Stats collection script for BIND 9.5 (and greater?)

Alexander Gall gall at
Tue Dec 16 09:38:49 UTC 2008

On Mon, 15 Dec 2008 15:44:39 -0800, JINMEI Tatuya / $(B?@L at C#:H(B <Jinmei_Tatuya at> said:

> At Mon, 15 Dec 2008 17:18:21 +0100,
> Alexander Gall <gall at> wrote:

>> >
>> This looks useful, thanks.  However, ISC has chosen to change some
>> tags in 9.6.0rc1 (nsstats -> nsstat, zonestats -> zonestat, resstats ->
>> resstat).  Unfortunately, they didn't bump the version of their XML
>> schema (it's still reported as 1.0 in the <isc> tag), so it's hard to
>> do right.  I hope this is going to be fixed in the final 9.6.0
>> release.

> 9.6.0rc2 (and 9.5.1rc2) will soon be released (and I don't think we
> can make a change to these versions unless it's a fatal bug).

> So, if this is a crucial fix to be incorporated in the final release
> we don't have much time.  Just out of curiosity, how crucial do you
> think it is to bump the version?  My understanding is that bumping the
> version will help if we keep supporting the first version for a
> reasonably long period, while chasing newer versions.  Realistically,
> however, I suspect most users (especially those using advanced
> management tools like this script) will move to 9.5.1 (or 9.6.0)
> anyway, since 9.5.0 and its P1/P2 variants have other problems:
> vulnerability to the Kaminsky attack (in case of vanilla 9.5.0),
> performance issues due to file descriptor limitations, ACL
> crash/memory leak, cache memory management bug, etc.

> *If* we could assume no one realistically continues to stick to 9.5.0,
> wouldn't it be simpler just to drop the old format and migrate to the
> new one? (I understand the migration itself is a pain, and I apologize
> for the inconvenience.  As I said in a response to a different thread,
> we'll try to avoid happening in future releases).

Yes, if that assumption is correct, I don't see a big problem sticking
to version 1.0 for the new format in 9.6.0 and 9.5.1.  Still, if you
do manage to get this particular change in (I guess it would be
extremely localized and probably doesn't require any actual code
change at all) that would be preferrable, IMHO.


More information about the bind-users mailing list