Thank you for your suggestions.
We are on the way to fix all those bugs. But, a minor issue observed by me when clustering a sample application like "Hello world".
When the 2 cluster nodes are running and when one of them which is interacting with the browser currently(at that time) suddenly fails, until the other one is about to receive this status, the application is not being focussed to be clustered well. Of course, it is a very minute period. But the target of clustering is that, the end user must not identify the switching of the server..
Let me know if any alternative for this, like the server which is not currently active may take some secondary back up, so that, it may act immediately, keeping the customer transperant about this switching.