RE: FW: WPD-6, WAA-8, and WAA-9 of draft-ietf-v6ops-ipv6-cpe-router-04

["Wes Beebee (wbeebee)" <wbeebee@cisco.com>]

> Please explain how you would get a connected route? Even though the SP
router may create a 
> NC entry for the address, I don't why it would also consider the
address to be onlink. 
> Unless you consider every address with a NC entry to be onlink. 

Considering every address with a NC entry to be onlink is, exactly, what
RFC 4861 does until
http://www.ietf.org/id/draft-ietf-6man-ipv6-subnet-model-10.txt came
along, which updates RFC 4861 (and is currently in the IESG queue).
However, I believe that, in this case, even though the SP router WILL
observe and possibly react to the NS(DAD), and may (if not updated by
subnet-model yet), create a connected route to that address, I believe
that in this particular case, the forwarding answer is the same whether
it considers the destination to be off-link or on-link - because the
destination of a forwarded packet is the same - in this particular,
quirky case (not in general).  And, the fact that the interface will now
respond to an NS with an NA (whereas it didn't before), I think will be
OK - because the behavior is a strict superset of what it had before.
So - I believe that in this particular case, where you assign an address
from the off-link pool to the interface that would otherwise get the
packet if considered off-link, will actually work. 

You got lucky this time, but in general, when you mess around with
on-link determination you usually end up with no connectivity.

- Wes