2 Replies Latest reply on Oct 30, 2013 6:27 AM by neha gupta

    JBoss Cluster: MDB behaviour

    neha gupta Newbie

      I am using JBoss 5.1 and have implemented a cluster consisting of two nodes.I have deployed my EAR containing the MDBs on both the nodes.These MDBs are listening to a clustered queue. The MDB picks up message from this queue and processes them. This processing consists of execution of a 4 stepped job.


      If a MDB on a particular node is in between of job execution (say it has completed 2 steps out of 4) and I bring down this node (by killing the process), what is the expected behaviour?

      Should the second node resume the job processing from where node 1 completed?

      As of now, my second node does not continue from where node 1 left and the job is left incomplete in between. Can anyone tell me what could be the issue?

        • 1. Re: JBoss Cluster: MDB behaviour
          Wolf-Dieter Fink Master

          Hi Neha,

          welcome to the forum.

          It is not clear to me what "a 4 stepped job" is. Do you have 4 messages or does a job create a new message to continue?

          Please share more details.

          • 2. Re: JBoss Cluster: MDB behaviour
            neha gupta Newbie

            Let me illustrate the processing of MDB in my application with an example.

            I have two nodes- node1 and node2 in a cluster.Each node has an MDB deployed on it.When the MDB picks a message from the queue, it starts processing it. Suppose the MDB on node1 has picked a message.It starts processing it by executing the following 4 steps:

            1) Step A

            2) Step B

            3) Step C

            4) Step D

            The above 4 steps are related to business processing.

            Now, suppose the MDB is currently at Step C and I stop this server.

            I want to know whether the MDB on node2 will continue from Step C onwards for processing this message?

            Currently, in my case, node2 is not continuing the processing left incomplete by node1.