We’re running 2.1 in a clustered environment.
We’ve noticed that for at least 1 cluster, all of the instances in that cluster appear to issue connection resets to ALL connected clients every 60 minutes. It’s not synchronized between instances (i.e. They issue the resets at different times) but each instance does it every 60 minutes.
The result is the high traffic app we have making persistent / keep alive connections to the instance sees ALL OF THEM simultaneously disconnected every 60 minutes.
Does anyone have any idea what could cause this?
With have keep alive enabled on the http service but A) the timeout is 30 seconds and B) the persistenct client connections are actively doing stuff) so obviously this doesn’t explain ALL connections simultaneously resetting every 60 minutes.
This is somewhat related to the other question I just asked about disabling the admin timeout - I can’t imagine it would cause what I’m seeing but the fact I’m seeing this EXACTLY every 60 minutes and that’s also the value of the <das-config admin-session-timeout-in-minutes="60" ... /> is at least a coincidence. And I have no idea why Glassfish would do this so it has me wondering if it’s not just a coincidence.
--
Sean Comerford, Staff Engineer
ESPN.com Site Architecture Group
Office: 860.766.6454 Cell: 860.329.5842