1 Reply Latest reply on Aug 1, 2007 6:05 PM by konkimalla

    HA-JNDI problem in the cluster environment

    konkimalla

      I am very new to Jboss A/S and clustering; I have read complete clustering jboss document before jumping to the configuration. I am little stuck with HA-JNDI part.. I have done the following things in Jboss 4.2 A/S:

      1) Created a stateful bean that simply computes a sum (used @Clustered)
      2) Added a stand along java application, client, outside the A/S and successfully invoked the bean in a loop.
      3) I successfully created a cluster of two nodes
      4) While the client is getting responses from A/S, I killed it and thought I would still get the responses as the second node takes over. But, the client died.


      I have just used @clustered annotation for bean and nothing more than that. I have used the following jndi properties in the client to invoke the bean:

      props.put("java.naming.factory.initial", "org.jnp.interfaces.NamingContextFactory");
      props.put("java.naming.provider.url", "jnp://localhost:1099");
      props.put("java.naming.factory.url.pkgs", "org.jboss.naming:org.jnp.interfaces");


      I have even used 1100 but i n vain. After going through wiki and forum, I understand that may be I did not registere the beans in the HA-JNDI service.
      Shoud I register the bean to the HA-JNDI service. If so, how to register to bean using HA-JNDI in 4.2 app server?

      If we need not to register the bean in the HA-JNDI service, is clustering not working because I am using client outside the A/S?

      I would really appreciate if you can direct me a way as this is very important for me to proceed further.

        • 1. Re: HA-JNDI problem in the cluster environment
          konkimalla

          Well, I solved my problem by starting the server using "run -c all -b (ipaddress)". At the client, I just configured to localhost:1100. I am not sure whether there is an xml entry to "Bind address for all JBoss services" with the node ip address.

          I successfully tested both fail over and loadbalancing, which can be easily tested by running several client applications that continuously access the ejb in a loop.