Memory Leak Causing Crashes Multiple Times a Day

Installation Home Assistant OS
1

I am running a RPi 4 4gb on 2022.8.5 (I have been having the issue since 2022.7), and for the last month I have had a memory leak crashing my system several times a day. I think I have it narrowed down to pychromecast because at the same timestamp when the memory use starts climbing, I have errors in my log such as:

[Living Room Chromecast(10.40.1.186):8009] Error communicating with socket, resetting connection

4:54:10 PM – (WARNING) /usr/local/lib/python3.10/site-packages/pychromecast/socket_client.py - message first occurred at 4:08:09 PM and shows up 3 times

[Soundbar(10.40.1.185):8009] Heartbeat timeout, resetting connection

4:54:10 PM – (WARNING) /usr/local/lib/python3.10/site-packages/pychromecast/socket_client.py - message first occurred at 4:08:03 PM and shows up 2 times

[Living Room Chromecast(10.40.1.186):8009] Error reading from socket.

4:54:10 PM – (ERROR) /usr/local/lib/python3.10/site-packages/pychromecast/socket_client.py - message first occurred at 4:08:09 PM and shows up 2 times

Does anyone have any idea how to fix? I have set up DHCP reservations for each device, turned off auto discovery for my google cast integration, spent DAYS googling and can’t seem to figure this out.

image
image2243×403 13.5 KB

This is my memory usage graph from today. Each of the large drops is from a reboot.

2

Disable the integration and see if the situation improves.

3

That could be a symptom rather than a cause, but do what @tom_l suggests.

4

I had a similar issue a week ago. Unstable memory and reboots. The symptoms were the same and the logs were directing me down a rabbit hole. By process of elimination over days, I determined that the problem was not being caused by those integrations at all. But what the logs were telling me was they were having trouble getting communications as the hub started grinding to a halt.

I validated this by watching the memory in real time and watching the log, right before it chugs it would throw an error about not being able to write to the db and there were 40,000 writes queued and it would stop writing. Then my graphs would flatline and the hub crash.

I too thought I had a memory leak. The crash would always coincide steep memory climbs. Somebody said it was a bad DB; I wiped it and still had issues. I restored it to 2 days before symptoms and everything’s been fine since. The conclusion i came to was I rushed some steps during and upgrade and whatever I did, caused the symptoms you mentioned. I could immediately tell the hub behavior was different the second time I upgraded after the restore.

5

Thank you all for the suggestions. I narrowed it down to a different integration dropping connection, causing exactly what Sergeantpup described