Might have to move my HA to another system

Installation
1

Hey all,

I currently have my HA installed in a docker on my unraid system. The same system is hosting plex, sonarr, radarr, nzbget ect. I have been dealing with weird HA behaviour such as false device tracking triggers which cause unwanted automations. I finally realized it was when the cpu was at 100% while downloading, and extracting media. So im thinking it might be best to put HA back on its own system.

I have a old phenom 2 system that I have proxmox on that I have been using just to mess around with. Im thinking of throwing it on there. Just not sure if I should keep proxmox or wipe it and just use ubuntu server.

My question is how are you guys fairing with Hass on a vm vs a host directly? I do plan to maybe get a intel NUC in the future but just need to stop the weird behaviour im getting for now.

2

Running great in docker on unraid for me. Why don’t you restrict the other services to certain cores and leave HA on its own? Or similar restrict CPU usage in other containers.

3

Same here, running just fine on an oldish unRAID in docker with LetsEncrypt, DuckDNS, InfluxDB, Grafana, Mosquito, NodeRed, Netdata, Plex and some other dockers as well.

I don’t seem to have these issues, but the CPU restriction/separation would by my first step as well.

4

I honestly didnt realize you could do that in unraid, not sure why,

Il have to look into the correct parameter to be entered. Is mostly Sonarr causing the issue when downloading full seasons of something and its moving the files to the correct folder after download. Its not just Hass, the whole system gets slow and unresponsive. It never used to do that too.

My system is running an i3-6100 which is only dual core so maybe thats an issue too.

5

CPU Pinning and memory limitation, add the following to the “extra parameters” section of the container in unRAID:

--cpuset-cpus=0,1,2 -m="8G"

Should pin both physical and logical cores - open your “Dashboard” and it will show which cores are paired.

6

pairs are

0,2
1,3

So maybe for sonarr, radarr and nzbget il do --cpuset-cpus=0,2 -m="4G"

Il give it a go and test for a few days and see what happens.

7

Is it the moving or the unrarring?

8

I think its a bit of both but mostly moving when the cpu can stay consistant at 100%

9

So should have HA on its own pair as well?

10

doesn’t have to be. Could leave it on all cores and that way it can utilize them if sonarr/radarr aren’t pegged and it needs it.

11

Im not sure if im doing aything wrong but when giving my other containers cpu 0,2 it still seems to max out all cores and threads from time to time and then I guess HA hangs up a bit which is when I get the fault automation.