Clustering - Load Balance Fails

[jtsimikas]

I am running JBOSS 3.0.6 w/ clustering enabled.
run -c all

I have built a network w/ 2 PCs for testing purposes. I can deploy an ejb to the farm directory of either server and it gets undeployed and deployed to the other server.

I built a test app that executes one of the clustered Session Beans. When I do this I always hit the same server. The Round Robin policy doesn not seem to work. It never load balances.

I am NOT reusing the remote reference, I obtain a new one every time.

Does anybody have any idea why the load balancing does not work as I would expect.

Also, the fail over works but not all the time. I think this depends on which server is shutdown.

My jndi.properties file looks like this:
java.naming.factory.initial=org.jnp.interfaces.NamingContextFactory
java.naming.provider.url=
java.naming.factory.url.pkgs=org.jboss.naming:org.jnp.interfaces


Please help

John

[slaboure]

known behaviour. Fixes:
- use the 3.2.x branch, or
- don't re-acquire a new remote for each call, or
- use the random load-balancing policy

[jonhaug]

> known behaviour. Fixes:
> - use the 3.2.x branch, or
> - don't re-acquire a new remote for each call, or
> - use the random load-balancing policy


I have the same problem. (At least I need to force two threads in a client test program to use different servers in order check transactional behaviour.) Anyway, I cannot find your random load-balancing policy. Checking the jar file jbossha.jar, I find no such class.

Maybe I can get this to work anyway. But I just want to know what you really meant here.

--
Jon Haugsand

[slaboure]

It is in 3.2 only. However, if you client is not in the same VM as the server, it will work.


sacha