Tornoci Laszlo <torl...@xenia.sote.hu> a écrit :
Hi,
I have a Horde/IMP system that behaves strangely since I upgraded
from 3.1.5/4.1.5 to 3.3.4/4.3.4. I have about 400 users with 50-70
concurrent sessions usually. The IMAP mailstore (Cyrus) and the MTA
are on separate boxes. The system is dual Xeon (quad core) 2GHz, 2GB RAM.
We had excellent response times before the upgrade. The problem is,
we still have excellent response times (practically instantaneous
mailbox lists with > 3000 mails in one folder) most of the time, BUT
sometimes for long periods of time, I get miserable response times
(30-80 sec waiting) for the same large mailbox to be listed. "top"
show the http process to shoot up to 100% cpu while waiting, but
nothing else remarkable. Restarting http or mysqld doesn't help.
Logging in is not slow. The larger a mailbox, the worse the problem
is.
After some time, response times get back to normal.
It's not the IMAP box. Using IMAP clients, direct access to the IMAP
box shows absolutely no difference.
Access is always fine if I use the "dynamic" mode. This problem
appears only if using the "traditional" mode.
Can't see any unusual network or process activity. There is no
obvious correlation with the number of concurrent users either.
Any ideas, anyone?
Do you have any hook enabled ?
Ronan.
--
IMP mailing list - Join the hunt: http://horde.org/bounties/#imp
Frequently Asked Questions: http://horde.org/faq/
To unsubscribe, mail: imp-unsubscr...@lists.horde.org
