2 Replies Latest reply on May 23, 2010 2:22 AM by Alex Harvey

    cache replication issues on centos

    Rishabh Chandra Newbie

      Hi all,

      I am having issues in cache replication on centos using jboss tree cache 3.2.1GA.

      I have set the multicast port as 45588 and bind_addr as 45587. The broadcast is done to 228.10.10.10.

      I have added fire wall rules to let UDP and TCP happen on both the ports still i am having issues replicating.

      Is there any way the cache can be using any other ports than the 2 mentioned above ?

      Thanks.

        • 1. Re: cache replication issues on centos
          Alex Harvey Newbie

          Hi all, I am the sysadmin working with Rishabh on this.

           

          I'd like to add that during the Tomcat startup we see two SYN packets from node A to node B being dropped by the local firewall (i.e. iptables) on an apparently random port (e.g. DPT=58317), another six from node A to node B being dropped using a different apparently random port (e.g. DPT=48800), and finally two SYN packets from node B to node A (i.e. in the remote firewall, e.g. DPT=53675).

           

          We got everything to work by following the instructions in the following document: http://community.jboss.org/wiki/RunTwoInstanceOfJBossInFedora which suggests we just allow all TCP and UDP ports between the two nodes. That is, however, not ideal from our security policy perspective!

          • 2. Re: cache replication issues on centos
            Alex Harvey Newbie

            Hi All,

             

            In case anyone follows this thread, we still have not found a solution and still find this puzzling. We continue to allow all TCP/UDP but feel there must be a better solution. Unmentioned in the original request is that we have a JCS cache running as well, although none of the random-looking ports seem to be connected with that application either. I'll update this thread in the event that I find a solution.

             

            Best, Alex