What is your hypothesis of the results of performing a rolling update of the JBOSS clustered web servers by shutting down just one JBOSS instance, upgrading it and then restarting it and then going to the next?? I m afraid that small changes in the structure of the application would make this fail. The problem then is that we have to create a WebUI/MidTier outage for the production servers when we push updates.
Any thoughts or ideas on this?