> > There was also some question in the past of what to do when the M and > O bits *change* between router advertisements. I can't remember who > brought it up. I don't recall seeing any resolution to this. > While secure RA sounds really neat, I suspect it will not be widely > usable on networks where ad-hoc connection is the norm (e.g., SOHO > and WiFi hotspots). If my gut feeling on this is correct, then how > we deal with the case where we get an RA with bad information really > is important. If a bad RA kills my IP connectivity, that's a big > problem. > >
This can happen in the following cases. 1. Temporary unavaialbility of DHCPv6 server 2. Inconsistency of Routers configuration 3. Host changes the subnet and the new subnet may not support DHCPv6 The 3rd case can be handled gracefully as there is a link changes, so the bits will be reset. First two cases may be difficult to handle as the host does not know exactly the reason could be for the bits transition. Even if SEND is used, the inconsistency in router configuration can cause a problem. Nay be a host can do one of the two following. 1. Host keeps the state for all routers for the M/O bits: if the same router is sending the bits with different values then host can accept the change. 2. When the host see the transition of bits 1 to 0 then it can check the availability of DHCPv6 server by explicity sending a solicit (?) message to reconfirm the bits transition. -Syam Bits may go from 1 to 0 > > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > ipv6@ietf.org > Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------