On Mon, May 24, 2004 at 03:43:39PM +0200, Joerg Over Dexia wrote:
You can also do a "nslookup BlueWaveDirect.com" and find that the
MTA is correct. The MTA can probably be configured to accept
unknown sender domains, but I'd recommend against that. It's an
antispam feature.
Opinion is divided.
If you consider it an "anti-spam" feature and are using fetchmail then in
principle you could configure your local MTA to give a 5xx response to allow
fetchmail to bounce the message, otherwise it will stay on your POP3 server
forever. But this is a very bad idea as it risks bouncing mail from people
whose DNS is temporarily down. Much better just to accept the mail and then
perform content-filtering on it (i.e. get spamassassin or whatever to
validate the domain for you, as one of its many spam-assessment tests)
I don't believe this sort of protection belongs on a second tier of SMTP
relay anyway. In other words:
SMTP POP3 SMTP
sender --------> ISP mailbox ------> fetchmail --------> end user
(A) (B)
If you want to validate sender domains, it should be done at point (A), not
point (B).
It's bad enough rejecting the mail at point (A); if the mail was relayed via
a smarthost at the sending ISP, as the vast majority is, then it can't be
returned to sender anyway. So actually I believe this sort of validation
really belongs at the smarthost, which is the only machine which can
successfully reject a mail submission from a misconfigured sender.
There seems to be a silent acceptance of the tenet that "malformed mail =
spam", when actually there are plenty of legitimate users whose
configuration is less than perfect.
In any case, most spammers send out mail using forged but genuine sender
addresses, which will tend towards 100% as such domain validation is done
more widely.
Regards,
Brian.