I have a habit of routinely check my servers, and today I noticed that one of my OrangePis, a Zero one, it was at 400% for over 5 hours, since 3 am. I connected to it, and found out that its tmpfs /var/log and /var/log.hdd directories were full. Manual check returned no results, I couldn’t diagnose the issue.
Also, there were multiple rsync processes from root. I couldn’t understand what’s up.
The temp was 75 ºC and I believe it was that low — and not 95 ºC as it was after a brief cpu-intensive session — only because I set the CPU to powersave from the default ondemand by
echo "powersave" | sudo tee /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
This modifies current behaviour, which would return back to ondemand after a reboot. For permanent change, add echo "powersave" > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor to /etc/rc.local
I didn’t know how to mitigate the issue. In hindsight, I realised I should have killed these multiple rsync processes, and possibly that could solve the issue of CPU being at 400%. But anyway, reboot would be needed anyway. So I had to reboot the server.
The issue is, this board, Orange Pi Zero, it has some weird bug somewhere, it cannot reboot. It only turns off. It’s not a big deal, when your SBC isn’t far (I have my at home), so I’ll turn it on manually in a while, I’ll let it cool down first. But I have an Orange Pi Zero at a remote location, and if I’ll face the same issue there, there’s no chance of manually re-enabling the board. I’ll have to either do a lot of unnecessary moves, or I’ll have to wait till the place have its next power outage. Which happens sometimes.
But what if there’s a very remote location with no power outages at all?