One issue I had though is how is the best way to fight spam?
I tried everything! Grey lists worked but caused a bit of a delay in email delivery. Real time blacklists worked fantastically (once I figured how to properly set them up). However, they tend to block so well that it didn’t give my Bayesian enough time to learn what is spam and what are genuine emails.
Eventually, I came to the conclusion that RBL’s combined with a Bayesian filter did the trick for me. But it required me to think outside the box and several days of monitoring my email server logs.
But rather than block the mail that hit the RBL, I set it to a custom event. e.g. ‘RBL Triggered’. I was then able to use the content filter to redirect any mail with this custom event to a specific email address on the mail server. I was then able to look through these emails and put them in a spam folder to manually train the Bayesian filter.
Before long I was able to train the Bayesian filter and then set another content filter rule to redirect any mail with a Bayesian score of 55 and over to redirect to the mail address dedicated for spam.
I then tweaked these settings again so the RBL’s reject if three or more blacklists are hit and modified the content filter rule to send a copy to the spam email if one blacklist is hit.
This way I can still check for spam, but rule out false positives. I also found that some Gmail email addresses can make their way to blacklists due to spammers abusing free email services!
So which RBL’s are best? I found these ones work quite well for me: