To:
Bill Manning <bmanning@ISI.EDU>
cc:
cathym@arin.net, dufberg@nic-se.se (Mats Dufberg), dnsop@cafax.se
From:
Robert Elz <kre@munnari.OZ.AU>
Date:
Thu, 10 May 2001 02:09:30 +0700
In-Reply-To:
<200105091818.f49II1Z05093@zed.isi.edu>
Sender:
owner-dnsop@cafax.se
Subject:
Re: Should a nameserver know about itself?
Date: Wed, 9 May 2001 11:18:01 -0700 (PDT)
From: Bill Manning <bmanning@ISI.EDU>
Message-ID: <200105091818.f49II1Z05093@zed.isi.edu>
| In fact, the "CNAME HACK" can be done w/ NS and A RRs.
The CNAME HACK isn't relevant - that is just a method to get
a whole bunch of names which seem as if they have to go in one
zone file into different zones - it is only the data in the zone
that's affected, nothing related to NS's (which is good, as
NS's and CNAMES don't mix well at all).
And while I could delegate junk.cs.mu.oz.au to servers in some
in-addr.arpa zone, that would be unlikely to require glue (only
if the servers for the in-addr.arpa zone were inside junk.cs.mu.oz.au)
The likely case is nameservers for the in-addr.arpa inside the domain
being delegated. And on other lists recently (and other times) one
DNS implementor has made it known that he recommends doing exactly that,
so much so, that his software will do it for you by default if you don't
tell it where else to put the servers (or that is how I understood it).
Whether that happens for in-addr.arpa domains or not I am not sure, but
if you believe the reasoning for doing this (which is largely to force
glue to be provided) then it probably should.
kre