Re: [Nolug] Mimedefang Spamassassin Bayesian learning

On Monday 15 December 2003 03:10 pm, J. Kent Busbee, Jr. wrote:
> I am using Mimedefang/Spamassassin on a FreeBSD box with some success.
>
> However, some nasty little spams slip through. So, I have been trying
> to use the sa-learn to get the Bayesian checker to correct the problem
> with no luck. I assumed that Mimedefang is calling up Spamassassin
> using user defang, which I created during the install. Mimedefang is
> running under user defang. After many sa-learns under the user defang,
> similar spams seem to keep getting through.
>
> Is it right to assume that Mimedefang/SA use defang user to run SA? How
> can I Know?
>
> Is Bayesian turned off for some reason? How can I know?
>
> There is one other setting I am unsure of, somewhat unrelated. In
> /etc/mail/spamassassin/sa-mimedefang.cf there is the following lines:
>
> # By default, SpamAssassin will run RBL checks. If your ISP already
> # does this, set this to 1.
>
> skip_rbl_checks 1
>
> Should I comment this line so that SA DOES RBL checks which I think is
> something to do with Reverse Domain Lookup? Uh, what is it? Should I
> do it? and what is the downsides/upsides?
>
Do you have something like the following in your sa-mimedefang.cf file?

# Enable the Bayes system
use_bayes 1

# Enable Bayes auto-learning
auto_learn 1
bayes_path /var/spool/spamassassin/bayes
bayes_file_mode 0640
auto_learn_threshold_spam 5.0
auto_learn_threshold_nonspam 0.0

Plus, you need to manually learn 200 hams and 200 spams to "prime" it
before auto-learning will actually begin, even if enabled in the config file.

Andy

___________________
Nolug mailing list
nolug@nolug.org
Received on 12/15/03