Nigel Metheringham <Nigel.Metheringham@???> writes
> If both A & B are local then I think that the loop resolution and
> collapsing of repeated addresses will cause this to be rather less
> effective [a quick test shows that this is the case and the victim
> address gets only one copy]
There's an amusing variant of this that sometimes catches out users
here. We have a cam.ac.uk domain consisting only of redirections of
the form user@??? => user@???; users can control
their own redirection. The MX records for cam.ac.uk point at our
central mail switches; however internally we use an iplookup router
that short-circuits the forwarding through these hosts.
Now a user creates a "loop" by having ~spqr/.forward on cus.cam.ac.uk
(say) point at spqr@???, and having that redirected back to
spqr@???. Normally, the iplookup works, Exim on cus.cam
determines the address is local, re-processes it (skipping the
forwardfile director this time), and delivers it to the user's inbox.
Much later, the fast lookup service used by the iplookup router is
down for a while, and messages to the user bounce to and fro between
cus.cam and the central switches 30 times, and are then failed.
"But I didn't change anything", says the user...
Chris Thompson University of Cambridge Computing Service,
Email: cet1@??? New Museums Site, Cambridge CB2 3QG,
Phone: +44 1223 334715 United Kingdom.
