You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Bradley Walker <ba...@modemnet.net> on 2006/01/19 18:41:02 UTC
Exim 4.60 & SpamAssassin 3.1.0 Problems
Hello all,
I'm writing this list in regards to an issue that has developed after I
upgraded to Exim 4.60 & SpamAssassin 3.1.0. Originally I posted this on the
Exim's user mailing list where I got numerous replies, but nothing concrete
answer wise to what the cause is or the solution is for this specific
problem. Most often I was told "Contact the SA developers, it's their
problem". Currently my platform is running SpamAssassin 3.1.0 on CentOS 3.4
with Exim 4.60 as my MTA.
*** continued from the orginial email ***
After upgrading to 4.60 I also upgraded to SpamAssassin 3.1.0 so as to
continue offering my webhosting customers the best spam protection I could.
However within hours of making this upgrade customers started calling me
nonstop that their email "wasn't working". It finally took me a while to
discover that what they were saying is that the email they were sending out
was bouncing back with the following text listed:
****************************************************************************
********************************************
This message was created automatically by mail delivery software.
A message that you sent could not be delivered to one or more of its
recipients. This is a permanent error. The following address(es) failed:
<ma...@teazles.com> test-address@teazles.com
local delivery failed
The following text was generated during the delivery attempt:
------ <ma...@teazles.com> test-address@teazles.com ------
An error was detected while processing a file of BSMTP input.
The error message was:
421 Lost incoming connection
The SMTP transaction started in line 0. The error was detected in line 3.
0 previous messages were successfully processed. The rest of the batch was
abandoned.
421 Lost incoming connection
Transaction started in line 0 Error detected in line 3
****************************************************************************
********************************************
The more research I did, I couldn't figure out what in the world was causing
this. I downgraded back to Exim 4.53 and SpamAssassin 3.0.4 (this setup
worked fine before) and the problems continued. Finally after killing the
SpamAssassin process, Exim 4.53 ran fine for a few days. Then I upgraded to
4.54 and finally 4.60 after each upgrade of Exim was confirmed to work.
However, again this evening once I recompiled SpamAssassin 3.1.0 and started
the process with the command /usr/bin/spamd -d -c -m 5 the same issues
immediately started.
Customers reported that it was very sporatic and random. They could email
someone and get the above 421 error, then reemail them and it'd work.
Several people who tried forwarding me copies of their error messages were
unable to get copies of the emails to me. I immediatedly started checking
my Exim logs and noticed BSMTP related errors there such as the following:
2006-01-10 23:05:23 SMTP connection from mail lost while reading message
data (header)
2006-01-10 23:05:23 1EwX9q-00060G-ML < <ma...@TEAZLES.COM>
test-address@TEAZLES.COM>: spamcheck transport output: An error was detected
while processing a file of BSMTP input.
2006-01-10 23:05:23 1EwX9q-00060G-ML ** <ma...@teazles.com>
test-address@teazles.com F=< <ma...@EFFORTSLIST.ORG>
owner-EFFORTS@EFFORTSLIST.ORG> R=spamcheck_director T=spamcheck: Child
process of spamcheck transport returned 2 from command: /usr/sbin/exim
(preceded by transport filter timeout while writing to pipe)
2006-01-10 23:05:23 1EwXEh-00062B-1P <= <> R=1EwX9q-00060G-ML U=mail P=local
S=34000 T="Mail delivery failed: returning message to sender" from <> for
<ma...@EFFORTSLIST.ORG> owner-EFFORTS@EFFORTSLIST.ORG
2006-01-10 23:05:23 1EwX9q-00060G-ML Completed
I really don't know what to say outside of this appears to be an issue that
Exim and SpamAssassin are having issues working together. All of my
webhosting customers are expecting quality spam filtering and protection. I
don't know what to do if SpamAssassin and Exim aren't working now, nor are
they working if I downgrade either. My temporary solution has been to run
Exim 4.60 but leave SA 3.1.0 turned off until I can get a solution
implimented.
Does ANYONE have any ideas on what direction to take??!
Thanks
Brad
Re: Exim 4.60 & SpamAssassin 3.1.0 Problems
Posted by Stanislaw Halik <we...@tehran.lain.pl>.
Stanislaw Halik <we...@tehran.lain.pl> wrote:
> my temporary solution would be to put:
> timeout_defer
> ignore_status
> into the SA exim router.
d'oh, sorry. i've meant the transport, not the router.
--
Stanisław Halik, http://tehran.lain.pl
Re: Exim 4.60 & SpamAssassin 3.1.0 Problems
Posted by Stanislaw Halik <we...@tehran.lain.pl>.
Bradley Walker <ba...@modemnet.net> wrote:
> However within hours of making this upgrade customers started calling
> me nonstop that their email "wasn't working".
