On Sun, Jun 13, 2004 at 04:28:43PM -0400, Greg A. Woods wrote:
> I.e. I have never witnessed nor seen any evidence whatsoever for there
> being any benefit to delaying the reject _at_all_, at least not when the
> client-SMTP is any true MTA (as I said all bets are off with MUA
> submission implementations).
There is a benefit where the additional data gathered is useful. Some
sites like to maintain statistics of how many mails you've prevented
from going to a specific user/customer for instance. There's no way to
tell unless you wait until the DATA stage. It can be very useful from
a customer-relationship point of view to go "hey, this month we
prevented N spams from getting into your inbox".
Other administrators (greylisters, exiscan + spamassassin users and so
on) base their allow/deny on a few of the parameters. It's not hard to
see a situation where mail is denied at the DATA stage because of a
single word server announcement, or literal IP address server
announcement, an alphabetically ordered recipient list and an RBL'd IP
address.
Obviously these approaches may open you to the spammers wasting more of
your CPU time, but if an admin judged the benifits worth it I don't
think that's invalid. The 5XX at the DATA stage should be as descriptive
as possible, and the vast majority of DSN's I've seen include the server
error string.
--
Colm MacCárthaigh / HEAnet, Teach Brooklawn, / Innealtóir Ghréasáin
+353 1 6609040 / Bóthar Shelbourne, BÁC, IE / http://www.hea.net/
