Re: [v6ops] Making RDNSS a MUST?

Le 04/04/2017 à 15:12, Tim Chown a écrit :
> Hi,
>
>> On 4 Apr 2017, at 13:55, Lorenzo Colitti <lorenzo@google.com
>> <mailto:lorenzo@google.com>> wrote:
>>
>> On Tue, Apr 4, 2017 at 2:21 AM, Ted Lemon <mellon@fugue.com
>> <mailto:mellon@fugue.com>> wrote:
>>
>>     This is frustrating.   It seems clear to me that requiring support
>>     for RDNSS would improve the situation.
>>
>>
>> +1
>>
>> As a reminder: the original question in this thread was very simple:
>> do we have rough consensus to say that hosts MUST implement RDNSS?
>>
>> Given the strong opposition from major host OS developers, and others
>> on this thread such as Barbara and James, I really don't see us
>> getting to consensus that hosts MUST implement DHCPv6. Therefore, in
>> practice there are only two feasible options:
>>
>>  1. Don't say anything, and have no mandatory mechanism that operators
>>     can rely on. Some network/device combinations won't work.
>>  2. Say that RDNSS is a MUST, and provide a mandatory mechanism that
>>     network operators can rely on, even if it's not the one they like
>>     the most.
>>
>> The question is not whether we prefer DHCPv6 or RDNSS. The question is
>> whether we want to do #1 or #2.
>
> I’m seeing strong support for RFC6801 RDNSS as a MUST. Personally, I
> would like to see this, for clients and routers, in RFC6434-bis
> and draft-ali-ipv6rtr-reqs respectively.
>
> And that would give one MTI method to (on paper at least!) break the
> current deadlock.
>
> But that doesn’t rule out a statement on *stateless* DHCPv6 support,
> does it? RFC8601 explicitly discusses what to do in the presence of both
> mechanisms, and iirc says to place a resolver address learnt by DHCP
> ahead of any learnt by the RDNSS option, effectively preferring
> stateless DHCPv6 if present. While I personally believe the RA-based
> option is technically better, I also see enterprise network admins
> running IPv6 today who are wondering how they’ll configure (say) NTP
> server settings, or other settings only available by stateless DHCPv6
> and not via RAs. What’s the answer for them, as we move to IPv6-only
> operation?

I agree with the question.

I would also like to suggest make clearer that any additional 
RA-specific recommendation (like DNS-in-RA) does not impede DHCPv6-PD, 
in addition to stateless DHCPv6.

Alex
>
> Tim
>
>
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>