raulbe wrote:
What I did was turn off auto_learn and that cleared up the error. So would
doing this would it affect how bayes works or?
Well, yes. All Bayes training will now have to be done manually.
What's usually a slightly better idea for a global Bayes database is to
set these Bayes options:
bayes_learn_to_journal 1
bayes_auto_expire 0
and make sure to set up a cron job to run "sa-learn --sync
--force-expire" periodically (I've been finding daily is a good idea; YMMV).
So all thats left is this error
spamd[29330]: prefork: child states: BBBBBIBI
This isn't an error, it's an informational message telling you what
spamd's children are doing.
If you really want to get rid of it, you'll either have to shift the
logging threshold for the "mail" facility in syslog until it goes away
(thus likely losing a great deal of valuable log data from both SA and
other sources) or fiddle the SA source so that those messages get sent
with a "debug" priority, and set syslog to not log debug-priority messages.
Unless you're *really* horribly short on disk (I can recall a few
systems I might have considered that on), it's not worth worrying about.
-kgd
