I'm currently using JBoss 3.2.1 and trying to develop a clusterable system, that uses JMS to enable async. processing. I understand that clustering is not possible, however I have a couple of questions:
1. Is it acceptable to have each node of the cluster running a queue with the same name (e.g. test1) in order to get each cluster to produce and consume messages locally in a P2P manner?
2. If yes, would configuring the persistance manager of each queue to point to a common database table or file system directory work?
What I'm trying to achieve is to have the rest of the nodes to continue the processing of a node that would potentially crash.
No this won't work.
But you can hotdeploy jms onto a new server to
pickup where the crashed server left off,
if it uses the same db. It doesn't work
for durable subscriptions unless you already
have them specified in jbossmq-state.xml
There are plans to use the HASingleton to make
this cold failover automatic in 3.2