The MPLS WG Archive

[Arunima Dasgupta <adasgupt@osf1.gmu.edu>]

Hi Peter,
   Please see my comments below....


----- Original Message -----
From: De Schrijver Peter <schrijvp@rc.bel.alcatel.be>
To: <mpls@UU.NET>
Sent: Monday, May 29, 2000 6:46 AM
Subject: LDP notification


> Hi,
>
> I have some questions about the interpretation of an LDP notification
> message.
> What is an LSR supposed to do when it receives a notification
> with an unknown status data field, the E-bit clear and the F-bit set ?

The unknown TLV is ignored and the rest of the message is processed as if
the unknown TLV did not exist.  Then the unknown TLV is forwarded with the
containing message.  Please see "3.3. Type-Length-Value Encoding" of
draft-ietf-mpls-ldp-07.txt.

> The message ID refers to a label request. Is the LSR supposed to
> teardown the LSP or not ?


If E-bit is set, the LSR closes the TCP connection and discards all label
mappings learned via the session.  If E-bit is clear, the processing of the
message will depend on U and F bits.  Please see "1.4. LDP Error Handling"
of draft-ietf-mpls-ldp-07.txt.

Should the LSR forward the notification
> message and also include TLV's unknown to the LSR but with both U and F
bits
> set ?    

   Yes.

> Peter.


regards
Arunima