Is there any consensus on the best approach to running HA on a turris omnia? I’ve tried the different approaches:
HassOS: Actually just tested to get acquainted with HA. No use on a Turris Omnia.
Core: I actually tried in an Alpine LXC container. I don’t like the huge buildenv which is needed. Might try to slim it down perhaps by using a Debian container.
I’ve not tested this on Turris OS yet but I think this needs the skopeo, umoci and jq tools installed. Could be an approach to running HA. I’ve tried it and it gives no errors but starting the container did not start HA. Could be because the comtainer did not have a network but looking into this.
This what you linked is LXD commands not LXC so not supported on TurrisOS.
The idea is to run Supervised version of HASS.io in a container. Like that you can add addons. If you need very basic HA then container approach is one command.
I successfully runned container setup in docker. But cannot get into Supervised install.
I haven’t tested it on Turris but the lxc-create <<name>> -t oci commands work on my lxc install (no lxd). However the created container of HA doesn’t really work. I think it’s missing some things.
For now I’m happy with the container of HA. I’m quite new to HA anyway. If I need more I think I will move to a core install in a LXC debian container.
comment sed -i 's/#lxc.include/lxc.include/g' ${CONTAINER_HOME}/config or container will not start
use docker from source
curl -fsSL https://get.docker.com -o get-docker.sh
sh get-docker.sh
and so dpkg -i homeassistant-supervised.deb
after some cofee i could access HA throuth my local network, install esphome, motion eye and my discovered iot esphome nodes.
edit : reload network did not resolve the network error
edit2: network error will not disapear but with a cron at startup network error will not limit updates
$ crontab -e
add this line : @reboot /usr/bin/sleep 1200 && /usr/bin/ha network reload
as i reload my router each night for updates sleep will wait 20 min for updates and full boot to reload network.
ps network error seems to appear because no networkmanager exists in openwrt, same as apparmor so cgroup.
I have created a fork of @bezda 's script, updated it with @rosbeef 's notes and updated it in general, added lxc-auto start etc. and turned it into actually runnable script (run commands in container using lxc-attach -- ...). The only remaining interaction is the “qemuarm” architecture choice and the final wait for it to load followed by the network reload command, otherwise everything works for me on TOS 7.1.3 on TOW6.
you can edit my crontab line with : @reboot /usr/bin/sleep 30 && while [ $(ha info | grep -e running | wc -l) -lt 1 ] ; do sleep 5 ;done ; ha network reload;
as sad (but also, understandable) as the dropped 32bit support is, with 2gb ram and the cpu speed being somewhere inbetween a pi3 and a pi4, it´s probably better to move HA over to a faster, 64bit platform like the pi4. i just recently moved it over to a pi5 for reasons of “snappiness”. the omnia can still host my voice server and some other small stuff.
PS: i´m sure @venouseek you meant wifi 6 - otherwise i´d like to know how wifi 7 works for you. i´ve ordered some cards to test (haven´t got them, yet), but i´m sure drivers aren´t working well.