On Thu, Aug 19, 2004 at 03:11:42PM -0400, Bound, Jim wrote: > Hi Tim, > > Hints Node Reqs I agree. Hints in others I never agreed to at all. > That does not mean that is not the case but lets be clear here people > are running agendas and that is a fact. DHCPv6 will be used by users > and required. If we do draft language that states its MAY and we feel > we have consensus that is fine but I will keep I-TOLD-YOU-SO cards on > all who supported it when the market barfs on it :--). Seeing the mail > I believe it was still for debate. Declare victory Tim and we can move > on but don't be so naieve to believe the entire market is going to use > stateless because we say so that will not happen. At this point I > prefer completed drafts to RFC is my priority. Regarding the matrix m/o > all are possible and should be supported.
Hi Jim, I fully support that fact that managed DHPCv6 environments will be required, and our site is/will be one of them :) So there's no "victory" here until support for full DHCPv6 is out there, for clients, servers, relay agents, etc... if you believe I'm advocating all-stateless that's not true. Right now Section 4 of RFC2462-bis draft says : Router Advertisements contain two flags indicating what type of stateful autoconfiguration (if any) is available. A "managed address configuration (M)" flag indicates whether hosts can use stateful autoconfiguration [RFC3315] to obtain addresses. An "other stateful configuration (O)" flag indicates whether hosts can use stateful autoconfiguration [RFC3736] to obtain additional information (excluding addresses). So it is a hint, but to me there's nothing there that doesn't suggest an implementor shouldn't support RFC3315. I would expect Microsoft, Sun, Linux, BSD, Mac OS/X and HP :) system to all support that. The hint is in the availability, not the support (or that's the way I read it :) Stig raises the interesting question that many sites (again, ours included) will want/need to have nodes able to configure statelessly. At present to do that you must have RFC3736 support (for DNS, etc), so at present I hope that all systems support RFC3736. The interesting case is where nodes that are managed and stateless are on the same link; how does then one "enforce" managed for some nodes? It's not clear yet where the line will be drawn between systems (clients or servers) that provide RFC3315 support, those that provide RFC3736 support, and those that have no support (so end systems can only get basic address and router info, and require manual configuration or a dual-stack crutch to get access to other services). I'm guessing we'll see all server, laptop and PDA OS's doing RFC3315, just because they may be deployed in that environment, and such systems all have DHCPv4 functionality as a basic provision now. Do you see that being any different with IPv6? Tim -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
