This could be due to a bug changed in JGroups 2.2.5 (from the release notes):
- Fixed bugs #943480 and #938584. STABLE handles SUSPEND_STABLE and
RESUME_STABLE events, and STATE_TRANSFER
sends SUSPEND_STABLE before fetching the state and RESUME_STABLE after
receiving the state, to prevent
message garbage collection from going on during a state transfer
- Fixed bug #938584 (NAKACK.retransmit problem)
The other JIRA issue I recently fixed is http://jira.jboss.com/jira/browse/JGRP-108. This is in JGroups 2.2.9 (not yet released).
The problem with JBoss 3.2.2 is that it comes with JGroups 2.2.0, which is not compatible with JGroups 2.2.5, so you cannot simply replace JGroups.
You will probably need to upgrade, here's a list of JGroups versions shipped with JBossAS: http://wiki.jboss.org/wiki/Wiki.jsp?page=JGroupsVersionInJBossAS
Thanks for the Prompt reply.
Can you please educate us on the impact of this bug? Could this be result in server node getting out of cluster?
Currently we see that the server nodes come out of cluster environment and we are not sure what could be the problem. Can you please some provide us the pointers so that we can investigate more on this issue.