Re: about draft-arkko-multi6dt-failure-detection-00.txt

[Erik Nordmark <erik.nordmark@sun.com>]

Francis Dupont wrote:

> => IMHO we don't need a bit but simply to add a threshold condition, for
> instance if the preferred time jumps from at least X hours to zero so
> existing communications may be affected. And add an advice to use
> smooth decreasing to zero in RAs for the standard semantics.

IMHO adding a bit is a lot clearer than trying to overload the lifetimes.

In the case of overloading, what would happen when a host is unreachable 
for some time, during which the deprecated lifetime announced by the 
routers decreases in real time (due to some planned renumbering).
When the host becomes reachable it sees that
	the last RA had lifetime = X
	the new RA has lifetime Y which is a lot less than X
	the remaining lifetime in its prefix list is close to Y

Figuring out robust rules for how the host says that the above is not a 
signal of a failed prefix, but other cases are, would be non-trivial.

*If* we are going to pursue this, then just defining a bit would be 
trivial.

   Erik