Re: [EXTERNAL] IPv6 certification - IPv6 Router Advertisement Lifetime 0 and Reachable time 10 seconds

[Bob Hinden <[e-mail]>]

Isaac,

> On Jan 24, 2021, at 9:40 PM, Isaac <[e-mail]> wrote:
> 
> Bert,
> 
> In a way, yes. Do you think this combination of RA lifetime 0 and reachable time 10 seconds need to be mandated for all vendors?

As many people have said, they are separate parameters.   Support for them needs to be implemented that way.

> 
> I have a question to the IETF group here, when RA lifetime is 0 (which means the the router is no longer to be used as gateway by hosts) and prefixes are provided

The key word is not being used as a "default router".  It can still be used as a router for a specific prefix.

> (prefixes have their own lifetime and flags and in this rare scenario there is going to be a second router acting as gateway), why do we need reachable time of 10 seconds for the 1st router?

It’s not that rare.

I think you have gotten the answer to your question(s).

Bob


> 
> Thanks,
> Isaac.
> 
> On Mon, Jan 25, 2021 at 10:45 AM Manfredi (US), Albert E <[e-mail]> wrote:
> From: ipv6 <[e-mail]> On Behalf Of Isaac
> 
> > Ole/Tim Winters/IETF team,
> >
> > Yes, we understand these knobs but we wanted to understand more on the scenario/topology. More importantly we wanted to understand the real world scenario when this combination of RA lifetime 0 and reachable time 10 seconds is used and the technical merit of it for which we did not get clear response (especially in the modern global IPv6 networks context). It's surprising that the certification bodies haven't clearly mandated only common/practical (although IETF has mentioned that these paramers need to be configurable but never said explicitly that all permutation/combination of values need to be supported. Vendors (definitely want) comply to RFCs but do not want allow impractical values) use cases but have listed even the corner scenario which may never be used. We understand that there are thousand vendors who have implemented this combination. But we fear that these are extra burden for vendors considering that vendors go ahead for certification without questioning the certification body itself believing that the certification body does its job of validating the modern technical relevance. Ideally, we expect the certification body (if not IETF) to re visit all the tests periodically to understand the relevancy as time passes and modify if required (which is the purpose of the certification body we believe). Sorry to have spilled certain discussions pertaining to certification body in this forum. But we do not have much option as we want technical answer from the IETF group. Let's not stop with the high statements in RFC. The reason we approcahed IETF is to go one level deep (especially in the context of modern day global networks) to undertand the relevance of RA lifetime 0 and reachable time 10 seconds whether it makes sense to support. These are our 2 cents contribution to the community (if there is someone to listen!)
> 
> Isaac, I'm trying to understand your point. At least one scenario in which the RA lifetime is set to 0, and reachable time is set to 10 seconds, was explained a couple of times. It can be used for a router to provide the IPv6 prefix, for example for SLAAC, but for that same router NOT to be used as the default router, for the subnet in question.
> 
> Is it that this scenario seems unrealistic and unnecessary, to you?
> 
> Thanks,
> Bert
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------