I am pretty frustrated… I do my best to not let it shine through in the following lines.
The CPU Load I am talking about is the one from
- platform: systemmonitor
- type: processor_use
HA is running on a VM inside a Windows Hyper-V.
Since end of last week I ocasionaly (at least once per day) have the CPU load suddently jump from the usualy < 10% up to 100% and stay there for hours (indefinetely?).
This happens at various times per day and I have not pattern there.
Slow, slugish behavior, missed triggers or actions are the consequence. Overall the C&C becomes unreliable, which is no suprise for an asynchronous system starved of CPU Cycles.
I allready stopped, commented out, disabled various Integrations, Addons, Sensors, Scripts etc. but nothing caused the CPU Load to come back down.
Plus the problem was not there last week and no new Integrations, Addons, etc. have been installed.
Only significant change was the last HA Update, but I have doubts that it is that.
Also I could not find any timewise correlation to anything in the LogFiles, or any Triggers or Conditions in my Automations or Scripts.
The amount of Network Traffic (as reported by HA) is not significant when this happens either. It always stays at a few kbps.
Also the top command is not providing any insight:
As you can see it shows the high CPU Usage in UserMode and Kernel Mode but none of the Processes consume ANY CPU Cycles ???!!! Wtf!?
What I figured out so, is that Restarting Home Assistant does nothing for the CPU Load, but Rebooting the System lets the CPU drop back to the usualy values.
Sometimes it starts again a 10 or so minutes later, but usually the problem disapears for the next hours or rest of the day.
What could that be?
What is the designated way to diagnose which Component/Process/Driver is consumping CPU Cycles on a HA VM?
What tools can I use?
What diagnostic options is HA providing?
Any insights into the diagnostic options available on HA is very much appreciated.
I could not find anything in that regards in the documentatation !?.