IAITS but it seems silly to query webservice APIs on every restart. For one many of them have rate limits, and secondly it just seems a bit silly for stuff that doesn’t change very often anyway.
So, on startup it needs to validate/hook up with data again. Which kinda makes sense for (most) integrations.
The real question here: Why the heck do you restart that often, that this is a concern?
I might restart very often sometimes when I try some new stuff or just mess around. Maybe too often
I restart when addin new sensors, configuration.yaml changes, known devices changes etc. So it’s not really a matter of wanting but must… I would also like to avoid restarting, mainly cause it looses all my last changed times…
I restart ALL the time. Add a template sensor? Restart. Add a sensor to Xiaomi Aqara Integration? Restart. ZHA starts being weird? Restart. HACS updated for the 23rd time today? Restart. Decide I want to use the cloud coverage sensor from the openweathermap integration? Restart. Add a new template light or switch? Restart. Didn’t get the code for that template switch just right? Restart 10 more times.
I don’t think my Home Assistant instance has ever had a week of solid uptime since I started using it 4 or 5 years ago. Not because it crashes. Because I tweak something that requires a restart.
Tell me about it! I still use the home automation software I started with 13 years ago (alongside Home Assistant). I have restarted Home Assistant more often in the last 2 years than the other one in 13. You can change anything without even requiring a reload let alone a full restart. I don’t think that degree of ‘non-stop’ operation is in Home Assistant’s near future but a few more reloadable integrations would be appreciated.
I’ve also hit rate limits on API’s when doing lots of restarts, it can be annoying.
I’m not really sure of a good solution to this, right now I’m just using workarounds. Maybe have API querys to not be automatic query, and trigger queries with automations? So use an automation to trigger a script?
For now I query most REST API’s using NodeRED and then feed the results into HA, not the best workaround, but it’s simple and it works; at lease I’m not hitting rate limits now.