MQSeries.net :: View topic - MQ5.3 and "pulsing" of message batches in error

ColinRenouf

I hope someone can help...

We have a live 12-way AIX4.3.3 RS/6000 with WMQ5.3 with no CSDs. The setmqcap command was not run (which itself is a problem but updating a live box requires proof of a fix in our business) and the environment has some funny quirks. I believe that CPU 0 is getting most of the processing but haven't proved that conclusively.

The problem is that the reply transmission queue for messages back from ou zSeries box gets a build up of messages without the RS/6000 taking them off, and a "pulsing" of responses happens where all of the messages suddenly get taken off the queue in a large batch. This greatly affetcs response times.

IBM haven't got any ideas, and I have a theory as to the CPU0 being the only one used because of a lack of capacity units, but this isn't proven.

Any ideas?
_________________
Colin Renouf