Best Practices:

Getting Started With Informix

Connection Manager

Thomas Beebe
Advanced DataTools Corporation
[email protected]
 About This Talk
• Created to give an entry point for getting
going with connection manager
• Supplement the rather sparse documentation
on the topic
• Document some of the confusion I ran into
configuring this for clients
• Assistance provided from others actively
using connection manager in production
environments

2
 What Is Connection Manager

• Introduced in 11.5 (revamped in 11.7)

• Bundled with the engine, also included with the csdk
• Standalone program that runs and passes connections
to the correct server or group
• Can be used as a central broker for connections
• Can be run solo or with a group of CM servers working
together
• Intended for HA, but very useful in any replicated
environment
• OAT module to monitor Connection Manager

3
 Topology

Client Connection
Manager

Client
Group 1 Group 2

Stand
Primary Secondary
Alone

4
 Setting Up Access

• Can be as simple as one server and one CM

– Client -> Connection Manager -> Instance
• It can be configured as a replication set
– Client -> Connection Manager -> Primary + HDR + RSS
• It can be configured with multiple groups each with their own
rules
• Clients connect to the CM on the listener that will tell the
manager where to redirect them and what rules to use
• If the connection manager server is not trusted to the Informix
server, use an encrypted password to authenticate

5
 Relevant Files
• SQLHOSTS
• Specifies the groups, also controls the CM
listeners
• $INFORMIXDIR/bin/oncmsm
• Connection Manager Binary
• $INFORMIXDIR/etc/cmsm.cfg.sample
• Sample config file, there are several versions
• $CMALARMPROGRAM
• Program, defined in configs that triggers on a
failover
6
 ONCONFIG Parameters
• DRAUTO – Controls failover
processing, 3 means it will rely on the
connection manager to initiate failovers.
• HA_FOC_ORDER – Default order
(HDR,RSS,SDS) to fail over servers.

7
 oncmsm

• Startup:
• $INFORMIXDIR/bin/oncmsm –c $INFORMIXDIR/etc/cmsm.cfg
• Optional environmental variable $CMCONFIG
• Commands:
• oncmsm –k –c <config file>
• Shutdown
• oncmsm –r –c <config file>
• Reload with updated config file

8
 Basic Example - SQLHOSTS
#Instance Config port 9088
instance1_tcp onsoctcp host_ip sqlexec
#Connection Manager Listener
report_group onsoctcp host_ip 9090

9
 Basic Example – cmsm.cfg
NAME samplecsm
LOGFILE
${INFORMIXDIR}/tmp/cmsm.log

CLUSTER samplecluster {
INFORMIXSERVER instance1_tcp
SLA report_group
DBSERVERS=primary
}
10
 Base CM Config Sample
NAME cm_1
LOGFILE
${INFORMIXDIR}/tmp/cmsm.log
LOG 1
CM_TIMEOUT 300
<Connection Info>

11
 cmsm.cfg Parameters
• NAME – Must be unique across the cluster

• LOG – log level, 1 is on

• LOGFILE – Path to CM log

• CM_TIMEOUT – Number of seconds to wait for a response before promoting the next
highest ranked connection manager. (60 default)

• EVENT_TIMEOUT – Number of seconds to wait before failover occurs of Informix servers.

If a secondary triggers ‘primary offline’ it will also trigger failover before the timeout (default
60)

• SECONDARY_EVENT_TIMEOUT – Seconds to wait before disconnecting from a

secondary (Default 60)

• SQLHOSTS – If it should use a local, remote or both SQLhosts files to find instances.
(Default local + remote)
• LOCAL_IP – Optional, can be used to tie CM to a specific IP address to listen for database
status changes

• MACRO – Used to create variables to be used in other parts of the script

12
 Connection Types
• CLUSTER – Group or selection of
servers to connect to that support HDR
failover
• GRID – ER Grid to connect to
• REPL_SET – ER replicate set to
connect to
• SERVERSET – Unrelated servers that
do not use failover
13
 CLUSTER Example
CLUSTER cluster_1
{
INFORMIXSERVER repl1_tcp
SLA report_1 DBSERVERS=(PRI,HDR) \
POLICY=WORKLOAD
FOC ORDER=ENABLED \
PRIORITY=1
CMALARMPROGRAM $INFORMIXDIR/etc/CMALARMPROGRAM.sh
}

