It's much easier for a script to parse a number than a date.
Well, I tried to use time_msec instead of time_wall_msec() in an attempt to
see how the output would look like.
To me, the original proposal of printing wall clock time
--
[root@rtr-29-225-196-232 ~]# ovs-appctl bfd/show
---- port3.4094 ----
Forwarding: true
Detect Multiplier: 3
Concatenated Path Down: false
TX Interval: Approx 500ms
RX Interval: Approx 500ms
Detect Time: now -1116ms
Next TX Time: now -436ms
Last TX Time: now +19ms
Last Flap Time: 2015-06-25 00:36:22.372 <<<
[..] <snip>
made lot more sense than printing something like
[sabyasse@sabyasg-lnx-90 sabyasse]$ sudo ovs-appctl bfd/show
---- enp0s10.4094 ----
Forwarding: true
Detect Multiplier: 3
Concatenated Path Down: false
TX Interval: Approx 2000ms
RX Interval: Approx 2000ms
Detect Time: now -4944ms
Next TX Time: now -728ms
Last TX Time: now +1012ms
Last Flap Time: now +2617332ms <<<<
[..]
Again, the whole purpose for this was to give some kind of heads up to the
user about flap and when that happened. (FWIW, I was also thinking of
printing flap_count in bfd/show as well if that made sense).
So.... if there is agreement in printing wall clock time, I'm fine with
printing GMT as suggested in this thread. Otherwise, if following existing
convention for displaying msecs only in bfd/show is a concern, I'm also fine
in NOT displaying "Last Flap Time" at all from ovs-appctl bfd/show, but just
saving the wall clock time in ovsdb as scripts can still read it from
there.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
