Kai Schaetzl wrote:
Chris Barnes wrote on Mon, 09 Mar 2009 12:06:10 -0500:
I have purged my bayes db and issued sa-learn to rebuild it.
How?
sa-learn --clear
But the
Bayes_00 score persists.
Are you learning those very messages as spam? I find that just learning a
message as spam *once* gives it a BAYES_99 on the next scan.
Yes, daily.
autolearn=ham
as long as your Bayes is "garbled" you shouldn't autolearn at all.
I have since turned autolearn off in /etc/spamassassin/local.cf
Q1: how did these addresses (which are all pretty obviously spam and
none of which are in our own domain) get into the AWL to begin with?
They came in and were delivered. You don't know what AWL thinks about
them. Why do you think it should only take mail from your domain?
At one point I had "whitelist_from *...@physics.tamu.edu" in my local.cf
file, I have long since removed that as well.
Q2: Is there a system-wide AWL and/or Bayes db that I need to
purge/clean up?
*you* should know if there is one.
Was this comment supposed to be a helpful comment? :-/
John Hardin wrote:
> Chris is probably being confused (as many have been) by the
> "whitelist" part of "auto-whitelist".
>
> Chris, AWL is a score averager. The name is misleading. It has nothing
> to do with trying to automatically score mail associated with your
> domain as ham.
Ok (this was helpful).
