Hi Nick.
Nick 'Sharkey' Moore wrote:
On 2004-06-03, Greg Daley wrote:
RFC2461's Section 7.2.3 describes the router's own recovery from this incorrect state, by sending subsequent router or neigbour advertisements.
Considering that the device doing optimistic DAD which erroneously causes the IsRouter flag to be unset has already sent a DAD NS to the address owner (in this case the router), the router will schedule an NA to all-nodes within a second of this NS's reception.
Aha! The draft suggests an RA will reset it, but on rereading it I realized this was wrong. Seems like the NA will fix the problem ...
Please also be aware there is no issue for default router selection on hosts, (which is what the IsRouter flag is for) since they never receive the RS in the first place.
... insofar as it is a problem in the first place. Anyone know of a situation in which a temporary toggling of the IsRouter bit _on another router_ will affect anything? Alternately, we could consider that last clause of 2461 6.2.6 to be a bug (why _shouldn't_ a router send a RS if it wants to?) and this whole thing a rare enough case to not worry about.
I'd prefer not to rely on changes to 2461. 2461 nodes are our 'deployed base'.
In that case I think I'll correct the text and make it clearer, leaving the option to send with a source address but no SLLAO in there for these hurried types.
It doesn't solve the 'four signals' problem (RS - NS - NA - RA), but improving the RS/RA exchange is beyond the scope of this particular draft I feel.
For the case of wireless hosts (as in DNA), I think that there may be a case where a neighbour cache entry for the host already exists.
This is because a nice host may may be doing (optimistic?) DAD on the link-local address when it checks if the current configuration is valid. If it sends an RS without a SLLAO, then the router with an existing NC entry for the host can immediately send back an RA.
For new nodes 4 signals would be required for unicast delivery, except if the router and host both supported Erik's tentative (existing NC entry safe) source link-layer address option.
In any case, a router may choose to multicast a response. If it does, theres no way to make it send the RA faster, nor is there a need for neighbour solicitation.
Greg
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
