Add a note in 6.3.3. Naming Discovery in Clustered Environments, Admin and Config Guide

Add a note in 6.3.3. Naming Discovery in Clustered Environments, Admin and Config Guide

There is a not well documented behavior about HA-JNDI discovery cache.

I got a customer who has multiple clusters within same network and each cluster has unique multicast address. They observed an odd behavior when trying to perform HA-JNDI discovery from within one cluster to the another cluster, by specifying jnp.discoveryGroup of the target remote cluster. It unexpectedly returns its own local cluster HAJNDI because JBoss has an HA-JNDI cache and local node is cached with default partition name.

http://docs.redhat.com/docs/en-US/JBoss_Enterprise_Application_Platform/5/html-single/Administration_And_Configuration_Guide/index.html#The_Naming_InitialContext_Factories-Naming_Discovery_in_Clustered_Environments

Proposed text to add:

Note that if you have multiple clusters within same network, make sure each cluster has unique partition name (-g option) and specify jnp.partitionName when perform HA-JNDI discovery. There is a cache for HA-JNDI discovery and the partition name is used as a key of the cache. It would create wrong cache entry if multiple clusters have identical name.