The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] [mpls] comments on draft-thomas-mpls-ldp-capabilities-00
Hi Bob, Thank-you for writing draft-thomas-mpls-ldp-capabilities draft. Couple of comments: 1) RFC3031 has a MUST statement that says label distribution protocols must negotiate exp-null capabilities. It would be good to get this into LDP ;-). However, I would agree I don't know anybody that doesn't support exp-null in frame-mode. To be backward compatible, it might be preferable to have a default behavior of supporting exp-null (like the ipv4 fec support mentioned in the draft), and non support would have to be explicitly disabled by signaling. RFC3031 3.16 " Initial label distribution protocol negotiations MUST allow each LSR to determine whether its neighboring LSRS are capable of popping the label stack. A LSR MUST NOT request a label distribution peer to pop the label stack unless it is capable of doing so." 2) Not sure how this is usually done, but perhaps a vendor specific range (or high order bit) for capability codes? or is this usually done after the IANA engagement ? 3) >From reading the draft it is unclear: At session establishment, if a capability ack is requested, is an explicit cap-ack is sent for each cap? Or will a single keepalive will suffice. 4) section 5. Let's say CAP-enabled LDP peer sends CAP-LIST-TLV with INIT message. * In the case the peer sends back a keepalive, this is seen as an 'implicit acknowledgement for any announced capabilities that may require acknowledgement'. However, a peer that does not support CAP would also send back a keepalive (when the U bit == 1 ). The above condition is resolved by setting the U bit to 0, but the draft leaves open the above confusing possibility where it is not really known if the peer even supports CAP-TLV at all. Is there any benefit in this openness, and should we have a recommendation? 5) It is unclear (from a quick read of RFC3036) what kind of notification (fatal or otherwise) is sent in the case U bit == 0 and peer is not CAP enabled. If it _is_ a fatal error, should the session reestablishment backoff timer be ignored? Regards, -- Aamer Akhter / aakhter@cisco.com NSITE, cisco Systems _______________________________________________ mpls mailing list mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls |
|