First unlearn that one message. Then read the rest of this message and
use sa-learn properly for your mail storage format.
Then do it correctly. You're obviously running mbox format for your spam
store. So you need to include --mbox as an option on sa-learn.
{^_^}
----- Original Message -----
From: <[EMAIL PROTECTED]>
Hello all! I am mostly a lurker here on this list. I have been using
spamassassin for about 2 months now. I use it in a site-wide config on
an OES-Linux server (based on SLES9). I use hula as my email server with
spamd. I noticed that my emails are not marked with any "spam" headings,
so i looked in the log file and see where the spam are getting
auto-learned as ham. So, i turned off "auto-learn" and ran sa-learn
--spam /path to spam. 1 message was learned as spam when i have over 200
spam in the spam folder. So, i am thinking the bayes database is
corrupt? Is it ok to --clear the database and start over again without
auto-learn on? Thanks
Chris