14
 Cluster Example
CLUSTER cluster_1 – Unique name of cluster, needs to be identical on
other connection managers
{
INFORMIXSERVER repl1_tcp -- The sqlhost entry (group or server) the
connection manager will listen on
SLA report_1 \ -- the SLA is for the report_1 group in sqlhosts
DBSERVERS=(PRI,HDR) \ -- This is the order it will
maintain
POLICY=WORKLOAD – This is the type of SLA policy it users
FOC ORDER=ENABLED \ -- Says to use the failover order above
PRIORITY=1 – Says this connection manager is the first
one to handle failover for this SLA
CMALARMPROGRAM $INFORMIXDIR/etc/CMALARMPROGRAM.sh –
If failover fails after 8 attempts it calls this program.
}

15
 INFORMIXSERVER
• Works with all 4 types of connection
• Specify the group of servers or
standalone server this SLA should
service
• This is what the connection manager
will connect to when it comes up to
establish the replication status

16
 SLA
• Service Level Agreement
• This is the directive of how a connection
manager should treat a particular group
of servers
• Any linked connection mangers should
have similar settings and the same
name for a group
• Each SLA will have its own listener port
17
 SLA - DBSERVERS
• List of servers to connect to, and the
order to connect to them in
• Can use server names, group names,
server aliases, server types (HDR, SDS,
ANY)

18
 DBSERVERS - Cluster
Keywords
• PRI, PRIMARY
• HDR – Secondary
• SDS – Shared disk secondary
• RSS
• ANY

19
 SLA - MODE
• Redirect – (Default) this will redirect the
client directly to the server, only works with
versions later then CSDK 3.0 and JDBC
3.5.1
• PROXY – Will pass all data through the
connection manager directly, allows for
older clients to be supported. Also use this
if the client cannot directly access the
Informix server.
20
 Topology - Redirect

Client Connection
Manager

Client
Group 1 Group 2

Stand
Primary Secondary
Alone

21
 Topology - PROXY

Client Connection
Manager

Client
Group 1 Group 2

Stand
Primary Secondary
Alone

22
 SLA - USEALIASES
• On – Default, this will add any entries in
DBSERVERALIASES into the mix.
• OFF – Only will use DBSERVERNAME
none of the aliases

23
 SLA - WORKERS
• Numerical Value – Default is 4.
• Used to define how many worker
threads the CM is allocated

24
 SLA - POLICY
• WORKLOAD – (Default) – Assigns the work to the least busy server
at the time

• ROUNDROBIN – Rotates between all of the available servers

• FAILURE – Requests pointed to the server with the fewest apply

failures. (Replset and GRID only)

• LATENCY – Redirects to the server with the lowest transaction

latency (Replset and GRID only)

• SECAPPLYBACKLOG:<num of pages> -- Stops sending requests

to the secondary after it exceeds the number of pages in a backlog.
CLUSTER only. (version 12.10xc2 or 11.70xc8 required)

25
 FOC
• Is used as a stand alone element inside
a connection type to define how failover
occurs
• Also used to specify the priority
between servers

26
 FOC - ORDER
• If not defined, the primary server’s
HA_FOC_ORDER parameter is used
• Default if neither are set is SDS, HDR,
RSS
• If enabled it will use the order defined
by the DBSERVERS in the SLA
• ENABLED – Means the connection
manager will allow failover
27
 FOC - PRIORITY
• Defines the priority between connection
managers.
• Must be a positive number, the lower
the number the higher the priority
• Required for CLUSTER types

28
 FOC - TIMEOUT
• Additional time before a failover occurs
• Adds to the value of EVENT_TIMEOUT
• Defaults to 0

29
 SQLHOSTS

• Connection manager will use SQLHOSTS like any other

Informix tool
• By default will read the local SQLHOSTS and if a server is not
found it will probe the remote sqlhosts for other relevant hosts
• Will use the INFORMIXSERVER directive in the SLA section
to determine the primary server or group to connect to
• The SLA name will be the connection name for the connection
manager listener for that SLA.
• Best practices is to always use groups when doing replication
rather then individual servers

