Re: [Exim] Ignoring bounce messages to alias - How?

Hi Alun, on Fri, 06 Feb 2004 13:35:14 +0000 you wrote:

> On outgoing mail, add a header containing a tag which changes every
> hour (e.g. MD5(secret . int(time/3600)))
> Keep a list of recent (say a month's worth?) tags that have been used.
> When a message from <> comes in, search it for any used tag and, if not
> present, drop the bounce.
<snip>

> The question is, how many mail systems bounce mail without including
> the headers?

I don't know, but I suspect a not-insignificant number. This actually ties
in with a discussion the other day about catching mail loops.

> I'm pretty sure I'm not the first person to think of this, so there must
> be a big flaw.

The main one that I see is that if someone sends mail from elsewhere using
an address handled by your server, and that mail causes a bounce, the
bounce will be lost even though it's "legit".

I suppose you could make the argument that the disadvantage of a few lost
bounces is outweighed by the advantage of stopping heaps of bogus bounces,
but that would be a policy decision I guess. Personally, tempting though a
plan like that is, I'm trying my best to avoid making my mail system
unreliable in that way.

I suppose a compromise might be to freeze any bounces that don't have the
"magic tag", for inspection by an administrator. That might cause a great
load though, depending on how many "not matching" bounces you get. The
bogus-virus-warnings filter should help to minimise the number caused by
viruses and similar, though obviously won't help in the case of joe-jobs,
or if a large number of your users send mail from outside your mail
system, using their aber.ac.uk address.



Tim