1 Reply Latest reply on Apr 24, 2013 11:57 AM by wdfink

    How to set up a 2node failover cluster with JBoss5.1

    molek

      Hi!

       

      How to set up a cluster where:

       

      both the nodes are up and running, but at any given time only one node will serve the requests and the other remains inactive. When the active node goes down, the stand-by node takes up and start serving the requests,

       

      It is reasonable to me to set up such cluster because there is no problem with the another  server being mostly idle (virtualization - and one server is more than enough to serve all clients )

      and  - what is much more important to me - there is no problem with logs  being dispersed among two servers. I have found how this set up can be  achieved for HTTP - but is there

      a way to set it for EJB calls? I  have standalone clients that make EJB calls to cluster.  The four  policies (FirstAvailable, RoundRobin etc)  does not contain the one I am  looking for - and I would like to avoid implementing my own.

       

      If the only way is to implement custom policy where can I find info about it and how to configure it properly. I sam some people were doing this and it looks it involved quite a lot of configuration exercises

       

      Regards

       

       

        • 1. Re: How to set up a 2node failover cluster with JBoss5.1
          wdfink

          If you use a RoundRobinPolicy it should work.

           

          In opposite to http requests, which are not load balanced, the ejb proxy will have this LB feature.

          What you need to do is instead of jnp://host.1099 you need to use jnp://host:1100 to have a JNDI-HA context.

          If you lookup the EJB you need to keep that proxy and invoke methods with this proxy. In that case both server will be used