MQ agent crashing just after its start up

Body

It may happen that as soon as you start the MQ agent to monitor a QMGR, it crashes.
It happen in several customer's scenarios and I will report here what we found in a recent case.
We have reviewed the RAS1 log file of the MQ agent and found the following interesting portion when the agent crashed:

(5C07561F.0013-D:kmqcutsi.cpp,425,"PrintCallStack") Thread 40 (Thread 0x57f2db40 (LWP 1536)):

(5C07561F.0014-D:kmqcutsi.cpp,425,"PrintCallStack") #0 0x55577430 in __kernel_vsyscall ()

...

(5C07561F.001A-D:kmqcutsi.cpp,425,"PrintCallStack") #6 0x082a0adc in QuiesceHandler ()

(5C07561F.001B-D:kmqcutsi.cpp,425,"PrintCallStack") #7 <signal handler called>

...

(5C07561F.001F-D:kmqcutsi.cpp,425,"PrintCallStack") #11 0x0822f14f in ErrLogCollector::scanMessages(_IO_FILE*) ()

so when the signal handler was called the agent was working/executing code about the Error Log feature.
The suggestion I gave to the customer was to disable the Error Log feature by setting
ERRLOGCYCLE(0)
in the agent cfg file called <hostname>_mq_<qmgr_name>.cfg file in SET MANAGER statement.

See
https://www.ibm.com/support/knowledgecenter/SS3JRN_7.2.1.1/com.ibm.itcama.doc_7.2.1.1/kmq_userguide/r_setmanager.html?cp=SS3JRN_7.2.1.2
 

A restart of the MQ agent is needed. Once this was done the agent could be started successfully and it stayed up.

In this case customer had another MQ agent instance which was working fine even before we disabled Error Log feature. And so why it happens?
The problem may be related to the actual content of the AMQERR log for the QMGR monitored by the affected instance.
Maybe there are strange/un-handled chars or something in this QMGR that broke the MQ agent and they may not be present in the other QMGR.

In this case the customer was at V730 FP1 level for the MQ agent but there are some APARs in v730 FP2 that are related to Error Log feature.
I could find the following APARs but there may be others, so I would strongly suggest to install this maintenance.

   IV80959 WEBSPHERE MQ AGENT "ERROR LOG" WORKSPACE SHOWS WRONG DATE

   IV79995 ERROR LOG WORKSPACE IS BLANK IN RUSSIAN ENVIRONMENT

   IV79961 MQ AGENT FAILS TO PARSE THE MESSAGEID STRING ACROSS LINES IN THE ERROR LOG

Hope this helps
Walter
 

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