Proxmox shutdown last night, again.

[gfngfn256]

It appears you have no scheduled jobs that should cause your problem. (I will assume you aren't using ZFS - even though you haven't directly informed us of this).

I'd look in the HA logs (Settings, System, Logs & right-hand side drop-down HOST) for this period of time & see what it shows. Also check the main HA core & supervisor logs & general logbook, you may find some activity there.

Even if you had (with setup) an API call from HA to shutdown the node host, this should show up in the PVE logs.

As another test - shutdown the HA VM next week before that 7/8am period, leaving the node up, & see what happens.

 

[gfngfn256]

Forgot to mention in the stress-ng you should also test the network category. I've had enough mini pc's that play up that way.
Have a look at this wiki on the stress-ng usages/options.

 

[esi_y]

I have a similar problem,

If this is also a NUC (which one?) ...

however I don't think my box is shutting down due to temp. The interesting thing is that it always shutdown exactly on Sunday morning, without fail. I've looked at the logs for couple weeks now. It's always just after pam_unix, as if somethin in those cron jobs are triggering a shutdown.

what's worse, the BIOS setting for S0 state don't restart it afterwards.

Despite what it looks like, I would try kernel param intel_idle.max_cstate=1 for a while, if not helpful, then 0 (to disable the driver).