How do I diagnose problem?

[David Relson]

On Mon, 31 Jan 2005 21:36:36 +0000
rkimber at ntlworld.com wrote:

> I just noticed that bogofilter was taking up 99.9% of one of my CPUs for
> what seemed like a rather long time.  It might of course only have been
> a minute or so, but I've never seen this before. No doubt I did the
> wrong thing, but I killed the bogofilter process.
> 
> The questions are how do I find out what went wrong and have I done any
> damage? Version 0.93.4, DB 4.2.52, Suse 9.1 opteron.

Hi Richard,

The damage question is the easier one, so I'll answer it first with "It
depends".  The database is corruptible during writing with
non-transactional bogofilter.  Reading is totally safe and transactional
bogofilter works with the database so that recovery happens in the event
of bogofilter (or system) problems.  You can check the database's state
with "bogoutil --db-verify=/your/bogo/path"

I don't have an answer to "what went wrong?"  Too much is not known :-<
If you can find the problem message and reproduce the problem, just zip
the message, send it to me, and I'll be glad to take a look at it. 

Are you running bogofilter and BerkeleyDB in 32-bit or 64-bit mode?  Not
being experienced with bogofilter and 64-bit opterons, there's always a
bit of a question in my mind.  However the likelihood of that being the
problem is slight. 

FWIW, there have been several messages over the last 2 1/2 yrs that
cause bogofilter long runs.  Typically they involve malformatted mime
sections.   At present all such problems have been fixed.

Sorry I can't give a definitive answer.

David