4 Replies Latest reply on May 13, 2011 6:47 AM by wdfink

    JBoss boot messages and confirming nodes communicate

    emmmcl01

      Hello.

      I'm unclear whether two machines on the same cluster are communicating with each other.

       

      These are the instructions that get executed in $%BOSS_HOME%\bin\service.bat :

       

      call run.bat -c vNode1 -g vCluster -b 178.31.17.114 -Djboss.messaging.ServerPeerID=1 >run_vNode1.log

      call run.bat -c vNode2 -g vCluster -b 178.31.17.115 -Djboss.messaging.ServerPeerID=2 >run_vNode2.log

       

      I'm confused by these messages that appear in the JBoss boot.log file:

       

      1. [QuartzScheduler] Scheduler JBossQuartzScheduler_$_NON_CLUSTERED started.

      2. ARJUNA-12332 "Failed to establish connection to server"

       

      The comments suggest that JBoss is starting in a non-clustered mode - but maybe
      the messages are JBossQuartzScheduler and Arjuna specfic - and have nothing to do with the
      cluster starting.

       

      Aside from bring down one of the nodes and confirming the cluster VIP is still responding -
      What can I do to confirm that the nodes of the cluster are communicating?

       

      Do I have to insert statements into my webapp which write user session variables to the log file
      and confirm those variables are being preservered when one of the nodes comes down?

       

      Thanks.

        • 1. JBoss boot messages and confirming nodes communicate
          wdfink

          Hi Emmett,

          your start commands looks correct.

          The first message is not related to the clustering, the second I don't know ATM.

          What version of JBoss do you use?

           

          You should see in your logfile a Messages like 'New Member' or 'Number of members', here you see whether the nodes come together.

          Also you can check via JMX console the JGroups state.

           

          If it will not work see the wiki http://community.jboss.org/wiki/TestingJBoss to check, it might be a network configuration issue.

           

          I suppose you try to cluster a webapp, in this case there is no automatic loadbalancing, like for EJB calls, you have to have a balancer in front of JBoss, e.g. Apache.

          1 of 1 people found this helpful
          • 2. JBoss boot messages and confirming nodes communicate
            emmmcl01

            Nothing appears in either log file about "New Members".

             

            Were using JBossAS 6.0.Final and Windows Server 2008.

             

            Is there a way to confirm the ports that the servers communicate on are open (not disabled by a phantom firewall configuration)?

             

            Thanks.

            • 3. JBoss boot messages and confirming nodes communicate
              emmmcl01

              BTW - there is a load balancer and a Virtual IP. There are no EJBs.

               

              In the TestingJBoss link where it says "If you work with dedicated bind addresses (run.sh -b <ip>) you should start the jgroups test with the additional property '-Djgroups.bind_addr=192.168.128.5', it should be equal to you final JBoss cluster. The bind address will make a difference for the network configuration!"

               

              Would the dedicated bin address be the address of the Virtual IP? Or is that a default Jboss IP address stored in a configuration file?

              • 4. JBoss boot messages and confirming nodes communicate
                wdfink

                -Djgroups.bind_addr should be the ip address that you set by run.sh -b XXX

                The mcast_addr within the XML config should be the same as you use as the run.sh -u XXX, the default shoulf be 228.11.11.11, see server/all/deploy/cluster/channelfactory.sar/META-INF/jgroups-channelfactory-stacks.xml

                 

                In this case the test use the same network routes as the jboss installation.