Metrics Server

What is Metrics Server

KubeSphere supports Horizontal Pod Autoscalers (HPA) for Deployments. In KubeSphere, the Metrics Server controls whether the HPA is enabled. You use an HPA object to autoscale a Deployment based on different types of metrics, such as CPU and memory utilization, as well as the minimum and maximum number of replicas. In this way, an HPA helps to make sure your application runs smoothly and consistently in different situations.

Enable the Metrics Server before Installation

Installing on Linux

When you implement multi-node installation of KubeSphere on Linux, you need to create a configuration file, which lists all KubeSphere components.

  1. In the tutorial of Installing KubeSphere on Linux, you create a default file config-sample.yaml. Modify the file by executing the following command:

    1. vi config-sample.yaml

    Note

    If you adopt All-in-One Installation, you do not need to create a config-sample.yaml file as you can create a cluster directly. Generally, the all-in-one mode is for users who are new to KubeSphere and look to get familiar with the system. If you want to enable the Metrics Server in this mode (e.g. for testing purposes), refer to the following section to see how the Metrics Server can be installed after installation.

  2. In this file, navigate to metrics_server and change false to true for enabled. Save the file after you finish.

    1. metrics_server:
    2.   enabled: true # Change "false" to "true"

  3. Create a cluster using the configuration file:

    1. ./kk create cluster -f config-sample.yaml

Installing on Kubernetes

As you install KubeSphere on Kubernetes, you can enable the Metrics Server first in the cluster-configuration.yaml file.

  1. Download the file cluster-configuration.yaml and edit it.

    1. vi cluster-configuration.yaml

  2. In this local cluster-configuration.yaml file, navigate to metrics_server and enable it by changing false to true for enabled. Save the file after you finish.

    1. metrics_server:
    2.   enabled: true # Change "false" to "true"

  3. Execute the following commands to start installation:

    1. kubectl apply -f https://github.com/kubesphere/ks-installer/releases/download/v3.1.0/kubesphere-installer.yaml
    3. kubectl apply -f cluster-configuration.yaml

    Note

    If you install KubeSphere on some cloud hosted Kubernetes engines, it is probable that the Metrics Server is already installed in your environment. In this case, it is not recommended that you enable it in cluster-configuration.yaml as it may cause conflicts during installation.

Enable the Metrics Server after Installation

  1. Log in to the console as admin. Click Platform in the top-left corner and select Cluster Management.

    clusters-management

  2. Click CRDs and enter clusterconfiguration in the search bar. Click the result to view its detail page.

    Info

    A Custom Resource Definition (CRD) allows users to create a new type of resources without adding another API server. They can use these resources like any other native Kubernetes objects.

  3. In Resource List, click the three dots on the right of ks-installer and select Edit YAML.

    edit-yaml

  4. In this YAML file, navigate to metrics_server and change false to true for enabled. After you finish, click Update in the bottom-right corner to save the configuration.

    1. metrics_server:
    2.   enabled: true # Change "false" to "true"

  5. You can use the web kubectl to check the installation process by executing the following command:

    1. kubectl logs -n kubesphere-system $(kubectl get pod -n kubesphere-system -l app=ks-install -o jsonpath='{.items[0].metadata.name}') -f

    Tip

    You can find the web kubectl tool by clicking the hammer icon in the bottom-right corner of the console.

Verify the Installation of the Component

Execute the following command to verify that the Pod of Metrics Server is up and running.

  1. kubectl get pod -n kube-system

If the Metrics Server is successfully installed, your cluster may return the following output (metrics-server-5ddd98b7f9-jjdln):

  1. NAME                                           READY   STATUS    RESTARTS   AGE
  2. calico-kube-controllers-59d85c5c84-m4blq       1/1     Running   0          28m
  3. calico-node-nqzcp                              1/1     Running   0          28m
  4. coredns-74d59cc5c6-8djtt                       1/1     Running   0          28m
  5. coredns-74d59cc5c6-jv65g                       1/1     Running   0          28m
  6. kube-apiserver-master                          1/1     Running   0          29m
  7. kube-controller-manager-master                 1/1     Running   0          29m
  8. kube-proxy-6qjz7                               1/1     Running   0          28m
  9. kube-scheduler-master                          1/1     Running   0          29m
  10. metrics-server-5ddd98b7f9-jjdln                1/1     Running   0          7m17s
  11. nodelocaldns-8wbfm                             1/1     Running   0          28m
  12. openebs-localpv-provisioner-84956ddb89-dxbnx   1/1     Running   0          28m
  13. openebs-ndm-operator-6896cbf7b8-xwcth          1/1     Running   1          28m
  14. openebs-ndm-pf47z                              1/1     Running   0          28m
  15. snapshot-controller-0                          1/1     Running   0          22m