Metrics-server: error: metrics not available yet

[leeruibin]

ENV:
kubernetes: v1.13.2
docker: 18.06.1-ce
system: ubuntu 18.04
architecture: arm-64
metrics-server: v0.3.1

Before I apply the 1.8+/ development, I try the command
kubectl top nodes
Then I got the error:
Error from server (NotFound): the server could not find the requested resource (get services http:heapster:)

After that I download the heapster from https://github.com/kubernetes-retired/heapster and apply the deploy.
The I run the

kubectl apply -f 1.8+/

and check the pods by
kubectl get pods --all-namespaces

The result are as follows:

NAMESPACE NAME READY STATUS RESTARTS AGE
default back-flask-86cbc75cd9-jdgwr 1/1 Running 0 3h53m
default demo-flask-756f45c879-68fvk 1/1 Running 0 3h53m
kube-system coredns-86c58d9df4-c4bv4 1/1 Running 0 4h16m
kube-system coredns-86c58d9df4-cbdvq 1/1 Running 0 4h16m
kube-system etcd-robbon-vivobook 1/1 Running 0 4h15m
kube-system heapster-f64999bc-2klfs 1/1 Running 0 91s
kube-system kube-apiserver-robbon-vivobook 1/1 Running 0 4h15m
kube-system kube-controller-manager-robbon-vivobook 1/1 Running 0 4h15m
kube-system kube-flannel-ds-amd64-8fxvb 1/1 Running 0 4h13m
kube-system kube-flannel-ds-amd64-xjsk6 1/1 Running 0 4h14m
kube-system kube-proxy-4smxt 1/1 Running 0 4h13m
kube-system kube-proxy-vv5xh 1/1 Running 0 4h16m
kube-system kube-scheduler-robbon-vivobook 1/1 Running 0 4h15m
kube-system kubernetes-dashboard-697f86d999-jk4b7 1/1 Running 0 3h33m
kube-system metrics-server-7d8db6b444-tzgbk 1/1 Running 0 19s
kube-system monitoring-grafana-57cbc5c659-zj6fq 1/1 Running 0 91s
kube-system monitoring-influxdb-7f897d5cc8-c2fsh 1/1 Running 0 91s

But when I run kubectl top nodes to check, I get another error:

error: metrics not available yet

So, what's wrong with this?

[ymrsmns]

Run
helm upgrade --install metrics stable/metrics-server --namespace kube-system
Or
kubectl create -f https://raw.githubusercontent.com/kubernetes/kops/master/addons/monitoring-standalone/v1.7.0.yaml

Add
Removal (Kubernetes 1.13)
Heapster will be migrated to the kubernetes-retired organization. No new code will be merged by the Heapster maintainers.
https://github.com/kubernetes-retired/heapster/blob/master/docs/deprecation.md

[yanghaichao12]

Does your apiserver open the aggregation layer?

--requestheader-client-ca-file=/etc/kubernetes/certs/proxy-ca.crt --requestheader-allowed-names=aggregator --requestheader-extra-headers-prefix=X-Remote-Extra- --requestheader-group-headers=X-Remote-Group --requestheader-username-headers=X-Remote-User --proxy-client-cert-file=/etc/kubernetes/certs/proxy.crt --proxy-client-key-file=/etc/kubernetes/certs/proxy.key

and how about your rolebinding?

[latchmihay]

its probably the self signed cert on your cluster.
Try enabling --kubelet-insecure-tls

As in

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: metrics-server

      containers:
      - name: metrics-server
        image: k8s.gcr.io/metrics-server-amd64:v0.3.3
        imagePullPolicy: Always
        args: [ "--kubelet-insecure-tls" ]
        volumeMounts:
        - name: tmp-dir
          mountPath: /tmp

[tomsherrod]

I had the "error: metrics not available yet" message. @latchmihay pointer fixed it. Thank you.

[sreedharbukya]

I have similar problem. It says metrics not available for node or pod. It still reports

kubectl top nodes
error: metrics not available yet

Here is my configuration.

