MQSeries.net :: View topic

vlucian · Posted: Fri Jan 28, 2011 7:53 am Post subject:

Hi
I have the following situation:
- MQ 6.0.2.5 on AIX 5.3
- queue manager file system goes full while application was reading/writing from queues
- I increase the file system and everything return to normal
- 1 hour later I try to save one queue (lets name it Q, it is a q where messages gather for later backup on disk) with WMQTool and I receive an error; I try to reconnect but the qmanager is dead (ended preemptively)
- I restart the qm; Q appear to be damaged, delete and recreate is the only solution
- first fdc contains AMQ6090: WebSphere MQ was unable to display an error message 7D9, probe id ZX092016
- next fdc AMQ9510, probe id RM220005
- then AMQ5008:An essential WebSphere MQ process 116674
(amqrrmfa) cannot be found and is assumed to be terminated , probe id ZX005025;
My questions:
- was it normal to die like this?
- what could be the cause? my supposition is that Q get damaged when the file system went full (maybe some uncommited 'put' operations?) and when I was reading it with WMQTool something went wrong and the qmanager ended. does it make any sens? anyway, it's strange because: 1. after increasing filesystem and qmanager returned to normal, more messages were succesufully written in Q.
2. before recreating the Q, I saved its q file and I identified there the very last message that coud be written before "file system full" event.
3. then, I looked with notepad in the .mqt file created by WMQTool (when qm died) and guess what ... last message that could be saved was the same message
- where else should I look?
Thanks.