ptimizing qmail/SpamAssassin to catch more Spams

When you first install SpamAssassin from either source, rpm or Ports, The default setup for SpamAssassin still allows a lot of spams to still get through. The following are a few recommended things to use to help SpamAssassin filter more spams out for you.

If you happen to get a lot of spams to invalid users, I would highly suggest taking a look at john Simpsons validrcptto patch. This is included with his combined patch as well. If you use his patch, you will also need to replace the original qmail-smtpd/run script with Johns as it breaks smtp-auth. Take a look at the following URLs

If you would like to enable jgreylist, follow this website:

The next thing I would recommend is to enable the Bayes Database. Take a look at:

I would suggest adding rules to SpamAssassin to mark up messages better. The only thing I have found about some of these rules is that when you run spamassassin --lint, You do get some errors from time to time.


Another good thing to use is SpamAssassin Auto-Learning with Site-Wide Bayes and User Feedback. This will allow your users to send spams or hams via forwarding as an attachment to report them to bayes as ham or spams.

If you implemented the site-wide bayes with feeedback, adding the Squirrelmail spam or ham reporting option makes reporting via Squirrelmail really simplistic.

There are some new spams going around that are gifs. Here is a link on how to get rid of those types of spams: