5 Replies Latest reply on Nov 30, 2009 6:17 AM by Trustin Lee

    Consumer Hanging after optimizations....

    Clebert Suconic Master

      There is something weird going on....


      I am running jms/perfSender perfListener with this following configuration:

      I - Changing persistence-enabled to false on hornetq-configuration.xml:

      <persistence-enabled>false</persistence-enabled>


      II - Running the perfSender/perfListener with the following properties:

      num-messages=500000
      num-warmup-messages=0
      message-size=1024
      durable=false
      transacted=false
      batch-size=2000
      drain-queue=false
      destination-lookup=perfQueue
      connection-factory-lookup=/ConnectionFactory
      throttle-rate=-1
      dups-ok-acknowledge=false
      disable-message-id=true
      disable-message-timestamp=true
      



      i.e. durable = false, Transaction=true, batchSize = 2000 and Null Persistency on the server.


      The sender will work fine, however the listener will hang around 50K messages (sometimes around 20K).

      After some time I start to get Connection failed messages on the server. The Listener will eventually finish but I will get a few connection failure on the server.
       [java] [hornetq-failure-check-thread] 18:59:00,978 WARNING [org.hornetq.core.server.impl.ServerSessionImpl] Client connection failed, clearing up resources for session 432d4774-dd4b-11de-bcac-0022faff08c8
       [java] [hornetq-failure-check-thread] 18:59:01,041 WARNING [org.hornetq.core.server.impl.ServerSessionImpl] Cleared up resources for session 432d4774-dd4b-11de-bcac-0022faff08c8
      


      Just a guess.. maybe there is something wrong on the FrameDecoder?


      I also suspect the Bridge issues we are having on the testsuite are related to this.



      This all worked fine when I reverted my SVN copy to r8430 (which is before optimization merges).