On Fri, 2004-03-12 at 16:52 -0500, Hector Santos wrote:
All CBV systems need to stop using non-NULL return paths, such as a FQDN
postmaster@ return path address or another other FQDN in a CBV session .
This can cause a potential loop if connecting to another CBV system who also
has a non-NULL return path.
There exists such a possibility, yes. I think CBV with non-empty
reverse-path is very dangerous and stupid.
Some people refuse to do CBV properly because that prevents them from
receiving mail from broken domains which reject all MAIL FROM:<>.
Personally, I think these people are insane, but they exist.
I've seen two CBV systems that uses postmaster@ so far. No problem when the
usage was small.
If I encounter a small site doing it, I'd be inclined to report it to
the abuse contact at their network provider as a denial of service
attack in waiting; just as I do autoresponders with non-empty
reverse-path and _all_ autoresponders to viruses.
In fact, the only one I've come across is pobox.com. They _really_ ought
to know better. If they really do feel they have to accept mail from
domains in dsn.rbl-ignorant.org, then they could at least _start_ with
MAIL FROM:<> and fall back to using postmaster@ if the former is
rejected at the MAIL FROM: stage.
Our CBV system watches for postmaster to stop potential loops. Are you
saying I should look at some special "SRS0" ?? address to avoid a CBV?
No, I was suggesting that those implementing SRS could, instead of
rejecting _all_ mail to SRS0 adresses from a non-empty reverse-path,
also work around the brain-damaged CBV discussed above by accepting mail
from postmaster(_at_)* to those addresses too.
More to the point, I was suggesting that this could have been the
problem which Jeremy was experiencing. I was guessing, since he didn't
show any actual SMTP transactions or much else to go on.
--
dwmw2