How does the automatic failover among members of a cluster works?
When the cluster failover support is activated, the client that is making use of that cluster get automatic failover among its members. A client will get access to one of the cluster members when it resolves against a cluster. If the cluster member is alive, then everything is fine. However, once the cluster member crashes or becomes unavailable, the orb will transparently switch to the next available cluster member based on the criterion. This will continue as long as the orb attempts to rebind. The failover feature allows you to use the next active cluster member when the current cluster member fails. If all the cluster members are dead, then it is useless!