Cannot create namespaces in AWS Elastic Kubernetes Service

I am trying to host an application in AWS Elastic Kubernetes Service(EKS). I have configured the EKS cluster using the AWS Console. Configured the Node Group and added a Node to the EKS Cluster and everything is working fine.

In order to connect to the cluster, I had spin up an EC2 instance (Centos7) and configured the following:

1. Installed docker, kubeadm, kubelet and kubectl.
2. Installed and configured AWS Cli V2.

To authenticate to the EKS Cluster, I had attached an IAM role to the EC2 Instance having the following AWS managed policies:

1. AmazonEKSClusterPolicy
2. AmazonEKSWorkerNodePolicy
3. AmazonEC2ContainerRegistryReadOnly
4. AmazonEKS_CNI_Policy
5. AmazonElasticContainerRegistryPublicReadOnly
6. EC2InstanceProfileForImageBuilderECRContainerBuilds
7. AmazonElasticContainerRegistryPublicFullAccess
8. AWSAppRunnerServicePolicyForECRAccess
9. AmazonElasticContainerRegistryPublicPowerUser
10. SecretsManagerReadWrite

After this, I ran the following commands to connect to the EKS Cluster:
1. aws sts get-caller-identity
2. aws eks update-kubeconfig --name eks-cluster --region ap-south-1

When I ran kubectl cluster-info and kubectl get nodes, I got the following:

However, when I try to run kubectl get namespaces I am getting the following error:

I am getting the same kind of error when I try to create Namespaces in the EKS cluster. Not sure what I'm missing here.

Error from server (Forbidden): error when creating "namespace.yml": namespaces is forbidden: User "system:node:ip-172-31-43-129.ap-south-1.compute.internal" cannot create resource "namespaces" in API group "" at the cluster scope

As an alternative, I tried to create a user with Administrative permission in IAM. Created AWS_ACCESS_KEY and AWS_SECRET_KEY_ID. Used aws configure to configure credentials within the EC2 Instance.

Ran the following commands:
1. aws sts get-caller-identity
2. aws eks update-kubeconfig --name eks-cluster --region ap-south-1
3. aws eks update-kubeconfig --name eks-cluster --region ap-south-1 --role-arn arn:aws:iam::XXXXXXXXXXXX:role/EKS-Cluster-Role

After running kubectl cluster-info --kubeconfig /home/centos/.kube/config, I got the following error:

An error occurred (AccessDenied) when calling the AssumeRole operation: User: arn:aws:iam::XXXXXXXXXXXX:user/XXXXX is not authorized to perform: sts:AssumeRole on resource: arn:aws:iam::XXXXXXXXXXXX:role/EKS-Cluster-Role

Does anyone know how to resolve this issue??

CodePudding user response：

Check your cluster role binding or user access to EKS cluster

---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: eks-console-dashboard-full-access-clusterrole
rules:
- apiGroups:
  - ""
  resources:
  - nodes
  - namespaces
  - pods
  verbs:
  - get
  - list
- apiGroups:
  - apps
  resources:
  - deployments
  - daemonsets
  - statefulsets
  - replicasets
  verbs:
  - get
  - list
- apiGroups:
  - batch
  resources:
  - jobs
  verbs:
  - get
  - list
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: eks-console-dashboard-full-access-binding
subjects:
- kind: Group
  name: eks-console-dashboard-full-access-group
  apiGroup: rbac.authorization.k8s.io
roleRef:
  kind: ClusterRole
  name: eks-console-dashboard-full-access-clusterrole
  apiGroup: rbac.authorization.k8s.io

Check the config map inside the cluster having proper user IAM mapping

kubectl get configmap aws-auth -n kube-system -o yaml

CodePudding user response：

I found the solution to this issue. Earlier I created the AWS EKS Cluster using the root user of my AWS account (Think that could have been the initial problem). This time, I created the EKS Cluster using an IAM user with Administrator privileges, created a Node Group in the cluster and added a worker node.

Next, I configured AWS Cli using the AWS_ACCESS_KEY_ID and AWS_SECRET_KEY_ID of the IAM user within the Centos EC2 instance.

Just like before, I logged into the Centos EC2 instance and ran the following commands:

1. aws sts get-caller-identity
2. aws eks describe-cluster --name eks-cluster --region ap-south-1 --query cluster.status (To check the status of the Cluster)
3. aws eks update-kubeconfig --name eks-cluster --region ap-south-1

After this, when I ran kubectl get pods, kubectl get namespaces, There were no more issues. In fact, I was able to deploy to the cluster as well.

So, as a conclusion, I think we should not be creating an EKS cluster using the root user in AWS.

Thanks!