Shutdown cluster node: Timing problem?
kevin.lohmann Nov 11, 2010 5:48 AMHi folks,
I'm using JBoss 5.1.0 EAP and I've set up a cluster with 2 nodes.
From time to time, when I shutdown one node (B with ip-address XXX), the other node (A, with ip-address YYY) logs this messages:
2010-11-11 11:14:08,565 ERROR [org.jgroups.blocks.ConnectionTable] failed sending data to XXX:7900: java.net.SocketException: Socket closed 2010-11-11 11:14:08,628 INFO [org.jboss.messaging.core.impl.postoffice.GroupMember] org.jboss.messaging.core.impl.postoffice.GroupMember$ControlMembershipListener@16d478b got new view [YYY:55066|2] [YYY:55066], old view is [YYY:55066|1] [YYY:55066, XXX:55200] 2010-11-11 11:14:08,628 INFO [org.jboss.messaging.core.impl.postoffice.GroupMember] I am (YYY:55066) 2010-11-11 11:14:08,628 INFO [org.jboss.messaging.core.impl.postoffice.GroupMember] Dead members: 1 ([XXX:55200]) 2010-11-11 11:14:08,628 INFO [org.jboss.messaging.core.impl.postoffice.GroupMember] All Members : 1 ([YYY:55066]) 2010-11-11 11:14:10,628 INFO [org.jboss.ha.framework.interfaces.HAPartition.lifecycle.cluster] New cluster view for partition cluster (id: 2, delta: -1) : [YYY:3045] 2010-11-11 11:14:10,628 INFO [org.jboss.ha.framework.server.DistributedReplicantManagerImpl.cluster] I am (YYY:3045) received membershipChanged event: 2010-11-11 11:14:10,628 INFO [org.jboss.ha.framework.server.DistributedReplicantManagerImpl.cluster] Dead members: 1 ([XXX:3045]) 2010-11-11 11:14:10,628 INFO [org.jboss.ha.framework.server.DistributedReplicantManagerImpl.cluster] New Members : 0 ([]) 2010-11-11 11:14:10,628 INFO [org.jboss.ha.framework.server.DistributedReplicantManagerImpl.cluster] All Members : 1 ([YYY:3045])
server.log from node B (the one, that is shutting down):
2010-11-11 11:14:08,669 ERROR [org.jgroups.blocks.ConnectionTable] failed sending data to YYY:7900: java.net.SocketException: Socket closed 2010-11-11 11:14:10,669 ERROR [org.jgroups.protocols.UNICAST] XXX:55200: sender window for YYY:55066 not found
After restarting node B the two nodes build a cluster again without any error or warning messages.
I am wondering if these errors are just a timing problem or if there is a really problem I should carry about.
I would appreciate any hints.
Thanks in advance,
Kevin