How do you guys run home assistant?
I tried out home assistant using virtual box on my main pc and I enjoyed what I could do with it so far. So, I ordered a mini pc (https://www.amazon.com/dp/B0C89TQ1YF/) with the intention of running home assistant’s OS so that it’s always running regardless of Windows doing what it wants with my PC.
I run other things all the time on my main pc too, plex, isponsorblocktv, tartube, things like that. Is running just home assistant on that mini pc going to be overkill? Should I just put Linux on it and run home assistant on there some other way so that I can also run some of these other things there? Will home assistant take a noticeable performance hit?
You must log in or register to comment.
The house has to work damn near 100% of the time, so I run it on a dedicated Raspberry Pi 4 that has Home Assistant OS with the full stack on it. Works great!
Home assistant in a podman container uses only about 400mb memory and .05% of cpu on my home server.
Put Linux on your mini PC and you can run dozens of services on it without it breaking a sweat.
Same just running in docker have DNS and few other services running side by side on a pi4
RasPi 4B with the dedicated HassOS distro.
Installation was easy. Upgrades run regularly and very stable.
On a Pi4.
I was running it on a VM on the home server but then any downtimes that machine had were also HA downtimes. Decided that mattered enough to run it on it’s own hardware.
Also Pi4, with a RaspBee Zigbee thingy on the io. I think 0 unplanned downtime so far after a year and a half on that device
I have that Beelink and while I don’t run Home Assistant on it (I run that from a VM on my NAS) , it does run a whole bunch of stuff, including Plex, and it’s more than capable.
I think you have a couple of options here:
-
Run HAOS on the bare metal and use Home Assistant addons to add the other functionality you want. Addons are HA managed Docker containers and there’s lots of them out there, including Plex. What I don’t know is whether you can access hardware acceleration this way, which you can do via regular Docker (see below).
-
Install something like Unraid, Proxmox or whatever flavour of Linux you prefer - literally anything that supports full blown VMs and Docker at the same time. Install HAOS in a VM and use Docker for everything else. Passthrough /dev/dri to any Docker containers that use hardware acceleration (Plex) and you’re golden.
It’s a great little box. Enjoy!
-
I personally run it on a MiniPC that runs proxmox. You could run other stuff on it through proxmox as well. I would definitely recommend running always on server stuff on something other than your main daily PC.
You could probably run all the other stuff on there as well, although transcoding might slow things down a bit. HA itself should be fine as once you’ve set up all your automations it’s not like you need to interact with it directly and you’ll just get a little bit of lag at worst.
I’m running it on a pi5. It was able to handle everything including voice easily although the voice thing out of the box is quite limited. I was able to do a lot with it through automations with custom commands though. It was pretty much able to replace my Google assistant system.
I’ve now upgraded it with ollama and Whisper on a network device (old laptop with i7 CPU and GTX1050 mobile GPU) and that works perfectly with llama3.2:3b. Pretty accurate and works well enough. Especially useful coupled with the local home assistant command processing since that by passes the LLM.
Proxmox HA cluster with a SAN. VM migrations go wheeeeeeeeee.
I’d just run HA on the mini PC. There are a boatload of add-ons that you can install which will allow you to make better use of the hardware.
Pretty much any cheap minipc will do. A very power efficient N100 is more than enough, and probably cheaper than a Pi board now.
HAOS on a virtual machine inside on unraid on home server made from old pc parts.
3b+ here. I’d like to upgrade to a 4 or 5 but haven’t had any pressing reason to do it.
Wow! That’s unexpected, because of the memory limitation. I have a Pi3 with Fedora IoT with a target on its back for a Home Assistant container. (I see another comment here notes 400MB of memory use so it seems much smaller than I expected.)
Could I ask how many devices it’s dealing with, or because I’m not running Home Assistant yet, whether routine number/complexity is a better metric? (That question went bad somewhere, excuse me.)
An Insteon integration with a hub and a bit more than a dozen switches, water sensors, etc.
Maybe 4 or so z-wave devices connected via a USB dongle on the Pi.
The Alexa integration for the direct Wi-Fi bulbs. Maybe 6-ish. Also for wife-acceptance.
So- small and unorganized. I really want to wipe and start fresh to be more organized. I also want to learn how to do cross-integration triggers and actions.
Thank-you.
I’m not dealing with much more. Maybe twenty IoT devices in total but quite a few routines spread over Alexa and Smart Life.
One of a few other docker containers on VM via Proxmox on a cheap Chinese mini PC.
I have an Intel NUC (3rd gen I think - it’s several years old by now) which runs Proxmox, which runs several VMs including Home Assistant on HAOS. The only thing I did was upgrade the RAM as the VMs eat this quickly…
Other services I run on this small box are AdGuard, Paperless-ngx, KitchenOwl, tt-rss and two Nightscout instances.
I have a “server” cobbled together out of old PC parts. I have proxmox running on it and Home Assistant is one of the VMs running on that machine.
HA Supervised running under Debian 12 on a mini laptop with an N200 processor takes about 3% CPU. With the Frigate NVR add-on running 2 security cameras it uses about 10%.
