Maintenance Best Practices

[David Relson]

Dave,

Something's off in your setup.  BerkeleyDB can happily handle databases 
greater than 50Mb.  Unfortunately, I can't tell you exactly what you'll 
need to change.  I suggest you check out the SleepyCat site.

I seem to recall an earlier report about problems at 51MB.  Unfortunately I 
don't recall the details.  A search through the bogofilter archives should 
reveal it.  Please let us know when you find it.

As you add tokens to the wordlists, they will grow.  Adding a token to a 
full database "page" forces the db to split the page and create 2 
pages.  That's probably the main cause for the fast growth you're reporting.

Simply running "bogoutil -d old | bogoutil -l new" will give a minimal 
database.  My mail server has 18,823 spam and 39,561 ham in wordlists of 
11M and 17M respectively.  I have yet to delete anything.

As I said above, something's odd about your setup.

David