On Tue, Oct 23, 2001 at 07:31:54PM +0100, Roger Burton West wrote:
> On Tue, Oct 23, 2001 at 06:45:16PM +0001, Ken Jackson wrote:
> >So the empty <> is nothing to do with 'spam' and is actually a desired thing
> >to have? Or am I reading it incorrectly?
>
> The sequence goes:
>
> Spammer sends message to non-existent user at your domain.
>
> Your mailserver tries to send back "that user doesn't exist", with the
> correct MAIL FROM: <>.
>
> This fails, because the spam doesn't contain a valid return address.
>
> Thus a delivery failure report has itself failed; so you end up with
> a frozen message from <>.
Running big mail servers, where I get more of those than I'd wish, I've
thought that the problem would be fixed by having secondary MXes not accept
MAIL FROM: <> and give back some temporary error, with the hope that
the sender will eventually connect to the primary server, which can
authoritavely say "no, I don't know this user" and refuse the mail at SMTP
time or accept the mail if it is directly deliverable.
This means you do not have to generate a bounce and you do not have problems
with unbounceable mail.
I do realize that it's stretching the RFCs a bit, but so does SMTP callback,
so...
Philip, would you consider that as a feature for exim (whether you're the
one who writes it or not) or is this misguided?
Marc
--
Microsoft is to operating systems & security ....
.... what McDonalds is to gourmet cooking
Home page:
http://marc.merlins.org/ | Finger marc_f@??? for PGP key
