On Fri, Oct 18, 2002 at 01:18:57PM -0400, Vladislav Yasevich wrote: > > - Sections 2.1 and 2.2 are not very clear. For example, in > section 2.2 there are two bullets that modify section 5.4.3, > but these 2 bullets are very ambigiouse as to what parts of > [rfc2462 sec. 5.4.3] they apply to. It would be better if > the draft spelled out the particular modified behavior.
Okay, I'll look into making this less ambiguous. The difficulty is that the changes are relatively small compared to the main text (eg, 2461/2462), and so I've tried to emphasize the changes rather than providing the full rules and making people work out the difference! > Where are these link-layer addresses taken from? > The router will ignore both NS and NA for the following reasons: > - NS is send without the Source Link-Layer Address option. > - NA is sent with the O-bit = 0. Well, more to the point, S=0. So if it doesn't have an entry, it will most likely ignore it. Hesham points out that this is probably an assumption of Fast Handovers type behaviour on my part: guilty as charged. There's nothing more effective than peer review to find your assumptions! In the Fast Handovers case, the router probably has some traffic waiting for the MN. I'll consider more carefully what will happen if this is not the case and post again on Monday. > The only place so far the node can get the address of the router is from > the RA, but Source Link-Layer Address option is optional in RAs > (it's a MAY in 2462). Ummm. Interesting. 2461 says "A router MAY omit this option in order to enable inbound load sharing across multiple link-layer addresses.", but I'll admit I'd never considered it! > For the simple case, there is a very minor effect on the operation > of non-optimistic nodes. An optimistic node will force the state > of all REACHABLE entries to transition to STALE and may force NUD. > This is a very minor effect, but it should be noted. A NA with S=0,O=0 for a REACHABLE entry will have no effect on the entry according to the Appendix C state machine, and these are the only NAs which the ON will sent to All Nodes which Tentative. Admittedly an NA with S=1,O=0 will reset a REACHABLE entry to STALE, but why is the correspondent soliciting for a REACHABLE entry? (Or have I missed something elsewhere?) > Also, the initial delay of the first NUD probe is 5 seconds which is > rather substantial amount of time that the packets may be blackholed > in the case of proxy NAs. Yep. That's probably the worst outcome on the list, but I think it is acceptable as it is both unlikely (new connection on colliding address) and recoverable (by NUD). Thanks for your feedback ... you've certainly given me lots to think about on Monday! -----Nick -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------