Improve HttpSessionListener handling in clustered environment

Suppose we have a web application which uses stateful session beans, and we have a HttpSessionListener like this:

public void sessionDestroyed(HttpSessionEvent event) {
event.getSession().getAttribute("sfsb").remove(); // Make sure to call SFSB's @Remove method
}

This behavior is what Seam does.

The HttpSessionListener.sessionDestroyed() would be called in session invalidation, session timeout and during undeploy/shutdown for cleaning up local resources. This happens in both cases, clustered and non-clustered .

However, in clustered environment, the sfsb.remove() method call in HttpSessionListener becomes a cluster-wide operation. So when we shutdown one node, HttpSession is still available on the other node for fail over but SFSB is missing. As a result we will get javax.ejb.NoSuchEJBException: Could not find stateful bean.