Hi,
I went through a lot of documents on how to install home assistant and where to install it.
Right now it’s running as a VM on hyper-v , I was thinking to move it to wither a raspberry pi 4 (I have one for that) or to an old computer with an i7 processor. The reason I want to move it is because I want to be able to use usb dongles for zigbee and zwave, and as you may know hyper-v doesn’t support that.
My question is it worth it to install it on the old i7 computer? Or raspberry pi 4 is enough? I don’t want to struggle with performance, but I don’t want to have something overkill either.
From what I’ve read, if you install it on your own on a computer, you will have to deal yourself with all packages install and update, while on a raspberry pi it’s all handled by the image.
Hi @Pedro1x, I’m running mine in a container (LXC/Proxmox) with 2 cores and 512Mb but I don’t have that much devices/entities/automations yet so the load is very low.
I would think that an i7 would be serious overkill and advice you to try out your RPi.
The only thing there is the SD card <=> possible corruption, which you can avoid by using a SSD.
Thanks guys for your answers.
So in theory I could install debian and run home assistant in a docker.
Would I be able to use usb dongles through the docker?
I missed that detail in your original post
I am using Hyper-V with a usb dongle for Z-Wave. The server sees it using the Windows drivers for the stick and the resulting COM port is mapped to a Hyper-V VM.
I am really interested how you managed that? It’s a known fact hyper-v doesn’t work with dongle unless you use a third party software for a usb redirection.
I have a dedicated RaspPi 4, 8GB with a 120 SSD that works great, with 35 plus Zigbee devices (using a Nortek dongle) as well as a similar qty of Kasa Wifi lights, dimmers and switches plus misc devices like my iRobot, Door Lock and thermostat. I use Node-red for all my automations and everything runs great. I did run on a NUC for a bit (Celeron processor) and it also ran well but that seemed to be overkill, even with a Celeron processor.