Cluster nodes not discovered

[sureshrk19]

Hi,

I'm trying to setup 2 node cluster;
Jboss 4.2.3
Jdk 1.5
APache 2.2 with mod_jk 1.2.x

On local desktop everything works fine; I configured 2 nodes (all and copy_of_all) on single machine and configured cluster. Application (.war file) is copied to '/all/farm/' directory and the same is notified to second node on startup. Loadbalancing, fail-over all works fine.

Now, I'm trying the same setup on remote linux boxes i.e., 2 Jboss nodes on 2 different boxes. In this case, the nodes are not recognizing the other cluster node.

On both the nodes, I see the following messages...

---------

15:18:02,378 INFO [DefaultPartition] Number of cluster members: 1
15:18:02,378 INFO [DefaultPartition] Other members: 0
15:18:02,378 INFO [DefaultPartition] Fetching state (will wait for 30000 milliseconds):
15:18:02,379 INFO [DefaultPartition] State could not be retrieved (we are the first member in group)

---------
As I copied the application to "/farm" directory of node1 (of machine 1)... the application is served by the same node all the time. If I kill the node1 then, node2 throws 404 (As the application is not deployed on that)

As I remember I didn;t do anything extra while setting up locally. Do I need to setup something like multicasting for remote machines??

Am I missing anything?

Thanks,
Suresh

[frankthetank]

Can they ping each other?
Is there a firewall running?

[ozguy]

could be something do with the way cluster is discovered.
if both nodes are in the same subnet than they will be able to discover each other but if they in different subnet, they you need extra config at the router level etc to get the UDP discovery work correctly.

or you can use TCP instead of UDP for cross network cluster. just change the UDP config to TCP in the jgroups settings,