Then create an external jndi context that binds the root of the sole jndi server and have all reference bound into it. The existence of the different servers will not be transparent to the client from a connection standpoint, only where they do the initial lookup. The HAJNDI server gives this behavior for free.
I am a new on JBoss. Does this mean that i have to make a cluter of all the JBoss servers having the beans and give the IPAddress of any one, so that a failover or redeployment of beans in another JBoss server gets reflected in the JNDI. Is there a single JNDI service running in the cluster? Because as per my knowledge even in a cluster the JMSProvider for each server runs seperately(as in JBoss 3.2.3) so i am not quite sure about JNDI service behaviour running in clustered environment. Thanx for ur reply.