I have a supervised install and this latest supervisor killed my install too. I am on the latest docker 20.10.17, on a pure Debian system. I check every box for a supported system.
Anyway, when supervisor starts the log is spammed with this message:
s6-ipcserver-socketbinder: fatal: unable to create socket: Permission denied
Other hassio containers seem to start for a few minutes, then randomly restart, but when I use docker logs to get a log, it returns nothing. Anybody have a clue how to fix it?
I can start the Home Assistant container separately on the same system, and it starts up. But the supervisor manages the mariadb for the recorder, so I canāt get that going, not to mention the other add-ons.
Iām not sure for 100% what helps but what I did was stopped my plex and nextcloud containers, additionally pull directly 2022.07.0 supervisor from repoā¦ and that helps. After two reboots and many supervisor restarts still working good.
If anyone has my problem, it seems that apparmor kernel modules were not loading for some reason. I switched to an older kernel, reinstalled the latest kernel, and rebooted. It came back up after that.
Thank you, this worked for me. Not sure how the hell you figured it out based on the error messages but Iām glad you are way better at this than I am.
Hi @kevincw01@bkprath@CentralCommand , I am pretty much facing the exact same problem and I am using HomeAssistant OS on RaspberryPi,
so no apt-get cmds that are applicable in the docker/containerized environment.
My docker version is 19.03 and current supervisor is, I think 2022.06.2,
I am hearing from some discussion by @GSzabados that this new supervisor would need docker version 20+ to install the 2022.07.0 supervisor. This seems like a deadlock to meā¦ I
How can I manage to solve this then?
NOTE: I do have access to the files in HAssOS. Is there something ugly that can be done?
then docker pull homeassistant/aarch64-hassio-supervisor:2022.06.2
or ha supervisor upgrade --version 2022.06.2
But I still that will try to install 2022.07 and fail because of your docker version. reimage. Frankly, even if you did get the supervisor under control, you donāt know if the rest of your install is corrupted. You havenāt even got there yet.
The folks on the discord advised me to update hass os from the local command line incrementally (6, then 7 and finally latest 8.x) with ha os update --version 6.0 . This didnāt work for me so they suggested downloading the latest hass os image and flashing it to my SSD.
This did work.
This issue (for users with hass os, not a custom supervisor docker on Linux) was likely caused because I was on an old hass os (v5.13). Many of us were not updating the OS because there were bootloader issues with rpi and USB 3 SSD after that version and we were afraid to upgrade. Luckily this update did not cause a bootloader issue for me and my system was already down so I was willing to take the risk since I had nothing to lose.
The reflash of the SSD wipes your config. Luck for me I have been using the samba backup addon to save nightly ha backups to my nas. After the SSD flash and I did the initial web login to ha, I restored my configuration from a backup by uploading the most recent backup file from nas. Initially hass didnāt recover but I tried a supervisor restart from the local command line and that fixed it.
Damn this sucked a little. My HASS was acting weird and supervisor was down so I rebooted my NUC and everything died. After googling, I found this thread and this didnāt work for me because I was on ubuntu 19.04 and docker 20 doesnāt support that -___-. So I had to manually update ubuntu to 20.04 and then running these fixed it for me. Thanks all!
sudo apt-get update
sudo apt-get upgrade
sudo apt-get install docker-ce docker-ce-cli containerd.io -y
docker pull homeassistant/amd64-hassio-supervisor
sudo systemctl restart hassio-supervisor.service
hm okay now this is weird. Anyone seeing this? If I reboot my system, the supervisor doesnāt come up nicely and I have to sudo systemctl restart hassio-supervisor.service to get my home assistant URL to work.
Iām on Ubuntu docker on NUC, and I know its not supported
Why you would like to install 2022.06.2? That is the one which is causing the issue. 2022.07.0 has already the code to verify if CgVersion is set or not, and it does not error if the CgVersion is not set.
But if you are running HA OS, then I would follow @kevincw01ās suggestion to update the OS as it should have been done a long time ago.
@OkayTrain121, generally as going forward, you will need docker version 20+, but the main issue comes from supervisor 2022.06.2 where it errors if the CgVersion is not set. But 2022.07.0 has a check for CgVersion and if it is not set, then it will not error and fail.
Thanks for this update.
I just came from vacation to a dead HA and after some hopeless poking I have found your post which saved me for today.
I am still on docker 19.03 without easy update path (OS upgrade is on my TODO list for quite some time ), so I tried going straight to 2022.07 and it works!
So simply deleting all present supervisor docker images and then
I was able to get Home Assistant working by updating as everyone says. But I shouldnāt have needed to do so in the first place!! Iām echoing the request for allowing auto-updates on supervisor be disabled. I have my setup working as I need it to, and shouldnāt wake up to an installation broken by an update I didnāt trigger!
Please.
I am in the same boat as everyone else here - luckily: sudo apt-get install docker-ce docker-ce-cli containerd.io
Fixed it - but this is yet another broken HASS headache - and should be FRONT and CENTER on the Forums for the thousands affected.