANALYZE_PARTIAL ( task_name =>TNAME, begin_snapshot =>43, end_snapshot =>44, instance_numbers=>'1,2,3'); END; /
/* Displaying an ADDM Report */ SET LONG 1000000 PAGESIZE 0; SELECT DBMS_ADDM.GET_REPORT('ADDM DB 01012009') FROM DUAL;
/* Deleting already created tasks */ exec DBMS_ADDM.DELETE(task_name=>'ADDM INST 01012009');
Using Automatic SQL Tuning Advisor

When SQL Tunining Advisor is running as an automatic task, it avoids analysing the following: Parallel queries Ad hoc queries Recursive statements SQL statements that use the INSERT and DELETE statements SQL statements that use DDL statements such as CREATE TABLE AS SELECT /* Obtain Info about Automatic SQL Tuning Advisor */ select OWNER, TASK_ID, TASK_NAME, EXECUTION_NAME, DESCRIPTION, EXECUTION_TYPE, EXECUTION_TYPE#, EXECUTION_START, EXECUTION_END, ADVISOR_NAME, STATUS, STATUS_MESSAGE, ERROR_MESSAGE from DBA_ADVISOR_EXECUTIONS order by task_id desc -- all SQL compilation and execution statistics DBA_ADVISOR_SQLSTATS
Page 371
DBA_ADVISOR_SQLPLANS
/* Configuring Automatic SQL Tuning */ -- Automatic SQL Tuning Advisor job runs for a maximum of one hour by default -- in seconds exec dbms_sqltune.set_tuning_task_parameter ('SYS_AUTO _SQL_TUNING_TASK', 'TIME_LIMIT', 14400) -- other parameters to set ACCEPT_SQL_PROFILES: TRUE/FALSE whether the database must automatically accept a SQL profile MAX_AUTO_SQL_PROFILES: maximum number of automatic SQL profiles allowed on the system, in sum MAX_SQL_PROFILES_PER_EXEC: maximum number of SQL profiles that can be automatically implemented per execution of the task. -- to view current parameter values: COLUMN parameter_value FORMAT A30 SELECT parameter_name, parameter_value FROM dba_advisor_parameters WHERE task_name = 'SYS_AUTO_SQL_TUNING_TASK' AND parameter_name IN ('ACCEPT_SQL_PROFILES', 'MAX_SQL_PROFILES_PER_EXEC', 'MAX_AUTO_SQL_PROFILES');
/* Enabling and Disable the automatic advisor */ -- enable it in all maintenance windows begin dbms_auto_task_admin.enable ( client_name => 'sql tuning advisor', operation => 'NULL', window_name='NULL'); end; / -- in a specific window begin dbms_auto_task_admin.enable ( client_name => 'sql tuning advisor', operation => 'NULL', window_name='monday_night_window'); end; /
/* View the Report */ -- display report of most recent execution: VARIABLE l_report CLOB; BEGIN :l_report := DBMS_SQLTUNE.report_auto_tuning_task( begin_exec => NULL, end_exec => NULL, type => DBMS_SQLTUNE.type_text, -- 'TEXT'
Page 372
level section object_id result_limit END; / print :l_report
=> => => =>
DBMS_SQLTUNE.level_typical, -- 'TYPICAL' DBMS_SQLTUNE.section_all, -- 'ALL' NULL, NULL);
Page 373
Implementing Automatic Memory Management

When you configure the database to use MEMORY_TARGET, you should take into consideration the following: The parameter STATISTICS_LEVEL must be set to TYPICAL If you set the parameters SGA_TARGET and PGA_TARGET, Oracle will consider the values as the minimum values for SGA and PGA.
Note
You may set PGA_AGGREGATE_TARGET to a value and then a command issued that is heavy on PGA, Oracle will try to expand the actual PGA size to serve the command. If Oracle could not find enough free memory in the machine, the following error will be returned: ORA-04030: out of process memory when trying to allocate 8192 bytes (sort subheap,sort key) Resolving the issue can be done by reducing the PGA to a lower value. Also, high value of SORT_AREA_SIZE may cause it. Consider reducing its value (to something like 1MB).
If you do not set the parameters SGA_TARGET and PGA_TARGET (or set them to zero), no minimum value is considered by Oracle for the SGA and PGA. When the instance starts, it assigns 60 percent to SGA and 40 percent to the PGA. When MEMORY_TARGET is configured, the following components are auto tuned: DB BUFFER CACHE, SHARED POOL, JAVA POOL, LARGE POOL and STREAMS POOL.
Note
On Linux systems, if you receive the following error after setting the MEMORY_TARGET parameter, most likely the reason is that the /dev/shm is allocated a size (can be known by issuing the command df -k) less than SGA_MAX_SIZE: ORA-00845: MEMORY_TARGET not supported on this system. Resolving the issue can be done by the following OS commands: #umount /dev/shm #mount -t tmpfs shmfs -o *size=><xx>m* /dev/shm
show parameter MEMORY_TARGET show parameter MEMORY_MAX_SIZE -- dynamic parameter ALTER SYSTEM SET MEMORY_TARGET = 410M ;
-- To set a proper value to the parameter MEMORY_TARGET (11g): V$MEMORY_TARGET_ADVICE. SELECT * FROM V$MEMORY_TARGET_ADVICE order by MEMORY_SIZE desc;
-- To display current status of the memory components, use the following query: col COMPONENT format a30 SELECT COMPONENT, ROUND(CURRENT_SIZE/1024/1024) CURRENT_SIZE , ROUND(MIN_SIZE/1024/1024) MIN, ROUND(MAX_SIZE/1024/1024) MAX FROM V$MEMORY_DYNAMIC_COMPONENTS order by CURRENT_SIZE desc;
Page 374
-- To know how Oracle has modified the memory area sizes by time (11g): select START_TIME, END_TIME, STATUS, COMPONENT, OPER_TYPE, OPER_MODE, PARAMETER, INITIAL_SIZE/1024/1024 INITIAL_SIZE_MB, TARGET_SIZE/1024/1024 TARGET_SIZE_MB, FINAL_SIZE/1024/1024 FINAL_SIZE_MB from V$MEMORY_RESIZE_OPS order by START_TIME, END_TIME
Page 375
Configuring DB_nK_CACHE_SIZE
ALTER SYSTEM SET DB_16K_CACHE_SIZE =1024M; CREATE TABLESPACE big_block_tbs DATAFILE '/test01/app/oracle/big_block_01.dbf' SIZE 1000M BLOCKSIZE 16K; ALTER TABLE names2 MOVE TABLESPACE big_block_tbs;
Page 376
Managing Optimizer Operations

Setting the Optimizer Mode
OPTIMIZER_MODE: ALL_ROWS (= CHOOSE)maximum throughput (good for patch processing) FIRST_ROWS_n where n 1,10,100,1000 maximum response time (good for interactive apps) FIRST_ROWS (deprecated) same as above SHOW PARAMETER OPTIMIZER_MODE select value from v$parameter where upper(name)='OPTIMIZER_MODE' ;
-- instance level: ALTER SYSTEM SET optimizer_mode = first_rows_10; -- session level (PL/SQL blocks won't be affected) ALTER SESSION SET optimizer_mode = first_rows_10;
-- statement level select /*+ first_rows(10) */ SELECT /*+ ALL_ROWS */ ...
..
Defining Access Paths and Joins for the Query Optimizer

Access Paths: Full Table Scans (FULL) Rowid Scans Index Scans: o o o o Index Unique Scans (INDEX): using UNIQUE or PRIMARY KEY indexes Index Range Scans (INDEX, INDEX_ASC, INDEX_DESC, NO_INDEX_RS): <,>,=, Like on index key Index Range Scans Descending (INDEX_DESC): when an order by descending clause can be satisfied by an index Index Skip Scans (INDEX_SS, INDEX_SS_ASC, INDEX_SS_DESC): when there are few distinct values in the leading column of the composite index and many distinct values in the nonleading key of the index. For example: INDEX(sex, emp_id) and select ... where emp_id=100 Full Scans: it is available if a predicate references one of the columns in the index or all of the columns in the table referenced in the query are included in the index. It read single blocks. Fast Full Index Scans (INDEX_FFS, INDEX_FFS_ASC, INDEX_FFS_DESC, NO_INDEX_FFS): when the index contains all the columns that are needed for the query, and at least one column in the index key has the NOT NULL constraint. It reads the entire index using multiblock reads, unlike a full index scan, and can be parallelized.
Page 377
o o
Index Joins (INDEX_JOIN): it is a hash join of several indexes that together contain all the table columns that are referenced in the query. Bitmap Indexes (INDEX_COMBINE or INDEX): the optimizer uses a bitmap index to get the rowids. It requires EE license.
Cluster Access (CLUSTER): when a table is stored in an indexed cluster. Hash Access (HASH): is used to locate rows in a hash cluster, based on a hash value. Sample Table Scans: when using SAMPLE [BLOCK] keyword in the select statement.
Joins: Nested Loop Join (USE_NL, No_USE_NL): for every row in the outer table, Oracle accesses all the rows in the inner table. Hash Join (USE_HASH, NO_USE_HASH): used when joining one large table with a small one. The smaller data source is built into memory as hash table. It then scans the larger table, probing the hash table to find the joined rows. Euality condition is a must. Sort Merge (USER_MERGE, NO_USE_MERGE): It performs well when no order is required on the data source (or already sorted) and the join condition is NOT equality. Outer Joins: extends the result of a simple join: o o o o Nested Loop Outer Joins: The outer table will be the one with rows that are being preserved. Hash Join Outer Joins Sot Merge Outer Joins Full Outer Join
Cartesian: two or more tables with no joins. /* Access Path Hints */ -- Full Table Scans select /* FULL(table_alias) */ -- "order by" might be served by the index select /*+ index_DESC(n names3i2) */ * from names3 n where name like '%JZ' order by NAME DESC -- you can indicate whether to save the block in buffer cache select /* CACHE . select /* NOCACHE .
-- Index Unique Scan Hints -- Index Unique Scan as access path is auto used by the optimizer -- you can define the hint to use the index though: select /* INDEX(table_alias index_name) */ ... select /*+ INDEX(e1 emp_emp_id_pk) .. */ from employees e1, .. SELECT /*+ index(t1 t1_abc) index(t2 t2_abc) */ COUNT(*) FROM t1, t2 WHERE t1.col1 = t2.col1;
Page 378
-- Index Range Scan Hints -- to instruct the optimizer to use a specific index (to avoid fts): select /* INDEX(table_alias index_name) */ ... select /*+ INDEX(e1 emp_emp_id_pk) .. */ from employees e1, ..
-- Index Range Scan Descending Hints -- use it when an order by descending clause can be satisfied by an index select /*+ INDEX_DESC(t indexC) */ * from try t where C < sysdate order by C desc -- exclude the index range scan access path SELECT /*+ NO_INDEX_RS(e emp_name_ix) */ last_name FROM employees e WHERE first_name BETWEEN 'A' AND 'B';
-- Fast Full Index Scan Hints select /*+ INDEX_FFS(t indexb) */ b from try t where t.b='b' select /*+ index_ffs_asc(n i1) */ name from names2 n order by name SELECT /*+ NO_INDEX_FFS(i pk_serv_inst) NO_INDEX_FFS(i ix_serv_inst) */ latitude FROM servers s, serv_inst i WHERE s.srvr_id = i.srvr_id;
-- Index Join Hints SELECT /*+ INDEX_JOIN(e emp_manager_ix emp_department_ix) */ department_id FROM employees e WHERE manager_id < 110 AND department_id < 50;
-- Bitmap Index -- bitmap indexes should be there SELECT /*+ INDEX_COMBINE(e emp_manager_ix emp_department_ix) */ * FROM employees e WHERE (manager_id = 108) OR (department_id = 110);
-- Cluster CREATE CLUSTER sc_srvr_id ( srvr_id NUMBER(10)) SIZE 1024;
Page 379
CREATE INDEX idx_sc_srvr_id ON CLUSTER sc_srvr_id; CREATE TABLE cservers CLUSTER sc_srvr_id (srvr_id) AS SELECT * FROM servers; CREATE TABLE cserv_inst CLUSTER sc_srvr_id (srvr_id) AS SELECT * FROM serv_inst; SELECT /*+ CLUSTER(cservers) */ srvr_id FROM cservers WHERE srvr_id = 503 GROUP BY srvr_id;
-- Hash CREATE CLUSTER sthc_si (srvr_id NUMBER(10)) SIZE 1024 SINGLE TABLE HASHKEYS 11 TABLESPACE uwdata; CREATE TABLE si_hash CLUSTER sthc_si (srvr_id) AS SELECT * FROM serv_inst; SELECT /*+ HASH(si_hash) */ srvr_id FROM si_hash WHERE srvr_id = 503 GROUP BY srvr_id;
-- Sample data -- SAMPLE (1) : read 1% of rows -- SAMPLE BLOCK (1) : read 1% of blocks SELECT * FROM employees SAMPLE BLOCK (1);
/* Join Operations Hints */ -- Nested Loops select /*+ use_nl(e d) use_nl(e j) */ -- or No_use_nl employee_id , department_name, salary, j.job_title from employees e , departments d, jobs j where e.department_id = d.department_id and e.job_id = j.job_id and d.department_id>10 /
-- Hash Join SELECT /*+ USE_HASH(o l) */ o.customer_id, l.unit_price * l.quantity FROM orders o ,order_items l WHERE l.order_id = o.order_id;
Page 380
-- Sort Merge select /*+ use_merge(e g) */ employee_id , last_name, grade from employees e, grades g where e.salary between g.ssal and g.esal;
-- Full Outer Join SELECT d.department_id, e.employee_id FROM employees e FULL OUTER JOIN departments d ON e.department_id = d.department_id ORDER BY d.department_id;
Gathering Optimizer Statistics

Gathering Object Statistics See Collecting Object Statistics. Preferences can be set. See next section. Collecting Dictionary Object Stats, see below. Recommended every week in off beak times. It may take considerable time (hours).
/* Retreive Gathered Stats */ -- table stats select OWNER, TABLE_NAME, PARTITION_NAME, PARTITION_POSITION, SUBPARTITION_NAME, SUBPARTITION_POSITION, OBJECT_TYPE, NUM_ROWS, BLOCKS, EMPTY_BLOCKS, AVG_SPACE, CHAIN_CNT, AVG_ROW_LEN, AVG_SPACE_FREELIST_BLOCKS, NUM_FREELIST_BLOCKS, AVG_CACHED_BLOCKS, AVG_CACHE_HIT_RATIO, SAMPLE_SIZE, LAST_ANALYZED, GLOBAL_STATS, USER_STATS, STATTYPE_LOCKED, STALE_STATS from DBA_TAB_STATISTICS where owner not in ('SYS','SYSTEM','DBSNMP','SYSMAN','OUTLN','TSMSYS','WMSYS','EXFSYS','CTXSYS',' XDB','ORDSYS','MDSYS','OLAPSYS','WKSYS','WK_TEST','IX') and owner NOT LIKE 'FLOWS_%' order by OWNER, LAST_ANALYZED DESC; -- col stats select OWNER, TABLE_NAME, COLUMN_NAME, NUM_DISTINCT, LOW_VALUE, HIGH_VALUE, DENSITY, NUM_NULLS, NUM_BUCKETS, LAST_ANALYZED, SAMPLE_SIZE, GLOBAL_STATS, USER_STATS, AVG_COL_LEN, HISTOGRAM from DBA_TAB_COL_STATISTICS where owner not in ('SYS','SYSTEM','DBSNMP','SYSMAN','OUTLN','TSMSYS','WMSYS','EXFSYS','CTXSYS',' XDB','ORDSYS','MDSYS','OLAPSYS','WKSYS','WK_TEST','IX') and owner NOT LIKE 'FLOWS_%' order by OWNER, LAST_ANALYZED DESC;
/* Collecting data dicionary stats */ -- for fixed tables conn / as sysdba EXECUTE DBMS_STATS.GATHER_FIXED_OBJECTS_STATS -- for real dictionary tables: -- SYS and SYSTEM users as well as the owners of all database components
Page 381
conn / as sysdba -- it may take hours to finish EXECUTE DBMS_STATS.GATHER_DICTIONARY_STATS Gathering System Statistics Gathering mode controlled by GATHERING_MODE parameter: o o No-Workload Mode: implemented by passed NOWORKLOAD. It gathers general I/O stats. Workload Mode: START will manually start the gathering. STOP will manually stops it. INTERVAL gathering is controlled by INTERVAL parameter.
See Managing OS statistics. -- start gathering EXECUTE dbms_stats.gather_system_stats('start');
-- stop gathering and save the stats EXECUTE dbms_stats.gather_system_stats('stop');
-- for three minutes interval EXECUTE dbms_stats.gather_system_stats('INTERVAL',3);
-- view gathered stats SELECT SNAME, PNAME, decode(PNAME, 'IOTFRSPEED','I/O transfer Speed (B/ms)', 'IOSEEKTIM','Seek+Latency+OS Overhead Time (ms)', 'SREADTIM','Single-Block Read Average Time (ms)', 'MREADTIM','MBRC Block Sequential Average Read Time (ms)', 'CPUSPEED','Average Number of CPU Cycles Captured for the Workload', 'CPUSPEEDNW','Average Number of CPU Cycles Captured for the Non-Workload', 'MBR','Average Multiblock Read Count for Sequential read (in Blocks)', 'MAXTHR','Maximum I/O System Throughput (B/s)', 'SLAVETHR','Average Slave I/O Throughput (B/s)', PNAME) Description, PVAL1, PVAL2 FROM sys.aux_stats$;
Changing Statistics Preferences

The function DBMS_STATS.GET_PARAM is used to in Oracle 10g to return the default values of parameters of the DBMS_STATS package. This function is now obsolete in Oracle 11g and replaced with GET_PREFS procedure. Following is an example: SET SERVEROUTPUT ON declare v_value varchar2(100); begin v_value := DBMS_STATS.GET_PREFS ( PNAME =>'STALE_PERCENT',
Page 382
OWNNAME =>'HR', TABNAME =>'EMPLOYEES'); DBMS_OUPTPUT.PUT_LINE(v_value); end; Regarding the GET_PREFS function, consider the following: PNAME parameter indicates the preference name and can take one of the following values: CASCADE, DEGREE, ESTIMATE_PERCENT, METHOD_OPT, NO_INVALIDATE, GRANULARITY, PUBLISH, INCREMENTAL and STALE_PERCENT. If the OWNNAME and TABNAME are provided and a preference has been entered for the table, the function returns the preference as specified for the table. In all other cases it returns the global preference if it has been specified, otherwise the default value is returned.
SET_GLOBAL_PREFS, SET_DATABASE_PREFS, SET_SCHEMA_PREFS, SET_TABLE_PREFS procedures are used to set the statistics preferences for the global, database, schema or table levels respectively. Following is an example: begin DBMS_STATS.SET_GLOBAL_PREFS ( PNAME =>'ESTIMATE_PERCENT', PVALUE =>'75'); end; Similarly, the procedures DELETE_*_PREFS are used to delete current statistics preferences. EXPORT_*_PREFS and IMPORT_*_PREFS procedures are used to export and import statistics preferences. Following is an example: begin DBMS_STATS.EXPORT_DATABASE_PREFS( STATTAB =>'mytable', -- table name to where statistics should be exported STATID =>'prod_prefs', -- identifier to associate with these statistics STATOWN =>'HR'); -- Schema containing stattab (if other than ownname) end;
Managing Pending and Published Statistics

Starting with Oracle 11g, when gathering statistics, you have the option to automatically publish the statistics at the end of the gather operation (default behavior), or to have the new statistics saved as pending. Saving the new statistics as pending allows you to validate the new statistics and publish them only if they are satisfactory. You can check whether or not the statistics will be automatically published checking the value of the PUBLISH attribute using the DBMS_STATS package as in the following example: SELECT DBMS_STATS.GET_PREFS('PUBLISH') PUBLISH FROM DUAL; You can change the PUBLISH setting at either the schema or table level. Following are examples to do so: -- setting PUBLISH at schema level exec DBMS_STATS.SET_SCHEMA_PREFS('HR', 'PUBLISH', 'FALSE'); -- setting PUBLISH at table level exec DBMS_STATS.SET_TABLE_PREFS('HR','EMPLOYEES', 'PUBLISH', 'FALSE'); Published statistics are stored in data dictionary views, such as DBA_TAB_STATISTICS and USER_IND_ STATISTICS. Pending statistics are stored in views such as DBA_TAB_PENDING_STATISTICS and DBA_IND_PENDING_STATISTICS. select NUM_ROWS, BLOCKS, AVG_ROW_LEN, SAMPLE_SIZE, LAST_ANALYZED from DBA_TAB_PENDING_STATISTICS where OWNER='HR' AND TABLE_NAME ='EMPLOYEES'; By default, the optimizer uses the published statistics stored in the data dictionary views. If you want the optimizer to use the newly collected pending statistics, set the initialization parameter Page 383
OPTIMIZER_PENDING_STATISTICS to TRUE (the default value is FALSE), and then run a workload against the table or schema or just gather its statistics: ALTER SESSION SET OPTIMIZER_PENDING_STATISTICS = TRUE; The optimizer will use the pending statistics (if available) instead of the published statistics when compiling SQL statements. If the pending statistics are valid, they can be made public by executing the following statement: -- for the whole database exec DBMS_STATS.PUBLISH_PENDING_STATS(null, null); -- publishing specific database object pending statistics exec DBMS_STATS.PUBLISH_PENDING_STATS('HR','EMPLOYEES'); If you do not want to publish the pending statistics, delete them by executing the following statement: exec DBMS_STATS.DELETE_PENDING_STATS('HR','EMPLOYEES');
Managing Extended Statistics

MultiColumn Statistics
When multiple columns from a single table are used together in the where clause of a query (multiple single column predicates), Oracle optimizer in previous versions (before 11g) does not understand the relationship between the columns. In Oracle 11g, statistics on these columns as a group (column group) can be gathered and thus resulting in high enhancement in CBO cardinality estimation. You can also create column groups manually by using the CREATE_EXTENDED_STATS function in the DBMS_STATS package. This function creates a column statistics entry in the system for a user specified column group or an expression in a table. This function returns a system-generated name of this newly created entry for the extension. Note: The optimizer will only use MultiColumn statistics with equality predicates. Following is an example: declare V_NAME VARCHAR2(30); begin -- stats of the combined columns will be collected next time the stats is gathered V_NAME := DBMS_STATS.CREATE_EXTENDED_STATS( OWNNAME => NULL, TABNAME => 'EMPLOYEES', EXTENSION => '(STATE_ID,COUNTRY_ID)'); -- you can then issue the gathering process DBMS_STATS.GATHER_TABLE_STATS (NULL, 'EMPLOYEES', skewonly' ); end; METHOD_OPT='for all columns size
The DBMS_STATS.GATHER_TABLE_STATS procedure can also be used to create column group and gather its statistics all in one step. The keyword FOR COLUMNS is used in this case as shown in the following example: begin DBMS_STATS.GATHER_TABLE_STATS ('HR', 'EMPLOYEES', METHOD_OPT=>'for all columns size skewonly for columns (STATE_ID,COUNTRY_ID)' ); end;
Page 384
Note
The default value of METHOD_OPT is 'FOR ALL COLUMNS SIZE AUTO' which makes Oracle create column groups for a table, based on the workload analysis, similar to how it is done for histograms.
You can use the methods in the following code examples to retrieve information on column groups that have been created: -- you can query the data dictionary USER_STAT_EXTENSIONS select EXTENSION_NAME, EXTENSION from USER_STAT_EXTENSIONS where TABLE_NAME='EMPLOYEES'; -- you can query USER_TAB_COL_STATISTICS (extension name appears as COLUMN_NAME) select COLUMN_NAME, NUM_DISTINCT, HISTOGRAM from USER_TAB_COL_STATISTICS where TABLE_NAME = 'EMPLOYEES'; -- you can use DBMS_STATS.SHOW_EXTENDED_STATS_NAME function select DBMS_STATS.SHOW_EXTENDED_STATS_NAME(OWNNAME => 'HR', TABNAME => 'EMPLOYEES', EXTENSION => 'STATE_ID,COUNTRY_ID') AS E_NAME from dual; After gathering the multi-column statistics as show in the example, when you check the explain plan for a query of a where condition like "STATE_ID = 'CA' AND COUNTRY_ID = 'US'", you will notice that the optimizer has retrieved the correct number of expected retrieved rows. Practically, this will lead to a significant improvement in the statement execution. Following is how to drop a column group that you have previously defined: exec DBMS_STATS.DROP_EXTENDED_STATS('HR', 'EMPLOYEES', '(STATE_ID,COUNTRY_ID)');
Expression Statistics
In Oracle 11g, you can create statistics on an expression. Following are examples to do that: declare V_NAME VARCHAR2(30); begin -- to create expression extended stats (not statistics are yet gathered) V_NAME := DBMS_STATS.CREATE_EXTENDED_STATS( OWNNAME => NULL, TABNAME => 'EMPLOYEES', EXTENSION => '(lower(last_name))'); end; begin -- to create expression extended stats and gather the statistics in one step DBMS_STATS.GATHER_TABLE_STATS(OWNNAME=>NULL,TABLE_NAME=>'EMPLOYEES', METHOD_OPT=>'for all columns size skewonly for columns (lower(last_name))'); end; After executing the code above, if you issue a query from EMPLOYEES table with a condition like LOWER(LAST_NAME)='ABC', the optimizer has statistics about the retrieved rows and thus will be able to estimate the correct number of returned rows. Consequently, the optimizer will most likely create a more efficient plan than if those statistics were not present. Use the DBA_STAT_EXTENSIONS data dictionary view to retrieve information on expression statistics that have been created in the database. select EXTENSION_NAME, EXTENSION from USER_STAT_EXTENSIONS where TABLE_NAME='EMPLOYEES'; Following is an example of the removal of an extended expression statistic: exec DBMS_STATS.DROP_EXTENDED_STATS(null, 'EMPLOYEES', '(lower(lat_name))' ); Note that you will not be able to drop an extended expression statistics, if a function-based index is dependent on that statistic (ORA-20000 error will be returned). Page 385
A Simple Approach to Tuning SQL Statements

Identify Problem Statements o o Using V$SQL or SQL Tracing AWR report and the ADDM analysis
Locate the Source of the Inefficiency o o Show the EXPLAIN PLAN of the statement Use the SQL Access Advisor and SQL Tunining Advisor
Page 386
Using Application Tracing Tools

Using the SQL Trace Facility and TKPROF
1. Set init parameters for trace file management. They are all system and session level STATISTICS_LEVEL = Typical or ALL Even if STATISTICS_LEVEL is set to TYPICAL or ALL, you can keep the database from tracing by using: ALTER SESSION statement to set TIMED_STATISTICS to FALSE. show parameter USER_DUMP_DEST show parameter MAX_DUMP_FILE_SIZE 2. Enable the SQL Trace facility for the desired session. -- current session ALTER SESSION SET sql_trace=true; begin sys.dbms_session.set_sql_trace(true); end; /
-- specific session exec DBMS_SYSTEM.SET_SQL_TRACE_IN_SESSION(SID=>123, SERIAL#=>567, SQL_TRACE=> true); exec DBMS_MONITOR.SESSION_TRACE_ENABLE(session_id => 27, serial_num => 60, waits => TRUE, binds => FALSE);
3. Run TKPROF to translate the trace file into a readable output file and optionally create a SQL script that can be used to store the statistics in a database. tkprof finance_ora_16340.trc test.txt sys=no explain=y AGGREGATE =yes|no WAITS =yes|no
4. Interpret the output file. Row Source Operations cr specifies consistent reads performed by the row source specifies physical reads performed by the row source w specifies physical writes performed by the row source time specifies time in microseconds
Using the Event 10046 to Trace SQL Code

This method is more advanced than the SQL Trace utility. alter alter alter alter
Page 387
session session session session
set set set set
timed_statistics = true; statistics_level=all; max_dump_file_size=unlimited; events '10046 trace name context forever, level 12'; Oracle DBA Code Examples
run the code to trace ALTER SESSION SET EVENT='10046 trace name context off'; -- trace file will be in USER_DUMP_DEST (usually udump folder) -- not recommended ALTER SYSTEM SET EVENT='10046 trace name context forever, level 4' SCOPE=spfile; ALTER SYSTEM SET EVENT='' SCOPE=spfile;
-- Setting the 10046 Event For a Session Other than Your Own -- Method 1 -- SPID known (OR next) oradebug setospid 1864; -- PID known oradebug setpid 21; -- then oradebug event 10046 trace name context forever, level 12 oradebug event 10046 trace name context off -- Method 2 Dbms_system.set_ev ( si binary_integer, -- SID se binary_integer, -- Serial# ev binary_integer, -- Event code or number to set. le binary_integer, -- Usually level to trace cm binary_integer -- When to trigger (NULL = context forever.)
Tracing End to End Application

/* To specify module and action names */ exec DBMS_APPLICATION_INFO.SET_MODULE(MODULE_NAME=>'FINANCIAL',ACTION_NAME=>'PAYROL L'); exec DBMS_APPLICATION_INFO.SET_ACTION(ACTION_NAME=>'PAY SLIP');
/* To add your own marker to the trace file names */ -- so you can more easily find the generated files. ALTER SESSION SET TRACEFILE_IDENTIFIER ="fin_payrol";
/* Enabling Collection of Client and Service Statistics */ For client-Level Statistics use: DBMS_MONITOR.CLIENT_ID_STAT_ENABLE(<client_id>) For Service-Level Statistics: SELECT NAME FROM V$ACTIVE_SERVICES ; select service_name from v$session where username='HR'; -- service name case sensitive DBMS_MONITOR.SERV_MOD_ACT_TRACE_ENABLE(<service_name>,<module_name>, <action_name>) Oracle DBA Code Examples
Page 388
For example: DBMS_MONITOR.SERV_MOD_ACT_TRACE_ENABLE( service_name=>'APPS1',module_name =>'PAYROLL') To enable tracing for a Service named APPS1: DBMS_MONITOR.SERV_MOD_ACT_TRACE_ENABLE('APPS1',DBMS_MONITOR.ALL_MODULES, DBMS_MONITOR.ALL_ACTIONS,TRUE,FALSE,NULL) To enable tracing for a session: DBMS_MONITOR.SESSION_TRACE_ENABLE (SESSION_ID=>139, SERIAL_NUM=>53, WAITS=>TRUE, BINDS=>FALSE); To enable trace in the whole database DBMS_MONITOR.DATABASE_TRACE_ENABLE To enable trace in the instance level DBMS_MONITOR.DATABASE_TRACE_ENABLE (INSTANCE_NAME=>'RAC1') To disable tracing: DBMS_MONITOR.CLIENT_ID_STAT_DISABLE(<Client_id>) DBMS_MONITOR.SERV_MOD_ACT_TRACE_DISABLE('APPS1') DBMS_MONITOR.DATABASE_TRACE_DISABLE(INSTANCE_NAME=>'RAC1')
/* Using the TRCSESS Tool to Analyze Trace Files */ show parameter USER_DUMP_DEST trcsess output="hr_report.trc" service="APPS1" module="PAYROLL" action="bulk load" You can then run TKPROF against the consolidated trace file to generate a report: tkprof hr_report.trc output=hr_trc_report sys=np SORT=(EXEELA, PRSELA, FCHELA)
Enabling and Disabling Statistic Gathering for End to End Tracing

-- first make sure the trcing is enabled, see sections above.
-- Statistic Gathering for Client Identifier select CLIENT_IDENTIFIER from V$SESSION where username='SA'; EXECUTE DBMS_MONITOR.CLIENT_ID_STAT_ENABLE(client_id => 'OE.OE'); EXECUTE DBMS_MONITOR.CLIENT_ID_STAT_DISABLE(client_id => 'OE.OE');
-- Statistic Gathering for Service, Module, and Action EXECUTE DBMS_MONITOR.SERV_MOD_ACT_STAT_ENABLE(service_name => 'ACCTG', module_name => 'PAYROLL'); EXECUTE DBMS_MONITOR.SERV_MOD_ACT_STAT_ENABLE(service_name => 'ACCTG', module_name => 'GLEDGER', action_name => 'INSERT ITEM'); EXECUTE DBMS_MONITOR.SERV_MOD_ACT_STAT_DISABLE(service_name => 'ACCTG', module_name => 'GLEDGER', action_name => 'INSERT ITEM');
/* Viewing the New Statistics */

Page 389
DBA_ENABLED_AGGREGATIONS enabled statistics aggregation DBA_ENABLED_TRACES enabled traces in the system V$CLIENT_STATS statistics on a client level (CLIENT_IDENTIFIER based) V$SERVICE_STATS Displays basic performance statistics V$SERV_MOD_ACT_STATS statistics for a combination of serve /module/action names.
Page 390
Writing Efficient SQL

Efficient WHERE Clauses o Select with high selectivity gets advantage from index. FTS performs better with low selectivity queries. o Whe using SQL functions, consider using function-based indexes. Using the Right Joins o Performing filtering operations early reduces the number of rows to be joined in later steps. o Always consider usinjg the equi join. o Join in the order that will produce the least number of rows as output to the parent step. Combine Multiples Scans with CASE Statements o When you need to calculate multiple aggregates from the same table, avoid writing a separate query for each aggregate. Use CASE statement instead. -- the following selects: SELECT COUNT (*) FROM employees WHERE salary < 2000; SELECT COUNT (*) FROM employees WHERE salary BETWEEN 2000 AND 4000; SELECT COUNT (*) FROM employees WHERE salary>4000; -- can be re written as: SELECT COUNT (CASE WHEN salary < 2000 THEN 1 ELSE null END) count1, COUNT (CASE WHEN salary BETWEEN 2001 AND 4000 THEN 1 ELSE null END) count2, COUNT (CASE WHEN salary > 4000 THEN 1 ELSE null END) count3 FROM employees; Efficient Subquery Execution o Oracle recommends using the IN clause if the subquery has the selective WHERE clause. o If the parent query contains the selective WHERE clause, use the EXISTS clause rather than the IN clause. -- the follwoing (selective predicate in the sub query) WHERE EXISTS (SELECT 1 FROM orders o WHERE e.employee_id = o.sales_rep_id AND o.customer_id = 144); -- can be converted into: WHERE e.employee_id IN (SELECT o.sales_rep_id FROM orders o WHERE o.customer_id = 144); -- in the following (selective predicate in the parent query): WHERE e.department_id = 80 AND e.job_id = 'SA_REP' AND e.employee_id IN (SELECT o.sales_rep_id FROM orders o); -- can be rewritten as: WHERE e.department_id = 80 Oracle DBA Code Examples
Page 391
AND e.job_id = 'SA_REP' AND EXISTS (SELECT 1 FROM orders o WHERE e.employee_id = o.sales_rep_id) Avoid Transformed Columns in the WHERE Clause such as: charcol = numexpr, col1 = NVL (:b1,col1), NVL (col1,-999) = ., TO_DATE(), TO_NUMBER(), and so on. Add the predicate versus using NVL() technique. SELECT employee_num, full_name Name, employee_id FROM mtl_employees_current_view WHERE (employee_num = NVL (:b1,employee_num)) AND (organization_id=:1) ORDER BY employee_num; Write Separate SQL Statements for Specific Tasks: make a very complex statement slightly less complex by using the UNION ALL operator or use PL/SQL blocks, if possible. -- this code won't use the index bcz somecol in both sides .. WHERE ... AND somecolumn BETWEEN DECODE(:loval, 'ALL', somecolumn, :loval) AND DECODE(:hival, 'ALL', somecolumn, :hival); -- it can be rewritten as .. WHERE ... AND somecolumn BETWEEN :loval AND :hival AND (:hival != 'ALL' AND :loval != 'ALL') UNION ALL .. WHERE ... AND (:hival = 'ALL' OR :loval = 'ALL'); If possible, define the Join Order using ORDERED or STAR hints. Table with the lowest percentage of retireved compared to its tolal it candidate to be the driving table. -- in this example, a is the SELECT info FROM taba a, tabb b, tabc c WHERE a.acol BETWEEN 100 AND AND b.bcol BETWEEN 10000 AND AND c.ccol BETWEEN 10000 AND AND a.key1 = b.key1 AND a.key2 = c.key2; driving table.
200 20000 20000
SELECT /*+ LEADING(e2 e1) USE_NL(e1) INDEX(e1 emp_emp_id_pk) USE_MERGE(j) FULL(j) */ e1.first_name, e1.last_name, j.job_id, sum(e2.salary) total_sal FROM employees e1, employees e2, job_history j WHERE e1.employee_id = e2.manager_id AND e1.employee_id = j.employee_id AND e1.hire_date = j.start_date GROUP BY e1.first_name, e1.last_name, j.job_id ORDER BY total_sal; Modifying or Disabling Triggers and Constraints Restructuring the Data: like using vitual columns, adding new columns or using partitions. Maintaining Execution Plans Over Time: by using stored statistics or SQL plan baselines. Use DML with RETURNING Clause: INSERT, UPDATE, or DELETE... RETURNING modify and then return the data in one call.
Page 392
var bnd1 NUMBER var bnd2 VARCHAR2(30) var bnd3 NUMBER UPDATE employees SET job_id ='SA_MAN', salary = salary + 1000, department_id = 140 WHERE last_name = 'Jones' RETURNING salary*0.25, last_name, department_id INTO :bnd1, :bnd2, :bnd3; Consider using Test Case Builder. Consider using Bitmap Join Indexes on a star model query. Selecting the Best Join Order: If youre joining three tables, the one with the more restrictive filter (driving table) should be joined first (after the FROM keyword) to one of the other two tables. -- this statement is less efficient than the following one select /*+ ordered */ order_date, order_total, line_item_id id, product_name ,quantity, quantity*unit_price item_tprice from order_items i, PRODUCT_INFORMATION p, orders o where order_date between to_date('01-01-2010','dd-mm-yyyy') and to_date('31-032010','dd-mm-yyyy') and o.order_id=i.order_id and i.product_id=p.product_id;
-- more efficient select /*+ ordered */ order_date, order_total, line_item_id id, product_name ,quantity, quantity*unit_price item_tprice from orders o, order_items i, PRODUCT_INFORMATION p where order_date between to_date('01-01-2010','dd-mm-yyyy') and to_date('31-032010','dd-mm-yyyy') and o.order_id=i.order_id and i.product_id=p.product_id;
Page 393
Improving SQL Processing Techniques

Following are the tips to improve SQL processing, even if you cannot change the code: Using Partitioned Tables Using Compression Techniques CREATE INDEX item_product_x ON order_items(product_id) TABLESPACE order_items_indx_01 COMPRESS; CREATE INDEX item_product_x ON order_items(product_id) TABLESPACE order_items_indx_01 COMPRESS; Using Materialized Views Using Stored Outlines to Stabilize the CBO (and SQL Plan Management in Oracle 11g). Using Parallel Execution
Page 394
Using SQL Tuning Advisor

The advisor possible inputs are: a single statement SQL tuning set or STS a SQL identifier from the AWR a SQL identifier from the cursor cache To manage the Automatic SQL Tuning Advisor, see Using Automatic SQL Tuning Advisor.
-- required priv grant advisor to youruser;
/* A single SQL Statement */ DECLARE my_task_name VARCHAR2(30); my_sqltext CLOB; BEGIN my_sqltext := 'SELECT /*+ ORDERED */ * FROM employees e, locations l, departments d WHERE e.department_id = d.department_id AND l.location_id = d.location_id AND e.employee_id < :bnd'; my_task_name := DBMS_SQLTUNE.CREATE_TUNING_TASK( sql_text => my_sqltext, bind_list => sql_binds(anydata.ConvertNumber(90)), user_name => 'HR', scope => 'COMPREHENSIVE', time_limit => 60, task_name => 'my_sql_tuning_task', description => 'Task to tune a query on a specified employee'); END; /
BEGIN DBMS_SQLTUNE.EXECUTE_TUNING_TASK( task_name => 'my_sql_tuning_task' ); END; /
SET LONG 1000 SET LONGCHUNKSIZE 1000 SET LINESIZE 100 SELECT DBMS_SQLTUNE.REPORT_TUNING_TASK( 'my_sql_tuning_task') FROM DUAL; -- you can query: DBA_ADVISOR_TASKS DBA_ADVISOR_FINDINGS DBA_ADVISOR_RECOMMENDATIONS DBA_ADVISOR_RATIONALE DBA_SQLTUNE_STATISTICS DBA_SQLTUNE_PLANS
Page 395
Part 6 Oracle Automatic Storage Management (ASM)
Page 396
Managing Oracle ASM

Obtaining Information about ASM Instance
-- asm: all diskgroups mounted to by db select * from V$ASM_DISKGROUP; -- asm: dbs connected to the asm select * from V$ASM_CLIENT; -- asm: discovered disks select * from V$ASM_DISK; -- asm files select * from V$ASM_FILE; -- info about asm templates select * from V$ASM_TEMPLATE; db: diskgroups mounted in asm and connected
db: asms connected by db
db: disks used by db
Creating an ASM Instance

The procedure below does not apply in 11g R2.
-- required services <orahome>\bin\localconfig add -- Creating the ASM Instance Manually (on Windows XP) 1. Building the ASM Candidate "disks": for testing or development purpose mkdir c:\asmdisks ASMTOOL -create c:\asmdisks\asmdiska1 150 -- you can use DBCA and avoid 2,3,4 2. Create a pfile with the name "init+ASM.ora" in the folder <ORACLE_HOME>\database. Insert the following parameters in the file: INSTANCE_TYPE=ASM _ASM_ALLOW_ONLY_RAW_DISKS = FALSE DB_UNIQUE_NAME = +ASM ASM_DISKSTRING ='C:\asmdisks\*' LARGE_POOL_SIZE = 16M
3. Create the ASM instance service: ORADIM -NEW -ASMSID +ASM -STARTMODE auto 4. Startup the instance SET ORACLE_SID=+ASM SQLPLUS / AS SYSDBA SQL> STARTUP FORCE SQL> SELECT PATH, MOUNT_STATUS FROM V$ASM_DISK; SELECT instance_name FROM v$instance;
Page 397
LSNRCTL STATUS
-- starting ASM in restricted mode: DB cannot access it (11g) -- in the ASM instance level SQL>STARTUP RESTRICT; -- in the diskgroup level SQL>ALTER DISKGROUP DATA MOUNT RESTRICTED; -- check status of diskgroups SQL>SELECT NAME,STATE FROM V$ASM_DISKGROUP;
-- creating a tablespace in an ASM diskgroup CREATE TABLESPACE sample DATAFILE '+dgroup1';
Managing Disk Groups and Disks in ASM

Fundementals of Managing Disk Groups and Disks /* Basics */ SELECT PATH, MOUNT_STATUS FROM V$ASM_DISK; SQL> STARTUP NOMOUNT SQL> CREATE DISKGROUP dgroup1 NORMAL REDUNDANCY FAILGROUP controller1 DISK 'c:\asmdisks\asmdiska1' name firsta size 150M, 'c:\asmdisks\asmdiska2' FAILGROUP controller2 DISK 'c:\asmdisks\asmdiskb1', 'c:\asmdisks\asmdiskb2' / in the pfile: ASM_DISKGROUPS=dgroup1 SQL> select NAME, SECTOR_SIZE, BLOCK_SIZE, ALLOCATION_UNIT_SIZE, STATE, TYPE, TOTAL_MB from V$ASM_DISKGROUP restart -- rebalancing speed show parameter ASM_POWER_LIMIT ALTER DISKGROUP dgroup1 REBALANCE POWER 5;
-- adding extra disks ASMTOOL -create c:\asmdisks\asmdiska3 150 ASMTOOL -create c:\asmdisks\asmdiskb3 150 SQL> ALTER DISKGROUP dgroup1 ADD FAILGROUP controller1 DISK 'c:\asmdisks\asmdiska3' FAILGROUP controller2 DISK 'c:\asmdisks\asmdiskb3' /
Page 398
ALTER DISKGROUP dgroup1 REBALANCE POWER 10 /
/* Dropping a disk */ ALTER DISKGROUP group1 DROP DISK disk5; -- to cancel a pending drop: ALTER DISKGROUP group1 UNDROP DISKS; /* Dropping a Diskgroup */ in Oracle 10g: # the group must be mounted on exactly one node # non-empty dgroup DROP DISKGROUP DATA INCLUDING CONTENTS; # empty dgroup DROP DISKGROUP DATA; # If the diskgroup could not be mounted and the disks # were to be repurposed for non-ASM uses use dd command in unix # If the disks were to be repurposed as an ASM disk for a new diskgroup, # then use the FORCE option in the CREATE DISKGROUP in Oracle 11g: # use FORCE if the group is not mounted (try mount it first) DROP DISKGROUP DATA FORCE INCLUDING CONTENTS FORCE;
/* Checking Diskgroup */ -- 11g: you can validate the internal consistency of ASM diskgroup metadata. Summary of errors is logged in the ASM alert log file. -- check specific diskgroup with automatic repair SQL>ALTER DISKGROUP data CHECK; -- disable automatic repair SQL>ALTER DISKGROUP data CHECK NOREPAIR; SQL>ALTER DISKGROUP data CHECK REPAIR; Managing Disk Groups Attributes
Oracle Database 11g introduces a new concept called ASM attributes at the diskgroup level. The attributes for the diskgroup can be established at create diskgroup time or can be modified using the ALTER DISKGROUP command later. Following are the attributes you can set: Allocation unit (AU) sizes. The compatible.rdbms attribute. The compatible.asm attribute. disk_repair_time in units of minute (M) or hour (H) and is set by the ALTER DISKGROUP command The redundancy attribute for a specific template. The stripping attribute for a specific template.
All of the diskgroup attributes can be queried from the V$ASM_ATTRIBUTE view. Consider the following examples:
Page 399
CREATE DISKGROUP data disk '/dev/raw/raw1', ... attribute 'au_size' = '16M', 'compatible.asm' = '11.1'; ALTER DISKGROUP data SET ATTRIBUTE 'compatible.asm' = '11.1.0.0.0'; select NAME, VALUE from V$ASM_ATTRIBUTE where GROUP_NUMBER=1; Variable AU Sizes The default size of Allocation Unit (AU) is 1 MB which is sufficient for most regular databases. However, when you have databases with TB sizes, you will have enormous number of AUs. With Oracle 11g, AU size can be specified at diskgroup creation time to 1, 2, 4, 8, 16, 32, or 64MB in size. You can check the AU size through the following query: select NAME, ALLOCATION_UNIT_SIZE from V$ASM_DISKGROUP; Compatibility Settings Compatibility in ASM is controlled in three ways, as shown below: COMPATIBLE initialization parameter The compatible initialization parameter can be set for either ASM or the database instance. It takes one of the following values: 10.1, 10.2, or 11.1. Setting the initialization parameter to a lesser value than the software release will exclude availability of the new features introduced in the new release. This is a diskgroup-level compatibility and is specified by setting the COMPATIBLE.RDBMS attribute. This attribute determines the minimum COMPATIBLE database initialization parameter setting for any database instance that uses the disk group. Its default value is 10.1. This is a diskgroup-level compatibility and is specified by setting the COMPATIBLE.ASM attribute. It determines the minimum software version for an ASM instance that uses the disk group.
RDBMS Compatibility
ASM Compatibility
If you assign any of the compatibility setting to a higher value, you cannot later reverse it to a lower value. Following are some queries to obtain information about the compatibility settings: -- diskgroup compatibility setting select NAME, BLOCK_SIZE, ALLOCATION_UNIT_SIZE AU_SIZE, STATE, COMPATIBILITY ASM_COMP, DATABASE_COMPATIBILITY DB_COMP from V$ASM_DISKGROUP; -- Compatibility of the database clients that use the ASM select DB_NAME, STATUS,SOFTWARE_VERSION,COMPATIBLE_VERSION from V$ASM_CLIENT; ASM Fast Mirror Resync Any problems that make a failure group temporarily unavailable are considered transient failures that can be recovered by the ASM fast mirror resync feature. Disk path malfunctions; such as cable failures, host bus adapter failures, controller failures, or disk power supply interruptions; can cause transient failures. ASM fast resync keeps track of pending changes to extents on an OFFLINE disk during an outage. The extents are resynced when the disk is brought back online. Following are the steps to enable and handle this feature: -- diskgroup compatibility must be set to 11.1 ALTER DISKGROUP dg1 SET ATTRIBUTE 'compatible.asm' = '11.1'; ALTER DISKGROUP dg1 SET ATTRIBUTE 'compatible.rdbms'='11.1'; -- specify the duration of the disk_repair_time (default is 3.6 hour) ALTER DISKGROUP dg1 SET ATTRIBUTE 'disk_repair_time' = '5H'; -- in hours ALTER DISKGROUP dg1 SET ATTRIBUTE 'disk_repair_time' = '40M'; -- minutes -- verify the attribute settings
Page 400
select NAME, VALUE from V$ASM_ATTRIBUTE; -- if you get an offline disk because of a transient failure, you can see the -- remaining time left in SECONDS before ASM drops an offline disk select NAME, HEADER_STATUS, MOUNT_STATUS, MODE_STATUS, STATE, REPAIR_TIMER/60 from V$ASM_DISK WHERE GROUP_NUMBER=1; -- while the fix is in progress, if you want to reset the elapsed time, just take -- the disk(s) offline ALTER DISKGROUP dg1 OFFLINE DISK d3_0001; ALTER DISKGROUP dg1 OFFLINE DISKS IN FAILGROUP f2; -- you can also make a disk offline with a repair time different from its -- disk_repair_time attribute ALTER DISKGROUP dg1 OFFLINE DISK d3_0001 DROP AFTER 50m; -- disks in a failure group (f2) can also be taken offline ALTER DISKGROUP dg1 OFFLINE DISKS IN FAILGROUP f2 DROP AFTER 5m; -- if the disk needs to be dropped immediately and before the repair time has expired -- Note: ALTER DISKGROUP DROP DISK will not work ALTER DISKGROUP dg1 OFFLINE DISK D3_0001 DROP AFTER 0m; -- after the disk(s) are fixed, you can bring them online ALTER DISKGROUP dg1 ONLINE ALL; ALTER DISKGROUP dg1 ONLINE DISK d3_0001;
Monitoring Long-Running Operations

The ALTER DISKGROUP DROP, RESIZE, and REBALANCE commands return before the operation is complete. To monitor progress of these long-running operations, you can query the V$ASM_OPERATION fixed view. GROUP_NUMBERDisk group OPERATION STATE POWER ACTUAL SOFAR EST_WORK EST_RATE Type of operation: REBAL State of operation: QUEUED or RUNNING Power requested for this operation Power allocated to this operation Number of allocation units moved so far Estimated number of remaining allocation units Estimated number of allocation units moved per minute
EST_MINUTES Estimated amount of time (in minutes) for operation termination
Page 401
Migrating a Database to ASM

You can use OEM or RMAN as follows: 1. Shut down the database in a consistent mode by using the SHUTDOWN IMMEDIATE command. 2. Add the DB_CREATE_FILE_DEST and DB_CREATE_ONLINE_LOG_DEST_n parameters, as well as the new flash recovery area initialization parameters, DB_RECOVERY_FILE_DEST and DB_ RECOVERY_FILE_DEST_SIZE, to your database parameter file so you can use an OMFbased file system. Make sure that the two OMF parameters refer to the disk groups that you want to use in your ASM system. DB_CREATE_FILE_DEST = '+dgroup1' DB_RECOVERY_FILE_DEST = '+dgroup2' 3. Delete the control file parameter from the SPFILE, since Oracle will create new control files in the OMF file destinations by restoring them from the non-ASM database control files. 4. Start the database with the STARTUP NOMOUNT command: RMAN> CONNECT TARGET; RMAN> STARTUP NOMOUNT; 5. Restore the old control file in the new location, as shown here: RMAN> RESTORE CONTROLFILE from '/u01/orcl/oradata/control1.ctl'; 6. Mount the database: RMAN> ALTER DATABASE MOUNT; 7. Use the following command to copy your database files into an ASM disk group: RMAN> BACKUP AS COPY DATABASE FORMAT +dgroup1; 8. Use the SWITCH command to switch all datafiles into the ASM disk group: RMAN> SWITCH DATABASE TO COPY; At this point, all datafiles will be converted to the ASM type. You still have your original datafile copies on disk, which you can use to restore your database if necessary. 9. Open the database with the following command: RMAN> ALTER DATABASE OPEN; 10. For each redo log member, use the following command to move it to the ASM system: RMAN> SQL "alter database rename '/u01/test/log1' to '+dgroup1' "; 11. Archive the current online redo logs, and delete the old non-ASM redo logs. Since RMAN doesnt migrate temp files, you must manually create a temporary tablespace using the CREATE TEMPORARY TABLESPACE statement or: ALTER TABLESPACE TEMP ADD TEMPFILE ALTER DATABASE TEMPFILE '/u1/temp1' DROP
Moving a Tablespace to ASM

SQL>select name from v$datafile where ts# = (select x.ts# from v$tablespace x where x.name='HRSTBS'); set oracle_sid=ora11g
Page 402
rman target / sql'alter tablespace hrstbs offline'; BACKUP AS COPY TABLESPACE hrstbs FORMAT '+dgroup1'; SWITCH TABLESPACE hrstbs TO COPY; sql'alter tablespace hrstbs online';
Accessing an ASM instance from DB Console

/* Method 1 */ drop and recreate the em repository: emca -deconfig dbcontrol db -repos drop emca -config dbcontrol db -repos create /* Method 2 */ - Edit your listener.ora and add an entry for the ASM instance in the SID_LIST (SID_DESC = (GLOBAL_DBNAME = My_DB_SID) (ORACLE_HOME=d:\oracle\product\10.2.0\db_2) (SID_NAME = +ASM) ) - Restart your listener - Access DBConsole (e.g. with SYS as SYSDBA) - In the "Administration" or "Server" tab, click on the link called "Migrate to ASM" - Fill in the required fields and press continue (it wont do anything yet. Its just the first step, to get ASM listen as a new OEM target). - If it ever complains about ORA-1017, you need to recreate the ASM instance paswordfile. Take a backup of it first if you feel better so. The syntax for recreating the passwordfile can be, for example: orapwd file=D:\oracle\product\10.2.0\db_2\database\PWD+ASM.ora password=my_password force=y (of course, you must replace the password and the path with the one that serves your needs) The ASM target should have been created. Now, you only need to: click on the "Database" link on the right top of the DB Console window, or access the DB Console page again. You should see the ASM target link below the listener target. You only need to click the link and youll get the ASM home administration page, from DBConsole. If ASM hyperlink is not directly shown in the main DBconsole page, you can still click on the Host name link in the left of the DBConsole main page (just over the listener name link), and once there, again click in the "Targets"
Page 403
link, and lastly in the "+ASM_your_DB_hostname" link, which then will take you to the ASM Administration main page. You also have the option of directly typing the following URL, after the ASM target is created: http://your_DB_hostname:your_DBConsole_port/em/console/database/osm/osmSitemap ?type=osm_instance&target=%2BASM_your_DB_hostname&event=doLoad
Managing ASM Files

Types of ASM Filenames 1. Fully Qualified ASM Filenames (System Alias) You use this fully qualified name for referencing existing ASM files. Heres the syntax of an ASM file using a fully qualified filename: +group/dbname/file_type/tag.file.incarnation 2. Numeric ASM Filenames ASM derives numeric filenames from fully qualified ASM filenames and uses them to refer to existing files. +group.file.incarnation 3. Alias ASM Filenames You can use ASM alias files both when creating new ASM files and when referring to existing files. Alias ASM filenames mean that the files are not OMF-managed files. Thus, Oracle wont automatically remove these files when it does not have any further need for them. +dgroup1/myfiles/control_file1 +dgroup2/mydir/second.dbf 4. Incomplete ASM Filenames You can use an incomplete ASM filename only when creating files. +dgroup1 +dgroup1(datafile) Alias Filename Management Creating Disk Group Directories for Alias Filenames You must create a directory structure to support your alias filenaming conventions. ALTER DISKGROUP dgroup1 ADD DIRECTORY '+dgroup1/mydir'; Using Templates with Aliases dgroup(template_name)/alias +dgroup1(spfile)/config1 Adding Aliases You can add a filename alias or rename an existing alias name, using the ADD ALIAS or RENAME ALIAS clause of the ALTER DISKGROUP statement. ALTER DISKGROUP dgroup1 ADD ALIAS '+dgroup1/mydir/second.dbf' FOR '+dgroupA/sample/datafile/mytable.342.3' You can retrieve created aliases using v$ASM_ALIAS. The REFERENCE_INDEX column is usable only for entries that are directory entries in the alias directory. For non-directory entries, it equals to zero. Dropping Files and Aliases from a Disk Group ALTER DISKGROUP dgroup1 DROP FILE '+dgroup1/payroll/compensation.dbf'
Page 404
Using ASMCMD Utility

set oracle_home=c:\oracle\oradb11g set oracle_sid=+ASM asmcmd help
Oracle 11g introduces new commands in the asmcmd utility and it also provides backward compatibility with Oracle Database 10g ASM instances. Following are summary of some of them:
Command Syntax Description and Examples
lsct [-gH] [group] lsdg [-gcH] [group] lsdsk [-ksptagcHI] [-d diskg_roup_name] [pattern]
Lists information about current ASM clients. >lsct dgroup1 lists all diskgroups and their attributes. >lsdg dgroup2 lists the disks that are visible to ASM by scanning the disk headers of the disks seen by the value of the ASM_DISKSTRING >lsdsk -k -d DATA *_001 >lsdsk -s -d DATA *_001 >lsdsk -t -d DATA *_001 >lsdsk -c -t -d DATA *_001 >lsdsk -g -t -d DATA *_001
cp [-ifr] [connect_string:]src_fname [connect_string:]tgt_fname cp [-ifr] [connect_string:]src_fnameN, src_fnameN+1 ... [connect_string:]tgt_directory The connect_string is in the form of: user_name@host_name[.port_number].SID -i interactive -f force overwrite (aliases cannot be overwritten) -r recursive remap
Enables you to copy files between ASM disk groups on local instances to and from remote instances. >cp +dg1/vdb.ctf1 /backups/vdb.ctf1 >cp /home/oracle/encrypted.dmp +dg1 >cp vdb.ctf1 /tmp # the target ASM instance must be registered with the LISTENER >cp +DATA/DBA11g/DATAFILE/DOCS_D1.289.631914611 sys@rac1.+ASM:+DATA/DBA11g1/datafile/xxx
Repairs a range of physical blocks on disk (only blocks exhibiting read disk I/O errors are repaired) excluding those with corrupted contents. Internally, it reads the blocks from a good copy of an ASM mirror and rewrites them to an alternate location on disk, if the blocks on the original location cannot be properly read. remap <disk group name> <disk name> <block range> > remap DISK_GRP1 DATA_0001 5000-5999
Backing up and Restoring Diskgroup Metadata The md_backup command captures information about ASM disks, diskgroup and failure group configurations, and template and alias directory structures, and stores them in a user-designated backup text file. Following is the basic syntax of the command: md_backup [-b <backup_file_path> ] [-g diskgroup_name [-g diskgroup_name ]] Following is an example of using the command: md_backup b /tmp/asm_backup.mdb -g dg1 g dg2 If the backup file already exists, you should remove it before issuing the command. If you issue the md_backup command without any option, it creates a file named as ambr_backup_intermediate_file which contains the metadata information of all the mounted diskgroups. Page 405
The md_restore command reads the backup file and restores a disk group. You can set its options to build a script file that contains the SQL statements required to rebuild the ASM components from the backup file. Following is the syntax of the command and description of its switches: md_restore -b <backup_file> [-li] [-t (full)|nodg|newdg] [-f <sql_script_file>] [-g '<diskgroup_name>,<diskgroup_name>,...'] [-o '<old_diskgroup_name>:<new_diskgroup_name>,...'] -t type of restore. full tag specifies that all the diskgroups should be re-created using the same configuration from the MDB backup file. nodg restore metadata only and skip the diskgroup creation. newdg create disk group with a different name and restore metadata; -o is required. This tag is used to allow the user to change diskgroup name. -f -o -i -l write SQL commands to <sql_script_file> instead of executing them. override option is used only with the newdg option to remap the diskgroup name, disk name, paths, and failure groups. ignore errors. By default, the command aborts when it encounters an error. log all messages to a log file.
Following are examples of using the command:
/* General Examples */ # To perform a restore of the dg1 diskgroup from the MDB backup file, use this: md_restore b /tmp/backupfile t full g dg1 -i # To just restore the metadata for the dg1 diskgroup (the diskgroup already exists). md_restore b /tmp/backupfile t nodg g dg1 i # To create a different diskgroup name: md_restore b /tmp/backupfile t newdg -o "DGNAME=dg1:dg3" i # To apply the override options as specified in the dg_over.txt file and restore # from the backup file: md_restore b /tmp/backupfile t newdg of /tmp/dg_override.txt i /* Scenario Example */ 1. Back up a tablespace exising in a disk group: RMAN> BACKUP TABLESPACE users; 2. Create a directory named test in the disk group DGROUPA. Also create an alias called +DGROUPA/test/users.f that points to the ASM datafile that contains the users tablespace: ASMCMD> mkdir +DGROUPA/test ASMCMD> mkalias TBSSRA.123.123456789 +DGROUPA/test/users.f 3. Back up the metadata for the disk group DGROUPA using the md_backup command: ASMCMD> md_backup g dgroupA The md_backup command stores the backup metadata in the text file named ambr_backup_ intermediate in the current directory. 4. Simulate a disk failure by dropping the disk group DGROUPA: SQL> ALTER DISKGROUP dgroup1 DISMOUNT FORCE; SQL> DROP DIKSGROUP dgroup1 FORCE INCLUDING CONTENTS; Oracle DBA Code Examples
Page 406
The DISMOUNT FORCE clause in the ALTER DISKGROUP command dismounts the disk group and force drops it. 5. Execute the md_restore command to restore the ASM metadata for the dropped disk group: ASMCMD> md_restore b ambr_backup_intermediate_file t full g data 6. Using the backup of the users tablespace from step 1, restore the users tablespace: RMAN>RESTORE TABLESAPCE users; 7. Exit from RMAN once the restore is completed.
Note that md_backup is a backup of the metadata of the ASM instance. The data is being backed up by RMAN. After the diskgroup is created, along with all the directories, you can restore the RMAN backup to the diskgroup. Bad Block Recovery If ASM cannot read a physical block from a disk, it considers that the block has IO error. In this case, ASM will automatically read a mirrored block and write a relocated copy to produce successful copy. However, you can manually repair blocks that have read disk I/O errors using the remap command. Following is the syntax of the command: remap <diskgroup name> <disk name> <block range>
Using SYSASM Privilege and OSASM Group

This feature introduces a new SYSASM privilege that is specifically intended for performing ASM administration tasks. Using the SYSASM privilege instead of the SYSDBA privilege provides a clearer division of responsibility between ASM administration and database administration. Following are code examples illustrating how to use this privilege: -- grant the privilege GRANT SYSASM TO firas; -- check the granted privilege SELECT * FROM V$PWFILE_USERS; -- ASM management commands are available to Adam CONNECT firas/his_password ALTER DISKGROUP dg1 DISMOUNT; ALTER DISKGROUP dg2 MOUNT; .. and so on. Be aware that users with SYSOPER privilege have some ASM privileges. Following table shows available and restricted ASM privilege for users with SYSOPER privilege:
Avaiable ASM Privilege Restricted ASM Privilege
STARTUP AND SHUTDOWN ALTER DISKGROUP MOUNT ALTER DISKGROUP DISMOUNT ALTER DISKGROUP ONLINE DISK ALTER DISKGROUP OFFLINE DISK ALTER DISKGROUP REBALANCE ALTER DISKGROUP CHECK
CREATE DISKGROUP / DISK DROP DISKGROUPS / DISKS ALTER DISKGROUP / DISK RESIZE
OSASM is a new operating system group that is used exclusively for ASM. Members of the OSASM group can connect as SYSASM using operating system authentication and have full access to ASM.
Page 407
Manually Upgrading Oracle AS from 10g to 11g

Following are the steps you follow to upgrade an existing Oracle 10g ASM to 11g: 1. Install the Oracle Database 11g software to a new ORACLE_HOME directory. 2. Update the location. /etc/oratab or /var/opt/oracle/oratab file with the new ASM ORACLE_HOME
3. Copy the ASM initialization file from the old ORACLE_HOME to the new one. 4. Edit any directory-based parameters (such as diag and dump) in the ASM initialization file as required. 5. If you are upgrading a non-RAC ASM instance, you should reconfigure the Oracle CSS using the new ORACLE_HOME. You can do this by executing the localconfig command from the new home. Once the CSS configuration is complete, you need to change your ORACLE_HOME to the new Oracle version 11.1 ORACLE_HOME and start the ASM instance. cd $ORACLE_HOME/bin # ./localconfig reset 6. If you are upgrading a ASM instance in a RAC environments, you can modify the new ASM home within the OCR using the srvctl utility as follows: srvctl modify asm -n racnode1 -i +ASM1 -o /apps/oracle/product/11.1.0/asm -p init+ASM1.ora 7. Grant the SYSASM role to the SYS GRANT SYSASM to sys; 8. If you have obsolete initialization parameters, you can address them now. To get a listing of all the obsolete initialization parameters, refer to the ASM alert log file.
Verifying Manually ASM Device

# To verify asmlib status: ## Both should be [OK] /etc/init.d/oracleasm status Checking if ASM is loaded: Checking if /dev/oracleasm is mounted:
[ [ OK ]
OK
# To verify user setting in asmlib: uid=1001(grid) gid=1000(oinstall) groups=1000(oinstall) /usr/sbin/oracleasm configure ORACLEASM_ENABLED=true ORACLEASM_UID=grid ORACLEASM_GID=oinstall ORACLEASM_SCANBOOT=true ORACLEASM_SCANORDER="" ORACLEASM_SCANEXCLUDE=""
# To verify disk /etc/init.d/oracleasm listdisks DISKCLU ls -l /dev/oracleasm/disks .. brw-rw---1 grid oinstall
8,
33 Sep 16 09:41 DISKCLU
Page 408
# Disk DISKCLU is available and readable from above output. dd if=/dev/oracleasm/disks/DISKCLU of=/dev/null bs=1024k count=1 1+0 records in 1+0 records out
Page 409
Part 7
Oracle Real Application Cluster
Page 410
Oracle RAC Possible Installation Configurations

For Oracle Software homes, Voting and OCR, db files: NTFS_RAW_ASM NTFS_RAW_RAW NTFS_OCFS_OCFS OCFS_OCFS_OCFS
Page 411
Installing Oracle 10g R2 RAC on Enterprise Linux 4

Note: The metalink document RAC Starter Kit and Best Practices (Linux) [ID 811306.1] is a good source reference for this task.
Emulation software: VMWare Workstation ACE Edition 6.0.5 or VMWare Server 2. RAC Nodes: 2 nodes with 2 GB RAM each , 2 ethernet cards. OS: Oracle Linux Enterprise 4.5 for x86: kernel 2.6.9
Required Software
Oracle 10g R2 Clusterware for Linux x86 32-bit Oracle Database 10g Release 2 for Linux x86 32-bit
Used Hardware
In the VMWare: create one virtual machine (rac1) with the following specs: o 2 GB RAM o Two ethernet cards: both can be configured as bridged or host-only in VMware. o One local hardisk with 20 GB o CPU Count: 2 o Create a folder in the same directory structure level as the parent folder containing the created virtual machine. Give it a meaningful name like 'shared_disks'. Create in that folder three disks of 10 GB and two of 512 MB. All of them are of LSI Logic type and SCISI Persistent. Make sure they are on SCISI controller different from the SCSI controller of the local hardisk. For example, if the SCSI controller of local hardisk is SCISI0, make those disks on controller SCSI1.
Installation Plan
1. Preinstallation tasks: o o o 2. 3. 4. 5. 6. 7.
Page 412
Hardware requirements Software requirements Environment configuration
Oracle Clusterware installation Oracle Database 10g Software Installation Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software Install EM Agent in cluster nodes (if required) Configure Listeners Perform ASM installation Oracle DBA Code Examples
8. 9.
Perform cluster database creation Complete postinstallation tasks
10. Useful postinstallation tasks
1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Give the first ethernet card IP 192.168.4.11 and the second 192.168.0.11 and the hostname rac1.mydomain.com. Define a gateway. If it does not exist, make it same as the host IP address. o Insall the following packages: Desktop Environments o GNOME Desktop Environment Desktop o X Window System o Gnome
Page 413
Applications o Graphical Internet (optional)
Servers o Do not select anything in this group.
Development o Development Tools
System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
Miscellaneous o Do not select anything in this group.
Complete the installation Install further packages: # to know distribution and version of Linux cat /etc/issue # to know kernel version (and its errata level) uname -r # from CD 3 rpm -Uvh libaio* rpm -Uvh openmotif21-2.1.30-11.RHEL4.6.i386.rpm rpm -Uvh openmotif-2.2.3-10.1.el4.i386.rpm # those packages downloaded from http://rpm.pbone.net rpm -e compat-libstdc++-296-2.96-132.7.2 rpm -Uvh compat-libstdc++-7.3-2.96.128.i386.rpm rpm -Uvh compat-libstdc++-devel-7.3-2.96.128.i386.rpm rpm -Uvh compat-gcc-7.3-2.96.128.i386.rpm rpm -Uvh compat-gcc-c++-7.3-2.96.128.i386.rpm # confirm the required packages are installed: rpm -qa|grep gccrpm -qa|grep glibcrpm -qa|grep compat-dbrpm -qa|grep compat-gccrpm -qa|grep compat-gcc-c++rpm -qa|grep compat-libstdc++rpm -qa|grep compat-libstdc++-develrpm -qa|grep control-center-2.8.0 rpm -qa|grep openmotif21rpm -qa|grep setarch# SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Oracle DBA Code Examples
Page 414
# Install ASMLib 2.0 packages # install the library for your kernel and CPU type # oracleasm-*-version.cpu_type.rpm # check installed packages rpm -qa|grep asm # install the packages from CD3 rpm -Uhv oracleasm-support-2.0.3-2.i386.rpm rpm -Uhv oracleasm-2.6.9-55.0.0.0.2.EL-2.0.3-2.i686.rpm rpm -Uhv oracleasm-2.6.9-55.0.0.0.2.ELsmp-2.0.3-2.i686.rpm # download the package Userspace Library # from http://otn.oracle.com/software/tech/linux/asmlib/files/RPMS/rhel4/x86/2.0.4/or acleasmlib-2.0.4-1.el4.i386.rpm rpm -Uvh oracleasmlib-2.0.4-1.el4.i386.rpm Check the hardware requirements # Hardware Requirements (in cluster nodes) # At least 1 GB of physical memory grep MemTotal /proc/meminfo # swap space: twice the amount of physical memory grep SwapTotal /proc/meminfo # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 400 MB disk space in /tmp df -k /tmp # 4 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: # df -h /dev/shm/ Create the required network configuration (rac2 will be created later): # Network names Resolution # configure /etc/hosts if no domain server is used (both nodes) vi /etc/hosts 127.0.0.1 localhost.localdomain localhost #Public 192.168.4.11 rac1.mydomain.com rac1 192.168.4.12 rac2.mydomain.com rac2 #VIP 192.168.4.13 rac1-vip.mydomain.com rac1-vip 192.168.4.14 rac2-vip.mydomain.com rac2-vip #Inter-connect 192.168.0.11 rac1-priv.mydomain.com rac1-priv 192.168.0.12 rac2-priv.mydomain.com rac2-priv Oracle DBA Code Examples
Page 415
Note: To prevent network hangs with failovers from public to virtual IP addresses with RAC databases using NAS devices or NFS mounts, enter the following command as root to enable the Name Service Cache Daemon: /sbin/service nscd start Create and configure the required OS users and groups Note: userid and groupid must be the same in all nodes. You can check them by id oracle command. # all group and user ids on all the nodes must have identical id # inventory and OSDBA groups (if needed, use -g <number> to specify the id) # inventory group groupadd -g 501 oinstall groupadd -g 502 dba # oracle software owner user (take note of userid) /usr/sbin/useradd -u 200 -g oinstall -G dba oracle passwd oracle # make sure nobody user exists (if not there, create it useradd nobody) id nobody # The oracle User Environment # in /home/oracle/.bash_profile # export DISPLAY if required export ORACLE_BASE=/u01/app/oracle if [ $USER = "oracle" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi export EDITOR=vi export ORACLE_HOME=$ORACLE_BASE/product/10.2.0/db_1 export ORA_CRS_HOME=/u01/crs export ORACLE_PATH=$ORACLE_BASE/common/oracle/sql:.:$ORACLE_HOME/rdbms/admin export ORACLE_SID=rac1 export NLS_LANG=AMERICAN_AMERICA.WE8ISO8859P1 export NLS_DATE_FORMAT="mm/dd/yyyy hh24:mi:ss" export PATH=.:${PATH}:$HOME/bin:$ORACLE_HOME/bin:$ORA_CRS_HOME/bin export PATH=${PATH}:/usr/bin:/bin:/usr/bin/X11:/usr/local/bin export PATH=${PATH}:$ORACLE_BASE/common/oracle/bin export ORACLE_TERM=xterm export TNS_ADMIN=$ORACLE_HOME/network/admin export ORA_NLS10=$ORACLE_HOME/nls/data export TNS_ADMIN=$ORACLE_HOME/network/admin export ORA_NLS10=$ORACLE_HOME/nls/data export LD_LIBRARY_PATH=$ORACLE_HOME/lib export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:$ORACLE_HOME/oracm/lib export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/lib:/usr/lib:/usr/local/lib export CLASSPATH=$ORACLE_HOME/JRE export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/jlib export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/rdbms/jlib export CLASSPATH=${CLASSPATH}:$ORACLE_HOME/network/jlib export THREADS_FLAG=native # should NOT be on a shared disk export TEMP=/tmp export TMPDIR=/tmp
Page 416
Configure kernel parameters and shell limits Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Configuring kernel parameters and shell limits # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf oracle soft nproc 2047 oracle hard nproc 16384 oracle soft nofile 1024 oracle hard nofile 65536 oracle soft memlock 3145728 oracle hard memlock 3145728 vi /etc/pam.d/login session required /lib/security/pam_limits.so # they can be tuned for a production db # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf kernel.sem = 250 32000 100 128 # It should be equal to or larger than the largest SGA (max 4 GB) kernel.shmmax = 536870912 net.ipv4.ip_local_port_range = 1024 65000 net.core.rmem_default = 4194304 net.core.rmem_max = 4194304 net.core.wmem_default = 262144 net.core.wmem_max = 262144 # to take immediate effect /sbin/sysctl -p
Configure hangcheck-timer kernel module: # check hangcheck-timer Module Configuration # with this module, if the kernel hangs, the machine will reboot # verify the module is loaded /sbin/lsmod | grep -i hang # if not loaded, load it vi /etc/modprobe.conf options hangcheck-timer hangcheck_tick=30 hangcheck_margin=180 # execute and add in the file vi /etc/rc.local /sbin/modprobe hangcheck-timer
Partition the disks and prepare the raw disks # Partition the devices # for the disks /dev/sdb .. /dev/sdf fdisk /dev/sdb # answers: "n", "p", "1", "Return", "Return", "p" and "w" Note: if the following message appears after the "w" command: WARNING: Re-reading the partition table failed with error 16: Device or resource busy. then, you can avoid restarting the machine by the following command: partprobe
Page 417
# to make sure partions are created ls -lX /dev/sd* brw-r----- 1 root disk 8, 0 Dec 1 brw-r----- 1 root disk 8, 1 Dec 1 brw-r----- 1 root disk 8, 2 Dec 1 brw-r----- 1 root disk 8, 16 Dec 1 brw-r----- 1 root disk 8, 17 Dec 1 brw-r----- 1 root disk 8, 32 Dec 1 brw-r----- 1 root disk 8, 33 Dec 1 brw-r----- 1 root disk 8, 48 Dec 1 brw-r----- 1 root disk 8, 49 Dec 1 brw-r----- 1 root disk 8, 64 Dec 1 brw-r----- 1 root disk 8, 65 Dec 1 brw-r----- 1 root disk 8, 80 Dec 1 brw-r----- 1 root disk 8, 81 Dec 1 # binding sdisks to raw devices raw /dev/raw/raw1 /dev/sdb1 raw /dev/raw/raw2 /dev/sdc1 # list the raw devices raw -qa # add to the file vi /etc/sysconfig/rawdevices /dev/raw/raw1 /dev/sdb1 /dev/raw/raw2 /dev/sdc1
20:35 20:36 20:35 21:55 21:55 21:57 21:57 21:57 21:57 21:58 21:58 21:58 21:58
/dev/sda /dev/sda1 /dev/sda2 /dev/sdb /dev/sdb1 /dev/sdc /dev/sdc1 /dev/sdd /dev/sdd1 /dev/sde /dev/sde1 /dev/sdf /dev/sdf1
# Adjust the raw devices permission settings: # Run the following commands AND add them the /etc/rc.local file: # will be used by OCR chown root:oinstall /dev/raw/raw1 chmod 660 /dev/raw/raw1 # will be used by voting disk chown oracle:oinstall /dev/raw/raw2 chmod 644 /dev/raw/raw2 # will be used by ASM, if it is using raw device (NOT IN THIS CONFIGURATION) # chown oracle:oinstall /dev/raw/raw3 # chmod 660 /dev/raw/raw3 # start the service service rawdevices restart Create the required directories for the Oracle clusterware and database software # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab # in the example above, /u01 should be owned by the root user # and writable by group oinstall # directories must be same in all nodes: mkdir -p /u01/app/oracle/product/10.2.0/db_1 mkdir /u01/stage10g # clusterware must NOT be subdirectory of the ORACLE_BASE mkdir /u01/crs chown -R oracle:oinstall /u01/app/oracle
Page 418
chmod chown chown chmod
-R 775 /u01/app/oracle oracle:oinstall /u01/stage10g -R oracle:oinstall /u01/crs -R 775 /u01/crs
Disable screensavers on host & guest machines. o In Oracle Linux: Applications-> Preferences-> Screen Saver-> Mode: Disable Screen Saver o Do the same after logging off and logging on again as oracle user.
Shutdown rac1 Edit the VMware file (with vmx extensions) and add the following entry to allow sharing the disks (make sure the scsi controller number is the one you used): disk.locking = "FALSE" diskLib.dataCacheMaxSize = "0" diskLib.dataCacheMaxReadAheadSize = "0" diskLib.dataCacheMinReadAheadSize = "0" diskLib.dataCachePageSize = "4096" scsi1.sharedBus = "virtual" scsi1:0.deviceType = "disk" scsi1:1.deviceType = "disk" scsi1:2.deviceType = "disk" scsi1:3.deviceType = "disk" scsi1:4.deviceType = "disk" scsi1:5.deviceType = "disk"
Copy the folder containing rac1 into a new folder in the same directory structure level. Let's name it "rac2". This will be the second node in the cluster. Edit the VMware file of rac1 and edit the following: displayName = "rac2" Open rac2 then probe its network cards, change it's IP addresses: IP 192.168.4.12 and 192.168.0.12. It's hostname to rac2.mydomain.com. You can use system-config-networkgui. Activate the network and reboot the vm. In rac2, perform: # change the variable in the file vi /home/oracle/.bash_profile export ORACLE_SID=rac2
Start rac1 Configure ASM drivers: # as root ( on ALL NODES ) service oracleasm configure Default user to own the driver interface []: oracle Default group to own the driver interface []: dba Start Oracle ASM library driver on boot (y/n) [n]: y Fix permissions of Oracle ASM disks on boot (y/n) [y]: y Writing Oracle ASM library driver configuration: Creating /dev/oracleasm mount point: Loading module "oracleasm": Mounting ASMlib driver filesystem: Scanning system for ASM disks: # As the root user on node node1 (ONLY ONE NODE) service oracleasm createdisk DISK1 /dev/sdd1
[ [ [ [ [
OK OK OK OK OK
] ] ] ] ]
Page 419
service oracleasm createdisk DISK2 /dev/sde1 service oracleasm createdisk DISK3 /dev/sdf1 # if any of the commands above fails, you can delete the disk service oracleasm deletedisk DISK1 # in ALL NODES service oracleasm scandisks service oracleasm listdisks Configure SSH in all the nodes:
Note: scp and ssh must be located in the path /usr/local/bin. If not, then create a symbolic link in /usr/local/bin to the location where scp and ssh are found. # Configuring SSH for Remote Installation # make sure the ssh is running # the following command shoudl return ssh process id pgrep sshd vi /etc/hosts.equiv # add node names to /etc/hosts.equiv (all nodes) rac1 rac2 # as oracle in node1 # whenever prompted for a passphrase leave it empty and press return mkdir ~/.ssh chmod 700 ~/.ssh /usr/bin/ssh-keygen -t rsa # as oracle in node2 mkdir ~/.ssh chmod 700 ~/.ssh /usr/bin/ssh-keygen -t rsa # as oracle on node1 cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys ssh rac2 cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys scp ~/.ssh/authorized_keys rac2:/home/oracle/.ssh/ # Perform the following as the oracle user on node1 and then node2 # copy paste the following command in both nodes twice: ssh rac1 date;ssh rac2 date;ssh rac1.mydomain.com date;ssh rac2.mydomain.com date;ssh rac1-priv date;ssh rac2-priv date;ssh rac1-priv.mydomain.com date;ssh rac2-priv.mydomain.com date;ssh localhost.localdomain date;ssh localhost date # Note: the keys are session dependent. it's gone from memory on log off # Execute (to load the keys in the memory): exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add 2. Oracle Clusterware installation # make sure system time of rac2 should be behind the system time of rac1 date; ssh rac2 date # Identify the shared disks to use # /dev/sdb & /dev/sdc will be used for OCR & Voting by the Cluster software /sbin/fdisk l # unzip the Oracle 10g R2 Clusterware for Linux in the stage folder (rac1) # If you plan to use Oracle Clusterware on x86 (64-bit) (but not on # Linux Itanium 64-bit), then you must run the rootpre.sh on all nodes
Page 420
su cd /u01/stage10g/clusterware/rootpre ./rootpre.sh # install cluvfy (in all nodes) # in node1 su cd /u01/stage10g/clusterware/rpm rpm -q cvuqdisk-1.0.1 export CVUQDISK_GRP=dba rpm -ivh cvuqdisk-1.0.1-1.rpm # in node2 su export CVUQDISK_GRP=dba rpm -ivh cvuqdisk-1.0.1-1.rpm # Verify Cluster Setup with cluvfy # as oracle # if the user equivalency is not loaded exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add # then proceed: cd /u01/stage10g/clusterware/cluvfy export CV_NODE_ALL=rac1,rac2 ./runcluvfy.sh stage -pre crsinst -n rac1,rac2 -verbose # If only the VIP was the problem, then you're going in the right way!
# lunch OUI from the clusterware ( as oracle from node1) # if logged in in Genome using another user, log out and log in as oracle # to reload the user equivalency, if logged off: (as oracle) exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add # if not turned off, turn off the screensaver IN ALL NODES # then process: cd /u01/stage10g/clusterware ./runInstaller
>Inventory Directory (displayed only in first time) /u01/app/oracle/oraInventory oinstall >Home Details /u01/crs >Product-Specific Prerequisite Checks They all should succeed >Specify Cluster Configuration Select 'Add' and add the details for node rac2. Enter all details. 'OK' to continue. rac2.mydomain.com rac2-priv.mydomain.com rac2-vip.mydomain.com Click 'Next'
Page 421
>Specify Network Interface Usage Change the eth0 Interface Type to Public. 'Ok' to continue >Specify Oracle Cluster Registry (OCR) Location Select External Redundancy and specify the first raw device /dev/raw/raw1 as the OCR location 'Next' to continue. >Specify Voting Disk Location Select External Redundancy and specify the first raw device /dev/raw/raw2 as the voting disk location 'Next' to continue. >Summary Review the summary and select 'Install' to start the installation >Configuration Scripts as the root user on each node, run the scripts. Do not run the scripts simultaneously on both nodes. if there is an error in scripts execution, it's reported in /u01/ /crs/log/<hostname> output of running the script in node1: Checking to see if Oracle CRS stack is already configured /etc/oracle does not exist. Creating it now. Setting the permissions on OCR backup directory Setting up NS directories Oracle Cluster Registry configuration upgraded successfully assigning default hostname rac1 for node 1. assigning default hostname rac2 for node 2. Successfully accumulated necessary OCR keys. Using ports: CSS=49895 CRS=49896 EVMC=49898 and EVMR=49897. node <nodenumber>: <nodename> <private interconnect name> <hostname> node 1: rac1 rac1-priv rac1 node 2: rac2 rac2-priv rac2 Creating OCR keys for user 'root', privgrp 'root'.. Operation successful. Now formatting voting device: /dev/raw/raw2 Format of 1 voting devices complete. Startup will be queued to init within 90 seconds. Adding daemons to inittab Expecting the CRS daemons to be up within 600 seconds. CSS is active on these nodes. rac1 CSS is inactive on these nodes. rac2 Local node checking complete. Run root.sh on remaining nodes to start CRS daemons. output of running the script in node2: Checking to see if Oracle CRS stack is already configured /etc/oracle does not exist. Creating it now. Setting the permissions on OCR backup directory Setting up NS directories
Page 422
Oracle Cluster Registry configuration upgraded successfully clscfg: EXISTING configuration version 3 detected. clscfg: version 3 is 10G Release 2. assigning default hostname rac1 for node 1. assigning default hostname rac2 for node 2. Successfully accumulated necessary OCR keys. Using ports: CSS=49895 CRS=49896 EVMC=49898 and EVMR=49897. node <nodenumber>: <nodename> <private interconnect name> <hostname> node 1: rac1 rac1-priv rac1 node 2: rac2 rac2-priv rac2 clscfg: Arguments check out successfully. NO KEYS WERE WRITTEN. Supply -force parameter to override. -force is destructive and will destroy any previous cluster configuration. Oracle Cluster Registry for cluster has already been initialized Startup will be queued to init within 90 seconds. Adding daemons to inittab Expecting the CRS daemons to be up within 600 seconds. CSS is active on these nodes. rac1 rac2 CSS is active on all nodes. Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Waiting for the Oracle CRSD and EVMD to start Oracle CRS stack installed and running under init(1M) Running vipca(silent) for configuring nodeapps The given interface(s), "eth0" is not public. Public interfaces should be used to configure virtual IPs.
# you MUST run vipca in the LAST NODE (NODE2) before clicking OK in OUI # from $ORA_CRS_HOME/bin cd /u01/crs/bin ./vipca # in vipca, click on vip address to auto fill rac1 rac1-vip.mydomain.com 192.168.4.13 rac2 rac2-vip.mydomain.com 192.168.4.14 When the configuration is complete, the final result report will be: Configuration ResultsThe VIP Configuration Assistant has successfully created resource aplications for each cluster node Click 'Exit' Click 'OK' in the OUI. All the checks should succeed. Click 'Exit' # To verify the installation: -- 1 cd /u01/stage10g/clusterware/cluvfy ./runcluvfy.sh stage -post crsinst -n rac1,rac2 -verbose
Page 423
-- 2 ping rac1-vip ping rac2-vip -- 3 /u01/crs/bin/crsctl check crs # To Avoid Node Eviction # In ALL NODES as root: (IMPORTANT) vi /u01/crs/install/rootconfig At line 356, change CLSCFG_MISCNT="-misscount 60" to CLSCFG_MISCNT="-misscount 300" # and then by the command crsctl set css misscount 300 # check status of crs daemon processes (ALL NODES) ./crs_stat -t Name Type Target State Host -----------------------------------------------------------ora.rac1.gsd application ONLINE ONLINE rac1 ora.rac1.ons application ONLINE ONLINE rac1 ora.rac1.vip application ONLINE ONLINE rac1 ora.rac2.gsd application ONLINE ONLINE rac2 ora.rac2.ons application ONLINE ONLINE rac2 ora.rac2.vip application ONLINE ONLINE rac2 # if State of any of them is UNKNOWN, try restarting the deamons ./crsctl stop crs # then start again ./crsctl start crs # to check their status (keep checking till all damons are up) ./crsctl check crs # when they are all up, the output should be: CSS appears healthy CRS appears healthy EVM appears healthy 3. Oracle Database 10g Software Installation # make sure all clusterware processes are running /u01/crs/bin/crs_stat -t # you can restart them /u01/crs/bin/crsctl stop crs /u01/crs/bin/crsctl start crs # If not loaded, to load the keys in the memory: exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add -- extract DB software in /u01/stage10g/database in node1 -- as oracle mkdir /u01/stage10g/database -- start OUI cd /u01/stage10g/database ./runInstaller Welcome 'Next' to skip the Welcome screen
Page 424
Select Installation Type 'Next' to continue. Install Location Keep the default Oracle Base location /u01/app/oracle and default Oracle Home location /u01/app/oracle/product/10.2.0/db_1 'Next' to continue. Specify Hardware Cluster Installation Mode Keep the default 'Cluster Installation' selection and select both nodes. 'Next' to continue. Product-Specific Prerequisite Checks The OUI will now verify that the environment meets all the requirements. All pre-requisite steps should complete successfully. Select 'Next' to continue. Select Configuration Option Select 'Install Software Only'. 'Next' to continue. Privileged Operating System Groups Keep the default options dba, oinstall. 'Next' to continue Summary Review the summary and select 'Install' to start the installation. Configuration Script Once the installation is complete you will be prompted to run a script as the root user. Open a terminal window and execute the script as the root user on each node. Select 'Ok' to continue after the script has been run successfully on both nodes. End of Installation Once the installation is complete select 'Exit' to complete the installation and exit the OUI. 4. Apply Patchset 3 (10.2.0.4) for Clusterware and Database Software Note: This is the patchset applied for this environment. Generally speaking, if there is a newer version, use it instead of this version. Note: This patch includes the Oracle Clusterware Process Monitor Daemon (oprocd) which when it detects a system hang, it restarts the hung node. This may restart a node under heavy workload. Check Oracle Clusteware Installation documenation for more information. # extract 10g Release 2 (10.2.0.4) Patch Set 3 for Linux x86 to /u01/stage10g/patch10.2.0.4/Disk1 mkdir /u01/stage10g/patch10.2.0.4 # If not loaded, to load the keys in the memory: exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add # you apply the patch on (1) clusterware then on (2) database software
Page 425
# (1) Apply the patch on clusterware # as oracle in rac1 cd /u01/stage10g/Patch10.2.0.4 ./runInstaller Select CRS Home then ->Next Cluster Info displayed -> Next Checks -> Next ->Install After Installation is complete it will ask to apply a script on the nodes oneby-one. Following is a sample of the message that appears: To complete the installation of this patchset, you must perform the following tasks on each node: 1. Log in as the root user. 2. As the root user, perform the following tasks: a. Shutdown the CRS daemons by issuing the following command: /u01/crs/bin/crsctl stop crs b. Run the shell script located at: /u01/crs/install/root102.sh This script will automatically start the CRS daemons on the patched node upon completion. 3. After completing this procedure, proceed to the next node and repeat. # following is the implementation of the above su cd /u01/crs/bin ./crsctl stop crs cd /u01/crs/install ./root102.sh cd /u01/crs/bin ./crsctl query crs softwareversion # REPEAT IN NODE2 # then check status of deamons ./crs_stat -t # (2) Apply the patch on DB software # as oracle in node1 cd /u01/stage10g/patch10.2.0.4/Disk1 ./runInstaller Select path of ORACLE_HOME ->Next After Installation is complete it will ask to apply a script on both the nodes one-by-one. 5. Install EM Agent in cluster nodes (if required)
6. Configure Listeners # Run netca (not netmanager which is not clusteraware) from any node. # as oracle cd $ORACLE_HOME/bin ./netca & Select Cluster configuration and select both nodes. Configure Listener Then Configure Naming Methods: Local and Easy Connect
Page 426
After installing listener check with below commands whether listener working properly. # as root /u01/crs/bin/crs_stat -t 7. Perform ASM installation # as oracle (in node1) cd /u01/app/oracle/product/10.2.0/db_1/bin dbca & Welcome Keep the default selection Oracle RAC database. 'Next' to continue. Operations Select Configure ASM. 'Next' to continue. Node Selection Select ALL the nodes and 'Next' to continue. Create ASM Instance Select a SYS password for the ASM instance. Select IFILE parameter file to create. 'Next' to continue. Select OK to confirm creation of the ASM instances. ASM Disk Groups Select Create New to create new ASM disk groups. Enter dg1 as the first disk group name. Keep the default redundancy settings (Normal) and select the 2 data disks DISK1 and DISK2. 'Ok' to continue Select Create New again to add another disk group. Enter dg2 as the disk group name. This time select External for redundancy and select the remaining disk DISK3. This group will be used as the recovery arae 'OK' to continue. All the disk groups are now created. Finish to complete. # An ASM instance will be created in every node named as ASMn export ORACLE_SID=ASM1 sqlplus /nolog SQL>conn / as sysdba select name from v$asm_diskgroup ; ... # after checking, return ORACLE_SID to its original value export ORACLE_SID=rac1 8. Perform cluster database creation # If not loaded, to load the keys in the memory: exec /usr/bin/ssh-agent $SHELL /usr/bin/ssh-add # as oracle (in node1) cd /u01/app/oracle/product/10.2.0/db_1/bin dbca &
Page 427
Create Oracle RAC Database 'Next' to continue Operations Select Create a Database. 'Next' to continue. Node Select all the nodes 'Next' to continue. Database Templates Select the required template 'Next' to continue. Database Identification Enter rac as the global database name. 'Next' to continue. Management Options Keep the default settings Configure Enterprise Manager and Configure Database Control for local management selected, but Enable Alert Notifications and Enable Daily Disk Backup to Recovery Area deselected. 'Next' to continue. Database Credentials Select Use the same Administrative Password for All Accounts and enter a password. 'Next' to continue. Storage Options Select ASM for storage. 'Next' to continue.
ASM Disk Groups Select both disk groups dg1 and dg2. 'Next' to continue. Database File Locations Keep the default Use Oracle-Managed Files. Make sure +DG1 is entered as the Database Area. 'Next' to continue Recovery Configuration: Select Specify Flash Recovery Area and enter +DG2 as the Flash Recovery Area. Set its size. Select Enable archiving and click on the Edit Archive Mode Parameters button and make sure the Flash Recovery area is pointing to +DG2. 'OK' and then 'Next' to continue. Database Content Select Sample Schemas if you want to install them. Click 'Next'
Page 428
Database Services: Click Add button and enter Service Name such as: hrserv then click 'OK' Make sure it's set to 'Preferred' in both nodes and select Basic for TAF. Click 'Next' Initialization Parameters: Memory Size to 70%. You can leave all other settings as is. 'Next' to continue Creation Options Select Generate database creation scripts if you want to review these at a later stage. 'Finish' to review the installation. Summary Select Ok to close the review page and 'Finish' to start the installation. Once database creation is done a summary screen will be displayed. 'Exit' to exit the OUI. # check Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep 9. Postinstallation tasks As the oracle user edit the /etc/oratab file on both nodes. Replace the database name with the instance name for the rac database i.e. replace the rac keyword with rac1 or rac2 depending on the node. Furthermore add details for your clusterware home to this file. This will enable you to set the Clusterware home using the oraenv script. Once edited the /etc/oratab file should contain the following: vi /etc/oratab On node RAC1: +ASM1:/u01/app/oracle/products/10.2.0/db_1:N rac1:/u01/app/oracle/products/10.2.0/db_1:N crs:/u01/crs:N On node RAC2 +ASM2:/u01/app/oracle/products/10.2.0/db_1:N rac2:/u01/app/oracle/products/10.2.0/db_1:N crs:/u01/crs:N
# verify the Cluster Registry configuration srvctl config database -d rac # backup the root.sh script cp /u01/app/oracle/product/10.2.0/db_1/root.sh ~/root.sh.bak # Back up the voting disk dd if=/dev/raw/raw2 of=~/vdisk.bak # verify that OEM is working https://rac1.mydomain.com:1158/em # restart the dbconsole if required emctl status dbconsole emctl stop dbconsole
Page 429
emctl start dbconsole 10. Useful Postinstallation Tasks Following are tips to consider after the successful installation to make managing RAC easier. Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
/* Make crs_stat -t more readable */ /* copy the following script into ~/scripts/crstat.sh */ #!/usr/bin/ksh # # Sample 10g CRS resource status query script # # Description: # - Returns formatted version of crs_stat -t, in tabular # format, with the complete rsc names and filtering keywords # - The argument, $RSC_KEY, is optional and if passed to the script, will # limit the output to HA resources whose names match $RSC_KEY. # Requirements: # - $ORA_CRS_HOME should be set in your environment # suggested scrip name: crstat.sh RSC_KEY=$1 QSTAT=-u AWK=/usr/bin/awk
# then add the following in the .bashrc of oracle user # if the file was saved in ~/scripts/crstat.sh alias crstat='~/scripts/crstat.sh' /* Easy Acces to crs and db homes */ # it is common to access bin directories in clusterware and db homes # add the following to .bashrc of oracle user alias db='cd $ORACLE_HOME/bin' alias crs='cd $ORA_CRS_HOME/bin'
Page 430
Installing Oracle 11g R2 RAC on Enterprise Linux 5

Note: The metalink document RAC Starter Kit and Best Practices (Linux) [ID 811306.1] is a good source reference for this task.
Main Changes in Oracle 11g Release 2 RAC

SCAN: single client access name (SCAN) is a domain name used by all clients connecting to the cluster. It is registered to three IP addresses, either in the domain name service (DNS) or the Grid Naming Service (GNS). GNS: enables using DHCP, which must be configured in the subdomain in which the cluster resides. OCR and Voting can now be configured on an ASM diskgroup. Passwordless automatic SSH connectivity: the installer can configure SSH for you. Intelligent Platform Management interface (IPMI): provides a set of common interfaces to computer hardware and firmware that administrators can use to monitor system health and manage the system. Time sync: The new Oracle Cluster Time Synchronization Service is designed for organizations whose Oracle RAC databases are unable to access NTP services. Grid Infrastructure home: Clusterware and ASM share the same Oracle Home. Hangchecktimer replaced by the cluster synchronization service daemon Agent and Monitor to provide more accurate recognition of hangs and to avoid false termination.
Emulation software: VMWare Workstation 7 RAC Nodes: 2 nodes with 2.5 GB RAM each , 2 ethernet cards. OS: Oracle Linux Enterprise 5 for x86 32-bit
Required Software
Oracle Database 11g Release 2 for Linux x86 32-bit Oracle Database 11g Release 2 Grid Infrastructure (11.2.0.1.0) for Linux x86 32-bit
Used Hardware
In the VMWare: create one virtual machine (rac1) with the following specs: o 2.5 GB RAM o Two ethernet cards: both can be configured as bridged or host-only in VMware. o One local hardisk with 24 GB on SCSI 0:0. o CPU Count: 2
Page 431
o Create a folder in the same directory structure level as the parent folder containing the created virtual machine. Give it a meaningful name like 'shared_disks'. Create in that folder the following disks: Disk1: of 3 GB. Allocate its disk space. It will be used for OCR and Voting disk. Set it on controller SCSI 1:1. Disk2: of 4 GB. Allocate its disk space. It will be used for +Data. Set it on controller SCSI 1:2. Disk3: of 2 GB. Allocate its disk space. It will be used for +Flash. Set it on controller SCSI 1:3.
Installation Plan
11. Preinstallation tasks Hardware requirements Software requirements Environment configuration
12. Oracle Grid Infrastructure installation 13. Oracle Grid Infrastructure Patching 14. Oracle Database 11g R2 Software Installation 15. Oracle Database 11g R2 Software Patching 16. Install EM Agent in cluster nodes (if required)
Page 432
17. ASM Diskgroups Creation 18. RAC Database Creation 19. Complete postinstallation tasks 20. Useful postinstallation tasks Note: The installation is explained without GNS and IPMI Note: For this installation we will be using ASM for Clusterware and Database storage 1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Configure the swap area in the local hardisk to have 3 GB disk space. o Give the first ethernet card IP 192.0.2.100 and the second 172.0.2.100 and the hostname rac1.mydomain.com. Define a gateway. If it does not exist, make it same as the host IP address. o Insall the following packages: Desktop Environments o GNOME Desktop Environment Applications o Graphical Internet (optional) o Editors (optional) Development o Development Libraries o Development Tools Servers o Do not select anything in this group. Base System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
X Window System
Complete the installation. After the Installation compelets, RHEL 5.2 and below will hang on booting when it reaches to "starting udev" line. To solve this problem, shutdown the Vmware machine and change the CPU count and Core Count to only one. Implement the changes below, then shutdown the machine, set CPU count back to 2 and startup the machine. put the kernel command line parameters at the end of the "kernel" line: vi /boot/grub/grub.conf add divider=10 clocksource=acpi_pm
Page 433
For example: kernel /vmlinuz-2.6.18 .. clock=acpi_pm divider=10 For Vmware machines, install VMWare tools and set it to synchronize its time with the guest: vmwaretoolbox. Alternatvily, you can use Oracle Cluster Time Synchronization Service (ctssd) (metalink document 551704.1) Install further packages: # to know distribution and version of Linux (Red Hat Ent. 5.2 used) cat /etc/issue # to know kernel version (and its errata level) (2.6.18-92 or newer) uname -r # to list missed packages: rpm -q --qf '%{NAME}-%{VERSION}-%{RELEASE} (%{ARCH})\n' binutils \ compat-libstdc++-33 \ elfutils-libelf \ elfutils-libelf-devel \ gcc \ gcc-c++ \ glibc \ glibc-common \ glibc-devel \ glibc-headers \ ksh \ libaio \ libaio-devel \ libgcc \ libstdc++ \ libstdc++-devel \ make \ sysstat \ unixODBC \ unixODBC-devel # for missed packages, install them: rpm -Uvh libaio-devel-0.3.106-3.2.i386.rpm rpm -Uvh unixODBC* # Download the appropriate ASMLib RPMs from OTN. # to know the kernel verion: uname -rm # In this case we need: oracleasm-2.6.18-92.el5-2.0.5-1.el5.i686.rpm oracleasmlib-2.0.4-1.el5.i386.rpm oracleasm-support-2.1.3-1.el5.i386.rpm rpm -Uvh oracleasm*.rpm # SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Check the hardware requirements # Hardware Requirements (in cluster nodes) # At least 1.5 GB of physical memory but practically 1.5 is not fine grep MemTotal /proc/meminfo
Page 434
# swap space: same as the amount of physical memory grep SwapTotal /proc/meminfo # to display swap and memory in one command: free # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 1 GB disk space in /tmp df -h /tmp # 8 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: # df -h /dev/shm/ Create the required network configuration (rac2 will be created later): o o o Public and Private interface names must be the same for all nodes. This private hostname does not need to be resolvable through DNS and should be entered in the /etc/hosts file. SCAN VIPs must NOT be in the /etc/hosts file, it must be resolved by DNS. But here I've defined it as a single IP address in the "/etc/hosts" file, which is wrong and will cause the cluster verification to fail, but it allows me to complete the install without the presence of a DNS. If you are using a DNS, Oracle recommends that you add lines to the /etc/hosts file on each node, specifying the public IP, VIP and private addresses. If you configured the IP addresses in a DNS server, then, as the root user, change the hosts search order in /etc/nsswitch.conf on all nodes as shown: Old: hosts: files nis dns New: hosts: dns files nis o Then restart nscd daemon on each node: /sbin/service nscd restart
o o
# Network names Resolution # configure /etc/hosts if no domain server is used (both nodes) vi /etc/hosts 127.0.0.1 localhost.localdomain localhost #eth0 - PUBLIC 192.0.2.100 rac1.mydomain.com rac1 192.0.2.101 rac2.mydomain.com rac2 #VIP 192.0.2.102 rac1-vip.mydomain.com rac1-vip 192.0.2.103 rac2-vip.mydomain.com rac2-vip #eth1 - PRIVATE
Page 435
172.0.2.100 rac1-priv 172.0.2.101 rac2-priv # in real production: the follwing should not be there at all # SCAN: cluster_name-scan.GNS_subdomain_name 192.0.2.104 rac-scan.mydomain.com rac-scan Create and configure the required OS users and groups Note: userid and groupid must be the same in all nodes. You can check them by id oracle command. # all group and user ids on all the nodes must have identical id # Grid Infrastructure (GI) and the Oracle RDBMS home will # be installed using different users: /usr/sbin/groupadd -g 501 oinstall /usr/sbin/groupadd -g 502 dba /usr/sbin/groupadd -g 504 asmadmin /usr/sbin/groupadd -g 506 asmdba /usr/sbin/groupadd -g 507 asmoper /usr/sbin/useradd -u 501 -g oinstall -G asmadmin,asmdba,asmoper grid /usr/sbin/useradd -u 502 -g oinstall -G dba,asmdba oracle # set passwords passwd oracle passwd grid # make sure nobody user exists (if not there, create it useradd nobody) id nobody # define the env variables for oracle user vi /home/oracle/.bash_profile # Oracle Settings export EDITOR=vi TMP=/tmp; export TMP TMPDIR=$TMP; export TMPDIR ORACLE_HOSTNAME=rac1.mydomain.com; export ORACLE_HOSTNAME ORACLE_UNQNAME=RAC; export ORACLE_UNQNAME ORACLE_BASE=/u01/app/oracle; export ORACLE_BASE ORACLE_HOME=$ORACLE_BASE/product/11.2.0/db_1; export ORACLE_HOME ORACLE_SID=RAC1; export ORACLE_SID ORACLE_TERM=xterm; export ORACLE_TERM PATH=/usr/sbin:$PATH; export PATH PATH=$ORACLE_HOME/bin:$PATH; export PATH LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib; export LD_LIBRARY_PATH CLASSPATH=$ORACLE_HOME/JRE:$ORACLE_HOME/jlib:$ORACLE_HOME/rdbms/jlib; export CLASSPATH # shell startup file vi /etc/profile if [ $USER = "oracle" ] || [ $USER = "grid" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi
Page 436
# for C shell vi /etc/csh.login if ( $USER = "oracle" || $USER = "grid" ) then limit maxproc 16384 limit descriptors 65536 endif Configure kernel parameters and shell limits Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Kernel Parameters # to tune thme, refer to metalink document 169706.1 # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf # kernel.shmmax not stated in 11g R2 (max: 4G) (169706.1) kernel.shmmni = 4096 kernel.sem = 250 32000 100 128 fs.aio-max-nr = 1048576 fs.file-max = 6815744 net.ipv4.ip_local_port_range = 9000 65500 net.core.rmem_default = 262144 net.core.rmem_max = 4194304 net.core.wmem_default = 262144 net.core.wmem_max = 1048576 # to take immediate effect /sbin/sysctl -p # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf grid soft nproc 2047 grid hard nproc 16384 grid soft nofile 1024 grid hard nofile 65536 oracle soft nproc 2047 oracle hard nproc 16384 oracle soft nofile 1024 oracle hard nofile 65536 vi /etc/pam.d/login session required pam_limits.so Create the required directories for the Oracle software: # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab # Oracle Inventory Directory # as a root mkdir -p /u01/app/oraInventory chown -R grid:oinstall /u01/app/oraInventory chmod -R 775 /u01/app/oraInventory
Page 437
# Grid Infrastructure Home Directory mkdir -p /u01/11.2.0/grid chown -R grid:oinstall /u01/11.2.0/grid chmod -R 775 /u01/11.2.0/grid # Oracle Base Directory mkdir -p /u01/app/oracle #needed to ensure that dbca is able to run after the rdbms installation mkdir /u01/app/oracle/cfgtoollogs chown -R oracle:oinstall /u01/app/oracle chmod -R 775 /u01/app/oracle # Oracle mkdir -p chown -R chmod -R RDBMS Home Directory /u01/app/oracle/product/11.2.0/db_1 oracle:oinstall /u01/app/oracle/product/11.2.0/db_1 775 /u01/app/oracle/product/11.2.0/db_1
Shutdown the Vmware machine then edit the VMware file (with vmx extensions) and add the following entry to allow sharing the disks (make sure the scsi controller number is the one you used): disk.locking = "FALSE" diskLib.dataCacheMaxSize = "0" diskLib.dataCacheMaxReadAheadSize = "0" diskLib.dataCacheMinReadAheadSize = "0" diskLib.dataCachePageSize = "4096" scsi1.sharedBus = "virtual" scsi1:1.deviceType = "disk" scsi1:2.deviceType = "disk" scsi1:3.deviceType = "disk"
Startup the machine then partition the disks:
Note: On a real life storage, you would create a single whole-disk partition with exactly 1 MB offset on each LUN to be used as ASM Disk. In fdisk: u (to change units from cylinder to sectors), n, p, 1, 2048, w. # as a root, for the disks /dev/sdb .. /dev/sdd # confirm they are seen: ls /dev/sd* #partition the disks: fdisk /dev/sdb # answers: "n", "p", "1", "Return", "Return", "p" and "w" Note: if the following message appears after the "w" command: WARNING: Re-reading the partition table failed with error 16: Device or resource busy, then you can avoid restarting the machine by the following command: partprobe # to make sure partions are created ls -lX /dev/sd* Configure ASM drivers: Note: If you see that the shared disks are not synced between rac1 and rac2, one of the things you can examine is to see if there is any "debug" command in any of the nodes' vmx files. If you find one, shutdown the node, remove the command from the vmx file and restart. # as root
Page 438
(to be done in all nodes) Oracle DBA Code Examples
oracleasm configure -i Default user to own the driver interface []: grid Default group to own the driver interface []: asmdba Start Oracle ASM library driver on boot (y/n) [n]: y Fix permissions of Oracle ASM disks on boot (y/n) [y]: y Writing Oracle ASM library driver configuration: Creating /dev/oracleasm mount point: Loading module "oracleasm": Mounting ASMlib driver filesystem: Scanning system for ASM disks: # Load the kernel module using the following command: /usr/sbin/oracleasm init # If you have any problems, make sure you have the correct # version of the driver: /usr/sbin/oracleasm update-driver # mark the shared disks: (one node) /usr/sbin/oracleasm createdisk DISK1 /dev/sdb1 /usr/sbin/oracleasm createdisk DISK2 /dev/sdc1 /usr/sbin/oracleasm createdisk DISK3 /dev/sdd1 # check the disks are marked and seen: /usr/sbin/oracleasm listdisks # in other nodes: /usr/sbin/oracleasm scandisks /usr/sbin/oracleasm listdisks #If you need to unmark a disk that was used in a createdisk command: /usr/sbin/oracleasm deletedisk DISK1 /usr/sbin/oracleasm deletedisk DISK2 /usr/sbin/oracleasm deletedisk DISK3 Disable screensavers on host & guest machines. o In Oracle Linux: Applications-> Preferences-> Screen Saver o Do the same after logging off and logging on again as oracle and grid user. Shutdown rac1 Copy the folder containing rac1 into a new folder in the same directory structure level. Let's name it "rac2". This will be the second node in the cluster. Edit the VMware file of rac1 and edit the following: displayName = "rac2" Open rac2, then perform: o o o in a terminal issue: system-config-network-gui. Remove the devices with the "%.bak" nicknames. To do this, highlight a device, deactivate, then delete it. Highlight the "eth0" interface and click the "Edit" button. Change its IP addresses and gate way: IP 192.0.2.101. Click on the "Hardware Device" tab and click the "Probe" button. For eth1 set its ip address to 172.0.2.101. Do not define a gateway. In DNS tab, change hostname to rac2.mydomain.com. Activate the network cards. Oracle DBA Code Examples
[ [ [ [ [
OK OK OK OK OK
] ] ] ] ]
o o o
Page 439
In rac2, perform: # change the variable in the file vi /home/oracle/.bash_profile ORACLE_SID=RAC2; export ORACLE_SID ORACLE_HOSTNAME=rac2.localdomain; export ORACLE_HOSTNAME
Start rac1. Make sure the machines can see each other: ping ping ping ping -c -c -c -c 3 3 3 3 rac1 rac1-priv rac2 rac2-priv
2. Oracle Grid Infrastructure installation # in rac1: copy the software in a staging folder mkdir -p /u01/app/stage/ora11gr2gridinfra chown -R grid:oinstall /u01/app/stage/ora11gr2gridinfra chmod -R 775 /u01/app/stage/ora11gr2gridinfra mkdir -p /u01/app/stage/ora11gr2db chown -R oracle:oinstall /u01/app/stage/ora11gr2db chmod -R 775 /u01/app/stage/ora11gr2db # in all nodes, make sure the asm disks are accessible: /usr/sbin/oracleasm scandisks /usr/sbin/oracleasm listdisks # do not use cluvfy because SSH was not configured. # lunch OUI from the clusterware ( as grid from rac1) # if logged in in Genome using another user, log out and log in as grid cd /u01/app/stage/ora11gr2gridinfra ./runInstaller Installation Option >Select radio button 'Install and Configure Grid Infrastructure for a Cluster' >Next Installation Type >Select 'Advanced Installation' >Next Product Language >Accept 'English' as language' >Next Grid Plug and Play >cluster name: rac >SCAN name:rac-scan.mydomain.com >Make sure 'Configure GNS' is NOT selected >Next Cluster Node Information >Add button >Hostname:rac2.mydomain.com >Virtual IP Name: rac2-vip.mydomain.com >OK
Page 440
>"SSH Connectivity" button >Enter the password >Setup button >Test button Network Interface Usage >check the public and private networks are specified correctly >Next Storage Option >Select 'Automatic Storage Management (ASM)' >Next Creat ASM Disk Group >Disk Group Name: DGOCRVOTE (3GB disk: Disk1) >Redundancy: external >Next NOTE: If you see an empty screen for you candidate disks it is likely that ASMLib has not been properly configured. Try reconfigure them. If you are sure that ASMLib has been properly configured click on 'Change Discovery Path' and provide the correct destination. ASM Password >Specify and conform the password you want to use >Next Failure Isolation Support >Select NOT to use IPMI >Next Privileged OS Groups >Assign the correct OS groups for OS authentication (mostly default is OK) >Next Installation Location >ORACLE_BASE: /u01/app/oracle Software location: /u01/11.2.0/grid >Next Create Inventory >Specify the locations: /u01/app/oraInventory >Next Perform Prerequisite Checks >OUI performs certain checks >Check that status of all checks is Succeeded Note: in this example, NPS error can be ignored >Next Summary >Finish Execute Configuration Scripts >Run the scripts as instructed in the screen Note: The scripts must be run on one node at a time. >OK
Page 441
We expect the verification phase to fail with an error relating to the SCAN, assuming you are not using DNS. INFO: Checking Single Client Access Name (SCAN)... INFO: Checking name resolution setup for "rac-scan.localdomain"... INFO: ERROR: INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "rac-scan.localdomain" INFO: ERROR: INFO: PRVF-4657 : Name resolution setup check for "rac-scan.localdomain" (IP address: 192.168.2.201) failed INFO: ERROR: INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "rac-scan.localdomain" INFO: Verification of SCAN VIP and Listener setup failed Provided this is the only error, it is safe to ignore this >Next Message: The installation of the Grid Infrastructure was successfull. >Close Note: If your OS is SUSE Linux, shutting down on node will result in shutting the other nodes. To workaround: #cd /etc/rc3.d #ln -s /etc/init.d/ohasd K07ohasd 3. Oracle Grid Infrastructure Patching Apply patch set, if there is any. 4. Oracle Database 11g R2 Software Installation # make sure all clusterware processes are running /u01/crs/bin/crs_stat -t # as oracle ./runInstaller Configure Security Updates >Provide your e-mail address, if you want >Next Installation Options >Select 'Install Database software only' >Next> Install Type >Select 'Real Application Clusters database installation', and select all nodes. >Use the 'SSH Connectivity' button to configure/test the passwordless SSH connectivity. >Next Product Languages >Confirm 'English' >Next Database Edition
Page 442
>'Enterprise Edition' is ticked >Next Installation Location >Oracle Base: /u01/app/oracle Software Location: /u01/app/oracle/product/11.2.0/db_1 >Next Privileged OS Groups >OSDBA: dba >OSOPER: oinstall >Next Prerequisite Checks ..OUI performs prerequisite checks >Check that status of all checks is Succeeded >If you are sure the unsuccessfull checks can be ignored tick the box 'Ignore All' >Next Summary >Check summary info >Finish Install Product ..OUI installs the db software >as a root, run the root.sh script on the first node then the other nodes (One at a time) >OK Finish >Close 5. Oracle Database 11g R2 Software Patching
6. Install EM Agent in cluster nodes (if required)
7. ASM Diskgroups Creation Note: It is Oracle's Best Practise to have an OCR mirror stored in a second disk group. To follow this recommendation add an OCR mirror. Mind that you can only have one OCR in a diskgroup. To add OCR mirror to an Oracle ASM disk group, ensure that the Oracle Clusterware stack is running and ocrconfig -add +ORADATA ocrcheck # as grid user: start the ASM Configuration Assistant (ASMCA) #su - grid cd /u01/11.2.0/grid/bin ./asmca >Disk Groups tab >Create button >Disk Group Name: DGDATA
Page 443
>Redundancy: External >Disk2 >OK >Create button >Disk Group Name: DGFLASH >Redundancy: External >Disk3 >OK >Exit >Yes 8. RAC Database Creation # as oracle cd /u01/app/oracle/product/11.2.0/db_1/bin ./dbca Welcome Select 'Oracle Real Application Clusters database' >Next Operations > choose option 'Create a Database' >Next Database Template >Select General Purpose or any template >Next Database Identification >Configuration Type: Admin >Globale Database Name: rac >SID: rac >Select All button >Next Management Options >Select the option you want >Next Database Credentials >Set the password(s) >Next Database File Locations >Database Area: +DGDATA >Practically (but not in this case), you should define 'Multiplex Redo Logs and Control Files'. >Next /* Note: If you cannot see the diskgroups, perform the following (ID: 1177483.1): su cd <Grid_Home>/bin chmod 6751 oracle ls -l oracle
Page 444
-rwsr-s--x 1 grid oinstall */ ASM Credentials ..If you chose to set up EM, you will be asked about ASMSNMP password >Enter the password >Ok button Recovery Configuration >Flash recovery area: +DGFLASH >define the size: 2000 MB If the size is smaller than recommended a warning will popup. >Enable Archiving >Next Database Content >Select if you want to have sample schemas created in your database >Next Initialization Parameters >Review and change the settings for memory allocation, characterset etc. >Next Database Storage >Review the database storage settings and change as required >Next Creation Options >Make sure the tickbox 'Create Database' is ticked >Finish Summary >OK .. Database creation proceeding >after completion Exit # Confirmation # to show the current configuration and status of the RAC database srvctl config database -d rac # check OEM (if configured): https://rac1.mydomain.com:1158/em/ # if not started, you can start it: su - oracle cd /u01/app/oracle/product/11.2.0/db_1/bin export ORACLE_UNQNAME=rac ./emctl status dbconsole # check Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep 9. Postinstallation tasks # backup the root.sh script (on all nodes) cp /u01/app/oracle/product/11.2.0/db_1/root.sh ~/root.sh.bak
Page 445
10. General Useful Postinstallation Tasks in Linux Following are tips to consider after the successful installation to make managing RAC easier. Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
/* Make crs_stat -t more readable */ /* copy the following script into ~/scripts/crstat.sh */ #!/usr/bin/ksh # # Sample 10g CRS resource status query script # # Description: # - Returns formatted version of crs_stat -t, in tabular # format, with the complete rsc names and filtering keywords # - The argument, $RSC_KEY, is optional and if passed to the script, will # limit the output to HA resources whose names match $RSC_KEY. # Requirements: # - $ORA_CRS_HOME should be set in your environment # suggested scrip name: crstat.sh RSC_KEY=$1 QSTAT=-u AWK=/usr/bin/awk
# then add the following in the .bashrc of oracle user # if the file was saved in ~/scripts/crstat.sh alias crstat='~/scripts/crstat.sh' /* Easy Acces to crs and db homes */ # it is common to access bin directories in clusterware and db homes # add the following to .bashrc of oracle user alias db='cd /u01/app/oracle/product/11.2.0/db_1/bin' # add the following to .bashrc of grid user alias crs='cd /u01/app/oracle/crs/bin'
Page 446
Installing Oracle 10g R2 RAC on Windows

Note: The metalink document RAC Starter Kit and Best Practices (Windows) [ID 811271.1] is a good source reference for this task.
Installation Methods
NTFS_RAW_ASM (shown in this document) NTFS_RAW_RAM NTFS_OCFS_OCFS
Emulation software: VMWare Server 2 for Windows. RAC Nodes: 2 nodes OS: Windows 2003 Server Standard Edition 32-bit SP2
Required Software
Oracle 10g R2 Clusterware for Windows 32-bit Oracle Database 10g Release 2 for Windows 32-bit
Used Virtual Hardware

