db cachesize

[David Relson]

At 09:19 AM 2/4/03, Greg Louis wrote:

>On 20030204 (Tue) at 0902:13 -0500, David Relson wrote:
> > Greg,
> >
> > Does this look better?  I've set the default size to 0 and the default
> > action to "do nothing if size 0" and added config file option
> > "db_cachesize".
> >
>Looks good to me.  Does bogoutil read the config file too then?  It
>will need to.

Good point.  bogolexer also needs to do that to get parsing options.

>Some sort of informed commentary in the cf.sample file will be needed
>to explain to the bewildered newbie why one might want to use this
>feature.  Matthias is probably better able than I to describe the
>issue, but pro tem one could say something like:
>
>The default cache used by Berkeley DB is 256K in size.  Under most
>circumstances, increasing the cache size will result in faster database
>performance.  This may be particularly noticeable if the bogofilter
>directory is located on an ext3 partition and the database is being
>updated; in that case, a buffer large enough to hold the entire db will
>greatly improve throughput.  If you have enough RAM, you might want to
>try uncommenting the following line, which will set a cache size of 25
>megabytes:
># db_cachesize = 25

As I understand it, the caching problem doesn't become evident until the 
database reaches a certain size.  I'm thinking most newbies won't encounter 
the problem until they've been using bogofilter for a while.  Of course, 
there's always the guy with a gazillion messages that he feeds in right 
away :-(

bogofilter.cf.example isn't presently used for long explanations and I 
don't think it should be.  I'm thinking that bogofilter-faq.html is the 
place for an explanation like this and that bogofilter.cf.example can say 
"see bogofilter-faq.html for more info."