On Monday 30 October 2006 20:44, Wojciech Potrzebowski took the opportunity to say: > Thank you for your time in handling with this case! > I have atached two e-mails with headers from both servers. I can only > configure SA on my local server: iwonka.med.virginia.edu. I don't have > access to the other mail server.
As you can see, on your local server the spam hits BAYES_99:
> X-Spam-Status: Yes, score=5.5 required=4.0 tests=BAYES_99,NO_RECEIVED,
> NO_RELAYS,TO_WM_FROM_COM autolearn=no version=3.0.6
But on the other server bayesian-style (it's not pure bayesian, but modified
to be better) classifying isn't used at all, or isn't trained (they're using
Amavisd-new as the interface to SA, which (in a way) explains the slightly
different header format. In any case, you won't get the same results unless
both servers share the same bayes database.
--
Magnus Holmgren [EMAIL PROTECTED]
(No Cc of list mail needed, thanks)
pgpsV8RYWvryj.pgp
Description: PGP signature
