[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Notification #9: replay in the future consensus point



Martin Bjorklund wrote:
Andy Bierman <ietf@andybierman.com> wrote:
What if the stopTime in the future happens before the agent
is done sending the replay notifications?

Nothing special, since the stopTime just means that the agent is
supposed to send all notifications that were generated before that
time.  When that's done (whenever that happens) it reverts to normal
mode.


This seems like more trouble to implement
than it is worth.  I think I am not the only
one that does not see a need to expand the replay feature
into something that requires the agent to treat each corner
case differently.

The simple case -- the start to stop interval represents
log entries -- seems to be getting WG consensus.  I have not
heard much support for applying stopTime-in-the-future
as an additional mechanism to transition to live notifications
but stop after some of them.  I think WG consensus is
converging on the idea that the replay feature applies
only to logged entries that exist when the <create-subscription>
is received.

I also think WG consensus is converging towards allowing
interleaving.  I do not know yet if there is consensus
that an agent MUST support <close-session> in notification
mode, even if no other operations are supported.  I think
only one person wants to make sure the agent MUST NOT accept
any interleaved RPCs.  There seems to be consensus that
this option is not what the WG wants.  I think the "manager
SHOULD NOT send, and the agent MAY NOT process" language
is not inter-operable and does not allow for <close-session>.
This is very important text, so we have to keep working on it
until it is right.



/martin

Andy

--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>