No subject

time is being spent in dns_rbt_deletetreeflat, following pointers in a
tight loop.

I can probably live with things the way they are now, but if I find more
time to look at this I will probably attempt to build a profiling image to
confirm where the time is being spent.

Thanks for your help!

Regards,
-Chris


On Sat, 2 Oct 2004, JINMEI Tatuya / [ISO-2022-JP] ?@L at C#:H wrote:

> >>>>> On Sat, 02 Oct 2004 07:20:07 +0900,
> >>>>> JINMEI Tatuya <jinmei at isl.rdc.toshiba.co.jp> said:
>
> >> Yes - that is right.  The daemon shuts down in ~6:36 now.
>
> > Hmm, okay.  If you have time for an experimental, more-aggressive
> > solution, could you try the attached patch to see how it works (either
> > with or without ISC_MEM_USE_INTERNAL_MALLOC)?  It will completely skip
> > freeing actual memory at the shutdown procedure.
>
> Oops, sorry, the previous patch was not really correct (it had missing
> initialization).  Please use this one (attached below) instead.

------------------------------

From: phn at icke-reklam.ipsec.nu
Subject: Re: Restriction for large zones?
Date: Thu, 7 Oct 2004 04:41:24 +0000 (UTC)

Tom Schmitt <TomSchmitt at gmx.de> wrote:

> Hi,

> one of my dns-zones is growing large and has now nearly 100 thousands
> records. My question is: Is there a limit how large a zone could be?
> Or is there a limit of any kind at which the performance is no longer
> acceptable?

Small machiones ( 32-bits ) will have trouble when zone size 
approaches 2GB. Besides that it's your memory size that is 
the practical limit ( you don't want named to page ), so 
make shure you have physical memory enought for all zone data.

> Thanks,
> Tom.

> -- 
> +++ GMX DSL Premiumtarife 3 Monate gratis* + WLAN-Router 0,- EUR* +++
> Clevere DSL-Nutzer wechseln jetzt zu GMX: http://www.gmx.net/de/go/dsl



-- 
Peter Hekanson         
        IPSec  Sverige      ( At Gothenburg Riverside )
           Sorry about my e-mail address, but i'm trying to keep spam out,
	   remove "icke-reklam" if you feel for mailing me. Thanx.

------------------------------

Date: Thu, 07 Oct 2004 15:44:24 +0900
From: JINMEI Tatuya / =?ISO-2022-JP?B?GyRCP0BMQEMjOkgbKEI=?= <jinmei at isl.rdc.toshiba.co.jp>
Subject: Re: daemon shutdown time with large zones (bind 9.2.4)

>>>>> On Wed, 6 Oct 2004 14:50:18 -0700 (PDT), 
>>>>> Chris Timmons <cwt at networks.cwu.edu> said:

> It would appear that FreeBSD's free(3) isn't the main issue here; I built
> an image with the patch, and the performance is about the same as before.

Hmm.  Just checking,

- did you try that with or without ISC_MEM_USE_INTERNAL_MALLOC (or
  both)?  After re-checking the patch, I noticed it was not effective
  when ISC_MEM_USE_INTERNAL_MALLOC is enabled.  (sorry, this was my
  bad again).

- did you see the long period to shutdown the server even just after
  starting it?  If the bottleneck is not really in the memory
  management but in that code traversing the DB structure, the problem
  should occur even if the server does no actual work (i.e., only just
  loading and destroying the DB).

					JINMEI, Tatuya
					Communication Platform Lab.
					Corporate R&D Center, Toshiba Corp.
					jinmei at isl.rdc.toshiba.co.jp

------------------------------

End of bind-users Digest V6 #264
********************************