Hassio_supervisor restart leaves HA + MQTT in non working state

Running on Docker 18.09.3, I am seeing an issue where the hassio watchdog is restarting Home Assistant, I haven’t been able to catch the home assistant at that time to see if its really down… but this would be okay, but then what happens is that it restart the core_mosquitto image which then doesn’t restart with a Error port in use.

19-03-14 10:40:42 INFO (MainThread) [hassio.homeassistant] Updated Home Assistant API token

19-03-14 11:05:39 INFO (MainThread) [hassio.api.proxy] Home Assistant WebSocket API error: Received message 8:1000 is not str

19-03-14 11:05:39 INFO (MainThread) [hassio.api.proxy] Home Assistant WebSocket API connection is closed

19-03-14 11:05:42 ERROR (MainThread) [hassio.homeassistant] Error on call http://172.30.32.1:8123/api/: Cannot connect to host 172.30.32.1:8123 ssl:None [Connection refused]

19-03-14 11:05:42 WARNING (MainThread) [hassio.tasks] Watchdog miss API response from Home Assistant

19-03-14 11:05:44 INFO (MainThread) [hassio.api.proxy] Home Assistant WebSocket API request initialize

To get it working again I have to manually restart (hassio gui) restart the MQTT addon then the home assistant… driving me nuts… any ideas where I can should look?