30
 Sample Config
Name Of This Connection Manager
NAME connection_manager_1
LOG 1
LOGFILE $INFORMIXDIR/tmp/my_cm1_log.log
EVENT_TIMEOUT 20
Cluster Name, Shared Among Connection Managers
CLUSTER primary_cluster
{ Informix Group/Server To connect to
INFORMIXSERVER cluster1
SLA report_1 DBSERVERS=(PRI,HDR) \
POLICY=WORKLOAD Connection Manager Listener
FOC ORDER=ENABLED \
PRIORITY=1
}

31
 SQLHOSTS - Example
#HDR Pair of Servers
cluster_1 group - - c=1,e=repl2_tcp
repl1_tcp onsoctcp server1 9088 g=cluster_1
repl2_tcp onsoctcp server2 9088 g=cluster_1

#Group of connection managers that service the report

SLA
report group - - c=1,e=report_2
report_1 onsoctcp server1 10088 g=report
report_2 onsoctcp server2 10088 g=report

32
 Application Set Up
SQLHOSTS:
report group - -
c=1,e=report_2
report_1 onsoctcp server1 10088
g=report
report_2 onsoctcp server2 10088
g=report

• Set up your application to connect to report_1

33
 Oncmsm Log - Startup
22:45:41 listener report initializing
22:45:41 listener report_rr initializing
22:45:41 listener current_rss initializing
22:45:41 listener proxy_rss initializing
22:45:41 Listener report_rr DBSERVERS=(HDR,RSS) POLICY=ROUNDROBIN is active with 4
worker threads
22:45:41 Listener current_rss DBSERVERS=RSS
POLICY=SECAPPLYBACKLOG:5500+WORKLOAD is active with 4 worker threads
22:45:41 Listener report DBSERVERS=(HDR,RSS) POLICY=WORKLOAD is active with 4
worker threads
22:45:41 Listener primary_cm DBSERVERS=primary is active with 4 worker threads
22:45:41 Listener proxy_rss DBSERVERS=RSS POLICY=WORKLOAD MODE=PROXY is
active with 4 worker threads
22:45:42 Connection Manager successfully connected to maytcp
22:45:42 The server type of cluster aos_cluster server furytcp is Primary.
22:45:48 The server type of cluster aos_cluster server fury is Primary.
22:46:04 Connection Manager started successfully

34
 Oncmsm Log – Startup Cont

22:46:04 Connection Manager successfully connected to coulsontcp

22:46:04 Cluster aos_cluster Arbitrator FOC ORDER=ENABLED PRIORITY=1
22:46:04 Connection Manager successfully connected to furynosql
22:45:42 The server type of cluster aos_cluster server furytcp is Primary.
22:45:48 The server type of cluster aos_cluster server fury is Primary.
22:46:04 Connection Manager started successfully
22:46:04 Connection Manager successfully connected to coulsontcp
22:46:04 Cluster aos_cluster Arbitrator FOC ORDER=ENABLED PRIORITY=1
22:46:04 Connection Manager successfully connected to furynosql
22:46:04 Connection Manager successfully connected to furytcp
22:46:04 Connection Manager successfully connected to furyrest
22:46:35 CM cm_1 arbitrator for aos_cluster is active
22:46:35 Cluster aos_cluster Arbitrator FOC ORDER=SDS,HDR,RSS PRIORITY=1
TIMEOUT=0

35
 Oncmsm log - Connections
23:04:23 SLA report_rr redirect SQLI client from
10.10.20.60 to maytcp may.10088
23:30:25 SLA report_rr redirect SQLI client from
10.10.20.71 to furynosql fury.10098
23:30:25 SLA primary_cm redirect SQLI client from
10.10.20.63 to furynosql fury.10098

36
 onstat –g cmsm

• Onstat –g cmsm
• Onconfig params

37
 Best Practices
• Use groups rather then individual servers
• Run more than one connection manager
• Make sure the connection managers are on
different servers from the instances.
• If running PROXY mode make sure to have the
resources on the CM to handle the data.
• Make sure to set up applications to the use the
group of connection managers, or at least failover.
• Make sure applications reconnect with at least a
short delay.

