[ On Sunday, March 16, 2003 at 10:06:12 (+0100), Florian Weimer wrote: ]
> Subject: Re: [Exim] Reaction to rude 554 greeting
>
> "Greg A. Woods" <woods@???> writes:
> > <attribution lost>
> > >
> > > Some people use the 554 greeting to indicate "no mail under this
> > > domain", some to say "please try the next MX".
> >
> > Those who think the latter have not properly interpreted the RFCs.
>
> This may be the case, but their practical experience shows them that
> it works, most of the time.
Limited practical experience is never a good substitute for a sound
understanding of a protocol definition when it comes to managing
interoperability issues. The sooner people learn this the better.
> On my own MTAs, I won't play by their rules, and have no problem with
> the current Exim behavior. At work, I'd really like to maximize
> likelihood of mail delivery, by mimicking the behavior of other MTAs.
You are also misunderstanding the definition of the protocol. Your job
as postmaster is not to maximize the likelihood of mail delivery -- your
job as an SMTP postmaster is to ensure the reliable handling of e-mail
messages, and that includes bouncing them when they cannot, and as in
this case MUST NOT, be delivered.
> "Greg A. Woods" <woods@???> writes:
> > Any 5xx response is always a _permanent_ failure, even if it is given
> > instead of a 220 greeting response.
>
> Permanent failure for that host, yes. But it's a bit hard to tell if
> it applies to the message which is being delivered. After all, the
> receiving MTA has never seen the message.
No, it's not hard at all. The 5xx _always_ applies to the message which
the mailer is trying to deliver at the moment. Period. You seem to be
confusing the content of a message with the message envelope. The
message envelope is just and only the SMTP transaction which is used to
deliver the content. When that transaction starts out with a "5xx"
response to the TCP connection then that message _MUST_ be bounced
immediately (_and_ delivery attempts to the target SMTP server MUST NOT
be attempted for at least another 30 minutes).
> "Greg A. Woods" <woods@???> writes:
> > A 5xx response to a connection attempt simply means fail the current
> > delivery attempt (i.e. bounce just the one message from the queue).
> > Nothing more, and certainly nothing less. Do not pass go, do not
> > collect $200, and definitely do not go on to any other MX host!
>
> Again, the problem is that people use the 554 opening message
> successfully, with a different meaning.
No, that's _NOT_ the problem.
The problem is in people thinking they can do something about this
automaticaly in software without human intervention. The RFCs state
very clearly that human intervention is absolutely required here! Only
the human sender can ultimately determine the correct course of action!
--
Greg A. Woods
+1 416 218-0098; <g.a.woods@???>; <woods@???>
Planix, Inc. <woods@???>; VE3TCP; Secrets of the Weird <woods@???>
