SpamAssassin problem

N

nikdahl

Verified User
Joined
Mar 14, 2005
Messages
65
Location
Seattle, WA, USA
I used the script provided (spam.sh) to deploy SpamAssassin on one of our DirectAdmin servers.

I followed all the directions online for installing spamassassin, being, I ran the script, started spamd, editted the exim.conf, and then restarted exim.

Incoming mail is getting bounced back. I have since edited the exim.conf back to original form as a quickfix.. However, I would like to figure out what the problems are with what was set up..

Here are the error messages in the exim messages log:

2006-11-21 10:21:49 1GmaFg-00036g-GJ Completed
2006-11-21 10:22:58 SMTP connection from mail lost while reading message data (header)
2006-11-21 10:22:58 1GmaBy-00034C-1n <[email protected]>: spamcheck transport output: An error was detected while processing a file of BSMTP input.
2006-11-21 10:22:58 1GmaBy-00034C-1n ** [email protected] F=<[email protected]> 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)


Does anyone have any ideas? Are there other logs I can use to track this problem down?

TIA!
 
If you search these forums I think you'll find a lot of questions about BSMTP and no good answers.

Does anyone have a solution?

Jeff
 
Hi,

we have this same problem, if we mass mail on same server:

Using Debian 3.1 / Spamassassin 3.2.3 / Clam AV 91.2 / Exim 4.68

2007-12-18 12:16:34 SMTP data timeout (message abandoned) on connection from local process F=<[email protected]>
2007-12-18 12:16:35 1J4ZQE-0000zI-RO ** [email protected] F=<[email protected]> 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)

Also same configuration other servers, if we send same messaged there, not any problem, so very weird ...
 
I have had this problem before, for me it was happing for one user only.

That user had gone over their disk quota recently.

I fixed it by deleting their account & re-creating.

I had the same problem happen again recently. I noticed that the 'top' command showed a perl process gobbling up lots of CPU, this was spam assassin

I believe one of the spam assassin files was corrupt, in the:
/home/user/.spamassassin directory

The user_prefs & two bayes_ files. I deleted them, restarted spamc & exim, the files got recreated & now it works fine.

Just an FYI if anybody encounters the same problem, on a per user basis, and happens to be after they went over disk quota. I reckon one of the files got corrupted when SA was trying to write to it but the system denied it based on disk quota.
 
Recreating the account was overkill. Upping the quota or removing files might have fixed it. If not, then certainly recreating user_prefs and the bayes files should have.

Jeff
 
jlasman said:
Recreating the account was overkill. Upping the quota or removing files might have fixed it. If not, then certainly recreating user_prefs and the bayes files should have.

Jeff
Click to expand...

This was the first time I had encounted the problem and had no idea on how to fix it. Increasing the quota had been done, the issue remained [I suspect] because when SA last tried to write to the file, the write was incomplete due to quota restrictions.

I mention in my post that re-creating the account was what I did the first time, however since I have found what I believe to be a better fix. There are many posts around the net on what appears to be this exact problem, none with a fix.
 
damn,

I didn't intend to belittle your work; only to clarify for anyone searching for an answer that killing of the account isn't really necessary.

Thanks for adding to our knowledge.

Jeff
 
All good mate.


I can confirm that days later this account is still all good, so one of those files was certainly corrupt.
 
You must log in or register to reply here.
Back
Top