[Mailman-Developers] Re: and interesting loop

David Champion dgc@uchicago.edu
Mon, 30 Jul 2001 11:28:26 -0500

On 2001.07.30, in <Pine.GSO.4.33.0107301153060.23731-100000@alb-net.com>,
	"Mentor Cana" <mentor@alb-net.com> wrote:
> I've been running mailman since v 1.1rc1 (or something similar).
> Until today (v2.0.6), I had not seen the following scenario:
> A messages was sent out on one of the lists. The message was a bit large
> (about 300k). There are three e-mail addresses listed as admin for this
> list. Now, it just happened that one of the admin e-mail addresses was out
> of disk space so there was a bounce back to the -admin address, which was
> going back to the same admin.... and a loop got started....
> Is this the expected behavior in this scenario or perhaps is the result
> of a broken/misconfigured smtp server?

I'm not sure about Barry, but I expect it. :) This happens to me every
few weeks. 1100 admins, just too many for us to govern proactively.

It's not a traditional mail loop, since there's a new message with each
rebound. Mailman would need to detect bounces from an admin's address to
the -admin list address, I guess. Is that feasible using the existing
bounce-detection modules?

You could also configure the mail server to block mail to addresses
that receive mail too quickly -- like, say, half a gigabyte in under 10
minutes is a bad sign. Not sure that's possible in most MTAs without
some concurrent process watching the log and adjusting access rules on
the fly.

 -D.	dgc@uchicago.edu	NSIT	University of Chicago