[Top] [All Lists]

Re: 2821bis and address rewriting (was: Re: Fixing graylisting [was TBR])

2007-11-17 20:23:28

Hi John,
At 08:53 17-11-2007, John C Klensin wrote:
>I'm working this weekend on what I hope will be the last version of
>2821bis prior to IETF Last Call, so, if you think such a mention
>would be useful, please propose text and where it should go.

I have not come across any cases where the empty return path is

I've seen it happen quite a few times, although not so much recently. I'm
pretty sure Exchange used to do it unconditionally when autoforwarding. I'm not
sure if it still does.

If Ned thinks it might be a problem, some text could be added.

Well, the only reason I even found out that some agents do it is because we've
had customers have meltdowns because of it. The sorts of loop this causes
usually have messages that grow continually in size and in many cases the
number of messages in the loop grows geometrically over time.

In Section

   "In some types of reporting messages for which a reply is likely to cause a
    mail loop (for example, mail delivery and nondelivery notifications), the
    reverse-path may be null (see Section 3.6)."

That provides some guidance on how to avoid a mail loop.

In Section 4.5.5, Messages with a null reverse-path:

   "Implementors of automated email processors should be careful to make
    sure that the various kinds of messages with null reverse-path are
    handled correctly, in particular such systems SHOULD NOT reply to
    messages with null reverse-path."

I read that as a warning about the null reverse-path, i.e. not to
rewrite the address or else the message may generate a reply.

Agreed. But it needs to be clear that this applies to autoforwarders. How
about adding "and SHOULD NOT add a non-null reverse path to such a message
when forwarding".

I also note that there's a missing "a" in the above, it should read "with a
null reverse-path".