Re: Message delay warning messages

Top Page
Delete this message
Reply to this message
Author: Chris Thompson
Date:  
To: A.Sharaz
CC: exim-users
Subject: Re: Message delay warning messages
Alex Sharaz writes:
[...]
> I've got a number of scripts that generate a standard set of automatic replies
> for mail messages going to
>
> a). Invalid e-mail addresses
> b). duplicate addresses
> c). People who have specifically requested that they do not receive e-mail
> e). etc .....


You shouldn't think of things like this as "automatic replies". They are
delivery failure reports.

Well (a) and (c), anyway. I am not sure what you mean by (b).

> The scripts run on our PP mail hubs and they send an auto-reply to either
>
> The Reply-To address if it exists,
> The From Field if it exists,
> The address passed as an argument to the shell
> script.
>
> The list in question has a Reply-To field set up to be the sender of the
> message, while the 1st argument to the shell script is of the form
> owner-<listname>-l@........


That last thing what you is refering to is the "return path", a.k.a. the
"envelope sender". It is where delivery failure reports should be sent to.

> The list manager says that I should send all error to the owner... address and
> not the reply-to or from addresses.


She or he is absolutely right.

> What do other people do? If it's the LISTSERV d-list owner, do you just check
> for the "owner-XXX-l" string in the sender (?) header? If I have to check for
> messages from LISTSERV type list, are their any other d-list services I have to
> check for as well?


You shouldn't be doing anything different for "LISTSERV type messages" than
for any other message. Always send delivery failure reports to the return path
that came with the message (unless it is itself an error report, indicated
by having a null return path: in that case do not send a report at all).
It is all explained in RFC 822.

An example of something that is normally considered an "automatic reply",
rather than a delivery failure or warning report, is a vacation message.
Here the message does go to the Reply-To address, and the way that noise
from mailing list expansions is avoided is to send it only if the To (or
sometimes Cc) headers contain a recognisable personal e-mail address for
the recipient.

Chris Thompson               Cambridge University Computing Service,
Email: cet1@???    New Museums Site, Cambridge CB2 3QG,
Phone: +44 1223 334715       United Kingdom.