Re: [v6ops] new draft: draft-byrne-v6ops-clatip

["cb.list6" <[e-mail]>]

On Oct 31, 2013 7:16 AM, "Lorenzo Colitti" <[e-mail]> wrote:
>
> On Thu, Oct 31, 2013 at 11:04 PM, cb.list6 <[e-mail]> wrote:
>>
>> The address is never on the wire from CLAT, and that topology should
never occur since both 464xlat emit only ipv6, so there should be no case
where the transition happens twice AND that on-path address responds...
>>
>> But you made me think about the case of  tracerouting from 464xlat node
to a ds-lite node across the ipv6 internet... And  that b4 does not use the
specific addresses reserved for b4s.  In this case, the b4 can send a icmp
ttl exceeded to the clat, where the clat will see the packet with a
possible same source and destination
>
> But that can't happen, right? Nobody will ever see either the 464xlat IP
address or the DS-Lite IP address, because neither ever appears on the wire
- both terminals will see the public IPv4 address of the other side.
>

You are right.  In the case I was thinking of I overlooked that the aftr
would translate the ipv4 to public before returning towards the clat.

Furthermore, I doubt the range would be allowed to be translated to a
public ipv4 as a matter of policy on the aftr

CB
> Even if they pass each other their IP addresses through some other
mechanism (e.g., SIP), that's no different from the case where two machines
talk to each other and both have the IPv4 address 192.168.1.2.