[On Hold] Deprecating Home Assistant Supervised on generic Linux

I know.

My question was more based around the void that is now being left between a simple Pi install and the jump to learning and understanding VMs, and is there something that will/can fill the gap being left by removing the generic Linux install.


As a novice Linux user, I am pretty confused. I have been running Home Assistant (fomerly Hass.io) on my Ubuntu desktop machine for many years. I have ~10 addons running at all times, and manage them occasionally with portainer. I am assuming by the post and the comments, that my install is no longer supported, and I should migrate to an official install image, like a RPi or a NUC?

Your install will continue to work, but will not be officially supported if you encounter any problems.

Fu… much more work incoming
I really liked an trusted the add-ons - now I either search for all add-ons an alternative or I go for a hypervisor.

(graphic made with https://app.creately.com)

Adds just a unhealthy lot of unwanted complexity.
That also means that I have either get my father in law a second Pi or a full fledged server with hypervisor just that he ll be able to use other containers. :frowning: but hey maybe I ll add pf-sense :man_shrugging:
Anyway - I don’t like the change. But hey. That’s Home Assistant for you. I have to adapt.


This is quite ridiculous… Or I’m missing the point of containers in general.

May I ask the following:

What is the technical difference between the supervisor on Hass.io and Hass.os?

What are the host os dependencies concerning the supervisor on Hass.io? This should not be a problem with containers… So what’s the deal with those “dependencies”

If I deploy an “official” vm (up to today… I though Hass.io was the official way of running ha if Not using a rpi)… How is everything kept up to date?


I actually run HassOS on an i5 NUC. I went the HassOS route because I wanted it to be a dedicated device that I did not have to micro-manage. I think the guide I originally followed is:

And I have it running on an M2 flash drive install. It is initially a little pain, but Linux “Live CDs” (or bootable flash’s) are not hard to come by, and the setup was not too bad after that. Then you just have to update the os through the interface periodically. I have been very happy with the setup for the past couple of years.

Edit: Add the portainer add-on and you have a fairly functional docker host.

Thanks for the reply. It will continue to work … until it doesn’t, then I’ll need to come up with an alternative plan. Thanks again for the insight.

You’re aware that running HA on Docker is still one of the officially supported options?

1 Like

You’re not missing anything. Plenty of us just as confused over those statements.


I was trying to get NUT set up in a docker container (which I couldn’t get to work for some reason) to then link it to HA using the integration.

Then I just installed HA Supervised and installed the add-on which worked fine. Except for the fact that my regular non-supervised HA in docker on the same machine keeps re-discovering it as if it’s not setup yet. It’s done that three times now.

Maybe i can use the config from your linked post to set up the NUT in docker and get it to work that way. But I don’t know if I’ll mess with it as long as my other way is still working.

Thanks for the link tho! :slightly_smiling_face:

supervisor on hassio is like a proxy that manages addons and features you’ll not have with plain docker HA, just like hassOS that on the other way is a dedicated OS, not just an application on top of the OS you want to use, even for other things than just HA.

I think I may migrate off supervised to docker and manually handle my add-ons. I have server-grade hardware and do not wish to use a Pi, a NUC or a VM. I’m mostly 100% docker already.

Will docker remain a first class citizen indefinitely? I’d hate to migrate only to migrate again in future.


I’ve just started learning about HA and got my first one set up but sorry for being a noob.
I’m not sure if my version is the one listed. I have ubuntu running on a HyperV VM.
It is running on docker within the VM.

container is listed as 'homeassistant/amd64-hassio-supervisor’.

I believe I installed it using linux command ‘ttps://raw.githubusercontent.com/home-assistant/supervised-installer/master/installer.sh’

Is this the supervisor version that will not be unsupported? Do I need to set up a new vm using a vhdx listed on their page?

I’m a bit unsure.


You installed HA Supervised - the old generic Linux HA install. It is this method of installation that is depreciated, not the Supervisor itself. As long as it works, you are fine though there is no telling when this will be goosed without notice at the whim of the Devs!

What this did was create the old HassOS (now HA) install onto any base OS you wanted.

1 Like

This is something I think is being overlooked. Proxmox is not a trivial thing to manage. It’s very easy to break something and have it be unrepairable. Just trying to format and add a new drive into Proxmox can be a challenge for a user unfamiliar with the system, and can lead to complete data loss if you’re not careful.

There’s going to be a lot more people needing support by going this route, and I guarantee the questions will be asked in these forums, not Proxmox. I’ve found the Proxmox forums, for the most part, already assume you know a fair bit if you’re looking at installing Hypervisors.


ok thanks. That’s a shame.

I should be able to set up a new vm with the image listed on their site and use a HA snapshot to restore what I have created in my config files then.

i have a raspberry pi 4 with rasbian lite, boot from sd and OS on ssd, i have installed home assistant by doker, what can i do? I don’t want to install hass.os on sd, I don’t find it reliable

1 Like

The other factor to be aware of is that, if a change is made to Supervisor that breaks supervised installs, there is bugger all you can do as the supervisor updates automatically without an option to prevent the auto-update.

Yes, you are running the version that has been deprecated. It will continue to work until some change is made that will cause problems or outright failure

It it is no longer considered to be a supported version of installation or operation. Last week, before the deprecation announcement was posted, I reported an Issue in the Supervisor repo. I was informed to use Home Assistant Core and the Issue was promptly closed.

I followed up with a more detailed explanation but did not receive a reply. Basically, it was already known that the version I was using would be deprecated, and no longer supported, so the Issue was closed.

The announcement made it clear that users of this version are now on their own and must seek community assistance. Fortunately, I found a workaround for the bug.


I always ran core HA in docker and never used many add-ons, couple I wanted I just setup myself. Most of the add-ons make no sense if you manage the OS yourself, I don’t need container manager, or Unifi, or Mosquito, etc…, I manage all that, it’s why I want to manage the OS. I imagine the docker support will remain because it’s what they use to run core HA inside Hass.io and HassOS as well.

I always assumed if you wanted to maintain your own hardware, your not a novice and can manage your own add-ons, I didn’t like supervisor and losing any control, why I manage the server.

I always thought the only two good install methods were full install like NUC or RasPi, aka, full OS and dedicated hardware (I don’t want to deal with anything and want point and click), or docker on my OS and I manage everything.

I never liked penv and I have no idea why they are wasting time with VM solutions, I feel cutting the VM work would have saved more time than the supervisor container solution they cut, heck, cut VM support and go to run our OS or run Core HA in docker and manage things yourself. That’s the two options I would give and kill the rest.

I think the VM stuff is the worst of all these solutions.

I guess what I’m saying, and just a guess, I would do full OS or core docker and manage myself. I see VM support going away, I can’t imagine dealing with support for all that and not focusing on actual HA issues and features, their actual passion. So if I was investing now in migration, it would not be in a VM solution.