Spamassassin, mailscanner 3, etc.

Rose, Bobby brose at MED.WAYNE.EDU
Wed Jan 16 02:42:41 GMT 2002


In the contruction of the @wholemessage array should the headers get
pushed to the $dfilename? If the header is included, then the score from
spamassassin -t $dfilename is closer to what it gets from the
mailscanner hand off.

I think it might be a good idea to add the score into the header that
way we can see it for debugging since I think I know what maybe
occurring.

We're saving the message from the email client, eg pine and piping that
message off to spamassassin which seems to generate a higher score than
what the same message gets if processed by the MTAs.  

I'll setup procmail and see what the score is for the same messages.

-----Original Message-----
From: Rose, Bobby 
Sent: Tuesday, January 15, 2002 9:23 PM
To: MAILSCANNER at JISCMAIL.AC.UK
Subject: Re: Spamassassin, mailscanner 3, etc.


I added get_hits () to the sendmail.pl and also uncommented the line to
save to a queue for analysis (had to move it into the routine where
$dfilename was getting set though).

Since $dfilename doesn't include the header, I have to subtract 8.33
from both.

For a message that doesn't get marked as spam, it's score was 4.26 when
mailscanner passed it off.  If I went to the queue where I dumped
$dfilename and passed that same message thru spamassassin -t it scored
13.2-8.33=4.87

I then sent a message that did get tagged as spam thru.  It's score when
passed off from mailscanner was 5.92.  When I piped it's $dfilename to
spamassassin -t it's score was 13.79-8.33=5.46

I'm going to check tomorrow and see if the headers are actually not
included or if I have the line to save a copy $dfilename in the wrong
place.

-=B




-----Original Message-----
From: Julian Field [mailto:jkf at ECS.SOTON.AC.UK] 
Sent: Tuesday, January 15, 2002 4:16 AM
To: MAILSCANNER at JISCMAIL.AC.UK
Subject: Re: Spamassassin, mailscanner 3, etc.


At 05:41 15/01/2002, you wrote:
>I've been running the same testing and seeing the same results.

<AOL>Me too!</AOL>

I haven't had time to debug SpamAssassin, I know from the docs (and some
sample code supplied by its author) that I'm calling it correctly, and
most of the time it works as it should.

But it still gets it wrong some times. Their compile_now() method
definitely has bugs, it was causing some people's setups to report every
message as spam, and on my own systems was causing it to never read the
preferences file. I took out the call to compile_now() and all those
problems just vanished.

Anyone who feels like debugging SA, you are very welcome to the job!

>   I've
>been trying to narrow it down also.  It's definitely not a timing issue

>since I can pass the same message thru several times on my test box and

>SpamAssassin still returns 0 to mailscanner.  But if you run the
>cmdline program against it, it gets tagged as spam. I want to say that 
>it has something to with html formated messages and the perl 
>spamassassin apis though I'm trying to figure out how it functions 
>while trying to diag the issue.

--
Julian Field                Teaching Systems Manager
jkf at ecs.soton.ac.uk         Dept. of Electronics & Computer Science
Tel. 023 8059 2817          University of Southampton
                             Southampton SO17 1BJ



More information about the MailScanner mailing list