Cyrus Daboo wrote:
The effectiveness of reducing joe-job DSNs depends on exactly how the
original spam message is being submitted and sent. If the submission server
(MSA) is the client that connects to the final delivery server (MDA) where
SIEVE is being run, then the DSN/MDN is avoided. However, if there is one
or more intervening MTA's relaying the message, then a DSN will always be
generated by the client connecting to the MDA when the script does the SMTP
error refuse. So the question is how many messages fall into these two
categories: 'direct' delivery vs 'relayed' delivery. That will determine
the real effectiveness of refuse.
indeed. I claim that most spam is submitted directly to an MX
responsible for the final recipient. the spammer software will
obviously not produce an MDN in response to the DSN.
I want to make it possible to run a user's Sieve script on the border.
this can't be guaranteed in general, but a system designer can make sure
that the MTA on the border supports the same Sieve extensions as the
MDA.
--
Kjetil T.