3 Replies Latest reply on Apr 26, 2008 7:25 AM by timfox

    JBoss Messaging JMS connection usage leading to problems

    harish43


      I have a question regarding the topic below,

      http://www.jboss.com/index.html?module=bb&op=viewtopic&t=103695&postdays=0&postorder=asc&start=0

      Which version of JBoss Messaging contains the fix you mention?
      Also is there a corresponding fix in JBoss Remoting as well? If so in which version is it?

      We are experiencing similar issues as mentioned in the topic when load testing with JBoss Messaging with usage pattern of JMS connections being very similar (lookup factory, lookup queue, connect to queue, open session, send message, close connection) but we are using a JMS connection max-pool-size of 20 as defined in the tx-connection-factory in jms-ds.xml.

      Our environment is :
      JBM 1.4.0 SP3
      JBoss AS 4.2.2 GA
      JDK 1.5.0_06
      Redhat Linux
      MySQL for JMS persistence
      Sybase for data persistence

      As mentioned by Kevin Conner in the topic the BisocketServerInvoker$ControlMonitorTimerTask and BisocketClientInvoker and HashMaps are growing in number and size throughout the test and never getting GC'd.

      Any help is appreciated.

      Harish.

        • 1. Re: JBoss Messaging JMS connection usage leading to problems
          timfox

          If you're using the JCA connection pool as you say, then the old problem should never occur since the JCA layer will cache connections.

          • 2. Re: JBoss Messaging JMS connection usage leading to problems
            harish43

            I am not certain whether we are using the JCA connection pool. Assuming that we are not using the connection pool is there a fix for the issue mentioned by Kevin Conner?

            • 3. Re: JBoss Messaging JMS connection usage leading to problems
              timfox

               

              "harish43" wrote:
              I am not certain whether we are using the JCA connection pool. Assuming that we are not using the connection pool is there a fix for the issue mentioned by Kevin Conner?


              It's slightly worrying that you don't know if your using the JCA managed conection factory or not, since this will have a big effect on your applications operation.

              See the wiki for more details on how to configure this.

              If you want to know when a particular issues was fixed please look in JIRA - all the information is there.

              But please don't expect me to do the JIRA search for you in my free time when it's a sunny day outside, to save you a little time. :)