Re: [dhcwg] about draft-ietf-dhc-dhcpv4-over-ipv6-01.txt

[Peng Wu <[e-mail]>]

Hi Francis,

2012/3/22 Francis Dupont <[e-mail]>:
>  In your previous mail you wrote:
>
>>  If there do exists multiple CRAs on the same link, what about refering
>>  to the standard procedure of multiple relay on the same link, if there
>>  is any?
>
> => all the relays forward to all the configured servers, for responses
> they are sent to relays so there is no further duplication.
So you believe this is especially harmful in the over-IPv6 context?
I'm sorry that I'm still a little lost...
Could you explain why and how the flag you proposed work? Thanks

>
>>  Yes, the second usage doesn't match. About the first usage, what about
>>  we fake an IPv4 address and fill in the server id field? We could
>>  extract 32 bits from the interface id in the IPv6 address, or hash the
>>  128 bit IPv6 address to for the 32 bits. Is it good enough for server
>>  identification?
>
> => it is good but it breaks the second usage so it won't work...
Sorry, apologize that I misunderstand what you point in the previous mail.
Here are some thoughts on this issue,
1)For the case of CRA co-locating with host, this should not be an
issue because we'll use IPv6 unicast, no matter what is written in
this field. We should state it in the draft.
For the case of CRA on the link:
2)Make it mandatory that the client send out DHCP message in broadcast
manner in this scenario, or
3)Make the CRA to be the gateway and relay unicast DHCP messages;

>
> Regards
>
> [e-mail]