We have installed 2 nodes with Jboss 5.1.0.GA; include Jgroups 2.6.15.GA and JBM 1.4.3.GA ( With MySQL persistence )
Node A: Jboss
Node B: Jboss + MySQL
We need re-start Jboss Node after "Jboss Messaging Failover completed" happen in one of these nodes; due a network link have been down and restart again more later?
We are using one switch to connect the PCs where the Jboss nodes are running.
We have distributed a topic and a queue in the cluster.
We disconnected the link for the Node A, and we wait until "Jboss Messaging Failover completed" message is showed on the JMS Logs in the Node B.
We have observed in this point that the jbm_postoffice row for the queue in Node A in the Persistence Storage have been removed.
We have observed also that the jgroups have excluded the Node A from the cluster in the Node B view.
After that, we connected again the network link for the Node A.
We observed that jgroups re-join the cluster correctly, so the JBM in Node A or in Node B does not create the row for the queue for Node A again. So i think that the Node A cannot manage messages for the distributed queue anymore.
Does anyone know this behaviour? Is the re-start process the only action that we can do to recover the cluster after this failure?
Let me know for more information.
Thanks, Carlos Roman
I'm the in same situation: have you resolved this problem ?