Tibco Ems Cluster - JBoss MQ

tzman Sep 15, 2009 4:00 PM

JBoss - 4.2.3.GA

We are trying to deploy an application that was previously not clustered into a clustered environment. The application has heavy usage of tibco ems with both consumers and producers. When initially configuring the system, we used the wiki document on integrating JBoss with TibcoEms.

The problem now seems to be in enabling the singleton jms for the consumers. I am not entirely sure which portion of the configuration would need to go into deploy-hasingleton/jms versus deploy/jms. The setup for clustered jms suggests that all of the configuration go into deploy-hasingleton/jms which causes only the first node started to be able to find the queues. Adding the configuration to deploy/jms allows everything to start appropriately, however we then have multiple consumers trying to connect to tibco ems, which is not expected.

The clustering documentation states that we should configure a shared persistence provider for the queues. Is this used to enable JBoss to determine which of the consumers ( node ) is active? We did not configure the persistence for jms as tibco is handling this functionality.

I am not quite sure what additional information to provide. Is this supported? Should the configuration be split between the deploy-hasingleton/jms and deploy/jms? A pointer to which configurations I need to modify/provide would be greatly appreciated and my apologies if this does not make perfect sense.

Thanks in advance.

1. Re: Tibco Ems Cluster - JBoss MQ

brian.stansberry Sep 24, 2009 6:58 PM (in response to tzman)

So it sounds like you want your producers and consumers to only be deployed on the HASingleton master. So, seems like you should:

1) Follow the instructions on http://www.jboss.org/community/wiki/IntegrationWithTibcoEMS

And it sounds like you've done that and gotten it working.

As you've discovered, you don't want to put the Tibco configuration stuff discussed on the wiki in deploy-hasingleton. It needs to be deployed on every server.

2) Delete the contents of deploy-hasingleton/jms . That stuff is relevant to using JBoss MQ as your JMS server, which you're not doing. The wiki page talks about deleting JBoss MQ, but discusses it in terms of the "default" config; by deleting deploy-hasingleton/jms you're doing the equivalent thing to the "all" config.

3) Deploy your producers and consumers in deploy-hasingleton.

The "shared persistence provider for queues" discussion is relevant to JBossMQ. As you said, Tibco is handling this.
Actions