High cpu load and memory leak after update

Installation
1

anyone else have a high cpu load and memory leak that leads to an entire system crash after updating above 0.97.2? I have tried two different versions above that and both seem to cause the issue (most recently on 0.99.2)? ive looked through my log and cannot seem to find anything out of ordinary.
Thanks.

2

Nope. So start disabling your custom components, addons and custom cards until you find the culprit.

You’re not using an older version of the home feed card are you?

That had a high CPU use issue that has been fixed with the latest version.

3

hmm ok thanks, I’m not useing the home feed card. only thing I can think of that may be a culprit is a custom sensor for my unifi network ill give that a shot and see what happens.

Thanks

4

I’ve experienced this at my two homes, similar time frames as you describe. Both HASSIO instances used to run for months but now will exhaust memory after 4-5 days, locking up the system. I’ve also not seen a smoking gun in the logs. Database size remains constant, and I only keep 5 days of history.

To crutch this, I added a Node-Red flow to monitor free memory and initiate a reboot if it falls below a threshold.

My add-ons are Node-Red, Samba, Configurator, Let’s Encrypt and SSH & Web Terminal. Only Node-Red starts at boot-up - the other add-ons remain off unless I manually start them.

If anybody has advice on how to systematically isolate the cause of the gradual memory leak, that would be great.

5

Glances add-on is helpful when trying to figure out where the memory is being consumed. I used it on my setup since I run with 60-65% memory usage on a daily basis. In my setup is my influxdb instance which is running on the same pi right now.

6

Thanks Tom - wasn’t aware of this add-on. Will give it a ride.

7

I don’t know if this will be helpful for anyone but I was actually able to figure it out. I was using the statistics component to average some information obtained from a sensor I made and for some reason it decided it was going to constantly average even if no new information came in from the sensor. once I disabled the statistics component for that particular sensor it stopped using so much CPU and seems to be working fine. I am still trying to figure out what is causing it with that sensor and will report back when I do figure it out.

2 Likes