> Treating RA information that DHCPv6 servers aren't available as a > hint and ignoring the hint is an implementation of the ability to do > DHCP without having to configure routers.
I agree and I think most agree that the router RA determines if dhc is to be used at all. Thus that purpose of the "m" bit is valid and useful. Thus I support that if the router does not set the m bit that the client MUST not use dhc. > > This can be accomplished in two ways - have clients only send > > Information-Request OR allow them to send Solicits and > "fix" DHCPv6 to > > allow only other configuration parameters to be communicated in an > > Advertise. Of course, the latter has come > interoperatability issues > > with > > existing implementations. > > Right. If the M bit is set then it is up to the dhc client per the dhc server what the client is permitted to do as dhc client. This implies to a network site setting the m bit means giving all dhc control for policy to the dhc server as to what the client can do or not do. Which to me makes sense. This means we don't need the o bit is my view. Now as that affects ND and Addrconf and all I care about for this discussion, not dhc for a moment, here is my suggestion, if we did reaach consensus we only need the m bit. Deprecate the o bit in some proper IETF way, and leave that bit position for 1 year alone and not resusable for 1 year. This permits ND and Addrconf code bases to update and simply delete the code base for the o bit. How we do that in a "proper" way is important. Then that bit would be reusable for the IETF within ND after 1 year. My first concern is to reduce the pain to ND and addrconf implementations which is widely deployed. Then dhc can work on the means to set the policy which is dhc and I think backwards compatibility will not be painful, if we do this very quickly. Note very quickly like in 3 months. /jim -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
