2 Replies Latest reply on Aug 8, 2007 9:29 AM by nayyares

    cache replication node went out !

    nayyares Newbie

      Hi,

      We are running Jboss 4.0.4 on solaris , the Jboss is clustered in five nodes with cache replication and failover, we are using cisco loadbalance switch for loadbalance (round-robin), , we mostly encounter problems with cache replication, as often one or more nodes stop serving jboss , and if we restart Jboss on these failed nodes, it d'nt get in to cache cluster, while it work stand alone perfactly, for recovery of cluster we have to restart Jboss on all node !


      I am unable to figure out why is it so , any clue?

      or even how can i put them again in cluster, without restarting whole cluster.

      TIA

        • 1. Re: cache replication node went out !
          Manik Surtani Master

          There are a good many number of reasons why nodes don't form a cluster, only a handful of them relating to JBoss Cache. We'd need a lot more informtion than this. Do you have any logs of what happens when a node attempts to join the cluster? Specific stuff to look for would be JGroups messages about joining the cluster. Set your log level for org.jboss.cache to DEBUG and org.jgroups to INFO.

          Cheers
          Manik

          • 2. Re: cache replication node went out !
            nayyares Newbie

            Hi,

            Thanks for the reply, i restart Jboss with following command

            $/jboss-dir/bin/run.sh -c all

            the stuff i have noticed in log that can give some clue is,
            ----------------------------------------------------------------
            [INFO] [mycluster] Number of cluster member: 1
            [INFO] [mycluster] Other Members: 0
            [INFO] [mycluster] New cluster view for partition mycluster (id:0, delta:0)
            [INFO] [mycluster] Dead Members: 0 ([])
            [INFO] [mycluster] New Members: 0 ([])
            [INFO] [mycluster] All Members: 1 ([])
            .....
            ---------------------------------------------------------

            while at the time i have given ./run.sh -c all , command there were four other cluster node running !

            how can i check that whether other cluster nodes are visible to this server?

            any clue, thanks