messages backing up - help

[Chuck Rock]

Excelent, I will try some of these suggestions and let you know. FYI, I am
using a local install of Bind on the server that is flying through the mail.
I didn't think of that before. I am working on creating and using my own RBL
locally to help weed out offenders to my local network.

I thought the MailScanner went from LIFO (normal working) to FIFO when the
Max Normal Queue size was hit. At one point I had new mail flowing
immediately through and older messages delivered sometimes days later. I
lowered the queue size to make sure older messages were delivered before
newer ones. 

The server that's flying through mail also has these settings.

Max Children = 20
Max Unscanned Messages Per Scan = 80
Max Unsafe Messages Per Scan = 80
Max Normal Queue Size = 5

The load avg on that server is normally around 15 with about 600,000
messages through it in the last 34 hours.

I'll get back with you on my fix if something works.

Thank you all very much.

Chuck

-----Original Message-----
From: mailscanner-bounces at lists.mailscanner.info
[mailto:mailscanner-bounces at lists.mailscanner.info] On Behalf Of Michael
Baird
Sent: Wednesday, September 26, 2007 12:43 PM
To: MailScanner discussion
Subject: Re: messages backing up - help

Something is way off, those batch times are terrible, especially without 
spamassassin running. Have you tried running MailScanner in debug mode, 
and watching it process a batch interactively to see where all the time 
is being spent?

I'd lower Max Children way down, say 5 and work up from there, and see 
if it speeds up your batch time, too many simultaneous scan's of 
mailqueue could be at issue, or too many clamav processes.

Regards
Michael Baird

> I am running FreeBSD 6.2 with MailScanner 4.61.7 and ClamAV 0.91.2 on a
Dual
> Xeon 3.0GHz with 2 Gig RAM server.
> 
> This box is just a pre-filter to do AV scan and the basic RBL and phishing
> scan. No spamassasin filtering on this server.
> 
> I have been messing with the Max Children, Max Unscanned Messages Per Scan
> and Max Unsafe Messages Per Scan to find the magic numbers to get mail to
> prcess through this server smoothly.
> 
> Max Children = 30
> Max Unscanned Messages Per Scan = 40
> Max Unsafe Messages Per Scan = 40
> Max Normal Queue Size = 5
> 
> Current queue has 5000 messages waiting to be scanned. 
> ----------------------------
> smtp1(1600):[/usr/local/etc/MailScanner]-#ps auxww | grep MailS | wc -l
>       17
> smtp1(1601):[/usr/local/etc/MailScanner]-#w
> 12:18PM  up 30 days,  2:38, 2 users, load averages: 1.47, 1.35, 1.37
> 
> Sep 26 12:18:25 smtp1 MailScanner[83495]: Batch (40 messages) processed in
> 52.51 seconds
> ---------------------------
> smtp1(1613):[/usr/local/etc/MailScanner]-#ps auxww | grep MailS | wc -l
>       32
> smtp1(1614):[/usr/local/etc/MailScanner]-#w
> 12:20PM  up 30 days,  2:40, 2 users, load averages: 0.53, 1.10, 1.27
> 
> Sep 26 12:20:40 smtp1 MailScanner[83502]: Batch (40 messages) processed in
> 182.66 seconds
> 
> I can't find it.
> 
> I grep the mail logs for the scan seconds and I find this.
> 
> Sep 26 12:00:49 smtp1 MailScanner[78986]: Batch (40 messages) processed in
> 221.82 seconds
> Sep 26 12:01:07 smtp1 MailScanner[78940]: Batch (40 messages) processed in
> 239.80 seconds
> Sep 26 12:01:09 smtp1 MailScanner[79037]: Batch (40 messages) processed in
> 237.63 seconds
> 
> Then: 
> 
> Sep 26 12:08:46 smtp1 MailScanner[79484]: Batch (40 messages) processed in
> 667.70 seconds
> Sep 26 12:08:53 smtp1 MailScanner[79207]: Batch (40 messages) processed in
> 691.09 seconds
> Sep 26 12:08:56 smtp1 MailScanner[79329]: Batch (40 messages) processed in
> 678.39 seconds
> 
> And this time will continue to climb. 
> 
> Earlier it was: 
> Sep 26 11:52:53 smtp1 MailScanner[64885]: Batch (40 messages) processed in
> 1272.42 seconds
> Sep 26 11:52:54 smtp1 MailScanner[65051]: Batch (40 messages) processed in
> 1291.63 seconds
> Sep 26 11:53:36 smtp1 MailScanner[66205]: Batch (40 messages) processed in
> 1186.13 seconds
> 
> I restarted MailScanner and it goes back down for a short time.
> 
> I have another server similar specs, FreeBSD 6.1, 4 Gig RAM, Dual Xeon
> 2.8GHz same MailScanner and Clam version. It's time also climbs, but takes
a
> long time and the /var/spool/mqueue.in directory doesn't fill up with
> messages waiting to scan. This server can handle about twice as much
> throughput as the other. I also have the MailScanner.conf settings as
above
> set higher. This one has 20 children with 80 messages per scan. The time
> rarely gets above 300 seconds in the mail logs.
> 
> How can I troubleshoot this? Obviously it's not going to be fixed by
> tweaking those MailScanner settings above. I just don't have any idea why
> the scan times creep up like that.
> 
> Any ideas?
> 
> Thanks,
> Chuck
> 

-- 
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner

Before posting, read http://wiki.mailscanner.info/posting

Support MailScanner development - buy the book off the website!