Question

How to run metrics-server in managed k8s?

Posted on November 3, 2020
DigitalOcean Managed Kubernetes DigitalOcean 1-Click Apps Marketplace Kubernetes
Asked by johnwook

I spent about 3 days with this issue :(

I tried to install metrcis-server(https://github.com/kubernetes-sigs/metrics-server) on my own cluster. Of course I tried to turn on the option of prefer-internalIP option, which mentioned in the article(https://www.digitalocean.com/community/tutorials/how-to-autoscale-your-workloads-on-digitalocean-kubernetes#step-3-—-installing-metrics-server)
(Because 1 failed) I tried to install metrics-server through marketplace one(https://cloud.digitalocean.com/marketplace/5d83d4f9cc248c518b58a38c?i=654b99)
Now I’m writing this…

No matter what I try to install metrics-server on my cluster, I can’t see CPU/memory usage of my pods. CPU/memory usage of my nodes is fine, but pods.

metric-server log says follwing (for every deployed pods)

1 reststorage.go:160] unable to fetch pod metrics for pod, ...: no metrics known for pod

My cluster runs

1.19.3-do.1 (most recent one)
metrics-server 0.3.7(docker.io/bitnami/metrics-server:0.3.7-debian-10-r89)

kubectl top nodes shows

NAME                   CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
pool-mypoolinfoblahb   130m         6%     1323Mi          42%
pool-mypoolinfoblahb   79m          3%     806Mi           25%

kubectl top pods shows

W1103 16:56:02.901596   25473 top_pod.go:265] Metrics not available for pod default/my-pod-info-blahblah, age: 8m1.901588s
error: Metrics not available for pod default/my-pod-info-blahblah, age: 8m1.901588s

Somebody please help me

Show comments

Submit an answer

This textbox defaults to using Markdown to format your answer.

You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!

These answers are provided by our Community. If you find them useful, show some love by clicking the heart. If you run into issues leave a comment, or add your own answer to help others.

johnwook • November 4, 2020

Accepted Answer

to whom it may concern, DO team would release new DO version to resolve the issue (yay!)

link: https://github.com/digitalocean/DOKS/issues/2#issuecomment-721544306

Wesley Vestjens • September 7, 2023

I apologize for bumping the topic, but the information mentioned by Chris Lenard is slightly out-of-date for the newer versions of the Bitnami metrics-server chart.

We deployed the chart using these values:

apiService:
  create: true
extraArgs:
- --kubelet-preferred-address-types=InternalIP
service:
  nodePorts:
    https: 8443

The divergent https port is required, because out-of-the-box DigitalOcean does not (or not any longer, perhaps) allow binding on privileged ports for containers.

Full command used:

helm -n <your-namespace> install metrics-server bitnami/metrics-server --atomic --set 'extraArgs[0]=--kubelet-preferred-address-types=InternalIP' --set apiService.create=true --set service.nodePorts.https=8443

Chart: https://github.com/bitnami/charts/tree/main/bitnami/metrics-server

Chris Lenard • April 20, 2021

I ran into a similar issue on DOKS 1.20.2. I was deploying using the bitnami metrics server helm chart. I had to set the command arg kubelet-preferred-address-types=InternalIP and apiService.create=true because the api-service isn’t created already by DO.

For reference: https://github.com/bitnami/charts/tree/master/bitnami/metrics-server#parameters