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: metrics-server
  namespace: kube-system
  labels:
    k8s-app: metrics-server
spec:
  selector:
    matchLabels:
      k8s-app: metrics-server
  template:
    metadata:
      name: metrics-server
      labels:
        k8s-app: metrics-server
    spec:
      serviceAccountName: metrics-server
      volumes:
      # mount in tmp so we can safely use from-scratch images and/or read-only containers
      - name: tmp-dir
        emptyDir: {}
      containers:
      - name: metrics-server
        image: k8s.gcr.io/metrics-server-amd64:v0.3.3
        args: [ "--kubelet-insecure-tls" ]
        imagePullPolicy: Always
        volumeMounts:
        - name: tmp-dir
          mountPath: /tmp

[sreedharbukya]

I got it working.

[regardfs]

@sreedharbukya I have the same error and try all the method above mentioned but still not work, may u post ur solution?

[sreedharbukya]

@regardfs, Please follow below instructions.

If you are using kops for creating the cluster.

kops edit cluster --name {cluster_name}

edit following part

    kubelet:
    anonymousAuth: false
    authenticationTokenWebhook: true
    authorizationMode: Webhook

you have run following commands

kops update cluster --yes
kops rolling-update cluster--yes

Watch out your logs.

This should resolve the issue.

Next is update your metrics-server deployments to following.

apiVersion: extensions/v1beta1 kind: Deployment metadata: annotations: deployment.kubernetes.io/revision: "" kubectl.kubernetes.io/last-applied-configuration: | {"apiVersion":"extensions/v1beta1","kind":"Deployment","metadata":{"annotations":{},"labels":{"k8s-app":"metrics-server"},"name":"metrics-server","namespace":"kube-system"},"spec":{"selector":{"matchLabels":{"k8s-app":"metrics-server"}},"template":{"metadata":{"labels":{"k8s-app":"metrics-server"},"name":"metrics-server"},"spec":{"containers":[{"image":"k8s.gcr.io/metrics-server-amd64:v0.3.3","imagePullPolicy":"Always","name":"metrics-server","volumeMounts":[{"mountPath":"/tmp","name":"tmp-dir"}]}],"serviceAccountName":"metrics-server","volumes":[{"emptyDir":{},"name":"tmp-dir"}]}}}} creationTimestamp: "2019-07-19T10:31:07Z" generation: 2 labels: k8s-app: metrics-server name: metrics-server namespace: kube-system resourceVersion: "8715" selfLink: /apis/extensions/v1beta1/namespaces/kube-system/deployments/metrics-server uid: 5145dcbb-aa10-11e9-bd85-06b142917002 spec: progressDeadlineSeconds: 2147483647 replicas: 1 revisionHistoryLimit: 10 selector: matchLabels: k8s-app: metrics-server strategy: rollingUpdate: maxSurge: 1 maxUnavailable: 1 type: RollingUpdate template: metadata: creationTimestamp: null labels: k8s-app: metrics-server name: metrics-server spec: containers: - command: - /metrics-server - --v=2 - --kubelet-preferred-address-types=InternalIP - --kubelet-insecure-tls=true image: k8s.gcr.io/metrics-server-amd64:v0.3.3 imagePullPolicy: Always name: metrics-server resources: {} terminationMessagePath: /dev/termination-log terminationMessagePolicy: File volumeMounts: - mountPath: /tmp name: tmp-dir dnsPolicy: ClusterFirst restartPolicy: Always schedulerName: default-scheduler securityContext: {} serviceAccount: metrics-server serviceAccountName: metrics-server terminationGracePeriodSeconds: 30 volumes: - emptyDir: {} name: tmp-dir status: availableReplicas: 1 conditions: - lastTransitionTime: "2019-07-19T10:31:07Z" lastUpdateTime: "2019-07-19T10:31:07Z" message: Deployment has minimum availability. reason: MinimumReplicasAvailable status: "True" type: Available observedGeneration: 2 readyReplicas: 1 replicas: 1 updatedReplicas: 1

[davydkov]

@sreedharbukya Thank you!

I have an AWS EKS Cluster and default installation of metrics server from Helm.
These two parameters solved the problem:

--kubelet-preferred-address-types=InternalIP
--kubelet-insecure-tls=true

[girol]

As pointed in the documentation, using --kubelet-insecure-tls=true is not recommended for production environments.

That said, how do we deploy metrics-server securely using tls?

