The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-May> msg00216



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Receive Label Receive !!

  • From: Abhijit M Gadgil <amg@sasken.com>
  • Date: Mon, 14 May 2001 12:27:15 +0530 (IST)
  • cc: <mpls@UU.NET>


manoj,

If u have a look at rfc3031, pg 15, Scope and Uniqueness of the Label, It
gives a clear picture as to when Rd can use same label for different FECs
for Upstream LSR(s). Let us suppose the Label Ld resides in some label
space LS. This will not let u distribute Ld for different FECs as it will
not be possible to distinguish the FEC from the label value. If on the
other hand the Label Ld was there in two label spaces (say per interface),
then upon receiving the label release for that Label from Upstream, u will
delete the label from the corrosponding label space and Ld can still be
used by packets for other FECs..

Hope that helps.

-abhijit

On Sun, 13 May 2001, Manoj Dutta wrote:

> Hi,
>
> According to RFC3036 Section A.1.4, a label should be removed from
> forwarding/switching use when an upstream peer releases a label
> (subsection LRl.10). Now, consider a case where a downstream LSR Rd
> propagates label Ld to upstream peer Ru for a set of fec elements {f1,
> f2....}. It seems like Rd will remove label Ld from forwarding/switching
> for traffic from Ru even when Ru releases the label Ld for only one
> fec,say f1.  Consequently, bindings of label Ld to the remaining fec
> elements {f2,f3....} is also lost and all the packets bearing the label
> Ld from Ru will have to be either discarded or forwarded using ip
> forwarding. I think that Rd should retain label Ld till Ru releases
> label Ld for all the fecs {f1, f2...}. Let me know if this is correct.
>
> Thanks in advance
> Manoj
>