Spurious ceph-mgr failovers during mon elections
This is essentially the mgr equivalent of CephFS's http://tracker.ceph.com/issues/19706
Beacons that should be forwarded get held up on a peon, and the leader thinks the mgr daemon has died.
It shows up as test failures like this:
[WRN] Manager daemon x is unresponsive. No standby daemons available." in cluster log