very urgent problem with MailScanner + postfix

David Vosburgh vosburgh at DALSEMI.COM
Fri Nov 7 18:38:37 GMT 2003


Try setting "Debug = yes" and "Debug SpamAssassin = yes" in your
MailScanner.conf file.  Alternatively, you could try to just run
spamassassin manually on a message in debug mode: spamassassin -D -t <
message_file.  If you choose the latter, make sure your (if run as root)
/.spamassassin/user_prefs is the same as your MS
spam.assassin.prefs.conf so you get the same behavior.

Dave

Ron E. wrote:

>Well, interestingly enough, after the backlog cleared out, I restarted the
>incoming postfix and very rapidly developed another backlog - and now I'm
>having the same problem you had - ie - SpamAssassin timed out and was
>killed...
>
>Any idea how to get SpamAssassin to log everything it's doing so I can see
>where it's bombing out - I've got DCC/razor/pyzor going but I've disabled
>the RBLs temporarily - just going really, really slow.
>
>I will try the lower scanning settings as below...
>
>-Ron
>
>On Fri, 7 Nov 2003, David Vosburgh wrote:
>
>
>
>>I had a different message - "SpamAssassin timed out and was killed,
>>consecutive failure 1 of 20", but the same general problem (extreme
>>slowness delivering mail).  My actual problem turned out to be an
>>ex-user that was forwarding corporate email to a hotmail account that
>>had hit it's quota. His email was getting bounced back and forth,
>>growing all the time.  After we changed his alias to /dev/null and
>>restarted MS, the queue cleared very quickly.  I've since reduced the
>>number of messages scanned in one pass, Max Unscanned/Unsafe Messages
>>Per Scan, from 30 (default) to 10.  FWIW, it does appear that my system
>>(Sun 220R w/ 2x450Mhz,1GB RAM) is "happier" with this setting.
>>
>>Dave
>>
>>Ron E. wrote:
>>
>>
>>
>>>Great, thanks Dave - yes I just found Monkeys still in my RBL list so I'll
>>>check further but just based on that my performance has gone way up
>>>already.
>>>
>>>Did you also have the problem with duplicate truncated messages being
>>>delivered? And messages in the maillog about messages "still being
>>>delivered"?
>>>
>>>
>>>-Ron
>>>
>>>On Fri, 7 Nov 2003, David Vosburgh wrote:
>>>
>>>
>>>
>>>
>>>
>>>>I had a similar problem several weeks ago.  I'd first take a look at
>>>>your mail log file for signs of any delays while running MS.  In my
>>>>particular case, there were SA timeouts during MS processing.  If you
>>>>want to jump the gun, you could just disable SA in the mailscanner
>>>>config file, restart MS, and see if things speed up.  If it is SA, then
>>>>turn on debugging to see where things are slowing down.  You should look
>>>>at the various net checks going on there (razor/pyzor/dcc and any
>>>>RBL's).  You could selectively disable each one at a time in the
>>>>spam.assassin.prefs.conf file to see which one is causing problems.
>>>>Also, several RBL's have gone out of business in the last few
>>>>months...you should make sure you aren't using them.
>>>>
>>>>Dave
>>>>
>>>>Ron E. wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>Hopefully someone will see this very shortly.
>>>>>
>>>>>Several days ago I posted a message about the same problem where
>>>>>mailscanner begins to backlog and errors about messages "still being
>>>>>delivered" show up in the maillog, with hardly any messages passed by
>>>>>mailscanner to the mta.
>>>>>
>>>>>I've since upgraded to 4.24-5 due to a backlog starting to happen again
>>>>>but it never cleared up and is unfortunately much worse this time - (much
>>>>>larger backlog).
>>>>>
>>>>>I've temporarily stopped the incoming postfix to give mailscanner a chance
>>>>>to catch up, but need some debugging here to get the backlog cleared out
>>>>>as fast as possible and to fully fix this...
>>>>>
>>>>>Any help would be appreciated.
>>>>>
>>>>>Thanks,
>>>>>
>>>>>
>>>>>Ron
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>--
>>>>
>>>>Dave Vosburgh
>>>>Sr. Unix System Administrator
>>>>Dallas Semiconductor
>>>>vosburgh at dalsemi.com  972-371-4418
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>
>
>



More information about the MailScanner mailing list