Back

One lunchbreak later, that included setting up an Alpine Linux dev-VM, I can now run "dbus-monitor --system" from within my rootless docker container.

But H-A bluetooth setup is still getting rejected even though I've patched the source. Apparently there's still something left to figure out.

(Seems most people just shrug their shoulders and run their containers --privileged when they stumble upon this problem. I don't approve)

Alright, so H-A uses bluetooth-adapters which uses dbus-fast.

"dbus-fast plans to improve over other DBus libraries for Python in the following ways:

Zero dependencies and pure Python 3"

... ok. So what does the source files say?

"# The auth interface here is unstable."

Thanks - I'll have a look at your unstable code and see if that's what's causing the issues!

Well now H-A doesn't complain that it cannot initialize Bluetooth any longer. dbus-fast didn't seem to perform authentication the way I had read the bug report on libdbus should be done so I hacked something together to test.

My purpose wasn't to use bluetooth but just to get ESPHome to install (which has bt as a dependency) - but this seems important enough for other users that I should test to see that bt actually works now too.

... and switching back to H-A from this window showed that it had now found my bluetooth adapter so that's verification enough.

Hey @homeassistant ! It seems I've managed to figure out the reason for why people are having issues with Bluetooth adapters when running H-A in rootless Docker containers. The Python project dbus-fast isn't authenticating correctly towards dbus.

(see thread)

Now I'll move on to play with ESPHome which is what I wanted to do from the beginning ... this little deep dive has taken way too many hours as it is.