ESB fails to consume when message count increases in queue

[tvraghavan]

We have configured our messaging against a oracle persistence store. We are using jboss-esb 4.8. sometimes when we load test our env, we observe that the messages start queueing up on the gateway queue or esb internal queue. Once we cross a threshold (approx 2000), the esb consumer completely stops consuming the message.

Strangely, during this scenario, we observe that even other esb service stops consuming messaging of other queue that are deployed under the same infrastructure.

The only way of resolution that we have observed is to delete the records from the oracle persistence table, flush the inflight messages from JMX-console (and also delete the data directory) and restart the esb server.

Is there any way to gracefully recover from such scenario ?

Do we need to tweak any out of the box parameter to improve the performance of the esb services.

Thanks

[tcunning]

Can you post your jboss-esb.xml?     What JMS provider are you using?

[tvraghavan]

Hi Tom,

We are using jboss-messaging as the JMS provider. Attached jboss-esb.xml and jbm queue defn.

Thanks

[tcunning]

Are the queries to select a message off the queue hanging?     What do you see in Oracle when this happens?

I'd go through the JBoss Messaging documentation and make sure that you are using whatever the correct oracle persistence settings are also.     Are you using the version of the ojdbc JAR?    Tried upgrading to ESB 4.10?