Failures in beam_PostCommit_Py_ValCont due to exception in read_log_control_messages

Example

https://console.cloud.google.com/dataflow/jobsDetail/locations/us-central1/jobs/2018-06-26_11_04_04-14383422363721841333?project=apache-beam-testing

All of the failures have the same exception in the docker logs:
I  2018/06/26 18:05:12 Executing: python -m apache_beam.runners.worker.sdk_worker_main
I  Exception in thread read_log_control_messages:
I  Traceback (most recent call last):
I    File "/usr/local/lib/python2.7/threading.py", line 801, in __bootstrap_inner
I      self.run()
I    File "/usr/local/lib/python2.7/threading.py", line 754, in run
I      self._target(*self.args, **self._kwargs)
I    File "/usr/local/lib/python2.7/site-packages/apache_beam/runners/worker/log_handler.py", line 61, in <lambda>
I      target=lambda: self._read_log_control_messages(log_control_messages),
I    File "/usr/local/lib/python2.7/site-packages/apache_beam/runners/worker/log_handler.py", line 107, in _read_log_control_messages
I      for _ in log_control_iterator:
I    File "/usr/local/lib/python2.7/site-packages/grpc/_channel.py", line 344, in next
I      return self._next()
I    File "/usr/local/lib/python2.7/site-packages/grpc/_channel.py", line 335, in _next
I      raise self
I  _Rendezvous: <_Rendezvous of RPC that terminated with (StatusCode.UNAVAILABLE, Connect Failed)>
I  
I  Traceback (most recent call last):
I    File "/usr/local/lib/python2.7/runpy.py", line 174, in _run_module_as_main
I      "_main_", fname, loader, pkg_name)
I    File "/usr/local/lib/python2.7/runpy.py", line 72, in _run_code
I      exec code in run_globals
I    File "/usr/local/lib/python2.7/site-packages/apache_beam/runners/worker/sdk_worker_main.py", line 195, in <module>
I      main(sys.argv)
I    File "/usr/local/lib/python2.7/site-packages/apache_beam/runners/worker/sdk_worker_main.py", line 134, in main
I      worker_count=_get_worker_count(sdk_pipeline_options)).run()
I    File "/usr/local/lib/python2.7/site-packages/apache_beam/runners/worker/sdk_worker.py", line 104, in run
I      for work_request in control_stub.Control(get_responses()):
I    File "/usr/local/lib/python2.7/site-packages/grpc/_channel.py", line 344, in next
I      return self._next()
I    File "/usr/local/lib/python2.7/site-packages/grpc/_channel.py", line 324, in _next
I      raise self
I  grpc._channel._Rendezvous: <_Rendezvous of RPC that terminated with (StatusCode.UNAVAILABLE, Connect Failed)>

Looks like a race condition?
Harness log with startup message is appearing after docker log with connection exception.

Harness log:
2018-06-26 10:40:59.328 PDT Launched Beam Fn Logging service url: "localhost:12370"
Docker log:
2018-06-26 10:40:53.361 PDT Exception in thread read_log_control_messages: