MQSeries.net :: View topic

WBI_user · Posted: Wed May 14, 2008 4:29 am Post subject: AMQ7469 daignostic help

I am using WMB 61 csd 5 and MQ 6021 on AIX

We experienced an intermittent production failure of a message flow which uses Java node to get all messages in group from a queue. The error was MQRC 2241. Because we use syncpoint, we can rerun the flow and it will always run fine the second time. This seems to us the all messages are there on the queue.
Everytime we had the failure, we noticed an AMQ6479 error a few minutes before the failure.

The explaination fir the error is:
"The log space for the queue manager is becoming full. One or more long-running transactions have been rolled back to release log space so that the queue manager can continue to process requests."

We are using circular logging with 10 primary and 5 secondary log of 160K each. For a total space of 2.4G. The total number of messages handled by the flow around that time was about 20K with a total of only about 140M. I understand that there must be other things using the log and more diagnostic work must be done.

So I need some hints from anyone to help me to narrow things down a little more.
1. It was MQ that reports the AMQ7469 which says "One or more long-running transactions have been rolled back to release log space" But it does not tell us which one. How does MQ select which one to roll back ? Will it select the MQ uses one first (i.e. MQ processes which may have open UOW) before any user application? If it had select an user application, should I expect to see error reported by the application ? I asked all application users who may have application running at the time of error. But no one siad that they have problem.
2. If it is not user application. MQ must have rolled back some MQ process or broker processes. But I cannot find any other errors in the syslog of my AIX system.

Any help ?