Hello to everyone.
I have HA (docker version, Hassio plus add-on) on Rpi3+ with last version of HA. That runs without problem.
In the last days, I discovered an average cpu typically higher than the past.
I’m speaking about cpu average above 30% (normally was between 20-25%).
Analysing the reasons,I found that the cpu behaviour of Home assistant docker is primarily based on a profile with repeated cycles of peaks between 10-15 secs, moving by cpu around 10% to short spikes of more than 40/50%.
You can see what I mean in two attached sample pictures taken when there were really low/no entities defined (just basic configuration).
So I kindly ask your support and expertise to understand if that behaviour is correct and aligned to HA internal design or if there is something in my installation that is producing that cpu profile.
Probably abnormal. I would suggest looking at your log files to see if an error is being generated. Something might be trying to communicate, failing, and generating an error.
Really thank you for your suggestion. I checked a lot of time HA logs (with default setting in terms of level of details) but it’s clear and anyway any messages every 10/12 seconds. Do you suggest to do a deeper debugging with different log settings? And by the way does someone other have similar behaviour ? Everyone could quickly check with Portainer docker stats or with cli docker stats …I know also that there are tools like Python profilers to do deeper analysis but I’m not so expert in that area. Thank you
Hi Foxxy. No explicit sensor polling. The data I showed in the pictures were obtained just with a very simple basic configuration (no sensors, no switches, no mqtt integrations, no influxdb feeding,…). I’m thinking if someone of the core developers maybe could tell me/us if that I showed it’s a by design behaviour or not… then, having some more insight, maybe it could be useful to understand if there are tools to do a deeper analysis of the problem. Thank you very much
Thank you maxmaia… that is interesting… I’m running 0.108.6 now but 0.108.3 was issued on april 10th and I had this problem (discovered by a little but constant increase of cpu average, in the range 5-10%) from april 11th (I tipically update the version as soon as available…).
I’m showing in the attached picture the cpu average growth, without any other configuration change (no new sensors or new integrations).
I suppose that some developer needs to investigate… I don’t know how activate them …