[ On Wednesday, August 11, 1999 at 10:59:21 (-0400), Dave C. wrote: ]
> Subject: Re: [Exim] RBL rejection at processing time
>
> Ah ok.. I was just going by the transcript that was posted, and didnt
> actually check.. Is there perhaps a different 5xxx code though, that
> might produce a more generic interpretation by the sender-MTA?
No, 550 is the best error code for rejecting a connection due to RBL or
other security checks or policy reasons. It is recommended by both RFC
821 and the upcoming SMTP draft (though in the case of "RCPT TO:" 821
does say explicitly "If the recipient is unknown the receiver-SMTP
returns a 550 Failure reply.").
Sendmail is (or at least was -- I'm not sure if more versions more
recent than the 8.8.8 I have on hand have fixed this) extremely bad at
wrongly categorizing the meaning of SMTP replies (which is in fact
intended for the end user!) with its own interpretation, at least when
reading reply codes from SMTP "RCPT TO:" commands. 550, 551, and 553
are all treaded as EX_NOUSER and the reply buffer text is only saved to
be buried in the text of the bounce message. Any other 5xx code is
returned as EX_UNAVAILABLE, and I'm not sure what happens then....
I've found that bouncing mail for policy reasons works OK in terms of
getting the point across if you send a 550 to *every* command they send.
The only drawback is that some mailers won't immediately bounce a
message as required by the spec., especially if they get the 550 from
their HELO/EHLO greeting. This though has the potential benefit that it
really makes the sending user mad and if the reply text is carefully
worded you can direct all of their anger at their own postmaster and
usually that'll get at least some response from the person who is best
able to do something to fix/replace their broken software.
--
Greg A. Woods
+1 416 218-0098 VE3TCP <gwoods@???> <robohack!woods>
Planix, Inc. <woods@???>; Secrets of the Weird <woods@???>
