Session Timeout issue in clustered jboss

[pilania]

It happens on Jboss-4.0.1sp1 and Jboss-4.0.2

Two server cluster is running. User logs in through CSS (using stick sessions) and request is sent to server 1.

Session gets replicated to server 2 fine.

Web-app is distributable. Timeout is set to 3 mts on web-app.

After 3 mts even though user is working actively on server 1, server 2 sends a session destroyed and our listener (implementing HttpSessionListener) has implemented sessionDestroyed method which logs the user out.
Which prompts the user to shout "What the F!@#$"

[ben.wang]

Not sure if I understand you correctly. But there is no good policy for the listner unfrotunately regarding to the lifecycle events in a distributed sense. For example, when a session is invalidated on one node, it will replicate to invalidate on the other nodes. Should we or should we not trigger the lifecycle events?

[brian.stansberry]

There is a JIRA issue related to this:

http://jira.jboss.com/jira/browse/JBAS-2377.

I'm sure there will be a design discussion on the dev forum for this as well once I get done with training.