|
we do not use IMonitor - did several years ago, but have
had it turned off in recent years.
I'll try the reduction in threads in QM; went down that
road before, but will see if it will shed some light again.
I'll look at reducung the influx of dumped mail; however,
any mailserver that costs what this costs should be able to handle 500 emails
introduced at once.
Randy A.
GlobalWeb.net
I have had Queuemgr hand twice so far, both times in a stopping
state I believe. Early this week I also had SMTP hang in a stopping
state, but that service is totally redesigned in 8.2+
I saw instances
before when IMail Monitor would attempt to stop a service while things were at
100% and cause that service to hang, and maybe if you have IMail Monitor on,
you might disable the Queuemgr checking to prevent it from trying to restart
the service when it is non-responsive due to load. IMail might also be
trying to recover from an error apart from IMail Monitor, and as I said
before, it appears that they should kill hanging processes instead of just
trying to have Windows attempt the shutdown and restart because they can hang
in a stopping state. My SMTP crash earlier this week would have
recovered gracefully if they did this in 8.15. Maybe Kevin or Eric can
comment on this?
Regardless, you should slow down the delivery, it will
have no bad effects and will probably fix your issues. You might also
reduce the threads in Queuemgr in order to help prevent hangs.
Instability with IMail on my system is always associated with very high and
prolonged load.
Matt
GlobalWeb.net Webmaster wrote:
Matt,
the mystery continues....
We are on a RAID 5 already;
dual Xeon 3.0 w/ 2GB ram (processors not much slower than
yours)
Defrag using DisKeeper nightly
In Declude, I whitelist everything sent from this
app
I had the same issues with QM in 8.15 with Declude 2.x
(with or without Declude running); upgraded to Declude 3.x to see if that
would help, and it seemed to at first. upgraded back to 8.22 (after an
initial downgrade to 8.15 in june last year) and again worked fine at first,
then started all over again.
beats me....new platform is the only thing I can
rationalize at this point. I have spent 12 man-hours this week
requeing email. Another week of that and a new platform and
hardware is paid for more than
I generally
will reboot a server if something like this has crashed because I don't
trust the general state of that server. Also, when things like the
SMTP and Queuemgr services crash, at least with 8.15 HF2, they seem to die
in a "stopping" state and can't be restarted without killing the
process. Ipswitch should add code that attempts a standard service
restart, but monitor the process to see if it hangs and then kill it if
necessary and then start it again.
If you had the server in a fresh
state and ran these messages through without Declude configured as an
external app and still saw the issue, I would suspect that it was an IMail
issue, or at least not Declude naturally.
IMail could use it's own
throttling mechanism for these things, in fact in the old Declude command
line process, Declude had the overflow directory that would help to
stabilize IMail servers.
As a quick fix, you should at least have
the sending device slow down the delivery by putting a wait in between the
E-mails that is longer than their processing time in IMail/Declude.
It sounds like you are mail bombing yourself. 500 good E-mails would
take my dual 3.2 GHz processors 5 to 10 minutes to go through since every
test and virus scanner in Declude would be run. You have far less
processing power. Adding a wait wouldn't necessarily extend the
delivery time much, but it would help you from overwhelming your
server. Naturally whatever is crashing would be better not to crash
since E-mail is bursty by nature and it should be able to manage such
conditions. You might also want to look into tuning the settings in
the Queuemgr service, such as the threads setting, and try a lower
number. I believe they can be dropped very low without a measurable
effect on a generally low volume server. 86,400 messages is the same
as 1 per second. Maybe your disk I/O is also a bottleneck and could
be improved by doing things like defragmenting drives, using RAID 5, or
adding more drives.
Matt
randy armbrecht wrote:
Matt,
Thank you for you response....
Even if I change the SMTP processing back to soley IMail and take Declude completely out of the sequence, the QM does not restart unless I perform the processes listed earlier. I ran for a day last week without Declude and the same thing happened.....
Randy A.
Global Web Solutions, Inc.
804.346.5300
877.600.GLOBAL (4562)
Randy,
I have seen others mention similar things with Declude 3.x and IMail
8.2+. I wouldn't discount the possibility of this having something to
do with Declude. You should report it to them just to be safe.
To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/
|