[wajdi-datalvo]

I have the same error here "metrics not available yet" .. The pod logs is showing this message:

I0925 20:15:33.886593 1 serving.go:312] Generated self-signed cert (apiserver.local.config/certificates/apiserver.crt, apiserver.local.config/certificates/apiserver.key)
I0925 20:15:34.430030 1 secure_serving.go:116] Serving securely on [::]:443
E0925 20:15:46.541160 1 reststorage.go:135] unable to fetch node metrics for node "k8s-worker-1": no metrics known for node

[ledroide]

I'm facing the same problem for nodes, but not for pods. Kubelet Insecure TLS is set : no change.

$ kubectl top pod -n default
NAME                                        CPU(cores)   MEMORY(bytes)   
resource-consumer-5766d495c6-bm47z          15m           5Mi             
resource-consumer-5766d495c6-rtkzk           6m           5Mi             
unprivileged-hello-world-658f6f7f49-dq6js    0m           6Mi             

$ kubectl top node
error: metrics not available yet



md5-13d231eb47f48dae2e991b244869f505



```yaml
      - command:
        - /metrics-server
        - --logtostderr
        - --cert-dir=/tmp
        - --secure-port=8443
        - --kubelet-preferred-address-types=InternalIP
        - --kubelet-insecure-tls

[serathius]

no metrics known for node itself doesn't point to a problem. More details here: https://github.com/kubernetes-sigs/metrics-server/issues/349 for details.

@ledroide Could you provide metrics-server logs

[ledroide]

@serathius : Here are logs from metrics-server, filtered for matching "node" : https://gist.github.com/ledroide/0cbc6750cab7d6ae0371b88c97aee44e

Exemple of what I'm seeing in these logs :

E1105 13:30:06.940698       1 manager.go:111] unable to fully collect metrics: unable to fully scrape metrics from source kubelet_summary:kube-poc-compute3: unable to get CPU for container "sentinel" in pod webs/webs-sentinel-redis-slave-0 on node "10.150.233.53", discarding data: missing cpu usage metric
E1105 13:44:06.541969       1 manager.go:111] unable to fully collect metrics: [unable to fully scrape metrics from source kubelet_summary:kube-poc-compute4: unable to get CPU for container "sentinel" in pod webs/webs-sentinel-redis-slave-0 on node "10.150.233.54", discarding data: missing cpu usage metric, unable to fully scrape metrics from source kubelet_summary:kube-poc-compute3: unable to get CPU for container "metrics" in pod webs/webs-sentinel-redis-master-0 on node "10.150.233.53", discarding data: missing cpu usage metric]

[serathius]

Can you verify that kubelet is exposing cpu metrics?
kubectl get --raw /api/v1/nodes/kube-poc-compute3/proxy/stats/summary | jq '.node'

You can skip using | jq '.node', it just filters node data

[ledroide]

I have copied the output of the command kubectl get --raw /api/v1/nodes/kube-poc-compute3/proxy/stats/summary | jq '.node' at the end of the same gist.

Output looks good to me.

[serathius]

Looks like node exposes UsageNanoCores correctly and I don't see log that mentions broken node metrics (unable to get CPU for node) .

Can you verify that kubectl top node still fails?

[ledroide]

Of course, still failing :

$ kubectl top node
error: metrics not available yet

[serathius]

Last check before I would need to deepdive into code. Need to eliminate if the problem is with kubectl top and not the api.
Please run

kubectl get --raw /apis/metrics.k8s.io/v1beta1/nodes/kube-poc-compute3
kubectl version

Are you running heapster in cluster?

[ledroide]

I think we reach the problem here : the nodes list is empty

$ kubectl get --raw /apis/metrics.k8s.io/v1beta1/nodes/
{"kind":"NodeMetricsList","apiVersion":"metrics.k8s.io/v1beta1","metadata":{"selfLink":"/apis/metrics.k8s.io/v1beta1/nodes/"},"items":[]}

$ kubectl get --raw /apis/metrics.k8s.io/v1beta1/nodes/kube-poc-compute3
