Bind growth

Tom Nichols tomn at team.citx.net
Wed Mar 21 15:13:11 UTC 2001


Roy Arends wrote:

> On Wed, 21 Mar 2001, Tom Nichols wrote:
>
> > We have three basic DNS servers (BSDI 4.2) running Bind 8.2, 8.2, and
> > 9.1
>

The primary is 8.2.2-P5, the first secondary is 9.1.0, the third is 8.2.2-P5

>
> 8.2 as in 8.2.3-REL ?
> 9.1 as in 9.1.1-rc5 ?
>
> > The first is a primary, the other two are secondaries.
> >
> > We experience growth on the secondaries.
> >
> > To explain, running TOP on each system gives the following results:
> > (yesterday, today)
> >
> > 1st                     2nd                      3td
> > 5104   4996             6116    6250             2480   2520
> > 5104   4996             6440    6532             2736   2780
> >
> > The first line is yesterday morning's data. The second, today. As you
> > can see, the primary is unchanged. That server hasn't been rebooted in
> > over 60 days. As a primary, we add to the DNS database and perform
> > kill -SIGHUP xxx on a daily basis.
>
> Why not use the cleaner "ndc reload" on the primary ?
>

I'm old fashioned. From my day's with SCO, we kill processes.



> > The other two will dump named.core after a few days of run time.
>
> Are both secondary's configured so they won't cache anything ? (recursion
> no, fetch-glue no). Are both secondaries only slave for your master, or
> other masters as well ?
>
>

Both secondaries have recursion set to no, I don't have the fetch-glue line
in any of the .conf files.



> Regards
>
>

Thanks for the re:


> Roy Arends
> Nominum


-- Binary/unsupported file stripped by Listar --
-- Type: text/x-vcard
-- File: tomn.vcf
-- Desc: Card for Tom Nichols




More information about the bind-users mailing list