[jboss-user] [JBoss Messaging] - Re: LocalTx, ClientTransaction leak?

[ydzsidemiik]

I am using the JMS RA. I look up the JMS connection factory using java:/JmsXA, my performance is reasonable, and I can see in the JMX console that the pool has connections in it and that they are being used. From the beginning I based this design on the assumption that the JMS connections would be pooled. Otherwise, as you note, it wouldn't work very well at all.

I propose that the JMS resource adapter could not be allocating or holding references to JBM internal objects. The trace only goes as far as to identify JcaXAResourceWrapper.end as being on the path to where the actual allocations occur, although it cuts off there for reasons unknown to me but probably having to do with my profiler.

I propose that the allocation could not actually occur in JcaXAResourceWrapper.end because that method contains no allocations of any kind. In my initial investigation of the JBM source, I guessed that the following frames which we do not see may be MessagingXAResource.end -> MessagingXAResource.unsetCurrentTransactionId -> ResourceManager.createLocalTx, a plausible control flow to a function where the objects observed to be leaking are actually being allocated. However, I cannot say with certainty that this is the case.

As per your suggestion, I will try not closing my JMS sessions explicitly.


View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4118429#4118429

Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4118429