I noticed that when (in my case) I created a template sensor and mistakenly used the same Unique ID for 2 different sensors, when I did a Configuration Check it did not catch it. So after the reboot, the second sensor did not show up and the logs identified the issue as the ID already exists. It would be nice if that could validated.
It’s my understanding that, in order to improve the responsiveness of the UI-based configuration check, it is limited to testing the validity of YAML syntax. I believe the same check when executed from the command line is more thorough (based on the fact it takes much longer to finish).
In the example you mentioned, two options have the same value and that’s beyond a basic YAML syntax check.
After you created the Template Sensor (with a duplicate unique_id) and executed Reload Template Entities, did it report the duplication as an error in the Log?
Faster but way less useful?
Sounds like a great idea.
Really. Who comes up with this stuff and thinks it’s better?
Have you ever ran the config check via cli? It creates a container and basically starts a cloned HA to check the config. You want to wait a whole extra startup duration just to check the config? Probably not. And if you are willing to wait, good news, you can do it from cli.
what was wrong with the way it used to work?
it actually gave decent results and it didn’t take as long as you are describing.
I’m sure there’s a happy medium to be had somewhere.
It still does a more thorough check even now but it just buries it in the logs and/or in a persistent notification.
If you don’t notice that notification and restart it fails to load every aspect of HA that had the error. That doesn’t sound very noob friendly to me.
Just put the results under the config validation check and be done with it instead of hiding it away in the logs. I don’t think I’m mistaken when I say that most users expect the errors to be found and show up there instead of being hidden away.
But I’ve actually used the cli checker and I don’t remember it taking that long. It’s just way more inconvenient to need to open an ssh session and not get the results in the config validation box.
I really honestly don’t know what the fear is with showing an error in a more accessible way. it makes no sense.
No I actually did not try the Reload, I did a reboot instead after changing the ID.
Wasn’t there a recently depreciated configuration check add-on that did a complete check like the OP wants?
Nothing was wrong with the way it used to work which is exactly the same way it works right now. It’s always worked like this, it was never stripped down for speed.
Yes but that was custom. Nope it was official, it was deprecated because HA always starts now even with a bad config.
Perhaps a warning in “repairs” could be a reasonable way to flag this to users?
To save time, use Developer Tools > YAML > Reload Template Entities. It probably would have reported the duplicate unique_id. It’s faster than restarting Home Assistant and much faster than rebooting the computer.
It would have.
Ok thanks. I will try that next time.