Non-Octet Boundary Delegation (RFC 4183)

[Peter Rathlev]

On Thu, 2011-09-22 at 16:55 +0400, Станислав wrote:
> It's almos OK if mask is 21. but what if it's, for example 17? In that
> case I need to create 128 zones.
> I can use network masks in BIND (RFC 4183) but if I use such zone as
> 24-21.133.10.in-addr.arpa. DHCP doesn't recognize that 10.133.27.68
> belongs to this zone.

Ah, now I see where you're getting at. I could imagine that ISC DHCPd
will not easily handle those, though I haven't tested it. The server
probably doesn't know how to map e.g. 172.16.0.10 to the in RFC4183
suggested name "10.0-25.0.16.172.in-addr.arpa" instead of the
traditional name "10.0.16.172.in-addr.arpa", even though it might have
the zone declaration to cover the latter. I'm not certain, but I don't
think DNS updates follow chains of indirection.

Beware that RFC 4183 is strictly informational, and should be considered
a suggestion at best. As far as I understand the RFC, it demands special
attention from resolvers to actually resolve an address under one of
these zones. The RFC does not discuss the problems associated with
dynamic updates, and from the "Abstract" it seems that that the authors
primary goal was topology description.

I'm personally not convinced that the suggestion in the RFC is a good
idea at all, but that is of course not relevant to your inquiry.

I think the conclusion is that ISC DHCPd does not support the scheme
suggested in RFC 4183, and it probably will not support it any time
soon.

-- 
Peter