38
 Things To Be Cautious Of
• Split Brain
• Listeners missing info
• Alias issues
• Missing Trusted status
• Make sure DBSERVERNAME is the
TCP Port

39
 Split Brain
• Two Primary Servers on the same network
• Need to restore one of the servers and re-establish
HDR
• Reduce the chance of this situation by having a
reliable network connection
• Can use the Connection Manager Alarm Program
to shutdown the Primary server if the Secondary
cannot be reached AND can’t get to the network –
See article on IBM developerWorks:
https://www.ibm.com/developerworks/community/blogs/informix_admins_blog/e
ntry/preventing_hdr_split_brain_scenaro

40
 Auto Failover and Network Loss
Connection
• What if the PRIMARY server loses Manager

the network connection? Primary Secondary

Server Server

• Connection Manager will promote Connection

Manager

the secondary server, incorrectly

Primary Primary
assuming that the primary is down Server Server

Connection
• When network connectivity is Manager

restored, there will be TWO Primary Primary

primary servers – Split Brain

Server Server

41
 Listeners Missing Info
• Will get a -930 error
• Can be caused due to missing entries
from sqlhosts on the connection
manager server
• Can also get errors if the port is in use,
each SLA needs a unique port

42
 Alias Issues
• By default will probe all of the systems in
the cluster for DBSERVERALIASES
values
• Can be disabled by USEALIASES
• If it sees tcp connections that appear valid
but cannot be reached by client, such as a
private replication interface, it will still try to
offer them for REDIRECT connections
43
 Missing Trusted Sources
• If one of your servers does not trust the
connection manager server it will not be
able to connect successfully.
• This can be resolved by using the
password management process.

44
 Encrypted Password
• Set up a file defining the servers and
what passwords to use
• Use onpassword to encrypt the file
• Stored as
$INFORMIXDIR/etc/passwd_file
• Uses a key to encrypt/decrypt to make
changes in the future

45
 More Resources
• Connection Manager Manual Examples
https://www.ibm.com/support/knowledgecenter/SSGU8G_12.1.0/co
m.ibm.admin.doc/ids_admin_1437.htm

• Informix Replication Technologies:

http://www.ibm.com/developerworks/data/library/techarticle/dm-
0807geib/

• Andrew Ford’s Blog – Setting up HDR:

http://www.informix-dba.com/2010/08/informix-hdr-will-save-your-
butt.html

• Connection Manager Alert List

https://www.ibm.com/support/knowledgecenter/SSGU8G_11.70.0/c
om.ibm.adref.doc/ids_adr_1148.htm

46
 Questions?

Send follow-up questions to

[email protected]
 Next Webcast
Informix Best Practices
• Informix Auditing by Mike Walker
– Thursday, July 27, 2017 at 2:00pm EST
• Stay tuned for upcoming announcements about the next
set of webcasts

Please register for each webcast here at:

http://advancedatatools.com/Informix/NextWebcast.html

Informix Best 48
Practices
 Upcoming WAIUG Meeting
• Aug 8th 2017 – Tysons Corner VA
• 8:30-1pm
• What HCL Brings to the Informix World
– Stuart Litel
• Enterprise Replication – Shawn Moe
• Free
• http://www.waiug.org to RSVP
49
 Informix Training in 2017
September 11-14, 2017
Advanced Informix Performance Tuning

Sept 18-21, 2017

Informix for Database Administrators

• All courses can be taken online on the web from your

desk or at our training center in Virginia
• We guarantee to NEVER cancel a course and will teach
a course as long as one student is registered
 Informix Support and Training from the Informix Champions
Advanced DataTools is an Advanced Level IBM Informix Data Management Partner, and has been an authorized
Informix partner since 1993. We have a long-term relationship with IBM, we have priority access to high-level support
staff, technical information, and Beta programs. Our team has been working with Informix since its inception, and
includes 8 Senior Informix Database Consultants, 4 IBM Champions, 2 IIUG Director’s Award winners, and an IBM
Gold Consultant. We have Informix specialists Lester Knutsen and Art Kagel available to support your Informix
performance tuning and monitoring requirements!

• Informix Remote DBA Support Monitoring

• Informix Performance Tuning
• Informix Training
• Informix Consulting
• Informix Development
Free Informix Performance Tuning Webcast replays at:
http://advancedatatools.com/Informix/Webcasts.html

Call: (800) 807-6732 x101 or email: [email protected]

web: http://www.advancedatatools.com
 Thank You

Thomas Beebe
Advanced DataTools Corporation
[email protected]

For more information:

http://www.advancedatatools.com