I would like to ask for a way to start Home Assistant even if the configuration files are defective.
Actually I use hass.io and maybe that would be a place where something like this could be done.
But let me explain:
I think that Home Assistant could save a shadow copy of the configuration it last started with successfully and use that shadow configuration in case something went wrong during a following restart. I ask because I already succeeded in having a non-working configuration twice while I’m away from home. That means no warm water when I get home tonight… brrrr…
The funny thing is: check config reported a sane configuration, the restart went wrong and I’m shut out for good. At least I think so… not saying, that anything is buggy or so. I’m sure the mistake was done by myself at some point (maybe I saved a yaml-file after I did a config check).
I rarely reboot my HA instance anymore when I’m not at home or have an ssh tunnel available to the machine running the HA docker instance because of this.
Always use the hassio cli to check the config. The GUI misses failed configs. It reports a valid config but prints errors in the log. If you don’t look there you will miss them.
Well, sometimes I need to restart it (or want to do it), because some third party integration does not work properly anymore and I would like to use HA to its full extent. Of course I could refrain from changing configuration files remotely… But you know… the temptation is stronger
The home automation software I’ve used for over a decade (Premise) maintains a history of its last 8 configurations. If it fails to start with the most recent configuration, it tries the previous one. If that one is also defective, it continues to fall back until it finds a configuration that is valid.
If all 8 stored configurations are defective, it uses a simple, default configuration. In other words, it never fails to start up; it always starts up with something.
Funny thing is, I restarted my raspberry pi yesterday night turning it off and on again. And it started without a hitch. Meaning it was not even a configuration problem at all. It’s the first time it did show such a behaviour. Interesting!
Anyway I still think having a failsafe would be a great Idea. Especially if you can’t access your host by ssh and need to restart it manually.