I would like to have some clarification on ENABLE_SERVER_AFFINITY property during a Jndi lookup made from a WL Server to another WL server.
Basically, our enterprise application has a singleton instance that holds certain internal handles. Our APIs called from applications deployed to other managed servers involves this handles. In a single host clustered environment, there's no problem with this singleton instance replication.
However, In a multi-host clustered environment, this singleton instance doesn't gets replicated to the other servers of the cluster. Due to this, when weblogic routes further API calls (in round-robin fashion) to another server in the cluster, the API call fails.
As per this doc link, ENABLE_SERVER_AFFINITY (set during Initial Context lookup) and 'round-robin-affinity' (set as algorithm for cluster) seem to be applicable only for External Clients-> Server invocations.
Can you please let me know some hints on establishing the server affinity during server-to-server invocations? Defining these two elements in weblogic-ejb-jar.xml of our EJB doesn't work.