Interesting subdomain question.

[Barry Margolin]

In article <199911240007.LAA27920 at bsdi.dv.isc.org>,
 <Mark_Andrews at iengines.com> wrote:
>	Actually this is a bug but one that is impossible to correct
>	in BIND 4/8 due to internal data structures.
>
>	BIND 9 will correct this in that the zone transfer will
>	contain the contents of z.com with nothing mixed in from the
>	child zones.
>
>	As to why it works.  The NS records from the parent zone are
>	thrown away when the server serves both the parent and child
>	zones.  Because of this we cannot see the difference between
>	a parent zone that had NS records at bottom of zone and one
>	that didn't.  Outgoing zone transfers just use the NS RRset
>	from the child zone.

Why is this considered to be a bug?  Isn't the child domain zone file
considered "more authoritative" than the parent zone file?  Aren't "stub"
zones a special case of this, where the parent server does a zone transfer
and just incorporates the NS records into the parent zone?

About the only reason I can think of to do this is that there are many DNS
administrators who don't realize they need to add explicit NS records into
their zones.  They use MS DNS, which automatically creates an NS record for
the master server itself, and don't bother adding more.  Our servers are
authoritative for the reverse zone for our netblock, and slaves for the
customer's reverse zone that's delegated to them.  We list all the servers
in the delegation records, but I frequently have to remind customers that
they must also add them to their zones or they won't be effective.

-- 
Barry Margolin, barmar at bbnplanet.com
GTE Internetworking, Powered by BBN, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.