WebLogic handles Stateful Session Bean clustering in the same fashion as they handle HttpSession clustering. They use passive replication to determine a primary and secondary server and if the primary fails the secondary takes over and choses a new secondary. This strategy is sufficient in all cases except when the primary and secondary fail at the same moment. WebLogic trys to minimize this risk by choosing secondary servers based on configurable replication groups. The idea is that the primary and secondary servers should never be on the same physical machine (where a hardware crash would affect both instances).
I am not sure who told WebLogic does not handle this very well... I find their implementation works fine. Maybe
you should go back and ask them for a detailed why.
[ October 13, 2002: Message edited by: Chris Mathews ]