[snip]
> 2006-01-10 23:05:23 1EwX9q-00060G-ML **
> <ma...@teazles.com> test-address@teazles.com F=<
> <ma...@EFFORTSLIST.ORG> owner-EFFORTS@EFFORTSLIST.ORG>
> R=spamcheck_director T=spamcheck: Child process of spamcheck transport
> returned 2 from command: /usr/sbin/exim (preceded by transport filter
> timeout while writing to pipe)
> My temporary solution has been to run Exim 4.60 but leave SA 3.1.0
> turned off until I can get a solution implimented.
my temporary solution would be to put:
timeout_defer
ignore_status
into the SA exim router.
you should see what happens during that timeout. most probably it is SA
spinning the CPU for few minutes. are you using bayes? i've had such
problem when the token file got *really* big. debug might help, too.
and these two lines are useful even as something more than a workaround,
they insure that whenever SA timeouts/dies, a message is deferred
instead of bounced.
--
Stanisław Halik, http://tehran.lain.pl
Re: Exim 4.60 & SpamAssassin 3.1.0 Problems
Posted by Ronan <r....@qub.ac.uk>.
Loren Wilton wrote:
> > Does ANYONE have any ideas on what direction to take??!
> I can't specifically help with your problem, I don't use Exim and have
> never seein anything like this reported.
>
> However, if 3.0.5 will work for you that would certainly be a pretty
> gool alternative to 3.1.0 until whatever this problem is gets figured out.
>
> Loren
>
I am also having, what I think is load related issues with my MySQL
bayes DB and exim4.60 SA3.1 .
I have a bayes_seen DB with 4.5m records and 650MB in size. How should I
deal with this? I know I could just blow it away as my corpus is quite
accurate... Is there another (less destructive) way?
Ronan
Re: Exim 4.60 & SpamAssassin 3.1.0 Problems
Posted by Loren Wilton <lw...@earthlink.net>.
> Does ANYONE have any ideas on what direction to take??!
I can't specifically help with your problem, I don't use Exim and have never seein anything like this reported.
However, if 3.0.5 will work for you that would certainly be a pretty gool alternative to 3.1.0 until whatever this problem is gets figured out.
Loren
Re: Exim 4.60 & SpamAssassin 3.1.0 Problems
Posted by "George R. Kasica" <ge...@netwrx1.com>.
Bradley:
Fought the same battle here just last week literally. With the help of
Larry Rosenman from the SA/Exim lists we got it working VERY well
here. It's basically a machine load issue for me, and I'm guessing for
you as well.
First thing...with SA are you running either of these rules:
blacklist-uri.cafe
blacklist.cf
They are both HUGE CPU hogs, remove tem from your rule sets.
Second:
Are your cleaning up after exim/SA? If not this script will do it for
you, I'd run it nightly around midnight here check the path names and
correct to match your setup:
# more /usr/sbin/exim-cleanup
exim_dbdir=/var/spool/exim
exim_tidydb=/usr/local/exim/sbin/exim_tidydb
echo ""
echo "Tidying Exim hints databases:"
for db in "$exim_dbdir"/db/*.lockfile; do
echo ""
"$exim_tidydb" "$exim_dbdir" `basename $db .lockfile`
done
ll /usr/local/exim/exiscan/virusmails
ll /usr/local/exim/spool/scan
rm /usr/local/exim/exiscan/virusmails/*
rm -r /usr/local/exim/spool/scan/*
These 2 items basically took my system load from a 10-12 and put it at
.89 and my mail queue from HOURS of queue time to avg less than a
minute:
Plus these command lines (with appropriate editing will give some nice
stats:
Subject: Cron /usr/sbin/sa-stats -l /var/log/exim -f mail
From: root@netwrx1.com (Cron Daemon)
Date: Wed, 18 Jan 2006 23:50:47 -0600
Email: 22065 Autolearn: 2 AvgScore: 13.51 AvgScanTime:
32.73 sec
Spam: 10091 Autolearn: 1 AvgScore: 29.70 AvgScanTime:
32.59 sec
Ham: 11974 Autolearn: 1 AvgScore: -0.14 AvgScanTime:
32.85 sec
Time Spent Running SA: 200.62 hours
Time Spent Processing Spam: 91.34 hours
Time Spent Processing Ham: 109.28 hours
Subject: Cron /usr/local/exim/sbin/eximstats -ne -nr
/var/log/exim/mainlog
From: root@netwrx1.com (Cron Daemon)
Date: Wed, 18 Jan 2006 23:50:58 -0600
Exim statistics from 2006-01-15 05:05:09 to 2006-01-18 23:50:48
Grand total summary
-------------------
At least one
address
TOTAL Volume Messages Hosts Delayed Failed
Received 107MB 7712 1032 417 5.4% 113
1.5%
Delivered 266MB 36558 378
Deliveries by transport
-----------------------
Volume Messages
address_file 2518KB 544
address_pipe 11MB 1196
procmail 84MB 6393
remote_smtp 169MB 28425
Try these suggestions and let me know how it goes.
I'm happy to try to help out more.
George
===[George R. Kasica]=== +1 262 677 0766
President +1 206 374 6482 FAX
Netwrx Consulting Inc. Jackson, WI USA
http://www.netwrx1.com
georgek@netwrx1.com
ICQ #12862186