k8s-infra-metrics not working in Digitalocean kubernetes (sigNoz v0.19.0)

[kmpartner]

I have tried to use dashboards in sigNoz (v0.19.0) that is deployed in DigitalOcean Kubernetes.
When add k8s-metrics (cpu & memory, kubernetes-metrics) dashboard added, it show panels but most of contents of the panels did not display. Dashboard of hostmetrics-k8s.json seems to work and display most of data.
How can I solve this problems?

When I tried to use k8s-metrics dashboard in locally running minikube, part of metrics displayed (not all).

Are these problem in configuration of sigNoz or dashboards json contents?

After installing metrics-server in kubernetes, I could get memory and cpu usage data using kubectl top pod or kubectl describe podMetrics commands.
Are there ways to import data of kubectl top or kubectl describe podMetrics command or metrics-server into dashboard?

[prashant-shahi]

@kmpartner I have faced this issue as well in DigitalOcean Kubernetes.
kubeletstats receiver is unable to fetch metrics since the kubelet API port of the host is not accessible from within the container.

It would take some time to find proper solution around this.
Meanwhile, you can use cadvisor instead to get K8s metrics.
However, the same dashboard will likely not work since the metrics would be named differently.
https://github.com/google/cadvisor/tree/master/deploy/kubernetes

Tracking issue in OpenTelemetry: open-telemetry/opentelemetry-collector-contrib#22843

[kmpartner]

Thank you for response.
Although I managed to send cadvisor data to sigNoz via prometheus reciever, it is very difficult to handle the data in dashboard because of limited knowledge of signoz and prometheus.

[prashant-shahi]

@kmpartner There is an option to import Grafana dashboard in SigNoz.

You could try importing Grafana dashboard JSON from here.

[ankitnayan]

@kmpartner There is an option to import Grafana dashboard in SigNoz.

This usually never works as most of the dashboards have variables and SigNoz does not support converting Grafama dashboard variables today. Users will have to create dashboard variables in SigNoz using clickhouse query as guided at https://signoz.io/docs/userguide/manage-variables/

[kmpartner]

Thank you for responses,
After importing cadvisor metrics and using prometheus query in sigNoz dashboard, CPU and memory usage of pod or container started to be displayed in sigNoz dashboard.