I have an MDB in an EAR and deploy that to the farm directory of a cluster. Now the MDB is available on every node. When I send a message to JMS every node will process the message with it's copy of the MDB. This even happens if I annotate the MDB with @Cluster.
This behaviour is not useful, methinks. I cluster MDBs to distribute load. A message should only be delivered to one instance of the MDB. That means the deployed MDBs should be managed as a clustered pool instead of a local pool. Am I completely off tracks and this problem (load-balancing work that is JMS triggered) is normally solved in a different way? Is this just a configuration thing: what do I have to configure? I am a bit lost here and any suggestions are greatly appretiated.