Bayes issue

[Jay Ehrhart <yo...@ycoe.org>]

The Bayes score is not being used in the overall spam score.

My MailScanner/SpamAssassin has been working fine.  I wanted to wipe out the
Bayes files and have them recreate and learn again.  I did a rm bayes_*  and
it removed the files.  I restarted MailScanner and the files were recreated
and the time stamp is current and the size is slowly building.  I can see in
the log file "autolearn=spam" and "BAYES_99 1.89" and other Bayes_ scores in
the log.  I ran sa-learn -dump -D and spamassassin --lint -D and everything
looks fine.  I have had over 3,000 emails through since I did the rm on the
bayes files.

The problem is the Bayes score is not being used in the SpamAssassin
scoring.  Before the flush I would see the Bayes score first and other the
scores under it.  How do I get it to start using the Bayes score again?  I
have retarted MailScanner mulitple times.

Re: Bayes issue

[Matt Kettler <mk...@evi-inc.com>]

Jay Ehrhart wrote:

>The Bayes score is not being used in the overall spam score.
>
>My MailScanner/SpamAssassin has been working fine.  I wanted to wipe out the
>Bayes files and have them recreate and learn again.  I did a rm bayes_*  and
>it removed the files.  I restarted MailScanner and the files were recreated
>and the time stamp is current and the size is slowly building.  I can see in
>the log file "autolearn=spam" and "BAYES_99 1.89" and other Bayes_ scores in
>the log.  I ran sa-learn -dump -D and spamassassin --lint -D and everything
>looks fine.  I have had over 3,000 emails through since I did the rm on the
>bayes files.
>
>The problem is the Bayes score is not being used in the SpamAssassin
>scoring.
>
Please qualify *exactly* what you mean by "the bayes score is not being
used". What makes you think it's not being used? Be specific. Include
examples if possible.

In general, if the rule is in the list of hits, its score will be used
by spamassassin.

Nothing short of catastrophic failure of basic mathematics or a VERY
severe bug in your perl interpreter or spamassassin's scoring engine can
cause what you seem to be describing, thus I'm fairly sure I've got the
wrong understanding of the problem. 

(it sounds like you've got a X-MailScanner-SpamCheck header where
BAYES_99 shows up, but the total score comes up 1.89 points short.)