Thanks for your very prompt reply Mark!
If the problem is hardware then it is down to the 5 disks in the array,
as we moved them from their former chassis to another one (same
model/specs) when we thought the problem might be a bad mobo. Also,
there's some additional information I forgot to include:
- we currently have no quotas (you read that right) on disk usage and
despite warnings from I.T., many users' Inboxes had grown to between 400
and 600MB. Sent and Trash folders for several users > 1GB. Some users
have as many as 300 imap folders, so while there are just 60 users,
we're hitting the machine pretty hard (IMO).
- our tech support mail account is shared by about 15 people who may
attach to it at one time.
Since the problem arose two weeks ago we have worked with staff to prune
system folders down significantly and did think we saw improvement.
Today however we experienced the same failure mode.
Thanks again.
-Jeff
Mark Crispin wrote:
For what it's worth, this problem sounds to me like some underlying system
failure and not anything in imapd. A little system with a mere 60 users
should not get into this kind of trouble.
In addition to hardware issues, have you checked for other possibilities, such
as a denial of service attack?
-- Mark --
http://panda.com/mrc
Democracy is two wolves and a sheep deciding what to eat for lunch.
Liberty is a well-armed sheep contesting the vote.
Date: Wed, 22 Oct 2008 12:31:52 -0400
From: [EMAIL PROTECTED]
To: imap-uw@u.washington.edu
Subject: [Imap-uw] uninterruptible sleep, high load average, high iowait
Hello,
pretty critical issue here...
Email server specs: RHEL3, imap-2004 (yes, I know the system is old,
we're upgrading), Dell PowerEdge 2650, PERC3/di controller, 5 SCSI
drives in RAID 5 config, 60 users mostly running Thunderbird client (2.x
version).
We have an older email server that for the past two weeks started
experiencing the following issue:
Load average (viewed by running 'top' begins to climb (usually to about
50), iowait is pretty close to 100% and as a result all imapd processes
eventually change to stat 'D' (uninterruptible sleep). Email server then
stops responding to all current imap connections and any new connection
attempts.
At this point even if people close their email clients many times their
sleeping imapd processes hang around and have to be killed at the
server. Once these sleeping processes are killed the system is again
fine (load, iowait are normal).
Hardware caveat: we did have a drive fail in the array, replaced it and
everything 'seems' to work (above state notwithstanding).
Has anyone seen this behavior? Am I missing something simple?
fyi, we are in the process of building a new machine (PowerEdge 2950)
running Centos 5, imap-2007b, but we need to stabilize the current
production server asap.
Many thanks in advance.
-Jeff
--
Jeff Cohen
Manager, Information Systems
MÄK Technologies
68 Moulton St.
Cambridge, MA 02138
(617) 876-8085 x129
_______________________________________________
Imap-uw mailing list
Imap-uw@u.washington.edu
http://mailman2.u.washington.edu/mailman/listinfo/imap-uw
_________________________________________________________________
When your life is on the go—take your life with you.
http://clk.atdmt.com/MRT/go/115298558/direct/01/
_______________________________________________
Imap-uw mailing list
Imap-uw@u.washington.edu
http://mailman2.u.washington.edu/mailman/listinfo/imap-uw
