Score:0

Microk8s high cpu usage

jp flag

I recently installed microk8s on a Ubuntu 22.04 server (running on VirtualBox 7, windows 10 as host). The VM has 6 cores and 32gb of ram, so it has plenty of resources.

Microk8s is installed via snap from channel 1.27/stable.

I'm not running any pods and k8s is using ~20% of the system. I've read that disabling ha on a single host deployment made the situation a lot better, which I've done, but is 20% cpu usage on idle normal for microk8s?

For example, this is the output from sar

$ sar 5
Linux 5.15.0-73-generic (********)   14/06/23        _x86_64_        (6 CPU)

16:21:56        CPU     %user     %nice   %system   %iowait    %steal     %idle
16:22:01        all      1.08      0.00     17.31      0.56      0.00     81.05
16:22:06        all      1.30      0.00     15.57      0.59      0.00     82.54
16:22:11        all      1.29      0.00     29.30      0.47      0.00     68.94
16:22:16        all      1.28      0.00     17.21      0.52      0.00     80.99
16:22:21        all      1.30      0.00     15.12      0.51      0.00     83.07
16:22:26        all      0.94      0.00     15.85      0.31      0.00     82.90
16:22:31        all      1.20      0.00     19.19      0.28      0.00     79.33
16:22:36        all      1.11      0.00     16.27      0.48      0.00     82.14

From top I can see that the 2 main processes using CPU are kubelite and etcd. If I stop microk8s the usage drops under 1% of cpu.

I'm not a k8s expert, but this doesn't seem right. Any tips?

I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.