Comment 5 for bug 1572237

The above logs are for the controller at a TRACE level. In that case, I did not see the lease manager error.

I see entries for pinging the unit that was having problems:
2016-04-19 19:11:03 TRACE juju.state.presence presence.go:671 pinging "u#rabbitmq-server/0" with seq=45

But I looked at the logs again, and found this error hiding in plain sight shortly after the last of the above entries shows up in the log:

2016-04-19 19:11:03 INFO juju.state.presence presence.go:542 pinger loop failed: E11000 duplicate key error collection: presence.presence.pings index: _id_ dup key: { : "9f3fbe9b-f105-412c-85af-3487741c5159:1461093060" }

After that error (which really should be logged at ERROR not INFO), I see no more logs for pinging the unit that was showing up as lost.