Hi,
I've been using sa for a while and it's worked very well for me. In
the past 18 month or so, spammers have discovered my domain name and
are sending tons of email with frenchguys.com in the from
([EMAIL PROTECTED], [EMAIL PROTECTED], and so forth). They're
not using my smtp server, just faking the from.
This results in hundreds of mail error emails a week as you can
imagine. Many of the email addresses the spammers use are invalid and
as a result i get tons of email sent to addresses such as
[EMAIL PROTECTED] Now the problem is that I use a catchall
mechanism so that when I sign up for say, flickr, I signup as
[EMAIL PROTECTED] With this mechanism, I can see quickly who's
selling my email address to who. And believe me, it happens often.
It's hard to make procmail rules that will identify correctly all
mail errors so I was thinking I might be able to use Spam Assassin's
bayesian filter to teach it what a mail delivery issue message looks
like and have it mark it in a certain way. I'd have to filter out
real mail delivery issues of course, those sent to very valid
addresses such as [EMAIL PROTECTED] (I never send email as
[EMAIL PROTECTED] obviously).
Is this feasible ?
Thanks in advance,
Erik