In the VMWare: create one virtual machine (rac1) with the following specs: o 2 GB RAM o two ethernet cards: both can be configured as bridged or host-only in VMware. o one local hardisk with 20 GB on controller SCISI 0:0. o CPU Count: 2 o create a folder in the same directory structure level as the parent folder containing the created virtual machine. Give it a meaningful name like 'shared_disks'. Create in that folder two disks of 10 GB and two of 512 MB each. All of them are of LSI Logic type and SCISI Persistent. Make sure they are on SCISI controller different from the SCSI0. For example, set them on SCSI1.
Page 447
Installation Plan
1. 2. 3. 4. 5. 6. 7. 8. 9. Preinstallation tasks Oracle Clusterware installation Apply Patch Set 3 (10.2.0.4) on Clusterware software Oracle ASM 10g Software Installation Apply Patchset 3 (10.2.0.4) on ASM software Install EM Agent in cluster nodes (if required) Configure Listeners Create ASM Instance Install Oracle RAC Database Home Software
10. Apply Patchset 3 (10.2.0.4) on Oracle RAC Software Home 11. Perform cluster database creation 12. Useful postinstallation steps 1. Preinstallation tasks The local admin username and password must be the same on both nodes. Give the first ethernet card IP 192.168.4.11 and the second 192.168.0.11. Define a gateway. If it does not exist, make it same as the host IP address. Note: Gateway must be defined and its IP must be alive. Set the hostname to rac1.mydomain.com (Desktop-> right click My Computer-> Properties-> Computer Name-> Change-> type computer name: rac1 -> More-> type Primary DNS Suffix: mydomain.com Oracle DBA Code Examples
Page 448
/* Prepare the raw disks */ # enable raw disk auto mounting diskpart AUTOMOUNT ENABLE exit # then reboot # start disk management start-> run-> type: diskmgmt.msc-> Disk Initialization Wizard pops up-> Next-> mark all disks (1-4) to initialize-> make sure all disks are unmarked Next-> Finish # all disks must be in Basic mode (not dynamic) right click Disk1-> New Partition-> Next-> select the Extended partition radio button-> Next-> select the partition size to fill the disk-> Next-> Finish right click on the Accept the default select the "Do not Next-> select the "Do not partition and select the "New Logical Drive" -> Next-> partition size Next-> assign a drive letter or drive path" option-> format this partition" option-> Next-> Finish
Repeat the previous partitioning steps for the remaining disks.
/* Time Sync */ -- make sure the time is synched in Vmwaretools -- in Windows, in Date and Time Settings, there is an option "Automatically syn with Internet Server" which can be configured, if connected to Web Create the required network configuration (rac2 will be created later): # Network names Resolution # if no domain server is used (both nodes) notepad C:\WINDOWS\system32\drivers\etc\hosts 127.0.0.1 localhost #Public 192.168.4.11 rac1.mydomain.com rac1 192.168.4.12 rac2.mydomain.com rac2 #VIP 192.168.4.13 rac1-vip.mydomain.com rac1-vip 192.168.4.14 rac2-vip.mydomain.com rac2-vip #Inter-connect 192.168.0.11 rac1-priv.mydomain.com rac1-priv 192.168.0.12 rac2-priv.mydomain.com rac2-priv Disable Windows Media Sensing, which allows Windows to uncouple an IP address from a card when the link to the local switch is lost: in the registry: HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Tcpip\Parameters Add the following registry entry to the Parameters subkey: Name: DisableDHCPMediaSense Data type: REG_DWORD (Boolean) Value: 1 Open the "Network Connections" screen (Start > All Programs > Accessories > Communications > Network Connections). Rename the two connections to "public" and "private" respectively, making sure you apply the names to the appropriate connections. Oracle DBA Code Examples
Page 449
Ensure the public interface is first in the bind order: o o o Open the "Network Connections" dialog by right-clicking on the "My Network Places" icon and selecting the "Properties" menu option. Select the "Advanced > Advanced Settings..." menu option. On the "Adapters and Bindings" tab, make sure the public interface is the first interface listed. Otherwise, promote it.
Disable screensavers Shutdown rac1 Edit the VMware file (with vmx extensions) and add the following entry to allow sharing the disks: disk.locking = "FALSE" diskLib.dataCacheMaxSize = "0" diskLib.dataCacheMaxReadAheadSize = "0" diskLib.dataCacheMinReadAheadSize = "0" diskLib.dataCachePageSize = "4096" diskLib.maxUnsyncedWrites = "0" scsi1.sharedBus = "virtual" scsi1:1.deviceType = "disk" scsi1:2.deviceType = "disk" scsi1:3.deviceType = "disk" scsi1:4.deviceType = "disk" scsi1:5.deviceType = "disk"
Copy the folder containing rac1 into a new folder in the same directory structure level. Let's name it "rac2". This will be the second node in the cluster. Edit the VMware file of rac1 and edit the following: displayName = "rac2" Start rac2 then change it's IP addresses: IP 192.168.4.12 and 192.168.0.12. It's hostname to rac2.mydomain.com. Modify their default gateway accordingly. Change hostname to rac2 and restart the virtual machine. Start rac1 and make sure it can ping rac2. Perform Net Use test, to make sure files can be transfered: # on rac1 net use \\rac2\c$ # on rac2 net use \\rac1\c$
Note: The environment variables TMP & TEMP must point to the same directory in all RAC nodes, which is the case here. Stage Oracle software (Clusterware, DB and patch set) in rac1. Run the CVU to check the state of the cluster prior to the install of the Oracle Software. The messages 'The system cannot find the file specified' and 'Could not find a suitable set of interfaces for VIPs' can be ignored. C:\temp\OracleClusterware\cluvfy>runcluvfy stage -post hwos -n rac1,rac2 C:\..\cluvfy>runcluvfy stage -pre crsinst -n rac1,rac2 -verbose 2. Oracle Clusterware installation if you are using terminal services to perform the installation, be sure to invoke the terminal services in 'console' mode: mstsc -v:servername /F /console
Page 450
OR, for some (newer) versions of Windows: mstsc -v:servername /F /admin /* Stop Interrupting Services on both nodes */ # the MSDTC service may interrupt installation process # it can be started after installation stop the service: Distributed Transaction Coordinator # lunch OUI from the clusterware software (on rac1) cd C:\temp\OracleClusterware\ ./setup.exe
>Welcome messge >Next >enter Home Details OraCr10g C:\oracle\product\10.2.0\crs >Product-Specific Prerequisite Checks They all should succeed >Specify Cluster Configuration Select 'Add' and add the details for node rac2. Enter all details >OK rac2.mydomain.com rac2-priv.mydomain.com rac2-vip.mydomain.com >Next >Specify Network Interface Usage set the Public interface (subet net 192.168.4.*) >Ok >Specify Oracle Cluster Registry (OCR) Location >Highlight disk 1 and click the "Edit" >select the "Place OCR(Primary) on this Partition" option >OK >Highlight disk 2 and click the "Edit" >Select the "Place Voting Disk on this Partition" option >OK >Next and ignore the redundancy warnings >OK >Summary >Install >Wait while the configuration assistants run If the Configuration Assistant fails, there is a problem that must be fixed before proceeding. Metalink documents might help: 356535.1, 310791.1 VIPCA should fails, just click the "OK" button on the resulting error screen >Next button and accept the subsequent warning >Exit On the RAC1 virtual machine, run the VIPCA manually: cd c:\oracle\product\10.2.0\crs\bin vipca.bat
Page 451
>Welcome >Next >Highlight the "public" interface >Next >Enter the virtual IP alias and address for each node: rac1-vip.mydomain.com Once you enter the first alias, the remaining values should default automatically >Next >Summary >Finish
# To verify the installation: cd C:\oracle\product\10.2.0\crs\BIN -- 1 cluvfy stage -post crsinst -n rac1,rac2 -- 2 ping rac1-vip ping rac2-vip -- 3 crsctl check crs -- 4 # check status of crs daemon processes (ALL NODES) crs_stat -t Name Type Target State Host -----------------------------------------------------ora.rac1.gsd application ONLINE ONLINE rac1 ora.rac1.ons application ONLINE ONLINE rac1 ora.rac1.vip application ONLINE ONLINE rac1 ora.rac2.gsd application ONLINE ONLINE rac2 ora.rac2.ons application ONLINE ONLINE rac2 ora.rac2.vip application ONLINE ONLINE rac2 # if State of any of them is UNKNOWN, try restarting the deamons crsctl stop crs # then start again crsctl start crs 3. Apply Patch Set 3 (10.2.0.4) on Clusterware software There is a bug in Oracle Clusterware 10.2.0.1 in which sometime CRSS service is unable to access OCR disk. This issue is addressed by Patch Set 3. Therefore, to avoid any possible problem by this issue, you should apply the Patch Set at this stage. # (1) Apply the patch on clusterware # in ALL NODES # Stop the following services and make their startup type Manual (if running): Oracle Object Service OracleClusterVolumeService OracleCRService OracleCSService OracleEVMService
Page 452
Open task manager and kill ons processes, if there is any. # if any of the services is hang, you can change the startup type from regedit then reboot (don't forget to stop the Distributed Transaction service after reboot) : HKEY_LOCAL_MACHINE -> SYSTEM -> CurrentControlSet -> Services -> OracleCSService -> Select (Start) -> Edit -> Specify 3 in Value
# in RAC1: start oui cd C:\temp\Patchset_10204\Disk1> setup.exe Welcome >Next >select Oracle Clusterware home >Next >Confirm nodes >Next >Checking should pass >Next >Summary >Install >Exit # update the nodes # make sure all related Oracle services are stopped, otherwise stop them # apply the batch in node1 then node2: C:\oracle\product\10.2.0\crs\install\patch102.bat # it should end with the following message: Successful upgrade of this node to Oracle Cluster Ready Services # Oracle Process Manager service will automatically be installed # check the updated verion crsctl query crs activeversion # then check status of deamons crs_stat -t # change the startup mode of the serives to AUTOMATIC # startup Distributed Service, if it was stopped. 4. Oracle ASM 10g Software Installation # make sure all clusterware processes are up and running cd C:\oracle\product\10.2.0\crs\BIN crs_stat -t -- in rac1: start OUI cd C:\temp\OracleDB10gR2 setup.exe Welcome >Next Select Installation Type >Next
Page 453
Install Location >Oracle Home name: OraDb10g >Oracle Home location: C:\oracle\product\10.2.0\db_1 >Next Specify Hardware Cluster Installation Mode >select both nodes >Next Product-Specific Prerequisite Checks All pre-requisite steps should complete successfully. >Next Select Configuration Option >Install Software Only >Next Summary >Finish End of Installation >Exit Optionally, check and modify, if you wish, NLS_LANG in regedit. 5. Apply Patchset 3 (10.2.0.4) on ASM Software Note: This is the patchset applied for this environment. Generally speaking, if there is a newer version, use it instead of this version. In all cases, same patch set applied to Oracle software must also be applied to Clusterware first. #Apply the patch on DB software # in RAC1: start oui cd C:\temp\Patchset_10204\Disk1> setup.exe Welcome >Next >select Oracle Database home >Next >Confirm nodes >Next >Checking should pass >Next >Summary >Install >Exit # not required, but recommended: restart the nodes 6. Install EM Agent in cluster nodes (if required)
7. Configure Listeners # Run netca (not netmanager which is not clusteraware) from any node cd C:\oracle\product\10.2.0\db_1\bin netca.bat
Page 454
>Select Cluster configuration and select both nodes. >Configure Listener >Configure Naming Methods: Local and Easy Connect After installing listener check with below commands whether listener working properly C:\oracle\product\10.2.0\crs\bin\crs_stat -t 8. Create ASM Instance # in rac1 cd C:\oracle\product\10.2.0\db_1\bin dbca.bat Welcome Keep the default selection Oracle RAC database >Next Operations Select >Configure ASM >Next Node Selection >Select ALL the nodes >Next Create ASM Instance Select a SYS password for the ASM instance. Select IFILE parameter file to create >Next >OK (asm instances will be created) # here's an issue faced in a case after creating the ASM instances: # check that the Administrator (or the OS you're using) belong to ora_dba # group in ALL the Nodes. If not there, add the user to the group.
# Stamp the disks for ASM >Create New >Stamp Disks >Select the "Add or change label" option >Next >select the first Disk of size 10G >type DATA in the prefix text field (don't type ASM) >Next >Next >Finish >stamp the second 10g disk: repeat above for the second disk rac2 should see the changes: asmtool -list Note: if you want to clean the disk to redo the procedure above, you can: (1) in rac1: DISKPART, select <DISK NAMe>, clean all, create part ext, create part log (2) in rac2: remove letter drive assigned to the partition # create ASM Disk Groups >Enter dgdata1 as the first disk group name. >Set redundancy settings to External
Page 455
>select the disk ORCLDISKDATA0 >Ok Select Create New again to add another disk group named as dgfra. This group will be used as the recovery arae >OK >Finish to complete. Note: in an environment, I kept receiving ORA-15063 error. After making sure that all the pre-requisits were applied, the issue was resolved by allocating all the disk space at disk creation time. # An ASM instance will be created in every node named as ASMn set ORACLE_SID=ASM1 sqlplus /nolog SQL>conn / as sysdba select name from v$asm_diskgroup ; ... 9. Install Oracle RAC Database Home Software # make sure all clusterware processes are up and running cd C:\oracle\product\10.2.0\crs\BIN crs_stat -t
-- in rac1: start OUI cd C:\temp\OracleDB10gR2 setup.exe Welcome >Next Select Installation Type: Enterprise Edition >Next Install Location >Oracle Home name: OraDb10g2 >Oracle Home location: C:\oracle\product\10.2.0\db_2 >Next Specify Hardware Cluster Installation Mode >select both nodes >Next Product-Specific Prerequisite Checks All pre-requisite steps should complete successfully >Next Upgrade an Existing Database >select No >Next Select Configuration Option >Install Software Only >Next
Page 456
Summary >Finish End of Installation >Exit Optionally, check and modify, if you wish, NLS_LANG in regedit. 10. Apply Patchset 3 (10.2.0.4) on Oracle RAC Software Home #Apply the patch on DB software # in RAC1: start oui cd C:\temp\Patchset_10204\Disk1> setup.exe Welcome >Next >select Oracle RAC Database home >Next >Confirm nodes >Next Oracle Configuration Manager you can setup the Oracle Configuration Manager >Checking should pass >Next >Summary >Install >Exit 11. Perform cluster database creation # in rac1: from Oracle Database home (not ASM) c:\oracle/product/10.2.0/db_2/bin\dbca Create Oracle RAC Database 'Next' to continue Operations Select Create a Database. 'Next' to continue. Node Select all the nodes 'Next' to continue. Database Templates Select the required template: like "General Purpose" 'Next' to continue. Database Identification Enter rac as the global database name. 'Next' to continue. Management Options Keep the default settings "Configure Database Control for Configure Enterprise Manager" selected, Keep "Enable Alert Notifications" and "Enable Daily Disk Backup to Recovery Area" deselected. 'Next' to continue
Page 457
Database Credentials Select Use the same Administrative Password for All Accounts and enter a password. 'Next' to continue Storage Options Select ASM for storage 'Next' to continue ASM Disk Groups Select the disk groups dgdata1 'Next' to continue Database File Locations Keep the default Use Oracle-Managed Files. Make sure +DGDATA1 is entered as the Database Area. 'Next' to continue Recovery Configuration: Select Specify Flash Recovery Area and enter +DGFRA as the Flash Recovery Area. Set its size. Select Enable archiving and click on the Edit Archive Mode Parameters button and make sure the Flash Recovery area is pointing to +DGFRA 'OK' and then 'Next' to continue. Database Content Select Sample Schemas if you want to install them. Click 'Next' Database Services: Click Add button and enter Service Name such as: hrserv then click 'OK' Make sure it's set to 'Preferred' in both nodes and select Basic for TAF. Click 'Next' Initialization Parameters: Memory Size to 70%. You can leave all other settings as is. 'Next' to continue Database Storage Here you can review the placement of various database files 'Next' to continue Creation Options Select Generate database creation scripts if you want to review these at a later stage. 'Finish' to review the installation. Summary Select Ok to close the review page and 'OK' to start the installation. Once database creation is done a summary screen will be displayed. Copy the OEM URL into clipboard. Save it in a file or the Internet Browser 'Exit' to exit the OUI.
Page 458
12. Useful Postinstallation Steps Following are tips to consider after the successful installation to make managing RAC easier. # Create links to Oracle crs and database homes. notepad C:\WINDOWS\system32\crs.bat cd /d C:\oracle\product\10.2.0\crs\BIN notepad C:\WINDOWS\system32\db.bat cd /d C:\oracle\product\10.2.0\db_2\BIN notepad C:\WINDOWS\system32\asm.bat cd /d C:\oracle\product\10.2.0\db_1\BIN
Page 459
Cleaning Up Clusterware Installation on Windows

If clusterware installation fails, sometimes you need to clean up the installation and redo the installation again. Following procedure just does that. The procedure applies on Oracle 10g R1, R2 and Oracle 11g R1. 1. Stop the following services on each node and set them to Manual: OracleCSService OracleCRService OracleEVMService
2. To see the link names that have been assigned to, invoke the Oracle tool <CRS_HOME>\bin\GuiOracleObjManager.exe If you already remove the CRS_Home, you can download the tools from: - metalink ID: 341214.1 - http://www.ahmedbaraka.com/oracle/cleanup.zip
3. If you're using RAW disks, invoke the Oracle tool logpatformat.exe to reinitialize the headers of the disks: run logpartformat /q <link name as shown in guioracleobjmanager tool> For example: logpartformat /q \\.\ocrcfg Repeat this step for all link names listed in guioracleobjmanager) 4. If you're using OCFS: navigate to CRS_HOME\bin\ run logpartformat /q <DRIVELETTER>: For example: logpartformat /q P:
5. Remove all the assigned link names using the GUIOracleobjmanager.exe tool by clicking: Placing a check mark in the check box for the given partition, then choosing 'Commit' from the 'Options' menu.
6. Remove and recreate your logical drives on top of extended partitions from Windows Disk Management.
7. Use the OUI to remove the software from the CRS home
8. Remove Oracle binaries using Windows explorer, both the CRS home and the files located in: c:\program files\oracle
9. Check the registry on each node and ensure that the following services have been removed from (remove them, if not): HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services and HKEY_LOCAL_MACHINE\SYSTEM\ControlSetXXX\Services Services to be removed include: ocfs OracleCSService
Page 460
OracleEVMService OracleCRService OracleObjectService OracleClusterVolumeService OraFenceService
10. Using Windows explorer, remove the following driver files from: %systemroot%\windows\system32\drivers: ocfs.sys orafencedrv.sys orafenceservice.sys
11. Reboot all servers in your RAC configuration
Page 461
Single Instance to RAC Conversion

The Tools to Convert a Single Instance DB to RAC
DBCA: enables moving from host that will not be part of the RAC rconfig utility: the single instance db host will be part of the RAC Enterprise Manager
Conversion Prerequisites for Oracle 10g R2

Oracle Clusterware 10g Release 2 (10.2) is installed, configured, and running. Oracle Real Application Clusters 10g Release 2 (10.2) software is installed. The Oracle binary is enabled for RAC. Shared storage is available and accessible from all nodes. User equivalence exists for the oracle user.
Using rconfig Utitlity

The example below applies on Oracle 10g R2. Test rconfig before converting to RAC #make copy of $ORACLE_HOME/assistants/rconfig/sampleXMLs/ConvertToRAC.xml su - oracle cd $ORACLE_HOME/assistants/rconfig/sampleXMLs/ cp ConvertToRAC.xml Ora10gToRac.xml #edit the file: make changes as instructed in the comments # make sure to set Convert verify = "ONLY" # below is an example file: xsi:schemaLocation="http://www.oracle.com/rconfig"> <n:ConvertToRAC>  <n:Convert verify="ONLY">  <n:SourceDBHome>/u01/app/oracle/product/10.2.0/db_1</n:SourceDBHome>  <n:TargetDBHome>/u01/app/oracle/product/10.2.0/db_1</n:TargetDBHome>  <n:SourceDBInfo SID="ora10g"> <n:Credentials> <n:User>sys</n:User> <n:Password>ora10g</n:Password> <n:Role>sysdba</n:Role> </n:Credentials>
Page 462
</n:SourceDBInfo>  <n:ASMInfo SID="+ASM1"> <n:Credentials> <n:User>sys</n:User> <n:Password>welcome</n:Password> <n:Role>sysdba</n:Role> </n:Credentials> </n:ASMInfo>  <n:NodeList> <n:Node name="rac1"/> <n:Node name="rac2"/> </n:NodeList>  <n:InstancePrefix>rac</n:InstancePrefix>  <n:Listener port=""/>  <n:SharedStorage type="ASM">  <n:TargetDatabaseArea>+DG1</n:TargetDatabaseArea>  <n:TargetFlashRecoveryArea>+DG2</n:TargetFlashRecoveryArea> </n:SharedStorage> </n:Convert> </n:ConvertToRAC> </n:RConfig>
# run the tool for testing rconfig Ora10gToRac.xml
Using DBCA
The example below tested on an Oracle 10g R2 for Enterprise Linux x86 (version 4.5). The DBCA: o o o o Automates the configuration of the control file attributes Creates the undo tablespaces and the redo logs Makes the initialization parameter file entries for cluster-enabled environments Configures Oracle Net Services, Oracle Clusterware resources, and the configuration for RAC database management for use by Oracle Enterprise Manager or the SRVCTL utility.
Page 463
# (1) Create a preconfigured image of your single-instance database $ORACLE_HOME\bin\DBCA Welcome screen >Next Operations screen >Manage Templates >Next Template Management screen >select "Create a database" template >"From an existing database (structure as well as data)," >Next Source Database screen, >Select the database name >Next On the Template Properties screen >enter a template name in the Name field. >enter a description of the file in the Description field >change the template file location in the Template data file field if you want >Next Location of Database Related Files screen >"Maintain the file locations," so that you can restore the database to the current directory structure >Finish The DBCA generates two files: a database structure file (template_name.dbc) and a database preconfigured image file (template_name.dfb).
#(2) Copy the preconfigured database image to node1 on rac1 as oracle scp [email protected]:/u01/app/oracle/product/10.2.0/db_1/assistants/dbca/templa tes/Ora10gDB.* /u01/app/oracle/product/10.2.0/db_1/assistants/dbca/templates
#(3) Create the RAC DB from the supplied template In rac1, run DBCA to create a new database. On the DBCA Template Selection screen, use the template that you copied. Complete the installation. Note: When I tested this procedure, DBCA returns unclear TNS error. I let the DBCA generate the scritp but unfortunately I then realized that the generated script will create a single instance database!
Page 464
Administering RAC Database
Page 465
Administering Oracle Clusterware Components

Managing Cluserware Daemons and Processes
Displaying Clusterware Processes ps -ef | grep d.bin ps -ef | grep -E "opro|racgimon|evmd|crsd|ocssd" | grep -v grep Starting, Stopping, Enabling and Disabling crs Stack # as root # prevent CRS from starting on a reboot crsctl disable crs # start crs crsctl start crs # enable CRS on reboot: crsctl enable crs # disable CRS on reboot (not stopping current stack): crsctl disable crs # start crs crsctl stop crs
CSS Parameters
MISSCOUNT: Represents network heartbeat timeouts (600 s) DISKTIMEOUT: Represents disk I/O timeouts outside reconfiguration (200 s). Should always be less than I/O latency otherwise crs may trigger node eviction. # to set disktimeout: 1. Shut down Oracle Clusterware on all nodes but one. 2. As root on available node, use (where M is the I/O latency): crsctl get css disktimeout crsctl set css disktimeout M+1 3. Reboot available node. 4. Restart all other nodes. # to set misscount # same as above except: crsctl set css misscount 300
Administering Voting Disks in RAC

Multiplexing Voting Disks Use one voting disk if it is stored on a reliable disk. Multiplexed copies should be stored on independent devices. Oracle DBA Code Examples
Page 466
Use at least three multiplexed copies. A typical voting disk configuration comprises between three and five disks.
Dynamically Adding and Removing Voting Disks after Installing RAC Recommendation is to use symbolic links # if you have multiple voting disks, you can add and remove voting disks su crsctl delete css votedisk /dev/raw/raw2 crsctl add css votedisk /dev/raw/raw3 # if the command doesn't work online: su # in all nodes crsctl stop crs crsctl add css votedisk /dev/raw/raw3 -force # in all nodes crsctl start crs Backing up Voting Disks Do it when you add or remove a node. Perform it on every voting disk. # Back up the voting disk (can be online) # usually 4K block size is OK su # to list voting disks currently used crsctl query css votedisk # backup dd if=/dev/raw/raw2 of=~/vdisk.bak bs=4k # on Windows use ocopy Recovering Voting Disks # recoverying voting disk dd if=~/vdisk.bak of=/dev/raw/raw2 # if you have multiple voting disks, # you can add (multiplex) and remove voting disks su crsctl delete css votedisk /dev/raw/raw2 crsctl add css votedisk /dev/raw/raw2
Administering the Oracle Cluster Registry (OCR)

RAC environments do not support more than two OCRs (OCR and its mirror) Using mirrored OCR is optional but recommended when not stored on RAID storage. Note: The ocrconfig command cannot modify OCR configuration information for nodes that are shut down or for nodes on which the Oracle Clusterware is not running.
Page 467
Replacing the OCR If you receive from ocrcheck the message: "Device/File needs to be synchronized with the other device", it means OCR mirror is out of sync with the primary OCR. In this case, replace or relocate your failing OCR with a copy of the other healthy OCR. If it is the primary OCR file that is failing, and if your OCR mirror is still in good health, you can use the ocrconfig replace ocr <ocrfilename> Executing ocrconfig replace ocr|ocrmirror filename adds the primary or mirror OCR file to your environment if it does not already exist. Executing ocrconfig replace ocr|ocrmirror removes the primary or the mirror OCR. If you remove a primary OCR file, the mirror OCR file becomes primary. # 1) verify the other OCR is online # 2) verify crs is running in the node you are using to replace OCR crsctl check crs crs_stat -t # 3) run one of the following ocrconfig -replace ocr /dev/raw/raw5 ocrconfig -replace ocrmirror /dev/raw/raw6 # 4) In any node that is stopped in your RAC ocrconfig -repair ocrmirror /dev/raw/raw2 Adding and Removing the OCR # Adding an Oracle Cluster Registry ocrconfig -replace ocr /dev/raw/raw5 ocrconfig -replace ocrmirror /dev/raw/raw6 # Removing an Oracle Cluster Registry # If you remove a primary OCR, then the mirrored OCR becomes the primary OCR # to remove the ocr ocrconfig -replace ocr # to remove the mirrored ocrconfig -replace ocrmirror Repairing the OCR You may need to repair an OCR configuration on a particular node if your OCR configuration changes while that node is stopped. The OCR configuration information is stored in: /etc/oracle/ocr.loc on Linux and AIX /var/opt/oracle/ocr.loc on Solaris and HP-UX Registry key HKEY_LOCAL_MACHINE\SOFTWARE\Oracle\ocr on Windows # Reparing an OCR (repairs only the ocr configuration info, not its itegrity) # 1. the crs must be stopped su crsctl stop crs # 2. repair ocrconfig -repair ocrmirror /dev/raw/raw2 # 3. start crs crsctl start crs Making Physical Backups of the OCR Create manually copies of the automatically generated physical backups on daily basis. Export the OCR contents before and after making significant configuration changes such as adding or deleting nodes from your environment, modifying Oracle Clusterware resources, or creating a database. Oracle DBA Code Examples
Page 468
Do not perform an OCR restore as a correction to revert to previous configurations if some of these configuration changes fail. # backup OCR # backup the backups that are automatically generated in ( every 4 hrs ) $CRS_HOME/cdata/CLUSTER_NAME # the default directory can be obtained: ocrconfig -showbackup # it is a good idea to set the auto-backup directory to a shared storage: ocrconfig backuploc /shared/bak
Recovering the OCR using the Physical Backups # make recovery only when you are sure there is an error in the # ocr. use ocrcheck to check the ocr # in unix: # 1. Identify the OCR backups then review the contents of the backup # backups done timings ocrconfig -showbackup ls -lt /u01/crs/cdata/crs/ ./ocrdump -backupfile /u01/crs/cdata/crs/backup00.ocr less OCRDUMPFILE # 2. Stop the Oracle Clusterware software on ALL Nodes su crsctl stop crs # 3. Make sure that the OCR devices that you specify in the OCR configuration file (/etc/oracle/ocr.loc) exist. Restore from an OCR backup file from Step 1: ocrconfig -restore /u01/crs/cdata/crs/backup00.ocr # 4. Restart the Oracle Clusterware on all of the nodes su crsctl start crs # 5. Verify the OCR integrity cluvfy comp ocr -n all [-verbose] # in Windows # 1. Identify the OCR backups using the ocrconfig -showbackup ocrdump -backupfile <file_name> # 2. On all of the remaining nodes, disable the following OCR clients: OracleClusterVolumeService, OracleCSService, OracleCRService, and the OracleEVMService # 3. Apply the OCR backup file from Step 1 ocrconfig -restore c:\oracle\crs\cdata\crs\backup00.ocr # 4. Start all of the services that were stopped in step 2. Restart all of the nodes and resume operations in cluster mode. # 5. Verify the OCR integrity where the -n all cluvfy comp ocr -n all [-verbose] Making Logical Backups of the OCR (Exporting) Export the OCR contents before and after making significant configuration changes, such as adding or deleting nodes from your environment, modifying OracleClusterware resources, or creating a database. # to export OCR contents (output file is a binary not-to-edit file) ocrconfig export /u01/ocrbackup/ocr.bak Making Logical Backups of the OCR (Importing) # in Unix # 1. Identify the OCR export file that you want to import # 2. Stop Oracle Clusterware on all the nodes in your RAC database
Page 469
su crsctl stop crs # 3. Import the file ocrconfig -import file_name # 4. Restart Oracle Clusterware on all the nodes su crsctl start crs # 5. verify OCR integrity cluvfy comp ocr -n all # Import in Windows 1. Identify the OCR export file 2. Stop the following OCR clients on each node: OracleClusterVolumeService, OracleCMService, OracleEVMService, OracleCSService, and the OracleCRService. 3. Import the OCR export file ocrconfig -import ocrexport.dat 4. Restart all of the affected services on all nodes. 5. Verify the OCR integrity where node_list is a list of all of the nodes: cluvfy comp ocr -n all [-verbose] Diagnosing OCR Problems with the OCRDUMP and OCRCHECK Utilities OCRDUMP view OCR contents in a readable format. ls -lt /u01/crs/cdata/crs/ ./ocrdump -backupfile /u01/crs/cdata/crs/backup00.ocr less OCRDUMPFILE ./ocrdump dumpoutput -backupfile /u01/crs/cdata/crs/backup00.ocr ./ocrdump -stdout -backupfile /u01/crs/cdata/crs/backup00.ocr -xml # checks logs are reported in CRS_Home/log/hostname/client ./ocrcheck
Page 470
Administering Storage
Datafile Access in Real Application Clusters Redo Log File Storage in Real Application Clusters Each instance has its own online redo log groups which are referred to as an instance's thread of online redo. /* instance Thread */ -- thread is unique to every node select value from v$parameter where name='thread'; -- If you change the thread of an instance, it won't -- take effect till you enable it (otherwise db restart will fail): ALTER DATABASE ENABLE THREAD 3; -- following commands remove thread 3 and replace it with thread 2 ALTER SYSTEM SET thread = 2 SCOPE=SPFILE SID='rac2'; /u01/crs11g/bin/srvctl stop instance -d rac -i rac2 /u01/crs11g/bin/srvctl start instance -d rac -i rac2 connect / as sysdba alter database disable thread 3; -- every thread has at least two groups select thread#, group#, members, bytes/1024/1024 MB, status from v$log order by thread#, group#; select group#, member, status from v$logfile order by group#;
/* to -- if ALTER ALTER -- or alter group group group group
add a redo log group */ OMF or ASM is used DATABASE ADD LOGFILE THREAD 1 GROUP 5 SIZE 50M; DATABASE ADD LOGFILE THREAD 2 GROUP 6 SIZE 50M; database add logfile 5 ('/../_g5_m1.dbf', 6 ('/../_g6_m1.dbf', 7 ('/../_g7_m1.dbf', 8 ('/../_g8_m1.dbf', thread 2 '/../_g5_m2.dbf') '/../_g6_m2.dbf') '/../_g7_m2.dbf') '/../_g8_m2.dbf')
SIZE SIZE SIZE SIZE
128M 128M 128M 128M
REUSE, REUSE, REUSE, REUSE;
/* to drop a group */ -- make sure its inactive select thread#, group#, members, bytes/1024/1024 MB, status from v$log order by thread#, group#; -- switch log if required alter system switch loglife; -- start archiving, if required: specific instance alter system archive log instance 'rac1' next ; -- drop the group ALTER DATABASE DROP LOGFILE GROUP 5; ALTER DATABASE DROP LOGFILE GROUP 6; Automatic Undo Management in Real Application Clusters # display undo tablespace used by the SID SELECT VALUE FROM V$PARAMETER WHERE UPPER(NAME) in
Page 471
('UNDO_TABLESPACE','INSTANCE_NAME'); # change undo tablespace for an instance ALTER SYSTEM SET UNDO_TABLESPACE ='UNDORAC2' SID='rac2';
Administering ASM Instances with SRVCTL in RAC

# to start, stop, and obtain the status of an ASM instance srvctl start asm -n node_name [-i asm_instnace_name] [-o start_options] [-c <connect_str> | -q] srvctl stop asm -n node_name [-i asm_instance_name] [-o stop_options] [-c <connect_str> | -q] srvctl config asm -n node_name srvctl status asm -n node_name # to add configuration information about an existing ASM instance: srvctl add asm -n node_name -i asm_instance_name -o oracle_home # to remove an ASM instance: srvctl remove asm -n node_name [-i asm_instance_name] # to enable an ASM instance: srvctl enable asm -n node_name [-i ] asm_instance_name # to disable an ASM instance: srvctl disable asm -n node_name [-i asm_instance_name]
Page 472
Administering Cluster Databases

Displaying Current Instance in SQL*Plus Prompt
# in login.sql file (usually in $ORACLE_HOME/bin) add: SET SQLPROMPT '_USER@ _CONNECT_IDENTIFIER> '
Starting and Stopping Instances and RAC Databases

Starting Up and Shutting Down with SQL*Plus -- on an instance on your local node echo $ORACLE_SID CONNECT / AS SYSDBA SELECT VALUE FROM V$PARAMETER WHERE UPPER(NAME) ='INSTANCE_NAME'; -- the following command will wait for all trans in the db SHUTDOWN TRANSACTIONAL -- the following command will wait for the trans in the connected node only SHUTDOWN TRANSACTIONAL LOCAL -- connected instance will shutdown STARTUP MOUNT -- or CONNECT /@db1 AS SYSDBA .. Intermittent Windows Shutdown Issue in RAC Environments Normal Windows shutdown may hange, if the instance was shutdown using SQL*Plus. To avoid it, shutdown the following using SRVCTL: All services on the node. All database instances on the node. All ASM instances on the node. All node applications on the node.
Starting Up and Shutting Down with SRVCTL Shutting down a database means shutting down all its instances. # specific instance(s) # preferred and available services will also be started alongside srvctl start instance -d db_name -i "inst_name_list" [-o start_options] srvctl stop instance -d name -i "inst_name_list" [-o stop_options] srvctl stop instance -d rac -i "rac1,rac2" -o immediate # entire cluster database # when you start, only non-running instances will be started srvctl start database -d name [-o start_options] srvctl start database -d rac -o mount srvctl stop database -d name [-o stop_options]
Page 473
Customizing How Oracle Clusterware Manages RAC Databases

You may want to: prevent the Oracle Clusterware from restarting your RAC database when you restart the system avoid restarting failed instances more than once
# when you make a service unavailable, it won't run under the Oracle crs # for automatic startup, failover, or restart srvctl disable database -d rac srvctl disable instance -d rac -i "rac1,rac2" srvctl disable service -d rac -s hrserv,marketing # -n node name srvctl disable asm -n crmnode1 -i asm1
Switching Between the Database Automatic and Manual Policies When AUTO_START attribute is set to 2 (MANUAL) for a resource, crs will not automatically start it on reboot. # to display the current policy srvctl config database -d rac -a .. POLICY: AUTOMATIC # change the current policy to another one srvctl modify database d database_name -y AUTOMATIC|MANUAL
Customizing Resource Parameters (like AUTO_START) You can customize the following resource parameters for database or ASM instances, databases, services, and service members: # when 1 (0), it atuo-restarts on system reboot AUTO_START as # restart attempts before relocate (1) RESTART_ATTEMPTS ra
# application up-time to be considered stable by crs UPTIME_THRESHOLD ut
# 1. retreive resource name crs_stat t # 2. update the OCR with the right attribute values for your resources crs_register ora..inst -update -o as=2,ra=1,ut=7d crs_register ora..asm -update -o as=2,ra=1,ut=7d crs_register ora..db -update -o as=2,ra=1,ut=7d crs_register ora..cs -update -o as=2,ra=0 crs_register ora..svr -update -o as=2,ra=0
Page 474
Handling Initialization Parameter Files in RAC

Setting Server Parameter File Parameter Values for Real Application Clusters Practically, each instance uses its own PFILE file that contains only one parameter called SPFILE. The SPFILE parameter points to the shared SPFILE on your shared storage. /* about the parameters in spfile in RAC */ # to know wheather spfile or pfile was used when startup SELECT DECODE(value, NULL, 'PFILE', 'SPFILE:'|| value) "Init File Type" FROM sys.v_$parameter WHERE name = 'spfile'; -- spfile is shared by all instances cat $ORACLE_HOME/dbs/initrac1.ora SPFILE='+DG1/rac/spfilerac.ora' -- example ALTER SYSTEM SET OPEN_CURSORS=500 COMMENT='Changed by Ahmed for testing on 16-dec-09' SCOPE=BOTH SID='*'; ALTER SYSTEM SET OPEN_CURSORS=1000 SPFILE=BOTH SID='rac1'; -- to remove a paramter from spifle ALTER SYSTEM RESET timed_statistics SCOPE=SPFILE SID='*'; ALTER SYSTEM SET timed_statistics = '' SCOPE=SPFILE; Parameters Used in RAC Databases CLUSTER_DATABASE must be TRUE CLUSTER_DATABASE_INSTANCES CLUSTER_INTERCONNECTS DB_NAME DISPATCHERS INSTANCE_NAME SERVICE_NAMES SESSIONS_PER_USER If SESSIONS_PER_USER is set to 1 for a user, the user can log on to the database more than once as long as each connection is from a different instance. THREAD SPFILE MAX_COMMIT_PROPAGATION_DELAY deprecated in 10g R2
Parameters that Must Have Identical Settings on All Instances ACTIVE_INSTANCE_COUNT ARCHIVE_LAG_TARGET CLUSTER_DATABASE CLUSTER_DATABASE_INSTANCES CONTROL_FILES DB_BLOCK_SIZE DB_DOMAIN DB_FILES DB_NAME DB_RECOVERY_FILE_DEST DB_RECOVERY_FILE_DEST_SIZE
Page 475
number of instances in the RAC
DB_UNIQUE_NAME UNDO_MANAGEMENT Parameters That Must Have Unique Settings on All Instances THREAD ROLLBACK_SEGMENTS UNDO_TABLESPACE if automatic is used
Parameters that Should Have Identical Settings on All Instances It is highly recommended to set same value for parameters in the following list in all the instances: ARCHIVE_LAG_TARGET LICENSE_MAX_USERS LOG_ARCHIVE_FORMAT SPFILE TRACE_ENABLED UNDO_RETENTION ASM Instance Initialization Parameters and RAC CLUSTER_DATABASE must be TRUE ASM_DISKSTRING Multiple instances can have different values (not recommended) ASM_POWER_LIMIT Multiple instances can have different values export ORACLE_SID=+ASM1 sqlplus /nolog conn / as sysdba select name, value from v$parameter where upper(name) in ('CLUSTER_DATABASE','ASM_DISKSTRING','ASM_POWER_LIMIT');
Dropping a RAC Database

Option 1: After making sure the crs and the database are up and running, issue dbca and delete the database. Option 2: issue the commands below: startup mount exclusive restrict; drop database; # to delete archived log and the backups as well drop database including backups; srvctl remove instance -d rac -i rac1 srvctl remove database -d rac
Page 476
Workload Management in RAC

Types of Workload Distribution
Connection load balancing: o o o Client-side connect-time load balancing Client-side connect-time failover Server-side connect-time load balancing
Run-time connection load balancing can be implemented by using connection pools in middle tier.
Connection Load Balancing

Client-Side Load Balancing and Failover If you use connect-time failover, do not set GLOBAL_DBNAME in the listener.ora # to enable Client-Side Load Balancing ONLY (i.e., no failover): send request # randomly to one listener from the list. CSLB_ONLY = (DESCRIPTION = (LOAD_BALANCE = ON) (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip)(PORT = 1521)) (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = racdb) ) ) # use this settings for RAC database # to enable failover (using TCP/IP time-out, the client make sure the host is # alive before sending connection requrest) (in RAC, you use vip addresses) CSLB_ONLY = (DESCRIPTION = (LOAD_BALANCE = ON) (FAILOVER = ON) (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip)(PORT = 1521)) (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = racdb) ) )
Server-Side Load Balancing Target: which listener should take the connection. -- Configuring Server-side connection load balancing -- 1) define the service which will be used for the connection load balancing -- 2) Add entries in every client's TNSNAMES.ORA file for the new alias
Page 477
HRSERV = (DESCRIPTION = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip.mydomain.com)(PORT = 1521)) (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip.mydomain.com)(PORT = 1521)) ) (LOAD_BALANCE = ON) -- not related to server-side load balancing (FAILOVER = ON) -- not related to server-side load balancing (CONNECT_DATA = (SERVICE_NAME = hrserv) ) ) -- 3) Add entries in the TNSNAMES.ORA file of every node to -include the REMOTE_LISTENER setting LISTENERS_RAC = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip.mydomain.com)(PORT = 1521)) (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip.mydomain.com)(PORT = 1521)) ) -- 4) set the parameter to make PMON automatically register the database -with all the listeners in the nodes ALTER SYSTEM SET REMOTE_LISTENER =LISTENERS_RAC SID='*' SCOPE=BOTH;
-- setting the default connection load balancing goal for the service hrserv -- to LONG (long-lived apps like forms)(default), SHORT (short-lived apps) execute dbms_service.modify_service (service_name => 'hrserv' , clb_goal => dbms_service.clb_goal_long); execute dbms_service.modify_service (service_name => 'hrserv' , clb_goal => dbms_service.clb_goal_short); -- view it (name is case sensitive) SELECT CLB_GOAL FROM DBA_SERVICES WHERE NAME='hrserv';
Fast Application Notification (FAN)

FAN publishing load balancing advisory events and UP and DOWN info about services, instances and nodes to client applicaitons. You can take advantage of FAN events in the following three ways: 1. 2. Application can take advantage of FAN without any programmatic changes by utilizing an integrated Oracle Client like: Oracle Database 10g JDBC, ODP.NET, and OCI. Applications can take advantage of FAN programmatically by using the Oracle Notification Service (ONS) Application Programming Interface (API) to subscribe to FAN events and execute event handling actions upon the receipt of an event. You can implement FAN with server side callouts on your database tier.
3.
Using Fast Application Notification Callouts

FAN callouts are server side executables that RAC executes immediately when high availability events occur.
Page 478
# place an executable in the directory CRS_home/racg/usrco in all nodes vi /u01/crs/racg/usrco/callout.sh #! /bin/ksh FAN_LOGFILE=/home/oracle/log/rac_`hostname`.log echo $* "reported="`date` >> $FAN_LOGFILE & # place an executable in the directory CRS_home/racg/usrco in all nodes # example source is Oracle Documentation 10g R1 vi /u01/crs/racg/usrco/callout.sh #!/usr/bin/sh# # Description: wrapper script to enable RAC event logging and notification # to generic third-party systems. The script showcases two possible # methods to enable local or remote logging/notification of RAC# detected events.# AWK=/usr/bin/awk MY_CRS_HOME=/private/oracle/crs # Scan and parse arglist:# for ARGS in $*; do PROPERTY=ècho $ARGS | $AWK -F"=" '{print $1}'` VALUE=ècho $ARGS | $AWK -F"=" '{print $2}'` #> map EVTTYPE to EVENT_TYP, NODE to HOST: case $PROPERTY in #> EVENT_TYP is one of: NODE, DATABASE, INSTANCE, SERVICE, SERVICEMEMBER EVENT_TYP | event_typ) NOTIFY_EVENT_TYP=$VALUE ;; VERSION | version) NOTIFY_VERSION=$VALUE ;; SERVICE | service) NOTIFY_SERVICE=$VALUE ;; DATABASE | database) NOTIFY_DBNAME=$VALUE ;; INSTANCE | instance) NOTIFY_INSTANCE=$VALUE ;; HOST | host) NOTIFY_HOST=$VALUE ;; STATUS | status) NOTIFY_STATUS=$VALUE ;; TIMESTAMP | timestamp) NOTIFY_SVRLOGDATE=$VALUE ;; esac done
# # # # # #
################################################### [1] Notification Method 1: On-cluster file logging ################################################### This section simply writes one-line entries for each event published by RAC, and the log is written to standard RAC log directory. It will blindly record all RAC events, regardless of state (UP, DOWN or NOT_RESTARTING):
RACEVT_LOGFILE=$MY_CRS_HOME/racg/log/rac_${NOTIFY_SERVICE}_uptime.log echo RAC$v$NOTIFY_VERSION$: $NOTIFY_STATUS event, type "$NOTIFY_EVENT_TYP", \ ìf [ -n "$NOTIFY_SERVICE" ]; then \ echo "for service $NOTIFY_SERVICE" fi` \ \[ìf [ -n "$NOTIFY_INSTANCE" ]; then \ echo "inst: $NOTIFY_INSTANCE" fi` \ ìf [ -n "$NOTIFY_DATABASE" ]; then \ echo "db: $NOTIFY_DATABASE" fi` \ ìf [ -n "$NOTIFY_HOST" ]; then \ echo "db: $NOTIFY_HOST" fi` \
Page 479
\] received on $NOTIFY_SVRLOGDATE >>
$RACEVT_LOGFILE
# # # # # # # # # #
######################################################## [2] Notification Method 2: On-cluster program execution ######################################################## Let's assume you have a custom client program in /tmp (say logTicket) to which you can pass certain arguments. This program connects to a customer-service application that processes incident tickets for your IT department: % /tmp/logTicket {serverside_timestamp} \ {databasename} {servicename} \ {instancename} {hostname}#
# Let us also assume that a ticket would be logged only for NOT_RESTARTING # events, as they are the ones that exceeded RAC-monitored timeouts and # seriously need human intervention for full resolution.# # ------------------# ONE SOLUTION TO [2]: # ------------------if [ $NOTIFY_STATUS = "NOT_RESTARTING" -o $NOTIFY_STATUS = "not_restarting" ]; then /tmp/logTicket $NOTIFY_SVRLOGDATE $NOTIFY_DBNAME \ $NOTIFY_SERVICE \ $NOTIFY_INSTANCE $NOTIFY_HOST >> $RACEVT_LOGFILE fi
Configuring the Server-Side ONS

When useocr=on in ons.config, racgons add_config command store this data in OCR. # target: to add rac3 node in the RAC ONS # (1) obtain the remote nodes from rac1,rac2 # and set useocr=on cat $CRS_HOME/opmn/conf/ons.config # (2) in rac3: racgons add_config rac1:6200 rac2:6200 racgons add_config rac3:6200 # (3) in rac1, rac2: onsctl reconfig
Administering Load Balancing Advisory

The load balancing advisory has the task of advising the direction of incoming work to the RAC instances that provide optimal quality of service for that work. To test the load balancing, you can use the scripts in the appendix to apply load and the following section for monitoring. -- view load balancing gaol of a current service -- NONE means load balancing advisory is disabled SELECT NAME, GOAL, CLB_GOAL FROM DBA_SERVICES; -- SERVICE TIME: response-time based, like online shopping
Page 480
execute dbms_service.modify_service (service_name => , goal => dbms_service.goal_service_time , clb_goal => dbms_service.clb_goal_short); -- THROUGHPUT: rat at which jobs are completed, like execute dbms_service.modify_service (service_name => , goal => dbms_service.goal_throughput , clb_goal => dbms_service.clb_goal_long); -- NONE: LBA disabled execute dbms_service.modify_service (service_name => , goal => dbms_service.goal_none , clb_goal => dbms_service.clb_goal_long);
'hrserv' -
batch processing 'hrserv' -
'hrserv' -
Monitoring Load Balancing Advisory

V$SERVICEMETRIC : service metrics on the most 5-second and 1-minute intervals V$SERVICEMETRIC_HISTORY : recent history of the metric values SELECT service_name ,TO_CHAR(begin_time,'hh24:mi:ss') beg_hist ,TO_CHAR(end_time,'hh24:mi:ss') end_hist ,inst_id ,goodness ,delta ,flags ,cpupercall ,dbtimepercall ,callspersec ,dbtimepersec FROM gv$servicemetric WHERE service_name ='hrserv' ORDER BY service_name, begin_time DESC, inst_id; SELECT SERVICE_NAME, to_char(BEGIN_TIME,'hh24:mi:ss') BEGIN_TIME, to_char(END_TIME,'hh24:mi:ss') END_TIME, INTSIZE_CSEC, GROUP_ID "Service Metric Group", CPUPERCALL "CPU time per call", DBTIMEPERCALL "Elapsed time per call", CALLSPERSEC "User Calls/s", DBTIMEPERSEC "DB Time/s" from V$SERVICEMETRIC_HISTORY order by SERVICE_NAME, BEGIN_TIME desc; -- aggregated SELECT SERVICE_NAME, GROUP_ID "Service Metric Group", round(sum(CPUPERCALL),2) "CPU time per call", round(sum(DBTIMEPERCALL),2) "Elapsed time per call", round(sum(CALLSPERSEC),2) "User Calls/s", round(sum(DBTIMEPERSEC),2) "DB Time/s" from V$SERVICEMETRIC_HISTORY group by SERVICE_NAME, GROUP_ID order by SERVICE_NAME;
Page 481
Transparent Application Failover (TAF)

TAF Basic Configuration without FAN (From Client Side) FAILOVER_MODE must be configured in tnsnames.ora If using TAF, do not set the GLOBAL_DBNAME parameter in the listener.ora # when you create the service srvctl add service -d rac -s hrserv -r rac1,rac2 -P BASIC srvctl start service -d rac -s hrserv # in the client tnsnames.ora # TYPE: SELECT, SESSION hrserv = (DESCRIPTION =(FAILOVER=ON)(LOAD_BALANCE=ON) (ADDRESS=(PROTOCOL=TCP)(HOST=rac1-VIP)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=rac2-VIP)(PORT=1521)) (CONNECT_DATA = (SERVICE_NAME = hrserv) (FAILOVER_MODE = (TYPE=SESSION) (METHOD=BASIC) (RETRIES=100) (DELAY=5)))) TAF Basic Configuration with FAN (Server-Side) You not need to specify TAF parameters in the tnsnames.ora srvctl add service -d rac -s hrserv -r rac1,rac2 srvctl start service -d RACDB -s AP exec dbms_service.modify_service ( ,service_name => 'hrserv' ,aq_ha_notifications => true ,failover_method => dbms_service.failover_method_basic ,failover_type => dbms_service.failover_type_session ,failover_retries => 100, failover_delay => 5 ,clb_goal => dbms_service.clb_goal_long); hrserv = (DESCRIPTION =(FAILOVER=ON)(LOAD_BALANCE=ON) (ADDRESS=(PROTOCOL=TCP)(HOST=rac1-VIP)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=rac2-VIP)(PORT=1521)) (CONNECT_DATA = (SERVICE_NAME = hrserv))) TAF Preconnect Configuration Fast failover but more resources. The shadow service is always named using the format <service_name>_PRECONNECT. # configure the preferred and available nodes srvctl add service -d rac -s hrserv -r rac1 -a rac2 -P PRECONNECT srvctl start service -d rac -s hrserv HRSERV = (DESCRIPTION =(FAILOVER=ON)(LOAD_BALANCE=ON)
Page 482
(ADDRESS=(PROTOCOL=TCP)(HOST=rac1-vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=rac2-vip)(PORT=1521)) (CONNECT_DATA = (SERVICE_NAME = hrserv) (FAILOVER_MODE = (BACKUP=HRSERV_PRECONNECT) (TYPE=SESSION)(METHOD=PRECONNECT)))) HRSERV_PRECONNECT = (DESCRIPTION =(FAILOVER=ON)(LOAD_BALANCE=ON) (ADDRESS=(PROTOCOL=TCP)(HOST=rac1-vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=rac2-vip)(PORT=1521)) (CONNECT_DATA = (SERVICE_NAME = HRSERV_PRECONNECT))) Verifying TAF Configuration SELECT machine, failover_method, failover_type, failed_over, service_name, COUNT(*) FROM v$session GROUP BY machine, failover_method, failover_type, failed_over, service_name; select instance_name from v$instance;
Enabling Distributed Transaction Processing for Services

# For services that you are going to use for distributed transaction processing, # define only one instance as the preferred instance srvctl add service -d crm -s xa_01.service.us.oracle.com -r RAC01 -a RAC02, RAC03 # mark the service for distributed trans processing by setting DTP to TRUE: execute dbms_service.modify_service(service_name =>'xa_01.mydomain.com',DTP=>TRUE);
Page 483
Administering Services
Service Attributes
Global unique name Network name Load Balancing Advisory goal: best service quality (service response time), or best throughput (how much work is completed in a unit of time). Distributed transactions flag Advance queuing notification characteristics for OCI and ODP.NET clients Failover characteristics Connection load-balancing algorithm: SHORT: Use Load Balancing Advisory, LONG: Using session count by service Threshold: for response time and CPU consumption Priority: services to consumer groups mapping High-availability configuration: how the service is distributed across instances when the system first starts
Administering Services with DBCA

1. On the DBCA Welcome page, select the Oracle Real Application Clusters option ->Next 2. On the DBCA Operations: select Services Management ->Next 3. On the DBCA List of Databases: select the cluster database ->Next -- confirm select value from v$parameter where upper(name)='SERVICE_NAMES'
Administering Services with PL/SQL

/* Create a Service (not a recommended method in RAC)*/ -- some attributes can only be modified using PL/SQL -- service_name and network_name are mandatory -- others are optional will NULL defaults begin DBMS_SERVICE.CREATE_SERVICE( service_name =>'olapserv', network_name =>'olapserv', -- TNS name for connections to the service goal =>DBMS_SERVICE.GOAL_SERVICE_TIME, -- or GOAL_THROUGHPUT, GOAL_NONE dtp =>FALSE, -- is it for distributed transaction aq_ha_notifications =>TRUE, -- should HA events sent to AQ failover_method =>DBMS_SERVICE.FAILOVER_METHOD_BASIC, -- or _NONE failover_type =>DBMS_SERVICE.FAILOVER_TYPE_SESSION, -- or _NONE _SELECT failover_retries =>10, failover_delay=>1 , -- in seconds clb_goal=>DBMS_SERVICE.CLB_GOAL_SHORT -- or CLB_GOAL_LONG ); end; Oracle DBA Code Examples
Page 484
/ select name from dba_services ; /* Modify a Service Charactaristic */ -- some attributes can only be modified using PL/SQL -- if stopped, start it before modify begin DBMS_SERVICE.MODIFY_SERVICE( service_name =>'olapserv', goal =>DBMS_SERVICE.GOAL_SERVICE_TIME, -- or GOAL_THROUGHPUT, GOAL_NONE dtp =>FALSE, -- is it for distributed transaction aq_ha_notifications =>TRUE, -- should HA events sent to AQ failover_method =>DBMS_SERVICE.FAILOVER_METHOD_BASIC, -- or _NONE failover_type =>DBMS_SERVICE.FAILOVER_TYPE_SESSION, -- or _NONE _SELECT failover_retries =>10, failover_delay=>1 , -- in seconds clb_goal=>DBMS_SERVICE.CLB_GOAL_SHORT -- or CLB_GOAL_LONG ); end; /
/* Delete Service */ begin DBMS_SERVICE.DELETE_SERVICE( service_name =>'olapserv'); end; /
/* Start a service */ begin DBMS_SERVICE.START_SERVICE(service_name=>'hrserv', instance_name => DBMS_SERVICE.ALL_INSTANCES, -- if NULL current inst ); end; /
/* Stop Service */ begin DBMS_SERVICE.STOP_SERVICE(service_name=>'hrserv', instance_name => DBMS_SERVICE.ALL_INSTANCES, -- if NULL current inst ); end; /
/* Disconnect Sessions */ -- sessions connected to the service are terminated begin -- CAUTION: control doesn't return till all sessions are terminated DBMS_SERVICE.DISCONNECT_SESSION('hrserv'); end; /
Page 485
Administering Services with SRVCTL

Creating a Service srvctl add service -d dbname -s service_name -r preferred_list [-a available_list] [-P TAF_policy] taf policy: NONE, BASIC, or PRECONNECT /* Creating Services */ # it does NOT add the required tnsnames.ora settings srvctl add service d rac s GL -r RAC02 -a RAC01 srvctl add service d rac s ERP r RAC01,RAC02 -a RAC03,RAC04 /* Starting and Stopping Services */ srvctl start service -d dbname [-s service_name_list] [-i inst_name] [-o start_options] srvctl stop service -d dbname [-s service_name_list [-i inst_name]] [-c connect_str] /* Enabling and Disabling Services */ srvctl enable service -d dbname -s service_name_list [-i inst_name] srvctl disable service -d dbname -s service_name_list [-i inst_name] /* Relocating Services */ srvctl relocate service -d rac -s hrserv -i rac2 -t rac3 /* upgrading an available instance as a preferred instance for a Service*/ -- see also next section srvctl modify service d rac s hrserv -i rac3 r /* Obtaining the Statuses of Services */ srvctl status service -d rac -s hrserv /* Obtaining the Configuration of Services */ srvctl config service -d rac -s hrserv -a /* Removing Services */ rvctl stop service -d rac -s hrserv rvctl remove service -d rac -s hrserv
Controlling the Preferred and Available Instances

/* Upgrading an Available instance as a Preferred instance for a Service */ -- -r upgrade instance to preferred srvctl modify service d rac s hrserv -i rac3 r
/* the scenario below makes: - i1-i4 preferred, i5-i6 available for dwserv - i5-i6 preferred, i1-i4 available for hrserv */ -- changes take effect on next services restart -- if used, -f Disconnect all sessions during stop srvctl modify service d PROD s dwserv n i I1,I2,I3,I4 a I5,I6 srvctl modify service d PROD s hrserv n i I5,I6 a I1,I2,I3,I4 -- f stops the services globally on your cluster
Page 486
srvctl stop service d rac s dwserv,hrserv -f srvctl start service d rac s dwserv,hrserv
Using Services with Client Applications

/* TNS connection descriptor */ -- LOAD_BALANCE=on only enables a client-side connection load balancing ERP=(DESCRIPTION= (LOAD_BALANCE=on) (ADDRESS=(PROTOCOL=TCP)(HOST=node-1vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-2vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-3vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-4vip)(PORT=1521)) (CONNECT_DATA=(SERVICE_NAME=ERP))) /* JDBC connection description (1) */ url="jdbc:oracle:oci:@ERP" /* JDBC connection description (2) */ url="jdbc:oracle:thin:@(DESCRIPTION= (LOAD_BALANCE=on) (ADDRESS=(PROTOCOL=TCP)(HOST=node-1vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-2vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-3vip)(PORT=1521)) (ADDRESS=(PROTOCOL=TCP)(HOST=node-4vip)(PORT=1521)) (CONNECT_DATA=(SERVICE_NAME=ERP)))"
Services and the Scheduler

BEGIN DBMS_SCHEDULER.CREATE_JOB_CLASS ( job_class_name => 'my_jobs_class', SERVICE => 'hrserv'); END; / BEGIN dbms_scheduler.set_attribute( name => 'my_jobs_class', attribute => 'service', value => 'hrserv'); END; / begin dbms_scheduler.create_job ( job_name => 'My_Job', job_type => 'plsql_block', -- stored_procedure, plsql_block, executable, chain job_action => 'begin null; end;', job_class => 'my_jobs_class', start_date => SYSDATE+1, repeat_interval => 'FREQ=MINUTELY; INTERVAL=2'); end;
Page 487
/ BEGIN dbms_scheduler.set_attribute( name => 'My_Job', attribute => 'INSTANCE_STICKINESS', value => TRUE); END; / BEGIN DBMS_SCHEDULER.DROP_JOB('My_Job'); END; / BEGIN DBMS_SCHEDULER.DROP_JOB_CLASS('my_jobs_class'); END; / DBMS_SCHEDULER.CREATE_JOB_CLASS( JOB_CLASS_NAME => 'HOT_BATCH_CLASS', RESOURCE_CONSUMER_GROUP => NULL , SERVICE => 'HOT_BATCH_SERV' , LOGGING_LEVEL => DBMS_SCHEDULER.LOGGING_RUNS, LOG_HISTORY => 30, COMMENTS => 'P1 batch'); DBMS_SCHEDULER.CREATE_JOB( JOB_NAME => 'my_report_job', JOB_TYPE => 'stored_procedure', JOB_ACTION => 'my_name.my_proc();', NUMBER_OF_ARGUMENTS => 4, START_DATE => SYSDATE+1, REPEAT_INTERVAL => 5, END_DATE => SYSDATE+30, JOB_CLASS => 'HOT_BATCH_CLASS', ENABLED => TRUE, AUTO_DROP => false, COMMENTS => 'daily status');
Measuring Performance by Service Using the AWR

/* Enabling Gather statiscitc on Service-Module-Action combination */ -- to display currently connected services, modules and actions select SID, USERNAME, SERVICE_NAME, MODULE, ACTION from V$SESSION where SERVICE_NAME in ('hrserv','oeserv') -- service name and module name are mandatory begin DBMS_MONITOR.SERV_MOD_ACT_STAT_ENABLE(SERVICE_NAME => 'hrserv', MODULE_NAME=>'PAYROLL', ACTION_NAME => 'EXCEPTIONS PAY'); end; / -- gather stats for PAYROLL module and ACTION whose name is null begin DBMS_MONITOR.SERV_MOD_ACT_STAT_ENABLE(SERVICE_NAME => 'hrserv',
Page 488
MODULE_NAME=>'PAYROLL', ACTION_NAME => NULL); end; / -- gather stats for PAYROLL module and All its ACTIONs begin DBMS_MONITOR.SERV_MOD_ACT_STAT_ENABLE(SERVICE_NAME => 'hrserv', MODULE_NAME=>'PAYROLL', ACTION_NAME => '###ALL_ACTIONS'); end; / -- to view enabled monitorings -- types: SERVICE, SERVICE_MODULE, SERVICE_MODULE_ACTION select A.AGGREGATION_TYPE, A.PRIMARY_ID , A.QUALIFIER_ID1 , A.QUALIFIER_ID2 from DBA_ENABLED_AGGREGATIONS a -- to view gathered stats select S.AGGREGATION_TYPE, S.SERVICE_NAME, S.MODULE, S.ACTION, N.CLASS, decode(n.CLASS, '1','User','2','Redo','4','Enqueue','8','Cache','16','OS','32','RAC','64','SQL ','128','Debug', N.CLASS) STAT_CLASS, S.STAT_NAME, S.VALUE from V$SERV_MOD_ACT_STATS s, V$STATNAME n where S.STAT_ID = N.STAT_ID order by N.CLASS, S.STAT_ID -- call times and performance statistics views: V$SERVICE_STATS V$SERVICE_EVENTS V$SERVICE_WAIT_CLASSES V$SERVICEMETRIC V$SERVICEMETRIC_HISTORY /* To Disable Cumulative Stats */ -- stats will be removed from V$SERV_MOD_ACT_STATS begin DBMS_MONITOR.SERV_MOD_ACT_STAT_DISABLE(SERVICE_NAME => 'hrserv', MODULE_NAME=>'PAYROLL', ACTION_NAME => 'EXCEPTIONS PAY'); end; /
/* Service Quality Statistics */ -- script from Oracle documentation -- provides service quality statistics every five seconds SET PAGESIZE 60 COLSEP '|' NUMWIDTH 8 LINESIZE 132 VERIFY OFF FEEDBACK OFF COLUMN service_name FORMAT A20 TRUNCATED HEADING 'Service' COLUMN begin_time HEADING 'Begin Time' FORMAT A10 COLUMN end_time HEADING 'End Time' FORMAT A10 COLUMN instance_name HEADING 'Instance' FORMAT A10 COLUMN service_time HEADING 'Service Time|mSec/Call' FORMAT 999999999 COLUMN throughput HEADING 'Calls/sec'FORMAT 99.99 BREAK ON service_name SKIP 1 SELECT
Page 489
service_name , TO_CHAR(begin_time, 'HH:MI:SS') begin_time , TO_CHAR(end_time, 'HH:MI:SS') end_time , instance_name , elapsedpercall service_time , callspersec throughput FROM gv$instance i , gv$active_services s , gv$servicemetric m WHERE s.inst_id = m.inst_id AND s.name_hash = m.service_name_hash AND i.inst_id = m.inst_id AND m.group_id = 10 ORDER BY service_name , i.inst_id , begin_time ;
Service Thresholds and Alerts

For the services, you can set alerts for: o o ELAPSED_TIME_PER_CALL Elapsed time for each user call for each service (micro s) CPU_TIME_PER_CALL CPU time for each user call for each service (micro s)
Thresholds must be set on each instance supporting the service -- thresholds must be set on each instance supporting the service begin DBMS_SERVER_ALERT.SET_THRESHOLD( METRICS_ID => DBMS_SERVER_ALERT.ELAPSED_TIME_PER_CALL , WARNING_OPERATOR => dbms_server_alert.operator_ge , WARNING_VALUE => '500000' -- = 0.5 seconds , CRITICAL_OPERATOR => dbms_server_alert.operator_ge , CRITICAL_VALUE => '750000' = 0.75 seconds , OBSERVATION_PERIOD => 30 -- in mins , CONSECUTIVE_OCCURRENCES => 5 -- tolerance occurance before alerts , INSTANCE_NAME => NULL -- must be NULL in this case , OBJECT_TYPE => dbms_server_alert.object_type_service , OBJECT_NAME => 'hrserv'); end; / -- Verify the threshold configuration SELECT METRICS_NAME, INSTANCE_NAME, WARNING_VALUE, CRITICAL_VALUE, OBSERVATION_PERIOD from dba_thresholds where OBJECT_NAME => 'hrserv' ; -- most recent 60 s SELECT service_name, elapsedpercall, cpupercall FROM V$SERVICEMETRIC; -- last hour SELECT service_name, elapsedpercall, cpupercall FROM V$SERVICEMETRIC_HISTORY;
Page 490
Service Performance Views

