Workaround for Syncronization Bug in 6.1 EAP

[malichtenberg]

In the process up upgrading our 7.1.1 instances to 6.1 EAP, we experienced periodic NullPointerExceptions at ServerLocationImpl:546 when a high volume of web threads were attempting to send messages to a queue. We found bug [HORNETQ-1130] Synchronize ServerLocatorImpl properly to avoid race condition when concurrently creating and closing Cli… that looks to be the culprit, and was fixed in the 2.3.0 CR2 release of HornetQ. When I tried to update our hornetq module to the 2.3.0 Final, we ran into the same issue mentioned here - Re: How to upgrade HornetQ version in JBoss 7

Has 1) there been any work done on a backward-compatible version of hornetq that we could employ or 2) is there a configuration workaround we could use that would block creation of new concurrent pool connections to throttle the workload? Thanks so much

[jbertram]

Are you asking about what can be done on your AS 7.1.1 instances?  I ask because EAP 6.1 ships with HornetQ 2.3.1.Final (which has the fix for HORNETQ-1130) so the obvious solution is just to complete your move to EAP 6.1.  If you are hitting an NPE in EAP 6.1 then that is a different issue from HORNETQ-1130.

[malichtenberg]

I'm using 6.1 EAP Alpha (7.2) which shipped with HornetQ 2.3.0 CR1. Apologies for not being clear.

[jbertram]

So why not use EAP 6.1 Final instead of the beta or alpha you're using now?

[malichtenberg]

We're coming from the community-edition and as I understand the Alpha release is the only 6.1 EAP version we can use in production under the community license.

Apologies for the lag in response. We switched to using the JmsXA pooled connections, and our problems went away. This must bypass the bug referenced as it reuses connections vs creating them for each of our web threads. Thanks for your help!