3 Replies Latest reply on Dec 26, 2011 5:41 AM by Dan Berindei

    hotrod client configuration - transactions

    Prasanth Manchambhatla Newbie

      I have 2 hotrod servers, but, in client I gave only 1 server in the SERVER_LIST and set PING_ON_STARTUP to true. Since client downloads the topology shouldnt the client continue to work (with second server) even after the server I mentioned in SERVER_LIST goes down?

       

      I get connection refused errors, when the first server (which I mentioned in the SERVER_LIST) goes down.

        • 1. Re: hotrod client configuration
          Prasanth Manchambhatla Newbie

          It was just jgroups configuration error. It works now.

          • 2. Re: hotrod client configuration
            Prasanth Manchambhatla Newbie

            I am using the following configuration:

             

                <default>

                     <transaction syncCommitPhase="true" syncRollbackPhase="true"/>

                </default>

             

             

                <namedCache name="myCache">

                    <loaders shared="false">

                        <loader

                                class="org.infinispan.loaders.file.FileCacheStore"

                                fetchPersistentState="true" ignoreModifications="false"

                                purgeOnStartup="false">

                            <properties>

                                <property name="location" value="../store/"/>

                            </properties>

                        </loader>

                    </loaders>

                    <clustering mode="DIST_SYNC">

                        <sync/>

                        <hash numOwners="2" rehashWait="120000" rehashRpcTimeout="900000"/>           

                    </clustering>  

               </namedCache>

             

            With the above configuration, when I start 2 servers and perform a remoteCache.put(k1, v1) I am expecting the put to be blocked until it is both replicated and persisted on both the instances, otherwise throw an error when either of those fails.

             

            But, it does not throw an error even when I shut down 1 instance.

            • 3. Re: hotrod client configuration
              Dan Berindei Expert

              Prasanth, I believe you don't see any error because the client retried the operation on the other server and succeeded.

               

              DIST mode doesn't require numOwners nodes in order to work, as long as there is one node in the cluster your put will succeed.