Rejecting some recipients after DATA?

Vernon Schryver vjs@calcite.rhyolite.com
Sat Apr 20 21:39:46 UTC 2002


> From: "Brian J. Murrell" <dcc-list-in@interlinx.bc.ca>

> > There is a similar problem already in dccm.  If only some envelope
> > RCPT values of a message are white listed and the checksum counts for
> > message exceeds the system-wide thresholds, then DCCM silently removes
> > the non-white-listed RCPTs from the message.
>
> Ugh.  You have a recipie for disaster there.  In the case of something
> legitimately bulky, the recipient does not get the message and the
> sender does not get a bounce of any kind, leaving everyone wondering
> where the mail is going.

I hope "disaster" is too strong, because at least some of the mail
gets through and because the message is logged for those not whitelisted.

What alternative is there?  On a non-trivial system you cannot
reduce the "recipients buffer" to 1 even if don't care that
page 55 of RFC 2821 demands 100.  You can't partially succeed or
fail on the DATA command.  It would be worse to reject the message
for everyone, because you know at least some recipients want it.

> > If there were a way to trigger Delivery Status Notifications through
> > the Milter interface, then I think the problem could solved for dccm.
>
> Interesting idea.  If any recipients for a message are to be
> delivered, accept the message, complete with 2yz and then DSN for the
> recipients that don't want it.  In the case of forged sender spam, the
> postmaster will get a double-bounce, but in the case of legit
> bulkiness, the sender at least knows the recipient considered it
> unsolicited.

Exactly.

> Wouldn't it be nice if Sendmail (or other MTA) had a feature, even on
> the command line to take a message and create and send a DSN for it.

I don't think sendmail would need or could use such a command line
mechanism, but it seems to me that it would fit nicely with the Milter
interface.
(I'm really trying to not think of the obvious solution of having
dccm generate the DSN.)

> Anyone know any other tools for easily creating a DSN from a message?
> I am sure I could cobble one together, but if it's already been
> done...

One of the reasons I'm trying not to think that unthinkable thought is
that I've recently seen Internet-Drafts about DSNs.  I fear DSNs are an
infinitely deep and wide swamp and best handled by someone, anyone else.


Vernon Schryver    vjs@rhyolite.com



More information about the DCC mailing list

Contact vjs@rhyolite.com by mail or use the form.