The MPLS WG Archive

["Zhun Jonathan Qiu" <zhunq@usa.net>]

Thank you for your reply James! This helps a lot. However, one more question is, if at
this point the LSR hasn't advertised any label for this FEC to any peer, what would be the
value of "PrevAdvLabel" then?

Regards,
Jonathan Zhun Qiu


-----Original Message-----
From: James R. Leu [mailto:jleu@mindspring.com]
Sent: Wednesday, January 23, 2002 9:50 AM
To: Zhun Jonathan Qiu
Cc: mpls@UU.NET
Subject: Re: Question about LMp.21 in RFC3036


On Tue, Jan 22, 2002 at 08:25:22PM -0800, Zhun Jonathan Qiu wrote:
> Hi, there, I guess this may have been discussed before and if that's case please direct
me
> to the thread.
>
> In RFC3036, A.1.2 (Receive Label Mapping), when we reach LMp.21, we should have failed
the
> check in LMp.18, i.e. LSR has NOT previously sent a label mapping for FEC to the peer.
So
> at this point "PrevAdvLabel" should be undefined. Then how can LMp.21 uses it as a
> parameter to "Send_Message"?
>
> Do we need to allocate a new label and advertise at this point?
>
> I think I must be missing something obvious. Please help me out then. Thanks a lot!
>

My interpretation is:

An LSR may have advertised a label for this FEC to another peer.  In which
case the LSR can advertise the same label or advertise another label and
merge the two.

In other words the check in LMp.18 has nothing to do with 'PrevAdvLabel'.

Jim

> Regards,
> Jonathan Zhun Qiu

--
James R. Leu