Then use proxmox, ditch hyper-V. The plus side to proxmox is debian under the hood.
I have a proxmox VM host. I’ve been using Proxmox since 2009. I don’t use Hyper-V, I was elaborating on the comment he made.
My HA runs in docker on a dedicated NUC.
1 Like
Haha lol. Proxmox is dope. I don’t have a nuc 
couldn’t get the nuc image to work on my optiplex micro pc.
Using a raspberry pi and after spending an eternity getting my lovelace just right my SD card died! What was worse was it was in the middle of taking a hassio snapshot! So I looked into installing on my existing windows server (HP N54L with a SSD and 8gb ram) and after some advice on here went with using a virtual machine using The Hook Up guide here https://www.youtube.com/watch?v=vnie-PJ87Eg
Reboots are much faster, only get warnings about custom components in the log and it has more grunt to be able to use camera streams (I have a load of cheap Yi cameras which are being given a new lease of life with Yi Hack V4 allowing RTSP and MQTT).
Have moved away from using duckdns since I’m already paying for the cloud service and means that I don’t have to open any ports. Tried node red but couldn’t get the hang of it so sticking with YAML for now.
Only thing I haven’t been able to get working at the moment is zigbee2mqtt so I’ve left that running on my Pi (if anyone has any hints it would be appreciated)!
I don’t use hassio. I find no value in it for MY setup. I just use normal home assistant docker.
That’s nice that MariaDB increased performance even though you were running it on the same machine, I’m running MariaDB on a seperate machine but still using an RPI3 to run HassIO. I guess the mem caching may be what gives the performance boost?
I found also that going to MariaDB really boosted history graph performance, + grafana is faster also. Still get warnings on my log however: Update of sensor.status_sxx is taking over 10 seconds and I’d like to fix possible.
I may try HA on a spare Intel CPU based system using Venv Python without HassIO, just to see if it fixes my issues, that would be nice if it does, my RPi really is enough for me however, I don’t restart more than once a day and my DB is only 200Mb so history graphs are fast especially if I use grafana instead of HA.
Very likely so.
Just moved to older HP Z220S with I5. After getting a snapshot restored, just had to change a serial port assignment. Even Z-Wave stick worked without attention.
Running native NUC image/hassio.
eBay for about 100, then added SSD.
Easy change over, and very fast !!!
IMO the main value of hassio value vs plain hass on docker (even with docker compose) is that the addons often come preintegratef/configured for hass.
Also you get a certain standardisation.
Eg nodered is much harder to install (pull etc), add the home assistant plugins than to just install the addon.
Same for VS code.
Easy is good 
1 Like
Sure, but I don’t find that valuable. The Add-Ons were barely there when I started using Home Assistant on Docker, so I had to connect everything up myself anyway. Like I said, it’s not valuable for ME or MY setup.
I use Docker for lots of things, so standing up services for Home Assistant wasn’t really that difficult at all.
Just to chime in, I’m currently running the HassOS VMDK on a NUC with Proxmox and have had no major issues so far.
There are advantages and disadvantages to each approach, though. I like having the addon interfaces in hassio and being able to manage them there. However, there are times when something is needed outside of that, which the Portainer addon can handle. But then you have to maintain two separate ways of managing containers, which is a bit unfortunate.
I also ran into an issue where Portainer stats don’t work on the HassOS VMDK and the issue was closed with an alternative that isn’t really suitable. So I guess they’ll just never work there 
. This wouldn’t be an issue if you ran hassio on a regular Linux install rather than HassOS, but then you don’t get OS updates through the hassio menus.
1 Like
Not arguing, or trying to convince/convert you 
Just laying out my personal view. What works for you is most important!
I also run (a separate) docker instance on another vm with non-HA related software - plex, Unifi controller, Ubiquity UNMS, Caddy, watchtower etc orchestrated with docker-compose. Its great
For HA apps I use explicitly for HA (node-red, VS code, motioneye, ESPHome, Mosquitto etc) the integration that comes out of the box from the addon means that there is a great community for support who all use the same infra.
Yeah, noticed this the other day. All that is required is to add the -e flag option to the ps command in HassOS.
That good or bad?
It’s overkill. Unless you are running a lot of other stuff.
Good advice! Thanks
ran across this thread… thought I would share my experience.
I’m new to HA (going on 1 week plus now). Started with a RPI4 (4gb ram/32gb sdcard).
This system seems really quick and worked great.
I’m in the process of transitioning from samsung smartthings after several ill-experiences with their migrations failing…etc…etc (no need to kick a dead horse anymore). I’m about 50-60% complete on my transition and liking everything so far.
Anyhow… this past weekend I noticed the RPI4 was already starting to show signs of lag, most likely from MEM usage. I ended up resurrecting my openSense build and installed Debian 10, then Hass via docker on it. [i5-3470T CPU @ 2.90GHz w/4GB DDR3 1333Mhz , LiteOn 64GB mSata, Intel I350-T2]
While this might be slight overkill, the responsiveness is much better.
I was able to take a snapshot from the RPI4 and scp it to the new system and import it.
Afterwards I just needed to update the nginx proxy mgr, my port forward rules, and bring over my Zwave/Zigbee dongle (all lights were retained).
Only issue I am experiencing which shouldn’t be related to this migration is getting locked out when going mobile (wifi to cellular). Been reading a few articles on the forum on this.