Re: draft-gont-6man-slaac-renum-01

[Erik Kline <[e-mail]>]

Rolling up replies.

I think all these concerns are addressable.

Yes, this kind of loopback testing might not currently be doable on
some operating systems without more control levers.  But the kernel
could easily arrange to permit and inspect loopback'd packets before
discarding them because the source address matched a local address.

And to Bob's point: I think it's correct to say that the policy
applied to the result of a "does this work" test is in fact a policy,
and is something that probably ought to be configurable.  But that
doesn't the negate the value of learning that the prefix has become
unusable (at a minimum it would be an administratively log-worthy
event, so someone can come along later and diagnose).