Hello All. We have a smarthost running spamassassin in the permiter, this host relays email for our internal domains to an M$ Exchange server in our private LAN. Spamassassin is doing a great job in filtering most spams.
I get my MS Exchange users to drop any 'Undetected SPAM' to a folder which I daily feed to sa-learn on the smarthost. I can notice a big difference this makes to the spamassassin filter. After a time, I notice an email pattern in this folder that doesn't seem to improve from time to time. I did some investigations and concluded that most emails with ms-tnef have low rate spam detection. My guess is that because the emails that I feed to sa-learn is NO LONGER the original email due to ms-tnef. Below is a sample email from my Undetected Spam folder, as you can see, the mail body gets modified, hence sa-learn do not get to learn the original message and this kind of spam pattern just keeps coming in. I read a bit about ms-tnef on this website http://agamemnon.ucs.ed.ac.uk/faq/mstnef.html - I can't get my Exchange server to receive text only message (instead of rich text/html) due to internal politic issue, so I'm trying to find a way to get around this outside the Exchange box. Has anyone experience this problem before? Thanks in advance. -------------------------------------------------------------------------- [-- Attachment #2: winmail.dat --] [-- Type: application/ms-tnef, Encoding: base64, Size: 3.0K --] Content-Type: application/ms-tnef; name="winmail.dat" Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="winmail.dat" [-- application/ms-tnef is unsupported (use 'v' to view this part) --] --------------------------------------------------------------------------