There are more than 300 performance-related statistics that are tracked and visible in V$SYSSTAT. Of these, 28 statistics are tracked for services. To see the statistics measured for services, run the following query: SELECT DISTINCT stat_name FROM v$service_stats. For service metrics, see Monitoring Load Balancing Advisory -- general stats per service -- DB time (response time)= DB CPU + wait time select V.SERVICE_NAME, decode(n.CLASS, '1','User','2','Redo','4','Enqueue','8','Cache','16','OS','32','RAC','64','SQL ','128','Debug', N.CLASS) STAT_CLASS , V.STAT_NAME , VALUE from V$SERVICE_STATS V , V$STATNAME N where V.STAT_ID = N.STAT_ID order by 1,2;
-- aggregated waits (by wait class) SELECT * FROM V$SERVICE_WAIT_CLASS WHERE SERVICE_NAME NOT LIKE 'SYS$%' ORDER BY SERVICE_NAME, TIME_WAITED DESC
-- aggregated waits (by wait event) SELECT * FROM V$SERVICE_EVENT WHERE SERVICE_NAME NOT LIKE 'SYS$%' AND EVENT NOT IN ('SQL*Net message from client') ORDER BY SERVICE_NAME, TIME_WAITED DESC;
-- stats for a specific combination of service/module/action -- When statistics collection for specific modules and actions is enabled select * from V$SERV_MOD_ACT_STATS ORDER BY 1,2,3,4 -- information about enabled on-demand statistic aggregation select * from DBA_ENABLED_AGGREGATIONS; -- information about enabled SQL traces selct * from DBA_ENABLED_TRACES;
Restricted Session and Services

If the restricted instance is manually registered in the listener, the incoming connection requrests may be routed to the instance, refused by the instance and returns error. If the restricted instance is dynamically registered in the listener, the incoming connection requrests will not be routed to the instance, even for users with RESTRICTED SESSION privilege.
Page 491
Configuring Recovery Manager and Archiving

Backup Possible Distributions in RAC
Network backup server: dedicated backup server manages backups for the whole RAC. One local drive: one node has access to a local backup appliance. /* Multiple Drives CFS Backup Scheme */ -- cfs = all nodes have their archived log in shared storage -- one time setup CONFIGURE DEVICE TYPE sbt PARALLELISM 3; CONFIGURE DEFAULT DEVICE TYPE TO sbt; CONFIGURE CHANNEL 1 DEVICE TYPE sbt CONNECT 'user1/passwd1@node1'; CONFIGURE CHANNEL 2 DEVICE TYPE sbt CONNECT 'user2/passwd2@node2'; CONFIGURE CHANNEL 3 DEVICE TYPE sbt CONNECT 'user3/passwd3@node3'; -- backup is then distributed in all backup tapes in the cluster BACKUP DATABASE PLUS ARCHIVELOG DELETE INPUT; Multiple drives: each node has access to a local backup appliance.
RMAN Restore Scenarios for Real Application Clusters

Cluster File System Restore Scheme 1. 2. Make the Backup media and all the Archived Log files available to the available node. Then Restore using RESOTRE DATABASE and RECOVER DATABASE commands.
Non-Cluster File System Restore Scheme 1. 2. You must configure a network file system file so that the recovery node can read the archiving directories on the remaining nodes. Then Restore using RESOTRE DATABASE and RECOVER DATABASE commands.
RMAN and Oracle Net in Real Application Clusters

Connection server must be dedicated. Each net service name must specify only one instance.
Connecting to Specific Node

rman TARGET=SYS/ORACLE@RAC1 CATALOG=RMAN/RMAN@RMAN
Page 492
Instance Recovery in Real Application Clusters

Single Node Failure in Real Application Clusters When one instance performs recovery for another instance, the surviving instance reads online redo logs generated by the failed instance and uses that information to ensure that committed transactions are recorded in the database.
Multiple-Node Failures in Real Application Clusters If all instances of an Oracle RAC database fail, then Oracle automatically recovers the instances the next time one instance opens the database.
Configuring the RMAN Snapshot Control File Location

The snapshot control file is a temporary snapshot control file that RMAN creates to resynchronize from a read-consistent version of the control file. -- configured location of the snapshot control file SHOW SNAPSHOT CONTROLFILE NAME; -- to change it (globally in all nodes) CONFIGURE SNAPSHOT CONTROLFILE NAME TO '$ORACLE_HOME/dbs/scf/snap_prod.cf';
Configuring the RMAN Control File and SPFILE Autobackup Feature

# switch it on CONFIGURE CONTROLFILE AUTOBACKUP ON; # Auto control file and spfile backup must be configured in a shared location. CONFIGURE CONTROLFILE AUTOBACKUP FORMAT = '/u02/oradata/RAC/cf_%F'; CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '+FRA'; # in the recovery destination CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '%F';
Configuring Channels for RMAN in Real Application Clusters

Configuring Channels to use Automatic Workload Balancing CONFIGURE DEVICE TYPE [disk | sbt] PARALLELISM number of channels; # configure the SQL Net service name with load balancing turned on CONFIGURE CHANNEL DEVICE TYPE SBT CONNECT 'sys/pwd1@<service_name>' Configuring Channels to Use a Specific Instance # manual instance assignment # To configure channels to use a specific instance, use the following syntax: CONFIGURE DEVICE TYPE sbt PARALLELISM 2; CONFIGURE CHANNEL 1 DEVICE TYPE sbt CONNECT 'SYS/pss@rac1' CONFIGURE CHANNEL 2 DEVICE TYPE sbt CONNECT 'SYS/pss@rac2' # Automatic instance assignment
Page 493
# you can configure special service for rman jobs # service workload will be used CONFIGURE DEFAULT DEVICE TYPE TO sbt; CONFIGURE DEVICE TYPE sbt PARALLELISM 3; CONFIGURE CHANNEL DEVICE TYPE sbt CONNECT='sys/rac@rmanserv'; # also, parallel configuration will distribute the job on multiple instances CONFIGURE DEVICE TYPE sbt PARALLELISM 3;
# configuring channels manually RUN { ALLOCATE CHANNEL CH1 CONNECT 'user1/pwd1@node1'; ALLOCATE CHANNEL CH2 CONNECT 'user2/pwd2@node2'; ALLOCATE CHANNEL CH3 CONNECT 'user3/pwd3@node3'; BACKUP DATABASE PLUS ARCHIVED LOG; }
Node Affinity Awareness of Fast Connections

When deciding which channel to use to back up a particular datafile, RMAN gives preference to the nodes with faster access to the datafiles that you want to back up.
Archived Redo Log File Conventions in RAC

show parameter LOG_ARCHIVE_FORMAT alter system set LOG_ARCHIVE_FORMAT='log_%t_%s_%r.arc' scope=spfile sid='*';
Archive Redo Log Configuration Scenarios

Automatic Storage Management and CFS Archiving Scheme The preferred configuration for RAC is to use Automatic Storage Management (ASM) for a recovery area with a different disk group for your recovery set than for your datafiles. show parameter archive_dest show parameter recovery # each node archives to a directory with the same name on all instances sid1.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest" sid2.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest" sid3.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest" Non-Cluster File System Local Archiving Scheme In the non-cluster file system local archiving scheme, each node archives to a uniquely named local directory. The directory in each node can be read only mounted from the other nodes. # archived log stored locally and the other locations can be read by the node # (for recovery) sid1.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest_1" sid2.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest_2" sid3.LOG_ARCHIVE_DEST_1="LOCATION=/arc_dest_3"
Page 494
Changing the Archiving Mode in Real Application Clusters

select log_mode from v$database; srvctl stop database -d rac Back up the database set LOG_ARCHIVE_DEST_n show parameter LOG_ARCHIVE_DEST -- no instance in OPEN mode STARTUP MOUNT ALTER DATABASE ARCHIVELOG; SHUTDOWN IMMEDIATE Back up the database srvctl start database -d rac
Deleting Archived Redo Logs after a Successful Backup

# if automatic channels are configured DELETE ARCHIVELOG ALL BACKED UP n TIMES TO DEVICE TYPE device_type; # configuring the channels manually ALLOCATE CHANNEL FOR MAINTENANCE DEVICE ALLOCATE CHANNEL FOR MAINTENANCE DEVICE ALLOCATE CHANNEL FOR MAINTENANCE DEVICE DELETE ARCHIVELOG ALL BACKED UP n TIMES
TYPE DISK TYPE DISK TYPE DISK TO DEVICE
CONNECT 'SYS/oracle@node1'; CONNECT 'SYS/oracle@node2'; CONNECT 'SYS/oracle@node3'; TYPE device_type;
Monitoring the Archiver Processes

SELECT A.INST_ID, A.PROCESS, A.STATUS, A.LOG_SEQUENCE, A.STATE FROM GV$ARCHIVE_PROCESSES A WHERE STATUS<>'STOPPED' ORDER BY A.INST_ID # archived log files SELECT INST_ID , THREAD#, NAME FROM GV$ARCHIVED_LOG;
Log_Archive_Dest_1 Set To Default Even When DB_Recovery_File_Dest Is Set (Bug 6373164)

You should take this bug into consideration on Oracle 11.1.0.6. If you just leave DEST_10 pointing to the flash recovery areas (as the default case), DEST_1 will also point to the default location: ORACLE_HOME\RDBMS Here are the steps to reproduce the bug: When I disable LOG_ARCHIVE_DEST_1, the v$parameter and V$ARCHIVE_DEST are correctly updated. But after I restart the db, the LOG_ARCHIVE_DEST_1 is NULL in v$parameter but it points to C:\app\Administrator\product\11.1.0\db_1\RDBMS in V$ARCHIVE_DEST. DEST_10 is pointing to the recovery area but it's shown by V$ARCHIVE_DEST and not by v$parameter. I also noticed Oracle writes archive log files in the both destinations: DB_RECOVERY_FILE_DEST and the default one.
Page 495
Work around: Explicitly set LOG_ARCHIVE_DEST_1 to LOCATION=USE_DB_RECOVERY_FILE_DEST
SQL> select schedule, destination 2 from V$ARCHIVE_DEST 3 where dest_name='LOG_ARCHIVE_DEST_1'; SCHEDULE DESTINATION -------- ---------------------------------------------ACTIVE C:\app\Administrator\product\11.1.0\db_1\RDBMS
SQL> select name, value 2 from v$parameter 3 where upper(name) like 'LOG_ARCHIVE_DEST_1'; NAME VALUE ------------------------------ -----------------log_archive_dest_1
SQL> alter system set LOG_ARCHIVE_DEST_1='' scope=both; System altered.
SQL> select schedule, destination 2 from V$ARCHIVE_DEST 3 where dest_name='LOG_ARCHIVE_DEST_1'; SCHEDULE DESTINATION -------- --------------------------------INACTIVE
SQL> shutdown immediate Database closed. Database dismounted. ORACLE instance shut down. SQL> startup ORACLE instance started. Total System Global Area Fixed Size Variable Size Database Buffers Redo Buffers Database mounted. Database opened. 431038464 1333676 343934548 79691776 6078464 bytes bytes bytes bytes bytes
SQL> select name, value 2 from v$parameter 3 where upper(name) like 'LOG_ARCHIVE_DEST_1'; NAME VALUE ------------------------------ -----------------Page 496
log_archive_dest_1
SQL> select schedule, destination 2 from V$ARCHIVE_DEST 3 where dest_name='LOG_ARCHIVE_DEST_1'; SCHEDULE DESTINATION -------- ---------------------------------------------ACTIVE C:\app\Administrator\product\11.1.0\db_1\RDBMS
SQL> select schedule, destination 2 from V$ARCHIVE_DEST 3 where dest_name='LOG_ARCHIVE_DEST_10'; SCHEDULE DESTINATION -------- ---------------------------------ACTIVE USE_DB_RECOVERY_FILE_DEST SQL> select name, value 2 from v$parameter 3 where upper(name) like 'LOG_ARCHIVE_DEST_10'; NAME VALUE ------------------------------ ------------------log_archive_dest_10
Media Recovery in Real Application Clusters

Same as standalone database.
Parallel Recovery in Real Application Clusters

When parallelism is configured, Oracle automatically use the optimum degree of parallelism for recovery. # Disabling Instance and Crash Recovery Parallelism configure RECOVERY_PARALLELISM to 0 # Disabling Media Recovery Parallelism RMAN> RECOVER DATABASE NOPARALLEL SQL>ALTER DATABASE RECOVER NOPARALLEL
Using a Flash Recovery Area in RAC

Set the parameter DB_RECOVERY_FILE_DEST to a shared location and must be the same value on all instances.
Page 497
Managing Backup and Recovery
Page 498
Administrative Options
Using Enterprise Manager Grid Control to Discover Nodes and Instances
To discover targets if a database is created after agents are installed or if a database is not automatically discovered at agent install time: 1. 2. 3. Log in to Enterprise Manager and click the Targets tab. Click the Database tab to view all of the available targets. The column labeled Types shows the Oracle RAC databases using the entry "Cluster Database". Select the target name, then clicking Add. The Add Database Target: Specify Host page appears, which enables you to add databases, Listeners, and Automatic Storage Management (ASM) as monitored targets. Click the flashlight icon to display the available host names, select a host, then click Continue. The Add Database: Specify Source page appears. Either request Enterprise Manager to discover only single-instance databases and Listeners, or to discover all cluster databases, single-instance databases, and Listeners on the cluster, then click Continue. Enterprise Manager performs discovery to locate and display the cluster database and its associated instances. The Targets Discovered on Cluster page appears. If this procedure did not discover your reconfigured cluster database and all of its instances, you can use this page to manually configure your cluster databases and single-instance databases.
4. 5.
6.
Additional Information About SQL*Plus in RAC

How SQL*Plus Commands Affect Instances Most SQL statements affect the current instance. ALTER SYSTEM SWITCH LOGFILE affects only the current instance. o o To force a global log switch, use the ALTER SYSTEM ARCHIVE LOG CURRENT statement. The INSTANCE option of ALTER SYSTEM ARCHIVE LOG enables you to archive each online redo log file for a specific instance.
Displaying Running Instances SHOW INSTANCE SELECT * FROM V$ACTIVE_INSTANCES; Displaying Connect Identifier use SQLPROMPT Command _CONNECT_IDENTIFIER SET SQLPROMPT "_CONNECT_IDENTIFIER _USER > "
Page 499
Quiescing RAC Databases

The ALTER SYSTEM QUIESCE RESTRICTED and ALTER SYSTEM UNQUIESCE statements affect all instances in an Oracle RAC environment. ALTER SYSTEM QUIESCE RESTRICTED; To set the database in the QUIESCE state, the Database Resource Manager must be activated in all the instances. -- to make sure the resource manager is active (not null) select value from v$parameter where upper(name)='RESOURCE_MANAGER_PLAN'; Quiesced State and Cold Backups You cannot use the quiesced state to take a cold backup.
Transparent Data Encryption and Wallets in RAC

One wallet shared by all instances on shared storage. One copy of the wallet on each local storage: Local copies need to be synchronized each time master key is changed. The wallet must reside in the directory specified by the setting for the WALLET_LOCATION or ENCRYPTION_WALLET_LOCATION parameter in sqlnet.ora.
Administering System and Network Interfaces with oifcfg

Defining Network Interfaces with oifcfg Syntax and Commands for the oifcfg Command-Line Tool Oracle Interface Configuration oifcfg -help oifcfg iflist oifcfg getif [ [-global | -node nodename] [-if if_name[/subnet]] [-type if_type] ] # store the interface hme0, with the subnet 139.185.141.0, as a global # interface (to be used as an interconnect for all of the RAC instances in your cluster) oifcfg setif -global hme0/139.185.141.0:cluster_interconnect # create the cms0 interface in rac1 and rac2 oifcfg setif -node rac1 cms0/139.185.142.1:cluster_interconnect oifcfg setif -node rac2 cms0/139.185.142.2:cluster_interconnect # deleting an interface # CAUTION: when no option provided, all interfaces will be deleted: oifcfg delif -global qfe0/204.152.65.16 # all global interfaces will be deleted oifcfg delif -global
Page 500
Changing Public or Interconnect IP Subnet Configuration

A network interface can be stored as a global interface or as a node-specific interface. An interface is stored as a global interface when all the nodes of a RAC cluster have the same interface connected to the same subnet (recommended). It is stored as a node-specific interface only when there are some nodes in the cluster that have a different set of interfaces and subnets. su $CRS_HOME/bin/oifcfg getif oifcfg delif -global eth0 oifcfg setif global eth0/139.2.166.0:public oifcfg delif global eth1 oifcfg setif global eth1/192.168.1.0:cluster_interconnect oifcfg getif
Changing VIP Addresses

If only the IP address is changed, it is not necessary to make changes to the listener.ora, tnsnames.ora and initialization parameter files, provided they are using the virtual host names. If changing both the virtual host name and the VIP address for a node, it will be necessary to modify those files with the new virtual host name. For the listener.ora file,you can use netca to remove the old listener and create a new listener. In addition, changes will need to be made to the tnsnames.ora file of any clients connecting to the old virtual host name. 1. Confirm the current IP address for the VIP by running the ifconfig a command. On Windows, run the ipconfig /all command. ifconfig a 2. Stop First, the database instance, and then the ASM instance. When done, stop nodeapps. srvctl stop instance .. srvctl stop asm .. srvctl stop nodeapps 3. Verify that the VIP is no longer running by executing the ifconfig -a command again: ifconfig a If the interface still online, this is an indication that a resource which is dependent on the VIP is still running. The crs_stat -t command can help to show resources that are still online: crs_stat -t 4. Make any changes necessary to all nodes' /etc/hosts files (on UNIX), or \WINNT\System32\drivers\etc\hosts files on Windows, and make the necessary DNS changes, to associate the new IP address with the old host name. 5. Modify nodeapps and provide the new virtual IP address: su srvctl modify nodeapps -n rac1 -A 192.168.2.125/255.255.255.0/eth0 6. Start nodeapps again. srvctl start nodeapps -n rac1
Page 501
7. Repeat the same steps for all the nodes in the cluster. You can stay connected from the first node. 8. verifies the connectivity between all of the nodes. cluvfy comp nodecon -n all [-verbose] 9. Restart all of the instances and node applications
Page 502
Adding Nodes and Instances on UNIX-Based Systems

Adding Nodes to a RAC Environment
In adding nodes to a cluster, there are three procedures that can be used: Cloning: copy images of Oracle Clusterware and RAC software onto the other nodes that have identical hardware and software. OEM: GUI of cloning. Interactive or silent procedures using scripts and DBCA.
Cloning Oracle Clusterware and RAC Software in Grid Environments

Prerequisites Make the required prerequisites changes on OS to prepare the node for the software cloning. The steps stated in the Preinstallation Tasks illustrates the steps required for Oracle 10g R2 for Oracle Linux Enterprise 4.5 for x86. Metalink document ID 169706.1 is quick reference of OS Installation and Configuration Requirements. The DB Console should be installed so that the required Perl files are installed in $ORACLE_HOME/clone/bin If the server has more than one Perl version installed, it may be necessary to specify the PERL5LIB environment variable so that the versions of the Perl modules match with the Perl version used.
Cloning Procedure Steps Cloning Oracle10g Release 2 on Oracle Linux Enterprise 4.5 for x86 Using OUI: 1. To ensure the integrity of the copy, shutdown any databases, listeners, agents etc. that are running from the source home. 2. Clone the Oracle Clusterware home then Oracle Database home cd /u01/crs tar -cvf /tmp/sourcecrs.tar cd /u01/app/oracle/product/10.2.0 tar -cvf /tmp/sourcedb.tar 3. Make sure the required users and groups are there on the new node id oracle id oinstall 4. Extract the tar file cd /u01/crs tar -xvf /tmp/sourcecrs.tar cd /u01/app/oracle/product/10.2.0 tar -xvf /tmp/sourcedb.tar 5. Run the Oracle Universal Installer (OUI) in clone mode /* Method 1 */ cd $ORACLE_HOME/clone/bin perl clone.pl ORACLE_HOME="<target_home>" ORACLE_HOME_NAME="<unique_home_name>" /* Method 2 */ cd $ORACLE_HOME/oui/bin ./runInstaller -clone -silent -ignorePreReq ORACLE_HOME="<target_home>" ORACLE_HOME_NAME="<unique_home_name>" If necessary, add "-invPtrLoc <path>/oraInst.loc" or "-ignoreSysPrereqs" to the command line.
Page 503
6. Run the installation scripts $ORACLE_HOME/root.sh 7. 8. 9. 10. Repeat steps 1 to 5 on Oracle home with RAC software Run the Oracle Net Configuration Assistant (NETCA) on the new node to create a Listener. Make if necessary, any modification in the tnsnames.ora file. To start the dbconsole on the new server, refer to Metalink Note 467598.1 and Note 278100.1 for
Quick-Start Node and Instance Addition Procedures

Prerequisites Make the required prerequisites changes on OS to prepare the node for the software cloning. o The steps stated in Preinstallation Tasks illustrates the steps required for Oracle 10g R2 for Oracle Linux Enterprise 4.5 for x86. Be careful about the following changes in the steps: The examples in this sections assumes the node to add has the ip addresses public:192.168.4.15, vip:192.168.4.17, priv:192.168.0.15 with hostname of rac3. ORACLE_SID=rac3 in the .bash_profile file Be careful NOT to partition the disks as they are already partitioned and used by the other node(s). However, raw disks must be binded. Domain name resolution (/etc/hosts) must apply in all nodes Remote installation procedure must be redone to all nodes. Metalink document ID 169706.1 is quick reference of OS Installation and Configuration Requirements for the supported Operating Systems.
Adding an Oracle Clusterware Home to a New Node Practically, I noticed the procedure does not work unless all the existing nodes in the RAC are alive. # (1) run add node in rac1 # as oracle export CRS_HOME=/u01/crs cd $CRS_HOME/oui/bin ./addNode.sh # Specify the New Nodes then Next # Verify the new node names in the Summary page then Finish (copy size:80 MB) # run the scripts as indicated by OUI. They must be: # /u01/app/oracle/oraInventory/orainstRoot.sh in rac3 # /u01/crs/install/rootaddnode.sh in rac1 # in rac3: EDIT root.sh and fix the node list in its header BEFORE running it # /u01/crs/root.sh in rac3 # exit the OUI ## if root.sh failed, you can cancel its effect by running: # $CRS_HOME/install/rootdelete.sh # (2) rac1 as oracle: run the Oracle Notification Service (RACGONS) # obtain remote port number cat $CRS_HOME/opmn/conf/ons.config $CRS_HOME/bin/racgons add_config rac3:6200 #to verify ping rac3-vip crs_stat -t
Page 504
# if you're using vmware, make backup of the environment now. Adding an Oracle Home with RAC to a New Node OUI can be used interactively or in silent mode. This example shows the interactive mode steps. # (1) copy Oracle home # in rac1 as oracle cd $ORACLE_HOME/oui/bin ./addNode.sh # select node then Next # verify info in the Summary page then Next # about 1.5 GB will be copied to rac3 # run the required script # (2) add Listener # rac3 as oracle: run netca to add a listener in rac3 node (select rac3 ONLY) $ORACLE_HOME/bin/netca # (3) create db instance as descried in "Adding Database Instances to New Nodes" # (4) Perform Postinstallation Steps. Be aware to check the ASM instance name before editing /etc/oractab file: ps -ef | grep asm # (5) Review $ORACLE_HOME/network/admin/tnsnames.ora file and fix node # names errors, if any. # (6) preferably perform General Useful Postinstallation Tasks.
Detailed Node and Instance Addition Procedure
Step 1: Connecting New Nodes to the Cluster

Making Physical Connections Installing the Operating System Creating Oracle Users Verifying the Installation with the Cluster Verification Utility Checking the Installation
Step 2: Extending Clusterware and Oracle Software to New Nodes

Adding Nodes at the Vendor Clusterware Layer Adding Nodes at the Oracle Clusterware Layer
Step 3: Preparing Storage on New Nodes

Raw Device Storage Preparation for New Nodes
Page 505
Step 4: Adding Nodes at the Oracle RAC Database Layer
Step 5: Adding Database Instances to New Nodes

Using Enterprise Manager to Add Database Instances to New Nodes Using DBCA in Interactive Mode to Add Database Instances to New Nodes 1. Start dbca in rac1: $ORACLE_HOME/bin/dbca & 2. Follow: Real Application Cluster-> Instance Management-> Add Instanc-> Select RAC database-> Next-> Enter the Instance Name, select node-> Enter service info-> $CRS_HOME/bin/crs_stat -t srvctl start service -d rac -s "hrserv" -i rac3 srvctl start service -d rac -s "oeserv" -i rac3 Using DBCA in Silent Mode to Add Database Instances to New Nodes
Page 506
Deleting Nodes and Instances on UNIX-Based Systems

Option 1: Quick-Start Node and Instance Deletion Procedures
Oracle Database home is always deleted first, then the Clusterware home is deleted. Quick-Start node and instance deleteion procedure could be the convenient option when you want to delete one or two nodes.
Deleting an Oracle Home with RAC from an Existing Node 1. The target here is to delete the Oracle database home. OUI can be used in interactive or silent mode to acheive this target. Interactive method is shown here. The example below assumes that you remove Oracle Database home from a node named rac2: Remove rac2 from the available or preferred list of all the services:
# list the services srvctl status service -d rac # stop the services in rac2 srvctl stop service -d rac -s hrserv -i rac2 srvctl stop service -d rac -s oeserv -i rac2 # check rac2 is used by any service srvctl config service -d rac # my services available only in the remaining nodes srvctl modify service -d rac -s hrserv -n -i rac1 srvctl modify service -d rac -s oeserv -n -i rac1 # confirm srvctl config service -d rac 2. From rac1, remove DB instance in rac2 using DBCA (OEM can also be used). Follow the steps as in the sub-section "Using DBCA in Interactive Mode to Delete Database Instances from Existing Nodes". To confirm:
crs_stat -t 3. 4. Delete ASM instance as described in the sub-section "ASM Instance Clean-Up Procedures for Node Deletion". Use NETCA to remove the listener. To confirm:
crs_stat -t 5. Stop nodeapps from rac2 then remove them.
su cd /u01/crs/bin ./srvctl stop nodeapps -n rac2 ./crs_stat -t ./srvctl remove nodeapps -n rac2 6. On rac2, to make the OUI remove the software home only from rac2:
su - oracle echo $ORACLE_HOME export ORACLE_HOME=/u01/app/oracle/product/10.2.0/db_1

Page 507
cd $ORACLE_HOME/oui/bin ./runInstaller -updateNodeList ORACLE_HOME=$ORACLE_HOME "CLUSTER_NODES={rac2}" local 7. Run OUI from the home and deinstall this home. Make sure that you choose the home to be removed and not just the products under that home.
./runInstaller & 8. On rac1, run the following command where "CLUSTER_NODES={remaining_list}":
echo $ORACLE_HOME export ORACLE_HOME=/u01/app/oracle/product/10.2.0/db_1 cd $ORACLE_HOME/oui/bin ./runInstaller -updateNodeList ORACLE_HOME=$ORACLE_HOME "CLUSTER_NODES={rac1}" Deleting an Oracle Clusterware Home from an Existing Node 1. The target here is to delete the Oracle Clusterware home from a node. OUI can be used in interactive or silent mode to acheive this target. Interactive method is shown here. The example below assumes that you remove Oracle Clusterware home from a node named as rac2: Makre sure Oracle Database home was removed before going on the clusterware home deletion procedure. If you ran the Oracle Interface Configuration Tool (OIFCFG) with the -global flag during the installation, then skip this step. Otherwise, from a node that is going to remain in your cluster, from the CRS_home/bin directory, run the following command:
2.
./oifcfg delif node rac2 3. Obtain the remote port number, which you will use in the next step, using the following command from the CRS_HOME/opmn/conf directory:
echo $CRS_HOME export CRS_HOME=/u01/crs cd $CRS_HOME/opmn/conf cat ons.config 4. In rac1: run the Oracle Notification Service (RACGONS)
cd $CRS_HOME/bin ./racgons remove_config rac2:remote_port 5. On rac2 as root:
su export CRS_HOME=/u01/crs $CRS_HOME/install/rootdelete.sh 6. On rac1 as root: run ./rootdeletenode.sh node2,node2-number
su export CRS_HOME=/u01/crs # if you want to identify the node number $CRS_HOME/bin/olsnodes -n $CRS_HOME/install/rootdeletenode.sh rac2,2 $CRS_HOME/bin/olsnodes -n
Page 508
7.
On rac2:
su - oracle export CRS_HOME=/u01/crs cd $CRS_HOME/oui/bin ./runInstaller -updateNodeList ORACLE_HOME=$CRS_HOME "CLUSTER_NODES={rac2}" CRS=TRUE -local 8. On rac2, de-install the Oracle Clusterware using the OUI:
cd $CRS_HOME/oui/bin ./runInstaller & 9. On rac1, where "CLUSTER_NODES={remaining_nodes_list}":
su - oracle cd $CRS_HOME/oui/bin ./runInstaller -updateNodeList ORACLE_HOME=$CRS_HOME "CLUSTER_NODES={rac1}" CRS=TRUE crs_stat -t olsnodes -n
Option 2: Detailed Node and Instance Deletion Procedure

Detailed node and instance deleteion procedure could be more convenient when you want to delete high number of nodes.
Step 1: Deleting DB Instances from Real Application Clusters Databases

Using Enterprise Manager to Delete Database Instances from Existing Nodes Follow the links: Cluster Database Home page-> Maintenance-> Deployments section-> Delete Instance-> Deletion Wizard starts
Using DBCA in Interactive Mode to Delete Database Instances from Existing Nodes 1. 2. 3. 4. 5. 6. 7. On rac1, open DBCA from Oracle home. On the DBCA Welcome page select Oracle Real Application Clusters Database, click Next On the DBCA Operations page, select Instance Management, click Next On the Instance Management page, Select Delete Instance, click Next On the List of Cluster Databases page, select the Oracle RAC database. Enter a SYSDBA user name and password. Click Next On the List of Cluster Database Instances page, select an instance to delete and click Finish. If you have services assigned to this instance, then the DBCA Services Management page appears. Use this feature to reassign services from this instance to other instances in the cluster database. On the Summary page and click OK. DBCA removes the instance and the instance's Oracle Net configuration.
8. 9.
10. After completion, exit DBCA

Page 509
Using DBCA in Silent Mode to Delete Instance from Existing Nodes 1. General Syntax:
dbca -silent -deleteInstance [-nodeList node] -gdbName gdbname instanceName instname -sysDBAUserName sysdba -sysDBAPassword password -- if running form rac2: dbca -silent -deleteInstance -gdbName rac -instanceName rac2 -sysDBAUserName sys -sysDBAPassword syspassword -- if running from rac1: dbca -silent -deleteInstance -nodeList rac2 -gdbName rac -instanceName rac2 sysDBAUserName sys -sysDBAPassword syspassword
Step 2: Deleting Nodes from Real Application Clusters Databases

The example assumes rac2 is the node to delete and that Oracle home is local (not shared). 1. 2. 3. 4. If there are instances on the node that you want to delete, then perform one of the procedures in the previous section Setp1. Perform ASM clean up procedure as described in the sub-section "ASM Instance CleanUp Procedures for Node Deletion". Use NETCA to remove the listener from rac2. On rac2:
cd $ORACLE_HOME/oui/bin runInstaller -updateNodeList ORACLE_HOME=$ORACLE_HOME CLUSTER_NODES="" local 5. Run OUI from the home and deinstall this home. Make sure that you choose the home to be removed and not just the products under that home.
./runInstaller & 6. On rac1, where "CLUSTER_NODES={node_list}":
runInstaller -updateNodeList ORACLE_HOME=Oracle_home "CLUSTER_NODES={rac1}" 7. Remove node-specific interface configurations:
# to obtain port number echo $CRS_HOME export CRS_HOME=/u01/crs cd $CRS_HOME/opmn/conf cat ons.config racgons remove_config rac2:<remote_port> oifcfg delif -node rac2 8. On rac2, disable the Oracle Clusterware applications that are on the node:
echo $CRS_HOME export CRS_HOME=/u01/crs cd $CRS_HOME/install # If the ocr.loc file is on a local file system rootdelete.sh remote nosharedvar # If the ocr.loc file is on a shared file system rootdelete.sh remote sharedvar 9. On any remaining node in the cluster (rac1), to delete the nodes from the Oracle cluster and to update the Oracle Cluster Registry (OCR): Oracle DBA Code Examples
Page 510
su echo $CRS_HOME export CRS_HOME=/u01/crs # to display node numbers $CRS_HOME/bin/olsnodes -n. # to delete the node # general syntax: rootdeletenode.sh node1,node1-number,node2,node2-number,... $CRS_HOME/install/rootdeletenode.sh rac2,2 10. On rac2: $CRS_HOME/oui/bin/runInstaller -updateNodeList ORACLE_HOME=$CRS_HOME CLUSTER_NODES="" local CRS=true 11. Run OUI from the home and deinstall this home. Make sure that you choose the home to be removed and not just the products under that home. $CRS_HOME/oui/bin/runInstaller & 12. On rac1, where "CLUSTER_NODES={ramining_nodelist}": runInstaller -updateNodeList ORACLE_HOME=CRS_home "CLUSTER_NODES=rac1" 13. Verify node removal: cluvfy comp crs -n all [-verbose]
ASM Instance Clean-Up Procedures for Node Deletion

The target here is to remove ASM instance from a node. 1. 2. Stop, if any, all of the databases that use the ASM instance. If you follow this procedure as part of a node deletion, the Db instance must have been removed by this step. On rac2:
srvctl stop asm -n rac2 srvctl remove asm -n rac2 srvctl config asm -n rac2 srvctl config asm -n rac1 crs_stat -t 3. On rac2:
rm -r $ORACLE_BASE/admin/+ASM rm -f $ORACLE_HOME/dbs/*ASM*
Page 511
Adding and Deleting Nodes and Instances on Windows-Based Systems
Page 512
Cloning Oracle Clusterware and RAC Software in Grid Environments
Quick-Start Node and Database Instance Addition and Deletion Procedures

Adding an Oracle Clusterware Home to a New Node Adding an Oracle Home with RAC to a New Node Deleting an Oracle Home with RAC from an Existing Node Deleting an Oracle Clusterware Home from an Existing Node
Detailed Node and Database Instance Addition and Deletion Procedures
Overview of Node Addition Procedures
Step 1: Connecting New Nodes to the Cluster

Making Physical Connections Installing the Operating System Verifying the Installation with the Cluster Verification Utility Checking the Installation
Step 2: Extending Oracle Software to New Nodes at the Oracle Clusterware
Step 3: Preparing Storage on New Nodes

Raw Device Storage Preparation for New Nodes
Step 4: Adding Nodes at the Oracle RAC Database Layer
Step 5: Adding Database Instances to New Nodes

Using Enterprise Manager to Add Database Instances to New Nodes Using DBCA in Interactive Mode to Add Database Instances to New Nodes Using DBCA in Silent Mode to Add Database Instances to New Nodes Connecting to iSQL*Plus after Adding a Node
Page 513
Adding Nodes that Already Have Clusterware and Oracle Software to a Cluster
Overview of Node Deletion Procedures
Step 1: Deleting Instances from Real Application Clusters Databases

Using Enterprise Manager to Delete Database Instances from Existing Nodes Using DBCA in Interactive Mode to Delete Database Instances from Existing Nodes Using DBCA in Silent Mode to Delete Instance from Existing Nodes
Step 2: Deleting Nodes from Real Application Clusters Databases
Step 3: ASM Instance Clean-Up Procedures for Node Deletion
Page 514
Monitoring Performance
RAC Common Tuning Tips
Application tuning is often the most beneficial Resizing and tuning the buffer cache Increasing sequence caches to a high value: to avoid index leaf contention caused by high or batch inserts. Reducing long full-table scans in OLTP systems: to reduce GCS requests. Using Automatic Segment Space Management Using partitioning to reduce interinstance traffic Avoiding unnecessary parsing Minimizing locking usage Removing unselective indexes Configuring interconnect properly
Instance Recovery and RAC

_FAST_START_INSTANCE_RECOVERY_TARGET initialization parameters controls the time in seconds from the start of instance recovery to the time when GCD is open for lock requests. If FAST_START_MTTR_TARGET is also configured, the more aggressive will be used. Practically, using either of them is useless because one of the surviving instances is doing the recovery. You can however consider the following: o o o o Increase PARALLEL_EXECUTION_MESSAGE_SIZE (message size for parallelism and its buffer taken from shared pool) from its default of 2,148 bytes to 4 KB or 8 KB. Set PARALLEL_MIN_SERVERS to CPU_COUNT-1. This will prespawn recovery slaves at startup time. Using asynchronous I/O is one of the most crucial factors in recovery time. The firstpass log read uses asynchronous I/O. Instance recovery uses 50 percent of the default buffer cache for recovery buffers. If this is not enough, you will see it in the alert.log. In that case, increase the size of your default buffer cache.
Global Cache Wait Events

gc current/cr request gc [current/cr] [2/3]-way gc [current/cr] block busy gc [current/cr] grant 2-way: grant occurs before I/O gc current grant busy gc [current/cr] [block/grant] congested: congested means it is internally > 1ms gc [current/cr] [failure/retry] gc buffer busy
Page 515
Monitoring Performance in Enterprise Manager

Using the Cluster Database Performance Page Chart: Desc.: Cluster Host Load Average load on CPU in all cluster nodes
Vertical: No of CPUs in all the cluster Notice : If the load average is higher than the average of the total number of CPUs across all of the hosts in the cluster, then too many processes are waiting for CPU resources.
Chart: Global Cache Block Access Latency Desc.: end-to-end elapsed time or latency for a block request Vertical: latency in ms Notice : - interconnect delay - unoptimized SQL plans to acheive local cache hit ration. - to resolve: drill down to the Cluster Cache Coherency page
Chart: Global Cache Block Transfer Rate Desc.: number of data blocks received by all instances Notice: drid down till segment type Chart: Average Active Sessions Desc.: average number of active sessions per wait class. Notice: Consider tuning the database, if the Average Active Sessions chart displays a large number of sessions waiting, indicating internal contention, and throughput the Database Throughput charts is low. Chart : Database Throughput Charts Desc. : summarize any contentions that appear in the Average Active Sessions chart Notes : drill down to instance level then top consumers Chart : Top Segments Page Desc. : to identify hot tables or indexes in a database Notes : tracks the number of CR and current blocks received by an object. Chart : Database Locks Page Desc. : to determine whether multiple instances are holding locks for the same object. Using the Cluster Database Instance Performance Page To access it, Performance tab-> instance name in the bottom. Instance-level statistics including generating ADDM and ASH reports. AWR: gather statistics about each individual instance in the RAC and not the entire database. ADDM Report: create a new AWR snapshot and run ADDM on this and the previous snapshot. exec dbms_advisor.set_default_task_parameter('ADDM','DB_ACTIVITY_MIN',30) exec dbms_workload_repository.modify_snapshot_settings(interval=>600) exec dbms_workload_repository.create_snapshot
Page 516
ASH Report: create a performance data report of the database based on session-sampling data over a specified period of time. This report is very useful for diagnosing small (five- to ten-minute) performance spikes that might be averaged out and consequently hidden or minimized by other issues in the 1-hour AWR report. Using the Cluster Performance Page With this information, you can determine whether resources need to be added, suspended, or redistributed. Using the Cluster Interconnects Page Home page-> Interconnect Findings Home page-> Performance-> Cluster Cache Coherency-> Interconnects This page helps determine the load added by individual instances and databases on the interconnect. Sometimes you can immediately identify interconnect delays that are due to applications that are outside Oracle Database.
Page 517
Making Applications Highly Available Using Oracle Clusterware

Note: You can install the Oracle Clusterware high availability Application Programming Interface (API) from the Oracle Database 10g release 10.2 client installation media. Note: Your Listener.ora file should contain one entry for the virtual internet protocol (VIP) address, using the VIP's name, and another entry for the physical host, by IP address not name.
Making an Application Highly Available Examples

Example1: Making an Application Highly Available If the application is accessed by network, define a VIP per application. VIP is failed over to another node together with the application(s). From there, clients can still connect through the VIP.
1. Create an application profile by editing an ASCII file or by running the crs_profile command. # file name format resource_name.cap save file in $CRS_HOME\crs\profile $CRS_HOME\crs\public # Required Ones # resource-specific script (start, stop or check will be passed by crs) ACTION_SCRIPT # application name NAME # resource type (must be APPLICATION) TYPE # ordered list of cluster nodes (use names as listed by olsnodes) HOSTING_MEMBERS # Optional Ones # description of the resource DESCRIPTION='myapplication' # list of required resources (must be registered) REQUIRED_RESOURCES # placement policy: (balanced), favored, or restricted PLACEMENT # when 1 (0), crs re-evaluates the placement of a resource # during addition or restart of a cluster node ACTIVE_PLACEMENT=0 # when 1 (0), it atuo-restarts on system reboot AUTO_START=1 # check intervals in seconds (60)
Page 518
for root user for non-root user
CHECK_INTERVAL # failover interval (0) FAILOVER_DELAY # interval (s) during which crs applies the failure threshold FAILURE_INTERVAL # (max 20) number of failures within FAILURE_INTERVAL after which # the resource is marked offline FAILURE_THRESHOLD # space-delimited liste of resource used during placement decisions OPTIONAL_RESOURCES # restart attempts before relocate (1) RESTART_ATTEMPTS # maintained by crs RESTART_COUNT # timeout in seconds the script needs before returning an error (60) SCRIPT_TIMEOUT START_TIMEOUT STOP_TIMEOUT # application up-time to be considered stable by crs UPTIME_THRESHOLD # example 1 oracle$crs_profile -create postman -t application -B /opt/email/bin/crs_postman \ -d "Email Application" -r network1 -l application2 \ -a postman.scr -o ci=5,ft=2,fi=12,ra=2 # in $CRS_HOME/crs/public # will generate file postman.cap containing: NAME=postman TYPE=application ACTION_SCRIPT=/oracle/crs/script/postman.scr ACTIVE_PLACEMENT=0 AUTO_START=0 CHECK_INTERVAL=5 DESCRIPTION=email app FAILOVER_DELAY=0 FAILURE_INTERVAL=12 FAILURE_THRESHOLD=2 HOSTING_MEMBERS= OPTIONAL_RESOURCES=application2 PLACEMENT=balanced REQUIRED_RESOURCES=network1 RESTART_ATTEMPTS=2 SCRIPT_TIMEOUT=60 ... # and script file postman.scr # Create Application VIP # create network1 VIP address application # where eth0 is the public network nic # 138.3.83.78 is the vip address crs_profile create network1 -t application \ Oracle DBA Code Examples
Page 519
-a $CRS_HOME/bin/usrvip \ -o oi=eth0,ov=138.3.83.78,on=255.255.240.0 # as oracle, register network1 oracle$crs_register network1 # change the owner of the resource crs_setperm network1 o root # enable oracle user to run the script crs_setperm network1 u user:oracle:r-x # as oracle starts the vip crs_start network1 2. Register the application profile using the crs_register command. crs_register postman # If you modify postman profile, then update the OCR: crs_register -u postman 3. Run the crs_start command to initiate the application profile and then the Oracle Clusterware runs the action program command that you have included in the profile to start your application. crs_start postman # to start an application resource, even if one of the required resources # is offline: crs_start -f postman 4. The Oracle Clusterware periodically runs the action program command to check an applications status. 5. In the event of a check or node failure, the Oracle Clusterware recovers the applications either by restarting it on the current node or by relocating the application to another node. To manually relocate the application and its resources to another node: # the application and its required resources must be offline crs_relocate postman -c rac2 6. If you run the crs_stop command to stop the application, then the Oracle Clusterware runs the action program command to stop it. crs_stop postman # to stop an application that is required by an online resource crs_stop -f postman Example2: Making an Application Highly Available /* Making xclock program highly available */ # # 1) create the action script file # create following script in both nodes (in $CRS_HOME/crs/script) # make the file executable # you can then test it by passing start stop and check
Page 520
su export CRS_HOME=/u01/crs vi $CRS_HOME/crs/script/crsclock_action.scr chmod 777 $CRS_HOME/crs/script/crsclock_action.scr #!/bin/bash # start/stop/check script for xclock example # the script assumes xclock is there # and DISPLAY variable is set APP=/usr/X11R6/bin/xclock BIN_NAME=xclock LOG_DIR=/tmp export DISPLAY=:0.0 echo `date +"%M:%S"` $0 $* $$>>/tmp/mylog.log PID1=`ps -ef | grep $BIN_NAME | grep -v grep | grep -v xclock_app | awk '{ print $2 }'` case $1 in 'start') if [ "$PID1" != "" ] then status_p1="running" else if [ -x $APP ] then #umask 002 ${APP} & PID1=`ps -ef | grep $BIN_NAME | grep -v grep | grep -v xclock_app | awk '{ print $2 }'` echo `date +"%M:%S"` $* $PID1 $USER>>/tmp/mylog.log status_p1="started" else echo `basename $0`": $APP: Executable not found" fi fi echo "$APP: $status_p1" ;; 'stop') if [ "${PID1}" != "" ] then kill -9 ${PID1} && echo "$APP killed" else echo "$BIN_NAME: no running Process!" fi ;; 'check') if [ "$PID1" != "" ] then echo "running" exit 0 else echo "not running" echo `date +"%M:%S"` $0 $* "ERR">>/tmp/mylog.log exit 1
Page 521
fi ;; *) echo "Usage: "`basename $0`" {start|stop|check}" ;; esac
# # 2) create application resource profile named myClock # if you copy paste, fix the hyphen issue su $CRS_HOME/bin/crs_profile -create myClock -t application -a crsclock_action.scr -p favored -h "rac1-vip rac2-vip" -o ci=5,ra=2 # check the generated file cat $CRS_HOME/crs/profile/myClock.cap # # 3) Register the resource (in ONE node only) # must be done as root and then permission is granted to oracle su $CRS_HOME/bin/crs_register myClock $CRS_HOME/bin/crs_setperm myClock -u user:oracle:r-x $CRS_HOME/bin/crs_stat myClock $CRS_HOME/bin/crs_stat -t # # 4) Start the resource # as oracle su - oracle $CRS_HOME/bin/crs_start myClock $CRS_HOME/bin/crs_stat -t
Managing Automatic Oracle Clusterware Resource Operations for Action Scripts /* Preventing Automatic Database Instance Restarts */ In the application profile, set AUTO_START to (lowercase only): always, restore, never /* Automatically Manage Restart Attempts Counter for Resources */ This is controlled by the attributes RESTART_ATTEMPTS and RESTART_COUNT. Displaying Clusterware Application and Application Resource Status Information # application status crs_stat postman # all resources status crs_stat -t # specific resource status crs_stat -t ora.rac1.ASM1.asm # stats about the resources ( R=Restart, F=Fail) crs_stat -v # application profile crs_stat -p
Page 522
crs_stat -p ora.rac1.ASM1.asm Unregistering Applications and Application Resources su $CRS_HOME/bin/crs_unregister postman $CRS_HOME/bin/crs_stat -t
Page 523
RAC Troubleshooting
Diagnosing the Oracle Clusterware High Availability Components
Debugging Recommnedation Always make sure that your nodes have exactly the same system time. Use NTP.
Clusterware Log Files and the Unified Log Directory Structure # Cluster Ready Services Daemon (crsd) # It is archived every 10 MB (crsd.l01, crsd.l02, ) $CRS_HOME/log/hostname/crsd # Oracle Cluster Registry (OCR) $CRS_HOME/log/hostname/client # Cluster Synchronization Services (CSS) # is archived every 20 MB (cssd.l01, cssd.l02, ) $CRS_HOME/log/hostname/cssd # Event Manager (EVM) $CRS_HOME/log/hostname/evmd # RACG Log Files $CRS_HOME/log/hostname/racg $ORACLE_HOME/log/hostname/racg # SRVM (srvctl) and OCR (ocrdump, ocrconfig, ocrcheck) logs $ORA_CRS_HOME/log/<hostname>/client # crs alerts $ORA_CRS_HOME/log/<hostname>/alert<nodename>.log Dynamic Debugging /* to Enable Debugging */ # as root # for the Oracle Clusterware crsctl debug log crs "CRSRTI:1,CRSCOMM:2" # for EVM crsctl debug log evm "EVMCOMM:1" # for resources (1 is the debugging level) crsctl debug log res "resname:1" # example: crsctl debug log res "ora.rac1.vip:1" Component Level Debugging # to enable debugging for all of the modules # where level 1 (least) to 5 (max) set ORA_CRSDEBUG_ALL <level> # to enable tracing for a specific sub-module
Page 524
set ORA_CRSDEBUG_modulename # to list the sub-modules # where module is crs, evm, or css crsctl lsmodules <module> Oracle Clusterware Shutdown and Startup su crsctl stop crs crsctl start crs Enabling and Disabling Oracle Clusterware Daemons su crsctl enable crs crsctl disable crs Diagnostics Collection Script Generates the following files in the local directory: basData_<hostname>.tar.gz crsData _<hostname>. tar.gz ocrData _<hostname>. tar.gz oraData _<hostname>. tar.gz # when asked by Oracle support su export ORACLE_HOME=/u01/app/oracle/product/10.2.0/db_1 export ORA_CRS_HOME=/u01/crs1020 export ORACLE_BASE= =/u01/app/oracle cd $ORA_CRS_HOME/bin $CRS_HOME/bin/diagcollection.pl -collect The Oracle Clusterware Alerts # can be found in: $CRS_Home/log/hostname/alerthostname.log Resource Debugging # method 1: where 1 in the following example is the debugging level crsctl debug log res "ora.node1.vip:1" # method 2: export USER_ORA_DEBUG=1 # .. then issue crsctl start, stop, or check Checking the Health of the Clusterware crsctl check crs Troubleshooting the Oracle Cluster Registry
Troubleshooting Hostname Changes and CSS If you change the host name for ASM, then the Oracle CSS daemon will not start. In order to counter this problem, please use the following steps:
Page 525
Login as the root user Run localconfig delete to deconfigure CSS. This will remove any configuration related files on the system that referenced the old host name. Run localconfig add to reconfigure CSS using the new host name.
Enabling Additional Tracing for Real Application Clusters High Availability
Diagnosing Oracle Real Application Clusters Components
Where to Find Files for Analyzing Errors
Using Instance-Specific Alert Files in Real Application Clusters
Enabling Tracing for Java-Based Tools and Utilities in Real Application Clusters
Resolving Pending Shutdown Issues
Using the Cluster Verification Utility

Cluster Verify Locations Download it from OTN Oracle software DVD: ./cluvfy/runcluvfy.sh Oracle Clusterware home: $ORA_CRS_HOME/bin/cluvfy Oracle Home: $ORACLE_HOME/bin/cluvfy Cluster Verify Stages You can list verifiable stages with the cluvfy stage -list Valid stage options and stage names are (ordered): -post hwos: Postcheck for hardware and operating system -pre cfs: Precheck for OCFS setup -post cfs: Postcheck for OCFS setup -pre crsinst: Precheck for CRS installation -post crsinst: Postcheck for CRS installation -pre dbinst: Precheck for database installation -pre dbcfg: Precheck for database configuration
Page 526
Cluster Verify Components You can list verifiable CVU components with the cluvfy comp -list Verifiable CVU components: nodereach: Checks reachability between nodes nodecon: Checks node connectivity cfs: Checks Oracle Cluster File System integrity ssa: Checks shared storage accessibility space: Checks space availability sys: Checks minimum system requirements clu: Checks cluster integrity clumgr: Checks cluster manager integrity ocr: Checks OCR integrity crs: Checks CRS integrity nodeapp: Checks node applications existence admprv: Checks administrative privileges peer: Compares properties with peers CVU Component Verification Examples #verify the minimal system requirements on the nodes before installing Clusterware cluvfy comp sys -n node1,node2 -p crs -verbose
#check the system requirements before installing RAC: cluvfy comp sys -n node1,node2 -p database -verbose
#verify whether storage is shared among the nodes in your cluster database or to identify all of the storage that is available on the system and can be shared across the cluster nodes: cluvfy comp ssa -n all -s /dev/sda1
#check there is 5 GB free in all nodes: cluvfy comp space -n all -l /home/product -z 5G
#can node1 reach node2: cluvfy comp nodereach -n node2 -srcnode node1
#checks whether node1 and node2 can communicate through the eth0 network interface (without i, all interfaces are checked): cluvfy comp nodecon -n node1,node2 i eth0 -verbose
#verify user equivalence for all the nodes: cluvfy comp admprv -n all -o user_equiv -verbose
#verify existence of node applications, namely VIP, ONS, and GSD, on all the nodes: cluvfy comp nodeapp -n all -verbose
Page 527
#compares all the nodes and determines whether any differences exist between the values of preselected properties cluvfy comp peer -n all verbose | more Understanding CVU Commands, Help, Output, and Nodelist Shortcuts
Performing Various CVU Tests
Known Issues for the Cluster Verification Utility
Page 528
Part 8
Oracle RAC One Node
Page 529
Installing Oracle 11g R2 RAC One Node on Linux 5

