The subject tells it all, do we always have to accept MAIL FROM: <>?
I can see a problem if a RCPT acl fails while another server is
checking the sender address.
Suppose I see a misconfigured (incorrect reverse and all the rest)
server in my logs whose mail has been denied on those grounds, I
recognise the source and I email back to let them know of the
problem. If I send from postmaster all is well (I see now another
good reason to always accept postmaster), but if I have to send from
the recipient who was denied in the first place and the remote server
is checking senders I might end up in a catch22 situation, as their
sender callout will fail.
I admit this is an unlikely scenario, but it would be interesting to
be able to see incoming sender callouts, perhaps in a separate file
and controllable with some signal (a bit like SIGUSR1 for named).
It would also be useful to detect address scans with a null SMTP
transaction (that is QUIT or RSET, QUIT after RCPT) that are not
currently logged.
Thanks
Giuliano
--
H U M P H
|| |||
software
Java & C++ Server/Client/Human Interface applications on MacOS - MacOS X
http://www.humph.com/
