For the record, _within limits of current spec/implementation_ MS-Trickle parameter tuning can make the yo-yo effect arbitrarily rare. However, I think that the correct way of doing this would be either:
- Also not form neighbor links without indicator that the link is relatively stable; based on my experience, this is not 100% solution as some links exhibite loss that is function of time; however, this would limit the refloodings just to those links and they are most likely small subset of the whole. - Something non-link-statey, notably not having global reachability graph (this implies TTLs, and selective flooding, so it is relatively different from DNCP as it is written now). As 'common link' concept is needed only by non-adhoc links in HNCP, this would work for adhoc networks _with_ same TLV content but different state synchronization algorithm. We discussed this privately in some group earlier and it also seemed workable. Cheers, -Markus _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
