Re: [dhcwg] draft-ietf-dhc-dhcpv6-opt-nisconfig-05.txt

    Date:        Thu, 05 Feb 2004 15:59:30 -0800
    From:        Harald Tveit Alvestrand <harald@alvestrand.no>
    Message-ID:  <2108571886.1075996769@halvestr-w2k1>

  | as an occasional network configurer, I think it seems like a pain in the
  | backside to have to configure two separate configuration tools that speak
  | two separate configuration protocols in order to get two pieces of config
  | info (the ipv4 address and the ipv6 address) to a single application on a
  | network client.

Yes, but do remember that all this is intended to just be a transitional
artifact - a few years from now (well, perhaps a lot of years from now)
IPv4 should be gone, it would be absurd in the extreme for DHCP (the DHCPv6
name variant probably long forgotten) to be handing back long dead
address formats - or even to be capable of it.

The suggested solution to the issue raised is really not just the easy
one, it is also the only one that makes much sense.

If I were a node, making a v6 DHCP query, and the answer I got back told me
that to use NIS I should use this v4 address, I'd be astounded.   How do
I do that?   I have no v4 address, I can't talk v4 at all, and yet DHCP is
telling me to use a v4 address to communicate with NIS?   Absurd.

If there is only v4 NIS, then as far as a v6 node is concerned, there is
no NIS.   And not supplying any NIS information in the DHCP offer is  the
right thing to do, not just to fill in some random address from some random
protocol and hope that the client happens to know how to deal with it.

If there is both v6 and v4 information available, and the node is dual
stack, then it is being configured in both stacks - each configuration
will configure its own access to the servers (v6 v6 access, and v4 v4 access).
That's exactly as it should be.  That way everyone gets (as mush as
is available) exactly what they want.   How the dual stack node decides
which to use is up to it (in 99% of cases for this particular information,
it will make no difference at all).

Having to deal with both v4 and v6 at the minute is a nuisance - but the
solution to this isn't to make every other protocol aware that both exist in
parallel, and add lots of mechanism to every place else to allow use
preferences to be stated, it is to make v4 go away, completely, as soon
as possible, so we're back with only needing to deal with one set of
information again.

kre

ps: somewhere, in the DHCPv6 docs, if it isn't already stated, it should
be, that DHCPv6 returns only v6 (and protocol neutral) information, no v4
noise at all, anywhere, ever, in any format.

_______________________________________________
dhcwg mailing list
dhcwg@ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg