Re: How to dump Bayes

[Marc Dufresne <Ma...@parks.on.ca>]

I have posted my problem before Christmas concerning bayes
database(Subject: SA-LEARN Keeps crashing at DBM.pm Line 624)

It seems that the bayes database is corrupt at a certain location. How
do I dump the database to see where the corruption is, remove it, save
it, and perform an SA-LEARN --SYNC?

Interim solution
My database is quite large to just disregard. I've downloaded a new
starter DB, and renamed corrupted DB  files from;

bayes.mutex    bayes.mutex.org
bayes_seen      bayes_seen.org
bayes.toks        bayes.toks.org

Installed the new starter db, and I began feeding that new database
with all new SPAM I'm receiving. Sa-Learn doesn't crash with(DBM.pm Line
624) anymore with the new database.

Once that was completed, I renamed the new database files with a .new
extension. Then I renamed the .org files back to their original form to
continue to use my large database, until I find out how I can fix the
corruption.

Anyone have any ideas????




Marc Dufresne, Corporate IT Officer
St. Lawrence Parks Commission
13740 County Road 2
Morrisburg, ON  K0C 1X0

E-mail: Marc.Dufresne@parks.on.ca
Voice: 613-543-3704  Ext#2455
Fax: 613-543-2847
Corporate website: www.parks.on.ca

>>> Matt Kettler <mk...@evi-inc.com> 1/17/2006 11:57:32 AM >>>
Clay Davis wrote:
> How do I dump my Bayes db so I can see what the tokes are scored at?


sa-learn --dump

However, if you use SA 3.0.0 or higher you won't be able to actually
read the
tokens. They are stored in SHA hash form, and cannot be reversed back
into the
original text.

The hashing provides speed benefits (fixed size records) and privacy
benefits
(nobody can snoop your bayes DB and try to figure out what you've been
talking
about).


If you have a specific message and you want to see what it's tokens
score at,
you can run the message through spamassassin -D. The debug output will
show you
the tokens in plain text form (since it has the message) and what score
they got.