Rajiv,
Thanks. I agree that this covers it. There is a remaining editorial
issue: the draft actually refers to [LDPCap], which is the draft
of RFC5561, so that needs to be fixed. If there are no other changes
I'm sure that can be done by the RFC Editor as a routine action.
Brian
On 2009-07-28 04:22, Rajiv Asati (rajiva) wrote:
> Hi Brian,
>
> Thanks for the review. We believe that your comment is already addressed.
>
> What you are commenting is that if R1 (in the below topology)
> sends "Unrecog Notif" capability to R2 during session
> initialization, which has no clue about this capability, then
> the issue of R2 not understanding the capability would creep up
> during session initialization.
>
> R1---------R2
>
> If so, then you are right about this issue. And we believe that
> this issue is already addressed by the LDP capability draft
> [RFC5561] section 6, as you suggested:
>
> <snip>
> to 0 otherwise. If a speaker receives a message containng unsupported
> capability, it responds according to U-bit setting in the TLV. If U-
> bit is 1, then speaker MUST silently ignore the Capability Parameter
> and allow the session to be established. However, if U-bit is 0, then
> speaker SHOULD send a Notification message to the peer before
> terminating the session. The Status Code in the Status TLV of the
> </snip>
>
>
> Thankfully, RFC5561 is already a normative reference in our document, so, we
> don't think that our document needs any changing. Please let us know if you
> think otherwise. Would be glad to accommodate your suggestion.
>
> Thanks.
>
> Cheers,
> Rajiv
>
> PS: For everyone's reference, here is Brian's comment:
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> 3. Unrecognized Notification Capability
>>
>> An LDP speaker MAY include a Capability Parameter [LDPCap] in the
>> Initialization message to inform a peer that it ignores Notification
>> Messages that carry a Status TLV with a non-fatal Status Code unknown
>> to it.
>
> You use this to avoid the problem of a speaker not understanding "End-of-LIB".
> But what happens if the recipient does not understand "Unrecog Notif"? Doesn't
> that create a similar problem during initialization?
>
> Actually the general design principle of silent discard should apply here,
> to both these indications.
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>
>> -----Original Message-----
>> From: Brian E Carpenter [mailto:[email protected]]
>> Sent: Sunday, July 26, 2009 5:15 PM
>> To: [email protected]; General Area Review Team
>> Cc: [email protected]; Adrian Farrel
>> Subject: Gen-ART LC review of draft-ietf-mpls-ldp-end-of-lib-03.txt
>>
>>
>>
>
_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art
