On Thu, 30 Jul 2009 20:30:58 -0500 Chris Bennett <[email protected]> wrote:
> As I understand it, spamdb works by using the fact that a > large portion of spam email is sent only once, so putting > a delay by temporarily rejecting incoming emails from > unknown addresses can dump a lot of the spam with that > one rejection. Good email will always go through > eventually. Some spam too, of course. > > So, normal spam - temp rejection, goes away. > Good email - temp rejection, keeps trying till successful > and is whitelisted. > Extra bad spam - temp rejection, keeps trying till > successful and is whitelisted. > > Please correct me if I am wrong. Good emails and extra > bad spam have equal chance of getting through? > If this is true, would it be reasonable to significantly > reduce greylisting time after getting second try from a > non-listed IP? (To get new IP good emails faster) > > Is this a silly idea or not? It's not a silly idea. But some of the spammers will make multiple times over a period of several minutes before quitting. Shortening the time will permit several of these instances through. We use "-G 10:12:864" (accept after 10 minutes, expire after 12 hours, and keep for 864 hours). The default is "-G 25:4:864". The problem is that some of our users were getting upset when an expected e-mail didn't arrive promptly. With the lower setting, I get fewer complaints and there is no apparent increase in spam. Eric -- Eric <[email protected]> [demime 1.01d removed an attachment of type application/pgp-signature]
