How to I diagnose problem of Host Restarting every few hours?

In the past few days my Home-Assistant seems to restart every few hours on it’s own for some reason which is starting to be annoying as you can imagine.

I am running HA OS on rPi 3 - recently installed the latest version on a fresh SD card (and using restored snapshot which has been stable for months now with no issues).

supervisor-2021.09.0
core-2021.9.6
All integrations are up to date
I’ve not identified any significant errors in the logs (but they may be overwritten with every restart?)

Any idea how to diagnose the root cause of these restarts?
I am at a lose where to look for meaningful clues :frowning:

Have a look at system info see CPU/Ram usage and disk space

There’s a known issue with the iperf3 integration that causes restarts in certain situations. If you have that try disabling it.

Thanks @huu

On the host level stats seem pretty clear:

CPU monitoring also looks good I guess:
image

Memory seems a bit high and gradually raising all the time though:
image

How do I dig into that in case it’s needed?

Thanks @Tinkerer

I am not familiar with iperf3, is it related to Speedtest.net by any chance?
Where can I read up on that integration / issue?

Those spikes dont look normal? and both at the exact place memory and CPU? is that the time they restart at well?

The spikes happen on the restarts I believe, AFAIK that’s normal
is it different for you?

anyway - How would I identify the culprit in case it is a memory issue for example?

You mention restore snapshot you might restore some old integration or addons… try update them and host too.

It was actually a pretty recent snapshot, I just moved to another SD card since I thought the issue was with the old SD card - turns out it’s not.

all components are up-to-date as far as I can tell.

Let’s see if anyone else has similar experience recently OR if some instructions on how to diagnose and where to look further are posted by someone :crossed_fingers:

Looking over memory consumption in the past - it seems like an abnormal consistent increase since I upgraded to core-2021.9.X

Is there any known memory related issue on this version @Tinkerer ?

1 Like

I’ve a vague memory that somebody said something - but to be honest I haven’t looked at the open issues to see.