Evaluation: draft-ietf-ospf-hitless-restart-07.txt

[hardie@qualcomm.com]

Ted Hardie          [ ]     [  ]       [ x]      [   ]

Hopefully, this will be quick to clear up, since it is may well
be a misreading on my part.  I'm trying to figure out what
happens in the case where two adjacent routers both
wish to enter graceful restart.  From one reading, since
each maintains its pre-reload forwarding tables, the
condition in section 3.2 3 (change in link state database)
doesn't occur.  From another reading, either router sending
a grace-LSA is a change in the link state database (type
change), and both should stop graceful restart on receipt.
Reading this text, though:

               Specifically, if router Y installs a
               new LSA in its database with LS types 1-5,7 and having
               the following two properties, it should cease helping X.
               The two properties of the LSA are a) the contents of the
               LSA have changed; this includes LSAs with no previous
               link-state database instance and the flushing of LSAs
               from the database, but excludes periodic LSA refreshes
               (see Section 3.3 of [8]), and b) the LSA would have
               been flooded to X, had Y and X been fully adjacent.

the authors seem to want to limit changes to LS types 1-5,7
(which the grace-LSA are not).

What is the desired behavior in this case?  Is it the same for planned
and unplanned outages (the text in Section 5 seemed to imply a
more stringent test might be applied here, given that it discouraged
graceful restart in this case in general).