On 8-nov-2005, at 19:20, marcelo bagnulo braun wrote:
One of the issues that are still open is how to detect context loss
and recover it.
[...]
At this point, this would work as follows:
- Regular IPv6 communication is established between A and B
- Shim context is set up
- Peer A losses the context
- a failure occurs
I this case, A doesn't have any shim state so it can't do anything
except time out. B on the other hand will notice a lack of incoming
packets (unless the failure is unidirectional B -> A, that would be
bad here) so after a while a reachability exploration will be started.
It's been a while, so I'm not entirely sure if the way this is written
down now implies that a reachability exploration automatically gives A
all of B's addresses so that it's possible for control packets to
start flowing again from A to B, but if it isn't, obviously we can add
this.
At this point A will need to receive some context and security info
from B and we're back in business. (If B uses a different address B'
towards A then B has to prove that B' is the same host as B, but
that's what we have HBA for.)