saving spam for Bayes repopulation

[Dene Ulmschneider]

Hey all-

I am curious about how anyone on this list is saving spam for future
repopulation of Bayes if a database ever gets corrupted.

I am using RHL 7.3, SA, MS, and sendmail. I have created the "spam" and
"notspam" boxes as described by Julian and am "redirecting" all appropriate
messages tot hose boxes. I am wondering if those boxes are retaining all of
the emails that are sent there or if SA somehow deletes the messages after
they have been scanned by sa-learn and have served their immediate purpose.

If they are kept in their respective boxes- then is it safe to assume that
if Bayes ever needs to be rebuilt that just running the sa-learn command
again will properly scan all of those saved emails at once? If they are not
kept - what is the best way to save them for future use if they are ever
needed.

ps-
I have noticed that I have a file that is over 60 megs located at
/var/mail/spool/spam.cumulative as well as a
/var/mail/spool/notspam.cumulative that is not quite as large. Are these
the actual files all of the redirected messages are being saved in?

thanks

Dene Ulmschneider