2 Replies Latest reply on Jul 28, 2005 12:12 AM by Ruchi

    TreeCache TCP configuration

    Ruchi Newbie

      I have configured TCP as the transport protocol in cluster config.

      <attribute name="ClusterConfig">
       <config>
       <TCP start_port="9999" bind_addr="A"/>
       <TCPPING initial_hosts="B[10001],C[10002]" port_range="1" timeout="3000" num_initial_members="2" up_thread="true" down_thread="true"/>
       <FD shun="true" up_thread="true" down_thread="true" timeout="2500" max_tries="5" />
       <VERIFY_SUSPECT timeout="1500" up_thread="false" down_thread="false" />
       <pbcast.NAKACK gc_lag="100" retransmit_timeout="3000" up_thread="true" down_thread="true" />
       <pbcast.STABLE desired_avg_gossip="20000" up_thread="false" down_thread="false" />
       <pbcast.GMS join_timeout="5000" join_retry_timeout="2000" shun="false" print_local_addr="false" down_thread="true" up_thread="true" />
       <pbcast.STATE_TRANSFER up_thread="true" down_thread="true"/>
       </config>
      </attribute>


      What i have understood after reading jgroups documentation is that in initial_hosts attribute we specify all the nodes which are to form a cluster and in timeout we specify the time to wait for other members.

      I have the following queries -

      1. In one of the forum it was mentioned that all nodes mentioned must be running. What do u mean my running here? Do you mean that TreeCache has been started or you mean just that they are switched on?

      2. Suppose Node A and B are running(by running i mean treecache has started). What if Node C starts up after 3000 ms. Will it be part of cluster?

      3. I want another Node D to join the cluster without restarting A, B and C. How shld i do it.

        • 1. Re: TreeCache TCP configuration
          xavierpayne2 Newbie

          I'm interested in knowing this as well.

          We are unable to use multicast here because we have several developers on the same subnet running many versions of jboss. But when I use the TCP configuration jboss clustering itself works fine but jboss cache just seems to not work (as a cluster) at all. The updates to one host just don't get replicated to the other (I am using REPL_SYNC, and I get no exceptions). Funny thing is I see them find each other in the server log!

          Maybe I am just misunderstanding portions of the config file and the provided documentation.

          • 2. Re: TreeCache TCP configuration
            Ruchi Newbie

            The following properties must be set in treecache.xml for cache replication to take place -
            1. CacheMode= REPL_SYNC
            2. ClusterName - All the nodes which are to form a cluster must have the same name
            3. A proper cluster config