It’s a bit difficult when HA runs your house. You need time and a comprehensive wife…
And with breaking change you can’t just downgrade, you need to restore a backup which is a little longer
With the supervised version of Home assistant, if you discover an issue that affects your uncomprehensive wife, all you have to do is click the leave beta button and restart.
Seriously??!!
Then you have absolutely hit the nail on the head
I have been here three years and had no idea it could be this simple to not only try new releases early but help (by reporting) if things aren’t right.
(Is it really this simple and as risk free as it sounds?)
Well there is always risk. That’s what backups are for. Reading the breaking changes before updating goes a long way to managing this risk though.
Yeah I added that last line as an afterthought, it was probably a silly question.
And yes to this: (obviously)
There’s also this point:
Though manually undoing a breaking change edit would probably be quicker than restoring a snapshot.
Especially if you save the change to a text file, or have access to a previous stapshot you can copy and paste a section of config or a whole file from.
My wife is very comprehensive wife. She is extremely good looking (former model), very intelligent, extremely compassionate, a hard worker and she is a former professional blues, jazz and soul singer so I get a mini concert most days. Her sense of humour is similar to mine, as is her taste in food and wine. That is about as comprehensive as you can get. However she does get pissed off if the lights don’t come on when they should.
I think you might have meant to say “understanding”
Is that feature available in a docker install of HA?
Does that still work if your UI is not reachable?
I ask because in my case:
/config/dashboard
/developer-tools
/profile
/logbook
/history
these were all fully or partially unreachable through 3 quick looks into the 0.108.xx releases that I tried.
In the first release I tried even most of my lovelace UI was unreachable due to all the modbus chaos.
Only if it is home assistant supervised (I donlt actually know if you can run core in a docker container, I suppose you could).
Well no. It is a UI button in the supervisor system panel. In that case you would have to use the CLI to downgrade to the previous version.
In my opinion (from what I have experienced recently and in the past) then that would be a very risky way to try to encourage more beta testers. It will fail for them far too often.
It wouldn’t be forcing anyone to joining the program. Just an explanation of how competent users could contribute to the community by helping out with development process by testing the beta releases.
Benefits, risks and mitigation would all need to be explained.
It would be far less likely that you would have this sort of bad upgrade experience if there were more beta testers.
Yes
Even when you need to remove yaml from configuration and go GUI > Integration?
Problem with beta is also that in case of an issue, you grab logs, conf, etc. open an issue on GitHub with all informations and then you rollback (because it HA must run your home without issue).
Then the dev fix the issue, you need to update again, retest, see that the issue is still there, grab logs, rollback, etc.
Best way to test beta is to have a second (empty) house
It is possible to have 2 HA running in 1 home? I will be happy to test betas that way
So beta testing isn’t for you.
There may still be plenty of others that don;t realise it’s an option.
I have a test-HA instance running (although on a RbPi 1), but I doubt if I would be a good beta-tester. Aside from 2 tuya ceiling lights (that don’t have a ESP, so not possible to flash), and a few 433 sensors, all my devices use mqtt. The few other integrations I have are mostly software based ( google hangouts, google calendar, google assistant, weather, influx, …)
absolutely. That’s how I run mine.
Yes.
I actually have three (or sometimes four) instances of HA running off & on simultaneously for different testing of things.
The only thing you wouldn’t be able to test are things that require a dedicated connection to a hub/controller such as zwave or zigbee devices.
Software sensors need testing too.
It is even easier to switch to and from beta with docker/core. My docker command is
docker run --restart unless-stopped --init -d --name="home-assistant" -e TZ=Pacific/Auckland" -v /home/homeassistant/.homeassistant:/config --net=host homeassistant/home-assistant:latest
All I need to do is specify a version number instead of ‘latest’.
MQTT is ideal for this sort of testing. Run one broker and point both instances at it.