load balancing in Jboss 3.2.1

[chuang88]

Hi,

I have a cluster using Non-Multicast JavaGroup with the following hosts:
RedHat,
Mandrek
Win2K.

I deployed Jboss's template on RedHat and
could access its bean on Mandrek and RedHat.
But all the beans or it's methods were created/invoked on the RedHat. LoadBalancing didn't
seem working on my cluster.
I did saw the following in the server.log on the Mandrek.

DefaultPartition] pre methodName: DistributedReplicantManager._add
2003-05-30 13:13:55,431 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] handlerName: DistributedReplicantManager methodName: _add
2003-05-30 13:13:55,431 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] Handle: DistributedReplicantManager._add

By the way, I couldn't access the beans from the Win2K and got the error ,avax.naming.NameNotFoundException: ejb/TestSession.

jboss.xml and cluster-service.xml are attached.

Thanks

[slaboure]

Your cluster-service.xml file is strange as you still seem to use the multicast settings (in XML format). Which version of JBoss?

Are you sure that all know discover each other? What can you see in the logs of the first two nodes once you start the last node?

Cheers,


sacha

[chuang88]

I was using JBoss3.2.1 with Tomcat.
I could see 'cluster view' with all three hosts in the
server.log of RedHat and Mandrake, but not Win2K.

Thanks

[slaboure]

Can you check you don't have any firewall that would prevent your node from communicating (RH has one by default). that's strange as you say that two of the nodes say they see the w2k node

[chuang88]

I'll check the firewall on the RedHat. In the meantime,
does the following messages in server.log indicate
that clustering and loadbalancing were working?

DefaultPartition] pre methodName: DistributedReplicantManager._add
2003-05-30 13:13:55,431 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] handlerName: DistributedReplicantManager methodName: _add
2003-05-30 13:13:55,431 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] Handle: DistributedReplicantManager._add


Thanks

[slaboure]

yes, it may for example mean that the nodes were exchanging information about which EJB are clustered on the nodes.