In that case then can you make the corresponding change in lib/reconnect.c also?
On Fri, Apr 13, 2012 at 03:57:20PM -0700, Ethan Jackson wrote: > Yes it appears to fix the problem according to the unit tests. It > seems correct to fix it in both places to me though. > > Ethan > > On Fri, Apr 13, 2012 at 15:53, Ben Pfaff <b...@nicira.com> wrote: > > On Wed, Apr 11, 2012 at 10:02:09PM -0700, Ethan Jackson wrote: > >> The python reconnect library attempted to send a probe every 0 > >> milliseconds instead of disabling probing when the probe_interval > >> was zero. > >> > >> Signed-off-by: Ethan Jackson <et...@nicira.com> > > > > I think that this changes the Idle state so that, if there is no > > probe interval, the deadline never expires. That doesn't make sense > > to me, because if we've got no probe interval then we should never > > enter the Idle state: > >> @@ -112,7 +112,9 @@ class Reconnect(object): > >> > >> @staticmethod > >> def deadline(fsm): > >> - return fsm.state_entered + fsm.probe_interval > >> + if fsm.probe_interval: > >> + return fsm.state_entered + fsm.probe_interval > >> + return None > >> > >> @staticmethod > >> def run(fsm, now): > > > > But this looks like the real fix, to prevent the Idle state from ever > > being entered if there is no probe interval (because Active.deadline() > > returns None in this case): > >> @@ -504,7 +506,9 @@ class Reconnect(object): > >> connection is indeed in working order. (This will only be > >> returned if the "probe interval" is nonzero--see > >> self.set_probe_interval()).""" > >> - if now >= self.state.deadline(self): > >> + > >> + deadline = self.state.deadline(self) > >> + if deadline is not None and now >= deadline: > >> return self.state.run(self, now) > >> else: > >> return None > > > > So, if you just make the second change, does it still fix the problem? > > > > Thanks, > > > > Ben. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
