Hello Nigel,
I configured it in following way:
1. I created custom thread pool:
Thread Pool ID: thread-pool-tasks
Minimum Pool Size: 0
Maximum Pool Size: 50
Idle Timeout: 120
Number of Work Queues: 1
2. Next I created new config for jmsra:
asadmin create-resource-adapter-config --threadpools thread-pool-tasks jmsra
On my development server things worked rather ok (at least I haven’t noticed anything wrong). But after deployment on production server I observed this behavior (sometimes Glassfish didn’t read messages form one queue, sometimes from all of them – the same time Glassfish was able to send messages to these queues). For a while I was able to cause this behavior repeatable as I described before (but now after removing my jmsra’s config and applying it again, it stopped to be repeatable).
Both servers should have the same configuration (I can’t remember any difference in configuration).
My guess is that this behavior could be connected with restarting servers – I mainly noticed it after this kind of activities.
As a workaround I removed my configuration of jmsra and it looks that it works fine but my MDB don’t run in separate thread-pool.
If it would be helpful I can attach Open MQ dump done in moment when it was not working properly.
Greetings
Sławek
[Message sent by forum member 'slawek666' (slawek666)]
http://forums.java.net/jive/thread.jspa?messageID=352567