We have 2 JBoss servers (5.1.0GA) that operate as a cluster.
We have changed the default partition name and the partion udp group via the system properties on startup.
In our one environment all works fine. The first jboss server immediately detects the second one and the cluster view is updated when the second jboss box has started. I can see a log entry stating so and I can verify it in the jmx-console.
In another environment with an identical setup (except for the partion name and udp group) the first jboss server NAKACK discards message from non member with "sender x.x.x.x not found in xmit_table" "range is null" and then after about 1 hour RPCManagerImpl logs a "Received new cluster view: MergeView and the second jboss box joins the cluster successfully.
It happens consistently in this environment.
What is causing this long delay ?