[KLUG Members] simple pop access setup

Jamie McCarthy jamie at mccarthy.vg
Tue Nov 16 07:58:42 EST 2004


awilliam at whitemice.org (Adam Tauno Williams) writes:

> > 1) I wanted to install spamassassin.
> 
> My only advice regarding such packages (Spam Assasin and it's
> horde) is "don't bother".

Huh?  Spamassassin is the greatest thing since sliced bread.
Version 3 has two new features that catch a _lot_ of spam:  Pyzor
support and URLBL checking.  And it now runs by default as a daemon
with 5 child processes pre-spawned, so it's quite fast.  If you
haven't tried it lately, give it another try, it works great.

> I created a Spam folder in each users Maildir folder to put
> suspect messages. I am having a hard time figuring out how it
> fits together though. It seems I need to add some information
> to my maildrop since postfix sends messages to maildrop.

I've never used any of these programs :) but SA's install page

http://spamassassin.apache.org/doc.html

includes two walk-throughs for configuring SA with Postfix in two
different ways.  The one includes Procmail support which I don't
know if you could use or not (but it does show a Procmail rule for
catching the relevant header: "^X-Spam-Level: \*\*\*\*\*\*"...
basically at some level you will have to scan for that header with
some number of asteriskes depending on how aggressive you want to
be).

Also, if anyone wants the magic formula to make SA integrate well
with Exim, I can help there too, that's my MTA of choice.

If you want SA's Bayesian filtering to work well, you'll want to
provide a way for your users to seed it with about 1000 messages of
each type (spam and non-spam);  after that it will auto-train
reasonably well.  If you want suggestions for how to do that, I can
try to help.  But even if you don't use SA's built-in Bayesian
checking, SA also is very useful for modern mail clients that
provide Bayesian spam filtering.  Since it loads up both spam and
non-spam messages with helpful identifying tags in the headers,
those are used by the client, resulting in faster and more accurate
training.
-- 
  Jamie McCarthy
 http://mccarthy.vg/
  jamie at mccarthy.vg



More information about the Members mailing list