I've actually been eying tarpits for a while now, specifically leaders like MailChannels (which some high-volume colleagues of mine use, plus an award from the MIT Spam Conference). Their product is actually very similar to what could be added to milter-greylist with the implementation of a tarpit command. I wouldn't actually be surprised if tarpitting was in almost every way better than greylisting; I'd strongly consider reconfiguring my servers use tarpitting over greylisting in every case (perhaps with a 10s grey-time on Windows desktop OS "servers" before the tarpit). The sendmail configuration option "greet_pause" is a basic form of tarpitting with astounding affect. See the testimonial at http://www.acme.com/mail_filtering/sendmail_config.html#greet_pause and data at http://mailchannels.com/images/drop-off.png One important implementation note: if the connecting server drops the connection but then comes back later, the tarpit clock should have been counting from that first connection. (Otherwise, some noncompliant servers might never deliver mail.) One more note on tarpitting: the full-on implementation is actually that of a connection throttle; traffic is let through very very slowly. The idea of pausing often accomplishes the same thing, but it's easily interpreted as a lost connection. A good tarpit implementation would actually have variable dynamic throttle rates (or at least several bandwidth thresholds), and no tarpit mechanism should throttle longer than 300-500 seconds (see above linked image). As to taRgrey, it appears to bring two concepts to the table: the aforementioned tarpit ability, which is awesome, and S25C, which is some kind of botnet detector. After reading a bit on S25C, I'm quite dubious. No concrete data on false-positives is presented and the whitelist is MASSIVE. I've implemented S25C in SpamAssassin with near-zero scores to see what kind of impact it would have on my servers, but I doubt it will prove useful (since SA fires after greylisting). I suspect the "botnet" plugin for SpamAssassin is far more comprehensive, and I've already decided not to use it thanks to the fact that greylisting's main function is combating botnets. The same will probably go for S25R. Implementing S25R within milter-greylist once the tarpitting functionality is present should prove trivial, so I see no need to implement a "targrey" clause.
Message
Re: [milter-greylist] [RFC] implementing taRgrey
2009-07-06 by Adam Katz
Attachments
- No local attachments were found for this message.