I have a web application which uses http sessions deployed to a two node JBoss cluster. Both nodes show up in the mod_cluster-manager and belong to the same balancer.

After hitting the application once and establishing a sticky session I wanted to test the behavior of mod_cluster when I then disabled the context for the application on the node to which the session was stuck.

The result was a segmentation fault in apache error_log on the next http request.

[Thu Jul 09 09:31:22 2009] [notice] child pid 9412 exit signal Segmentation fault (11)

I assume the segmentation fault is a bug. Can anyone confirm this?

Also I anticipated that the request would be re-routed to the node that still had the context enabled. What should have been the correct behavior had I not received a segmentation fault?