Re: [vps-mail] Filtering catch-all's misses

[Abigail Marshall <abigail@xxxxxxxxxxxx>]

JS> I set a Procmail rule to get rid of all
JS> such spams.

JS> Now that I'm using Bayes, I'd like to somehow turn this
JS> into a SA rule so it tags each of these as spam (right
JS> now it sends them directly to spam folder without
JS> running through Bayes).

Jim, I have a slightly different approach to this. I have a
number of Procmail recipes running before SA that are
virtually certain to catch real spam; for example, spam to
recipients who never existed except as an invention of
spammers, or spam containing URL's to known spam sites.
After these recipes, I send a copy of the spam to a
temporary file called "spamfile".

This looks like this:

{
  LOG="Link to Spam Site "
  :0 c
  /var/tmp/spamfile
  :0
  /var/mail/spamdeath
}

Every night at about 1 am, I have a cronjob to run sa-learn
on the "spamfile". This is done via a shell script that
recycles the file as soon as the sa-learn job is completed.

So the commands look like this:

# Run sa-learn on secondary Spam file

  ulimit; /usr/local/bin/virtual /usr/local/bin/sa-learn --spam --mbox ~/var/tmp/spamfile

  cp ~/var/tmp/spamfile ~/var/tmp/spamfile.save

  rm ~/var/tmp/spamfile

NOTE: there are only 3 lines, they do NOT wrap. These are
for a VPS-1 - for a VPS-2 you would eliminate /usr/local/bin/virtual

So basically, all of the procmail-filtered email that is
certain to be spam is sent to Bayes via this route. This has
the advantage of feeding the spam to Bayes without adding to
the Spamassassin load.

-Abigail

======================================================================
This is <vps-mail@xxxxxxxxxxxx>       <http://www.perlcode.org/lists/>
Before posting a question, please search the archives (see above URL).