RFC 6303 vs. BIND: NS ... has no address records (A or AAAA)

[Chris Thompson]

On Jan 10 2012, Tony Finch wrote:

>Irwin Tillman <irwin at princeton.edu> wrote:
>>
>> What's the recommended approach?
>
>My empty zone is:
>
>@ SOA	localhost. root.localhost. 1 1h 1000 1w 1h
>  NS	localhost.
>
>I also have a "localhost." zone (RFC 2606) which is:
>
>@ SOA	localhost. root.localhost. 1 1h 1000 1w 1h
>  NS	localhost.
>  A	127.0.0.1
>  AAAA	::1
>
>In the reverse direction I have 1.0.0.172.in-addr.arpa and
>1.0.....0.ip6.arpa zones with the predictable contents:
>
>@ SOA	localhost. root.localhost. 1 1h 1000 1w 1h
>  NS	localhost.
>  PTR	localhost.

Here is what we use on our central nameservers:

empty zone:

@ 86400 SOA  localhost. . 0 28800 7200 604800 86400
@     0 NS   localhost.

localhost:

@ 86400 SOA  localhost. . 0 28800 7200 604800 86400
@     0 NS   localhost.
@ 86400 A    127.0.0.1
@ 86400 AAAA ::1

1.0.0.127.in-addr.arpa & 1.0.0.[etc].0.0.ip6.arpa:

@ 86400 SOA  localhost. . 0 28800 7200 604800 86400
@     0 NS   localhost.
@ 86400 PTR  localhost.

These are deliberately designed to look as much like BIND's automatic
empty zones as is reasonable, and to increase the similarity we also
use <<empty-server "localhost";>> in options.

RFC 6303 specifies using "nobody.invalid" as the SOA.rname, but BIND
still uses "." for empty zones (apparently even in 9.9.0rc1). I imagine
we will change that if/when BINS does.

-- 
Chris Thompson
Email: cet1 at cam.ac.uk