Re: Consensus call on adopting:draft-krishnan-6man-rs-mark-06.txt

[Wojciech Dec <[e-mail]>]

Hi Alan,

that wasn't quite the question I asked. DHCPv6 has a well defined mechanism
to periodically retry, while RS client sending simply timeout. This would
seemingly leave such clients in the proposed scheme with no connectivity.

-Woj.

On 18 August 2010 16:51, Alan Kavanagh <[e-mail]> wrote:

>  Hi Woj
>
> Its the same issue for DHCPv6, if the client dont send a DHCP_Solicit you
> dont get an address. Also, the RS similar to the DHCP_Solicit is used to
> "kick_start" the IP Sub session and as you know there are lots of hosts whom
> dont have a DHCPv6 client and will not have a DHCPv6 client.
>
> The RS LIO is used to cater for hosts who do not have a DHCPv6 Client. Also
> the LIO is used to identify the subscriber line and tie rules etc to this
> sub line.
>
> Alan
>
>  ------------------------------
> *From:* [e-mail] [mailto:[e-mail]] *On Behalf Of
> *Wojciech Dec
> *Sent:* August-18-10 10:46 AM
> *To:* Suresh Krishnan
> *Cc:* Brian Haberman; IPv6 WG Mailing List
> *Subject:* Re: Consensus call on
> adopting:draft-krishnan-6man-rs-mark-06.txt
>
> Hi Suresh,
>
> thanks for your reply. Continued inline...
>
> On 18 August 2010 16:03, Suresh Krishnan <[e-mail]>wrote:
>
>> Hi Woj,
>>  Thanks for your comments.
>>
>>
>> On 10-08-18 07:11 AM, Wojciech Dec wrote:
>>
>>> Hi,
>>>
>>> I have a question or two to the draft authors who can hopefully clarify
>>> the expected context and working of this scheme, which at the moment is a
>>> bit unclear.
>>> In essence the problem this draft appears to be trying to solve is using
>>> RS/RA messages to induce state into intermediate or IP edge devices like
>>> what is done for DHCP, with the LIO being used to induce such state. All
>>> this is presumably meant to take place following an RS message sent by a
>>> client. Thus, my questions are:
>>> How does this solution cope in a case where the client does not send an
>>> RS? (or the RS sending has timed out)?
>>>
>>
>> The first sign of life from the client is either an RS or a DHCPv6
>> message. If the network does not see either of the messages, there will be
>> no address allocated/prefix advertised to the client. The client will not
>> have any connectivity.
>>
>
> Hmm, but if the first sign of life is a DHCPv6 message from a client , then
> why would the RS LIO be needed ?
>
> Now, in the case of a non DHCPv6 client, given that such clients are do
> time out from sending RS messages, how does the solution cater to that? Just
> leaving clients with no connectivity seems like a highly undesirable
> proposition/outcome...
>
> -Woj.
>
>>
>> Thanks
>> Suresh
>>
>>
>