Slightly OT: Greylisting success or failure stories?

[Scot L. Harris]

On Fri, 2005-02-04 at 09:39, David Hoffman wrote:

> 
> Paul emphasizes part of the point I was hoping to make originally.
> Prior to using TMDA, I had LOADS of spam in several accounts. I did
> whatever i could to make those accounts as hidden as possible. I never
> used them in newsgroups or to register at web sites, but soon enough
> the spam came pouring in. Even with RBLs, I was still getting way too
> much spam. So I tried TMDA and it has worked great for me. If someone
> wants to get in touch with me, they almost certainly understand that
> clicking a link or sending a reply to a confirmation is a good thing
> and the only ever have to do it one time. The people who get upset
> about TMDA are the spammers.
> 
> That being said.... my intention, if I can find a viable solution, was
> to try using greylisting, followed by RBLs, and then possibly backed
> up with SpamAssassin or ClamAV. Basically take TMDA out of the picture
> altogether.
> 
> The reason I was asking for information was not to start a C/R flame
> war. It was because some articles on greylisting talk about how
> non-compliant MTAs can break the greylisting system by NOT sending
> back legitimate messages after the delay, or by seeing the delay
> response as an error and reporting it back to the original sender.
> 
> All I wanted to know is if anyone has seen issues like that and how to
> get around them.

I have not seen any of the potential problems people think they will
have with greylisting.  It actually worked far beyond what I had thought
it would.  Greylisting backed up with spamassassin is virtually 100%
solution for spam with the added benefit that it reduced by several
factors the CPU overhead I had when just using spamassassin.  BTW:
spamassassin by itself with a properly trained bayes database can get
you in the 95 to 98 percent range for blocking all spam.  But on a site
that gets many thousands of spam messages a day it can become resource
intensive.

Greylisting is well worth the small amount of time to setup and test. 
Then you can judge the results yourself.  I would be interested in
hearing how it does for you.

-- 
Scot L. Harris
webid at cfl.rr.com

Test-tube babies shouldn't throw stones.