Problems in failover

[dediana]

Hello,

I am trying to invoke a simple session bean in a cluster. I am running JBoss-3.2.3 in Win2k machines. I have seen other people with the same problem in this forum but none of the solutions worked here.
The two JBoss instances are able to find each other. The client can call the bean, everything works fine. But if the instance that the client is accessing is shut down, the client get exceptions with message 'Connection refused'.
I am using port 1100 on the client. I have all jars of the JBoss client directory in my client classpath. My machines are multihomed, but bind_addr makes no difference. I can see in jmx-console that HAJNDI is started (State = 3) and JRMPHA is started (but ServerAddress=0.0.0.0 and RMIObjectPort=0. Is it right). In the stack trace I can see JRMPInvoker being invoked, but not JRMPInvokerHA (I read in another thread Sacha saying that this is a problem - http://www.jboss.org/index.html?module=bb&op=viewtopic&t=33561).

Any hints?

[dediana]

More information:

Everytime I deploy the ear I get the message "*** EJB 'Agent' deployed as CLUSTERED but not a single clustered-invoker is bound to container ***".

Thanks.

[dediana]

I think I solved the problem.

When I stopped using local interfaces the message disappeared and the failover worked properly. The only problem is that now I have entity beans with remote interfaces exposed. Is it a big problem?

And another problem was solved turning the private fields in my SFSB public. Is it a problem?

Thanks.