What did I do wrong?

Chris Wilkes cwilkes-bf at ladro.com
Wed Feb 18 20:35:19 CET 2004


On Wed, Feb 18, 2004 at 02:00:34PM -0500, Eric Wood wrote:
> Greg McCann wrote:
> > To keep the database
> > from getting too big I set up a daily cron job to automatically
> > delete singletons or any key that had not been updated in the last 14
> > days:
> >
> > /usr/local/bin/bogoutil -c1 -a14 -m /home/bogofilter/wordlist.db
> 
> My wordlist is 124Meg.  I've never done any "maintenance" on it.   So I
> decided to run the above command, but it is taking too long and I see a lot
> of procmail processes getting backed up.  bogoutil must have the wordlist.db
> file locked down.
> 
> I'll schedule sendmail to stop and bogoutil to run in the middle of the
> night.  How long should this command normally approximately take on 1.4GHz
> athlon system?

You can make a copy of your wordlist at any time by doing a
  bogoutil -d ./wordlist.db | bogoutil -l ./new.db
provided that you're not constantly updating your wordlist (ie using the
-s flag).  Then just copy the new.db file to wordlist.db.

If you're paranoid shut down local delivery at that time or tell
procmail not to use the wordlist while you do the move.

Also keep in mind that if you're not constantly updating your wordlist
with every single email that comes in then the -a14 could harm you.  If
you're only making changes to your wordlist based on corrections then
the tokens from "earn spare money" emails that always get labeled as
spam will have old update dates and thus will be dropped off.  Then
suddenly you'll get some more spam, etc.

Chris




More information about the Bogofilter mailing list