The MPLS WG Archive

[David Charlap <David.Charlap@marconi.com>]

Hong Liao wrote:
> 
> In the rsvpte ResvTear msg, is there any label object in the FF/SE
> flow descriptor list?
> 
>  <ResvTear Message> ::= <Common Header> [<INTEGRITY>]
>                                          <SESSION> <RSVP_HOP>
>                                          [ <SCOPE> ] <STYLE>
>                                          <flow descriptor list>
> 
>              <flow descriptor list> ::= (see earlier definition)
> 
> For my understanding, the flow descriptor list here in ResvTear msg
> will be the same as the one in the Resv msg???, if I am correct,
> then the Label object should be in the ResvTear msg????   I
> observed that there is one vendor, it sends out the resvTear msg
> without the label object.

It is ambiguous, IMO.  RSVP-TE does not re-define ResvTear, but it does
re-define flow descriptors.  Nowhere is it explicitly stated whether the
redefined flow descriptors should retroactively apply to all RSVP
messages from other drafts/RFCs.

IMO, you should follow the internet mantra - be conservative in what you
send, be liberal in what you accept.

It is unlikely that you will require the label value in order to process
the ResvTear, so you should just ignore if it is missing.  This is being
liberal in what you accept.  If, for some reason, your implementation
can't process ResvTear without the LABEL object, then you will have to
reject the message, although you may want to see if you can change your
design to not require it.

When it's your turn to send out ResvTear messages, you should be able to
include the LABEL object, even if you did not receive one.  You should
do so if it is at all possible.  This is being conservative in what you
send.

-- David