Re: [mif] Comments on draft-mouton-mif-dhcpv6-drlo-00

[Alexandru Petrescu <[e-mail]>]

Le 13/09/2011 18:58, Ted Lemon a écrit :
> On Sep 13, 2011, at 12:18 PM, Alexandru Petrescu wrote:
>> It is good that lifetime of ND and of DHCP are coherent when used
>> on default route.
>
> Could you restate your point more explicitly? I find that I could
> interpret your response to mean either that you think that the major
>  extension to DHCP is required, or that it is not.

Allow me explain myself.

I am not sure what is meant by a major extension to DHCP.  Generally
speaking I oppose major rehauling of such vast software as DHCP
benefitting such huge installed base.

But I do believe a different option type is needed to communicate the
default route (different than the option type used in general for route
option for DHCP).

In the particular case of default route being communicated by DHCP (not
generally any routes communicated by DHCP) I believe that DHCP should
use a lifetime for default route in a way that is coherent with how ND
communicates that default route.

The lifetime fields that DHCP typically communicates are of different
length than the lifetime field used by ND about default router (32bit vs
16bit if I am not mistaken).

If that implies a major rehaul to DHCP implementation of timing in
general then I would oppose suggesting such a change.

If that implies that DHCP implementation does not change its way of
managing its 32bit timing for usual DHCP matters, but just stores 16bit
Default Route lifetime values in a conf file, and communicates these
fields to Client in a DHCP Message, then I would agree with it.

If that implies that ND should be modified such at the expiration of the
Default Router lifetime it fires a RS _and_ (optionally) a DHCP Request
ORO 64 asking anew the Default Router, then I would not oppose it either.

Does this explanation add to clarity of my initially short phrase?

Alex