>>>>> On Thu, 26 May 2005 10:17:06 -0400,
>>>>> "Soliman, Hesham" <[EMAIL PROTECTED]> said:
>> And the 03 version reads:
>>
>> 1) If no Router Advertisement is received on any interfaces, a
>> multihomed host will have no way of knowing which
>> interface to
>> send packets out on, even for on-link destinations.
>> One possible approach for a multihomed node would be
>> to attempt
>> to perform address resolution on all interfaces, a step
>> involving significant complexity.
>>
>> I'm not sure if this really addresses the issue. Actually, isn't the
>> "possible approach" (i.e., performing address resolution anyway, when
>> no RA is received) just a multi-homed version of the "on-link
>> assumption", which was removed from 2461bis? And, if so, is
>> there any
>> special reason for allowing the "removed" feature for the multihomed
>> case?
> => I don't think it's the same as the on-link assumption. It's
> suggested as one possible approach and is not mandated (on top of
> being in the appendix), but also, it's fine to attempt address
> resolution because those addresses might actually be
> onlink. I.e. the node does not automatically assume that the
> addresses are onlink, which was the case before. We need to somehow
> address both cases: a) there is no router and some of the addresses
> are in fact onlink, and b) there is no RA because there is no native
> v6 connectivity and everything should go through a tunnel.
Whether or not the text indicates the "on-link assumption", I still
don't think the current 03 text is appropriate in this context. In
fact, it's possible for a single-homed host to not receive any RA on
its only interface, and "it would be fine to attempt address
resolution because those addresses might actually be onlink". So, why
should we then mention that in the multi-homed context?
Additionally, though I'm not sure if this is related to this exact
context, I suspect "attempting address resolution anyway" does not
really help your case (b) above, since in this case the other end of
the tunnel is most likely a router, and what the host should do is
just to send the packet to the link, not trying address resolution for
the destination.
My primary suggestion is thus to remove "problem 1)" of Appendix A.
Or, if we still want to emphasize the "complexity" of the multi-homed
case, I'd rewrite it to:
1) If no Router Advertisement is received on any interfaces, a
multihomed host will have no way of knowing which interface to
send packets out on, even for on-link destinations. One
possible approach for a multihomed node would be to attempt to
perform address resolution on all interfaces. The same
argument applies to a singlehomed node that does not receive
any Router Advertisement, but the step in the multihomed case
involves significant complexity.
JINMEI, Tatuya
Communication Platform Lab.
Corporate R&D Center, Toshiba Corp.
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
