Hello - Just recently noticed that all of the mail coming through my system has the "autolearn=spam" tag in the SA header. I run SA through MailScanner and had been using 2.64 previous to 3.0. I recently noticed an increase in spam that has been getting through, and after some investigation saw that most messages, regardless of their SA score, were getting autolearned=spam.
Now, the kind of strange part is that I added bayes_auto_learn 0 and then even use_bayes 0 to my config. The autolearn message is still appearing though... seems like something about my config didn't jive with the upgrade. Any thoughts as to where to look? I'm afraid to turn the bayes stuff back on if it's training improperly... Here's an interesting line from the logs that demonstrates what I'm talking about: Sep 28 16:01:07 vps MailScanner[12314]: Message 1CCO9p-000831-Kx from 209.237.227.199 ([EMAIL PROTECTED]) to dlugosz.net is not spam, SpamAssassin (score=0.05, required 6.5, autolearn=spam, FORGED_RCVD_HELO 0.05) Thanks for any suggestions. -Ryan -- Ryan Dlugosz [EMAIL PROTECTED] http://dlugosz.net
