MDB and load balnce

Question

Appreciated for any input in advance.
          I am working on application that is running on a weblogic 9.2 cluster with two managered nodes. There two JMS servers, one for each node instance, and Uniformed Distributed Destination queue is configured for JMS servers (or targeted to cluster). MDB listen to this queue, and configured maxBeansInFreePool = "5", initialBeansInFreePool = "5".
          MDB uses customized connection factory with config
               Maximum Messages per Session = 1
               lient Acknowledge Policy = Previous
          I Sent 40 message to queue in cluster, then from weblogic console, we can see "Messages Current" is 15, and "Message Pending" 5 in the queue for each node, and "Bean In Use Count" is 5.
          MDBs in one node process all message very fast, and MDBs in another node are very slow. for instance, "Messages Current" and "Message Pending" is 0 in the node one, and "Messages Current" 15 and "Message Pending" 5 in node two. My question here is why Weblogic does not move some message from node two to node one? How I can fix this problem?
          Best Regards,
          James

Answer 1

Tom,
          Do you have any further information on the internal extension to do with
          ALSB?
          > I think our ALSB product can use an internal extension to force MDBs
          to consume from all queue members, rather than just the local queue
          members - so if you're using ALSB you might want to investigate this
          option.
          I've a problem at the moment getting an ALSB proxy service to consume
          from all queue members in a cluster - it only consumes from a single
          destination if you've one server in the alsb cluster, yet I've 4 jms
          member destinations.
          We're using ALSB3.0 and it has a proxy listening to a JMS distributed
          destination on a WLS8.1 server.
          Pete
          Tom Barnes wrote:
          > Hi,
          >
          > We don't move messages from slowly served destination members to fast served destinations. You might want to try setting things up to take advantage of the load balancing that queueing naturally provides when their are multiple receivers on the same member:
          >
          > I think our ALSB product can use an internal extension to force MDBs to consume from all queue members, rather than just the local queue members - so if you're using ALSB you might want to investigate this option.
          >
          > If you're not using ALSB, you can match this behavior by either (A) deploying the MDBs to a different cluster than the distributed queue, or (B) deploying two MDBs to the cluster: one that directly consumes from member1 and another that consumes from member2 (rather than a single MDB that consumes from the distributed queue's logical name).
          >
          > Tom