On Jul 15, 2009, at 2:14 AM, Cheong Hee (Gmail) wrote:
I believe most have observed the active session count in JavaMonitor
is not accurate when the session is ended due to time out. The
count reduced if the session terminate e.g. user log out. I have
not tried it out but it is quite obvious this count may be taken
from Application.activeSessionCount().
It posed an issue as the session count (due to timeout) seemed to
remain there forever, and it leaves us uncertain if we are to stop
and restart the instance. Has anyone got any to share so that a
more accurate active session count to rely on?
Cheers
Cheong Hee
The memory usage stats are meaningless, but the active session count
has been accurate in my experience. I suspect that your code has bugs
in it that are preventing session termination. Get a thread dump and
see if you have deadlocked threads.
