Please ignore this...
I was wrong with my analysis. seems spring + dwr layer is not problematic.
Anyway, I still have the problem unresolved... will post the details, once I get the fail-over working fine.
Further debugging, I found the issue is related to JAAS authentication at the application level.
On failover the user principal is reset to null on second node hence, I can;t access the secured resources.
My expectation is as I logged into application on node1 (user principal is set correctly)... the same principal should be replicated on node2.
Any pointers would be highly appreciated.
Enabling "ClusteredSingleSign" did the trick.
Enabling this replicating the principal on second node correctly.
Yep. The principal is not stored in the session, so replicating the session doesn't propagate it around the cluster. Using ClusteredSingleSignOn works around the problem.