Clustring HA-JMS on Internal Client (non-master node)

anandnatraj Jun 18, 2007 1:22 AM

I have this following scenario on my clustering of JBoss with JMS.

1. I have two ears placed on a two servers (App-A and App-B). The APP-A and APP-B are clustered for the HA-JNDI.
2. These two ears need a Queue myQueue which they look for during deployment
3. Since I am using HA-JNDI, i placed the queue definitions on the all/deploy-singleton/jms of App-A, considering that as my master node.
4. When I start App-A, the myQueue is Deployed and the ear is also able to pickup the queue during Deployment. No issues.
5. Now I start App-B, the myQueue is not deployed here, and there is no need since the App-A is already having it. But when my ear is getting deployed, it is not able to find the queue JNDI in this server App-B, and the ear deployment is failed: (exception : queue Not Found)
6. By any chance I shutdown the App-A, then the myQueue is replicated in the App-B and the ear is bound with it.

My question is : Is this the way the HA-JNDI of JMS is going to work when we have the client on one of the non-master node itself? Or am I missing something?

Moreover can I exclude the HA-JNDI deployment of JMS and handle two individual myQueue on both the server?

Some help / eye opener is really appreciated.

Anandh

1. Re: Clustring HA-JMS on Internal Client (non-master node)

ratrask Sep 24, 2007 11:51 AM (in response to anandnatraj)

Did you ever resolve this? I am having the same issue. THe first node comes up fine, but when the second node comes up the first node gets a barrier stop message, and undeploys all of the queues. My belief is that both think that the other is the master. I continue to search, but so far have found no documentation on where to specify the default master node, or anything about avoiding this issue
Actions