Overview This installation implements Oracle RAC One Node. Oracle RAC One Node is a single instance of Oracle RAC running on a node in a cluster. RAC One Node offers a form of virtualization where multiple databases can be consolidated on the same physical server, sharing a common pool of storage and server resources and thus reducing the physical IT footprints. Note: The metalink document RAC Starter Kit and Best Practices (Linux) [ID 811306.1] is a good source reference for this task. Note: The procedure applies on 11.2.0.1. Version 11.2.0.2 provides simplified procedure.
Emulation software: VMWare Workstation 7 RAC Node: 2 node with 2.5 GB RAM and 2 ethernet cards. OS: Red Hat Linux Enterprise 5.2 for x86 32-bit
Required Software
Oracle Database 11g Release 2 for Linux x86 32-bit Oracle Database 11g Release 2 Grid Infrastructure (11.2.0.1.0) for Linux x86 32-bit RAC One Node patch# 9004119 : RACONENODE_p9004119_112010_LINUX.zip
Used Hardware
In the VMWare: create one virtual machine (rac1) with the following specs: o 2.5 GB RAM o Two ethernet cards: both can be configured as bridged or host-only in VMware. o One local hardisk with 24 GB on SCSI 0:0. o CPU Count: 2 o Create a folder in the same directory structure level as the parent folder containing the created virtual machine. Give it a meaningful name like 'shared_disks'. Create in that folder the following disks: Disk1: of 3 GB. Allocate its disk space. It will be used for OCR and Voting disk. Set it on controller SCSI 1:1 Disk2: of 4 GB. Allocate its disk space. It will be used for +Data. Set it on controller SCSI 1:2 Disk3: of 2 GB. Allocate its disk space. It will be used for +Flash. Set it on controller SCSI 1:3
Page 530
Installation Plan
1. Preinstallation tasks Hardware requirements Software requirements Environment configuration
2. Oracle Grid Infrastructure installation 3. Oracle Grid Infrastructure Patching 4. Checking Oracle Grid Infrastructure Status 5. Oracle Database 11g R2 Software Installation 6. Oracle Database 11g R2 Software Patching 7. Install EM Agent in cluster nodes (if required) 8. ASM Diskgroups Creation 9. RAC Database Creation 10. Initialize the Database to RAC One Node 11. Complete postinstallation tasks 12. Useful postinstallation tasks Note: The installation is explained without GNS and IPMI Note: For this installation we will be using ASM for Clusterware and Database storage
Page 531
1. Preinstallation tasks Install Oracle Enterprise Linux in the first local hardisk. Install nothing in the remaining disks. Note: for a production system, consider becoming an Oracle Unbreakable Linux customer and register your server on the Unbreakable Linux Network. o Configure the swap area in the local hardisk to have 3 GB disk space. o Give the first ethernet card IP 192.0.2.100 and the second 172.0.2.100 and the hostname rac1.mydomain.com. Define a gateway. If it does not exist, make it same as the host IP address. o Insall the following packages: Desktop Environments o GNOME Desktop Environment Applications o Graphical Internet (optional) o Editors (optional) Development o Development Libraries o Development Tools Servers o Do not select anything in this group. Base System o Administration Tools o System Tools Add the package 'sysstat' by clicking on the Details link and selecting "sysstat - The sar an iostat system monitoring commands." from the Optional Packages list.
X Window System
Complete the installation. After the Installation compelets, RHEL 5.2 and below will hang on booting when it reaches to "starting udev" line. To solve this problem, shutdown the Vmware machine and change the CPU count and Core Count to only one. Implement the changes below, then shutdown the machine, set CPU count back to 2 and startup the machine. put the kernel command line parameters at the end of the "kernel" line: vi /boot/grub/grub.conf add divider=10 clocksource=acpi_pm For example: kernel /vmlinuz-2.6.18 .. clock=acpi_pm divider=10
For Vmware machines, install VMWare tools and set it to synchronize its time with the guest: vmwaretoolbox. Alternatvily, you can use Oracle Cluster Time Synchronization Service (ctssd) (metalink document 551704.1) Install further packages: # to know distribution and version of Linux (Red Hat Ent. 5.2 used) cat /etc/issue
Page 532
# to know kernel version (and its errata level) (2.6.18-92 or newer) uname -r # to list missed packages: rpm -q --qf '%{NAME}-%{VERSION}-%{RELEASE} (%{ARCH})\n' binutils \ compat-libstdc++-33 \ elfutils-libelf \ elfutils-libelf-devel \ gcc \ gcc-c++ \ glibc \ glibc-common \ glibc-devel \ glibc-headers \ ksh \ libaio \ libaio-devel \ libgcc \ libstdc++ \ libstdc++-devel \ make \ sysstat \ unixODBC \ unixODBC-devel # for missed packages, install them: rpm -Uvh libaio-devel-0.3.106-3.2.i386.rpm rpm -Uvh unixODBC* # Download the appropriate ASMLib RPMs from OTN. # to know the kernel verion: uname -rm # In this case we need: rpm -Uvh oracleasm-support-2.1.3-1.el5.i386.rpm rpm -Uvh oracleasm-2.6.18-92.el5-2.0.5-1.el5.i686.rpm rpm -Uvh oracleasmlib-2.0.4-1.el5.i386.rpm
# SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r Check the hardware requirements # Hardware Requirements (in cluster nodes) # At least 1.5 GB of physical memory but practically 1.5 is not fine grep MemTotal /proc/meminfo # swap space: same as the amount of physical memory grep SwapTotal /proc/meminfo # to display swap and memory in one command: free # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: Oracle DBA Code Examples
Page 533
dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap # 1 GB disk space in /tmp df -h /tmp # 8 GB of disk space for Oracle software df The size of the shared memory should be at least the greater of MEMORY_MAX_TARGET and MEMORY_TARGET for each Oracle instance on the computer. To determine the amount of shared memory available, enter the following command: # df -h /dev/shm/ Create the required network configuration (rac2 will be created later): o o o Public and Private interface names must be the same for all nodes. This private hostname does not need to be resolvable through DNS and should be entered in the /etc/hosts file. SCAN VIPs must NOT be in the /etc/hosts file, it must be resolved by DNS. But here I've defined it as a single IP address in the "/etc/hosts" file, which is wrong and will cause the cluster verification to fail, but it allows me to complete the install without the presence of a DNS. If you are using a DNS, Oracle recommends that you add lines to the /etc/hosts file on each node, specifying the public IP, VIP and private addresses. If you configured the IP addresses in a DNS server, then, as the root user, change the hosts search order in /etc/nsswitch.conf on all nodes as shown: Old: hosts: files nis dns New: hosts: dns files nis o Then restart nscd daemon on each node: /sbin/service nscd restart
o o
# Network names Resolution # configure /etc/hosts if no domain server is used (both nodes) vi /etc/hosts 127.0.0.1 localhost.localdomain localhost #eth0 - PUBLIC 192.0.2.100 rac1.mydomain.com rac1 192.0.2.101 rac2.mydomain.com rac2 #VIP 192.0.2.102 rac1-vip.mydomain.com rac1-vip 192.0.2.103 rac2-vip.mydomain.com rac2-vip #eth1 - PRIVATE 172.0.2.100 rac1-priv 172.0.2.101 rac2-priv # in real production: the follwing should not be there at all # SCAN: cluster_name-scan.GNS_subdomain_name 192.0.2.104 rac-scan.mydomain.com rac-scan Create and configure the required OS users and groups
Page 534
Note: userid and groupid must be the same in all nodes. You can check them by id oracle command. # all group and user ids on all the nodes must have identical id # Grid Infrastructure (GI) and the Oracle RDBMS home will # be installed using different users: /usr/sbin/groupadd -g 501 oinstall /usr/sbin/groupadd -g 502 dba /usr/sbin/groupadd -g 504 asmadmin /usr/sbin/groupadd -g 506 asmdba /usr/sbin/groupadd -g 507 asmoper /usr/sbin/useradd -u 501 -g oinstall -G asmadmin,asmdba,asmoper grid /usr/sbin/useradd -u 502 -g oinstall -G dba,asmdba oracle # set passwords passwd oracle passwd grid # make sure nobody user exists (if not there, create it useradd nobody) id nobody # define the env variables for oracle user vi /home/oracle/.bash_profile # Oracle Settings export EDITOR=vi TMP=/tmp; export TMP TMPDIR=$TMP; export TMPDIR ORACLE_HOSTNAME=rac1.mydomain.com; export ORACLE_HOSTNAME ORACLE_UNQNAME=ron; export ORACLE_UNQNAME ORACLE_BASE=/u01/app/oracle; export ORACLE_BASE ORACLE_HOME=$ORACLE_BASE/product/11.2.0/db_1; export ORACLE_HOME ORACLE_SID=RON1; export ORACLE_SID ORACLE_TERM=xterm; export ORACLE_TERM PATH=/usr/sbin:$PATH; export PATH PATH=$ORACLE_HOME/bin:$PATH; export PATH LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib; export LD_LIBRARY_PATH CLASSPATH=$ORACLE_HOME/JRE:$ORACLE_HOME/jlib:$ORACLE_HOME/rdbms/jlib; export CLASSPATH # shell startup file vi /etc/profile if [ $USER = "oracle" ] || [ $USER = "grid" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi umask 022 fi
# for C shell vi /etc/csh.login if ( $USER = "oracle" || $USER = "grid" ) then limit maxproc 16384 limit descriptors 65536 endif # define the env variables for oracle user
Page 535
vi /home/grid/.bash_profile # Oracle Settings export ORACLE_SID=+ASM1 export EDITOR=vi export ORACLE_HOME=/u01/11.2.0/grid export LD_LIBRARY_PATH=$ORACLE_HOME/lib export PATH=$ORACLE_HOME/bin:/bin:/usr/bin:/usr/sbin:/usr/local/bin Configure kernel parameters and shell limits Note: If you make a mistake with a parameter setting and your system does not start, then you must start Linux in the single-user runlevel (runlevel 1). At this runlevel, the /etc/sysctl.conf file is not run. # Kernel Parameters # to tune thme, refer to metalink document 169706.1 # Append the following to the /etc/sysctl.conf file as the root user: vi /etc/sysctl.conf # kernel.shmmax not stated in 11g R2 (max: 4G) (169706.1) kernel.shmmni = 4096 kernel.sem = 250 32000 100 128 fs.aio-max-nr = 1048576 fs.file-max = 6815744 net.ipv4.ip_local_port_range = 9000 65500 net.core.rmem_default = 262144 net.core.rmem_max = 4194304 net.core.wmem_default = 262144 net.core.wmem_max = 1048576 # to take immediate effect /sbin/sysctl -p # User Shell Limits # memlock is used to increase the per-process max locked memory vi /etc/security/limits.conf grid soft nproc 2047 grid hard nproc 16384 grid soft nofile 1024 grid hard nofile 65536 oracle soft nproc 2047 oracle hard nproc 16384 oracle soft nofile 1024 oracle hard nofile 65536 vi /etc/pam.d/login session required pam_limits.so Create the required directories for the Oracle software: # to know if there is an existing oracle inventory # from its output, ORACLE_BASE will be parent of oraInventory more /etc/oraInst.loc # to identify existing Oracle home directories more /etc/oratab # Oracle Inventory Directory # as a root mkdir -p /u01/app/oraInventory chown -R grid:oinstall /u01/app/oraInventory
Page 536
chmod -R 775 /u01/app/oraInventory # Grid Infrastructure Home Directory mkdir -p /u01/11.2.0/grid chown -R grid:oinstall /u01/11.2.0/grid chmod -R 775 /u01/11.2.0/grid # Oracle Base Directory mkdir -p /u01/app/oracle #needed to ensure that dbca is able to run after the rdbms installation mkdir /u01/app/oracle/cfgtoollogs chown -R oracle:oinstall /u01/app/oracle chmod -R 775 /u01/app/oracle # Oracle mkdir -p chown -R chmod -R RDBMS Home Directory /u01/app/oracle/product/11.2.0/db_1 oracle:oinstall /u01/app/oracle/product/11.2.0/db_1 775 /u01/app/oracle/product/11.2.0/db_1
Disable screensavers after logging as root, oracle and grid: o In Oracle Linux: Applications-> Preferences-> Screen Saver
In rac1, partition the disks:
Note: On a real life storage, you would create a single whole-disk partition with exactly 1 MB offset on each LUN to be used as ASM Disk. In fdisk: u (to change units from cylinder to sectors), n, p, 1, 2048, w. # as a root, for the disks /dev/sdb .. /dev/sdd # confirm they are seen: ls /dev/sd* #partition the disks: fdisk /dev/sdb # answers: "n", "p", "1", "Return", "Return", "p" and "w" Note: if the following message appears after the "w" command: WARNING: Re-reading the partition table failed with error 16: Device or resource busy, then you can avoid restarting the machine by the following command: partprobe # to make sure partions are created ls -lX /dev/sd* Shutdown rac1 and the Vmware machine soiftware. Then edit its VMware file of rac1 (with vmx extensions) and add the following entry to allow sharing the disks (make sure the scsi controller number is the one you used): disk.locking = "FALSE" diskLib.dataCacheMaxSize = "0" diskLib.dataCacheMaxReadAheadSize = "0" diskLib.dataCacheMinReadAheadSize = "0" diskLib.dataCachePageSize = "4096" scsi1.sharedBus = "virtual" scsi1:1.deviceType = "disk" scsi1:2.deviceType = "disk" scsi1:3.deviceType = "disk" Copy the folder containing rac1 into a new folder in the same directory structure level. Let's name it "rac2". This will be the second node in the cluster. Edit the VMware file of rac2 and edit the following: displayName = "rac2" Oracle DBA Code Examples
Page 537
Open rac2, then perform: o o o in a terminal issue: system-config-network-gui Remove the devices with the "%.bak" nicknames. To do this, highlight a device, deactivate, then delete it. Highlight the "eth0" interface and click the "Edit" button. Change its IP addresses and gate way: IP 192.0.2.101. Click on the "Hardware Device" tab and click the "Probe" button. For eth1 set its ip address to 172.0.2.101. Do not define a gateway. In DNS tab, change hostname to rac2.mydomain.com. Activate the network cards.
o o o
In rac2, perform: # change the variable in the file vi /home/oracle/.bash_profile # mark the following: # ORACLE_SID=RON1; export ORACLE_SID ORACLE_HOSTNAME=rac2.localdomain; export ORACLE_HOSTNAME vi /home/grid/.bash_profile export ORACLE_SID=+ASM2
Start rac1. Make sure the machines can see each other: ping ping ping ping -c -c -c -c 3 3 3 3 rac1 rac1-priv rac2 rac2-priv
In rac1, configure ASM drivers: Note: If you see that the shared disks are not synced between rac1 and rac2, one of the things you can examine is to see if there is any "debug" command in any of the nodes' vmx files. If you find one, shutdown the node, remove the command from the vmx file and restart. # as root oracleasm configure -i Default user to own the driver interface []: grid Default group to own the driver interface []: oinstall Start Oracle ASM library driver on boot (y/n) [n]: y Fix permissions of Oracle ASM disks on boot (y/n) [y]: y # In all nodes: Load the kernel module using the following command: /usr/sbin/oracleasm init # If you have any problems, make sure you have the correct # version of the driver: /usr/sbin/oracleasm update-driver # mark the shared disks: (one node) /usr/sbin/oracleasm createdisk DISK1 /dev/sdb1 /usr/sbin/oracleasm createdisk DISK2 /dev/sdc1 /usr/sbin/oracleasm createdisk DISK3 /dev/sdd1 # check the disks are marked and seen: /usr/sbin/oracleasm listdisks
Page 538
# in other nodes: /usr/sbin/oracleasm scandisks /usr/sbin/oracleasm listdisks #If you need to unmark a disk that was used in a createdisk command: /usr/sbin/oracleasm deletedisk DISK1 /usr/sbin/oracleasm deletedisk DISK2 /usr/sbin/oracleasm deletedisk DISK3 2. Oracle Grid Infrastructure installation # in rac1: copy the software in a staging folder mkdir -p /u01/stage/ora11gr2gridinfra chown -R grid:oinstall /u01/stage/ora11gr2gridinfra chmod -R 775 /u01/stage/ora11gr2gridinfra mkdir -p /u01/stage/ora11gr2db chown -R oracle:oinstall /u01/stage/ora11gr2db chmod -R 775 /u01/stage/ora11gr2db # do not use cluvfy because SSH was not configured.
# if you are installing 11.2.0.2: install the package cvuqdisk-1.0.9-1 su cd /u01/stage/ora11gr2gridinfra/stage/cvu/cv/remenv/ rpm -iv cvuqdisk-1.0.9-1.rpm # install it in rac2 as well
# lunch OUI from the clusterware ( as grid from rac1) # if logged in in Genome using another user, log out and log in as grid cd /u01/app/stage/ora11gr2gridinfra ./runInstaller Installation Option >Select radio button 'Install and Configure Grid Infrastructure for a Cluster' >Next Installation Type >Select 'Advanced Installation' >Next Product Language >Accept 'English' as language' >Next Grid Plug and Play >cluster name: rac >SCAN name:rac-scan.mydomain.com >Make sure 'Configure GNS' is NOT selected >Next Cluster Node Information >Add button >Hostname:rac2.mydomain.com >Virtual IP Name: rac2-vip.mydomain.com >OK
Page 539
>"SSH Connectivity" button >Enter the password >Setup button >Test button >Next Network Interface Usage >check the public and private networks are specified correctly >Next Storage Option >Select 'Automatic Storage Management (ASM)' >Next Creat ASM Disk Group >Disk Group Name: DGOCRVOTE (3GB disk: Disk1) >Redundancy: external >Next NOTE: If you see an empty screen for you candidate disks it is likely that ASMLib has not been properly configured. Try reconfigure them. If you are sure that ASMLib has been properly configured click on 'Change Discovery Path' and provide the correct destination. ASM Password >Specify and conform the password you want to use >Next Failure Isolation Support >Select NOT to use IPMI >Next Privileged OS Groups >Assign the correct OS groups for OS authentication (mostly default is OK) >Next Installation Location >ORACLE_BASE: /u01/app/oracle Software location: /u01/11.2.0/grid >Next Create Inventory >Specify the locations: /u01/app/oraInventory >Next Perform Prerequisite Checks >OUI performs certain checks >Check that status of all checks is Succeeded Note: in this example, NPS error can be ignored # 11.2.0.2 returns "PRVF-5150: Path ORCL: is not a valid path on all nodes" # in this case: verify manually the asm device. If it succeeds, ignore the # error. If it fails, run /etc/init.d/oracleasm configure -i on all nodes and # set grid user and oinstall as a group (Note ID 1210863.1) # For the error "PRVF-5636 : The DNS response time for an unreachable node # exceeded "15000" ms", the response time can be measured by the command: # time nslookup rac1
Page 540
>Next Summary >Finish /* For 64-bit, before you run root.sh apply patch# 9974223 (this patch should also be applied on RDBMS) : export PATH=$PATH:/u01/app/11.2.0/grid/OPatch opatch version cd /u01/app/oracle/admin/patches opatch lsinventory -detail -oh /u01/app/11.2.0/grid unzip p9974223_112020_Linux-x86-64.zip pwd /u01/app/oracle/admin/patches opatch napply -local -oh /u01/app/11.2.0/grid -id 9974223 opatch lsinventory -detail -oh /u01/app/11.2.0/grid */ Execute Configuration Scripts >Run the scripts as instructed in the screen Note: The scripts must be run on one node at a time. /* If you face problems in running root.sh and you want to deconfigure the previous run of root.sh: /u01/11.2.0/grid/crs/install/rootcrs.pl -verbose -deconfig -force */ >OK /* Note: when I ran root.sh in node2, I faced "ORA-15018: diskgroup cannot
be created". I deconfigured what made by root.sh, issued oracleasm stop, start and then ran root.sh again */
We expect the verification phase to fail with an error relating to the SCAN, assuming you are not using DNS. INFO: Checking Single Client Access Name (SCAN)... INFO: Checking name resolution setup for "rac-scan.localdomain"... INFO: ERROR: INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "rac-scan.localdomain" INFO: ERROR: INFO: PRVF-4657 : Name resolution setup check for "rac-scan.localdomain" (IP address: 192.168.2.201) failed INFO: ERROR: INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "rac-scan.localdomain" INFO: Verification of SCAN VIP and Listener setup failed Provided this is the only error, it is safe to ignore this >Next Message: The installation of the Grid Infrastructure was successfull. >Close
Note: If your OS is SUSE Linux, shutting down on node will result in shutting the other nodes. To workaround: #cd /etc/rc3.d #ln -s /etc/init.d/ohasd K07ohasd
Page 541
3. Oracle Grid Infrastructure Patching at time of this writing (Jan, 2011), no patches applied. 4. Checking Oracle Grid Infrastructure Status # Application resources crsctl stat res -t
# Oracle Cluster Ready Services (CRS) crsctl check crs
# Oracle Cluster Registry (OCR) and Voting DiskCluster Ready Services (CRS) ocrcheck
# Oracle High Availability Services autostart su /u01/11.2.0/grid/bin/crsctl config crs
# Oracle Automatic Storage Management (ASM) su - grid srvctl config asm -a
# Oracle Listener srvctl config listener -a
# Single Client Access Name (SCAN) srvctl config scan 5. Oracle Database 11g R2 Software Installation # as oracle cd /u01/app/stage/ora11gr2db/db unzip L11gR2_database_1of2.zip >/dev/null unzip L11gR2_database_2of2.zip >/dev/null cd database ./runInstaller
Configure Security Updates >Provide your e-mail address, if you want or leave them blank >Next Installation Options >Select 'Install Database software only' >Next> Install Type in 11.2.0.1: >Select 'Real Application Clusters database installation', and select all nodes.
Page 542
in in 11.2.0.2: >select Oracle RAC One Node installation >Use the 'SSH Connectivity' button to configure/test the passwordless SSH connectivity. >Next Product Languages >Confirm 'English' >Next Database Edition >'Enterprise Edition' is ticked Optional: Click on Select Options> Select "Oracle Partitioning" and "Oracle Real Application Testing" >OK >Next Installation Location >Oracle Base: /u01/app/oracle Software Location: /u01/app/oracle/product/11.2.0/db_1 >Next Privileged OS Groups >OSDBA: dba >OSOPER: oinstall >Next Prerequisite Checks ..OUI performs prerequisite checks >Check that status of all checks is Succeeded >If you are sure the unsuccessfull checks can be ignored tick the box 'Ignore All' >Next Summary >Check summary info >Finish Install Product ..OUI installs the db software >as a root, run the root.sh script on the first node then the other nodes (One at a time) >OK Finish >Close 6. Oracle Database 11g R2 Software Patching Patch 9004119 should be applied on 11.2.0.1. Do not apply it to 11.2.0.2. The patch just installes the following scripts which are used for RAC One Node: o o o o o
Page 543
raconefix raconeinit raconestatus racone2rac Omotion
Fixes metadata after an Omotion failure or failover Initialize the database to RAC One Node Check the status of RAC One Node database Upgrade RAC One Node database to RAC Migrate database online from one node to another Oracle DBA Code Examples
# at time of this writing, Patch #9004119 should be applied for RAC One Node /* Download and apply patch 9004119 */ -- the patch will be applied in a rolling forward style su - oracle export PATH=$PATH:$ORACLE_HOME/OPatch unzip p9004119_112010_LINUX.zip cd 9004119 # Verify the OUI Inventory is accessible by OPatch: opatch lsinventory
# apply the patch opatch apply # NOTE: the patch asks you to shutdown the instance before applying the patch. In this case, you do not need to do so because the patch just installs some scripts. Proceed with applying the script without stopping the instances in neither rac1 nor rac2. # make sure the scripts were installed in both nodes: cd $ORACLE_HOME/bin ls racone* Omotion 7. Install EM Agent in cluster nodes (if required)
8. ASM Diskgroups Creation Note: It is Oracle's Best Practise to have an OCR mirror stored in a second disk group. To follow this recommendation add an OCR mirror. Mind that you can only have one OCR in a diskgroup. To add OCR mirror to an Oracle ASM disk group, ensure that the Oracle Clusterware stack is running and ocrconfig -add +ORADATA ocrcheck # as grid user: start the ASM Configuration Assistant (ASMCA) #su - grid cd /u01/11.2.0/grid/bin ./asmca >Disk Groups tab >Create button >Disk Group Name: DGDATA >Redundancy: External >Disk2 >OK >Create button >Disk Group Name: DGFRA >Redundancy: External >Disk3 >OK >Exit >Yes
Page 544
9. RAC Database Creation After the database is created, create a service. # as oracle cd /u01/app/oracle/product/11.2.0/db_1/bin ./dbca Welcome 11.2.0.1: Select 'Oracle Real Application Clusters database' 11.2.0.2: Select 'Oracle RAC One Node Database' >Next Operations > choose option 'Create a Database' >Next Database Template >Select General Purpose or any template >Next Database Identification >Configuration Type: Admin >Globale Database Name: ron >SID: ron 11.2.0.2: >Serive Name: ronsrv 11.2.0.1: > Select the nodes: rac1. Make sure you select just ONE NODE. 11.2.0.2: > Select the nodes: rac1, rac2 >Next Management Options >Select the option you want. I selected "Configure Enterprise Manager" >Next Database Credentials >Set the password(s) >Next Database File Locations >Oracle-Managed Files >Database Area: +DGDATA >Practically (but not in this case), you should define 'Multiplex Redo Logs and Control Files'. >Next /* Note: If you cannot see the diskgroups, perform the following (ID: 1177483.1): su cd <Grid_Home>/bin chmod 6751 oracle ls -l oracle -rwsr-s--x 1 grid oinstall */
Page 545
ASM Credentials ..If you chose to set up EM, you will be asked about ASMSNMP password >Enter the password >Ok button Recovery Configuration >Flash recovery area: +DGFRA >define the size: 2000 MB If the size is smaller than recommended a warning will popup. >Next Database Content >Select if you want to have sample schemas created in your database >Next Initialization Parameters >Review and change the settings for memory allocation, characterset etc. >Next Database Storage >Review the database storage settings and change as required >Next Creation Options >Make sure the tickbox 'Create Database' is ticked >Finish Summary >OK .. Database creation proceeding >after completion Exit
/* Fix /home/oracle/.bash_profile */ vi /home/oracle/.bash_profile ORACLE_SID=RON_1
/* Check Confirmation */ # to show the current configuration and status of the RAC database srvctl config database -d ron srvctl status database -d ron sqlplus /nolog conn system@ron select instance_name, status from v$instance ; select name, db_unique_name from v$database;
# check OEM (if configured): https://rac1.mydomain.com:1158/em/ # if not started, you can start it: su - oracle cd /u01/app/oracle/product/11.2.0/db_1/bin export ORACLE_UNQNAME=ron export ORACLE_SID=RON_1
Page 546
./emctl status dbconsole ./emctl start dbconsole
# check Oracle processes: ps -eo pid -o command | grep ora_ | grep -v grep
/* In 11.2.0.1: Create the db service(s) */ OEM home-> Availability-> Cluster Managed Database Services-> enter credentials-> Create Service-> enter Service Name: hr_srv-> Ok button # check service status: srvctl config service -d ron 10. Initialize the Database to RAC One Node (11.2.0.1 Only) The raconeinit utility renames the db instance and creates the directories and files supporting the renamed instance. Renaming the instance name resulted in the DB Control to return the error: "ORA-12505: TNS:listener does not currently know of SID given in connect descriptor". Personally, the only solution I found was to recreate the DB Control configuration as follows. Refer to the "Administrator's Guide" for further details about administering DB Control. srvctl status database -d ron raconestatus raconeinit Output>> Candidate Databases on this cluster: # Database RAC One Node Fix Required === ======== ============ ============ [1] ron NO N/A Enter the database to initialize [1]: Database ron is now running on server rac1 Candidate servers that may be used for this DB: rac2 Enter the names of additional candidate servers where this DB may run (space delimited): rac2 Please wait, this may take a few minutes to finish....... Database configuration modified.
# check the new instance was created: srvctl status database -d ron srvctl config database -d ron
# if DB Control returns ORA-12505, you can recreate its repository as follows: /* Drop the DB Config Files and Rep Objects */ # Remove the following directories from your file system: # $ORACLE_HOME/hostname_sid # $ORACLE_HOME/oc4j/j2ee/OC4J_DBConsole_hostname_sid rm -r $ORACLE_HOME/rac1_ron rm -r $ORACLE_HOME/oc4j/j2ee/OC4J_DBConsole_rac1_ron # $ORACLE_HOME/sysman/admin/emdrep/bin/RepManager hostname listener_port sid action drop $ORACLE_HOME/sysman/admin/emdrep/bin/RepManager rac1 1521 ron_1 -action drop
Page 547
/* Add DB Control Config */ [oracle@rac1 ~]$ $ORACLE_HOME/bin/emca -config dbcontrol db -repos create cluster STARTED EMCA at Jan 4, 2011 12:18:02 PM EM Configuration Assistant, Version 11.2.0.0.2 Production Copyright (c) 2003, 2005, Oracle. All rights reserved. Enter the following information: Database unique name: ron Service name: ron Listener port number: 1521 Listener ORACLE_HOME [ /u01/11.2.0/grid ]: /u01/11.2.0/grid Password for SYS user: Password for DBSNMP user: Password for SYSMAN user: Cluster name: ron Email address for notifications (optional): Outgoing Mail (SMTP) server for notifications (optional): ASM ORACLE_HOME [ /u01/11.2.0/grid ]: /u01/11.2.0/grid ASM port [ 1521 ]: ASM username [ ASMSNMP ]: ASM user password: 11. Postinstallation tasks # backup the root.sh script (on all nodes) cp /u01/app/oracle/product/11.2.0/db_1/root.sh ~/root.sh.bak 12. General Useful Postinstallation Tasks in Linux Following are tips to consider after the successful installation to make managing RAC easier. Consider using rlwrap utility with SQL*Plus and RMAN: o o Using rlwrap Utility with RMAN in Unix-Based Systems Using rlwrap Utility with SQL*Plus in Unix-Based Systems
/* Easy Acces to crs and db homes */ # it is common to access bin directories in clusterware and db homes # add the following to .bashrc of oracle user alias db='cd /u01/app/oracle/product/11.2.0/db_1/bin'
Page 548
Instance Relocation using Omotion (11.2.0.1)

"Omotion" moves one instance from one node to another in a cluster. To migrate an instance without client interruption, implement FAN. You can then also enable TAF. You can also provide the database name and migration time as inputs on the command line: ./Omotion -d adm2 -m 5 # in client tnsnames.ora implement TAF: RONTAF = (DESCRIPTION = (ENABLE = BROKEN) (LOAD_BALANCE = OFF) (FAILOVER = ON) (ADDRESS = (PROTOCOL = TCP)(HOST = rac-scan.mydomain.com)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = ron) (FAILOVER_MODE = (TYPE = SELECT) (METHOD = BASIC) (BACKUP = RON) ) ) )
# check on which instance the db is running: srvctl status database -d ron
# to test the TAF, issue a long query in a client
# migrate the instance (-v verbose): ./Omotion -v
Page 549
Instance Relocation using Omotion (11.2.0.2)

srvctl relocate database -d rac1 -n node7 Coninue>>
Page 550
Part 9
Oracle Warehousing
Page 551
Oracle Warehouse Builder (OWB)

Oracle Warehouse Builder Architecture
Starting and Stoping the Service

Starting and Stopping OWB Service -- stop service sqlplus OWBSYS/<password> @<ORACLE_HOME>\stop_service.sql -- start service sqlplus OWBSYS/<password> @<ORACLE_HOME>\start_service.sql -- if start fails, check the problem using the following script: sqlplus OWBSYS/<password> @<ORACLE_HOME>\service_doctor.sql -- if OWBSYS user is locked or its password was modified, the start -- will fail. Store the changed password in the repository using: sqlplus OWBSYS/<password> @<ORACLE_HOME>\set_repository_password.sql
Configuring the Repository and Workspaces

Goto Start>Oracle Home> Warehouse Builder>Administration>Repository Assistant on UNIX, locate OWB_ORACLE_HOME/owb/bin/ and run reposinst.sh Select Manage Workspaces> Create a New Workspace> Workspace Owner> ... The user must have OWB_USER role. Select a base language for the repository (cannot be changed later) Select Workspace users
Page 552
Steps of Using Warehouse Builder

Design the target structure o Design a dimensional model and implement it in a database as: Relational implementation (star schema) Multidimensional implementation (OLAP) Create a project Tip: to delete a project, expand another project then right click on the project you want to delete. Create source module(s): o o Oracle module: specify hostname:port rather than using SQL*Net Connection. Non-Oracle module: create ODBC connection to the database then configure Oracle to connect to the ODBC connection: Copy the file ORACLE_HOME\hs\admin\initdg4odbc.ora to init<odbc connection>.ora HS_FDS_CONNECT_INFO = <odbc data_source_name> HS_FDS_TRACE_LEVEL = <trace_level> For example: HS_FDS_CONNECT_INFO = MYSQLSERVER HS_FDS_TRACE_LEVEL = 0 Edit the listener.ora then restart the listener: # PROGRAM=dg4odbc in Oracle 11g # PROGRAM=hsodbc in 10g SID_LIST_LISTENER= (SID_LIST= (S Global Explorer> Secutiryt>Users >NewID_DESC= (SID_NAME=mysqlserver) (ORACLE_HOME=D:\app\bob\product\11.1.0\db_1) (PROGRAM=dg4odbc) ) ) Right-click on the ODBC node in the Design Center and select New Edit the connection. Make sure that both username and password are enclosed in double quotes. Enter the Host where the Oracle database resides. Tip: to delete a connection, unregister it first from Control Manager then delete it from Design Center. Note: The transparent gateway agents must be purchased and installed separately from the Oracle Database, and then configured to support the communication with the external database. The generic connectivity agent is included with the Oracle Database. o Import source metadata from files: Files> New> define path (not file) On the created object> Import..> Sample>.. >Finish
Page 553
Import source metadata: by using the Import Metadata Wizard or manually. Manual method: under Tables node, New> Data Object Editor. Create the target user Global Explorer> Secutiryt>Users >New (always created in the db repository) Create the target module Create target dimensions in OWB: o o Dimension: Dimension Attributes, Levels, Level Attributes and Hierarchies Time dimension: Project Explorer> Databases> Ourproject> Dimensions> New> Using Time Wizard. A mapping and a sequence will be as a result created.
Create target cubes in OWB Design staging area table(s) Create a mapping from source to staging table: add source tables, add the staging target table then link from source to staging target. Linking may include: o o Joiner: edit it to add more Input Groups, link source tables to it, define join condition. Note: Oracle 10.2.0.4 returns "Bad expression return type" error (bug ID 7417869). Aggregator: link output of Joiner to its input. Set its Group By Clause setting. Add required aggregation functions (usually SUM) by right clicking on OUTGRP1> Open details> Output Attributes> Add button to add something like AMOUNT> OK> click on AMOUNT in the aggregator> Properties Window> Expression> set the function> OK Transformation
Validate the mapping Generate the mapping Deploy the mapping Execute the mapping
Mapping Operators
Source and Target Operators: Cube, Dimension, External Table, Table, Constant, View, Sequence, Construct (returns SYS_REFCURSOR) Data Flow Operators: Aggregator, Deduplicator (distinct SQL function), Expression, Filter, Joiner, Key Lookup, Pivot, Set Operation, Splitter, Transformation, Table Function Pre/Post Processing Operators: Mapping Input Parameter, Mapping Output Parameter, Post-Mapping Process, Pre-Mapping Process Pluggable Mappings: group of operators act as single operator
Page 554
Page 555
Part 10
Oracle Database Utilities
Page 556
Using SQL*Plus
Using SQL*Plus Command-Line Options
sqlplus [ [<option>] [<logon>] [<start>] ]
<start> is: @<URL>|<filename>[.<ext>] [<parameter> ...]
The silent option (-S): no output on screen The no-prompt logon option (-L): no username/password question after login fail The markup option (-M): HTML generation sqlplus help
Starting SQL*Plus Session

$export ORACLE_HOME= /u01/app/oracle/product/11.1.0/db_1 $ORACLE_SID=orcl $export ORACLE_SID NLS_LANG and ORA_NLS11 sqlplus oe/oe@//myserver.mydomain:1521/testdb sqlplus username/passwd sqlplus username/passwd AS SYSDBA sqlplus username/passwd@(DESCRIPTION =(ADDRESS=(PROTOCOL=tcp)(HOST=salesserver)(PORT=1521)(CONNECT_DATA= (SERVICE_NAME=orcl.mycompany.com)))
Controlling User Privileges in SQL*Plus

Product_user_profile table controls what commands a user can issue in SQL*Plus session. The following commands can be prevented by using this feature. SQL: ALTER, AUDIT, ANALYZE, CREATE, DELETE, DROP, INSERT, LOCK, NOAUDIT, RENAME, SELECT, UPDATE, VALIDATE, TRUNCATE, GRANT, REVOKE, SET ROLE, SET TRANSACTION PL/SQL:DECLARE, BEGIN SQL*PLUS: COPY, HOST, SET, EDIT, PASSWORD, SPOOL, EXECUTE, QUIT, START, EXIT, RUN, GET, SAVE You can use the RESTRICT command to prevent users from using certain operating system commands. It disables the commands even where there are no connections to the server. Command EDIT GET HOST Disabled Disabled Disabled SAVE SPOOL START STORE
Page 557
Level 1
Level 2
Level 3
Disabled Disabled Disabled Disabled
Disabled Disabled Disabled Disabled Disabled Disabled Disabled Oracle DBA Code Examples
SQL> DESC product_user_profile Name Null? Type -----------------------------PRODUCT NOT NULL USERID ATTRIBUTE case SCOPE NUMERIC_VALUE CHAR_VALUE DATE_VALUE LONG_VALUE
VARCHAR2(30) VARCHAR2(30) VARCHAR2(240) -- command to disable in upper VARCHAR2(240) NUMBER(15,2) VARCHAR2(240) -- role name to disable DATE LONG
-- no insert INSERT INTO product_user_profile VALUES ('SQL*PLUS','OE','INSERT',NULL,NULL,NULL,NULL,NULL); -- no OS command INSERT INTO product_user_profile (product,userid,attribute) VALUES ('SQL*Plus','myuser','HOST'); -- no set Role to DBA insert into product_user_profile(product, userid, attribute, char_value) values('SQL*Plus', 'APPS', 'ROLES', 'DBA'); -- Preventing access using PL/SQL insert into system.product_profile (product, userid, attribute, char_value) values ('SQL*Plus', 'AMAR', 'DECLARE', 'DISABLED'); insert into system.product_profile (product, userid, attribute, char_value) values ('SQL*Plus', 'AMAR', 'BEGIN', 'DISABLED'); sqlplus -RESTRICT 1
Setting the SQL*Plus Environment with the SET Command

help index help set list of available commands by typing variable possible values
AUTO[COMMIT] Specifies whether commits of transactions are automatic or manual. DEF[INE]{&/C/ON/OFF} Sets the prefix character used during variable substitutions. ECHO {OFF/ON} onscreen. LONG {80/n} when ON, each command will be displayed before its output
FEED[BACK] {OFF/ON} whether to show the number of records returned by your query. maximum width of the LONG, CLOB, NCLOB, and XMLType values NEWP[AGE] {1/n/none} Specifies the number of blank lines at the top of each new page. PAGES[IZE] {24/n} Specifies the number of lines in each page TI[ME] {OFF/ON} Displays time if set to on. TIMI[NG] {OFF/ON} Controls the display of timing for SQL commands. VER[IFY] {OFF/ON} Specifies whether SQL text is displayed after variable substitution.
Setting SQL*Plus Preferances

Globally glogin.sql User level login.sql: in Oracle 11g, executed even after connect command
Page 558
Logging SQL*Plus Errors

In Oracle 11g, SET ERRORLOGGING ON command writes all subsequent errors into a table called sperrorlog. SET ERRORLOGGING ON SHOW ERRORLOGGING
Key SQL*Plus "Working" Commands

Setting SQLPROMPT Command _USER _CONNECT_IDENTIFIER _PRIVILEGE as sysdba _DATE SET SQLPROMPT "_USER > " host HOST cp /u01/app/oracle/new.sql /tmp SPOOL { file_name[.ext] [CRE[ATE]|REP[LACE]|APP[END]]| OFF} SPOOL /u01/app/oracle/data/employees.txt; SELECT * FROM hr.employees; SPOOL OFF; -- ACCEPT and PROMPT Commands ACCEPT MYTITLE PROMPT 'Title: ' -- using PAUSE to check long listing output SQL> SHOW PAUSE PAUSE is OFF SQL> SET PAUSE ON SQL> very long list is displayed
Creating Command Files in SQL*Plus

SAVE status.sql -- set the search directory using env var ORACLE_PATH @myscriptfile.sql -- Creating a Windows Batch Script sqlplus username/password@connect_identifier @C:\temp\testscript.sql -- substitution vars DEFINE owner = '&1' SELECT segment_name,segment_type,extents FROM dba_segments WHERE owner = upper ('&owner');
Copying Tables with the COPY Command

commits while data are moved slower than CREATE TABLE .. AS SELECT Oracle DBA Code Examples
Page 559
SQL> COPY usage: COPY FROM <db> TO <db> <opt> <table> { (<cols>) } USING <sel> <db> : database string, e.g., hr/your_password@d:chicago-mktg <opt> : ONE of the keywords: APPEND, CREATE, INSERT or REPLACE <table>: name of the destination table <cols> : a comma-separated list of destination column aliases <sel> : any valid SQL SELECT statement set ARRAYSIZE 100 SQL> COPY FROM sysadm/sysadm1@finance1> CREATE test01 > USING SELECT * FROM employee;
Creating Web Pages Using SQL*Plus

SQLPLUS -MARKUP "HTML ON" SQL> SET MARKUP HTML ON SPOOL ON SQL> commands here . . . SQL> SET MARKUP HTML OFF SPOOL OFF
Using SQL to Generate SQL

SET ECHO OFF HEADING OFF FEEDBACK OFF SPOOL test.txt SELECT 'grant connect, resource to '||username||';' FROM dba_users; @test.txt
Enabling AUTOTRACE for a User

-- create plan_table (if not there) desc plan_table -- login as the user who wants to use the AUTOTRACE conn / as sysdba @d:\oracle\product\10.1.0\db_1\RDBMS\ADMIN\utlxplan.sql -- create plustrace role @d:\oracle\product\10.1.0\db_1\sqlplus\admin\plustrce.sql grant plustrace to scott; -- the following privilege is also required: grant select any dictionary to test; SET AUTOTRACE ON ON STATISTICS ON EXPLAIN TRACEONLY TRACEONLY STATISTICS OFF
Page 560
Using rlwrap Utility with SQL*Plus in Unix-Based Systems

rlwrap utility allows you to navigate history of commands in SQL*Plus and edit them. /* Install */ # Option 1: for rpm version: # can be downloaded from: # http://rpm.pbone.net # http://ivan.kartik.sk rpm -ivh rlwrap*.rpm
# Option 2: for tar version # download rlwrap-0.30.tar.gz (search the net or from http://www.ahmedbaraka.com/download/oracle/rlwrap-0.30.tar.gz ) # unzip the file and install su gunzip rlwrap-0.30.tar.gz tar -xvf rlwrap-0.30.tar cd rlwrap-0.30 ./configure make make install /* Recommended */ vi /home/oracle/.bashrc alias sqlpus='rlwrap sqlplus /nolog' # or echo "alias sqlpus='rlwrap sqlplus /nolog'" >> /home/oracle/.bashrc
Escaping Special Characters

/* Escape quotes */ Use two quotes for every one displayed. Examples: SELECT 'Frank''s Oracle site' AS text FROM DUAL; TEXT -------------------Franks's Oracle site SELECT 'A ''quoted'' word.' AS text FROM DUAL; TEXT ---------------A 'quoted' word. SQL> SELECT 'A ''''double quoted'''' word.' AS text FROM DUAL; TEXT ------------------------A ''double quoted'' word.
/* Escape wildcard characters */ The LIKE keyword allows for string searches. The '_' wild card character is used to match exactly one character, while '%' is used to match zero or more occurrences of any characters. These characters can be escaped in SQL. Examples: SELECT name FROM emp
Page 561
WHERE SELECT WHERE Escape
id LIKE '%/_%' ESCAPE '/'; name FROM emp id LIKE '%\%%' ESCAPE '\'; ampersand (&) characters in SQL*Plus
When using SQL*Plus, the DEFINE setting can be changed to allow &'s (ampersands) to be used in text: SET DEFINE ~ -- if SP2-0317 returned, try SET DEF & SELECT 'Laurel & Hardy' FROM dual; Other methods: Define an escape character: SET ESCAPE '\' SELECT '\&abc' FROM dual; Don't scan for substitution variables: SET SCAN OFF SELECT '&ABC' x FROM dual; Another way to escape the & would be to use concatenation, which would not require any SET commands SELECT 'Laurel ' || '&' || ' Hardy' FROM dual; /* Use the 10g Quoting mechanism: */ Syntax q'[QUOTE_CHAR]Text[QUOTE_CHAR]' Make sure that the QUOTE_CHAR followed by an ' doesn't exist in the text. SELECT q'{This is Orafaq's 'quoted' text field}' FROM DUAL;
Page 562
Using SQL*Loader Utility

Invoking SQL*Loader
SQLLDR keyword=value [,keyword=value,. . .] Parameters can be specified in a parameter file: sqlldr PARFILE=c:\...\myparfile.txt You specify the command-line parameters (which also could be specified in the control file): USERID = usrname/passwored CONTROL = '/../mycontrol.ctl' DATA = '/../mydata.dat' LOG = '/.../mylog.log' BAD = '/.../mybadfile.bad' DISCARD='/../mydiscard.dat' DISCARDMAX=100 SKIP = 235550 if not specified, Oracle will create one discarded data do not meet the criteria
in conventional path mode, bind size in bytes OR number of rows BINDSIZE = 512000 ROWS = 64000 DIRECT=true UNRECOVERABLE=Y ERRORS = 0 LOAD = 10000 SILENT = ALL PARALLEL=true (see direct load options in the following sections) no errors tolerated maximum number of logical records to be loaded into the table all generated message are NOT displayed only when DIRECT=true
RESUMABLE=true default is false RESUMABLE_NAME = finance1_load RESUMABLE_TIMEOUT = 3660 in seconds
Using SQL*Loader Control File

Datafile: o Input File o External Input file: INFILE='/a01/app/oracle/oradata/load/consumer.dat' Inline Data: INFILE = * then BEGINDATA ... CONCATENATE n RECLEN m every n rows of data is considered one logical row, and the row length is m characters CONTINUEIF THIS (1:4) = 'next' if SQL*Loader finds the four letters next at the beginning of a line, it should treat the data that follows as a continuation of the previous line CONTINUEIF LAST = '&' continue if last character is & specific record terminator: by default new line. To change it to something else like the pipe for example: infile "test.dat" "str '|\n'"
Physical and Logical Records
Record Format
Page 563
in this case, the following data are two records (multi-line fields): one line;hello dear world;| two lines;Dear world, hello!;| Loading: o o defining destination table(s): INTO TABLE .. INSERT | REPLACE | APPEND defining rejection condition: WHEN <condition> WHEN(activity_type <>'H') and (activity_type <>'T') Fixed record format: INFILE '..' "fix 12" (12-byte-size records) ahmed,1234, johnso,1234= "ahmed,1234, " and "12johnson,1234" Variable record format: INFILE '..' "var 2" 05ahmed12johnson,1234 = "ahmed" and "12johnson,1234"
Note: OR operator cannot be used in the WHEN condition. To workaraound, load into multiple tables. Check the examples. Table- and Field-Mapping o Fields position (either you use this or the delimiters) Relative: employee_name to start in position 7 and continue for 30 characters employee_id POSITION(*) NUMBER EXTERNAL 6 employee_name POSITION(*) CHAR 30 Absolute employee_id POSITION(1:6) INTEGER EXTERNAL employee_name POSITION(7:36) CHAR
Note: a field can be skipped using FILLER keyword. o Data Types: o INTEGER(n)binary integer, where n can be 1, 2, 4, or 8 SMALLINT CHAR INTEGER EXTERNAL FLOAT EXTERNAL DECIMAL EXTERNAL LOBFILE
Delimiters: using POSITION is faster than delimiters TERMINATED BY WHITESPACE TERMINATED BY "," FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"'
Data Transformation Parameters: SQL function(s) is specified after the data type and should be enclosed in double quotation marks field_name CHAR TERMINATED BY "," "SUBSTR(:field_name, 1, 10)" employee_name POSITION 32-62 CHAR "UPPER(:ename)" salary position 75 CHAR "TO_NUMBER(:sal,'$99,999.99')" commission INTEGER EXTERNAL "":commission * 100"
Command-Line Parameters in the Control File USERID = usrname/passwored CONTROL = '/../mycontrol.ctl' DATA = '/../mydata.dat' LOG = '/.../mylog.log' BAD = '/.../mybadfile.bad' DISCARD='/../mydiscard.dat' DISCARDMAX=100 if not specified, Oracle will create one discarded data do not meet the criteria
Page 564
SKIP = 235550 in conventional path mode, bind size in bytes OR number of rows BINDSIZE = 512000 ROWS = 64000 DIRECT=true ERRORS = 0 LOAD = 10000 SILENT = ALL PARALLEL=true (see direct load options in the following sections) no errors tolerated maximum number of logical records to be loaded into the table all generated message are NOT displayed only when DIRECT=true
RESUMABLE=true default is false RESUMABLE_NAME = finance1_load RESUMABLE_TIMEOUT = 3660 Generating Data o o o o o Constant loaded_by CONSTANT "sysadm" Expression: SQL or PL/SQL function (doesnt work in direct loading) column_name EXPRESSION "SQL string" Record Number in the datafile record_num RECNUM System Date loaded_date sysdate Sequence loadseq SEQUENCE(max,1) in seconds
/* Example 1 */ -- tabe separated with Data to Import: 1 Ahmed Baraka 2 John Rice 3 Emme Rak 4 King Size 5 Small Size
nulls in the data 1000 5000 2500 2700 3000 1.87 2.4 2.34 1-1-2000 10-5-1998
31-3-2001
Table Structure PERSONS (ID NUMBER, PNAME VARCHAR2(100), BALANCE NUMBER, RATE NUMBER, JOIN_DATE DATE ); Control File: OPTIONS ( ERRORS=0) LOAD DATA INFILE 'C:\temp\data\persons.dat' BADFILE 'C:\temp\data\persons.bad' DISCARDFILE 'C:\temp\data\persons.dsc' INTO TABLE "HR"."PERSONS" REPLACE FIELDS TERMINATED BY X'9' TRAILING NULLCOLS
Page 565
(ID INTEGER EXTERNAL, PNAME CHAR, BALANCE INTEGER EXTERNAL, RATE FLOAT EXTERNAL, JOIN_DATE date 'dd-mm-yyyy')
/* Example 2 */ -- positional columns load data infile * replace into table departments ( dept position (02:05) char(4), deptname position (08:27) char(20) ) begindata COSC COMPUTER SCIENCE ENGL ENGLISH LITERATURE MATH MATHEMATICS POLY POLITICAL SCIENCE /* Example 3 */ -- data transformation LOAD DATA INFILE * INTO TABLE modified_data ( rec_no "my_db_sequence.nextval", region CONSTANT '31', time_loaded "to_char(SYSDATE, 'HH24:MI')", data1 POSITION(1:5) ":data1/100", data2 POSITION(6:15) "upper(:data2)", data3 POSITION(16:22)"to_date(:data3, 'YYMMDD')" ) BEGINDATA 11111AAAAAAAAAA991201 22222BBBBBBBBBB990112 LOAD DATA INFILE 'mail_orders.txt' BADFILE 'bad_orders.txt' APPEND INTO TABLE mailing_list FIELDS TERMINATED BY "," ( addr, city, state, zipcode, mailing_addr "decode(:mailing_addr, null, :addr, :mailing_addr)", mailing_city "decode(:mailing_city, null, :city, :mailing_city)", mailing_state, move_date "substr(:move_date, 3, 2) || substr(:move_date, 7, 2)" )
/* Example 4 */
Page 566
-- loading from multiple input files LOAD DATA INFILE file1.dat INFILE file2.dat INFILE file3.dat APPEND INTO TABLE emp ( empno POSITION(1:4) INTEGER EXTERNAL, ename POSITION(6:15) CHAR, deptno POSITION(17:18) CHAR, mgr POSITION(20:23) INTEGER EXTERNAL )
/* Example 5 */ -- loading into multiple tables -- skipping columns (FILLER) -- POSITION(1:4) in the example is a must to resent the pointer back -- to the beginning of the row -- In delimited formats, use "POSITION(1)" after the first -- column to reset the pointer LOAD DATA INFILE * INTO TABLE tab1 WHEN tab = 'tab1' ( tab FILLER CHAR(4), col1 INTEGER ) INTO TABLE tab2 WHEN tab = 'tab2' ( tab FILLER POSITION(1:4), col1 INTEGER ) BEGINDATA tab1|1 tab1|2 tab2|2 tab3|3 -- another example LOAD DATA INFILE 'mydata.dat' REPLACE INTO TABLE emp WHEN empno != ' ' ( empno POSITION(1:4) INTEGER EXTERNAL, ename POSITION(6:15) CHAR, deptno POSITION(17:18) CHAR, mgr POSITION(20:23) INTEGER EXTERNAL ) INTO TABLE proj WHEN projno != ' ' ( projno POSITION(25:27) INTEGER EXTERNAL, empno POSITION(1:4) INTEGER EXTERNAL )
/* Example 6 */ -- work around on being unable to use OR in the WHEN condition LOAD DATA INFILE 'mydata.dat' BADFILE 'mydata.bad' DISCARDFILE 'mydata.dis'
Page 567
APPEND INTO TABLE my_selective_table WHEN (01) <> 'H' and (01) <> 'T' ( region CONSTANT '31', service_key POSITION(01:11) call_b_no POSITION(12:29) ) INTO TABLE my_selective_table WHEN (30:37) = '20031217' ( region CONSTANT '31', service_key POSITION(01:11) call_b_no POSITION(12:29) )
INTEGER EXTERNAL, CHAR
INTEGER EXTERNAL, CHAR
/* Example 7 */ -- load records with multi-line fields -- doesn't work with inline data load data infile "test.dat" "str '|\n'" into test_table fields terminated by ';' TRAILING NULLCOLS ( desc, txt ) test.dat: one line;hello dear world;| two lines;Dear world, hello!;|
/* Example 8 */ -- loading binary files (word, images, video... etc) CREATE TABLE image_table ( image_id NUMBER(5), file_name VARCHAR2(30), image_data BLOB); Control File: LOAD DATA INFILE * INTO TABLE image_table REPLACE FIELDS TERMINATED BY ',' ( image_id INTEGER(5), file_name CHAR(30), image_data LOBFILE (file_name) TERMINATED BY EOF ) BEGINDATA 001,image1.gif 002,image2.jpg
Page 568
003,image3.jpg
/* Example 9 */ -- using specified characterset LOAD DATA CHARACTERSET WE8EBCDIC500 INFILE data.ebc "fix 86 buffers 1024" BADFILE data.bad' DISCARDFILE data.dsc' REPLACE INTO TABLE temp_data ( ... /* Example 10 */ -- Loading a Sequence Number LOAD DATA INFILE '/u01/app/oracle/oradata/load/testload.txt' INSERT INTO TABLE test123 (test_seq.nextval,. . .)
Loading Excel File into a Table using SQL*Loader

1. If any cell data has newline characters (Alt+ENTER), remove them using the following Excel script:
' Removing tabs and carriage returns from worksheet cells Sub CleanUp() Dim TheCell As Range On Error Resume Next For Each TheCell In ActiveSheet.UsedRange With TheCell If .HasFormula = False Then .Value = Application.WorksheetFunction.Clean(.Value) End If End With Next TheCell End Sub 2. 3. Save Excel file as CSV Use SQL*Loader to load from CSV.
Loading Large Fields into a Table

If you try to load any field larger than 255 bytes into a table, even if the table column is defined as VARCHAR2(2000) or a CLOB, "Field in datafile exceeds maximum length" error will be returned. You need to specify the size of the table column in the control file: LOAD DATA INFILE '/u01/app/oracle/oradata/load/testload.txt' INSERT INTO TABLE test123
Page 569
FIELDS TERMINATED BY ',' (text CHAR(2000))
Using Direct Load Options

DIRECT = true DATA_CACHE: (default 1000) used if you have duplicate data and timestamp values in your data. ROWS: save data every number of ROWS UNRECOVERABLE=Y (default N) used with direct=true and disables redo log generation SKIP_INDEX_MAINTENANCE=true (default false) do not bother maintaining the indexes during the load. COLUMNARRAYROWS: number of rows loaded before the building of the stream buffer STREAMSIZE: the size of the stream buffer MULTITHREADING: (default true) parallel conversion of column arrays to stream buffers and stream buffer
Page 570
Data Pump Export and Import

Data Pump Components
The DBMS_DATAPUMP package: this is the main engine of the Data Pump utilities. It contains procedures that do the export and import actions. The DBMS_METADATA package: this package is used to extract and modify data dictionary metadata. The command-line clients, expdp and impdp.
Data Pump Export Interfaces

Using the Command Line
expdp system/manager directory=dpump_dir1 dumpfile=expdat1.dmp
Using a Parameter File

expdp parfile=myfile.txt paramerer file may contain: SCHEMAS=HR DIRECTORY=dpump_dir1 DUMPFILE=system1.dmp SCHEMAS=hr ...
Using Interactive-command Data Pump Export
This mode is enabled by pressing [Ctrl] + [C] during an export operation started with the commandline interface or the parameter file interface.
Using EM Database Control

Start the Database Control and go to the Maintenance | Utilities page.
Export Modes Parameters
FULL (requires EXPORT_FULL_DATABASE role), SCHEMAS, TABLES, TABLESPACES, TRANSPORT_TABLESPACES, and TRANSPORT_FULL_CHECK
Required Rrivileges
# basic privileges: grant create session, create table, create procedure to datapump_user; # if you want to do any of the following: - to run a full database Export or - to run a transport_tablespace job or - to run an Export DataPump job with the TRACE parameter or - to run an operation that exports a different schema. grant exp_full_database, imp_full_database to datapump_user;
Invoking Export Data Pump Examples

# the command requires CREATE ANY DIRECTORY privileges
Page 571
CREATE DIRECTORY dpump_dir1 AS '/u01/mydir'; GRANT READ, WRITE ON DIRECTORY dpump_dir1 TO baraka;
# schema mode is the default expdp baraka/password DIRECTORY=dpump_dir1 dumpfile=testexp01.dmp LOGFILE=dpump_dir2:mylog.log # generate filenames based on date and time (sed used to get rid of spaces) expdp sa/s directory=dpdir dumpfile=sa`date +%d-%m-%y_%k-%M | sed 's/[[:space:]]//'`.dmp logfile=dpdir:sa`date +%d-%m-%y_%k-%M | sed 's/[[:space:]]//'`.log # tables mode expdp baraka/password tables=employees DIRECTORY=dpump_dir1 dumpfile=testexp01.dmp expdp system/password tables=hr.employees .. # nolog file (by default export.log is generated) expdp nologfile=y # overwrite existing dumpfiles (11g) expdp REUSE_DUMPFILES=y # compression: ALL, DATA_ONLY, METADATA_ONLY, NONE expdp COMPRESSION=NONE
Export Filtering Parameters

/* EXCLUDE & INCLUDE */ EXCLUDE=INDEX EXCLUDE=PROCEDURE EXCLUDE=TABLE:"LIKE 'EMP%'" EXCLUDE=SCHEMA:"='HR'" INCLUDE=TABLE:"IN ('EMP', 'DEPT')" # When used in command line, use slashes to escape single and double quotation: expdp .. schemas=SCOTT EXCLUDE=TABLE:\"=\'EMP\'\" /* QUERY */ QUERY=OE.ORDERS: "WHERE order_id > 100000 ORDER BY order_date desc"
Export Remapping Parameters

/* REMAP_DATA (11g) */ -- the remapping function shouldnt have commit or rollback REMAP_DATA=[schema.]tablename.column_name:[schema.]pkg.function expdp hr/passwd DIRECTORY=dp_dir DUMPFILE=remap.dmp TABLES=hr.employees REMAP_DATA=hr.employees.last_name:hr.remap_pckg.modifychar
Page 572
Sampling Export Data
When you sample a parent table, the child table may contain rows unreferenced by the parent. In such case, impdp will generate "ORA-02298: cannot validate (<foreign key constraint name>)parent keys not found"
SAMPLE=[[schema_name.]table_name:]sample_pct SAMPLE=10 SAMPLE="HR"."EMPLOYEES":50
Export Encryption Parameters
(11g): To secure the exported dump file, the following new parameters are presented in Oracle 11g Data pump: ENCRYPTION, ENCRYPTION_PASSWORD and ENCRYPTION_ALGORITHM. To enable encryption, you must specify either the ENCRYPTION or ENCRYPTION_PASSWORD parameter, or both.
ENCRYPTION = {all | data_only | encrypted_columns_only | metadata_only | none} ENCRYPTION_ALGORITHM = { AES128 | AES192 | AES256 } ENCRYPTION_MODE = { DUAL | PASSWORD | TRANSPARENT }
expdp hr DUMPFILE=dp_dir.hr_enc.dmp JOB_NAME=enc ENCRYPTION=data_only ENCRYPTION_PASSWORD=mypassword expdp hr DIRECTORY=dp_dir DUMPFILE=hr_enc.dmp ENCRYPTION=all ENCRYPTION_PASSWORD=mypassword ENCRYPTION_ALGORITHM=AES256 ENCRYPTION_MODE=dual
Export Estimating Parameters

ESTIMATE={BLOCKS | STATISTICS} ESTIMATE_ONLY=y
expdp system/pswd estimate_only=y
Export Network Link Parameter

You can initiate an export job from your server and have Data Pump export data from a remote database to dump files located on the instance from which you initiate the Data Pump export job. READ ONLY DB can still be loaded from.
expdp hr/hr DIRECTORY=dpump_dir1 NETWORK_LINK=source_database_link DUMPFILE=network_export.dmp -- more detailed steps: -- scenario: I will take a data pump export from database ORCL -- and dumpfile will be written to database TIGER sqlplus sa/a@tiger create database link orcl.net using 'ORCL'; OR Create database link orcl.net connect to sa identified by a using '(DESCRIPTION=(ADDRESS = (PROTOCOL = TCP)(HOST =10.4.x.x) (PORT=1521)) (connect_data=(service_name=orcl)))'; select * from [email protected]; $expdp arju/a@tiger directory=d schemas=arju dumpfile=arju_dump_from_orcl.dmp
Page 573
network_link=orcl.net
Export PARALELL Parameter
You should specify number of dump files equal to the PARALLEL value.
expdp system/manager full=y parallel=4 dumpfile= DIR1:full1%U.dat, DIR2:full2%U.dat, DIR3:full3%U.dat, DIR4:full4%U.dat filesize = 2G
Import Modes Parameters
You can perform a Data Pump import in various modes, using the TABLE, SCHEMAS, TABLESPACES, and FULL
File- and Directory-Related Parameters

The Data Pump import utility uses the PARFILE, DIRECTORY, DUMPFILE, LOGFILE, and NOLOGFILE commands in the same way as the Data Pump export utility. SQLFILE This parameter enables you to extract the DDL from the export dump file, without importing any data. impdp salapati/sammyy1 DIRECTORY=dpump_dir1 DUMPFILE=finance.dmp SQLFILE=dpump_dir2:finance.sql REUSE_DATAFILES This parameter tells Data Pump whether it should use existing datafiles for creating tablespaces during an import.
Using TABLE_EXISTS_ACTION Parameter

TABLE_EXISTS_ACTION=SKIP,APPEND,TRINCATE,REPLACE
Import Filtering Parameters
Same as export filtering parameters
Import Remapping Parameters

/* REMAP_TABLE (11g) */ -- rename table during export or import -- it wont work if table already exists impdp dumpfile=dp_dir:docs.dmp REMAP_TABLE=hr.docs:docs2 userid=hr/password impdp dumpfile=dp_dir:docs.dmp REMAP_TABLE=hr.docs.part1:docs3 userid=hr/password
Page 574
REMAP_SCHEMA impdp system/manager dumpfile=newdump.dmp REMAP_SCHEMA=hr:oe
REMAP_DATAFILE
Changes the name of the source datafile to the target datafile name in all SQL statements where the source datafile is referenced: CREATE TABLESPACE, CREATE LIBRARY, and CREATE DIRECTORY. Remapping datafiles is useful when you move databases between platforms that have different file naming conventions.
impdp hr/hr FULL=y DIRECTORY=dpump_dir1 DUMPFILE=db_full.dmp REMAP_DATAFILE='DB1$:[HRDATA.PAYROLL]tbs6.f':'/db1/hrdata/payroll/tbs6.f'
REMAP_TABLESPACE
This parameter enables you to move objects from one tablespace into a different tablespace during an import.
impdp hr/hr REMAP_TABLESPACE='example_tbs':'new_tbs' DIRECTORY=dpump_dir1 PARALLEL=2 JOB_NAME=cf1n02 DUMPFILE=employees.dmp NOLOGFILE=Y
Ignoring Nondeferred Constraints
(11g): setting the DATA_OPTIONS parameter to SKIP_CONSTRAINT_ERRORS will cause the import program to skip errors generated by the nondeferred database constraints. In the case of deferred constraints, imports will always be rolled back.
impdp Robert/robert DIRECTORY=data_pump_dir DUMPFILE=remap.dmp tables=ROBERT.NAMES data_options=SKIP_CONSTRAINT_ERRORS
Import Network Link Parameter

Get data from a remote database through db link. Read Only DB can still be loaded from. When the NETWORK_LINK parameter is used in conjunction with the TABLES parameter, only whole tables can be imported (not partitions of tables). If the USERID that is executing the import job has the IMP_FULL_DATABASE role on the target database, then that user must also have the EXP_FULL_DATABASE role on the source database. The only types of database links supported by Data Pump Import are: public, fixed-user, and connected-user. Current-user database links are not supported.
impdp hr/hr TABLES=employees DIRECTORY=dpump_dir1 NETWORK_LINK=source_database_link EXCLUDE=CONSTRAINT
Import Flashback Parameters
FLASHBACK_TIME and FLASHBACK_SCN enable you to import data consistent as of the flashback time you specify in your import job.
impdp system/manager FLASHBACK_TIME="TO_TIMESTAMP('01-06-2009 07:00:00;','DDMM-YYYY HH24:MI:SS')"
Page 575
Import PARALELL Parameter

You should specify number of dump files equal to the PARALLEL value. If you received the errors: ORA-39029, ORA-31671, ORA-39078, try increasing STREAMS_POOL_SIZE: alter system set STREAMS_POOL_SIZE=100M;
impdp system/manager directory = MYDIR parallel = 4 dumpfile = full1%U.dat,full2%U.dat, full3%U.dat,full4%U.dat
Monitoring a Data Pump Jobs

# data pump jobs SELECT * FROM dba_datapump_jobs; # data pump sessoins SELECT sid, serial# FROM v$session s, dba_datapump_sessions d WHERE s.saddr = d.saddr; # viewing job progress SELECT opname, target_desc, sofar, totalwork FROM v$session_longops;
Page 576
LogMiner
Types of Supplemental Logging
Unconditional supplemental log group: column values are always logged. ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (UNIQUE) COLUMNS; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (FOREIGN KEY) COLUMNS; Conditional supplemental log group: column values are logged if changed.
Levels of Supplemental Logging

Database Level: more resourse consuming, least required level is: ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS; ALTER DATABASE ADD SUPPLEMENTAL LOG DATA (UNIQUE) COLUMNS; Table Level: ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS; ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS; ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG DATA (UNIQUE) COLUMNS; Table-Level User-Defined Supplemental Log Groups o User-defined unconditional log groups ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG GROUP emp_parttime (EMPLOYEE_ID, LAST_NAME, DEPARTMENT_ID) ALWAYS; o User-defined conditional supplemental log groups -- always keyword isn't used: ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG GROUP emp_fulltime (EMPLOYEE_ID, LAST_NAME, DEPARTMENT_ID); ALTER TABLE HR.EMPLOYEES ADD SUPPLEMENTAL LOG GROUP emp_parttime( DEPARTMENT_ID NO LOG, EMPLOYEE_ID); Note: you must enable at least minimal supplemental logging prior to generating log files which will be analyzed by LogMiner.
Disabling Database-Level Supplemental Logging

ALTER DATABASE DROP SUPPLEMENTAL LOG DATA (PRIMARY KEY) COLUMNS; ALTER DATABASE DROP SUPPLEMENTAL LOG DATA (UNIQUE) COLUMNS; ALTER DATABASE DROP SUPPLEMENTAL LOG DATA;
Page 577
LogMiner Dictionary Options

Using the Online Catalog: use it, if you believe no DDL is performed during Analysis. Extracting a LogMiner Dictionary to the Redo Log Files: good if DDL might be there during analysis Extracting the LogMiner Dictionary to a Flat File: forget this option!
Redo Log File Options

Automatically: required redo log is automaitcally detected and used by the utility. Manually: you provide list of the redo files to analyse.
OPTIONS possible values in DBMS_LOGMNR.START_LOGMNR:

DICT_FROM_ONLINE_CATALOG : Using the Online Catalog DICT_FROM_REDO_LOGS : Start LogMiner CONTINUOUS_MINE : Redo Log File Options COMMITTED_DATA_ONLY : Showing Only Committed Transactions SKIP_CORRUPTION : Skipping Redo Corruptions NO_SQL_DELIMITER : Formatting Reconstructed SQL Statements for Reexecution PRINT_PRETTY_SQL : Formatting the Appearance of Returned Data for Readability NO_ROWID_IN_STMT : Formatting Reconstructed SQL Statements for Reexecution DDL_DICT_TRACKING : Tracking DDL Statements in the LogMiner Dictionary (not with DICT_FROM_ONLINE_CATALOG)
Obtaining LogMiner Operational Information

-- specified redo log files -- STATUS: 0: the log file will be processed 1: the file will be the first redo log file to be processed 2: the file will not be processed by LogMiner (pruned) 4: the file is missing select * from V$LOGMNR_LOGS; -- optional LogMiner parameters select * from V$LOGMNR_PARAMETERS; -- current settings for supplemental logging select SUPPLEMENTAL_LOG_DATA_FK, SUPPLEMENTAL_LOG_DATA_ALL, SUPPLEMENTAL_LOG_DATA_UI, SUPPLEMENTAL_LOG_DATA_MIN from V$DATABASE; select * from select * from select * from DBA_LOG_GROUPS; ALL_LOG_GROUPS; USER_LOG_GROUPS;
Page 578
select * from select * from select * from
DBA_LOG_GROUP_COLUMNS; ALL_LOG_GROUP_COLUMNS; USER_LOG_GROUP_COLUMNS;
Examples of Using LogMiner

