Re: [dhcwg] Relaxing text in RFC3633 [recap]

[Ole Troan <[e-mail]>]

Ted,

>> with the following exception: the requesting router MUST
>> NOT assign any delegated prefixes or subnets from the delegated
>> prefix(es) to the link through which it received the DHCP message
>> from the delegating router.
> 
> When I read this text, the way it first appeared to me was that it was saying the RR should not do PD on the link on which it got the prefix, but it's not saying that, so I guess my main objection to changing this text goes away.
> 
> However, if we add the ability to knock out a /64 from a delegated prefix, the need to relax this restriction goes away--it makes much more sense for the DR to assign the RR an address out of the /64 than it does for the DR to *delegate* the /64 to the RR.

indeed.

> What's the scenario where the RR would get a *delegation* of a /64 and then assign it to the interface on which it received the delegation?   How would that even work?   I think this change is simply unnecessary.

this discussion started off by claims that the "MUST NOT assign any..." hindered using a part of a delegated prefix on the DR-RR link. changing the text will not change any behaviour. it is purely a clarification, one that makes it even clearer that authority of the address space is transferred as a part of the protocol. the new "owner" of that prefix should have no restrictions on its use. I think the current text has led to some misunderstandings, which is why I would like to put in the erratum.

you are right in that this will not result in any changes on either RR or DR behaviour. which is why it is an erratum not 3633bis.

cheers,
Ole