Re: [v6ops] I-D Action: draft-ietf-v6ops-reducing-ra-energy-consumption-02.txt

[Lorenzo Colitti <[e-mail]>]

On Thu, Nov 5, 2015 at 10:47 AM, Erik Nordmark <[e-mail]> wrote:
>
> However, in 01 there was a section added (4.2 - now 5.2) with device side
> recommendations/requirements, with with I have serious reservations at
> several levels:
> 1. The text uses terms like "maintain IPv6 connectivity" and "disconnect
> from the IPv6 network" which are not defined. I would not know what impact
> this has on an implementation with those terms being defined. E.g., does
> "disconnect" mean to discard all IPv6 addresses and prefixes and release
> all DHCPv6 leases? Kill all connections? What?
>

"Maintain IPv6 connectivity" means "maintain the ability to correctly send
and receive IPv6 packets as configured by the network".

Example: suppose that while a node is asleep a network sends out a number
of RAs setting the lifetime of the current default router to 0 and adding
another default router, or changing the on-link prefix. The node's IPv6
connectivity is broken until such time that it receives an RA.

"Disconnect from the IPv6 network" means whatever process the device would
follow when disconnecting from the link.

Happy to clarify that in the draft if necessary.

2. Section 5.2 changes (or at least clarifies) the IPv6 standards which
> seems to not be in scope in the v6ops charter.

Thus I think section 5.2 should be removed and that discussion be had in
> the 6man WG.
>

Section 5.2 changes does not clarify the standards. It provides operational
guidance to implementers. It points out that if implementations do not
listen to RAs when asleep, then they will lose the ability to correctly
send packets off-link, and says that if implementations want avoid that,
then they need to listen to RAs.