0 Replies Latest reply on Mar 15, 2011 6:47 AM by williamuk

    JBoss JMS cluster - load balancing

    williamuk Newbie
       
           down vote     favorite        
         
           
                          
                   

      I've setup a JBoss cluster (2 nodes) in a similar manner discussed here, using JBoss version 5.0.1.

      I have a JBoss ESB application listening to the JMS queues, which is deployed to both nodes.  This is not a HA Singleton.

      However another JBoss ESB application is also deployed to both nodes which is a HA Singleton (this doesn't listen to any of the queues, so I presume  should not have any relevance to the issue described below).

      The throughput of messages is not massively high - perhaps a few a second.

      Should I expect to see both nodes processing messages in  approximately equal numbers? Or will one process all of them unless the  load becomes too high (or it fails).

      What is actually happening is that one node (the 2nd to join the  cluster, in case that's relevant) is processing virtually all of them. 

      However the 1st node does very occasionally process one or two (i.e.  as I write this, in the past 24hrs it did nothing but just randomly  processed 4/5 messages a few minutes ago). 

      There is nothing in the logs, such as exceptions, etc, to indicate that there is anything wrong with the 1st node.

                 

      I've setup a JBoss cluster (2 nodes) in a similar manner discussed here, using JBoss version 5.0.1.

       

      I have a JBoss ESB application listening to the JMS queues, which is deployed to both nodes. This is not a HA Singleton.

       

      However another JBoss ESB application is also deployed to both nodes which is a HA Singleton (this doesn't listen to any of the queues, so I presume should not have any relevance to the issue described below).

       

      The throughput of messages is not massively high - perhaps a few a second.

       

      Should I expect to see both nodes processing messages in approximately equal numbers? Or will one process all of them unless the load becomes too high (or it fails).

       

      What is actually happening is that one node (the 2nd to join the cluster, in case that's relevant) is processing virtually all of them.

       

      However the 1st node does very occasionally process one or two (i.e. as I write this, in the past 24hrs it did nothing but just randomly processed 4/5 messages a few minutes ago).

       

      There is nothing in the logs, such as exceptions, etc, to indicate that there is anything wrong with the 1st node.