MQ Monitoring agent: CTIRA_HOSTNAME vs SET AGENT NAME

Body

Often there are problems with MQ agent showing an unexpected behavior like frequent disconnections/re-connections (online/offline status in TEP) and/or MQ agent instances shown with a wrong format in TEP physical tree (eg only "MQSERIES" navigator item created instead of "MQSERIES - <qmgr_name>" as it should in case only one queue manager is being monitored).

Basically the problem is often related to the presence of CTIRA_HOSTNAME in the mq.ini (Unix/Linux) or KMQENV (Windows) file.
This variable cannot be used for MQ agents and should be removed because it may lead to unpredictable results.
The same results of using CTIRA_HOSTNAME can be achieved by setting SET AGENT NAME in the cfg file.
See:
http://www-01.ibm.com/support/knowledgecenter/SSTFXA_6.3.0/com.ibm.itm.doc_6.3/install/common_var.htm
CTIRA_HOSTNAME:
Used by many, but not all, agents to provide an alternate host name qualifier (subsystem:hostname:nodetype) for the published agent managed system name.
Used to remove a long network domain name (such as acme.us.com) from the default agent hostname.
Not honored by all agents. For some agents, CTIRA_HOSTNAME might cause unpredictable results in the Tivoli Enterprise Portal navigation tree..
 
Along with the usage of SET AGENT NAME statement on the .cfg on the agent's machine, you need to remember to use
KFW_TOPOLOGY_CLUSTER_LIST=AFF_MVS_MQM
in cq.ini for the TEPS.
See http://www-01.ibm.com/support/docview.wss?uid=swg21326517

So the wrong behaviors can be corrected by removing CTIRA_HOSTNAME from mq.ini/KMQENV files and use the SET AGENT NAME(xxxx) in the .cfg file for each affected MQ agent's instance.
You can start with the instances that are not working now and check that it gives more stable results.

Subscribe and follow us for all the latest information directly on your social feeds: