Rod Whitworth wrote: >>> Anybody run into this kind of logic before? >> Yes, that's part of how greytrapping works: >> http://www.openbsd.org/cgi-bin/man.cgi?query=spamd#GREYTRAPPING > > No. That is NOT how greytrapping works. RTFM more carefully. > > spamd NEVER issues a 2xx code, because it NEVER accepts any mail.
I did RTFM carefully. I don't see anything in the spamd manpage that indicates one way or another what response is sent in the specific case of greytrapping. So I assumed it did, because that's the way I've seen other greytrapping systems whose code I've read worked. Perhaps you can point out my mistake. But your comment got me curious, so I poked at the source, and it looks like it never lets the sender get far enough in the DATA to be done before issuing a 450/550 (per -4/-5); it only issues 2xx codes (and it's not "NEVER") to string the connection along. >> I've seen other implementations do greytrapping for *every* invalid >> address that comes through, too. > > And that's a great way to blacklist a genuine sender who misheard an > email address and so misspelled it. S/he will never get a 5xx that > flags the problem. John Brooks asked if anyone had run into this before. Yes, I have. Hell, I'm pretty sure this approach has been presented at LISA before. -- Matthew Weigel hacker unique & idempot.ent
