I believe this is caused by the same underlying issue as #45310.  Note
that both slaves were reconfigured at 29-Jun-2017 22:13:30 and lots of
"zone <name>: (slave) removed" messages appeared immediately afterwards.