[bug] Why no data in CPU / Memory usage?

[k-szortyka]

Describe the bug

I don't have metrics for Namespace(s) usage on total cluster CPU/Memory in % and Memory / CPU usage by Pod,
I have changed image="" with image!="" and it not resolve the issue.

sum(rate(container_cpu_usage_seconds_total{namespace=~"$namespace", image!="", pod=~"${created_by}.*", cluster="$cluster"}[$__rate_interval])) by (pod)

The same with Namespace(s) usage on total cluster CPU in %

How to reproduce?

No response

Expected behavior

I would like to have metrics of usage CPU and memory of cluster.

Additional context

No response

[dotdc]

Hi @k-szortyka,
Do you have any series when looking at raw container_cpu_usage_seconds_total in Grafana explore ?
Can you give more information: Kubernetes distribution, Kubernetes version, deployment method/chart...

[k-szortyka]

@dotdc I have any series but I can't filter by cluster or namespace which is crucial.

I use AWS EKS with the following version: 1.29 and kube-metrics 2.11.

[dotdc]

Do you have the machine_cpu_cores metric ?
Did you install node_exporter (helm chart) ?

[k-szortyka]

@dotdc Yes I have but without cluster and namespace filters.

I don't have node_exporter. I'm using kube-state-metrics and metrics-server.

[dotdc]

@k-szortyka non-existing label shouldn't be an issue.
Can you try with node-exporter ?