Without Sepecifying the Redo Files Example: LogMiner Dictionary: Online Catalog, Redo Log File Options: Automatically -- check Supplemental Logging is enabled (see the examples above) -- start LogMiner ALTER SESSION SET NLS_DATE_FORMAT = 'DD-MON-YYYY HH24:MI:SS'; begin DBMS_LOGMNR.START_LOGMNR( STARTTIME => '01-Jan-2003 08:30:00', ENDTIME => '01-Jan-2003 08:45:00', -- future value can be set SYSDATE+5/24 OPTIONS => DBMS_LOGMNR.DICT_FROM_ONLINE_CATALOG + DBMS_LOGMNR.CONTINUOUS_MINE + DBMS_LOGMNR.COMMITTED_DATA_ONLY + DBMS_LOGMNR.PRINT_PRETTY_SQL ); end; /
-- query V$LOGMNR_CONTENTS SELECT USERNAME AS usr,(XIDUSN || '.' || XIDSLT || '.' || XIDSQN) as XID, SQL_REDO FROM V$LOGMNR_CONTENTS WHERE SEG_OWNER IS NULL OR SEG_OWNER NOT IN ('SYS', 'SYSTEM') AND TIMESTAMP > '10-jan-2003 15:59:53';
SELECT OPERATION, SQL_REDO, SQL_UNDO FROM V$LOGMNR_CONTENTS WHERE SEG_OWNER = 'OE' AND SEG_NAME = 'ORDERS' AND OPERATION = 'DELETE' AND USERNAME = 'RON'; -- Querying Based on Column Values SELECT SQL_REDO FROM V$LOGMNR_CONTENTS WHERE SEG_NAME = 'EMPLOYEES' AND SEG_OWNER = 'HR' AND OPERATION = 'UPDATE' AND DBMS_LOGMNR.MINE_VALUE(REDO_VALUE, 'HR.EMPLOYEES.SALARY') DBMS_LOGMNR.MINE_VALUE(UNDO_VALUE, 'HR.EMPLOYEES.SALARY');
-- end the Miner session EXECUTE DBMS_LOGMNR.END_LOGMNR;
Page 579
Wit Sepecifying the Redo Files Example: LogMiner Dictionary: Using the LogMiner Dictionary in the Redo Log Files, Redo Log File Options: list of files provided Redo log file that contains the end of the dictionary extract must have been created before the redo log file that you want to analyze, but should be as recent as possible. Assume: you want to analyze file no 210
-- check Supplemental Logging is enabled (see the examples above)
- TO extract the data dictionary to the redo logs (must be done before the redo to analyze) EXECUTE sys.DBMS_LOGMNR_D.build( OPTIONS => sys.DBMS_LOGMNR_D.store_in_redo_logs); - Find a redo log file that contains the end of the dictionary extract SELECT NAME, SEQUENCE#, DICTIONARY_BEGIN d_beg, DICTIONARY_END d_end FROM V$ARCHIVED_LOG WHERE SEQUENCE# = (SELECT MAX (SEQUENCE#) FROM V$ARCHIVED_LOG WHERE DICTIONARY_END = 'YES' and SEQUENCE# <= 210); - Find the redo log file that contains the start of the data dictionary extract that matches the end of the dictionary found in the previous step: SELECT NAME, SEQUENCE#, DICTIONARY_BEGIN d_beg, DICTIONARY_END d_end FROM V$ARCHIVED_LOG WHERE SEQUENCE# = (SELECT MAX (SEQUENCE#) FROM V$ARCHIVED_LOG WHERE DICTIONARY_BEGIN = 'YES' and SEQUENCE# <= 208); - Specify the list of the redo log files of interest. Order doesn't matter: EXECUTE DBMS_LOGMNR.ADD_LOGFILE(LOGFILENAME => '/usr/oracle/data/db1arch_1_210_482701534.dbf', OPTIONS => DBMS_LOGMNR.NEW); EXECUTE DBMS_LOGMNR.ADD_LOGFILE(LOGFILENAME => '/usr/oracle/data/db1arch_1_208_482701534.dbf'); EXECUTE DBMS_LOGMNR.ADD_LOGFILE(LOGFILENAME => '/usr/oracle/data/db1arch_1_207_482701534.dbf'); - Query the V$LOGMNR_LOGS : SELECT FILENAME AS name, LOW_TIME, HIGH_TIME FROM V$LOGMNR_LOGS;
- Start LogMiner: EXECUTE DBMS_LOGMNR.START_LOGMNR(OPTIONS => DBMS_LOGMNR.DICT_FROM_REDO_LOGS + DBMS_LOGMNR.COMMITTED_DATA_ONLY + DBMS_LOGMNR.PRINT_PRETTY_SQL); - Query the V$LOGMNR_CONTENTS: SELECT USERNAME AS usr, SQL_REDO FROM V$LOGMNR_CONTENTS WHERE SEG_OWNER IS NULL OR SEG_OWNER NOT IN ('SYS', 'SYSTEM') AND TIMESTAMP > '10-jan-2003 15:59:53'; - to display all the DML statements that were executed as part of the CREATE Oracle DBA Code Examples
Page 580
TABLE DDL statement: SELECT SQL_REDO FROM V$LOGMNR_CONTENTS WHERE XIDUSN = 1 and XIDSLT = 2 and XIDSQN = 1594; -- end the Miner session EXECUTE DBMS_LOGMNR.END_LOGMNR;
Page 581
Part 11
Miscellaneous Oracle Database Topics
Page 582
Managing Oracle Database Control

Configuring and Using the Database Control

Note ID 1099271.1 Master Note for Enterprise Manager Configuration Assistant (EMCA) in Single Instance Database Environment is a good reference. In one situation, I faced the error "System error 1060 has occurred." when I issued the command emctl start on Oracle 10g for Windows Server. To resolve it: Issue the command and get from its output the Agent process ID. Usin the Process manager, kill the corresponding emAgent.exe process. Then issue emctl start command. If error presists, try killing all the emAgent.exe processes.
-- ORACLE_SID should be set export ORACLE_SID=mydb -- manually configuring db control emca -config dbcontrol db emctl status dbconsole emctl start dbconsole emctl stop dbconsole http://localhost:5500/em $ORACLE_HOME/install/portlist.ini isqlplusctl start
Implementing EM Database Control Auto Startup

# (1) as root create oraemctl file su vi /etc/init.d/oraemctl #!/bin/bash # # oraemctl Starting and stopping Oracle Enterprise Manager Database Control. # Script is valid for 10g and 11g versions. # # chkconfig: 35 80 30 # description: Enterprise Manager DB Control startup script # Source function library. . /etc/rc.d/init.d/functions ORACLE_OWNER="oracle" ORA_HOME="/u01/app/oracle/product/10.2.0/db_1" case "$1" in start) echo -n $"Starting Oracle EM DB Console:" su - $ORACLE_OWNER -c "$ORA_HOME/bin/emctl start dbconsole" echo "OK"
Page 583
;; stop) echo -n $"Stopping Oracle EM DB Console:" su - $ORACLE_OWNER -c "$ORA_HOME/bin/emctl stop dbconsole" echo "OK" ;; *) echo $"Usage: $0 {start|stop}" esac
# (2) change permissions and runlevels chmod 750 /etc/init.d/oraemctl chkconfig --add oraemctl --level 0356
Dropping and Recreating the Management Repository

-- Oracle 11g R2: cd /u01/app/oracle/product/11.2.0/db_1/sysman/admin/emdrep/bin ./RepManager rac1 1521 ron_1 -sys_password qwer5 -action drop -repos_user sysman ./RepManager rac1 1521 ron_1 -sys_password qwer5 -action create -repos_user sysman
Page 584
Installing Oracle 10g R5 (10.2) Enterprise Manager Grid Control for Linux x86
Following are the basic installation steps to install Oracle 10.2.0.5 Enterprise Manager Grid Control on Linux on a new database. Note: for any installation, you should check the Release Notes documenation before taking any practical step. Note: Since Enterprise Manager 10g Grid Control Release 4 or higher are patch sets, you need to use the 'Installing Software-Only and Configuring Later' installation method as indicated below. This method is not supported by the interactive OUI.
Emulation software: VMWare Workstation 7 for Windows OS: Redhat Enterprise Linux 5.2 for x86 32-bit
Required Software
Oracle Enterprise Manager Grid Control 10.2.0.1 for Linux x86 32-bit Oracle Enterprise Manager Grid Control 10.2.0.5 for Linux x86 32-bit The interim RDBMS patch# 4329444 Oracle Database Patch Set Notes 10g Release 1 (10.1.0.5) Patch Set for Linux x86
Used Hardware
In the VMWare: create one virtual machine with the following specs: o 2 GB RAM o One ethernet card configured as bridged or host-only in VMware o CPU Count: 2 o Disk1: 40 GB on SCSI 0:0 used to install the OS and software
Installation Steps
1. Hardware ans Software Requirements Note: A reference to the requirements is Note ID 419646.1. # At least 2 GB of physical memory grep MemTotal /proc/meminfo # OS and Kernel version Oracle Enterprise Manager Grid Control Certification Checker [ID 412431.1] For OEM 10.2.0.1 in Linux 5.2, only software will be installed (no config) # swap space: if RAM=2G, swap=4GB
Page 585
grep SwapTotal /proc/meminfo # to display swap and memory in one command: free # if you don't have enought swap, # you can add swap space by creating a temporary swap file. # let's say about 500MB: dd if=/dev/zero of=tempswap bs=1k count=500000 chmod 600 tempswap mke2fs tempswap mkswap tempswap swapon tempswap
# required packs rpm -q --qf '%{NAME}-%{VERSION}-%{RELEASE} (%{ARCH})\n' binutils \ glibc- \ make- \ gcc- \ libaio- \ glibc-common- \ setarch- \ pdksh- \ openmotif22 \ sysstat- \ libstdc++- \ libstdc++-devel- \ compat-libstdc++- \ compat-db- \ control-center- \
# install missed packages rpm -Uvh libXp-1.0.0-8.1.el5.i386.rpm rpm -Uvh openmotif22-2.2.3-18.i386.rpm rpm -Uvh compat-db-4.2.52-5.1.i386.rpm rpm -Uvh compat-gcc-34-3.4.6-4.i386.rpm rpm -Uvh compat-gcc-34-c++-3.4.6-4.i386.rpm Note: for pdksh-5.2.14-36.el5.i386.rpm, it isn't used in EL 5.2
2. Configure kernel parameters and shell limits vi /etc/sysctl.conf # mark existing ones kernel.shmall = 2097152 kernel.shmmax = 536870912 kernel.shmmni = 4096 # semaphores: semmsl, semmns, semopm, semmni kernel.sem = 250 32000 100 128 # the following setting is removed because default value is higher # fs.file-max = 65536 # old values (1024 65000) not recommended net.ipv4.ip_local_port_range = 9000 65500 net.core.rmem_default = 4194304 net.core.rmem_max = 4194304
Page 586
net.core.wmem_default = 262144 net.core.wmem_max = 262144 net.ipv4.tcp_wmem = 262144 262144 262144 net.ipv4.tcp_rmem = 4194304 4194304 4194304 # then run: /sbin/sysctl -p
vi /etc/security/limits.conf * soft nproc 2047 * hard nproc 16384 * soft nofile 1024 * hard nofile 65536
vi /etc/pam.d/login session required
pam_limits.so
# in Linux 5.X run: ln -s /usr/lib/libgdbm.so.2.0.0 /usr/lib/libdb.so.2 # SELINUX must be disabled cat /etc/selinux/config | grep SELINUX= vi /etc/selinux/config SELINUX=disabled shutdown -h now -r 3. Create the required network configuration: # /etc/hosts file must contain a fully qualified name: # <IP-address> <fully-qualified-machine-name> <machine-name> vi /etc/hosts 127.0.0.1 localhost.localdomain localhost 192.168.4.107 srv107.localdomain srv107 ping srv07 4. Create and configure the required OS users, groups and directories groupadd -g 501 oinstall groupadd -g 502 dba # oracle software owner user /usr/sbin/useradd -u 200 -g oinstall -G dba oracle passwd oracle # oracle parent direcotry # Oracle homes will be created as subdirectories under this parent directory mkdir -p /u01/app mkdir /u01/stage chown -R oracle:oinstall /u01 chmod 775 /u01/app chmod 775 /u01/stage # will be used instead of /tmp by oracle mkdir /home/oracle/oratemp chown -R oracle:oinstall /home/oracle/oratemp
Page 587
vi /home/oracle/.bash_profile TMP=/home/oracle/oratemp; export TMP TMPDIR=$TMP; export TMPDIR ORACLE_BASE=/u01/app/oracle; export ORACLE_BASE ORACLE_HOME=/u01/app/oracle/product/10.2.0/db10g; export ORACLE_HOME AGENT_HOME=/u01/app/oracle/product/10.2.0/agent10g; export AGENT_HOME OMS_HOME=/u01/app/oracle/product/10.2.0/oms10g; export OMS_HOME ORACLE_HOSTNAME=srv107.localdomain; export ORACLE_HOSTNAME ORACLE_SID=oemdb; export ORACLE_SID ORACLE_TERM=xterm; export ORACLE_TERM PATH=/usr/sbin:$PATH; export PATH PATH=$ORACLE_HOME/bin:$PATH; export PATH LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib; export LD_LIBRARY_PATH CLASSPATH=$ORACLE_HOME/JRE:$ORACLE_HOME/jlib:$ORACLE_HOME/rdbms/jlib; export CLASSPATH
vi /etc/profile if [ $USER = "oracle" ]; then if [ $SHELL = "/bin/ksh" ]; then ulimit -p 16384 ulimit -n 65536 else ulimit -u 16384 -n 65536 fi fi 5. Install the Software # edit the response file vi /u01/stage/gc10.0.2.1/Disk1/response/em_with_new_db.rsp FROM_LOCATION="../rdbms/Disk1/stage/products.xml" BASEDIR="/u01/app/oracle/product/10.2.0" INSTALLATION_NAME="OEMGC 10.2.0.5" s_gdbName="oemdb" s_mountPoint="/u01/app/oracle/product/10.2.0/oradata" s_operGroup="dba" s_adminGroup="dba" s_securePassword="qwer5" s_securePasswordConfirm="qwer5" b_lockedSelected=false b_passwordsDifferent=false b_passwordsSame=true s_reposPwd="qwer5" s_reposPwdConfirm="qwer5" UNIX_GROUP_NAME="dba"
# invoke the installer: # Note: if the folder exists and not empty, add -force switch cd /u01/stage/gc10.0.2.1/Disk1 ./runInstaller -noconfig -silent -responseFile /u01/stage/gc10.0.2.1/Disk1/response/em_with_new_db.rsp | tee output.txt
# as a root user in a new terminal:

Page 588
# (if this is the first Oracle product you just installed on the host): /u01/app/oraInventory/orainstRoot.sh # execute (answer n to y/n questions): /u01/app/oracle/product/10.2.0/db10g/allroot.sh
# stop all the OPMN processes: /u01/app/oracle/product/10.2.0/oms10g/opmn/bin/opmnctl stopall # verify the processes were stopped: ps -ef | grep opmn 6. Apply the patches on the repository database home # Apply the interim RDBMS patch# 4329444 to the Oracle home of the database export ORACLE_HOME=/u01/app/oracle/product/10.2.0/db10g cd /u01/stage/4329444 export OBJECT_MODE=32_64 export PATH=$PATH:/u01/app/oracle/product/10.2.0/db10g/OPatch opatch apply
# Apply 10.1.0.5 Patch Set cd /u01/stage/db10.1.0.5/Disk1 ./runInstaller -ignoreSysPrereqs >select db10g home >next, next till you finish applying the patch set 7. Apply the patch set 10.2.0.5 on OMS and Agent homes # as root: run the following command: ln -s /usr/lib/libgdbm.so.2.0.0 /usr/lib/libdb.so.2 /* Apply the patch on oms */ # edit the response file: vi /u01/stage/gc10.0.2.5/3731593/Disk1/response/patchset.rsp UNIX_GROUP_NAME="dba" b_softwareonly=true s_sysPassword="qwer5" sl_pwdInfo={ "qwer5" } ORACLE_HOME="/u01/app/oracle/product/10.2.0/oms10g" oracle.iappserver.st_midtier:szl_InstanceInformation={"ias_password"} SHOW_SPLASH_SCREEN=false SHOW_SUMMARY_PAGE=false SHOW_INSTALL_PROGRESS_PAGE=false SHOW_REQUIRED_CONFIG_TOOL_PAGE=false SHOW_CONFIG_TOOL_PAGE=false SHOW_XML_PREREQ_PAGE=false HOW_END_OF_INSTALL_MSGS=false SHOW_ROOTSH_CONFIRMATION=false SHOW_END_SESSION_PAGE=false # apply the patch set cd /u01/stage/gc10.0.2.5/3731593/Disk1 ./runInstaller -noconfig -silent -responseFile /u01/stage/gc10.0.2.5/3731593/Disk1/response/patchset.rsp -force | tee outputoms.txt
Page 589
# open terminal as root and run (no output expected): /u01/app/oracle/product/10.2.0/oms10g/root.sh
/* Apply the patch on agent */ # change only the following parameters in the file: vi /u01/stage/gc10.0.2.5/3731593/Disk1/response/patchset.rsp ORACLE_HOME="/u01/app/oracle/product/10.2.0/agent10g" # apply the patch on agent: cd /u01/stage/gc10.0.2.5/3731593/Disk1 ./runInstaller -noconfig -silent -responseFile /u01/stage/gc10.0.2.5/3731593/Disk1/response/patchset.rsp -force | tee outputagent.txt # as root run: /u01/app/oracle/product/10.2.0/agent10g/root.sh 8. Configure the Enterprise Manager Grid Control export PERL5LIB=/u01/app/oracle/product/10.2.0/oms10g/perl/lib/5.6.1 /u01/app/oracle/product/10.2.0/oms10g/perl/bin/perl /u01/app/oracle/product/10.2.0/oms10g/sysman/install/ConfigureGC.pl /u01/app/oracle/product/10.2.0 # to verify the configured OEM release: /u01/app/oracle/product/10.2.0/oms10g/bin/emctl status oms /u01/app/oracle/product/10.2.0/agent10g/bin/emctl status agent 9. Running Enterprise Manager Grid Control 10.2.0.5 #find the EM port: cat /u01/app/oracle/product/10.2.0/oms10g/install/portlist.ini | grep "Enterprise Manager Central Console Port"
# access the URL: http://srv107.localdomain:4889/em/
# login as sysman 10. Deploy Management Agent on Target Machines and Add them to the Grid Control To add a target host to OEM, you need to install Management Agent on the target machine. The steps below uses agentDownload Script method. /* Install the Agent Software on the Target machine */ # Documenation of the script can be viewed: http://<OMS_host>:<OMS_port>/agent_download/agent_install_readme.html http://srv107.localdomain:4889/agent_download/agent_install_readme.html OR <OMS_HOME>/sysman/agent_download/agent_install_readme.html # The agentDownload script is located at OMS_HOME/sysman/agent_download/<version>/<platform>
Page 590
# Download the Management Agent software # from OTN OR the "Download Agent Software" feature in the Grid Control # console Linux_Grid_Control_agent_download_10_2_0_5_0.zip
# In OEM server, copy the Agent downloaded file to: <OMS_HOME>/sysman/agent_download/10.2.0.5.0
# unzip the file: unzip Linux_Grid_Control_agent_download_10_2_0_5_0.zip
# Download the agentDownload script to the target host from the Management Service URL http://srv107.localdomain:4889/agent_download/10.2.0.2.0/<platform>/agentDownl oad.<OS> http://srv107.localdomain:4889/agent_download/10.2.0.5.0/linux/agentDownload.l inux OR scp [email protected]:/u01/app/oracle/product/10.2.0/oms10g/sysman/agent_downlo ad/10.2.0.5.0/linux/agentDownload.linux agentDownload.linux chmod 744 agentDownload.linux
# ensure the wget is there and included in the PATH ls /usr/local/bin/wget echo $PATH | grep /usr/local/bin
# ensure the hostnames of the OEM server and target are accessbile from each # other ping srv107 ping oradb1
# export the required env variable export ORACLE_HOSTNAME=oradb1 # RUN THE SCRIPT # agent10g will be created as a subdirectory ./agentDownload.linux -b /u01/app/oracle/product/10.2.0 -m srv107 -r 4889
# run the script su root /u01/app/oracle/product/10.2.0/agent10g/root.sh exit
# If the agent you are installing is not secure, # you must execute the following command after the installation is complete: <Agent_Home>/bin/emctl secure agent <password>
Page 591
# status of the agent can be verified by: cd /u01/app/oracle/product/10.2.0/agent10g/bin # Note: in the output of the following command, the Agent URL should not be # https://localhost:1830/emd/main/. If it is so, OEM won't find the target. # You need to re-install the Agent and make sure you follow all the steps # above. ./emctl status agent # also you can check: https://oradb1:3872/emd/main/ /* Add the Target to OEM */ # login to the OEM and you'll see the target already added: http://srv107.localdomain:4889/em/ # click on Congiure button to configure the database connection info.
Page 592
Remote Diagnostic Agent (RDA)

Using Remote Diagnostic Agent (RDA)
Refer to note 314422.1
Page 593
Connect Oracle to SQL Server

Configuring and Using the Database Control
To connect to SQL*Server from Oracle using Hetergeneous Services: 1. Install ODBC drivers for the SQL*Server database on the server that contains the Oracle code tree. 2. Setup the ODBC connection to the SQL Server 3. Test the ODBC conection 4. Ensure the GLOBAL_NAMES parameter is set to FALSE in the Oracle database. show parameter GLOBAL_NAMES 5. Configure the Hetergeneous services. This consists of creating an initodbc.ora file in <ORACLE_HOME>\ hs\admin For example: You may create a file named as initclveodbc.ora containing the following: # # HS init parameters # HS_FDS_CONNECT_INFO = clve HS_FDS_TRACE_LEVEL = off # # Environment variables required for the non-Oracle system # #set <envvar>=<value> 7. Modify the Listener.ora file : SID_LIST_LISTENER = (SID_LIST = (SID_DESC = (SID_NAME = PLSExtProc) (ORACLE_HOME = D:\oracle\product\10.1.0\db_1) (PROGRAM = extproc) ) (SID_DESC = (SID_NAME = CLVEODBC) (ORACLE_HOME = D:\oracle\product\10.1.0\db_1) (PROGRAM = HSODBC) ) (SID_DESC = (SID_NAME = HKCLVEODBC) (ORACLE_HOME = D:\oracle\product\10.1.0\db_1) (PROGRAM = HSODBC) ) ) 8. Modify the TNSNAMES.ORA file to point to the proper code tree. CLVE.MYDOMAIN.LOCAL = (DESCRIPTION = Oracle DBA Code Examples
Page 594
(ADDRESS_LIST = (ADDRESS = (PROTOCOL = TCP)(HOST = au-syd-dw01)(PORT = 15220)) ) (CONNECT_DATA = (SERVICE_NAME = CLVEODBC) ) (HS = OK) ) 9. Reload the listener lsnrctl reload 10. Create a database link on the Oracle installation. CREATE DATABASE LINK "CLVE.MYDOMAIN.LOCAL" CONNECT TO "sqlusername" IDENTIFIED BY "sqluserpassword" USING 'CLVE.MYDOMAIN.LOCAL'; 11. Run a Select statement for the Oracle installation using the database link select sysdate from [email protected] ;
Page 595
Part 12
PL/SQL Samples
Page 596
PL/SQL Basics
PL/SQL Data Types
/* character */ -- max length in 11g: 32767 DECLARE v_string VARCHAR2(10); DECLARE v_string VARCHAR2(10 CHAR); ...
/* number */ BINARY_DOUBLE BINARY_FLOAT BINARY_INTEGER NUMBER PLS_INTEGER
2147483647 to +2147483647
/* data time */ -- TIMESTAMP select SYSTIMESTAMP from dual; -- TIMESTAMP WITH TIME ZONE SET SERVEROUTPUT ON DECLARE v_datetime TIMESTAMP (3) WITH TIME ZONE := SYSTIMESTAMP; BEGIN DBMS_OUTPUT.PUT_LINE(v_datetime); END; / -- TIMESTAMP WITH LOCAL TIME ZONE SET SERVEROUTPUT ON DECLARE v_datetime TIMESTAMP (0) WITH LOCAL TIME ZONE := SYSTIMESTAMP; BEGIN DBMS_OUTPUT.PUT_LINE(v_datetime); END; /
/* inerval */ INTERVAL YEAR TO MONTH INTERVAL DAY TO SECOND DECLARE v_college_deadline TIMESTAMP; BEGIN
Page 597
v_college_deadline := TO_TIMESTAMP('06/06/2004', 'DD/MM/YYYY') + INTERVAL '12-3' YEAR TO MONTH + INTERVAL '19 9:0:0.0' DAY TO SECOND; DBMS_OUTPUT.PUT_LINE('My daughter leaves for college in ' || v_college_deadline); END; /
/* REF CURSOR */ -- a pointer to a result set CREATE OR REPLACE PROCEDURE cust_sel ( cv_results IN OUT SYS_REFCURSOR) IS BEGIN OPEN cv_results FOR SELECT customer_id, cust_last_name from customers; END; / VARIABLE x REFCURSOR EXEC cust_sel(:x) PRINT x /* REF */ REF value simply as a pointer to an object instance in an object table or object view.
Controlling Compile-Time Displayed Warnings

PLSQL_WARNINGS is a system and session levels. ={ ENABLE | DISABLE | ERROR }: { ALL | SEVERE | INFORMATIONAL | PERFORMANCE | { integer | (integer [, integer ] ...) }} SHOW PARAMETER PLSQL_WARNINGS -- it can take one value PLSQL_WARNINGS='ENABLE:PERFORMANCE' -- it can take multiple values PLSQL_WARNINGS='ENABLE:PERFORMANCE', 'ENABLE:SEVERE' -- more examples: PLSQL_WARNINGS = 'ENABLE:SEVERE', 'DISABLE:INFORMATIONAL'; PLSQL_WARNINGS = 'DISABLE:ALL'; PLSQL_WARNINGS = 'DISABLE:5000', 'ENABLE:5001', 'ERROR:5002'; PLSQL_WARNINGS = 'ENABLE:(5000,5001,5002)', 'DISABLE:(6000,6001)'; -- setting the parameter using DBMS_WARNING SELECT DBMS_WARNING.GET_WARNING_SETTING_STRING() WARNING_LEVEL FROM dual; CALL DBMS_WARNING.SET_WARNING_SETTING_STRING('ENABLE:ALL', 'SESSION'); CALL DBMS_WARNING.SET_WARNING_SETTING_STRING('DISABLE:ALL', 'SESSION');
Page 598
Catching Returned Errors

.. EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE('Exception: '||sqlerrm); ..
Hiding Code
wrap iname=input_file.sql oname=output_file.plb
Controlling Program Flow

/* IF */ IF condition THEN action ELSIF condition THEN action ELSE action END IF; /* CASE */ CASE expression WHEN test1 THEN action; WHEN test2 THEN action; ... ELSE ... END CASE; CASE v_category WHEN 'Oracle Basics' THEN v_discount := .15; WHEN 'Oracle Server' THEN v_discount := .10; ELSE v_discount := .5; END CASE;
/* Searched CASE */ CASE WHEN <exp> THEN actions; WHEN <exp> THEN actions; ELSE actions; END CASE;
/* Case Expressions */
Page 599
appraisal := CASE grade WHEN 'A' WHEN 'B' WHEN 'C' ELSE 'No END;
THEN THEN THEN such
'Excellent' 'Very Good' 'Good' grade'
appraisal := CASE WHEN grade = 'A' THEN 'Excellent' WHEN grade IN ('B','C') THEN 'Good' ELSE 'No such grade' END;
/* Loop */ LOOP .. EXIT [WHEN condition] END LOOP; -- loops and lablels BEGIN <<Outer_loop>> LOOP counter := counter+1; EXIT WHEN counter>10; <<Inner_loop>> LOOP ... EXIT Outer_loop WHEN total_done = 'YES'; -- Leave both loops EXIT WHEN inner_done = 'YES'; -- Leave inner loop only ... END LOOP Inner_loop; ... END LOOP Outer_loop; END; /
/* Numeric FOR Loop */ FOR counter IN [reverse] low_number .. high_number LOOP action; END LOOP; begin for i in reverse 1..10 loop dbms_output.put_line(i); end loop; end; /
Page 600
/* WHILE Loop */ WHILE condition LOOP .. END LOOP;
/* Goto and Labels */ BEGIN DBMS_OUTPUT.PUT_LINE('BEGINNING OF BLOCK'); GOTO l_Last_Line; DBMS_OUTPUT.PUT_LINE('GOTO didn''t work!'); RETURN; <<l_Last_Line>> DBMS_OUTPUT.PUT_LINE('Last Line'); END; /
Page 601
Using Cursors
CURSOR cursor_name [parameter_list] [RETURN return_type] IS query [FOR UPDATE [OF (column_list)][NOWAIT]]; OPEN cur1; OPEN cur1(50); FETCH cursor_name INTO variable_name(s) | PL/SQL_record; CLOSE cursor_name; Cursor Attributes: %BULK_ROWCOUNT number of rows changed during the operation %ROWCOUNT number of rows fetched from the cursor at any given time %FOUND %ISOPEN %NOTFOUND -- example 1 SET SERVEROUTPUT ON DECLARE v_first_name AUTHORS.FIRST_NAME%TYPE; v_last_name AUTHORS.LAST_NAME%TYPE; v_row_count PLS_INTEGER := 0; v_book_count PLS_INTEGER := 0; CURSOR auth_cur IS SELECT a.first_name, a.last_name, count(b.title) FROM authors a, books b WHERE a.id = b.author1 OR a.id = b.author2 OR a.id = b.author3 GROUP BY a.first_name, a.last_name HAVING count(b.title) > 0 ORDER BY a.last_name; BEGIN DBMS_OUTPUT.ENABLE(1000000); OPEN auth_cur; LOOP FETCH auth_cur INTO v_first_name, v_last_name, v_book_count; EXIT WHEN auth_cur%NOTFOUND; v_row_count := auth_cur%ROWCOUNT; DBMS_OUTPUT.PUT_LINE(v_row_count||' rows processed so far'); DBMS_OUTPUT.PUT_LINE(v_last_name ||', ' ||v_first_name ||' wrote ' ||v_book_count ||' book(s).'); END LOOP; CLOSE auth_cur; IF auth_cur%ISOPEN = FALSE THEN DBMS_OUTPUT.PUT_LINE('Cursor closed'); ELSE DBMS_OUTPUT.PUT_LINE('The cursor is still open'); END IF; EXCEPTION
Page 602
WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE(SQLERRM); END; /
-- example 2 .. WHILE auth_cur%FOUND LOOP DBMS_OUTPUT.PUT_LINE(v_author.last_name); FETCH auth_cur INTO v_author; END LOOP; ..
-- example 3 SET SERVEROUTPUT ON DECLARE CURSOR auth_cur IS SELECT * FROM authors; BEGIN FOR v_author IN auth_cur LOOP DBMS_OUTPUT.PUT_LINE(v_author.last_name); END LOOP; END; /
/* Cursor Variables */ SET SERVEROUTPUT ON DECLARE TYPE book_typ IS REF CURSOR RETURN BOOKS%ROWTYPE; cv_books book_typ; v_books BOOKS%ROWTYPE; BEGIN DBMS_OUTPUT.ENABLE(1000000); OPEN cv_books FOR SELECT * FROM books WHERE isbn = '78824389'; FETCH cv_books INTO v_books; DBMS_OUTPUT.PUT_LINE(v_books.title||' is '||v_books.price); CLOSE cv_books; END; /
CREATE OR REPLACE PROCEDURE authors_sel ( cv_results IN OUT SYS_REFCURSOR) IS BEGIN OPEN cv_results FOR SELECT id, first_name, last_name FROM authors; END; / VARIABLE x REFCURSOR EXEC authors_sel(:x) PRINT x
Page 603
/* Handling Implicit Cursors */ SET SERVEROUTPUT ON BEGIN DBMS_OUTPUT.ENABLE(1000000); UPDATE books SET price = price * .90 WHERE isbn = '78824389'; DBMS_OUTPUT.PUT_LINE(SQL%ROWCOUNT||' rows updated'); IF SQL%NOTFOUND THEN DBMS_OUTPUT.PUT_LINE('Unable to update isbn 78824389'); END IF; COMMIT; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE(SQLERRM); END; /
/* Cursor Subqueries */ SET SERVEROUTPUT ON DECLARE cv_author SYS_REFCURSOR; v_title BOOKS.TITLE%TYPE; v_author AUTHORS%ROWTYPE; v_counter PLS_INTEGER := 0; CURSOR book_cur IS SELECT b.title, CURSOR (SELECT * FROM authors a WHERE a.id = b.author1 OR a.id = b.author2 OR a.id = b.author3) FROM books b WHERE isbn = '78824389'; BEGIN OPEN book_cur; LOOP FETCH book_cur INTO v_title, cv_author; EXIT WHEN book_cur%NOTFOUND; DBMS_OUTPUT.PUT_LINE('Title from the main cursor: '||v_title); LOOP FETCH cv_author INTO v_author; EXIT WHEN cv_author%NOTFOUND; v_counter := v_counter + 1; DBMS_OUTPUT.PUT_LINE('Author'||v_counter||': ' ||v_author.first_name||' '||v_author.last_name); END LOOP; END LOOP; CLOSE book_cur; END; /
Page 604
/* Using Where Current of */ SET SERVEROUTPUT ON DECLARE v_isbn INVENTORY.ISBN%TYPE; v_amount INVENTORY.AMOUNT%TYPE; CURSOR inventory_cur IS SELECT isbn, amount FROM inventory WHERE status = IN STOCK AND isbn IN (SELECT isbn FROM books WHERE price > 40) FOR UPDATE OF amount; BEGIN FOR y IN inventory_cur LOOP FETCH inventory_cur INTO v_isbn, v_amount; EXIT WHEN inventory_cur%NOTFOUND; DBMS_OUTPUT.PUT_LINE(v_isbn||Amount IN STOCK before: ||v_amount); v_amount := v_amount + 250; UPDATE inventory SET amount = v_amount WHERE CURRENT OF inventory_cur; DBMS_OUTPUT.PUT_LINE(v_isbn||Amount IN STOCK after: ||v_amount); END LOOP; COMMIT; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE(SQLERRM); ROLLBACK; END; /
Page 605
Using Records
DECLARE TYPE individual_record IS RECORD (individual_id INTEGER ,first_name VARCHAR2(30 CHAR) ,middle_initial individuals.middle_initial%TYPE ,last_name VARCHAR2(30 CHAR)); -- Define a variable of the record type. individual INDIVIDUAL_RECORD; BEGIN -- Initialize the field values for the record. individual.individual_id := 2; individual.first_name := 'John'; individual.middle_initial := 'P'; individual.last_name := 'Morgan'; -- Insert into the table. INSERT INTO individuals VALUES (individual.individual_id ,individual.first_name ,individual.middle_initial ,individual.last_name); -- Commit the work. COMMIT; END; /
-- nested records DECLARE -- Define a record type. TYPE individual_record IS RECORD (individual_id INTEGER ,first_name VARCHAR2(30 CHAR) ,middle_initial VARCHAR2(1 CHAR) ,last_name VARCHAR2(30 CHAR)); -- Define a record type. TYPE address_record IS RECORD (address_id INTEGER ,individual_id INTEGER ,street_address1 VARCHAR2(30 CHAR) ,street_address2 VARCHAR2(30 CHAR)
Page 606
,street_address3 VARCHAR2(30 CHAR)); -- Define a record type of two user-defined record types. TYPE individual_address_record IS RECORD (individual INDIVIDUAL_RECORD ,address ADDRESS_RECORD); -- Define a user-defined compound record type. individual_address INDIVIDUAL_ADDRESS_RECORD; BEGIN -- Initialize the field values for the record. individual_address.individual.individual_id := 3; individual_address.individual.first_name := 'Ulysses'; ...
-- Record Types can be explicitly defined as Object Type as well
-- Defining and Using Record Types as Return Values DECLARE -- Define a record type. FUNCTION get_row (individual_id_in INTEGER) RETURN INDIVIDUAL_RECORD IS -- Define a cursor to return a row of individuals. CURSOR c (individual_id_cursor INTEGER) IS SELECT * FROM individuals WHERE individual_id = individual_id_cursor; BEGIN -- Loop through the cursor for a single row. FOR i IN c(individual_id_in) LOOP -- Return a %ROWTYPE from the INDIVIDUALS table. RETURN i; END LOOP; END get_row;
Page 607
Using Table Functions

/* Table Function (not pipelined) */ -- not pipelined: row set will be returned when all rows are fetched /* */ CREATE TYPE BookType AS OBJECT ( isbn CHAR(10), title VARCHAR2(100) ) / CREATE TYPE BookTypes AS TABLE OF BookType; / CREATE OR REPLACE FUNCTION SomeBooks(p_Category IN books.category%TYPE) RETURN BookTypes AS v_ResultSet BookTypes := BookTypes(); CURSOR c_SomeBooks IS SELECT isbn, title FROM books WHERE category = p_Category; BEGIN FOR v_Rec IN c_SomeBooks LOOP v_ResultSet.EXTEND; v_ResultSet(v_ResultSet.LAST) := BookType(v_Rec.isbn, v_Rec.title); END LOOP; RETURN v_ResultSet; END SomeBooks; /
SELECT SomeBooks('Oracle Basics') FROM dual; COLUMN title FORMAT a60 SELECT * FROM TABLE (SomeBooks('Oracle Basics'));
-- Since SomeBooks will always return the same output given the same -- input, we can use the DETERMINISTIC keyword. -- its supposed positive impact is not proved though CREATE OR REPLACE FUNCTION SomeBooks(p_Category IN books.category%TYPE) RETURN BookTypes DETERMINISTIC AS
/* Example of using pipelined table functions in data transformation */ -- with pipelined table functions, each row will be returned as soon --- as it is created -- sample destination table CREATE TABLE yearly_store_sales (store_name VARCHAR2(25), sales_year NUMBER, total_sales NUMBER); -- (1) create type CREATE TYPE yearly_store_sales_row AS OBJECT(
Page 608
store_name varchar2(25), sales_year number, total_sales number); -- (2) create PL/SQL table of the type CREATE TYPE yearly_store_sales_table AS TABLE OF yearly_store_sales_row; -- (3) create package, or function CREATE OR REPLACE PACKAGE sales_package AS TYPE sales_cursor_type IS REF CURSOR RETURN sales_data%ROWTYPE; FUNCTION modify_sales_data (INPUTDATA IN sales_cursor_type) RETURN yearly_store_sales_table PIPELINED; END; / CREATE OR REPLACE PACKAGE BODY sales_package AS FUNCTION modify_sales_data(inputdata IN sales_cursor_type) RETURN yearly_store_sales_table PIPELINED IS inputrec sales_data%ROWTYPE; outputrow_2001 yearly_store_sales_row := yearly_store_sales_row(NULL,NULL,NULL); outputrow_2002 yearly_store_sales_row := yearly_store_sales_row(NULL,NULL,NULL); BEGIN LOOP FETCH inputdata INTO inputrec; EXIT WHEN inputdata%NOTFOUND; IF INPUTREC.SALES_2001 IS NOT NULL THEN outputrow_2001.store_name := inputrec.store_name; outputrow_2001.sales_year := 2001; outputrow_2001.total_sales:= inputrec.sales_2001; pipe row (outputrow_2001); END IF; IF INPUTREC.SALES_2002 IS NOT NULL THEN outputrow_2002.store_name := inputrec.store_name; outputrow_2002.sales_year := 2002; outputrow_2002.total_sales:= inputrec.sales_2002; pipe row (outputrow_2002); END IF; END LOOP; RETURN; END; END sales_package; / -- (4) usage SELECT * FROM TABLE(sales_package.modify_sales_data( CURSOR(select store_name,sales_2001,sales_2002 FROM sales_data)));
Page 609
Using Collections
Varrays Nested tables Fixed: may not be used in tables Fixed: may be used in tables Dynamic
Associative arrays (index-by PL/SQL tables)
Multiset Operator MULTISET EXCEPT MULTISET INTERSECT MULTISET UNION SET
Its Equivalent in SQL MINUS INTERSECT UNION ALL DISTINCT
Using VARRAYS
Although VARRAYS can be stored as part of a table, it is best suited for PL/SQL proccessing.
TYPE type_name IS {VARRAY | VARYING ARRAY} (size_limit) OF element_type [ NOT NULL ]; -- initialized in the declaration section DECLARE TYPE integer_varray IS VARRAY(3) OF INTEGER; Declare and initialize a varray that allows nulls. varray_integer INTEGER_VARRAY := integer_varray(NULL,NULL,NULL); BEGIN Assign values to subscripted members of the varray. varray_integer(1) := 11; varray_integer(2) := 12; varray_integer(3) := 13; FOR i IN 1..3 LOOP dbms_output.put_line('Integer Varray ['||i||'] '|| '['||varray_integer(i)||']'); END LOOP; END; / -- initialized with zero rows in the declaration -- then dynamic initialization in the execution DECLARE Define a varray of integer with 3 rows. TYPE integer_varray IS VARRAY(3) OF INTEGER; Declare and initialize a null set of rows. varray_integer INTEGER_VARRAY := integer_varray(); BEGIN Loop through the three records. FOR i IN 1..3 LOOP
Page 610
Initialize row. varray_integer.EXTEND; Assign values to subscripted members of the varray. varray_integer(i) := 10 + i; END LOOP; Loop through the records to print the varrays. FOR i IN 1..3 LOOP Print the contents. dbms_output.put ('Integer Varray ['||i||'] '); dbms_output.put_line('['||varray_integer(i)||']'); END LOOP; END; /
-- Varrays as Object Types CREATE OR REPLACE TYPE integer_varray AS VARRAY(3) OF INTEGER;
-- initializing by Collection API CREATE OR REPLACE TYPE integer_varray AS VARRAY(100) OF INTEGER NOT NULL; / DECLARE Declare and initialize a null set of rows. varray_integer INTEGER_VARRAY := integer_varray(); BEGIN Loop through all records. FOR i IN 1..varray_integer.LIMIT LOOP Initialize row. varray_integer.EXTEND; END LOOP; dbms_output.put ('Integer Varray Initialized '); dbms_output.put_line('['||varray_integer.COUNT||']'); END; /
-- Using Varrays as Column Data Types in Tables -- it doesn't produce meaningful output CREATE OR REPLACE TYPE address_varray AS VARRAY(3) OF VARCHAR2(30 CHAR); / CREATE TABLE addresses (address_id INTEGER NOT NULL .. ,street_address ADDRESS_VARRAY NOT NULL
Page 611
.. ); INSERT INTO addresses VALUES (..,address_varray('MyAddress','450 West Paseo Redondo','Suite 200'),..); -- you can insert only one element of the varray INSERT INTO addresses VALUES (12 ,12 ,address_varray('Office of Senator Kennedy') ,..); -- it gives though unmeaningful output SELECT street_address FROM addresses; STREET_ADDRESS ADDRESS_VARRAY('MyAddress','450 West Paseo ... UPDATE addresses SET street_address = address_varray('Your Address' ,'Suite 1150') WHERE address_id = 11;
,'2400 E. Arizona Biltmore Cir.'
-- to update a portion of a varray column, you must use pl/sql DECLARE Define a record type for a row of the addresses table. TYPE address_type IS RECORD .. ; Define a variable of the addresses table record type. address ADDRESS_TYPE; Define a cursor to return the %ROWTYPE value. CURSOR get_street_address (address_id_in INTEGER) IS SELECT * FROM addresses WHERE address_id = address_id_in; BEGIN Open the cursor. OPEN get_street_address(11); Fetch a into the record type variable. FETCH get_street_address INTO address; Close the cursor. CLOSE get_street_address; Reset the first element of the varray type variable. address.street_address(1) := 'Office of Senator John McCain'; Update the varray column value. UPDATE addresses
Page 612
SET street_address = address.street_address WHERE address_id = 11; END;
Using Nested Tables

Nested tables may be used in table, record, and object definitions. They may be accessed in SQL and PL/SQL.
CREATE OR REPLACE TYPE type_name AS TABLE OF element_type [ NOT NULL ]; -- basic example DECLARE -- Define a nested table of variable length strings. TYPE card_table IS TABLE OF VARCHAR2(5 CHAR); -- Declare and initialize a nested table with three rows. cards CARD_TABLE := card_table(NULL,NULL,NULL); BEGIN -- Assign values to subscripted members of the varray. cards(1) := 'Ace'; cards(2) := 'Two'; cards(3) := 'Three'; END; /
-- dynamic initialization and assignment in the execution section DECLARE -- Define a nested table of variable length strings. TYPE card_suit IS TABLE OF VARCHAR2(5 CHAR); -- Declare and initialize a null set of rows. cards CARD_SUIT := card_suit(); BEGIN -- Loop through the three records. FOR i IN 1..3 LOOP -- Initialize row. cards.EXTEND; -- Assign values to subscripted members of the varray. IF i = 1 THEN cards(i) := 'Ace'; ELSIF i = 2 THEN cards(i) := 'Two'; ELSIF i = 3 THEN cards(i) := 'Three'; END IF; END LOOP; END; /
-- Nested Tables as Column Data Types in Tables CREATE OR REPLACE TYPE address_table AS TABLE OF VARCHAR2(30 CHAR) NOT NULL;
Page 613
/ CREATE TABLE addresses ( .. ,street_address ADDRESS_TABLE ...) NESTED TABLE street_address STORE AS nested_street_address; -- any number of elements can be inserted INSERT INTO addresses VALUES ( .. ,address_table('Office of Senator McCain' ,'450 West Paseo Redondo','Suite 200'),...);
-- arrange generated output -- the normal output: SELECT street_address FROM addresses; ADDRESS_TABLE('Office of Senator McCain', '450 West Paseo ... -- to arrange it: SELECT column_value FROM THE (SELECT street_address FROM addresses WHERE id = 1);
-- updating a nested-table using pl/sql CREATE OR REPLACE FUNCTION many_to_one (street_address_in ADDRESS_TABLE) RETURN VARCHAR2 IS Define a return variable and initial it. retval VARCHAR2(4000) := ''; BEGIN Loop from the beginning to end of the nested table. FOR i IN 1..street_address_in.COUNT LOOP Append the next value and a line break. retval := retval || street_address_in(i) || CHR(10); END LOOP; RETURN retval; END many_to_one; /
-- You can update a portion of a nested table column directly in SQL: UPDATE THE (SELECT street_address FROM addresses WHERE address_id = 21) SET column_value = 'Office of Senator John McCain' WHERE column_value = 'Office of Senator McCain';
Using Associative Arrays

Associative arrays cannot be used in tables. They are the key to using the FORALL statement or BULK COLLECT clause. This collection provides faster and simpler lookups
CREATE OR REPLACE TYPE type_name
Page 614
AS TABLE OF element_type [ NOT NULL ] INDEX BY [ PLS_INTEGER | BINARY_INTEGER | VARCHAR2(size) ] -- index by integer DECLARE -- Define an associative array of strings. TYPE card_table IS TABLE OF VARCHAR2(5 CHAR) INDEX BY BINARY_INTEGER; cards CARD_TABLE; BEGIN cards(2) := 'test'; cards(10) := 'test'; cards(1) := 'test'; DBMS_OUTPUT.PUT_LINE(cards.first); DBMS_OUTPUT.PUT_LINE(cards.last); DBMS_OUTPUT.PUT_LINE(cards.count); END; /
-- index by string .. -- Loop through all the associative array elements. FOR i IN 1..calendar.COUNT LOOP -- Check if the first element in the loop. IF i = 1 THEN -- Assign the first character index to a variable. current := calendar.FIRST; -- Use the derived index to find the next index. element := calendar(current); ELSE -- Check if next index value exists. IF calendar.NEXT(current) IS NOT NULL THEN -- Assign the character index to a variable. current := calendar.NEXT(current); -- Use the derived index to find the next index. element := calendar(current); ELSE -- Exit loop since last index value is read. EXIT; END IF; END IF; END LOOP; ..
-- Using Associative Arrays with BULK COLLECT and FORALL -- BULK COLLECT to retrieve a record set into associative arrays or nested tables -- FORALL to send DML statements in batches -- using FORALL CREATE TABLE bulk_numbers (number_id NUMBER ,CONSTRAINT number_id_pk DECLARE
PRIMARY KEY (number_id));
Page 615
TYPE number_table IS TABLE OF bulk_numbers.number_id%TYPE INDEX BY BINARY_INTEGER; number_list NUMBER_TABLE; BEGIN FOR i IN 1..10000 LOOP -- Assign number value. number_list(i) := i; END LOOP; -- Loop through all to do a bulk insert. FORALL i IN 1..number_list.COUNT INSERT INTO bulk_numbers VALUES (number_list(i)); COMMIT; END; / -- Using a BULK COLLECT DECLARE TYPE number_table IS TABLE OF bulk_numbers.number_id%TYPE INDEX BY BINARY_INTEGER; number_list NUMBER_TABLE; BEGIN -- Check if calendar has no elements. SELECT number_id BULK COLLECT INTO number_list FROM bulk_numbers ORDER BY 1; FOR i IN number_list.FIRST..number_list.LAST LOOP -- print only the first and last two IF i <= 2 OR i >= 9999 THEN DBMS_OUTPUT.PUT_LINE('Number ['||number_list(i)||']'); END IF; END LOOP; END; /
Using Collection API

API Function or Procedure Notes Return Type
COUNT DELETE(n) DELETE(n,m) EXISTS(n) EXTEND EXTEND(n) not in Associative Tables n = number of elements to add n= minimum m=maximum
PLS_INTEGER NONE NONE TRUE or FALSE NONE NONE
Page 616
EXTEND(n,i) FIRST LAST LIMIT NEXT(n) PRIOR(n) TRIM TRIM(n)
i= number of replicated elements the lowest subscript the highest subscript If there is no next element, it will return null. If there is no previous element, it will return null. removes the highest subscripted value from a collection removes the number or elements passed
NONE INTEGER or VARCHAR2 INTEGER or VARCHAR2 INTEGER INTEGER or VARCHAR2 INTEGER or VARCHAR2 NONE NONE
Errors returned by Collections: COLLECTION_IS_NULL NO_DATA_FOUND An attempt to use a null collection. An attempt to use a subscript that has been deleted or is a nonexistent unique string index value in an associative array. This error applies only to varrays and nested tables.
SUBSCRIPT_BEYOND_COUNT SUBSCRIPT_OUTSIDE_LIMIT
-- using DELETE method DECLARE -- Define a nested table type of INTEGER. TYPE number_table IS TABLE OF INTEGER; -- Define a variable of the nested table type. number_list NUMBER_TABLE; -- Define a local procedure to check and print elements. PROCEDURE print_list (list_in NUMBER_TABLE) IS BEGIN -- Loop through the possible index values of the list. FOR i IN list_in.FIRST..list_in.LAST LOOP -- Check if the subscripted element is there. IF list_in.EXISTS(i) THEN -- Print the element. DBMS_OUTPUT.PUT_LINE('List ['||list_in(i)||']'); END IF; END LOOP; END print_list; BEGIN -- Check if a subscript element of one does not exists. IF NOT number_list.EXISTS(1) THEN -- Construct the collection. number_list := number_table(1,2,3,4,5); END IF; -- Print a title. DBMS_OUTPUT.PUT_LINE('Nested table before a deletion'); DBMS_OUTPUT.PUT_LINE('------------------------------'); -- Print the list.
Page 617
print_list(number_list); -- Delete an element. number_list.DELETE(2,4); -- Print a title. DBMS_OUTPUT.PUT_LINE(CHR(10)|| 'Nested table after a deletion'); DBMS_OUTPUT.PUT_LINE('-----------------------------'); -- Print the list. print_list(number_list); END; /
-- using EXTEND SET SERVEROUTPUT ON SIZE 1000000 DECLARE -- Define a nested table type of INTEGER. TYPE number_table IS TABLE OF INTEGER; -- Define a variable of the nested table type. number_list NUMBER_TABLE := number_table(1,2); -- Define a local procedure to check and print elements. PROCEDURE print_list (list_in NUMBER_TABLE) IS BEGIN -- Loop through the possible index values of the list. FOR i IN list_in.FIRST..list_in.LAST LOOP -- Check if the subscripted element is there. IF list_in.EXISTS(i) THEN -- Print the element. DBMS_OUTPUT.PUT_LINE('List ['||list_in(i)||']'); END IF; END LOOP; END print_list; BEGIN -- Print a title. DBMS_OUTPUT.PUT_LINE('Nested table before extension'); DBMS_OUTPUT.PUT_LINE('-----------------------------'); -- Print the list. print_list(number_list); -- Allocate two null elements. number_list.EXTEND(2); -- Allocate three elements and copy element two number_list.EXTEND(3,2); -- Print a title. DBMS_OUTPUT.PUT_LINE(CHR(10)|| 'Nested table after extension'); DBMS_OUTPUT.PUT_LINE('----------------------------'); -- Print the list. print_list(number_list); END; /
-- using LIMIT Method (function) DECLARE -- Define a varray type of INTEGER. TYPE number_varray IS VARRAY(5) OF INTEGER;
Page 618
-- Define a variable of the varray type. number_list NUMBER_VARRAY := number_varray(1,2,3); -- Define a local procedure to check and print elements. PROCEDURE print_list (list_in NUMBER_VARRAY) IS BEGIN -- Loop through the possible index values of the list. FOR i IN list_in.FIRST..list_in.COUNT LOOP -- Print the element. DBMS_OUTPUT.PUT_LINE( 'List Index ['||i||'] '|| 'List Value ['||list_in(i)||']'); END LOOP; END print_list; BEGIN -- Print a title. DBMS_OUTPUT.PUT_LINE('Varray after initialization'); DBMS_OUTPUT.PUT_LINE('---------------------------'); -- Print the list. print_list(number_list); -- Extend null element to maximum limit. number_list.EXTEND(number_list.LIMIT - number_list.LAST); -- Print a title. DBMS_OUTPUT.PUT(CHR(10)); DBMS_OUTPUT.PUT_LINE('Varray after extension'); DBMS_OUTPUT.PUT_LINE('----------------------'); -- Print the list. print_list(number_list); END; /
-- using TRIM Method DECLARE -- Define a varray type of INTEGER. TYPE number_varray IS VARRAY(5) OF INTEGER; -- Define a variable of the varray type. number_list NUMBER_VARRAY := number_varray(1,2,3,4,5); -- Define a local procedure to check and print elements. PROCEDURE print_list (list_in NUMBER_VARRAY) IS BEGIN -- Loop through the possible index values of the list. FOR i IN list_in.FIRST..list_in.COUNT LOOP -- Print the element. DBMS_OUTPUT.PUT_LINE( 'List Index ['||i||'] '|| 'List Value ['||list_in(i)||']'); END LOOP; END print_list; BEGIN -- Print a title. DBMS_OUTPUT.PUT_LINE('Varray after initialization'); DBMS_OUTPUT.PUT_LINE('---------------------------'); -- Print the list. print_list(number_list);
Page 619
-- Extend null element to maximum limit. number_list.TRIM; -- Print a title. DBMS_OUTPUT.PUT(CHR(10)); DBMS_OUTPUT.PUT_LINE('Varray after a single element trim'); DBMS_OUTPUT.PUT_LINE('----------------------------------'); -- Print the list. print_list(number_list); -- Extend null element to maximum limit. number_list.TRIM(3); -- Print a title. DBMS_OUTPUT.PUT(CHR(10)); DBMS_OUTPUT.PUT_LINE('Varray after a three element trim'); DBMS_OUTPUT.PUT_LINE('---------------------------------'); -- Print the list. print_list(number_list); END; /
Page 620
Handling Errors
Predefined Exceptions
Oracle Error ORA-0001 ORA-0051 ORA-1001 ORA-1012 ORA-1017 ORA-1403 ORA-1410 ORA-1422 ORA-1476 ORA-1722 ORA-1725 ORA-6500 ORA-6501 ORA-6502 ORA-6504 ORA-6511 ORA-6530 ORA-6531 ORA-6532 ORA-6533 Equivalent Exception DUP_VAL_ON_INDEX TIMEOUT_ON_RESOURCE INVALID_CURSOR NOT_LOGGED_ON LOGIN_DENIED NO_DATA_FOUND SYS_INVALID_ROWID TOO_MANY_ROWS ZERO_DIVIDE INVALID_NUMBER USERENV_COMMITSCN_ERROR 1 STORAGE_ERROR PROGRAM_ERROR VALUE_ERROR ROWTYPE_MISMATCH CURSOR_ALREADY_OPEN ACCESS_INTO_NULL COLLECTION_IS_NULL SUBSCRIPT_OUTSIDE_LIMIT SUBSCRIPT_BEYOND_COUNT Description Unique constraint violated. Time-out occurred while waiting for resource. Illegal cursor operation. Not connected to Oracle. Invalid user name/password. No data found. Conversion to a universal rowid failed. A SELECT.INTO statement matches more than one row. Division by zero. Conversion to a number failed; for example, '1A' is not valid. Incorrect usage of the USERENV('COMMITSCN') function. Internal PL/SQL error raised if PL/SQL runs out of memory. Internal PL/SQL error. Truncation, arithmetic, or conversion error. Host cursor variable and PL/SQL cursor variable have incompatible row types. Attempt to open a cursor that is already open. Attempt to assign values to the attributes of a NULL object. Attempt to apply collection methods other than EXISTS to a NULL PL/SQL table or varray. Reference to a nested table or varray index outside the declared range (such as 1). Reference to a nested table or varray index higher than the number of elements in the collection. Caller of a pipelined function does not need more rows. No matching WHEN clause in a CASE statement is found. Attempt to call a method on a null object instance.
ORA-6548 ORA-6592 ORA-30625
NO_DATA_NEEDED1 CASE_NOT_FOUND2 SELF_IS_NULL
Using User-Defined Exceptions

It is impossible for an exception handler to be defined for more than one exception simultaneously; i.e.: WHEN exception1 AND exception2 will raise a compilation error. Examining the Error Stack: Use SQLCODE and SQLERRM OR: DBMS_UTILITY.FORMAT_ERROR_STACK provides a function FORMAT_ERROR_STACK that returns the same information as SQLERRM, also limited to 2000 bytes. Oracle DBA Code Examples
Page 621
DECLARE -- Exception to indicate an error condition e_DuplicateAuthors EXCEPTION; -- IDs for three authors v_Author1 books.author1%TYPE; v_Author2 books.author2%TYPE; v_Author3 books.author3%TYPE; BEGIN /* Find the IDs for the 3 authors of 'Oracle9i DBA 101' */ SELECT author1, author2, author3 INTO v_Author1, v_Author2, v_Author3 FROM books WHERE title = 'Oracle9i DBA 101'; /* Ensure that there are no duplicates */ IF (v_Author1 = v_Author2) OR (v_Author1 = v_Author3) OR (v_Author2 = v_Author3) THEN RAISE e_DuplicateAuthors; END IF; EXCEPTION WHEN e_DuplicateAuthors THEN INSERT INTO log_table (info) VALUES ('Oracle9i DBA 101 has duplicate authors'); WHEN OTHERS THEN INSERT INTO log_table (code, message, info) VALUES (NULL, SUBSTR(DBMS_UTILITY.FORMAT_ERROR_STACK, 1, 200),'Oracle error occurred'); END; /
/* The EXCEPTION_INIT Pragma */ DECLARE e_MissingNull EXCEPTION; PRAGMA EXCEPTION_INIT(e_MissingNull, -1400); BEGIN INSERT INTO authors (id) VALUES (NULL); EXCEPTION WHEN e_MissingNull then INSERT INTO log_table (info) VALUES ('ORA-1400 occurred'); END; /
Using RAISE_APPLICATION_ERROR
RAISE_APPLICATION_ERROR(error_number, error_message, [keep_errors]); error_number is a value between 20,000 and 20,999 The error_message must be fewer than 512 characters If keep_errors is TRUE, the new error is added to the list of errors already raised (if one exists). If it is FALSE, which is the default, the new error will replace the current list of errors. ..
Page 622
IF v_AuthorCount = 0 THEN RAISE_APPLICATION_ERROR(-20001, 'Author1 ' || p_Author1 || ' does not exist'); ..
Page 623
Autonomous Transactions
Autonomous transactions are started by a parent, or main, transaction but operate independently of the parent for transaction control. If a commit or rollback is used in the autonomous or main transaction, or if a failure occurs for any reason, it does not impact the other transaction. CREATE OR REPLACE PROCEDURE logging_ins ( i_username IN VARCHAR2, i_datetime IN TIMESTAMP) IS PRAGMA AUTONOMOUS_TRANSACTION; BEGIN INSERT INTO logging (username, datetime) VALUES (i_username, i_datetime); commit; -- commit is a must here in Autonomous proc othewise ORA-06519 returned END; /
Page 624
Some Stored Subprobrams Concepts

Serially Reusable Packages
Serially Reusable Packages Run-time state is kept in shared memory and is freed after every database call. The maximum memory used is proportional to the number of concurrent users of the package. CREATE OR REPLACE PACKAGE TEST_PCK IS -- display number of rows passed PROCEDURE DISPLAY_ROWS(P_N IN NUMBER); END; / NonSerially Reusable Packages Run-time state is kept in process memory and lasts for the life of the database session. The maximum memory used is proportional to the number of concurrently logged-on users, which is typically much higher.
CREATE OR REPLACE PACKAGE BODY TEST_PCK IS -- this cursor keeps opening in the session life CURSOR CR IS SELECT CUSTOMER_ID FROM CUSTOMERS ORDER BY CUSTOMER_ID; PROCEDURE DISPLAY_ROWS(P_N IN NUMBER) IS V_ID NUMBER; I NUMBER := 0; V_DONE BOOLEAN := FALSE; BEGIN IF NOT CR%ISOPEN THEN OPEN CR; END IF; -- display only passed number of rows WHILE NOT V_DONE LOOP FETCH CR INTO V_ID; IF CR%NOTFOUND THEN CLOSE CR; V_DONE := TRUE; ELSE I := I + 1; DBMS_OUTPUT.PUT_LINE(V_ID); IF I >= P_N THEN V_DONE := TRUE; END IF; END IF; END LOOP; END DISPLAY_ROWS; END TEST_PCK; /
-- if you add PRAGMA SERIALLY_REUSABLE, CR will at reset every call
Page 625
CREATE OR REPLACE PACKAGE TEST_PCK IS PRAGMA SERIALLY_REUSABLE; .. CREATE OR REPLACE PACKAGE BODY TEST_PCK IS PRAGMA SERIALLY_REUSABLE; ..
Stored Subprograms and Roles

If a stored subprogram refers to an object in another schema, that object should be granted directly to the subprogram owner, not via a role. CONN SA/s CREATE ROLE R1; GRANT SELECT ON CUSTOMERS TO R1; GRANT R1 TO USER1; CONN USER1/U CREATE OR REPLACE PACKAGE USER1.TEST_PCK IS PROCEDURE DISPLAY_ROW; END; / -- the following will not compile because CUSTOMERS granted -- to USER1 via role (not directly) CREATE OR REPLACE PACKAGE BODY USER1.TEST_PCK IS PROCEDURE DISPLAY_ROW IS N NUMBER; BEGIN SELECT CUSTOMER_ID INTO N FROM SA.CUSTOMERS WHERE CUSTOMER_ID=101; END DISPLAY_ROW; END TEST_PCK; /
Invokers vs. Definers Rights

