mdsd constantly restarts in Service Fabric

I'm seeing mdsd constantly restarting in a Service Fabric deployment; it's happening roughly twice a second. The configuration is a Service Fabric application on Azure with the stock Ubuntu 16.04 image on the scaleset. Here's an excerpt from /var/log/syslog:

```
Dec 6 02:34:18 isoapi00000W systemd[1]: mdsd.service: Service hold-off time over, scheduling restart.
Dec 6 02:34:18 isoapi00000W systemd[1]: Stopped Azure MDS daemon (on systemd).
Dec 6 02:34:18 isoapi00000W systemd[1]: Starting Azure MDS daemon (on systemd)...
Dec 6 02:34:18 isoapi00000W mdsd[4086]: * Starting Azure MDS Daemon:
Dec 6 02:34:18 isoapi00000W mdsd[4086]: ...done.
Dec 6 02:34:18 isoapi00000W systemd[1]: Started Azure MDS daemon (on systemd).
Dec 6 02:34:18 isoapi00000W systemd[1]: mdsd.service: Main process exited, code=exited, status=1/FAILURE
Dec 6 02:34:18 isoapi00000W mdsd[4145]: * Stopping Azure MDS Daemon:
Dec 6 02:34:18 isoapi00000W mdsd[4145]: No mdsd found running; none killed.
Dec 6 02:34:18 isoapi00000W mdsd[4145]: ...fail!
Dec 6 02:34:18 isoapi00000W systemd[1]: mdsd.service: Control process exited, code=exited status=1
Dec 6 02:34:18 isoapi00000W systemd[1]: mdsd.service: Unit entered failed state.
Dec 6 02:34:18 isoapi00000W systemd[1]: mdsd.service: Failed with result 'exit-code'.
Dec 6 02:34:21 isoapi00000W systemd[1]: mdsd.service: Service hold-off time over, scheduling restart.
Dec 6 02:34:21 isoapi00000W systemd[1]: Stopped Azure MDS daemon (on systemd).
Dec 6 02:34:21 isoapi00000W systemd[1]: Starting Azure MDS daemon (on systemd)...
Dec 6 02:34:21 isoapi00000W mdsd[4170]: * Starting Azure MDS Daemon:
Dec 6 02:34:21 isoapi00000W mdsd[4170]: ...done.
Dec 6 02:34:21 isoapi00000W systemd[1]: Started Azure MDS daemon (on systemd).
Dec 6 02:34:21 isoapi00000W systemd[1]: mdsd.service: Main process exited, code=exited, status=1/FAILURE
Dec 6 02:34:21 isoapi00000W mdsd[4231]: * Stopping Azure MDS Daemon:
Dec 6 02:34:21 isoapi00000W mdsd[4231]: No mdsd found running; none killed.
Dec 6 02:34:21 isoapi00000W mdsd[4231]: ...fail!
Dec 6 02:34:21 isoapi00000W systemd[1]: mdsd.service: Control process exited, code=exited status=1
Dec 6 02:34:21 isoapi00000W systemd[1]: mdsd.service: Unit entered failed state.
Dec 6 02:34:21 isoapi00000W systemd[1]: mdsd.service: Failed with result 'exit-code'.
```

I understand that this is not terribly much to work with; what other debug information would be useful to track down the root cause and get a fix?