1 Reply Latest reply on May 12, 2004 2:14 PM by Hernan Terzian

    Only Fail-over

    mkyaj Newbie

      Hi,

      I'm new to Clustering and also to JBOSS. We want to use clustering only in the case of a fail-over, not for any load-balancing. We will have 2 nodes. If the main Masternode is down then only the requests has to go to the other node. But all the HTTP sessions and EJB requests has to be replicated on the secondary node so that if the node1 is down, node2 will start functioning.

      I have gone thorugh all the documentation of clustering and many postings of Clustering. But i am so much confused of where to start from.
      I need a small help from anyone of you to start.

      We are using
      -- JBOSS-3.2.3
      -- only localhome and local interefaces
      -- struts action classes which will lookup for the local interfaces and do the corresponding job.

      So we have all these action classes and EJB's everything in a single war.
      But i came to know that setting the list of the nodes in the property object before lookup is one of the ways

      Properties props = new Properties();
      props.setProperty("java.naming.factory.initial", "org.jnp.interfaces.NamingContextFactory");
      props.setProperty("java.naming.factory.url.pkgs", "jboss.naming:org.jnp.interfaces");
      props.setProperty("java.naming.provider.url, "node1:2000,node2:2000");

      But in our case this lookup code and the EJB's will reside under the same JVM. So if the node is down means the lookup code also is down.

      As i understood we can use some Load-balancer like Apache/Jetty so that it will redirect the requests to only the primary node and using HASingleton will solve our purpose.

      Is clustering possible in this case?
      Please guide me in starting this.

      Thanks,
      Murali.