Reposting on ccamp: Clarification on RSVP Restart Procedure

[Anca Zamfir <ancaz@cisco.com>]

> Hi,
> I have a couple of questions regarding the RSVP restart procedure:
>
> In draft-ietf-mpls-generalized-rsvp-te-07.txt, section 9.5.2. "Procedures 
> for the Restarting node", one of its paragraph reads:
> "In the special case where a restarting node also has a resta(r)ting 
> downstream neighbor, a Recovery_Label object should be used instead of a 
> Suggested_Label object."
>
> This implies that the restarting node - N1 is able to detect that its 
> downstream neighbor - N2 has also restarted, that is N1, N2 is the 
> restarting order.
>
> For the LSPs established from N2 to N1, since N2 cannot detect that N1 has 
> restarted, N2 will send Path messages using the Suggested Label.
> In this case N1 will not be able to differentiate the case where this 
> message is a new Path message sent with Suggested Label or a Path message 
> for an old LSP for which recovery happens. Is N1 expected to handle the 
> two cases in the same way? If yes, why was the Recovery Label "invented" 
> and why wasn't the Suggested Label considered good for the RSVP restart 
> procedure?
>
> Thanks,
> Anca