pod cpu metrics not reported correctly #19
Description
Bug report
Describe the bug
The stats reported by vz-kubelet seem to be the correct data but kubernetes is reporting essentially 100% cpu utilization. For example I have a dormant VM reporting very low usage (per the commands executed to gather stats and fill out the struct). However, running something like this shows essentially 100% cpu utilization:
kubectl describe podmetrics macos
Name: macos
Namespace: default
Labels: <none>
Annotations: <none>
API Version: metrics.k8s.io/v1beta1
Containers:
Name: macos
Usage:
Cpu: 3983005841n
Memory: 352348Ki
Kind: PodMetrics
Metadata:
Creation Timestamp: 2025-05-13T16:20:38Z
Timestamp: 2025-05-13T16:20:23Z
Window: 15.064s
Events: <none>
The above pod has cpu request/limits of 4, the stats data reports something like:
{
"name": "",
"startTime": null,
"cpu": {
"time": "2025-05-13T16:28:53Z",
"usageNanoCores": 259200000,
"usageCoreNanoSeconds": 2364000000000
},
"memory": {
"time": "2025-05-13T16:28:53Z",
"usageBytes": 1135652864,
"workingSetBytes": 895881216,
"rssBytes": 895868928
}
}
To Reproduce
Steps to reproduce the behavior:
Simply run kubectl top pod <podname> or the command above. Or use something like k9s which shows current utilization.
Expected behavior
The kubernetes api reports cpu utilization the roughly matches what's going on in the VM.