The MPLS WG Archive

["Kishore Tiruveedhula" <tiruveedhula@avici.com>]

Here is one more issue:
   In case of loopdetection enabled in Downstream Unsolicited mode, need to
address the following issue.

  Router A             Router B
  --------             --------
  Send Label Mapping
                       Recv Label Mapping
                       Send Label Release (due to Loop found).
  NHOP Changes
  Send Label Mapping
  (with new pv tlvs)

  Recv Label Release
                       Recv Label Mapping.

In the above scenario, "A" ends up with no label advertised to "B", But "B"
has received label from "A".

 Probably, the Label Release message should carry the message id (mandatory)
to indicate exactly this is an ack for which message.

Thanks,
Kishore


-----Original Message-----
From: mpls-bounces@lists.ietf.org [mailto:mpls-bounces@lists.ietf.org]On
Behalf Of Ina Minei
Sent: Tuesday, August 24, 2004 6:43 PM
To: Mark Duffy
Cc: mpls@ietf.org
Subject: Re: [mpls] Requesting your feedback -
issues/errors/clarificationsfor RFC3036



	Mark,

	Thank you for your feedback, please see inline below.

>
> 1.  It would be nice to fold the content of
> draft-ietf-mpls-ldp-mtu-extensions-02.txt into the main LDP standard
document.

	From a standards point of view, this cannot be done, since the mtu
extensions is not yet an RFC.

	Regarding the second point, I will send all comments together to
the list for review/discussion after receiving all the feedback.

				Ina
>
> 2.  Re LDP discovery (RFC 3036 section 2.4):  Two types of peer discovery
> are described: Basic and Extended.  Basic as described will work on
> broadcast and point-point media but not in general on NBMA media (e.g.
> non-LC-ATM when used in a multipoint mode).
>
> Extended can work on NBMA but its semantics are quite different than
Basic:
> it allows "discovery" of peers that are more than 1 hop away, and it does
> not bind the Hello adjacency to a particular interface.  Both of these
> differences can be undesirable in a context where one is looking to use
LDP
> to distribute outer labels with adjacent peers on a non-LC NBMA
> network.  Extended discovery appears to be primarily intended for
> distributing "inner" labels with non-adjacent peers.
>
> I think the successor to 3036 should either point out that basic discovery
> on NBMA is not provided for, or it should specify a mechanism semantically
> similar to basic discovery (1 hop limit, bind the hello adjacency to an
> interface) except using preconfigured neighbor addresses.
>
> Thanks, Mark
>

_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls



_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls