On Jul 11, 2005, at 11:54, Elwyn Davies wrote:
Late breaking thought...Paragraph 2 of Section 7.2.1 of draft-ietf-ipv6-2461bis-03.txt says that joining and leaving the solicited-node multicast group SHOULD be done using MLD (and the reference is to RFC2710 i.e., MLDv1). We now have MLDv2.. RFC3810.. Should something be said about this?
Since the MLD reference is Informative, it could be augmented to mention both 2710 and 3810.
If I understand the compatibility section RFC3810 correctly, if hosts continue to generate MLDv1 Listener Reports the routers will be forced to operate in MLDv1 compatibility mode for the solicited-node multicast addresses, perpetuating this mode for ever. Is this desirable? Presumably this can only be handled by providing a configuration flag which would tell an interface to send MLDv2 Listener Reports instead of MLDv1 ones.
The main, current benefit of MLDv2 is to support source-specific multicast
(SSM). However, the NDP use of multicast is non-SSM (ASM or any-sourcemulticast) and does not need MLDv2 capability in order to function correctly.
I would support adding an additional reference to 3810 in order to encourage
implementors to move up in capability. Regards, Brian
smime.p7s
Description: S/MIME cryptographic signature
-------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
