My hass.io is running in a VirtualBox machine, on a Win 10 system.
The hardware is a Z8300 CPU with 2GB RAM (64 bit).
The hass.io VM has 784mb memory allocated, 2 processors with hardware acceleration enabled.
Once in a while (2-3 weeks) the system hangs and I see on the console an error msg that says: “end Kernel panic - not syncing: System is deadlocked on memory”, and then the system is locked until I force a shutdown.
HA runs well on a 1GB RAM pi. I know Win 10 takes up some memory but I have 1.1GB memory available on the machine before starting the VM.
How do I proceed to debug this? What could be the source?
I had the same error with VirtualBox an a Mac mini server.
I made an automation that restarts HA everyday at night. That works for the kernel panic but creates other problems.
I ended up installing hassio on docker inside a Latte Panda Alpha (similar to intel NUC) with ubuntu. Works great for month now.
I’m experiencing the deadlock on memory too. I’m running hass.io on a NUC. Same as most others: VM with Virtualbox using 2 of the 4 cores and 3GB of RAM(!!!).
In the beginning I experienced this only once every few months. But since a few weeks it’s happening every few days. I’m monitoring the CPU and memory use inside hass.io and made an automation to restart hass.io if it’s running out of memory (>90% memory use = restart). This didn’t change anything.
Hass.io uses only 20% of the available RAM. It doesn’t rise over time at all, according to the graph.
Does anyone have an idea of what could cause it then?