Error from server (NotFound): nodemetrics.metrics.k8s.io "kube-poc-compute3" not found



md5-a74ec8613827977ce0bea1cac086477e



$ kubectl version
Client Version: version.Info{Major:"1", Minor:"16", GitVersion:"v1.16.2", GitCommit:"c97fe5036ef3df2967d086711e6c0c405941e14b", GitTreeState:"clean", BuildDate:"2019-10-17T17:16:09Z", GoVersion:"go1.12.10", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"16", GitVersion:"v1.16.2", GitCommit:"c97fe5036ef3df2967d086711e6c0c405941e14b", GitTreeState:"clean", BuildDate:"2019-10-15T19:09:08Z", GoVersion:"go1.12.10", Compiler:"gc", Platform:"linux/amd64"}

$ kubectl api-versions | grep metrics
metrics.k8s.io/v1beta1

$ kubectl api-resources -o wide | egrep '^NAME|metrics'
NAME                              SHORTNAMES   APIGROUP                        NAMESPACED   KIND                             VERBS
nodes                                          metrics.k8s.io                  false        NodeMetrics                      [get list]
pods                                           metrics.k8s.io                  true         PodMetrics                       [get list]

[ledroide]

Solved.
I removed metrics-server and reinstalled it (using Kubespray).
It is ugly, but smoother or cleaner methods did not work at all.
Now it works properly. Thank you very much @serathius for you precious help :

$ kubectl get --raw /apis/metrics.k8s.io/v1beta1/nodes/kube-poc-compute3
{"kind":"NodeMetrics","apiVersion":"metrics.k8s.io/v1beta1","metadata":{"name":"kube-poc-compute3","selfLink":"/apis/metrics.k8s.io/v1beta1/nodes/kube-poc-compute3","creationTimestamp":"2019-11-12T09:42:21Z"},"timestamp":"2019-11-12T09:41:57Z","window":"30s","usage":{"cpu":"849755531n","memory":"8506004Ki"}}

[neerajjain92]

Solved by following steps :

1) Clone https://github.com/kubernetes-sigs/metrics-server
2) Disable the metrics-server addon for minikube in case it was enabled
minikube addons disable metrics-server
3) Deploy the latest metric-server
kubectl create -f deploy/1.8+/
4) Edit metric-server deployment to add the flags

# args:
# - --kubelet-insecure-tls
# - --kubelet-preferred-address-types=InternalIP,ExternalIP,Hostname
$ kubectl edit deploy -n kube-system metrics-server

6) Wait for 1 minute and boom metrics are available.

[oceaneLonneux]

Hello, I have the same problem...
I have deleted the metrics-server then re-installed it with the args mentioned by @neerajjain92 but I'm afraid nothing is happening.
1 desired | 1 updated | 1 total | 0 available | 1 unavailable

[illagrenan]

I had the same problem with Kubernetes 1.16 installed by kops on AWS. I found that a heapster is still installed on the server. When I removed the heapster (it was installed using helm), the command kubectl top nodes started to work.

[terencebor]

Thanks @latchmihay your solution works for me

[Guervyl]

The problem happened to me and none of the suggestion above worked because the problem wasn't metrics. The problem was because the 443 and 80 was busy. I had apache server running. If yo have any application running on those ports try to kill them before anything.

[ricpf]

@regardfs, Please follow below instructions.

If you are using kops for creating the cluster.

kops edit cluster --name {cluster_name}

edit following part

    kubelet:
    anonymousAuth: false
    authenticationTokenWebhook: true
    authorizationMode: Webhook

you have run following commands

kops update cluster --yes
kops rolling-update cluster--yes

Watch out your logs.

This should resolve the issue.

Next is update your metrics-server deployments to following.

kind: Deployment
metadata:
  annotations:
    deployment.kubernetes.io/revision: ""
