6 Replies Latest reply on Dec 20, 2002 9:59 AM by slaboure

    howto restart a jboss cluster/ farm member correctly?

    meatjboss

      Hi,
      I noticed that a jboss deletes the files within the farm folder when shutting down.
      To make it behave as it did before (deploy the farm folder content) I have to follow these steps before restarting it:

      1. remove the farm-service.xml
      2. copy the content of some other farm members farm folder to the restart candidate
      3. restart it
      4. deploy the farm-service.xml file again

      Is there a solution that works more simple without copying/removing/adding these files?

      thxia, gk.

        • 1. Re: howto restart a jboss cluster/ farm member correctly?
          slaboure

          If you feel this is a bug, do not hesitate to report a bug on SF.

          I appreciate. Thank you.

          Cheers,


          Sacha

          http://sourceforge.net/tracker/?group_id=22866&atid=376685

          • 2. Re: howto restart a jboss cluster/ farm member correctly?
            meatjboss

            I think it' is ok to empty the farm folder on shutdown - but not adequate.

            What I am actually missing is a feature that synchronizes the content of the farm folder on startup with that of the running farm members in the cluster.

            If a jboss farm member gets killed somehow and cannot empty the farm folder and gets restarted - the content of the farm folder might not be in sync with the others... and deploy older versions of services which is imho worse than deploying nothing.

            Is there a feature in jboss 3.0 that syncs the farm folder on startup with the living farm members?

            thxia, g.

            • 3. Re: howto restart a jboss cluster/ farm member correctly?

              Maybe this should be a new topic but the argument is the same.
              I've noticed that in JBoss 3.0.4 farming does not delete files from the farm directory, and when the server starts, components from other nodes are downloaded.
              But now the problem is that during the boot process of the server the components that are already in the farm directory get deployed "too early". This means that JBoss tries to deploy entity beans before the relative data store is initialized (I'm using the defaultDS for now) or servlets that use session beans that are not yet deployed. What I get is a lot of binding exceptions.
              On the other hand, if I deploy the ear file on runtime, it works perfectly. It also works fine if I boot the server with the ear file in the deploy directory.
              I don't know if this is this an error in my configuration or a bug?

              Jaksa

              • 4. Re: howto restart a jboss cluster/ farm member correctly?
                slaboure

                That's strange. I haven't tested this with EAR files though (only with WAR and JAR) and it works correctly.

                which kind of stacktrace do you get?

                cheesr,



                sacah

                • 5. Re: howto restart a jboss cluster/ farm member correctly?

                  Hi Sacha.
                  I'm sending you the whole server.log file because I think the question is to see WHERE do the exceptions occur. If you want, I can send you the .ear file (13kb) and the sources also.
                  Thanx,

                  Jaksa

                  • 6. Re: howto restart a jboss cluster/ farm member correctly?
                    slaboure

                    Hello,

                    Can you please try to modify the file deploy/farm-service.xml and add a tag to the FarmMemberService MBean, something like that:
                    jboss.deployment:type=DeploymentScanner,flavor=URL

                    Cheers,


                    sacha