The only one reason why I’ve using Home Assistant Supervised, because there’s no docker config for HA with supervisor. Having addon store is really needed for me (and the confort of it). However to use a full VM is soo inefficient in my point of view (both in time and resource). If I need to restart the machine or something happens I don’t want to wait until 2 OS boots up and then HA starts, especially when I build around HA it’s a huge key to have up and running in the shortest period of time. At the end it’s really good that it will stay around this installation method, at least until there will be some easier install for HA + supervisor in docker.
FYI, Supervised uses Docker. You don’t install it in Docker.
Yes, I know it using Docker, but I was not able to install them like an Apache + MySQL + PHP bundle (same container or separate) for instance. I’ve run the scrip which seems like does something like I wanted and some additional thing.
Thanks also from me that the Home Assistant Supervised (HA-S) configuration remains supported. But I have to say, I remain confused.
I carefully read all HA update bulletins, but was caught out because updates to Supervisor are not included in those bulletins. Yes, I should probably have known better by monitoring the forums, but there must be plenty of other people who had no idea that HA-S was deprecated, then put back again.
I would love it if the HA devs could include notes relevant to Supervisor in the HA update bulletins. It would be very helpful. I know that the release cadence is different (and automated), but I don’t think this is a reason not to provide updates.
I would also like to know whether my HA-S build is only temporarily supported by stay of execution, or whether it is for the long term.
I am running HA-S on a Debian 10, built on an Intel NUC. I got a shock yesterday when I saw a new message in the Supervisor log saying that I was running in an unsupported environment. After carefully studying ADR-0014, I discovered that I didn’t have docker configured for journald logging. I was able to fix that, and now I don’t get the error message any more.
I don’t have any obsession about staying with an HA-S build, but when I initially built my NUC, this appeared to be the most straight-forward, well documented way to get going. Still today, I can’t find a better supported installation for an Intel NUC with NVME system drive (there are some hackish looking workarounds, but not much more).
I don’t love building a OSs for fun. I’d really appreciate not being caught out again. Some better communication would help.
Do you have a succinct howto for that, or a page to point to? @kanga_who should probably include that point in his howto (if he hasn’t already).
I think this is the same. @mjoshd posted it in the Debian guide. I’m in the process of testing it in a Vanilla Debian VM to see if I can follow the ADR14 rules to make the guide 100% compliant.
# open/create a file with nano
sudo nano /etc/docker/daemon.json
# insert the following content then save (ctrl + s) and exit (ctrl + x)
{
"log-driver": "journald",
"storage-driver": "overlay2"
}
# restart docker
sudo systemctl restart docker
# wait ~10-15 seconds then restart Home Assistant
sudo ha core restart
Sorry @nickrout - what @kanga_who said. Note that my “out of the box” docker install already had the storage-driver set to overlay2, so I didn’t include that line in the daemon.json file.
If you’re not sure how your docker is configured, just run a docker info:
$ docker info
Client:
Debug Mode: false
Server:
Containers: 19
Running: 16
Paused: 0
Stopped: 3
Images: 17
Server Version: 19.03.12
Storage Driver: overlay2
Backing Filesystem: extfs
Supports d_type: true
Native Overlay Diff: true
Logging Driver: journald
Cgroup Driver: cgroupfs
...
They’ll just change something again to shaft that concept 
.
The only 100% compliant way, is their way (ie HA OS).
4 Likes
Untrue. Stop spreading FUD.
The script sets this as the default config, so there should not be a need to do this anymore https://github.com/home-assistant/supervised-installer/blob/928e46db76912daaee8bf1eb3b8fd1f9af95ce5c/installer.sh#L49-L50
1 Like
Excellent. Just changed overnight I see. 
Ok so it’s worth pointing out that IF you don’t have that logging driver set as per the above to journald that you will get an error (red text saying unsupported state with a link to ADR14) and I fully expect this will prevent future updates but I guess we will see when 0.114.2 is released. Instead of the error I am not seeing:
On existing installs you mean?
yes existing supervised installs… I just ran up an old VM now and updating the supervisor to see what happens…
Was about to do the same. Might actually do a fresh VM install to test. At least we’ll know and can post how to troubleshoot and fix if/when updates won’t work.
1 Like
looks ok. When I updated the supervisor I now see:
But I do have the update button in the dashboard and it seems to be updating
Yes it updated ok…
1 Like
It is an observation and the activity of the last few months seems to indicate otherwise. If it wasn’t true, changes that would affect this would be more widely communicated. They are not so it looks like there is a deliberate policy of making use of this method as difficult as possible. If it isn’t, and you want Supervised to live, you need to communicate breaking changes better.
3 Likes
I noted “You are running an unsupported installation”. Well I was running Hass.io on Proxmox->Debian9 -> Docker-> Hass.io (well do not ask as why I have installed that way as I did my set up long before people in HA community started using Proxmox and I changed my hardware 3 times- just restore the backup from old machine to new and it all worked and even did not need to change IP or USB passthroughs I have x2).
I though the message is due to I was not on Debian 10 and latest docker so I upgraded the system in my back up machine (Dell i7-6700 where I have proxmox 6.2 and other VMs). Now
System Health
Chasis vm
dev false
docker true
docker_version 19.03.12
hassio true
host_os Debian GNU/Linux 10 (buster)
installation_type Home Assistant Supervised
os_name Linux
os_version 4.19.0-10-amd64
python_version 3.8.3
supervisor 235
timezone Australia/Adelaide
version 0.114.4
virtualenv false
Lovelace
dashboards 1
mode storage
resources 2
views 7
Well no luck- it is showing as unsupported.
Well not sure what I need to do. I think I do not need to worry. As every time things fail I have to spend time to fix it reading and researching so what does the unsupported mean? With my install I have full control and I have migrated system hardware several times and it still work. So not sure if I should worry at all.
As the upgrade to Debian 10 did not do any change to make it supported may be I will keep my Ryzen 9 3900x (my main system for many VMs and Gaming machine) with Debian 9 and docker 18 and continue to use HA there. Any suggestions most welcome
I think this is correct.
There is a post about logging in docker - this needs to be changed somehow, but I cannot find the post now!
sudo nano /etc/docker/daemon.json
make sure :
{
"log-driver": "journald"
}
1 Like