I’ve recently setup an Intel NUC8I5BEH with an i5-8259U CPU. I installed k3os on it, basically vanilla (I’ve only installed a bitwarden helm chart, but this problem exists before that’s installed). It’s not the beefiest thing in the world, but I expect it to run this fairly easily, especially at idle.
The problem: looking at CPU usage in top, k3s-server idles at around 6-8% which seems really high. It also means that the fans in the NUC have to ramp up a bit every now and then meaning it’s a lot noisier than I’d expect. CPU usage also very occasionally spikes to ~20% cpu. All of this is spread across all the cores.
PID USER PR NI VIRT RES %CPU %MEM TIME+ S COMMAND
3245 root 20 0 1311436 565748 8.0 3.5 299:32.38 S k3s-server
4053 root 20 0 140000 27908 0.7 0.2 5:04.00 S local-path-prov
12929 root 20 0 187060 78940 0.7 0.5 19:37.91 S traefik
4492 root 20 0 746916 42932 0.7 0.3 11:59.90 S coredns
3379 root 20 0 822576 139560 0.7 0.9 22:29.98 S containerd
Is this actually high or does it seem reasonable? The node running the containers has low usage (1% via kubectl top node). Anything else I can do to see what’s going on? /var/log/* doesn’t have any messages when idling.