Import an AWS EKS Cluster

This tutorial demonstrates how to import an AWS EKS cluster through the direct connection method. If you want to use the agent connection method, refer to Agent Connection.

Prerequisites

• You have a Kubernetes cluster with Kube AI Hub installed, and prepared this cluster as the host cluster. For more information about how to prepare a host cluster, refer to Prepare a host cluster.
• You have an EKS cluster to be used as the member cluster.

Import an EKS Cluster

Step 1: Deploy Kube AI Hub on your EKS cluster

You need to deploy Kube AI Hub on your EKS cluster first. For more information about how to deploy Kube AI Hub on EKS, refer to Deploy Kube AI Hub on AWS EKS.

Step 2: Prepare the EKS member cluster

1. In order to manage the member cluster from the host cluster, you need to make jwtSecret the same between them. Therefore, get it first by executing the following command on your host cluster.

   kubectl -n kubesphere-system get cm kubesphere-config -o yaml | grep -v "apiVersion" | grep jwtSecret
   

   The output is similar to the following:

   jwtSecret: "QVguGh7qnURywHn2od9IiOX6X8f8wK8g"
   

2. Log in to the Kube AI Hub console of the EKS cluster as admin. Click Platform in the upper-left corner and then select Cluster Management.

3. Go to CRDs, enter ClusterConfiguration in the search bar, and then press Enter on your keyboard. Click ClusterConfiguration to go to its detail page.

4. Click on the right and then select Edit YAML to edit ks-installer.

5. In the YAML file of ks-installer, change the value of jwtSecret to the corresponding value shown above and set the value of clusterRole to member. Click Update to save your changes.

   authentication:
     jwtSecret: QVguGh7qnURywHn2od9IiOX6X8f8wK8g
   

   multicluster:
     clusterRole: member
   

   Note

   Make sure you use the value of your own jwtSecret. You need to wait for a while so that the changes can take effect.

Step 3: Create a new kubeconfig file

1. Amazon EKS doesn’t provide a built-in kubeconfig file as a standard kubeadm cluster does. Nevertheless, you can create a kubeconfig file by referring to this document. The generated kubeconfig file will be like the following:

   apiVersion: v1
   clusters:
   - cluster:
       server: <endpoint-url>
       certificate-authority-data: <base64-encoded-ca-cert>
     name: kubernetes
   contexts:
   - context:
       cluster: kubernetes
       user: aws
     name: aws
   current-context: aws
   kind: Config
   preferences: {}
   users:
   - name: aws
     user:
       exec:
         apiVersion: client.authentication.k8s.io/v1alpha1
         command: aws
         args:
           - "eks"
           - "get-token"
           - "--cluster-name"
           - "<cluster-name>"
           # - "--role"
           # - "<role-arn>"
         # env:
           # - name: AWS_PROFILE
           #   value: "<aws-profile>"
   

   However, this automatically generated kubeconfig file requires the command aws (aws CLI tools) to be installed on every computer that wants to use this kubeconfig.

2. Run the following commands on your local computer to get the token of the ServiceAccount kubesphere created by Kube AI Hub. It has the cluster admin access to the cluster and will be used as the new kubeconfig token.

   TOKEN=$(kubectl -n kubesphere-system get secret $(kubectl -n kubesphere-system get sa kubesphere -o jsonpath='{.secrets[0].name}') -o jsonpath='{.data.token}' | base64 -d)
   kubectl config set-credentials kubesphere --token=${TOKEN}
   kubectl config set-context --current --user=kubesphere
   

3. Retrieve the new kubeconfig file by running the following command:

   cat ~/.kube/config
   

   The output is similar to the following and you can see that a new user kubesphere is inserted and set as the current-context user:

   apiVersion: v1
   clusters:
   - cluster:
       certificate-authority-data: LS0tLS1CRUdJTiBDRVJUSUZ...S0tLQo=
       server: https://*.sk1.cn-north-1.eks.amazonaws.com.cn
     name: arn:aws-cn:eks:cn-north-1:660450875567:cluster/EKS-LUSLVMT6
   contexts:
   - context:
       cluster: arn:aws-cn:eks:cn-north-1:660450875567:cluster/EKS-LUSLVMT6
       user: kubesphere
     name: arn:aws-cn:eks:cn-north-1:660450875567:cluster/EKS-LUSLVMT6
   current-context: arn:aws-cn:eks:cn-north-1:660450875567:cluster/EKS-LUSLVMT6
   kind: Config
   preferences: {}
   users:
   - name: arn:aws-cn:eks:cn-north-1:660450875567:cluster/EKS-LUSLVMT6
     user:
       exec:
         apiVersion: client.authentication.k8s.io/v1alpha1
         args:
         - --region
         - cn-north-1
         - eks
         - get-token
         - --cluster-name
         - EKS-LUSLVMT6
         command: aws
         env: null
   - name: kubesphere
     user:
       token: eyJhbGciOiJSUzI1NiIsImtpZCI6ImlCRHF4SlE5a0JFNDlSM2xKWnY1Vkt5NTJrcDNqRS1Ta25IYkg1akhNRmsifQ.eyJpc3M................9KQtFULW544G-FBwURd6ArjgQ3Ay6NHYWZe3gWCHLmag9gF-hnzxequ7oN0LiJrA-al1qGeQv-8eiOFqX3RPCQgbybmix8qw5U6f-Rwvb47-xA
   

   You can run the following command to check that the new kubeconfig does have access to the EKS cluster.

   kubectl get nodes
   

   The output is simialr to this:

   NAME                                        STATUS   ROLES    AGE   VERSION
   ip-10-0-47-38.cn-north-1.compute.internal   Ready    <none>   11h   v1.18.8-eks-7c9bda
   ip-10-0-8-148.cn-north-1.compute.internal   Ready    <none>   78m   v1.18.8-eks-7c9bda
   

Step 4: Import the EKS member cluster

1. Log in to the Kube AI Hub console on your host cluster as admin. Click Platform in the upper-left corner and then select Cluster Management. On the Cluster Management page, click Add Cluster.

2. Enter the basic information based on your needs and click Next.

3. In Connection Method, select Direct connection. Fill in the new kubeconfig file of the EKS member cluster and then click Create.

4. Wait for cluster initialization to finish.