More and more depends on my HA installation. But what of it crashes or freezes??? It rarely occurs, but if it does I am in big trouble… Would it be possible to create a notification service in case a specific component freezes or becomes unresponsive? Ultimately it would be great if there would even be a notification in case HA becomes unresponsive altogether. I think it would be a very useful and powerful addition to this great project.
Problem is, if ha is not responsive it’s not likely to be in a good position to be sending out alerts …
You’d be better to use something external that monitors key stats such as processor (by core) and memory load. Better still something that navigates some pages in ha and times it.
Personally, in 12 months, I’ve not done across a situation where ha is unresponsive, but YMMV.
Hassio has a watchdog on the docker container. When it becomes unresponsive, it automatically reboots.
I’ve had this watchdog triggered a few times in the past.
I use uptimerobot.com to monitor my hassio from an external / independent perspective. An alert when I’m out of the house usually indicates an issue with the home broadband or that the power is out, but it’s good to get the heads up so I can try remote access to other servers, see if Plex is down too, etc
I do sometimes wonder if ignorance is bliss and I would be better off not finding out until I get home, but I do also need to consider the rest of my family who might be wandering around the house in the dark wondering why everything is acting screwy.
You cannot expect/program/configure a running device to tell you when it is not running. The only solution is to have another device ask first device if it is okay or even listening. A 5 minute single ping command from another computer should solve that.
try streaming a hacked dafang camera - instant lock up after 5 seconds of video. 
Ha, that is exactly the scenario in my setup that led to multiple notifications from uptimerobot…
1 Like
Good to know this! How can you check if the watchdog has been triggered?
I’ll look into this uptime robot thing too! Thanks 
It’s stated in the log of homeassistant.
I don’t know if you can attach a sensor to it.
You could use the Docker Swarm solution that someone posted around here. If you are indeed in big trouble when your HA fails, why not build in a redundant node that takes over the moment the first node fails. See the HAHA topic.