We are running wildfly-8.0.0.Final and have 2 servers that are supposed to be clustered and a 3rd server that makes remote calls to an ejb deployed on the cluster. We also have our own implementation of the org.jboss.ejb.client.ClusterNodeSelector interface and is specified in the jboss-ejb-client.xml. Our ClusterNodeSelector gets initialized, but it is never called nor is the default RandomClusterNodeSelector. I do see a log message from the LocalEJBReceiverPreferringDeploymentNodeSelector but it always selects the same node in the cluster. This all worked under JBoss 7, but it looks like I must be missing something in our WildFly configuration. We are using a slightly modified standalone-ha.xml which I have attached.
One thing I have noticed is that on JBoss 7 I used to see log messages like this:
2014-01-14 16:24:30,180 INFO [org.jboss.as.clustering] (Incoming-1,shared=udp) JBAS010225: New cluster view for partition ejb-atp-matching (id: 1, delta: 1, merge: false) : [ape-atpmatcher01/ejb-atp-matching, ape-atpmatcher02/ejb-atp-matching]
2014-01-14 16:24:30,182 INFO [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-1,shared=udp) ISPN000094: Received new cluster view: [ape-atpmatcher01/ejb-atp-matching|1] [ape-atpmatcher01/ejb-atp-matching, ape-atpmatcher02/ejb-atp-matching]
I don't see any "cluster view" type log messages now when we start wildfly, but I am seeing warning messages like this:
2014-04-21 14:43:22,753 WARN [org.jboss.as.ejb3] (MSC service thread 1-6) JBAS014267: The <clustering xmlns="urn:clustering:1.0"/> element will be ignored.
I'm assuming this is coming from the jboss-ejb3.xml deployment descriptor which has not changed since JBoss 7 and just looks like this:
It looks like the cluster is not getting initialized correctly, but I'm haven't been able to spot what is wrong.
wf800-xxi-ha.xml 26.3 KB