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"extensions/v1beta1","kind":"Deployment","metadata":{"annotations":{},"labels":{"k8s-app":"metrics-server"},"name":"metrics-server","namespace":"kube-system"},"spec":{"selector":{"matchLabels":{"k8s-app":"metrics-server"}},"template":{"metadata":{"labels":{"k8s-app":"metrics-server"},"name":"metrics-server"},"spec":{"containers":[{"image":"k8s.gcr.io/metrics-server-amd64:v0.3.3","imagePullPolicy":"Always","name":"metrics-server","volumeMounts":[{"mountPath":"/tmp","name":"tmp-dir"}]}],"serviceAccountName":"metrics-server","volumes":[{"emptyDir":{},"name":"tmp-dir"}]}}}}
  creationTimestamp: "2019-07-19T10:31:07Z"
  generation: 2
  labels:
    k8s-app: metrics-server
  name: metrics-server
  namespace: kube-system
  resourceVersion: "8715"
  selfLink: /apis/extensions/v1beta1/namespaces/kube-system/deployments/metrics-server
  uid: 5145dcbb-aa10-11e9-bd85-06b142917002
spec:
  progressDeadlineSeconds: 2147483647
  replicas: 1
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      k8s-app: metrics-server
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
    type: RollingUpdate
  template:
    metadata:
      creationTimestamp: null
      labels:
        k8s-app: metrics-server
      name: metrics-server
    spec:
      containers:
      - command:
        - /metrics-server
        - --v=2
        - --kubelet-preferred-address-types=InternalIP
        - --kubelet-insecure-tls=true
        image: k8s.gcr.io/metrics-server-amd64:v0.3.3
        imagePullPolicy: Always
        name: metrics-server
        resources: {}
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
        volumeMounts:
        - mountPath: /tmp
          name: tmp-dir
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      serviceAccount: metrics-server
      serviceAccountName: metrics-server
      terminationGracePeriodSeconds: 30
      volumes:
      - emptyDir: {}
        name: tmp-dir
status:
  availableReplicas: 1
  conditions:
  - lastTransitionTime: "2019-07-19T10:31:07Z"
    lastUpdateTime: "2019-07-19T10:31:07Z"
    message: Deployment has minimum availability.
    reason: MinimumReplicasAvailable
    status: "True"
    type: Available
  observedGeneration: 2
  readyReplicas: 1
  replicas: 1
  updatedReplicas: 1```

didn't had to update the deployment tough. --insecure tls and kops update cluster did the trick for me thank you

[mbabauer]

I'm facing the same problem for nodes, but not for pods. Kubelet Insecure TLS is set : no change.

$ kubectl top pod -n default
NAME                                        CPU(cores)   MEMORY(bytes)   
resource-consumer-5766d495c6-bm47z          15m           5Mi             
resource-consumer-5766d495c6-rtkzk           6m           5Mi             
unprivileged-hello-world-658f6f7f49-dq6js    0m           6Mi             

$ kubectl top node
error: metrics not available yet

$ kubectl get deploy metrics-server -o yaml | grep -B6 insecure

      - command:
        - /metrics-server
        - --logtostderr
        - --cert-dir=/tmp
        - --secure-port=8443
        - --kubelet-preferred-address-types=InternalIP
        - --kubelet-insecure-tls

This setup worked for me, except I left the secure-port at 4443, which was how it was installed via the deployment script. Now I can get top on both node and pods.

[mayquanxi]

its probably the self signed cert on your cluster.
Try enabling --kubelet-insecure-tls

As in

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: metrics-server

      containers:
      - name: metrics-server
        image: k8s.gcr.io/metrics-server-amd64:v0.3.3
        imagePullPolicy: Always
        args: [ "--kubelet-insecure-tls" ]
        volumeMounts:
        - name: tmp-dir
          mountPath: /tmp

I had the same problem, and I had fix it by add args as you, thankyou

[regardfs]

