[lacnog] El reino de lo insano (Fwd: Re: [v6ops] New I-D: SLAAC and DHCPv6 (Fwd: New Version Notification for draft-gont-v6ops-host-configuration-00.txt))

[Fernando Gont]

Estimados,

FYI: <https://goo.gl/fJTukL>. El thread empieza aca: <https://goo.gl/LUi6Km>

Otra "perlita": una cantidad de gente (incluyendo el ya "abonado"
Lorenzo Colitti de Google), argumentando en el v6ops wg de IETF que la
solución para solucionar problemas de configuración de DNS uno debe...
usar IPv4.


El desarrollo de estandares de IPv6 creo que ha llegado a uno de sus
puntos mas cannnabicos. :-)


Sería interesante que quienes tengan mas o menos idea del tema, se sumen
a la discusión.

P.S.: Si... entiendo el dilema para quien se encuentra desplegando o
intentando desplegar IPv6: Llorar vs tomar whisky vs dejar todo y
dedicarse a otra cosa.

P.S.: Para quienes nos gustan los protocolos, pensar que IPv6 es el
protocolo de Internet cuyo desarrollo ha tmado mas de 20 años para
llegar a este estado nos da... cierta verguenza ajena.



-------- Forwarded Message --------
Subject: 	Re: [v6ops] New I-D: SLAAC and DHCPv6 (Fwd: New Version
Notification for draft-gont-v6ops-host-configuration-00.txt)
Date: 	Tue, 28 Feb 2017 22:45:53 +0900
From: 	Lorenzo Colitti <lorenzo en google.com>
To: 	Fernando Gont <fgont en si6networks.com>
CC: 	Erik Kline <ek en google.com>, IPv6 Operations <v6ops en ietf.org>



On Tue, Feb 28, 2017 at 9:45 PM, Fernando Gont <fgont en si6networks.com
<mailto:fgont en si6networks.com>> wrote:

    >   * When IPv4 goes away, replace DHCPv4 with stateless DHCPv6.

    Problem benig that not all host implementations support DHCPv6.

    >   * Until then, use RDNSS.

    Problem being that not all host implementations support RDNSS.


Actually, no problem. All implementations support DHCPv4, so for as long
as you run DHCPv4 with RDNSS things work. Then, when you replace DHCPv4
with DHCPv6, all implementations support either DHCPv6 or RDNSS and
things still work.

    > Stateless DHCPv6 is worse than RDNSS in many ways.

    Are you suggesting that we provide guidance on this topic? -- If so,
    that'd be fine to me. -- So far, the current version of the doc just
    aims at having  both options on the table. But I'd be fine with
    providing guidance.


Personally I'm fine with providing guidance that all implementations
should support RDNSS, but a) I doubt we'll reach consensus, and b) I
don't think it will matter even if we do.

    > I don't see why we should require hosts to implement stateless DHCPv6.

    1) so that for networks in which they want to convey DNS info in DHCPv6,
    they can.


"Hosts should do something so that networks can do what they want"? Is
that going to be sufficient motivation? The incentives don't seem to
align very well.