CREATE [OR REPLACE] FUNCTION function_name [parameter_list] RETURN return_type [AUTHID {CURRENT_USER | DEFINER}] {IS | AS}
Page 626
Pinning an Programunit in the Shared Pool

To pin an object: DBMS_SHARED_POOL.KEEP(name VARCHAR2, flag CHAR DEFAULT 'P') flag Determines the type of the object: P Package, function, or procedure Q Sequence R Trigger T Object type (Oracle8 and higher) JS Java source (Oracle8i and higher) JC Java class (Oracle8i and higher) JR Java resource (Oracle8i and higher) JD Java shared data (Oracle8i and higher) C SQL cursor
To echo the contents of the shared pool of objects greater than minisize to the screen:
DBMS_SHARED_POOL.SIZES(minsize NUMBER)
Page 627
Using Triggers
General Syntax of creating any trigger: CREATE [OR REPLACE] TRIGGER trigger_name {BEFORE | AFTER | INSTEAD OF} triggering_event [referencing_clause] [WHEN trigger_condition] [FOR EACH ROW] trigger_body Restrictions on Triggers A trigger may not issue any transaction control statementsCOMMIT, ROLLBACK, SAVEPOINT, or SET TRANSACTION. The trigger body cannot declare any LONG or LONG RAW variables. Code in a trigger body may reference and use LOB (Large OBject) columns, but it may not modify the values of the columns. This is also true for object columns.
Using DML Triggers

CREATE OR REPLACE TRIGGER UpdateCategoryStats AFTER INSERT OR DELETE OR UPDATE ON books DECLARE CURSOR c_Statistics IS SELECT category, COUNT(*) total_books, AVG(price) average_price FROM books GROUP BY category; BEGIN DELETE FROM category_stats; FOR v_StatsRecord in c_Statistics LOOP INSERT INTO category_stats (category, total_books, average_price) VALUES (v_StatsRecord.category, v_StatsRecord.total_books, v_StatsRecord.average_price); END LOOP; END UpdateCategoryStats; /
CREATE OR REPLACE TRIGGER GenerateAuthorID BEFORE INSERT OR UPDATE ON authors REFERENCING new AS new_author FOR EACH ROW BEGIN /* Fill in the ID field of authors with the next value from author_sequence. Since ID is a column in authors, :new.ID is a valid reference. */ SELECT author_sequence.NEXTVAL INTO :new_author.ID
Page 628
FROM dual; END GenerateAuthorID; /
-- using WHEN keyword CREATE OR REPLACE TRIGGER CheckPrice BEFORE INSERT OR UPDATE OF price ON books FOR EACH ROW WHEN (new.price > 49.99) BEGIN .. END; /
-- using Trigger Predicates CREATE OR REPLACE TRIGGER LogInventoryChanges BEFORE INSERT OR DELETE OR UPDATE ON inventory FOR EACH ROW DECLARE v_ChangeType CHAR(1); BEGIN /* Use 'I' for an INSERT, 'D' for DELETE, and 'U' for UPDATE. */ IF INSERTING THEN v_ChangeType := 'I'; ELSIF UPDATING THEN v_ChangeType := 'U'; ELSE v_ChangeType := 'D'; END IF; .. END LogInventoryChanges; /
Using Instead-of Triggers

All instead-of triggers are row level, whether or not the FOR EACH ROW clause is present. CREATE OR REPLACE TRIGGER InsertBooksAuthors INSTEAD OF INSERT ON books_authors DECLARE v_Book books%ROWTYPE; v_AuthorID authors.id%TYPE; BEGIN -- Figure out the ID of the new author BEGIN SELECT id INTO v_AuthorID FROM authors WHERE first_name = :new.first_name AND last_name = :new.last_name; EXCEPTION WHEN NO_DATA_FOUND THEN -- No author found, create a new one INSERT INTO authors (id, first_name, last_name)
Page 629
VALUES (author_sequence.NEXTVAL, :new.first_name, :new.last_name) RETURNING ID INTO v_AuthorID; END; SELECT * INTO v_Book FROM books WHERE isbn = :new.isbn; -- Figure out whether the book already has 1 or 2 authors, and update -- accordingly IF v_Book.author2 IS NULL THEN UPDATE books SET author2 = v_AuthorID WHERE isbn = :new.isbn; ELSE UPDATE books SET author3 = v_AuthorID WHERE isbn = :new.isbn; END IF; END InsertBooksAuthors; /
Using System Triggers

System trigger fires on two different kinds of events: DDL or database. Because system triggers are generally committed anyway, declaring them as autonomous will not have any effect. Required privilege on database triggers: ADMINISTER DATABASE TRIGGER
CREATE [OR REPLACE] TRIGGER [schema.]trigger_name {BEFORE | AFTER} {ddl_event_list | database_event_list} ON {DATABASE | [schema.]SCHEMA} [when_clause] trigger_body; Database Events:
When Trigger Fires When the database is opened. Conditions Restrictions None allowed No database operations allowed in the trigger. Return status ignored. SHUTDOWN Just before the server starts the shutdown of an instance. This lets the cartridge shutdown completely. For abnormal instance shutdown, this triiger None allowed No database operations allowed in the trigger. Return status ignored. Transaction Starts a separate transaction and commits it after firing the triggers. Attribute Functions ora_sysevent ora_login_user ora_instance_num ora_database_name
Event STARTUP
Starts a separate transaction and commits it after firing the triggers.
ora_sysevent ora_login_user ora_instance_num ora_database_name
Page 630
Event
When Trigger Fires might not fire.
Conditions Restrictions
Transaction
Attribute Functions
DB_ROLE_CH ANGE
When the database is opened for the first time after a role change.
None allowed
Return status ignored.
Starts a separate transaction and commits it after firing the triggers. Starts a separate transaction and commits it after firing the triggers.
ora_sysevent ora_login_user ora_instance_num ora_database_name
SERVERERRO R
When the error eno occurs. If no condition is given, then this trigger fires whenever an error occurs. The trigger does not fire on ORA-1034, ORA1403, ORA-1422, ORA1423, and ORA-4030 because they are not true errors or are too serious to continue processing. It also fails to fire on ORA-18 and ORA-20 because a process is not available to connect to the database to record the error.
ERRNO = eno
Depends on the error. Return status ignored.
ora_sysevent ora_login_user ora_instance_num ora_database_name ora_server_error ora_is_servererror space_error_info
Schema or Client Events: When Trigger Fires When a catalog object is altered. Attribute Functions ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_type ora_dict_obj_name ora_dict_obj_owner ora_des_encrypted_password (for ALTER USER events) ora_is_alter_column (for ALTER TABLE events) ora_is_drop_column (for ALTER TABLE events) ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_type ora_dict_obj_name ora_dict_obj_owner ora_sysevent ora_login_user ora_instance_num ora_database_name Oracle DBA Code Examples
Event BEFORE ALTER AFTER ALTER
BEFORE DROP AFTER DROP
When a catalog object is dropped.
BEFORE ANALYZE AFTER ANALYZE
When an analyze statement is issued
Page 631
Event
When Trigger Fires
Attribute Functions ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner
BEFORE ASSOCIATE STATISTICS AFTER ASSOCIATE STATISTICS
When an associate statistics statement is issued
ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_dict_obj_name_list ora_dict_obj_owner_list ora_sysevent ora_login_user ora_instance_num ora_database_name ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_type ora_dict_obj_name ora_dict_obj_owner ora_is_creating_nested_table (for CREATE TABLE events) ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_dict_obj_name_list ora_dict_obj_owner_list ora_sysevent ora_login_user ora_instance_num ora_database_name Oracle DBA Code Examples
BEFORE AUDIT AFTER AUDIT BEFORE NOAUDIT AFTER NOAUDIT BEFORE COMMENT AFTER COMMENT
When an audit or noaudit statement is issued
When an object is commented
BEFORE CREATE AFTER CREATE
When a catalog object is created.
BEFORE DDL AFTER DDL
When most SQL DDL statements are issued. Not fired for ALTER DATABASE, CREATE CONTROLFILE, CREATE DATABASE, and DDL issued through the PL/SQL subprogram interface, such as creating an advanced queue. When a disassociate statistics statement is issued
BEFORE DISASSOCIATE STATISTICS AFTER DISASSOCIATE STATISTICS
BEFORE GRANT AFTER GRANT
When a grant statement is issued
Page 632
Event
When Trigger Fires
Attribute Functions ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_grantee ora_with_grant_option ora_privileges
BEFORE LOGOFF
At the start of a user logoff
ora_sysevent ora_login_user ora_instance_num ora_database_name ora_sysevent ora_login_user ora_instance_num ora_database_name ora_client_ip_address ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_owner ora_dict_obj_type ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner ora_revokee ora_privileges ora_sysevent ora_login_user ora_instance_num ora_database_name ora_server_error ora_is_servererror space_error_info ora_sysevent ora_login_user ora_instance_num ora_database_name ora_dict_obj_name ora_dict_obj_type ora_dict_obj_owner
AFTER LOGON
After a successful logon of a user.
BEFORE RENAME AFTER RENAME
When a rename statement is issued.
BEFORE REVOKE AFTER REVOKE
When a revoke statement is issued
AFTER SUSPEND
After a SQL statement is suspended because of an out-of-space condition. The trigger must correct the condition so the statement can be resumed.
BEFORE TRUNCATE AFTER TRUNCATE
When an object is truncated
CREATE OR REPLACE TRIGGER LogCreations AFTER CREATE ON SCHEMA BEGIN INSERT INTO ddl_creations (user_id, object_type, object_name, object_owner, creation_date) VALUES (USER, ORA_DICT_OBJ_TYPE, ORA_DICT_OBJ_NAME,
Page 633
ORA_DICT_OBJ_OWNER, SYSDATE); END LogCreations;
Handling Mutating Tables in Triggers

SQL statements in a trigger body may not: o o o Read from or modify any mutating table of the triggering statement. This includes the triggering table itself. Read from or modify the primary-, unique-, or foreign-key columns of a constraining table of the triggering table. They may, however, modify the other columns if desired. These restrictions apply to all row-level triggers. They apply for statement triggers only when the statement trigger would be fired as a result of a DELETE CASCADE operation.
If an INSERT statement affects only one row, the before- and after-row triggers for that row do not treat the triggering table as mutating. This is the only case where a row-level trigger may read from or modify the triggering table. Statements such as INSERT INTO table SELECT ... always treat the triggering table as mutating, even if the subquery returns only one row. -- the following trigger will return ORA-4091 CREATE OR REPLACE TRIGGER LimitMajors BEFORE INSERT OR UPDATE OF major ON students FOR EACH ROW DECLARE v_MaxStudents CONSTANT NUMBER := 5; v_CurrentStudents NUMBER; BEGIN SELECT COUNT(*) INTO v_CurrentStudents FROM students WHERE major = :new.major; -- If there isn't room, raise an error. IF v_CurrentStudents + 1 > v_MaxStudents THEN RAISE_APPLICATION_ERROR(-20000, 'Too many students in major ' || :new.major); END IF; END LimitMajors; /
To workaround: create two triggers: one statement level and the other row level. The statement trigger store the SELECT result into a PL/SQL table in a package. The row-level trigger will read that data from the package. CREATE OR REPLACE PACKAGE StudentData AS TYPE t_Majors IS TABLE OF students.major%TYPE INDEX BY BINARY_INTEGER; TYPE t_IDs IS TABLE OF students.ID%TYPE INDEX BY BINARY_INTEGER; v_StudentMajors t_Majors; v_StudentIDs t_IDs; v_NumEntries BINARY_INTEGER := 0; END StudentData; /
Page 634
CREATE OR REPLACE TRIGGER RLimitMajors BEFORE INSERT OR UPDATE OF major ON students FOR EACH ROW BEGIN StudentData.v_NumEntries := StudentData.v_NumEntries + 1; StudentData.v_StudentMajors(StudentData.v_NumEntries) := :new.major; StudentData.v_StudentIDs(StudentData.v_NumEntries) := :new.id; END RLimitMajors; / CREATE OR REPLACE TRIGGER SLimitMajors AFTER INSERT OR UPDATE OF major ON students DECLARE v_MaxStudents CONSTANT NUMBER := 2; v_CurrentStudents NUMBER; v_StudentID students.ID%TYPE; v_Major students.major%TYPE; BEGIN /* Loop through each student inserted or updated, and verify that we are still within the limit. */ FOR v_LoopIndex IN 1..StudentData.v_NumEntries LOOP v_StudentID := StudentData.v_StudentIDs(v_LoopIndex); v_Major := StudentData.v_StudentMajors(v_LoopIndex); -- Determine the current number of students in this major. SELECT COUNT(*) INTO v_CurrentStudents FROM students WHERE major = v_Major; -- If there isn't room, raise an error. IF v_CurrentStudents > v_MaxStudents THEN RAISE_APPLICATION_ERROR(-20000, 'Too many students for major ' || v_Major || ' because of student ' || v_StudentID); END IF; END LOOP; -- Reset the counter so the next execution will use new data. StudentData.v_NumEntries := 0; END SLimitMajors;
Dropping and Disabling Triggers

DROP TRIGGER <triggername>; ALTER TRIGGER triggername {DISABLE | ENABLE}; ALTER TABLE authors {DISABLE | ENABLE} ALL TRIGGERS;
Page 635
Using Dynamic SQL

Working with Native Dynamic SQL
Using Dynamic SQL Working with Native Dynamic SQL -- DDL statement := 'CREATE SEQUENCE '||sequence_name||CHR(10) || ' INCREMENT BY 1' ||CHR(10) || ' START WITH 1' ||CHR(10) || ' CACHE 20' ||CHR(10) || ' ORDER';
-- encapsulates a PL/SQL block SELECT-INTO PROCEDURE increment_sequence ( sequence_name IN VARCHAR2, sequence_value IN OUT NUMBER ) IS -- Define local native dynamic SQL variables. statement VARCHAR2(2000); BEGIN -- Build dynamic SQL statement as anonymous block PL/SQL unit. statement := 'BEGIN' ||CHR(10) || ' SELECT hr.'||sequence_name||'.nextval'||CHR(10) || ' INTO :retval' ||CHR(10) || ' FROM DUAL;' ||CHR(10) || 'END;'; -- Execute dynamic SQL statement. EXECUTE IMMEDIATE statement USING OUT sequence_value; END increment_sequence;
-- DML without bind variables (less performance) PROCEDURE insert_into_table ( table_name IN VARCHAR2 , table_column_value1 IN NUMBER , table_column_value2 IN VARCHAR2 , table_column_value3 IN VARCHAR2) IS -- Define local variables. statement VARCHAR2(2000); BEGIN -- Build dynamic SQL statement. statement := 'INSERT ' || 'INTO '||table_name||' ' || 'VALUES (' || ''''||table_column_value1||''',' || ''''||table_column_value2||''',' || ''''||table_column_value3||''')';
Page 636
-- Execute the NDS statement. EXECUTE IMMEDIATE statement; -- Commit the records. commit; END insert_into_table;
-- a DML with ordered bind variables PROCEDURE inserts_into_table ( table_name IN VARCHAR2 , table_column_value1 IN NUMBER , table_column_value2 IN VARCHAR2 , table_column_value3 IN VARCHAR2) IS -- Define local variables. statement VARCHAR2(2000); BEGIN -- Build dynamic SQL statement. statement := 'INSERT ' || 'INTO '||table_name||' ' || 'VALUES (:col_one, :col_two, :col_three)'; -- Execute the NDS statement. EXECUTE IMMEDIATE statement USING table_column_value1 , table_column_value2 , table_column_value3; -- Commit the records. commit; END inserts_into_table;
-- select single row , single col (DQL) PROCEDURE single_row_return IS -- Define local variables. statement VARCHAR2(2000); value_out VARCHAR2(1); BEGIN -- Build dynamic SQL statement. statement := 'SELECT ''A'' FROM DUAL'; -- Use NDS to query a static string. EXECUTE IMMEDIATE statement INTO value_out; END single_row_return;
-- select single row, mutiple col (DQL) PROCEDURE single_row_return ( table_name VARCHAR2 , column_name1 VARCHAR2 , column_name2 VARCHAR2 , column_name3 VARCHAR2 ) IS -- Define local variables.
Page 637
statement VARCHAR2(2000); cvalue_out1 VARCHAR2(20); cvalue_out2 VARCHAR2(30); nvalue_out NUMBER; BEGIN -- Build dynamic SQL statement. statement := 'SELECT ' || column_name1 ||',' || column_name2 ||',' || column_name3 ||' ' || 'FROM '|| table_name; EXECUTE IMMEDIATE statement INTO nvalue_out, cvalue_out1, cvalue_out2; END single_row_return;
-- SELECT multiple row of single col (DQL) -- Create a Varray of a one character string -- it can also be a local type defined in the procedure -- it can also be any PL/SQL Collection like PL/SQL table CREATE OR REPLACE TYPE varchar2_table1 IS VARRAY(100) OF VARCHAR2(1); / PROCEDURE multiple_row_return IS -- Define local variables. statement VARCHAR2(2000); value_out VARCHAR2_TABLE1; BEGIN -- using an anonymous block is mandatory, otherwise ORA-03001 statement := 'BEGIN ' || 'SELECT ''A'' ' || 'BULK COLLECT INTO :col_val ' || 'FROM DUAL;' || 'END;'; -- Use Bulk NDS to query a static string. EXECUTE IMMEDIATE statement USING OUT value_out; -- Use a range loop to read the values. FOR i IN 1..value_out.COUNT LOOP -- Print output message. dbms_output.put_line(value_out(i)); END LOOP; END multiple_row_return;
-- multiple row with columns (DQL) -- NDS can use bulk collections only from within an anonymous-block CREATE OR REPLACE TYPE card_number_varray IS VARRAY(100) OF NUMBER; CREATE OR REPLACE TYPE card_name_varray IS VARRAY(100) OF VARCHAR2(2000); CREATE OR REPLACE TYPE card_suit_varray IS VARRAY(100) OF VARCHAR2(2000);
Page 638
PROCEDURE multiple_row_return ( table_name VARCHAR2 , column_name1 VARCHAR2 , column_name2 VARCHAR2 , column_name3 VARCHAR2 ) IS -- Define local Native Dynamic SQL variables. statement VARCHAR2(2000); cvalue_out1 CARD_NAME_VARRAY; cvalue_out2 CARD_SUIT_VARRAY; nvalue_out CARD_NUMBER_VARRAY; BEGIN statement := 'BEGIN ' || 'SELECT ' || column_name1 ||',' || column_name2 ||',' || column_name3 ||' ' || 'BULK COLLECT INTO :col1, :col2, :col3 ' || 'FROM '|| table_name ||';' || 'END;'; -- Execute native dynamic SQL. EXECUTE IMMEDIATE statement USING OUT nvalue_out, OUT cvalue_out1, OUT cvalue_out2; FOR i IN 1..nvalue_out.COUNT LOOP dbms_output.put_line('Value from ['||column_name1||'] '|| 'is: ['||nvalue_out(i)||']'); dbms_output.put_line('Value from ['||column_name1||'] '|| 'is: ['||SUBSTR(cvalue_out1(i),1,20)||']'); dbms_output.put_line('Value from ['||column_name1||'] '|| 'is: ['||SUBSTR(cvalue_out2(i),1,30)||']'); END LOOP; END multiple_row_return;
Using DBMS_SQL
DBMS_SQL still has a major feature that is not delivered in NDS. It does not need to know beforehand the number and types of arguments it will receive and process. To use dbms_sql: GRANT EXECUTE ON dbms_sys_sql TO SYSTEM WITH GRANT OPTION; GRANT EXECUTE ON dbms_sql TO SYSTEM WITH GRANT OPTION;
-- Working with DDL and DML Without Bind Variables -- Procedure to close DBMS_SQL open cursor. PROCEDURE close_open_cursor ( c IN OUT INTEGER) IS BEGIN IF dbms_sql.is_open(c) THEN dbms_sql.close_cursor(c); END IF; END close_open_cursor; PROCEDURE create_sequence ( sequence_name IN VARCHAR2) IS c INTEGER := dbms_sql.open_cursor; fdbk INTEGER;
Page 639
statement VARCHAR2(2000); BEGIN -- Build dynamic SQL statement. statement := 'CREATE SEQUENCE '||sequence_name||CHR(10) || ' INCREMENT BY 1' ||CHR(10) || ' START WITH 1' ||CHR(10) || ' CACHE 20' ||CHR(10) || ' ORDER'; -- Parse and execute the statement. dbms_sql.parse(c,statement,dbms_sql.native); fdbk := dbms_sql.execute(c); -- Close the open cursor. dbms_sql.close_cursor(c); END create_sequence;
-- encapsulates a PL/SQL block SELECT-INTO. PROCEDURE increment_sequence ( sequence_name IN VARCHAR2 , sequence_value IN OUT NUMBER ) IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); BEGIN /* || || || || || || || || || || || */ Debugging Tip: ============= When you are using a SELECT-INTO-FROM within DBMS_SQL, which is a reserved PLSQL syntax not directly supported by DBMS_SQL. You need to encapsulate it in a PLSQL wrapper. When you use a PLSQL wrapper, the semicolons must be used in the statement and the PLSQL block because DBMS_SQL adds a single semicolon to execute the PLSQL block. If you forget to encapsulate the SQL in a PLSQL wrapper, you will raise the following error message. ------------------------------------------------------------------ORA-01006: bind variable does not exist
-- Build dynamic SQL statement as anonymous block PL/SQL unit. statement := 'BEGIN'||CHR(10) || ' SELECT PLSQL.'||sequence_name||'.nextval'||CHR(10) || ' INTO :retval'||CHR(10) || ' FROM DUAL;'||CHR(10) || 'END;'; -- Parse the statement. dbms_sql.parse(c,statement,dbms_sql.native); /* || || || ||
Technical Note: ============== The BIND_VARIABLE procedure is returning a NUMBER and does not require parameter four.
Page 640
*/ -- Bind variable retval to an output sequence value. dbms_sql.bind_variable(c,'retval',sequence_value); -- Execute the dynamic cursor. fdbk := dbms_sql.execute(c); -- Copy the variable value from the bind variable. dbms_sql.variable_value(c,'retval',sequence_value); dbms_sql.close_cursor(c); dbms_output.put('Sequence <'||sequence_name||'> '); dbms_output.put_line('Value <'||sequence_value||'>'); END increment_sequence;
-- a DML with ordered bind variables PROCEDURE insert_into_table ( table_name IN , table_column_value1 IN , table_column_value2 IN , table_column_value3 IN
VARCHAR2 NUMBER VARCHAR2 VARCHAR2) IS
-- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); BEGIN /* || Debugging Tip: || ============= || Statement strings are terminated by a line return CHR(10) to || ensure that a space is not missing between concatenated segments. || Using a BIND variable provides efficiencies in SQL statements || because it avoids the reparsing of the statement. Therefore, || they should be used as follows for performance gains: || || SQL STATEMENTS PREDICATES || ----------------------|| SELECT WHERE || UPDATE SET || WHERE || DELETE WHERE || || Error Explanations: || -----------------|| 1. An explicit size is always required for a VARCHAR2 variable || and the overloaded procedure has an output size variable in the || fourth position that you may need to use. The output length is || provided below to demonstrate it. || 2. A bad bind variable message typically means the identifier is || outside of the VARCHAR2 string and treated as a session level || undefined bind variable.
Page 641
|| || || || || || || || || || || || */
3. A "missing SELECT keyword" can occur on an insert statement if you put bind variables into the INTO clause for column names. 4. If you have quote marks around VARCHAR2 bind variables, you may raise the "bind variable does not exist" error. If you need to use that syntax, you can encapsulate the DML in a PLSQL wrapper. ------------------------------------------------------------------1. ORA-06502: PL/SQL: numeric or value error 2. PLS-00049: bad bind variable 3. ORA-00928: missing SELECT keyword 4. ORA-01006: bind variable does not exist
-- Build dynamic SQL statement. statement := 'INSERT ' || 'INTO '||table_name||' ' || 'VALUES ' || '( :table_column_value1' || ', :table_column_value2' || ', :table_column_value3)'; -- Parse the statement. dbms_sql.parse(c,statement,dbms_sql.native); -- Bind each bind variable. dbms_sql.bind_variable(c,'table_column_value1',table_column_value1); dbms_sql.bind_variable(c,'table_column_value2',table_column_value2); dbms_sql.bind_variable(c,'table_column_value3',table_column_value3); fdbk := dbms_sql.execute(c); dbms_sql.close_cursor(c); commit; dbms_output.put_line('Value inserted <'||table_column_value1||'>'); dbms_output.put_line('Value inserted <'||table_column_value2||'>'); dbms_output.put_line('Value inserted <'||table_column_value3||'>'); END insert_into_table;
-- a DML with ordered bind variables. PROCEDURE inserts_into_table ( table_name IN VARCHAR2 , table_column_values1 IN DBMS_SQL.NUMBER_TABLE , table_column_values2 IN DBMS_SQL.VARCHAR2_TABLE , table_column_values3 IN DBMS_SQL.VARCHAR2_TABLE) IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); BEGIN -- Build dynamic SQL statement. statement := 'INSERT ' || 'INTO '||table_name||' '
Page 642
|| || || || || || ||
'( card_number ' ', card_name ' ', card_suit)' 'VALUES ' '( :card_number' ', :card_name' ', :card_suit)';
-- Parse the statement. dbms_sql.parse(c,statement,dbms_sql.native); -- Bind each bind variable. dbms_sql.bind_array(c,'card_number',table_column_values1); dbms_sql.bind_array(c,'card_name',table_column_values2); dbms_sql.bind_array(c,'card_suit',table_column_values3); fdbk := dbms_sql.execute(c); dbms_sql.close_cursor(c); commit; -- Use a for-loop to print values. FOR i IN 1..table_column_values1.COUNT LOOP dbms_output.put_line( 'Value inserted <'||table_column_values1(i)||'>'); dbms_output.put_line( 'Value inserted <'||table_column_values2(i)||'>'); dbms_output.put_line( 'Value inserted <'||table_column_values3(i)||'>'); END LOOP; END inserts_into_table;
-- multiple row DQL PROCEDURE multiple_row_return IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); value_out VARCHAR2(1); BEGIN statement := 'SELECT ''A'' FROM DUAL'; dbms_sql.parse(c,statement,dbms_sql.native); -- Define the column mapping to the value_out variable. dbms_sql.define_column(c,1,value_out,1); fdbk := dbms_sql.execute(c); LOOP -- Exit when no more rows to fetch. EXIT WHEN dbms_sql.fetch_rows(c) = 0; -- Copy the contents of column #1 to the value_out variable. dbms_sql.column_value(c,1,value_out); dbms_output.put_line('Value from COLUMN_VALUE <'||value_out||'>'); END LOOP;
Page 643
-- Close the open cursor. dbms_sql.close_cursor(c); END multiple_row_return;
-- multiple row with columns DQL. PROCEDURE multiple_row_return ( table_name VARCHAR2 , column_name1 VARCHAR2 , column_name2 VARCHAR2 , column_name3 VARCHAR2 )IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); cvalue_out1 VARCHAR2(2000); cvalue_out2 VARCHAR2(2000); nvalue_out NUMBER; BEGIN -- Build dynamic SQL statement. statement := 'SELECT ' || column_name1 ||',' || column_name2 ||',' || column_name3 ||' ' || 'FROM '|| table_name; -- Parse dynamic SQL statement. dbms_sql.parse(c,statement,dbms_sql.native); /* || || || || || || || */
Debugging Tip: ============= Define the column values and DO NOT forget to assign a size parameter for a string datatype, like VARCHAR2; however, if you forget, the error message is: ------------------------------------------------------------------PLS-00307: too many declarations of 'DEFINE_COLUMN' match this call
-- Define the column mapping to the value_out variable. dbms_sql.define_column(c,1,nvalue_out); dbms_sql.define_column(c,2,cvalue_out1,2000); dbms_sql.define_column(c,3,cvalue_out2,2000); -- Execute dynamic SQL statement. fdbk := dbms_sql.execute(c); -- Use a loop to read all rows. LOOP -- Exit when no more rows to fetch. EXIT WHEN dbms_sql.fetch_rows(c) = 0; -- Copy the contents of column #1 to the value_out variable. dbms_sql.column_value(c,1,nvalue_out);
Page 644
dbms_sql.column_value(c,2,cvalue_out1); dbms_sql.column_value(c,3,cvalue_out2); dbms_output.put_line( 'Value from ['||column_name1||'] '|| 'is: ['||nvalue_out||']'); dbms_output.put_line( 'Value from ['||column_name1||'] '|| 'is: ['||SUBSTR(cvalue_out1,1,5)||']'); dbms_output.put_line( 'Value from ['||column_name1||'] '|| 'is: ['||SUBSTR(cvalue_out2,1,8)||']'); END LOOP; dbms_sql.close_cursor(c); END multiple_row_return;
-- single row DQL. /* || Demonstrate a single row return using the DEFINE_COLUMN and COLUMN_VALUE || program unit, as you would in an explicit cursor. */ PROCEDURE single_row_return IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); value_out VARCHAR2(1); BEGIN -- Build dynamic SQL statement. statement := 'SELECT ''A'' FROM DUAL'; -- Parse the dynamic SQL statement. dbms_sql.parse(c,statement,dbms_sql.native); /* || || || || || || || || || || || || */
Debugging Tip: ============= Define the column values and DO NOT forget to assign a size parameter for a string datatype, like VARCHAR2; however, if you forget, the error message is: ------------------------------------------------------------------PLS-00307: too many declarations of 'DEFINE_COLUMN' match this call This is the message returned because the DEFINE_COLUMN procedure is overloaded and it doesn't know how to implicitly cast without the OUT_VALUE_SIZE argument. Only CHAR, RAW and VARCHAR2 support a fourth argument.
-- Define the column mapping to the value_out variable. dbms_sql.define_column(c,1,value_out,1); fdbk := dbms_sql.execute_and_fetch(c);
Page 645
-- Copy the contents of column #1 to the value_out variable. dbms_sql.column_value(c,1,value_out); dbms_output.put_line( 'Value from COLUMN_VALUE <'||value_out||'>'); -- Close the open cursor. dbms_sql.close_cursor(c); END single_row_return;
-- single row DQL. PROCEDURE single_row_return ( table_name VARCHAR2 , column_name1 VARCHAR2 , column_name2 VARCHAR2 , column_name3 VARCHAR2 ) IS -- Define local DBMS_SQL variables. c INTEGER := dbms_sql.open_cursor; fdbk INTEGER; statement VARCHAR2(2000); cvalue_out1 VARCHAR2(20); cvalue_out2 VARCHAR2(30); nvalue_out NUMBER; BEGIN -- Build dynamic SQL statement. statement := 'SELECT ' || column_name1 ||',' || column_name2 ||',' || column_name3 ||' ' || 'FROM '|| table_name; -- Parse the dynamic SQL statement. dbms_sql.parse(c,statement,dbms_sql.native); /* || || || || || || || || || || || || */
Debugging Tip: ============= Define the column values and DO NOT forget to assign a size parameter for a string datatype, like VARCHAR2; however, if you forget, the error message is: ------------------------------------------------------------------PLS-00307: too many declarations of 'DEFINE_COLUMN' match this call This is the message returned because the DEFINE_COLUMN procedure is overloaded and it doesn't know how to implicitly cast without the OUT_VALUE_SIZE argument. Only CHAR, RAW and VARCHAR2 support a fourth argument.
-- Define the column mapping to the value_out variable.
Page 646
dbms_sql.define_column(c,1,nvalue_out); dbms_sql.define_column(c,2,cvalue_out1,20); dbms_sql.define_column(c,3,cvalue_out2,30); -- Execute dynamic SQL statement. fdbk := dbms_sql.execute_and_fetch(c); -- Copy the contents of column #1 to the value_out variable. dbms_sql.column_value(c,1,nvalue_out); dbms_sql.column_value(c,2,cvalue_out1); dbms_sql.column_value(c,3,cvalue_out2); -- Print output message. dbms_output.put_line('Value from COLUMN_VALUE <'||nvalue_out||'>'); dbms_output.put_line('Value from COLUMN_VALUE <'||cvalue_out1||'>'); dbms_output.put_line('Value from COLUMN_VALUE <'||cvalue_out2||'>'); dbms_sql.close_cursor(c); END single_row_return;
Page 647
Calling Java from PL/SQL

Pre-requisits: o set CLASSPATH should contain: . current directory Oracle 11g: %ORACLE_HOME%/jdbc/lib/ojdbc5.jar (or 6 but not both) ORACLE_HOME/jlib/orai18n.jar Oracle 10g: %ORACLE_HOME%/jdbc/lib/classes12.zip Java JDK or J2SE 5.n or 6.n In Oracle 11g, the oracle.jdbc.driver.* classes, the ojdbc4.jar file, and the OracleConnectionCacheImpl class are no longer supported or available. o Make sure PATH points to JDK home.
/* to create a Thick Java client program to Oracle 11g */ -- JDBCExample.java import java.sql.*; import oracle.jdbc.pool.OracleDataSource; public class JDBCExample { public static void main(String args[]) throws SQLException /* Declare the type of Oracle Driver you are using */ { DriverManager.registerDriver(new oracle.jdbc.driver.OracleDriver()); /* Create a database connection for the JDBC program */ Connection conn= DriverManager.getConnection("jdbc:oracle:thin:@srv01:1521:ora11g","HR","h"); Statement stmt = conn.createStatement(); /* Pass a query to SQL and store the results in the result set rs */ ResultSet rs = stmt.executeQuery("select employee_id, last_name from employees"); /* Using the while loop, result set rs is accessed row by row */ while(rs.next()){ int number = rs.getInt(1); String name= rs.getString(2); System.out.println(number+" "+name); } /* Close the JDBC result set and close the database connection */ rs.close(); conn.close(); } }
-- load that into the database : loadjava -r -f -o -user HR/h JDBCExample.class
Page 648
Configuring Oracle Database to Use External Routines

o For further details refer to the documentation or Oracle Database 10g PL/SQL Programming by Ron Hardman, Michael McLaughlin and Scott Urman, Oracle Press.
Configure one listener for the database and one for the the extproc agent #(1) In listener.ora # remove the IPC protocol from the standard settings LISTENER = (DESCRIPTION_LIST = (DESCRIPTION = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = TCP) (HOST = srv01) (PORT = 1521) ) ) ) ) SID_LIST_LISTENER = (SID_LIST = (SID_DESC = (SID_NAME = ora11g) (ORACLE_HOME = E:\oracle\OraDB11g) ) )
# another listener. "extproc" lowercase CALLOUT_LISTENER = (DESCRIPTION_LIST = (DESCRIPTION = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = IPC) (KEY = extproc) ) ) ) )
# "PLSExtProc" case sensitive # (ENV = "EXTPROC_DLLS=ONLY:<custom_dll_directory>/<custom_shared_library>,LD_LIBRARY_P ATH=E:\oracle\OraDB11g\LIB") SID_LIST_CALLOUT_LISTENER = (SID_LIST = (SID_DESC = (SID_NAME = PLSExtProc) (ORACLE_HOME = E:\oracle\OraDB11g) (PROGRAM = extproc) (ENV = "EXTPROC_DLLS=ONLY:C:\myfiles\dll/C:\myfiles\lib,LD_LIBRARY_PATH=E:\oracle\Ora
Page 649
DB11g\LIB") ) ) # if there is an ASM instance (SID_DESC = (GLOBAL_DBNAME = ora11g.srv01) (ORACLE_HOME=E:\oracle\OraDB11g\database) (SID_NAME = +ASM) )
#(2) Add the following in to the tnsnames.ora EXTPROC_CONNECTION_DATA = (DESCRIPTION = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = IPC) (KEY = extproc)) ) (CONNECT_DATA = (SID = PLSExtProc) (PRESENTATION = RO) ) )
#(3) rebuild the original listener service and create new one # In Windows lsnrctl stop # use NETCA to delete the original service and then re-create it. # build a new service for the second listener CALLOUT_LISTENER. A new linstener.ora will # be created so paste again the code above in the file. # In Unix lsnrctl stop LISTENER # backup the original linstener.ora and tnsnames files # copy the new ones lsnrctl start LISTENER lsnrctl start CALLOUT_LISTENER # verify ps ef | grep v grep | grep tnslsnr #(4) verify the configuration tnsping EXTPROC_CONNECTION_DATA If you get a TNS-12541 error when using tnsping, the likelihood is that there is a mismatch between the ADDRESS parameter values in the listener.ora and tnsnames.ora files. # the following command should return the following error: sqlplus plsql/plsql@EXTPROC_CONNECTION_DATA ORA-28547: connection to server failed, probable Oracle Net admin error
Working with a C Shared Library Define a C shared library:

Page 650
#include <stdio.h> /* Declare a writestr function. */ void writestr1(char *path, char *message) { /* Declare a FILE variable. */ FILE *file_name; /* Open the File. */ file_name = fopen(path,"w"); /* Write to file the message received. */ fprintf(file_name,"%s\n",message); /* Close the file. */ fclose(file_name); }
Unix C Compiler that supports the G option cc G o writestr1.so writestr1.c Unix C Compiler that supports the shared option cc shared o writestr1.so writestr1.c - OR gcc shared o writestr1.so writestr1.c
# Defining and Calling the PL/SQL Library Wrapper CREATE OR REPLACE LIBRARY library_write_string AS '<oracle_home_directory>/<custom_library>/<file_name>.<file_ext>'; / CREATE OR REPLACE PROCEDURE write_string (path VARCHAR2 ,message VARCHAR2) AS EXTERNAL LIBRARY library_write_string NAME "writestr" PARAMETERS (path STRING ,message STRING); / -- Available online as part of create_library1.sql BEGIN -- Call the external procedure. write_string('/tmp/file.txt','Hello World!'); END; /
Page 651
Using Large Objects (LOBs)

Creating LOB
To specify different storage options, use the following syntax in your create table statement: CREATE TABLE table_name ( .. lob_column lob_datatype ) LOB (lob_column,..) STORE AS [lob_segment_name] [(TABLESPACE tablespace {ENABLE|DISABLE} STORAGE IN ROW STORAGE storage_clause CHUNK int PCTVERSION int CACHE CACHE READS [[NO]LOGGING] NOCACHE [[NO]LOGGING]) PCTVERSION defaults to 10 NOCACHE is the default ENABLE STORAGE IN ROW, which is the default setting, says that if the LOB value is less than 4K (including control information), store it inline. CREATE TABLE book_samples ( book_sample_id NUMBER (10) PRIMARY KEY, isbn CHAR(10 CHAR), description CLOB, nls_description NCLOB, book_cover BLOB, chapter_title VARCHAR2(30 CHAR), chapter BFILE ) LOB (book_cover) STORE AS blob_seg ( TABLESPACE blob_ts CHUNK 8192 PCTVERSION 0 NOCACHE NOLOGGING DISABLE STORAGE IN ROW) LOB (description, nls_description) STORE AS ( TABLESPACE clob_ts CHUNK 8192 PCTVERSION 10 NOCACHE LOGGING ENABLE STORAGE IN ROW);
Using SQL with Internal LOBs

With SQL, you can insert, update, and delete internal LOBs. SQL does not yet support piecewise manipulation of LOBs. SET LONG 64000
Page 652
SELECT description FROM book_samples; -- empty_clob() creates a locator INSERT INTO book_samples ( book_sample_id, isbn, description, nls_description, book_cover, chapter) VALUES ( 1, '72230665', 'The ...', EMPTY_CLOB(), EMPTY_BLOB(), BFILENAME('BOOK_SAMPLES_LOC', '72230665.jpg')); UPDATE book_samples SET description = EMPTY_CLOB() WHERE description IS NOT NULL;
Using LOBs in PL/SQL

A opened BFILE should be closed after handling it. SESSION_MAX_OPEN_FILES controls the number of BFILEs that can be open at any one time. /* APPEND */ CREATE OR REPLACE PROCEDURE LOBAPPEND ( io_lob_source IN OUT NCLOB, io_lob_destination IN OUT NCLOB) AS BEGIN DBMS_LOB.OPEN(io_lob_source, DBMS_LOB.LOB_READONLY); DBMS_LOB.OPEN(io_lob_destination, DBMS_LOB.LOB_READWRITE); DBMS_LOB.APPEND(io_lob_destination, io_lob_source); -- always close lob after handling them DBMS_LOB.CLOSE(io_lob_source); DBMS_LOB.CLOSE(io_lob_destination); END; / SET SERVEROUTPUT ON DECLARE v_source_lob_loc NCLOB; v_destination_lob_loc NCLOB; v_combined_lob NCLOB; BEGIN -- for update is a must here SELECT nls_description INTO v_source_lob_loc FROM book_samples_nls FOR UPDATE;
Page 653
SELECT nls_description INTO v_destination_lob_loc FROM book_samples FOR UPDATE; LOBAPPEND(v_source_lob_loc, v_destination_lob_loc); SELECT nls_description INTO v_combined_lob from book_samples; DBMS_OUTPUT.PUT_LINE(SUBSTR(v_combined_lob, 1, 150)); DBMS_OUTPUT.PUT_LINE(SUBSTR(v_combined_lob, 151, 300)); END; /
/* COMPARE */ If 0 is returned, then the LOBs are the same. If 1 is returned, then they are different. FUNCTION COMPARE RETURNS NUMBER(38) Argument Name Type In/Out Default? - LOB_1 CLOB IN LOB_2 CLOB IN AMOUNT NUMBER(38) IN DEFAULT OFFSET_1 NUMBER(38) IN DEFAULT OFFSET_2 NUMBER(38) IN DEFAULT CREATE OR REPLACE PROCEDURE CLOB_COMPARE ( v_lob1 IN OUT CLOB, v_lob2 IN OUT CLOB) AS v_compare PLS_INTEGER := 0; BEGIN DBMS_LOB.OPEN(v_lob1, DBMS_LOB.LOB_READONLY); DBMS_LOB.OPEN(v_lob2, DBMS_LOB.LOB_READONLY); -- buffer size used 32K (the max) v_compare := DBMS_LOB.COMPARE(v_lob1,v_lob2, 32768, 1, 1); DBMS_OUTPUT.PUT_LINE('The value returned by COMPARE is: '||v_compare); DBMS_LOB.CLOSE(v_lob1); DBMS_LOB.CLOSE(v_lob2); END; / DECLARE v_lob1 CLOB; v_lob2 CLOB; v_lob3 CLOB; BEGIN SELECT description INTO v_lob1 FROM book_samples WHERE book_sample_id = 1;
Page 654
SELECT description INTO v_lob2 FROM book_samples WHERE book_sample_id = 2; SELECT description INTO v_lob3 FROM book_samples WHERE book_sample_id = 3; CLOB_COMPARE(v_lob1, v_lob2); CLOB_COMPARE(v_lob1, v_lob3); END; /
/* CONVERTTO..LOB */ PROCEDURE CONVERTTOBLOB DEST_LOB BLOB IN/OUT # in CONVERTTOCLOB the following is SRC_BLOB SRC_CLOB CLOB IN AMOUNT NUMBER(38) IN DEST_OFFSET NUMBER(38) IN/OUT SRC_OFFSET NUMBER(38) IN/OUT BLOB_CSID NUMBER IN LANG_CONTEXT NUMBER(38) IN/OUT WARNING NUMBER(38) OUT CREATE OR REPLACE PROCEDURE CONVERT_ME ( v_blob_or_clob IN NUMBER, -- if 0 TOBLOB, if 1 TOCLOB v_blob IN OUT BLOB, v_clob IN OUT CLOB, v_amount IN OUT NUMBER, v_blob_offset IN OUT NUMBER, v_clob_offset IN OUT NUMBER, v_lang_context IN OUT NUMBER, v_warning OUT NUMBER) AS BEGIN DBMS_LOB.OPEN(v_blob, DBMS_LOB.LOB_READWRITE); DBMS_LOB.OPEN(v_clob, DBMS_LOB.LOB_READWRITE); IF v_blob_or_clob = 0 THEN DBMS_LOB.CONVERTTOBLOB(v_blob, v_clob, v_amount, v_blob_offset, v_clob_offset, 1, v_lang_context, v_warning); ELSE
Page 655
DBMS_LOB.CONVERTTOCLOB(v_clob, v_blob, v_amount, v_clob_offset, v_blob_offset, 1, v_lang_context, v_warning); END IF; DBMS_LOB.CLOSE(v_blob); DBMS_LOB.CLOSE(v_clob); END; /
DECLARE v_clob_or_blob NUMBER; v_blob_locator BLOB; v_clob_locator CLOB; v_blob_offset NUMBER; v_clob_offset NUMBER; v_lang_context NUMBER := DBMS_LOB.DEFAULT_LANG_CTX; v_warning NUMBER; v_string_length NUMBER(10); v_source_locator BLOB; v_destination_locator BLOB; v_amount PLS_INTEGER; v_string CLOB; BEGIN -- CONVERT CLOB TO BLOB SELECT description INTO v_clob_locator FROM book_samples WHERE book_sample_id = 1 FOR UPDATE; SELECT misc INTO v_blob_locator FROM book_samples WHERE book_sample_id = 1 FOR UPDATE; v_string_length := DBMS_LOB.GETLENGTH(v_blob_locator); v_amount := DBMS_LOB.GETLENGTH(v_clob_locator); DBMS_OUTPUT.PUT_LINE('The initial length of the BLOB is: '||v_string_length); v_clob_or_blob := 0; -- Convert clob to blob v_clob_offset := 1; v_blob_offset := 1; CONVERT_ME(v_clob_or_blob,
Page 656
v_blob_locator, v_clob_locator, v_amount, v_blob_offset, v_clob_offset, v_lang_context, v_warning); v_string_length := DBMS_LOB.GETLENGTH(v_blob_locator); DBMS_OUTPUT.PUT_LINE('The length of the BLOB post-conversion is: '||v_string_length); -- COPY BLOB FOR ONE ROW TO BLOB IN ANOTHER v_source_locator := v_blob_locator; SELECT misc INTO v_destination_locator FROM book_samples WHERE book_sample_id = 2 FOR UPDATE; DBMS_LOB.COPY(v_destination_locator, v_source_locator, 32768, 1, 1); v_string_length := DBMS_LOB.GETLENGTH(v_destination_locator); DBMS_OUTPUT.PUT_LINE('The length of the BLOB post-copy is: '||v_string_length); -- COPY BLOB FOR RECORD 2 BACK TO A CLOB SELECT description INTO v_clob_locator FROM book_samples WHERE book_sample_id = 2 FOR UPDATE; SELECT misc INTO v_blob_locator FROM book_samples WHERE book_sample_id = 2 FOR UPDATE; v_string_length := DBMS_LOB.GETLENGTH(v_clob_locator); -- v_ammount must equal to BLOB size, otherwise ORA-22993 will return v_amount := DBMS_LOB.GETLENGTH(v_blob_locator); DBMS_OUTPUT.PUT_LINE('The initial length of the CLOB (record 2) is: '||v_string_length); v_clob_or_blob := 1; -- Convert blob to clob -- must be reset to 1 because its value changed by the -- previous call of CONVERT_ME v_clob_offset := 1; v_blob_offset := 1; CONVERT_ME(v_clob_or_blob, v_blob_locator,
Page 657
v_clob_locator, v_amount, v_clob_offset, v_blob_offset, v_lang_context, v_warning); v_string_length := DBMS_LOB.GETLENGTH(v_clob_locator); SELECT description INTO v_string FROM book_samples WHERE book_sample_id = 2; DBMS_OUTPUT.PUT_LINE('The length of the CLOB post-conversion is: '||v_string_length); DBMS_OUTPUT.PUT_LINE('The converted CLOB'); DBMS_OUTPUT.PUT_LINE('=================='); DBMS_OUTPUT.PUT_LINE(SUBSTR(v_string,1,150)); DBMS_OUTPUT.PUT_LINE(SUBSTR(v_string,151,300)); END; /
/* BFILE FILEEXISTS */ -- This function tests whether a file exists by the name specified in the insert statement INSERT INTO book_samples ( book_sample_id, isbn, description, nls_description, misc, bfile_description) VALUES ( 1, '72230665', EMPTY_CLOB(), EMPTY_CLOB(), EMPTY_BLOB(), BFILENAME('BOOK_SAMPLES_LOC', 'bfile_example.pdf')); CREATE OR REPLACE PROCEDURE CHECK_FILE ( v_bfile IN BFILE) AS v_exists PLS_INTEGER := 0; BEGIN v_exists := DBMS_LOB.FILEEXISTS(v_bfile); IF v_exists = 0 THEN DBMS_OUTPUT.PUT_LINE ('The file does not exists in the directory specified'); ELSE DBMS_OUTPUT.PUT_LINE ('The file exists and the directory valid!');
Page 658
END IF; END; / DECLARE v_bfile BFILE; BEGIN SELECT bfile_description INTO v_bfile FROM book_samples WHERE book_sample_id = 1; CHECK_FILE(v_bfile); END; /
/* BFILE FILEOPEN/OPEN */ -- Oracle recommends that OPEN be used instead of FILEOPEN. PROCEDURE OPEN Argument Name Type In/Out Default? - FILE_LOC BINARY FILE LOB IN/OUT OPEN_MODE BINARY_INTEGER IN DEFAULT OPEN_MODE: DBMS_LOB.LOB_READONLY or DBMS_LOB.LOB_READWRITE
/* BFILE FILEISOPEN/ISOPEN */ -- ISOPEN should be used in place of FILEISOPEN when possible. CREATE OR REPLACE PROCEDURE CHECK_STATUS ( v_bfile IN BFILE) AS v_isopen PLS_INTEGER := 0; BEGIN v_isopen := DBMS_LOB.ISOPEN(v_bfile); IF v_isopen = 0 THEN DBMS_OUTPUT.PUT_LINE ('The file is not open. You must open the'); ELSE DBMS_OUTPUT.PUT_LINE ('The file is open already.'); END IF; END; / DECLARE v_bfile BFILE; BEGIN SELECT bfile_description INTO v_bfile FROM book_samples WHERE book_sample_id = 1; CHECK_STATUS(v_bfile); END; /
Page 659
/* BFILE FILECLOSE/CLOSE/FILECLOSEALL */ FILECLOSE and CLOSE both close one BFILE at a time, while FILECLOSEALL closes all open BFILEs. It is recommended by Oracle that CLOSE be used rather than FILECLOSE for all new development. CLOSE can be used with all LOB types, not just BFILEs. CREATE OR REPLACE PROCEDURE CLOSE_ALL_FILES AS v_isopen PLS_INTEGER := 0; v_counter PLS_INTEGER := 0; v_bfile BFILE; CURSOR cur_bfile IS SELECT bfile_description FROM book_samples; BEGIN DBMS_OUTPUT.PUT_LINE('Open all BFILEs in the table'); OPEN cur_bfile; LOOP FETCH cur_bfile INTO v_bfile; EXIT WHEN cur_bfile%NOTFOUND; BEGIN v_counter := v_counter + 1; DBMS_LOB.OPEN(v_bfile); v_isopen := DBMS_LOB.ISOPEN(v_bfile); IF v_isopen = 0 THEN DBMS_OUTPUT.PUT_LINE ('File number '||v_counter||' is closed'); ELSE DBMS_OUTPUT.PUT_LINE ('File number '||v_counter||' is open'); END IF; END; END LOOP; CLOSE cur_bfile; DBMS_LOB.FILECLOSEALL(); DBMS_OUTPUT.PUT_LINE(' END; /
DONE
');
/* LOADFROMFILE/LOADCLOBFROMFILE/LOADBLOBFROMFILE */ load file contents to CLOB and BLOB columns. It is recommended that LOADCLOBFROMFILE and LOADBLOBFROMFILE be used for their specific datatypes rather than using the generic overloaded LOADFROMFILE. PROCEDURE LOADBLOBFROMFILE Argument Name -----------------------------DEST_LOB SRC_BFILE AMOUNT
Type ----------------------BLOB BINARY FILE LOB NUMBER(38)
In/Out Default? ------ -------IN/OUT IN IN
Page 660
DEST_OFFSET SRC_OFFSET PROCEDURE LOADCLOBFROMFILE Argument Name -----------------------------DEST_LOB SRC_BFILE AMOUNT DEST_OFFSET SRC_OFFSET BFILE_CSID LANG_CONTEXT WARNING set serveroutput on
NUMBER(38) NUMBER(38) Type ----------------------CLOB BINARY FILE LOB NUMBER(38) NUMBER(38) NUMBER(38) NUMBER NUMBER(38) NUMBER(38)
IN/OUT IN/OUT In/Out Default? ------ -------IN/OUT IN IN IN/OUT IN/OUT IN IN/OUT OUT
DECLARE v_dest_blob BLOB; v_dest_clob CLOB; v_source_locator1 BFILE := BFILENAME('BOOK_SAMPLES_LOC', 'bfile_example.pdf'); v_source_locator2 BFILE := BFILENAME('BOOK_SAMPLES_LOC', 'bfile_example.txt'); v_source_offset NUMBER := 1; v_dest_offset NUMBER := 1; v_lang_context NUMBER := DBMS_LOB.DEFAULT_LANG_CTX; v_warning PLS_INTEGER; BEGIN -- Empty the description and misc columns UPDATE book_samples SET description = EMPTY_CLOB(), misc = EMPTY_BLOB() WHERE book_sample_id = 1; -- Retrieve the locators for the two destination columns SELECT description, misc INTO v_dest_clob, v_dest_blob FROM book_samples WHERE book_sample_id = 1 FOR UPDATE; -- Open the BFILEs and destination LOBs DBMS_LOB.OPEN(v_source_locator1, DBMS_LOB.LOB_READONLY); DBMS_LOB.OPEN(v_source_locator2, DBMS_LOB.LOB_READONLY); DBMS_LOB.OPEN(v_dest_blob, DBMS_LOB.LOB_READWRITE); DBMS_LOB.OPEN(v_dest_clob, DBMS_LOB.LOB_READWRITE); DBMS_OUTPUT.PUT_LINE('Length of the BLOB file is: '||DBMS_LOB.GETLENGTH(v_source_locator1)); DBMS_OUTPUT.PUT_LINE('Length of the CLOB file is: '||DBMS_LOB.GETLENGTH(v_source_locator2)); DBMS_OUTPUT.PUT_LINE('Size of BLOB pre-load: '||DBMS_LOB.GETLENGTH(v_dest_blob)); DBMS_OUTPUT.PUT_LINE('Size of CLOB pre-load: '||DBMS_LOB.GETLENGTH(v_dest_clob)); -- Load the destination columns from the source DBMS_LOB.LOADBLOBFROMFILE(v_dest_blob, v_source_locator1, DBMS_LOB.LOBMAXSIZE, v_dest_offset, v_source_offset);
Page 661
DBMS_OUTPUT.PUT_LINE('Size of BLOB post-load: '||(v_dest_offset -1)); v_dest_offset := 1; v_source_offset := 1; DBMS_LOB.LOADCLOBFROMFILE(v_dest_clob, v_source_locator2, DBMS_LOB.LOBMAXSIZE, v_dest_offset, v_source_offset, DBMS_LOB.DEFAULT_CSID, v_lang_context, v_warning); DBMS_OUTPUT.PUT_LINE('Size of CLOB post-load: '||(v_dest_offset -1)); -- Close the LOBs that we opened DBMS_LOB.CLOSE(v_source_locator1); DBMS_LOB.CLOSE(v_source_locator2); DBMS_LOB.CLOSE(v_dest_blob); DBMS_LOB.CLOSE(v_dest_clob); EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE(SQLERRM); DBMS_LOB.CLOSE(v_source_locator1); DBMS_LOB.CLOSE(v_source_locator2); DBMS_LOB.CLOSE(v_dest_blob); DBMS_LOB.CLOSE(v_dest_clob); END; / SET LONG 64000 SELECT description FROM book_samples WHERE book_sample_id = 1;
Performance Considerations
Using Returning Clause RETURNING is a keyword added to the end of the INSERT statement allowing you to work with the LOB immediately, without any additional steps. DECLARE v_clob CLOB; BEGIN INSERT INTO book_samples ( book_sample_id, isbn, description, nls_description, book_cover, chapter) VALUES (
Page 662
1, '72230665', 'The ...', EMPTY_CLOB(), EMPTY_BLOB(), BFILENAME('BOOK_SAMPLES_LOC', '72230665.jpg')) RETURNING description INTO v_clob; COMMIT; DBMS_OUTPUT.PUT_LINE(v_clob); END; / Using CONTEXT Index You can apply many indexing types. The examples here show how to use CONTEXT index. Using the CONTEXT index, we are able to perform the following types of queries: o o o o o o Boolean searches AND, OR, NOT. Exact matches Search for the exact word or phrase inside the text. Inexact matches Search using stemming (a search for mice finds mouse), wildcard, soundex (one word sounds like another). Proximity A word is near another. Ranking A value is provided based on relevance to the keywords used in the query. Theme searches Search on what a document or text is about.
Refer to Oracle Text documentation for further details.

/* Creating CONTEXT Index */ -- 1) create named preferences -- If you want to index text in another language: eg JAPANESE_LEXER and JAPANESE_VGRAM_LEXER -- WORLD_LEXER that can detect the language of the text BEGIN ctx_ddl.create_preference ('lob_lexer', 'basic_lexer'); ctx_ddl.set_attribute ('lob_lexer', 'index_text', 'true'); ctx_ddl.set_attribute ('lob_lexer', 'index_themes', 'false'); END; / -- 2) create Wordlist BEGIN ctx_ddl.create_preference ('lob_wordlist', 'basic_wordlist'); ctx_ddl.set_attribute ('lob_wordlist', 'substring_index', 'true'); END; / -- 3) Create the index on the column -- role ctxapp and resource must be granted first -- STOPLIST refers to noise words. It can also take DEFAULT_STOPLIST CREATE INDEX lob_indx ON book_samples(description) INDEXTYPE IS CTXSYS.CONTEXT PARAMETERS ( 'lexer lob_lexer wordlist lob_wordlist stoplist ctxsys.empty_stoplist' ) /
To examine the tokens:

Page 663
set pages 9999 SELECT token_text FROM DR$LOB_INDX$I;
/* Using the Index */ SELECT SCORE(1), book_sample_id FROM book_samples WHERE CONTAINS(description, 'website', 1) > 0;
Migrating from LONGs to LOBs

ALTER TABLE long_to_lob MODIFY text CLOB;
Page 664
PL/SQL Performance Tuning Tips

Use PL/SQL Profiler
It is DBMS_HPROF in 11g and DBMS_PROFILER in 10g. Provides performance statistical information on an executed procedure. /* Required Privs */ GRANT EXECUTE ON dbms_hprof TO sa; CREATE OR REPLACE DIRECTORY profiler_dir AS '/home/oracle/temp'; GRANT READ, WRITE ON DIRECTORY profiler_dir TO sa;
/* Package Tables */ conn sa/s @?/rdbms/admin/dbmshptab.sql
/* Using the package */ BEGIN DBMS_HPROF.start_profiling ( location => 'PROFILER_DIR', filename => 'profiler.txt'); -- call the procedure to profile PROC3; DBMS_HPROF.stop_profiling; END; /
-- run the ANALYZE function to analyse the -- raw data and fill the Profiler tables SET SERVEROUTPUT ON DECLARE l_runid NUMBER; BEGIN l_runid := DBMS_HPROF.analyze ( location => 'PROFILER_DIR', filename => 'profiler.txt', run_comment => 'Test run.'); DBMS_OUTPUT.put_line('l_runid=' || l_runid); END; / l_runid=1
-- check the info: SELECT runid, run_timestamp, total_elapsed_time, run_comment FROM dbmshp_runs WHERE runid=1;
Page 665
-- to list the profiling info: SELECT symbolid, owner, module, type, function FROM dbmshp_function_info WHERE runid = 1 ORDER BY symbolid;
-- to list the profiling info in hierarchy: SELECT RPAD(' ', level*2, ' ') || fi.owner || '.' || fi.module AS name, fi.function, pci.subtree_elapsed_time, pci.function_elapsed_time, pci.calls FROM dbmshp_parent_child_info pci JOIN dbmshp_function_info fi ON pci.runid = fi.runid AND pci.childsymid = fi.symbolid WHERE pci.runid = 1 CONNECT BY PRIOR childsymid = parentsymid START WITH pci.parentsymid = 3;
/* Using plshprof Utility */ -- profiler.txt generated by the package plshprof -output plshprof_out profiler.txt
Use BULK COLLECT Clause

More efficient than traditional loop through curoser records /* loading ALL the returned recordset into one PL/SQL tables */ -- if the returned rows are more than 200000, make the rows -- be returned into batches using LIMIT clause DECLARE -- Define collection type and variables to be used by the -- BULK COLLECT clause TYPE student_id_type IS TABLE OF student.student_id%TYPE; TYPE first_name_type IS TABLE OF student.first_name%TYPE; TYPE last_name_type IS TABLE OF student.last_name%TYPE; student_id_tab student_id_type; first_name_tab first_name_type; last_name_tab last_name_type; BEGIN -- Fetch all student data at once via BULK COLLECT clause SELECT student_id, first_name, last_name BULK COLLECT INTO student_id_tab, first_name_tab, last_name_tab FROM student; FOR i IN student_id_tab.FIRST..student_id_tab.LAST LOOP DBMS_OUTPUT.PUT_LINE ('student_id: '||student_id_tab(i)); DBMS_OUTPUT.PUT_LINE ('first_name: '||first_name_tab(i)); DBMS_OUTPUT.PUT_LINE ('last_name: '||last_name_tab(i)); END LOOP;
Page 666
END;
/* Using the limit Option */ -- Note: limit option cannot be used directly with SELECT statemnet. Only on -- cursors -- batches are good for better performance when -- the returned rowset is very large ( roughly >500000) DECLARE -- to be used by the LIMIT (batch size) v_limit PLS_INTEGER := 50; CURSOR student_cur IS SELECT student_id, first_name, last_name FROM student; -- Define collection type and variables to be used by the -- BULK COLLECT clause TYPE student_id_type IS TABLE OF student.student_id%TYPE; TYPE first_name_type IS TABLE OF student.first_name%TYPE; TYPE last_name_type IS TABLE OF student.last_name%TYPE; student_id_tab student_id_type; first_name_tab first_name_type; last_name_tab last_name_type; BEGIN OPEN student_cur; LOOP -- Fetch 50 rows at once FETCH student_cur BULK COLLECT INTO student_id_tab, first_name_tab,last_name_tab LIMIT v_limit; -- exit is based on the the number of records in the collection EXIT WHEN student_id_tab.COUNT = 0; -- it is wrong to put it after the outer loop FOR i IN student_id_tab.FIRST..student_id_tab.LAST LOOP DBMS_OUTPUT.PUT_LINE ('student_id: '||student_id_tab(i)); DBMS_OUTPUT.PUT_LINE ('first_name: '||first_name_tab(i)); DBMS_OUTPUT.PUT_LINE ('last_name: '||last_name_tab(i)); END LOOP; END LOOP; CLOSE student_cur; END;
/* Loading into PL/SQL Multiple-Column Table */ -- good when there are so many cols to read -- Example 1 DECLARE CURSOR student_cur IS SELECT student_id, first_name, last_name FROM student;
Page 667
-- Define record type TYPE student_rec IS RECORD (student_id student.student_id%TYPE, first_name student.first_name%TYPE, last_name student.last_name%TYPE); -- Define collection type TYPE student_type IS TABLE OF student_rec; -- Define collection variable student_tab student_type; -- to be used by the LIMIT clause v_limit PLS_INTEGER := 50; BEGIN OPEN student_cur; LOOP -- Fetch 50 rows at once FETCH student_cur BULK COLLECT INTO student_tab LIMIT v_limit; EXIT WHEN student_tab.COUNT = 0; FOR i IN student_tab.FIRST..student_tab.LAST LOOP DBMS_OUTPUT.PUT_LINE('student_id: '||student_tab(i).student_id); DBMS_OUTPUT.PUT_LINE('first_name: '|| student_tab(i).first_name); DBMS_OUTPUT.PUT_LINE('last_name: '|| student_tab(i).last_name); END LOOP; END LOOP; CLOSE student_cur; END;
-- Example 2: DECLARE V_LIMIT PLS_INTEGER :=50; TYPE ORDERS_TYPE IS TABLE OF ORDERS%ROWTYPE; ORDERS_TB ORDERS_TYPE; CURSOR ORDERS_CRS IS SELECT * FROM ORDERS; BEGIN OPEN ORDERS_CRS; LOOP -- load 500 at once FETCH ORDERS_CRS BULK COLLECT INTO ORDERS_TB LIMIT V_LIMIT; -- exit when the collection is empty EXIT WHEN ORDERS_TB.COUNT=0; -- process the batch FOR I IN ORDERS_TB.FIRST .. ORDERS_TB.LAST LOOP NULL; END LOOP; END LOOP; CLOSE ORDERS_CRS; END; /
Page 668
/* Using BULK COLLECT with the DELETE */ DECLARE -- Define collection types and variables TYPE row_num_type IS TABLE OF NUMBER INDEX BY PLS_INTEGER; TYPE row_text_type IS TABLE OF VARCHAR2(10) INDEX BY PLS_INTEGER; row_num_tab row_num_type; row_text_tab row_text_type; BEGIN DELETE FROM TEST RETURNING row_num, row_text BULK COLLECT INTO row_num_tab, row_text_tab; DBMS_OUTPUT.PUT_LINE ('Deleted '||SQL%ROWCOUNT ||' rows:'); FOR i IN row_num_tab.FIRST..row_num_tab.LAST LOOP DBMS_OUTPUT.PUT_LINE ('row_num = '||row_num_tab(i)|| ' row_text = ' ||row_text_tab(i)); END LOOP; COMMIT; END;
Output may look like: Deleted 4 rows: row_num = 3 row_text = row_num = 4 row_text = row_num = 6 row_text = row_num = 8 row_text =
row row row row
3 4 6 8
/* Using BULK COLLECT with FORALL */ -- Example 1: with Insert DECLARE -- Declare collection types TYPE string_type IS TABLE OF VARCHAR2(100) INDEX BY PLS_INTEGER; TYPE date_type IS TABLE OF DATE INDEX BY PLS_INTEGER; -- Declare collection variables to be used by the FORALL statement zip_tab string_type; city_tab string_type; state_tab string_type; cr_by_tab string_type; cr_date_tab date_type; mod_by_tab string_type; mod_date_tab date_type; v_counter PLS_INTEGER := 0; v_total INTEGER := 0; BEGIN -- Populate individual collections SELECT * BULK COLLECT INTO zip_tab, city_tab, state_tab, cr_by_tab, cr_date_tab, mod_by_tab, mod_date_tab FROM zipcode WHERE state = 'CT';
Page 669
-- Populate MY_ZIPCODE table FORALL i in 1..zip_tab.COUNT INSERT INTO my_zipcode (zip, city, state, created_by, created_date, modified_by, modified_date) VALUES (zip_tab(i), city_tab(i), state_tab(i), cr_by_tab(i), cr_date_tab(i), mod_by_tab(i), mod_date_tab(i)); COMMIT; -- Check how many records were added to MY_ZIPCODE table SELECT COUNT(*) INTO v_total FROM my_zipcode WHERE state = 'CT'; DBMS_OUTPUT.PUT_LINE(v_total||' records were added to MY_ZIPCODE table'); END;
-- Example 2: with DELETE SET SERVEROUTPUT ON DECLARE TYPE order_id_type IS TABLE OF orders.order_id%TYPE; oi_in_tb order_id_type ; oi_out_tb order_id_type ; BEGIN -- Populate collection use in forall. SELECT order_id BULK COLLECT INTO oi_in_tb FROM orders WHERE order_id between 335332 and 335341; FORALL i IN oi_in_tb.first .. oi_in_tb.last DELETE FROM orders WHERE order_id = oi_in_tb(i) RETURNING order_id BULK COLLECT INTO oi_out_tb; DBMS_OUTPUT.put_line('Deleted IDs ROLLBACK; END; : ' || oi_out_tb.count || ' rows');
Set PLSQL_OPTIMIZE_LEVEL and Subprogram Inlining

The parameter controls level of optimization performed by PL/SQL compiler. It ranges from 0 (no optimization) to 3 (3 in 11g). Thed default (2) is generally acceptable. The pragma INLINE compiler directive specifies that a subprogram call is, or is not, to be inlined. It must appear immediately before the subprogram call. SET SERVEROUTPUT ON DECLARE v_num PLS_INTEGER := 1;
Page 670
v_result PLS_INTEGER; -- Following are used for elapsed time calculation v_start_time NUMBER; v_end_time NUMBER; -- Define simple function to test PRAGMA INLINE FUNCTION test_inline_pragma (in_num1 IN PLS_INTEGER, in_num2 IN PLS_INTEGER) RETURN PLS_INTEGER IS BEGIN RETURN (in_num1 + in_num2); END test_inline_pragma; BEGIN -- Test function with INLINE PRAGMA enabled v_start_time := DBMS_UTILITY.GET_TIME; FOR i in 1..10000000 LOOP PRAGMA INLINE (test_inline_pragma, 'YES'); v_result := test_inline_pragma (1, i); END LOOP; v_end_time := DBMS_UTILITY.GET_TIME; DBMS_OUTPUT.PUT_LINE ('Elapsed time when PRAGMA INLINE enabled: '|| (v_end_time-v_start_time)); -- Test function with PRAGMA INLINE disabled v_start_time := DBMS_UTILITY.GET_TIME; FOR i in 1..10000000 LOOP PRAGMA INLINE (test_inline_pragma, 'NO'); v_result := test_inline_pragma (1, i); END LOOP; v_end_time := DBMS_UTILITY.GET_TIME; DBMS_OUTPUT.PUT_LINE ('Elapsed time when INLINE PRAGMA disabled: '|| (v_end_time-v_start_time)); END;
-- output of the code above: Elapsed time when PRAGMA INLINE enabled: 46 Elapsed time when INLINE PRAGMA disabled: 147 PL/SQL procedure successfully completed.
Using Bind variables in Dynamic SQL

Using bind variables allows sharable SQL, reduces parse overhead and minimizes latch contention. -- less performance example: EXECUTE IMMEDIATE 'SELECT COUNT(*) FROM ' || p_mytable || ' WHERE ' || p_column || '=' || p_value INTO v_count -- more efficient EXECUTE IMMEDIATE 'SELECT COUNT(*) FROM ' || p_mytable || ' WHERE ' || p_column || '= :1' INTO v_count USING p_value
Page 671
Use NOCOPY Keyword

The NOCOPY clause causes a parameter to be passed by reference rather than by value. Use it, if possible, when the parameter is of a collection datatype. FUNCTION myfunc ( p_input_table IN OUT NOCOPY number_tab_type, p_row NUMBER, ...
Use Associative arrays

Associative arrays profives faster and simpler lookup than other collections. For further examples, see Using Associative arrays. TYPE custs_type IS TABLE OF NUMBER INDEX BY VARCHAR2(1000); -- the following array should be loaded by some code g_custs custs_type; Function get_cust_id ( p_cust_name VARCHAR2) RETURN NUMBER IS v_cust_id sh.customers.cust_id%type; BEGIN -- simple and fast lookup (no loops required) v_cust_id := g_custs(p_cust_name); RETURN (v_cust_id); END;
Use Server Result Cache

It suits deterministic but expensive functions, such as expensive table lookups on nonvolatile tables. For details, see Using Server Result Cache.
Page 672
PL/SQL Miscellaneous Topics

Accessing V$ Views from PL/SQL
grant select on V_$SESSION to <username> ; grant select on V_$PROCESS to <username>; -- following will NOT work: GRANT SELECT_CATALOG_ROLE TO <username>; GRANT SELECT_CATALOG_ROLE TO <username>; -- connect as <username> create procedure .. select .. from sys.V_$PROCESS p, sys.V_$SESSION s ..
Page 673
Part 13
Appendixes
Page 674
Program Units and Scripts Used in the Document

Return Parameter Value for Normal User
Used by normal user who do not have access on v$parameter. CREATE OR REPLACE FUNCTION GET_PAR ( P_PARAMETER VARCHAR2) RETURN VARCHAR2 IS parnam VARCHAR2(256); intval BINARY_INTEGER; strval VARCHAR2(256); partype BINARY_INTEGER; V VARCHAR2(1000); BEGIN partype := dbms_utility.get_parameter_value(P_PARAMETER, intval, strval); IF partype = 1 THEN RETURN 'TYPE:STRING - Value:' || STRVAL; ELSE RETURN 'TYPE:INTEGER - Value:' || TO_CHAR(intval); END IF; END; /
Page 675
Applying Random Load on Database Sample 1

Scripts to apply loads on Oracle DB on a single table. Applicable on Unix and Windows.
Setup Database Side /* Create Schema cotaining target objects */ create tablespace hrstbs ; create user hrs identified by h default tablespace hrstbs quota unlimited on hrstbs ; grant create table, create view, create procedure, create session, create sequence, plustrace to hrs; grant execute on DBMS_LOCK to hrs; grant execute on DBMS_RANDOM to hrs;
/* Create Schema Objects */ conn hrs/h CREATE TABLE NAMES ( ID NUMBER , NAME VARCHAR2(50), HDATE DATE, SAL NUMBER, REGION VARCHAR2(1)); CREATE SEQUENCE S cache 1000; /* Create Random Load Package */ -- Generates various different loads on the target database CREATE OR REPLACE PACKAGE LOAD_GENERATOR IS -- insert batch rows into NAMES table PROCEDURE INSERT_NAMES ( P_ROWS IN NUMBER); -- high CPU calls PROCEDURE SpinCPUs(P_ITERATION IN NUMBER); -- high CPU+DB calls PROCEDURE SpinCycles (P_ITERATION IN NUMBER); -- random query: from NAMES PROCEDURE RandomQuery(P_ITERATION IN NUMBER, P_MAX IN NUMBER); -- random DML on NAMES PROCEDURE RandomDML(P_ITERATION IN NUMBER, P_MAX IN NUMBER); END load_generator; / -- unmark the SLEEP function, if you wish CREATE OR REPLACE PACKAGE Body LOAD_GENERATOR IS -- generate random text: its lengnth between 4 and the passed value FUNCTION G_TEXT(P_SIZE IN NUMBER) RETURN VARCHAR2 IS V VARCHAR2(2000); BEGIN FOR I IN 1..DBMS_RANDOM.VALUE(4,P_SIZE) LOOP V := V || CHR(ROUND(DBMS_RANDOM.VALUE(65,90))); -- 122 END LOOP; RETURN V; END;
Page 676
PROCEDURE INSERT_NAMES ( P_ROWS IN NUMBER) IS V1 VARCHAR2(15); V2 VARCHAR2(15); BEGIN FOR I IN 1..P_ROWS LOOP V1 := G_TEXT(15); V2 := G_TEXT(15); INSERT INTO NAMES VALUES ( S.NEXTVAL, -- ID V1 || ' ' || V2, -- NAME TRUNC(SYSDATE-DBMS_RANDOM.VALUE(60,1800)), -- HDATE ROUND(DBMS_RANDOM.VALUE(1000,55000)), -- SAL DECODE( TO_CHAR(ROUND(DBMS_RANDOM.VALUE(1,4))), '1','N','2','W','3','E','4','S') ); -- REGION IF MOD(I,100) = 0 THEN COMMIT; END IF; END LOOP; COMMIT; END INSERT_NAMES; PROCEDURE SpinCPUs (P_ITERATION IN NUMBER) IS N NUMBER; BEGIN FOR I IN 1.. P_ITERATION LOOP -- pure CPU processing (no physical or logical read) N := SQRT(ROUND(DBMS_RANDOM.VALUE(1,1000))); -- DBMS_LOCK.SLEEP(round(DBMS_RANDOM.VALUE(0.01,0.05),2)); -- in seconds END LOOP; END SpinCPUs; PROCEDURE SpinCycles (P_ITERATION IN NUMBER) IS N NUMBER; BEGIN FOR I IN 1.. P_ITERATION LOOP FOR X IN 1.. ROUND(DBMS_RANDOM.VALUE(1,10)) LOOP N := SQRT(ROUND(DBMS_RANDOM.VALUE(1,1000))); END LOOP; SELECT COUNT(*) INTO N FROM NAMES; -- DBMS_LOCK.SLEEP(ROUND(DBMS_RANDOM.VALUE(0.01,1),2)); -- in seconds END LOOP; END SpinCycles; PROCEDURE RandomQuery(P_ITERATION IN NUMBER, P_MAX IN NUMBER) IS V_START NUMBER; V_END NUMBER; N NUMBER; BEGIN V_END := P_MAX; V_START := ROUND(DBMS_RANDOM.VALUE(1,V_END)); FOR I IN 1.. P_ITERATION LOOP SELECT COUNT(ID) INTO N FROM NAMES WHERE ID BETWEEN V_START AND V_END;
Page 677
-- DBMS_LOCK.SLEEP(ROUND(DBMS_RANDOM.VALUE(0.01,3),2)); END LOOP; END RandomQuery; PROCEDURE RandomDML(P_ITERATION IN NUMBER, P_MAX IN NUMBER) IS N NUMBER; M NUMBER; V_NEW_SAL NUMBER; V1 VARCHAR2(15); V2 VARCHAR2(15); BEGIN FOR I IN 1.. P_ITERATION LOOP N := ROUND(DBMS_RANDOM.VALUE(1,3)); IF N=1 THEN V1 := G_TEXT(15); V2 := G_TEXT(15); INSERT INTO NAMES VALUES ( S.NEXTVAL, -- ID V1 || ' ' || V2, -- NAME TRUNC(SYSDATE)-DBMS_RANDOM.VALUE(60,1800), -- HDATE ROUND(DBMS_RANDOM.VALUE(1000,55000)), -- SAL DECODE( TO_CHAR(ROUND(DBMS_RANDOM.VALUE(1,4))), '1','N','2','W','3','E','4','S') ); -- REGION ELSIF N=2 THEN M := ROUND(DBMS_RANDOM.VALUE(1,P_MAX)); V_NEW_SAL := ROUND(DBMS_RANDOM.VALUE(1000,55000)); UPDATE NAMES SET SAL = V_NEW_SAL WHERE ID = M; ELSIF N=3 THEN M := ROUND(DBMS_RANDOM.VALUE(1,P_MAX)); DELETE NAMES WHERE ID = M; END IF; -- DBMS_LOCK.SLEEP(ROUND(DBMS_RANDOM.VALUE(0.1,2),2)); COMMIT; END LOOP; END RandomDML; END load_generator; /
-- load some rows in names table execute load_generator.insert_names(10000); OS side: Unix -- in a folder create the following scripts -- (1) loadcpu1.sh #!/bin/bash # apply CPU load on Oracle DB # parameters: 1 connections, 2 Iterations users=$1 SRVC="hrserv" UNPW="hrs/h" SQLCMD="/home/oracle/scripts/load/loadcpu1.sql" x=1 y=$users ITER=$2
Page 678
while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER & x=èxpr $x + 1` done -- (2) loadcpu1.sql begin hrs.LOAD_GENERATOR.SpinCPUs(&1); end; / exit -- (3) loadcpu2.sh #!/bin/bash # apply CPU+DB Calls load on Oracle DB # parameters: 1 connections, 2 Iterations users=$1 SRVC="hrserv" UNPW="hrs/h" SQLCMD="/home/oracle/scripts/load/loadcpu2.sql" x=1 y=$users ITER=$2 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER & x=èxpr $x + 1` done -- (4) loadcpu2.sql begin hrs.LOAD_GENERATOR.SpinCycles(&1); end; / exit -- (5) loadquery.sh #!/bin/bash # apply random queries load on Oracle DB # parameters: 1 connections, 2 Iterations, 3 rows in names users=$1 SRVC="hrserv" UNPW="hrs/h" SQLCMD="/home/oracle/scripts/load/loadquery.sql" x=1 y=$users ITER=$2 MAX=$3 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER $MAX & x=èxpr $x + 1` done
Page 679
-- (6) loadquery.sql begin hrs.LOAD_GENERATOR.RANDOMQUERY (&1, &2); end; / exit -- (7) loaddml.sh #!/bin/bash # apply random DML load on Oracle DB # parameters: 1 connections, 2 Iterations, 3 rows in names users=$1 SRVC="hrserv" UNPW="hrs/h" SQLCMD="/home/oracle/scripts/load/loaddml.sql" x=1 y=$users ITER=$2 MAX=$3 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER $MAX & x=èxpr $x + 1` done -- (8) loaddml.sql begin hrs.LOAD_GENERATOR.RANDOMDML (&1, &2); end; / exit OS side: Windows -- in a folder create the following scripts -- (1) loadcpu1.bat REM apply CPU load on Oracle DB REM parameters: 1 connections, 2 Iterations set set set set set set set users=%1 SRVC=hrserv UNPW=hrs/h SQLCMD=C:\TEMP\load\loadcpu1.sql x=1 y=%users% ITER=%2
for /L %%i in (1,1,%y%) do (start /d "E:\oracle\OraDB11g" sqlplus -S %UNPW%@%SRVC% @%SQLCMD% %ITER%)
-- (2) loadcpu1.sql begin hrs.LOAD_GENERATOR.SpinCPUs(&1); end; / Oracle DBA Code Examples
Page 680
exit
-- (3) loadcpu2.bat REM apply CPU+DB Calls load on Oracle DB REM parameters: 1 connections, 2 Iterations set set set set set set set users=%1 SRVC=hrserv UNPW=hrs/h SQLCMD=C:\TEMP\load\loadcpu2.sql x=1 y=%users% ITER=%2
-- (4) loadcpu2.sql begin hrs.LOAD_GENERATOR.SpinCycles(&1); end; / exit
-- (5) loadquery.bat REM apply random queries load on Oracle DB REM parameters: 1 connections, 2 Iterations, 3 rows in names set set set set set set set set users=%1 SRVC=hrserv UNPW=hrs/h SQLCMD=C:\TEMP\load\loadquery.sql x=1 y=%users% ITER=%2 MAX=%3
for /L %%i in (1,1,%y%) do (start /d "E:\oracle\OraDB11g" sqlplus -S %UNPW%@%SRVC% @%SQLCMD% %ITER% %MAX%) -- (6) loadquery.sql begin hrs.LOAD_GENERATOR.RANDOMQUERY (&1, &2); end; / exit
-- (7) loaddml.bat REM apply random DML load on Oracle DB REM parameters: 1 connections, 2 Iterations, 3 rows in names set users=%1 set SRVC=hrserv
Page 681
set set set set set set
UNPW=hrs/h SQLCMD=C:\TEMP\load\loaddml.sql x=1 y=%users% ITER=%2 MAX=%3
for /L %%i in (1,1,%y%) do (start /d "E:\oracle\OraDB11g" sqlplus -S %UNPW%@%SRVC% @%SQLCMD% %ITER% %MAX%)
-- (8) loaddml.sql begin hrs.LOAD_GENERATOR.RANDOMDML (&1, &2); end; / exit Using the Load Generator Scripts /* in Unix and Windows */ ## Using the Load Generator # # 5 connections 1000 iterations 100000 number of rows in NAMES table loadcpu1 5 1000 loaddml 5 1000 100000
Page 682
Applying Random Load on Database Sample 2

Scripts to apply loads on Oracle DB on a schema named as SA. Tables in this schema taken from the default Oracle OE schema. Applicable on Unix and Windows. Following is the ERD of the SA schema:
Setup Database Side /* Create sa Schema (if not already there) */ create tablespace satbs datafile 'C:\ORACLE\ORADATA\ORA11G\satbs1.dbf' size 100m autoextend on next 12m maxsize 4g extent management local segment space management auto ; -- if ASM is used: CREATE TABLESPACE satbs DATAFILE '+DATA' size 50m autoextend on next 12m maxsize 1g; create user sa identified by s default tablespace satbs quota unlimited on satbs ; grant create table, create view, create procedure, create session, create sequence to sa; grant execute on DBMS_LOCK to sa; grant execute on DBMS_RANDOM to sa; grant SELECT_CATALOG_ROLE tO sa; grant alter session to sa; grant create role to sa;
Page 683
grant plustrace to sa; -- if role doesn't exist @C:\oracle\product\11.1.0\db_1\RDBMS\ADMIN\utlxplan.sql @C:\oracle\product\11.1.0\db_1\sqlplus\admin\plustrce.sql # in unix: @/u01/app/oracle/product/10.2.0/db_1/rdbms/admin/utlxplan.sql @/u01/app/oracle/product/10.2.0/db_1/sqlplus/admin/plustrce.sql
/* Create Schema Objects */ conn sa/s -- execute code in: @ http://www.ahmedbaraka.com/oracle/scripts/populatesa.sql
-- the package /* Create Random Load Package */ -- Generates various different loads on the target database CREATE OR REPLACE PACKAGE sa.LOAD_GENERATOR IS -- global vars G_MAX_ORDER_ID NUMBER; G_MIN_ORDER_ID NUMBER; -- initialize the package PROCEDURE INITIALIZE; -- insert batch rows into NAMES table PROCEDURE INSERT_ORDERS ( P_ROWS IN NUMBER, P_DAYS IN NUMBER DEFAULT 30 ); -- high CPU calls PROCEDURE SpinCPUs(P_ITERATION IN NUMBER); -- random query: from ORDERS PROCEDURE RandomQuery(P_ITERATION IN NUMBER); -- random DML on NAMES PROCEDURE RandomDML(P_ITERATION IN NUMBER); END load_generator; /
CREATE OR REPLACE PACKAGE Body sa.LOAD_GENERATOR IS PROCEDURE INITIALIZE IS BEGIN SELECT MAX(ORDER_ID) , MIN(ORDER_ID) INTO G_MAX_ORDER_ID, G_MIN_ORDER_ID FROM ORDERS; END INITIALIZE; PROCEDURE INSERT_ORDERS ( P_ROWS IN NUMBER, P_DAYS IN NUMBER DEFAULT 30 ) IS V_ORDER_ID NUMBER; V_ORDER_DATE DATE; V_ORDER_MODE VARCHAR2(8); V_CUSTOMER_ID NUMBER;
Page 684
V_ORDER_STATUS NUMBER; V_SALES_REP_ID NUMBER; V_PRODUCT_ID NUMBER; V_PROD_PRICE NUMBER; V_CHANGE_PRICE NUMBER; V_QUANTITY NUMBER; V_ORDER_TOTAL NUMBER; V_ITEMS_COUNT NUMBER; N NUMBER; M NUMBER; -- 97 customers TYPE CUST_ID_TYPE IS VARRAY(100) OF INTEGER; V_CUST_IDS CUST_ID_TYPE := CUST_ID_TYPE (341,342,343,344,345,346,347,348,349,350,351,352,360,361,363,378,380,447,448,4 49,450,451,452,453,454,458,463,466,467,468,470,473,474,475,476,477,478,479,480 ,481,482,483,487,488,492,496,605,606,607,609,615,621,627,712,713,715,717,719,7 21,727,729,731,754,755,756,757,766,767,768,769,770,771,772,782,825,826,827,828 ,829,830,831,832,833,834,835,836,837,838,839,840,841,842,843,844,845,846,847); -- 275 PRODUCTS TYPE PROD_ID_TYPE IS VARRAY(275) OF INTEGER; V_PROD_IDS PROD_ID_TYPE := PROD_ID_TYPE (1750,1755,1761,1763,1768,1769,1770,1772,1774,1775,1778,1779,1780,1781,1782,17 87,1788,1791,1792,1794,1797,1799,1801,1803,1804,1805,1806,1808,1820,1822,1825, 1910,1912,1940,1948,2004,2005,2030,2049,2056,2058,2091,2093,2144,2152,2211,223 1,2236,2243,2245,2252,2253,2254,2255,2257,2259,2260,2261,2262,2264,2266,2268,2 270,2272,2274,2276,2278,2289,2293,2299,2302,2308,2311,2316,2319,2322,2323,2326 ,2330,2334,2335,2336,2337,2339,2340,2350,2351,2359,2365,2370,2371,2373,2374,23 75,2377,2378,2380,2381,2382,2384,2387,2394,2395,2396,2400,2402,2403,2404,2406, 2408,2409,2410,2411,2412,2414,2415,2416,2417,2418,2419,2422,2423,2424,2430,243 9,2449,2452,2453,2457,2459,2462,2464,2467,2468,2470,2471,2492,2493,2494,2496,2 497,2522,2536,2537,2594,2596,2631,2638,2721,2722,2725,2751,2752,2761,2779,2782 ,2783,2808,2810,2870,2878,2879,2944,2976,2982,2986,2995,2999,3000,3001,3003,30 04,3020,3051,3054,3057,3060,3061,3064,3065,3069,3071,3072,3073,3077,3082,3083, 3086,3087,3088,3090,3091,3097,3099,3101,3106,3108,3110,3112,3114,3117,3123,312 4,3127,3129,3133,3134,3139,3140,3143,3150,3155,3163,3165,3167,3170,3171,3172,3 173,3175,3176,3177,3178,3179,3182,3183,3187,3189,3191,3193,3197,3204,3208,3209 ,3216,3220,3224,3225,3234,3245,3246,3247,3248,3250,3251,3252,3253,3255,3256,32 57,3258,3260,3262,3277,3290,3300,3301,3331,3334,3337,3350,3353,3354,3355,3359, 3361,3362,3391,3399,3400,3501,3502,3503,3511,3515); BEGIN FOR I IN 1..P_ROWS LOOP SELECT SEQ_ORDER_ID.NEXTVAL INTO V_ORDER_ID FROM DUAL; V_ORDER_DATE := TRUNC(SYSDATE- P_DAYS ); N := ROUND(DBMS_RANDOM.VALUE(1,2)); IF N = 1 THEN V_ORDER_MODE := 'direct'; ELSE V_ORDER_MODE := 'online'; END IF; V_CUSTOMER_ID := ROUND(DBMS_RANDOM.VALUE(1,96)); V_ORDER_STATUS:= ROUND(DBMS_RANDOM.VALUE(1,10)); V_SALES_REP_ID:= ROUND(DBMS_RANDOM.VALUE(153,163)); INSERT INTO ORDERS (ORDER_ID,ORDER_DATE,ORDER_MODE,CUSTOMER_ID,ORDER_STATUS, ORDER_TOTAL,SALES_REP_ID) VALUES ( V_ORDER_ID, V_ORDER_DATE, V_ORDER_MODE,
Page 685
V_CUST_IDS(V_CUSTOMER_ID), V_ORDER_STATUS, 0, V_SALES_REP_ID); V_ORDER_TOTAL := 0; V_ITEMS_COUNT := ROUND(DBMS_RANDOM.VALUE(1,10)); M := 1; FOR X IN 1..V_ITEMS_COUNT LOOP V_PRODUCT_ID:= ROUND(DBMS_RANDOM.VALUE(1,275)); SELECT LIST_PRICE INTO V_PROD_PRICE FROM PRODUCT_INFORMATION WHERE PRODUCT_ID=V_PROD_IDS(V_PRODUCT_ID); V_CHANGE_PRICE := ROUND(DBMS_RANDOM.VALUE(1,4)); -- discount (MAX 20%) may be given to 25% of entered items IF V_CHANGE_PRICE=4 THEN V_PROD_PRICE := ROUND(V_PROD_PRICE (ROUND(DBMS_RANDOM.VALUE(1,20))/100)*V_PROD_PRICE); END IF; V_QUANTITY:=ROUND(DBMS_RANDOM.VALUE(1,20)); V_ORDER_TOTAL := V_ORDER_TOTAL + V_QUANTITY*V_PROD_PRICE; BEGIN INSERT INTO ORDER_ITEMS(ORDER_ID,LINE_ITEM_ID,PRODUCT_ID,UNIT_PRICE,QUANTITY) VALUES(V_ORDER_ID,M,V_PROD_IDS(V_PRODUCT_ID),V_PROD_PRICE,V_QUANTITY ); M := M + 1 ; EXCEPTION WHEN OTHERS THEN IF UPPER(SQLERRM) LIKE '%ORDER_ITEMS_UK%' THEN NULL; ELSE RAISE; END IF; END ; END LOOP; -- X loop UPDATE ORDERS SET ORDER_TOTAL = V_ORDER_TOTAL WHERE ORDER_ID = V_ORDER_ID; IF MOD(I,100) = 0 THEN COMMIT; END IF; END LOOP; -- I loop COMMIT; END INSERT_ORDERS; PROCEDURE SpinCPUs (P_ITERATION IN NUMBER) IS N NUMBER; BEGIN FOR I IN 1.. P_ITERATION LOOP -- pure CPU processing (no physical or logical read) N := SQRT(ROUND(DBMS_RANDOM.VALUE(1,1000))); -- DBMS_LOCK.SLEEP(round(DBMS_RANDOM.VALUE(0.01,0.05),2)); -- in seconds END LOOP; END SpinCPUs; PROCEDURE RandomQuery(P_ITERATION IN NUMBER) IS V_START NUMBER; V_END NUMBER; N NUMBER;
Page 686
BEGIN V_END := G_MAX_ORDER_ID; V_START := ROUND(DBMS_RANDOM.VALUE(G_MIN_ORDER_ID,G_MAX_ORDER_ID)); FOR I IN 1.. P_ITERATION LOOP FOR R IN (SELECT O.ORDER_ID, O.ORDER_DATE, C.CUST_LAST_NAME, O.ORDER_TOTAL FROM ORDERS O , CUSTOMERS C WHERE O.CUSTOMER_ID = C.CUSTOMER_ID AND ORDER_ID BETWEEN V_START AND V_END) LOOP -- just retreive data NULL; END LOOP; -- DBMS_LOCK.SLEEP(ROUND(DBMS_RANDOM.VALUE(0.01,3),2)); END LOOP; END RandomQuery; PROCEDURE RandomDML(P_ITERATION IN NUMBER) IS V VARCHAR2(1); N NUMBER; V_ORDER_ID NUMBER; V_LINE_ITEM_ID NUMBER ; V_MAX_ORDERI_ID NUMBER; V_MIN_ORDERI_ID NUMBER; V_PRICE NUMBER; V_ORDER_LOCKED BOOLEAN := FALSE; BEGIN FOR I IN 1.. P_ITERATION LOOP N := ROUND(DBMS_RANDOM.VALUE(1,3)); IF N=1 THEN INSERT_ORDERS(1); ELSIF N=2 THEN -- pick up an order and modify the price of one of its items V_ORDER_ID := ROUND(DBMS_RANDOM.VALUE(G_MIN_ORDER_ID,G_MAX_ORDER_ID)); V_ORDER_LOCKED := FALSE; -- make sure the order isn't locked BEGIN SELECT 'X' INTO V FROM ORDERS WHERE ORDER_ID = V_ORDER_ID FOR UPDATE WAIT 3; EXCEPTION WHEN OTHERS THEN IF SQLCODE='-30006' THEN -- the row is locked V_ORDER_LOCKED := TRUE; ELSE RAISE; -- if there is another non-expected error, raise it END IF; END; IF V_ORDER_LOCKED THEN NULL; -- no need to make the session blocked ELSE -- update the item BEGIN SELECT MAX(LINE_ITEM_ID), MIN(LINE_ITEM_ID) INTO V_MAX_ORDERI_ID, V_MIN_ORDERI_ID FROM ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID; V_LINE_ITEM_ID := ROUND(DBMS_RANDOM.VALUE(V_MIN_ORDERI_ID,V_MAX_ORDERI_ID)); SELECT UNIT_PRICE INTO V_PRICE FROM ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID AND LINE_ITEM_ID=V_LINE_ITEM_ID; V_PRICE := V_PRICE + (ROUND(DBMS_RANDOM.VALUE(-20,20))/100)*V_PRICE; -- update if there is no lock on the item
Page 687
BEGIN SELECT 'X' INTO V FROM ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID AND LINE_ITEM_ID=V_LINE_ITEM_ID FOR UPDATE WAIT 3; UPDATE ORDER_ITEMS SET UNIT_PRICE=V_PRICE WHERE ORDER_ID=V_ORDER_ID AND LINE_ITEM_ID=V_LINE_ITEM_ID; UPDATE ORDERS SET ORDER_TOTAL=ORDER_TOTAL+V_PRICE WHERE ORDER_ID=V_ORDER_ID; EXCEPTION WHEN OTHERS THEN IF SQLCODE='-30006' THEN -- the row is locked NULL; -- do nothing ELSE RAISE; -- if there is another non-expected error, raise it END IF; END; EXCEPTION -- if order has no items WHEN NO_DATA_FOUND THEN NULL; END; END IF; -- V_ORDER_LOCKED ELSIF N=3 THEN -- pick up an item to delete V_ORDER_ID := ROUND(DBMS_RANDOM.VALUE(G_MIN_ORDER_ID,G_MAX_ORDER_ID)); SELECT MAX(LINE_ITEM_ID), MIN(LINE_ITEM_ID) INTO V_MAX_ORDERI_ID, V_MIN_ORDERI_ID FROM ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID; V_LINE_ITEM_ID := ROUND(DBMS_RANDOM.VALUE(V_MIN_ORDERI_ID,V_MAX_ORDERI_ID)); DELETE ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID AND LINE_ITEM_ID=V_LINE_ITEM_ID; -- re-order the itmes in the order DECLARE CURSOR CR IS SELECT LINE_ITEM_ID FROM ORDER_ITEMS WHERE ORDER_ID=V_ORDER_ID ORDER BY LINE_ITEM_ID FOR UPDATE; BEGIN N:=1; FOR R IN CR LOOP UPDATE ORDER_ITEMS SET LINE_ITEM_ID = N WHERE CURRENT OF CR; N:=N+1; END LOOP; END ; END IF; -- DBMS_LOCK.SLEEP(ROUND(DBMS_RANDOM.VALUE(0.1,2),2)); COMMIT; END LOOP; END RandomDML; Begin INITIALIZE; END load_generator; / -- initial data exec load_generator.insert_orders(100000); exec dbms_stats.gather_schema_stats('SA');
Page 688
OS side: Unix -- in a folder create the following scripts mkdir ~/scripts cd ~/scripts mkdir load cd load -- (1) loadcpu1.sh vi loadcpu.sh #!/bin/bash # apply CPU load on Oracle DB # parameters: 1 connections, 2 Iterations users=$1 SRVC="ora11gr2" UNPW="sa/s" SQLCMD="/home/oracle/scripts/load/loadcpu.sql" x=1 y=$users ITER=$2 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER & x=èxpr $x + 1` done -- (2) loadcpu.sql vi loadcpu.sql begin sa.LOAD_GENERATOR.SpinCPUs(&1); end; / exit
-- (3) loadquery.sh vi loadquery.sh #!/bin/bash # apply random queries load on Oracle DB # parameters: 1 connections, 2 Iterations, 3 rows in names users=$1 SRVC="ora11gr2" UNPW="sa/s" SQLCMD="/home/oracle/scripts/load/loadquery.sql" x=1 y=$users ITER=$2 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER & x=èxpr $x + 1` done -- (4) loadquery.sql vi loadquery.sql begin sa.LOAD_GENERATOR.RANDOMQUERY (&1);
Page 689
end; / exit
-- (5) loaddml.sh vi loaddml.sh #!/bin/bash # apply random DML load on Oracle DB # parameters: 1 connections, 2 Iterations users=$1 SRVC="ora11gr2" UNPW="sa/s" SQLCMD="/home/oracle/scripts/load/loaddml.sql" x=1 y=$users ITER=$2 while [ $x -le $y ] do sqlplus -s $UNPW@$SRVC @$SQLCMD $ITER & x=èxpr $x + 1` done -- (6) loaddml.sql vi loaddml.sql begin sa.LOAD_GENERATOR.RANDOMDML(&1); end; / exit
# make all sh files in current directory executables for f in *.sh; do chmod 744 $f; done OS side: Windows -- in a folder create the following scripts -- (1) loadcpu.bat REM apply CPU load on Oracle DB REM parameters: 1 connections, 2 Iterations set set set set set set set users=%1 SRVC=ora11g UNPW=sa/s SQLCMD=C:\TEMP\load\loadcpu.sql x=1 y=%users% ITER=%2
-- (2) loadcpu1.sql begin hrs.LOAD_GENERATOR.SpinCPU(&1); end; Oracle DBA Code Examples
Page 690
/ exit
-- (3) loadquery.bat REM apply random queries load on Oracle DB REM parameters: 1 connections, 2 Iterations set set set set set set set users=%1 SRVC=ora11g UNPW=sa/s SQLCMD=C:\TEMP\load\loadquery.sql x=1 y=%users% ITER=%2
for /L %%i in (1,1,%y%) do (start /d "E:\oracle\OraDB11g" sqlplus -S %UNPW%@%SRVC% @%SQLCMD% %ITER%) -- (4) loadquery.sql begin hrs.LOAD_GENERATOR.RANDOMQUERY(&1); end; / exit
-- (5) loaddml.bat REM apply random DML load on Oracle DB REM parameters: 1 connections, 2 Iterations, 3 rows in names set users=%1 set SRVC=ora11g set UNPW=sa/s set SQLCMD=C:\TEMP\load\loaddml.sql set x=1 set y=%users% set ITER=%2 for /L %%i in (1,1,%y%) do (start /d "E:\oracle\OraDB11g" sqlplus -S %UNPW%@%SRVC% @%SQLCMD% %ITER%)
-- (6) loaddml.sql begin sa.LOAD_GENERATOR.RANDOMDML (&1); end; / exit Using the Load Generator Scripts ## Using the Load Generator # # 5 connections 1000 iterations 100000 number of rows in NAMES table # in Windows: loadcpu 5 1000 loaddml 5 1000 # in Unix:
Page 691
./loadcpu.sh 5 1000
Page 692
SQL Usage Samples

Merge Command
/* Marge Command */ MERGE INTO products p /* Destination table USING product_changes s /* Source table ON (p.prod_id = s.prod_id) /* Search/join condition WHEN MATCHED THEN UPDATE /* Update if join SET p.prod_list_price = s.prod_new_price WHERE p.prod_status <> 'EXPIRED' /* Conditional update WHEN NOT MATCHED THEN INSERT /* Insert if not join SET p.prod_list_price = s.prod_new_price WHERE s.prod_status <> 'EXPIRED' /* Conditional insert -- using delete keyword with Merge command -- only rows to update will be considered by the delete clause MERGE INTO products p USING product_changes s ON (p.prod_id = s.prod_id) WHEN MATCHED THEN UPDATE SET p.prod_list_price = s.prod_new_price, p.prod_status = s.prod_new_status DELETE WHERE (p.prod_status = 'OLD_ITEM') -- it part of update WHEN NOT MATCHED THEN INSERT (prod_id, prod_list_price, prod_status) VALUES (s.prod_id, s.prod_new_price, s.prod_new_status);
Multitable Inserts
/* Multitable Inserts */ -- unconditional INSERT ALL INTO target1 VALUES (product_id, customer_id, sysdate, product_quantity) INTO target2 VALUES (product_id,sysdate,product_price,product_discount) SELECT s.product_id, s.customer_id, sysdate, s.product_quantity, s.product_price, s.product_discount FROM source s; -- conditional ALL rows -- a row can be on the two tables INSERT ALL WHEN product_id IN(SELECT product_id FROM primary) THEN INTO target1 VALUES (product_id, customer_id, sysdate, product_quantity) WHEN product_id IN (SELECT product_id FROM secondary) THEN INTO target2 VALUES (product_id, sysdate, product_price, product_discount) SELECT s.product_id, s.customer_id, sysdate, s.product_quantity, s.product_price, s.product_discount FROM source s; -- conditional first rows -- insert into table of first applied condition -> a row is in only one table INSERT FIRST WHEN (sum_quantity_sold > 10 AND prod_weight_class < 5) AND sum_quantity_sold >=1) OR (sum_quantity_sold > 5 AND prod_weight_class > 5)
Page 693
THEN INTO large_freight_shipping VALUES (time_id, cust_id, prod_id, prod_weight_class, sum_quantity_sold) WHEN sum_amount_sold > 1000 AND sum_quantity_sold >=1 THEN INTO express_shipping VALUES (time_id, cust_id, prod_id, prod_weight_class, sum_amount_sold, sum_quantity_sold) WHEN (sum_quantity_sold >=1) THEN INTO default_shipping VALUES (time_id, cust_id, prod_id, sum_quantity_sold) ELSE INTO incorrect_sales_order VALUES (time_id, cust_id, prod_id) SELECT s.time_id, s.cust_id, s.prod_id, p.prod_weight_class, SUM(amount_sold) AS sum_amount_sold, SUM(quantity_sold) AS sum_quantity_sold FROM sales s, products p WHERE s.prod_id = p.prod_id AND s.time_id = TRUNC(SYSDATE) GROUP BY s.time_id, s.cust_id, s.prod_id, p.prod_weight_class; -- Mixed Conditional and Unconditional Insert INSERT FIRST WHEN cust_credit_limit >= 4500 THEN INTO customers_special VALUES (cust_id, cust_credit_limit) ELSE INTO customers SELECT * FROM customers_new;
Parallel Insert
For further details see: Enabling Direct-Path INSERT /* Parallel loading into table */ -- must before the hint ALTER SESSION ENABLE PARALLEL DML; -- then use the hint INSERT /*APPEND NOLOGGING PARALLEL */ INTO sales_data SELECT product_id, customer_id, TRUNC(sales_date), discount_rate, sales_quantity, sale_price FROM sales_history; -- if triggers are there, hint won't be used
Deleting Duplicate Rows in a Table

-- delete duplicated rows in our_table based on col1, col2.. etc. DELETE FROM our_table WHERE rowid not in (SELECT MIN(rowid) FROM our_table GROUP BY column1, column2, column3...) ;
Adding a Primary Key to a Table and then Filling it

create sequence seq_OAS_BALANCE; alter table OAS_BALANCE add ( pkid number );
Page 694
declare n number; cursor crs is select rowid from OAS_BALANCE for update of pkid; begin for r in crs loop select seq_OAS_BALANCE.nextval into n from dual ; update OAS_BALANCE set pkid=n where current of crs; end loop; commit; end; /
Page 695
More DBA Scripts

Compare Table Sizes in Two Databases
--compare tables in current db with same tables in other db useful for checking after transferring too many tables
DECLARE N NUMBER; M NUMBER; BEGIN DBMS_OUTPUT.ENABLE(100000); FOR S IN ( SELECT USERNAME FROM SCHEMA_NAMES WHERE USERNAME <>'MOL' ORDER BY 1) LOOP FOR O in ( SELECT TABLE_NAME FROM DBA_TABLES@OLDORA11G WHERE OWNER = S.USERNAME ) LOOP BEGIN N := 0; M := 0; EXECUTE IMMEDIATE 'SELECT COUNT(ROWID) FROM ' || S.USERNAME ||'."'|| O.TABLE_NAME || '"@OLDORA11G' INTO N; EXECUTE IMMEDIATE 'SELECT COUNT(ROWID) FROM ' || S.USERNAME ||'."'|| O.TABLE_NAME || '"' INTO M; IF M <> N THEN DBMS_OUTPUT.PUT_LINE(S.USERNAME || CHR(9) || O.TABLE_NAME); END IF; EXCEPTION WHEN OTHERS THEN DBMS_OUTPUT.PUT_LINE(S.USERNAME || CHR(9) || O.TABLE_NAME); RAISE; END; END LOOP; END LOOP; END; /
Page 696
Using Linux for Oracle

Verifying the Kernel
# is of format: <Base>-<Errata>.<Mod><Type> # Mod= Modifier (developer) # Type: None: Uniprocessor, smp: SMP up to 16 GB, hugemem: SMP up to 64 GB uname -r 2.6.9-42.0.0.0.1.ELsmp # name of the release cat /etc/enterprise-release
Checking for a Tainted Kernel

Usually, tainted kernelt is not supported by Oracle. Some exceptions: OCFS (276450.1) and EMC Powerpath driver (284823.1) # 0: not tainted 1:tainted cat /proc/sys/kernel/tainted # to list the status of modules lsmod
Supported Hardware
Systems listed in Red Hat site (https://hardware.redhat.com) are supported by Oracle Linux.
Using Oracle Relink Utility

If the operating system or Oracle object files change, then Oracle executables should be relinked to create updated executables properly integrated with the operating system. If utility returns nothing, then it means it successfully finished. Relinking occurs automatically when a patch set is installed, but must be performed manually when: o An OS upgrade has occurred o o o A change has been made to the OS system libraries. This can occur during the application of an OS patch. A new installation failed during the relinking phase. An individual Oracle patch has been applied. Explicit relink instructions are usually either included in README or integrated into the patch install script.
1. Verify whether ORACLE_HOME is correctly set. 2. Set LD_LIBRARY_PATH to include $ORACLE_HOME/lib. 3. Execute the appropriate relink command to relink the appropriate binaries: relink all: All Oracle executables relink oracle Database executables relink network: Listener, cman, names
Page 697
relink oemagent: Intelligent Agent relink utilities: SQL*Loader, tkprof, rman, impdp, expdp, imp, exp relink ctx: Oracle Text
Certified and Supported File Systems

In Linux, Oracle certifies three file systems: o o ext3 Automatic Storage Management (ASM) Oracle Cluster File System/Oracle Cluster File System 2 (OCFS2)
Enterprise Linux Runlevels

Runlevel 0: Halt. It is used to shut down the system. Runlevel 1: Single-user (maintenance) mode. Only root may log in. Runlevel 2: Multiuser mode, text-based console only Runlevel 3: Multiuser mode with networking Runlevel 4: For custom use Runlevel 5: The default mode; multiuser with networking and active X session Runlevel 6: Reboot. This shuts everything down and then starts it back. Booting to a Nondefault Runlevel Enterprise Linux: >you press [E] then scroll down to select the kernel >press [E] again >press the spacebar >enter the runlevel you want to boot to >then press [B] # to print previous and current runlevels: runlevel
Using /etc/oratab File and dbstart Utility

Y or N in /etc/oratab indicates to the dbstart utility whether the database should be brought up at system boot time. ASM and database instances created without using DBCA will not appear in the oratab file. You should manually add them: +ASM:/u01/app/oracle/product/10.2.0/db_1:Y orcl:/u01/app/oracle/product/10.2.0/db_1:Y The dbstart script that is provided with 10.2.0.1.0 contains an error. The Metalink note 336299.1 documents this issue. To solve this problem, perform the following steps: 1. Edit $ORACLE_HOME/bin/dbstart 2. Go to line 78. 3. Replace the line with: Oracle DBA Code Examples
Page 698
ORACLE_HOME_LISTNER=$ORACLE_HOME 4. Save the file and exit the editor. 5. Execute dbstart.
Automating Jobs
Using cron /* security */ Only users listed in the /etc/cron.allow file are allowed to use cron Users listed in the /etc/cron.deny file cannot use cron. If neither of these files exist, only the superuser can use cron.
/* Managing crontab files */ crontab used to install, deinstall or list the tables (crontabs). -u user the user whose crontab is to be tweaked -l display the current crontab file -r remove the current crontab file -e edit. After exit from the editor, the modified crontab will be installed
/* Using cron */ crontab format: <minute> <hour> <day> <month> <weekday> <cmd> 1.minute (from 0 to 59) or special keywords: 2.hour (from 0 to 23) 3.day of month (from 1 to 31) 4.month (from 1 to 12) 5.day of week (from 0 to 6) (0=Sunday) Special keywords are: @reboot Run once, @yearly Run once @annually (same as @monthly Run once @weekly Run once @daily Run once @midnight (same as @hourly Run once
at startup a year @yearly) a month a week a day @daily) an hour
"0 0 1 1 *" "0 0 1 * *" "0 0 * * 0" "0 0 * * *" "0 * * * *"
If the administrator puts an executable script into one of the following directories, then cron runs the script at the appropriate interval: /etc/cron.hourly /etc/cron.daily /etc/cron.weekly /etc/cron.monthly
/* Mailing the crontab output */ By default cron saves the output in the user's mailbox on the local system. But you can also configure crontab to forward all output to a real email address: crontab MAILTO="[email protected]"
Page 699
If want to receive only one cronjob's output in your mail, make sure this package is installed: aptitude install mailx then change the cronjob like this: */10 * * * * /bin/execute/this/script.sh 2>&1 | mail -s "Cronjob ouput" yourname@yourdomain
/* Examples */ >every Friday 1AM 0 1 * * 5 /bin/execute/this/script.sh >14 minutes after 10 p.m. every Monday through Friday of every month 14 22 * * 1-5 su - oracle -c /usr/local/bin/backup.cmd >/dev/null 2>&1 >10 past after every hour on the 1st of every month 10 * 1 * * /bin/execute/this/script.sh >every 10 minutes 0,10,20,30,40,50 * * * * /bin/execute/this/script.sh */10 * * * * /bin/execute/this/script.sh >daily @daily /bin/execute/this/script.sh Using anacron Most system cron jobs will run during the night and will get done only if the computer is switched on. The anacron mechanism on the other hand assumes that the computer is not running continuously. After the computer is started anacron will execute jobs after a certain delay. anacron keeps track of the day a given job has been executed. So the shortest period a job can be executed is a day. The anacron daemon is usually started from an rc-script and will read an anacrontab which has the following format: Period Delay Job-Identifier Command The period is given in days, the delay is in minutes, the job-identifier can be any character and the command is usually of the form run-parts /etc/cron.daily but can be any command. Example 9.4. Sample /etc/anacrontab SHELL=/bin/sh PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin MAILTO=root 1 65 cron.daily run-parts /etc/cron.daily 7 70 cron.weekly run-parts /etc/cron.weekly 30 75 cron.monthly run-parts /etc/cron.monthly
Using at command The at command executes a command once at a specified time. By default at is restricted to the root user. To override this you must either have an empty /etc/at.deny or have a /etc/at.allow with the appropriate names. at jobs are saved in /var/spool/at/ Oracle DBA Code Examples
Page 700
at jobs are spooled in /var/spool/at/ at [time] Where time can be expressed as: now now + 1 min now + 1 hour now + 1 day 3am + 2days midnight 10:15 Apr 12 teatime For complete list of valid time formats see /usr/share/doc/at-xxx/timespec # execute the commands in the file at -f myjobs.txt now + 1 hour # list commands that are scheduled at -l atq # remove a job atrm <jobid>
Using batch command batch command executes commands when system load average drops below 0.8, or the value specified in the invocation of atrun. batch [-V] [-q queue] [-f file] [-mv] [TIME] Task Scheduler There are GUI task schedulers available that are front ends for cron in both the Gnome and the K(ommon) Desktop Environment (KDE).
Configuring Linux Memory for Oracle 32 bit Architecture: With Linux 32-bit, Oracle is able to use about 1.7GB of address space for its SGA. To increase it, refere to noted 200266.1. Be aware of the consequenced limitations. In Red Hat Enterprise Linux 3, 4 or 5 the smp kernel can be used on systems with up to 16 GB of RAM. The hugemem kernel is required in order to use all the memory on systems that have more than 16GB of RAM up to 64GB. However, it is recommend to use the hugemem kernel even on systems that have 8GB of RAM. In Red Hat Enterprise Linux 5, a 32 bit kernel is always a hugemem kernel so there is no need to install a special kernel.
Using Linux Performance Monitoring Tools

Note: OS Watcher uses most commonly used Linux tools and store their output in files. It is a nice frame work to implement using those tools. To learn more about using it, see Using OS Watcher.
Page 701
About Linux Tools top: useful for diagnosing CPU and I/O issues, though not as reliable for Oracle memory issues. Individual process memory numbers should not be relied upon. sar: reports system activity including memory, I/O, and CPU usage. sar samples the /proc file system at 10-minute intervals and records the data in /var/log/sa as a sa# file where # is the day of the month. A full months worth of statistics are retained so that the administrator can use sar to view past statistics as well as current. For example, to view the CPU statistics for the 23rd of the month: sar u 2 4 f /var/log/sa/sa29 -o [<fnmae>] X) -f [<fnmae>] save output to filename (default /var/log/sa/sadd ) (cannot be used with read from filename (default /var/log/sa/sadd )
-X pid | SELF | ALL Report statistics for the child processes of the process whose PID is pid. SELF statistics of the child processes of the sar process itself ALL statistics of the child processes of all the system processes. vmstat: activity on processes, memory, paging, block I/O, traps, and CPU activity (best for memory) iostat: disk activity System Log Files: /var/log/message ../syslog free command displays the total amount of free and used physical and swap memory in the system, as well as the shared memory and buffers used by the kernel. mpstat: reporter on CPU statistics. Graphical Measurement Tools
Using Linux Tools /* CPU */ # how many CPUs cat /proc/cpuinfo # use top to see averae load (now, 5 minutes ago, and 10 minutes ago). # load factor= load average/#CPUs # if lf<1: no load on CPU, # between 1 and 2: CPU running at capacity # >2, CPU may be a bottleneck (use sar and vmstat to investigate) # if you use "1" interactive command, you can see load per CPU top # CPU metrics # all appended by <interval> <count> vmstat mpstat -P <CPU>|ALL sar -u # queue size: waiting for CPU # (shouldn't be larger than CPU#) sar -q # display CPU stats only iostat -c # to know which session is consuming CPU
Page 702
# get its PID from top # then: select pid, spid, pname, username, program from v$process where spid=20052; select * from v$session where paddr= ( select addr from v$process where spid=20304);
/* Memory */ # Total Memory free -m, top, or cat /proc/meminfo # higher "Pages In" indicate RAM shortage top free cat /proc/meminfo # paging info sar -B # memory metrics sar -R
/* IO */ # storage vendors have their monitoring tools # io activity by disk # check avgrq-sz.await which gives the average wait time for requests iostat -d # by partition iostat -p # exetended output iostat -d -x # rtps/wtps requests per se sar -d 2 10 # about overall io stats vmstat
Checking Some General Guideline on Truning Oracle in Linux

Use an appropriately sized database block: general purpose:8K, OLTP:4K, DSS:16K. Use a larger block size if the application has very large indexes (15>levels), or consider using partitioned indexes. Use a larger redo log buffer than default: for OLTP systems Multiprocessing (SMP) machines. Setting PRE_PAGE_SGA = true causes each server process that starts to touch every page of the SGA, thus mapping all the SGA pages to physical memory. It may slow down connection time, though; so make sure you have enough free memory to accomodate all SGA. Use automatic memory management. Setting atime off: In Linux, by default, the last time-read attribute (atime) is updated every time a file is read. For database files, this is not an important statistic and you can turn it off to reduce the number of I/Os:
Page 703
# This parameter can be set for a file with: chattr +A <filename> # for a directory with: chattr -R +A <directory name> # To make it persistent across reboots: vi /etc/fstab file /dev/hdb7 /u3/app/oracle/ ext3 rw,noatime 1 1
Troubleshooting Oracle Database in Linux

Using OS Watcher (OSW) OS Watcher (OSW) is a collection of UNIX shell scripts intended to collect and archive operating system and network metrics to aid support in diagnosing performance issues. OSW operates as a set of background processes on the server and gathers OS data on a regular basis, invoking such Unix utilities as vmstat, netstat and iostat. OSW will need access to the OS utilities: top, vmstat, iostat, mpstat, netstat, and traceroute. On RAC, you can install it in every node. After generating the output files, consider generating a profile from oswg (next section). Obtain latest version from Note 301137.1 /* Managing OSW */ # installing OSW (directory osw will be created): tar xvf osw.tar
# Uninstalling OSW rm -rf osw
# Setting up OSW for tracing private networks (RAC): mv Exampleprivate.net private.net
# starting OSW ./startOSW.sh <interval in seconds> <hour# retention> # to startup in background: nohup ./startOSW.sh 30 12 &
# Stopping OSW ./stopOSW.sh
/* Diagnostic Data Output */ # archived files generated per hour with the following format: <node_name>_<OS_utility>_YY.MM.DD.HH24.dat
# there are 3 entries for each timestamp. You should always ignore # the first entry as this entry is always invalid. The second and
Page 704
# third entry will be valid but the second entry will be 1 sec later # than the timestamp and the third entry will be 2 seconds later than the timestamp.
# oswiostat # What to look for: - Average service times greater than 20msec for long duration. - High average wait times r/s Shows the number of reads/second w/s Shows the number of writes/second kr/s Shows the number of kilobytes read/second kw/s Shows the number of kilobytes written/second wait Average number of transactions waiting for service (queue length) actv Average number of transactions actively being serviced wsvc_t Average service time in wait queue, in milliseconds asvc_t Average service time of active transactions, in milliseconds %w Percent of time there are transactions waiting for service %b Percent of time the disk is busy device Device name
# oswmpstat What to look for - Involuntary context switches (this is probably the more relevant statistic when examining performance issues.) - Number of times a CPU failed to obtain a mutex. Values consistently greater than 200 per CPU causes system time to increase. - xcal is very important, show processor migration
# oswnetstat # the collisions should be 0.1 percent or less Network collision rate = Output collision / Output packets # Input Error Rate = Ierrs / Ipkts If the input error rate is high (over 0.25 percent), the host is excessively dropping packets. # segment retransmission rate should be low %segment-retrans=(tcpRetransSegs / tcpOutDataSegs) * 100
# oswprvtnet check Interface is up and responding
# oswps list of processes
# # # # #
oswtop examine average load Large run queue A process which is "hogging" CPU is always suspect number of processes usually do not change over time
# oswvmstat
Page 705
Using OS Watcher Graphs(OSWg) Using OSWg # to use Java shipped with Oracle $ cd $ORACLE_HOME/jre/1.4.2/bin export PATH=/u01/app/oracle/product/11.2.0/db_1/jdk/jre/bin:$PATH
# invoking the oswg: java -jar oswg.jar -i /home/oracle/osw/archive # if you see: java.lang.OutOfMemoryError, # you may have to increase the size of the java heap: $java -jar -Xmx512M oswg.jar -i /home/oracle/osw/archive Note: to generate a profile, mkdir profile under osw home # Using OSWg: Menu Option java -jar OSWg.jar -i <fully qualified path name of an osw archive directory> # other starting options: java -jar OSWg.jar -i <fully qualified path name of an osw archive directory> -P <name> -L <name> -6 -7 -8 -B <time> -E <time> -B <start time> Same as option T from the menu. The start time will allow the user to select a start time from within the archive of files to graph/profile. This overrides the default start time which is the earliest time entry in the archive directory. The format of the start time is Mon DD HH:MM:SS YYYY. (Example :Jul 25 11:58:01 2007). An end time is required if selecting this option. -E <end time> Same as option T from the menu. The end time will allow the user to select an end time from within the archive of files to graph/profile. This overrides the default end time which is the latest time entry in the archive directory. The format of the end time is Mon DD HH:MM:SS YYYY. (Example :Jul 25 11:58:01 2007). A start time is required if selecting this option. Using the On-Board Monitor (LTOM) LTOM is an embedded real-time data Collection and diagnostics platform developed by Center of Expertise in Oracle. LTOMg is embedded in LTOM. Its reference is Note 352363.1. Usage details are in README file. Automatic Hang Detection o it starts collecting statisics about hanged sessions o to use it for RAC, install LTOM in a single node. (Other features require installing it in all nodes) o uses a rule based hang detection algorithm configured in $TOM_HOME/init/hangDetect.properties System Profiler o The System Profiler provides the ability to continually collect data from both the operating system and oracle and provides an integrated snapshot of the overall health of the operating system together with the database. o Once the data is collected, the data can be parsed and analyzed through LTOMg.
Page 706
Automatic Session Tracing o Automatic Session Tracing uses a set of rules to determine when to turn on SQL trace for individual oracle sessions, using event 10046 level 12 trace. /* installing ltom */ # Running LTOM # os user must be member of dba group # db user required with full dba priv tar -xf ltom420.tar # the script creates "tom" user which can be dropped after installation cd ltom/tom_base/install # if default java cannot help, use Oracle's one (add it to .bash_profile) export PATH=/u01/app/oracle/product/11.2.0/db_1/jdk/jre/bin:$PATH ./autoinstall.sh
# uninstall rm -rf tom_base
/* TOOL USAGE */ # insertactive export TOM_HOME=/home/oracle/ltom/tom_base/tom cd /home/oracle/ltom/tom_base/tom ./startltom.sh # in the background: see the README file # running ltomg java -jar LTOMg.jar -i /home/oracle/ltom/tom_base/tom/recordings/profile/pro1284214950770.log java -jar -Xmx512M LTOMg.jar -i /u02/home/ltom/recordings/profile/pro1190754096675.log Using strace Strace is a utility that intercepts and records the system calls, which are called by a process, and the signals, which are received by a process. strace o /tmp/helloworld.out ./helloworld.sh strace aef Ttt o /tmp/date.out date strace p 1287 o /tmp/ora_pmon_orcl.out
Page 707

Oracle DBA Code Examples

Uploaded by

Document Informationclick to expand document informationDocument is edited to be a quick reference of code examples used to achieve specific Oracle DBA tasks. Document mainly demonstrates using SQL and PL / SQL code to achieve any task. It does not concentrate on using OEM to perform a task.

Document Informationclick to expand document information

Copyright:

Available Formats

Oracle DBA Code Examples

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Oracle DBA Code Examples

Uploaded by

Copyright:

Available Formats

Oracle DBA Code Examples

Version Date: 30 July 2011 Editor: Ahmed Baraka