James, The topic under discussion is how does a SLAAC PPP client acquire an IPv6 address using DHCPv6? Yes, stateless DHCPv6 can be used to dole out options to the SLAAC client but you cannot have the SLAAC client obtain an IPV6 address with stateless DHCPv6. Sorry, I wasn't clear like so in my earlier email.
Hemant -----Original Message----- From: James Carlson [mailto:[EMAIL PROTECTED] Sent: Monday, July 23, 2007 10:56 AM To: Hemant Singh (shemant) Cc: Ole Troan (otroan); [EMAIL PROTECTED]; JINMEI Tatuya / ????; ipv6@ietf.org; Dave Thaler Subject: RE: Neighbor Discovery and PPP links Hemant Singh (shemant) writes: > Let's clear the air. I agree with Srihari's email on IPV6CP and new > options. So do I. Those issues are just out of scope. > Now let's see what the PPP interop issues are: > > The sender PPP node is ND compliant when it issues a unicast NS for > NUD before sending any data. The PPP peer of this sending node has a > bug as to why the peer is not responding with an NA? Let's focus on > this interop issue rather than asking for 2461bis or PPP to be changed. Agreed. And unless there's text that supports this non-answering mode, that certainly sounds like a bug to me. I'm not sure that every implementation bug necessarily deserves to be enshrined as normative "don't make this mistake" text in an RFC. If others (particularly those who are refusing to send NAs when solicited) feel otherwise, then it doesn't bother me to have additional "must respond" text as long as it doesn't muddy the rest of the document by making it appear as though normal responses are otherwise optional in other cases. > It's an orthogonal issue to ask if one should change 2461bis that > requires an IPv6 node to perform reachability detection before > communicating with another node. As for any changes needed by PPP and > IPV6CP, do note, DHCPv6 cannot be used for other options and addresses > in PPP if the client is SLAAC. It seems to work fine for me. Set the RA 'M' bit to zero, the 'O' bit to one, and the prefix 'A' and 'L' bits to one. That gets me interface addresses via stateless address autoconfiguration along with configuration parameters (but not addresses) via DHCPv6. In fact, RFC 3315 states (among many notes on interoperation): stateful autoconfiguration. Compatibility with stateless address autoconfiguration is a design requirement of DHCP. It seems to have been designed to work that way. Why would it be assumed not to work? -- James Carlson, Solaris Networking <[EMAIL PROTECTED]> Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
