4 Replies Latest reply: Jul 29, 2012 10:28 PM by 804328 RSS

    Loadbalancer redirection only to one node on cluster

      We have application which was migrated from Weblogic 8.1 to Weblogic 10.3.4. After that we have session timeout errors - session is not replicated on cluster nodes (and can't be - bug in code does not allow to change it).
      Is possible to configure Weblogic loadbalancer to redirect user to node on which user was logged?
        • 1. Re: Loadbalancer redirection only to one node on cluster
          Kalyan Pasupuleti-Oracle

          Can you check your weblogic.xml file once.

          Under session-descriptor check for persistent-store-type

          increase the session time out value too.

          • 2. Re: Loadbalancer redirection only to one node on cluster
            Thank for answer. We have below configuration:


            I want have situation when user will log to application and all his request will be handled by only one node. Failover is not important for us.
            • 3. Re: Loadbalancer redirection only to one node on cluster
              murali veligeti - oracle
              Hi Adam,

              If your application doesn't allow session replication across clusters then for <persistent-store-type> replicated_if_clustered is not required. You can use memory for that.

              There is no <session-param> in WLS 10.3.4. Instead you can use <cookie-path>.

              Could you please let us know why you are using cookie-path ?
              Have you defined any session-timout in web.xml ?

              Yes you can use load balancer to route the requests to weblogic cluster in round robin fashion. The load balancer will make sure that it redirects user to node on which user was logged in and this is session stickiness. But failover will not happen because session replication will not happen.

              Refer below URL for more details


              Thanks & Regards,
              • 4. Re: Loadbalancer redirection only to one node on cluster
                We need to undertstand that Weblogic cluster in conjunction with Weblogic proxy plugin configured on the Web server allows session stickyness where by subsequent requests within the same browser session would be directed to the same WLS server instance and this is called as Session Stickyness.

                WLS cluster also allows failover where by if in the middle of processing a specific request,the managed server goes down,the request would be failed to another weblogic server instance in the cluster(determined by configuration of Replication groups,secondary Jsession ID) and the request would be processed successfully in the same session.

                In your case,the session times out and once the session times out the next request is considered as a new request which is likelyto be sent to another weblogic server instance as Round Robin is the default algorithm used by weblogic in a cluster.So,it is normal for you to see the next request (after the session has timed-out for the initial request) to be directed to the next server instance in the cluster.

                What is important to check from your perspective is finding out if subsequent requests from the same session (having the same jsession ID) is being redirected to different servers in the cluster or is it being handled by the same server instance until the termination of the session.