Slashdot Mirror

Then don't let them have an e-mail account. There is no perfect spam filter ... except you filter it by your own.

One solution offers the answer to both of these problems. Maia Mailguard. I'm a huge fan of that project and it is, in my opinion, the single most underpromoted open source app out there. It should be on every sys admin's (at least) radar.

With Mailguard you can set up customised filtering levels (based on spamassassin score). Want manual spam filtering for somebody's account? Set up two email addresses, one for the kid, one for you. Link their address to yours in Mailguard and set the spam threshold score to minus 100. Everything gets treated as spam, and you get a daily notification listing the messages and can go in and manually release the real stuff, and whitelist trusted senders. It even lists things in increasing order of spamminess so the legit stuff will be near the top.

If you are less paranoid, leave the spam detection score at a more reasonable level and let the stuff that is unlikely to be spam go through.

Our company forwards email to google (MX record in the DNS), where it runs through the spam filter and then a forwarding rule (an anything-but-spam rule) sends it on to our mailboxes.

Or you could just use Spamassassin, which properly configured is every bit as good as commercial offerings (and I have actually trialled them to do the comparison). If you put MAIA Mailguard on top of it, you have a solution that leaves the commercial offerings for dead - per user, server based sensitivity settings, quarantine, anti-virus and most importantly, no stupid bounces to the sender address of spam, since the sender address is almost always forged and if you are sending those stupid bounces you are the spammer.

Yes, I am sick of Messagelabs spamming me.

I'm surprised that nobody has mentioned "Maia Mailguard." http://www.maiamailguard.com/ I've never used it and would love to hear about it from folks who have. I was planning on getting around to test it one of these days...

A sysadmin I know runs maia mailguard (with spamassassin, clam-av[?]) on his small-midsize network, and since the users train it, and also get to see all their spam (if they want), they get to feel in control. Of course getting users to train it is a social issue. http://www.maiamailguard.com/

Anyone that's done much exploration of spam filtering already knows the basic architecture of self-learning filters. This article has nothing new.

Recently I've implemented greylisting on my mail server. The drop in spam has been enormous, though there have been a couple cases where email didn't go through on the first try.

Care to share your method that is so successfull? I'm sure a lot of other admins would love to know a system that results in very little spam and has a low false positive rate

If the ISP used Maia Mailguard, then they could indeed do just that.

(Disclaimer, I'm a Maia developer)