BIND 9.3.2 on FreeBSD 6.1 out of control

[patrick]

But when it reaches that limit, is there any reason why the named
process starts eating up all CPU time? The memory size, I can handle
(and control) -- it's the unexplained CPU usage that concerns me.

This server is a master for 143 domains, and will likely take on many
more. Is there a way I can see what's in the cache? I'd just like to
get an idea of how much memory each record takes up so I can do some
math for future planning on resource requirements.

Patrick



On 1/2/07, Mark Andrews <Mark_Andrews at isc.org> wrote:
>
> > I'm running BIND 9.3.2 on FreeBSD 6.1, and am noticing that it gets
> > out of control after running for a while.
> >
> >   PID    UID           THR PRI NICE   SIZE    RES STATE    TIME   WCPU COMMAN
> > D
> > 60480     53             1 132    0   195M   194M RUN     41.7H 75.54% named
> >
> > After restarting it, its CPU usage goes back down to what it should
> > be, as does its memory usage. I really don't want to babysit this
> > process, so I'm trying to find the cause of this. I have
> > "max-cache-size" set to "150M", as before I turned this on, this
> > process would just grow and grow until it hit FreeBSD's limit and
> > would stop responding all together not to mention eating up as much
> > CPU time as it could.
> >
> > I never had this problem at all with BIND 8, and am wondering if
> > there's something I'm doing wrong with BIND 9 to have this problem?
>
>         BIND 8 has a smaller memory foot print.
>
> > Has anyone else experienced this? On a related note, is there any way
> > to have BIND 9 keep its memory usage much lower automatically without
> > having to set an arbitrary/explicit max-cache-size? (ie. Have it clean
> > up its own cache for entries that haven't been used after a certain
> > period of time?)
>
>         It does do that based on ttl.  max-cache-size is an additional
>         constraint.  You can also adjust maximum retention times using
>         max-cache-ttl and max-ncache-ttl.
>
>         195M with a 150M max-cache-size sounds about right.  You have
>         150M of active cache and some overhead.
>
> > Any help would be greatly appreciated.
> >
> > Patrick
> >
> >
> --
> Mark Andrews, ISC
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE: +61 2 9871 4742                 INTERNET: Mark_Andrews at isc.org
>