Thanks to Mark Martinec on the amavisd-new list I've managed to narrow this down to my comcast address being assigned about 18 points by the AWL. Now I just have to figure out why. I read the (not very expansive) POD docs on the AWL, and it's clear that I don't really understand how it works. Why doesn't it like my address? Also, why does it like me so much less under 3.0.1 than it did under 2.64?
> -----Original Message----- > From: Aaron Grewell [mailto:[EMAIL PROTECTED] > Sent: Thursday, December 02, 2004 1:55 PM > To: users@spamassassin.apache.org > Subject: Upgrade to 3.0.1 results in false positives > > I've been using SA 2.6x to block spam at our site for some > time. With the release of 3.0.1 I decided to upgrade. > Unfortunately, once the upgrade was complete I found that my > test e-mails were marked as spam. I also received several > false positives from end-users as well. I could understand > some of the FP's, but when a simple test mail (subject "test > test" body "test test") is marked I'm in trouble. The test I > sent was from my Comcast account, and ended up with the > following header: > > X-Spam-Status: Yes, hits=17.8 tagged_above=2.0 required=3.0 > tests=AWL, BAYES_00, DNS_FROM_RFC_POST, DNS_FROM_RFC_WHOIS, > NO_REAL_NAME, RCVD_BY_IP, RCVD_DOUBLE_IP_LOOSE, RCVD_NUMERIC_HELO > > I added up the scores for these rules, and they didn't seem > to add up to 17.anything. I'm using a Postfix MTA setup with > Amavisd-new to call SA, and the only thing that has changed > is SA so I don't think this is due to Amavisd-new. Why would > a simple test mail from my Comcast account generate this? >