[Mimedefang] Mimedefang/Spamassassin/bayesian

[alan premselaar]

On 2/17/04 9:31 PM, "Paul Murphy" <pmurphy at ionixpharma.com> wrote:

...snip...  
> debug: bayes: 29638 tie-ing to DB file R/O /var/spool/spamassassin/bayes_toks
> debug: bayes: 29638 tie-ing to DB file R/O /var/spool/spamassassin/bayes_seen
> debug: bayes: found bayes db version 2
> debug: bayes: Not available for scanning, only 0 spam(s) in Bayes DB < 200
> debug: bayes: 29638 untie-ing
> debug: bayes: 29638 untie-ing db_toks
> debug: bayes: 29638 untie-ing db_seen
> 
> Since we get about 600 incoming messages per day of which around 50 are scored
> 10+, and 20 scored between 5 and 10, I'd expect to see this number be non-zero
> by now....
> 

the bayes learning process takes into account the score of the email without
network and whitelist/blacklist scores added.  So, that means that if the
email scored say a 5 or 6 with network tests, but the actual non-network
score is say a 2 or a 3 then it may not be learned automatically (keeping in
mind these figures aren't necessarily accurate, just being used as an
example) ... I recommend checking the docs for SpamAssassin for specifics
about how the bayes learning works.

> The relevant parts of my sa-mimedefang.cf are:
> 
> use_bayes 1
> auto_learn 1
> bayes_path /var/spool/spamassassin/bayes
> bayes_file_mode 0666
> 
> The bayes stuff looks like this:
> 
> localhost:/var/spool/spamassassin# ls -al
> total 24
> drwxr-xr-x    2 root     root         4096 Feb 17 12:29 .
> drwxr-xr-x   14 root     root         4096 Jan 19 17:30 ..
> -rw-rw-rw-    1 defang   root         4096 Jan 19 18:36 bayes_seen
> -rw-rw-rw-    1 defang   root        20480 Jan 19 18:36 bayes_toks
> 
> What's missing?

try running sa-learn --dump magic to show you relevant information regarding
the bayes database.

hope this helps

alan