We start the build of 0.44.2 while 0.44.1 had not fix the bug. I think in 2-3 hours you can see the update on your panel. To force fetch version, you can call “restart” on supervisor card. But call in first in 3 hours, before it will not change.
@pvizeli, Supervisor update worked. Now on 0.22.
HA Card shows current 0.44, latest 0.44.2.
After klick on Update it spins a while, then klick on Restart ends in the error postet above.
After hassio-supervisor restart HA is already on 0.44. Also host restart did not help.
Are there any logs about this?
Klick update and wait until you see the new version. The time for update is depens on speed of SD card and internet brandwith.
I think the restart buttin is brocken. It is only a service call. I will move that to supervisor. That will also help that is not possible to call restart on update.
Puuh, took some time for this.
Nothing helped. After removing the image and run manually in a terminal i found the problem.
root@ubu1604:~# docker run homeassistant/qemux86-64-homeassistant
Unable to find image 'homeassistant/qemux86-64-homeassistant:latest' locally
latest: Pulling from homeassistant/qemux86-64-homeassistant
...
docker: failed to register layer: Error processing tar file(exit status 1): write /usr/local/lib/python3.6/unittest/test/testmock/testpatch.py: no space left on device.
See 'docker run --help'.
The disk of my VM was a little to small for 0.44.2, but 0.44 worked.
Also install of Addons works now.
Thanks once again @pvizeli
Yes the system is a docker only construct. You can build local addons or use the appstore inside hass.io but it need to by a docker. On this way we can delivery ready to use application and it give high security between applications.
So Hass.io will only ever be able to run using Docker? Will Home Assistant continue to exist as a standalone version? I am running it in on FreeBSD (in a jail), and have no intention of moving over to docker.
Yes we stay on docker. I like FreeBSD with jails and have also work a long time with it, but it is not a docker with that functionality. Docker run on linux, mac osx, Windows prof and with linux layer also on freebsd. Jails run only on FreeBSD
I am new to docker, but is it possible to have hass.io on multiple raspberry pi’s in a cluster create some form of high availability? Or is the hass.io purely dedicated to a single machine?
The beta is for single only. But it is fix plane that Hass.IO support also snapshot/backups. If that feature inside, it should be easy to write a HA add-on that do that or maby it will be a core feature.
A real cluster is at the moment not possible. Mabye in future with some change on docker swarm it could be possible to run Hass.IO in that clustering. But there are no planes at the moment.
Just realised something, since all of hass.io is based on resinOS, i would love just the hability to run docker containers (based on resinOS images, to run on raspberry).
I noticed you are using FROM %%BASE_IMAGE%% in the dockerfile example for adons.
Also, one thing i did not understood : if i make a github repository, there is no point on having the dockerfile on dockerhub too ? (and no building process too, since that’s to generate multiple images for each platform, right ?)
Plus what does the homeassistant/docker-build-env do ? i mean there is no source available ?
Started up a new fresh install of Hass.io. After first upstart i was able to see the frontpage at hassio.local:8123
I went in to Hass.io settings tab and there i saw some text below “Installed Addons”.
I did shutdown the pi to add the “authorized_keys” file to be able to SSH into the pi.
At second upstart there is nothing below “Installed Addons”.
How do i add addons?
We only support this 4 add-ons arch types. A maintainer of a hass.io repository need to prebuild all add-ons to docker hub, in this case we make a arch support and need only 4 docker image to support all platform instead to build 30 machine based image. So you can use our build script https://home-assistant.io/hassio/addon_publishing/
But at the end, you can use any stuff and push it to docker hub. With image setting inside config, we load it.