Hello,
so today it was basically impossible to use Homeassistant after I updated it to 2022.03.03
the CPu Usage is suddenly off the charts, no logs to find out why. This is new, it wasn’t like this before.
Also new is how my dashboard suddenly freezes. The moment I fold in the navigation (sidebar) my Dashboard starts twitching and then it completely freezes. I can’t do anything anymore. Anyone else having these issues since the update?
Not using that much cpu but also >20% while idle’ing here on my Pi 3
I have the same issue, but for me it was before updating to 2022.03.03
(Was on 2022.03.01 while receiving an High CPU Load & Temp Warning)
Was thinking an update to latest .03 would resolve the issue, but the issue is here again rising to 90%+ Load after few minutes.
(Running on rp4-64, hassio)
Was watching the event-stream while listening to “*” but… nothing strange.
It has to be core-related, all addons behave normal.
I attach a screenshot:
It started exactly at 20:00 today. (All the drops are restarts or core-update trying to fix it)
(I have no automations or addon scripts like in node red or something starting at 20:00)
edit: there is also something that fills up the memory as its load is rising while CPU load is so high.
A restart does not resolve the issue. (after few minutes load is up to 99% again)
edit: also checked the logs: no anomaly
I will write an core issue on git when this persists. For now I added an automation, when CPU above 95% for 8 min straight: restart HA.
The script was restarting the pi 9 times this night:
It seems, that since 4:00 the CPU Load wasnt rising anymore. (at 5:00 there is my bi-daily restart)
@Timaelly is your issue ghostly gone again too?
Same here, i came back this morning to this
Absolutly no idea why, I even assigned the VM more space and another virtual CPU unit. the logs seem all fine. Any idea where else I could check?
Also, no it didn’t really get much better. I’m backing everything up now and rolling back to 2022.3.2 if that won’t help I’ll roll further back. Though I will try to uninstall all add ons first. Never had this problem before. So weird
Update
I removed Cardmod and Layout card and yep, those are the ‘culprits’. Which I don’t get since they worked fine just before the update. No massive RAm spikes or cpu spikes before.
Another Update:
I uninstalled the mods and reinstalled them. RAM is good, CPU is good but it’s still quite slow.
For example, I get stuck on loading the cards/code for a while, the dashboard freezes for a moment and everything slows down.
But it only happens when I open the dashboards and try to do something on there or go somewhere from there. It seems the update and the mods don’t mix well yet, I’m not sure how to post this issue in github ( thomaslovens mods) yet or maybe he’s already aware of it.
Last update here:
So in the end I rolled back to the core-2022.2.9 Version, everything is back to working perfectly fine.
I guess I found the solution. Some day ago I read somewhere that it‘s important to use “card_mod:” before “style:” now.
For example.
views:
- badges:
- entity: sensor.home_assistant_website
name: Update
card_mod:
style: |
@keyframes blinker { 75% { opacity: 0; } }
This seems to solve have impact on high docker CPU usage.
Well yeah, card_mod is needed for the card mod-addon to work. I’ve rolled back to the 2022.02.9 Version and have had no problems at all since then. This is definetly the new Update that isn’t mixing well with the Comunity Addons and it might be sometime before any fixes are out. Also the new Update messed with the input_select helper. You can’t delete the options anymore, that was also fixed by rolling back to 2.9
Can’t confirm that rolling back solves this issue. It seems to be a bit better but I guess it’s only an impression.
In my case 2022.2.9 has same irregular freezing issues. Any 2022.3 version seems to behave nearly the same and CPU behaves normally for several periods.
Didn’t have the time yet but it would be interesting to test dashboards without the card_mod. As I’m quite new to card_mod I’m not sure it there’s an possibility to increase components’ loglevel.
Same here. Started last month. Running HASSOS on VM in ProxMox. Reboot, and then CPU usage rises slowly over an hour to 65%.
Solved! Reloaded fresh HassOS using Proxmox script by TTech and restored from backup, didn’t work. Deleted/disabled integrations/add-ons and stripped the yaml configurations one by one. The culprit was either an Amcrest camera or a template sensor that was no longer functioning. 3% cpu nice and cool and 3GB Ram. Hooray!
