The MPLS WG Archive

[Eric Gray <eric.gray@sandburst.com>]

Vijay,

    Consider two cases: one, the targetted LSR _is_ a next hop on
some (one or more) LSP(s) corresponding to a link/interface(s) at
the local LSR; two, the targetted LSR _is_not_ a next hop in any
sense.  In the former case, LMp.12 takes you past LMp.13 and the
local LSR can use these labels when sending packets to the
targetted LSR over an LSP that acts as a direct link.  In the latter
case, the received label is not useful, anyway, and would naturally
be released.

    If an LSR receives a label from another LSR that is not directly
connected to it - by some definition of 'directly connected' - then
I cannot use the label because no directly connected LSR will be
able to interpret it.

--
Eric Gray

You wrote:

> Hello all,
> I have a question regarding the label mapping procedure(A.1.2) of
> rfc3036(LDP).
>
> In Lmp.11,12,13 the next Hop for FEC is checked and if message source is not
> the next hop, the label release procedure is executed.
>
> Suppose the peer which sent the label mapping is a targetted peer, then it
> will not be the next hop for the FEC but still, it may be a valid mapping if
> the mapping has come through the correct next hop interface. So, I think the
> check should do more than just compare the message source against the FEC
> Next Hop.Going one step further, what to do if there are 2 or more targetted
> peers on the same Next Hop interface for the FEC and both of them send
> Unsolicited label mappings. How will we decide which to retain and which to
> drop?
>
> Please correct me if I am wrong.
>
> Regards,
> Vijay