You have WebSphere MQ running in an MQ cluster environment. When you issue a SUSPEND QMGR command, the repository manager (amqrrmfa) terminates and an FDC file is generated to report probe id XC130004 due to a SIGBUS. This only occurs if the locale is set to "ja" (CCSID 5050) prior to creating and starting the queue manager (qmgr). After the queue manager is stopped and restarted the problem does not reoccur.
The procedure for altering the queue manager CCSID was not followed. After the queue manager was created with CCSID 5050, the queue manager CCSID is altered (using runmqsc) to 943.
When the command "SUSPEND QMGR CLUSNL(CLNM_LIST)" is issued, the corresponding PCF command is translated from CCSID 943 to 5050, which causes corruption. This eventually leads to the SIGBUS error in amqrrmfa.
When the queue manager is restarted, all processes are initialized with 943 CCSID, thus preventing the CCSID conversion. This is the reason why no SIGBUS was observed after queue manager restart.
Resolving the problem
To modify an existing queue manager CCSID, follow this procedure:
1. Start MQSC commands by typing:
2. Display the existing queue manager CCSID, using the MQSC command:
display qmgr ccsid
3. Change the CCSID to the new CCSID with the MQSC command:
alter qmgr ccsid (new.ccsid)
where new.ccsid is the number of the new CCSID.
4. Stop MQSC commands by typing:
5. Stop the queue manager, and then restart the queue manager and any channels that it uses.
It should be noted that the issues described in this problem were because Step 5 was not followed after changing the queue manager CCSID and before the suspend qmgr was issued .