@sreedharbukya thanks a lot!!!

[kevinsingapore]

Of course, still failing :

$ kubectl top node
error: metrics not available yet

do u have solve it?

[serathius]

@kevinsingapore

I recommend to create separate issue, as in this issue we didn't even manage to fix original problem that reporter had.

error: metrics not available yet

Signifies that there is a problem with metrics-server, but says nothing about what's the problem. Meaning that this issue isn't really helpful for others (e.g. @ledroide just recreated MS).

Could try some suggestions from this issue, still it would be better to tackle each setup separately.

[rpolnx]

I had a similiar problem on minikube running on windows 10 home. Your tutorial helped me.
Thank you, @neerajjain92 !

[Rock981119]

Solved by following steps :

1. Clone https://github.com/kubernetes-sigs/metrics-server
2. Disable the metrics-server addon for minikube in case it was enabled
   minikube addons disable metrics-server
3. Deploy the latest metric-server
   kubectl create -f deploy/1.8+/
4. Edit metric-server deployment to add the flags

# args:
# - --kubelet-insecure-tls
# - --kubelet-preferred-address-types=InternalIP,ExternalIP,Hostname
$ kubectl edit deploy -n kube-system metrics-server

1. Wait for 1 minute and boom metrics are available.

It worked, Thx.

[zhou880220]

I had the same problem，this is my config:
[root@k8s-master data]# kubectl get pods --all-namespaces
NAMESPACE NAME READY STATUS RESTARTS AGE
default whoami-7c8d684c55-vmc9x 1/1 Running 1 20h
kube-system coredns-6d56c8448f-g98fb 1/1 Running 3 2d
kube-system coredns-6d56c8448f-nhmk7 1/1 Running 3 2d
kube-system etcd-k8s-master 1/1 Running 5 2d
kube-system heapster-7fcb4d8889-nm2ln 1/1 Running 0 15h
kube-system kube-apiserver-k8s-master 1/1 Running 5 2d
kube-system kube-controller-manager-k8s-master 1/1 Running 8 47h
kube-system kube-flannel-ds-662nb 1/1 Running 1 21h
kube-system kube-flannel-ds-8njs6 1/1 Running 0 21h
kube-system kube-flannel-ds-d2z4w 1/1 Running 3 21h
kube-system kube-proxy-2blhp 1/1 Running 3 46h
kube-system kube-proxy-6hjpl 1/1 Running 1 47h
kube-system kube-proxy-zsvrh 1/1 Running 4 2d
kube-system kube-scheduler-k8s-master 1/1 Running 8 47h
kube-system monitoring-grafana-6d69444f6-fr9bp 1/1 Running 2 41h
kube-system monitoring-influxdb-64596c7b6d-jzskg 1/1 Running 1 41h
kubernetes-dashboard dashboard-metrics-scraper-7b59f7d4df-4mkbb 1/1 Running 0 16h
kubernetes-dashboard kubernetes-dashboard-665f4c5ff-9v8kf 1/1 Running 0 16h

[root@k8s-master k8s]# kubectl version
Client Version: version.Info{Major:"1", Minor:"19", GitVersion:"v1.19.2", GitCommit:"f5743093fd1c663cb0cbc89748f730662345d44d", GitTreeState:"clean", BuildDate:"2020-09-16T13:41:02Z", GoVersion:"go1.15", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"19", GitVersion:"v1.19.2", GitCommit:"f5743093fd1c663cb0cbc89748f730662345d44d", GitTreeState:"clean", BuildDate:"2020-09-16T13:32:58Z", GoVersion:"go1.15", Compiler:"gc", Platform:"linux/amd64"}

[root@k8s-master k8s]# kubectl top po
error: Metrics API not available

heapster logs:

W0924 01:44:25.002300 1 manager.go:152] Failed to get all responses in time (got 0/3)
E0924 01:45:05.001192 1 manager.go:101] Error in scraping containers from kubelet:192.168.28.143:10250: failed to get all container stats from Kubelet URL "https://192.168.28.143:10250/stats/container/": "https://192.168.28.143:10250/stats/container/" not found
E0924 01:45:05.007179 1 manager.go:101] Error in scraping containers from kubelet:192.168.28.141:10250: failed to get all container stats from Kubelet URL "https://192.168.28.141:10250/stats/container/": "https://192.168.28.141:10250/stats/container/" not found
E0924 01:45:05.015769 1 manager.go:101] Error in scraping containers from kubelet:192.168.28.142:10250: failed to get all container stats from Kubelet URL "https://192.168.28.142:10250/stats/container/":

[root@k8s-master data]# netstat -nultp
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name
tcp 0 0 0.0.0.0:31303 0.0.0.0:* LISTEN 126245/kube-proxy
tcp 0 0 127.0.0.1:10248 0.0.0.0:* LISTEN 1059/kubelet
tcp 0 0 127.0.0.1:10249 0.0.0.0:* LISTEN 126245/kube-proxy
tcp 0 0 192.168.28.141:2379 0.0.0.0:* LISTEN 126075/etcd
tcp 0 0 127.0.0.1:2379 0.0.0.0:* LISTEN 126075/etcd
tcp 0 0 192.168.28.141:2380 0.0.0.0:* LISTEN 126075/etcd
tcp 0 0 127.0.0.1:2381 0.0.0.0:* LISTEN 126075/etcd
tcp 0 0 127.0.0.1:10257 0.0.0.0:* LISTEN 15888/kube-controll
tcp 0 0 127.0.0.1:10259 0.0.0.0:* LISTEN 15947/kube-schedule
tcp 0 0 0.0.0.0:30101 0.0.0.0:* LISTEN 126245/kube-proxy
tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN 1056/sshd
tcp 0 0 127.0.0.1:41111 0.0.0.0:* LISTEN 1059/kubelet
tcp 0 0 0.0.0.0:31001 0.0.0.0:* LISTEN 126245/kube-proxy
tcp 0 0 127.0.0.1:25 0.0.0.0:* LISTEN 1206/master
tcp 0 0 0.0.0.0:30108 0.0.0.0:* LISTEN 126245/kube-proxy
tcp6 0 0 :::10250 :::* LISTEN 1059/kubelet
tcp6 0 0 :::10251 :::* LISTEN 15947/kube-schedule
tcp6 0 0 :::6443 :::* LISTEN 125805/kube-apiserv
tcp6 0 0 :::10252 :::* LISTEN 15888/kube-controll
tcp6 0 0 :::10256 :::* LISTEN 126245/kube-proxy
tcp6 0 0 :::22 :::* LISTEN 1056/sshd
tcp6 0 0 ::1:25 :::* LISTEN 1206/master
udp 0 0 10.244.0.1:123 0.0.0.0:* 838/ntpd
udp 0 0 10.244.0.0:123 0.0.0.0:* 838/ntpd
udp 0 0 172.17.0.1:123 0.0.0.0:* 838/ntpd
udp 0 0 192.168.28.141:123 0.0.0.0:* 838/ntpd
udp 0 0 127.0.0.1:123 0.0.0.0:* 838/ntpd
udp 0 0 0.0.0.0:123 0.0.0.0:* 838/ntpd
udp 0 0 0.0.0.0:8472 0.0.0.0:* -
udp6 0 0 fe80::98d8:4aff:fef:123 :::* 838/ntpd
udp6 0 0 fe80::888f:dfff:feb:123 :::* 838/ntpd
udp6 0 0 fe80::886a:32ff:fed:123 :::* 838/ntpd
udp6 0 0 fe80::c89e:ecff:fed:123 :::* 838/ntpd
udp6 0 0 fe80::20c:29ff:febc:123 :::* 838/ntpd
udp6 0 0 ::1:123 :::* 838/ntpd
udp6 0 0 :::123 :::* 838/ntpd

is namespace question?

[debu99]

if you use bitnami metrics-server, enable this even k8s > 1.18
apiService:
create: true

[Type1J]

I only needed --kubelet-preferred-address-types=InternalIP to fix it. With the Terraform Helm Provider the value must be an array, which uses { and } and values are separated by ,.

resource "helm_release" "metrics-server" {
  create_namespace = true
  namespace        = "platform"
  name             = "metrics-server"
  repository       = "https://charts.helm.sh/stable"
  chart            = "metrics-server"

  set {
    name  = "args"
    value = "{\"--kubelet-preferred-address-types=InternalIP\"}"
  }
}

[kingdonb]

Edit: this report definitely does not belong here, sorry for contributing to the noise

I was able to resolve it by switching from the bitnami helm chart for metrics server, to kustomized deploy of metrics-server from this repo, with very similar to "test" kustomize manifests. Thank you for providing this.

I am on kubeadm v1.20.2 with a matching kubectl and I had to set apiService.create: true as @debu99 suggested.

This is in conflict with the docs in values.yaml which are maybe incorrect

## API service parameters
##
apiService:
  ## Specifies whether the v1beta1.metrics.k8s.io API service should be created
  ## This should not be necessary in k8s version >= 1.8, but depends on vendors and cloud providers.
  ##
  create: false

Else I ran into error: Metrics API not available

While that's not the subject of this issue report, this issue is one of the top results for "error: Metrics API not available" and it helped me, so I am highlighting it here.

~I am not sure if this information belongs here, I'm using bitnami/metrics-server which has a repo of its own in https://github.com/bitnami/charts/ and so~ I guess the new issue report should go there, if there is a problem.