I’ve been running HAOS (currently OS 11.1, Supervisor 2023.11.3 and Core 2023.11.2) on an RPi3 for a couple of years, gradually adding more devices, integrations and automations, and it’s been great. I love how easy Supervisor makes it to install and manage add-ons.
Not surprisingly, though, the RPi3 is straining under the load, and I’ve had to disable some add-ons to keep it healthy. An RPi4 would certainly do better, but they’ve been hard to get, and even that might not handle things like a local Assist pipeline as well as I’d like. I want to keep HA Core on the Pi, as it is dedicated to that purpose, and things like my Zigbee and Z-wave adapters work well, there.
It would be nice to run at least some add-ons on the beefier server sitting next to the Pi, but have them integrated and managed just as smoothly as they would be if they ran on the Pi, with the Supervisor on the Pi coordinating it all. My server is running Ubuntu 22.04 and docker, and I’m certainly willing to add any other software or services needed to allow Supervisor to use it as a resource. A switch to a different Linux distro wouldn’t be out of the question, either.
I’ve tried researching this, but I can’t find anything that describes this sort of approach. If my Google-fu has simply failed me, I’d greatly appreciate a pointer to how this could be done. If this isn’t a thing… well, is this something the HA devs have considered, as a way to provide flexibility for distributing the HA load to multiple computing devices? Is there a different way to do this that I’m not considering?