best way to combat porn email

[Steve Freegard]

Raymond Norton wrote:
> 
> I apologize for posting this much info, but I keep tripping the pastebin
> spam filter when pasting my info. I have included the mailscanner info,
> and matching headers.

Two things I immediately noticed:

> 93.196.6.57      p5DC40639.dip0.t-ipconnect.de

smf at laptop-smf:~$ host 57.6.196.93.zen.spamhaus.org
57.6.196.93.zen.spamhaus.org has address 127.0.0.11

Consider using the Spamhaus Zen RBL at the MTA level - both of these
spam messages were delivered to your MTA by consumer DSL lines which
this will prevent.  It will also mean that you machine have less mail to
scan.

> -9.00    BAYES_00    Bayesian spam probability is 0 to 1%

The default score for BAYES_00 is not -9; and that coupled with
autolearning has likely compounded your problem.

Hitting BAYES_00 on this message means that your bayes database is
garbage and has learnt messages incorrectly.  You might as well nuke it
and start over (sa-learn --clear).

I ran just the headers you posted into a well trained bayes database
here with simply 'Test' in the message body and it scored BAYES_60.

Regards,
Steve.