I've been struggling with my EKS setup all day. I created my cluster successfully, but when I try to create a Node Group, it gets stuck in the 'creating' state and the instances aren't able to join. The EC2 instances are running, and my IAM roles include the AmazonEKS_CNI_Policy, AmazonEC2ContainerRegistryReadOnly, and AmazonEKSWorkerNodePolicy. I'm also using the same add-ons for my cluster: Amazon VPC CNI, CoreDNS, and kube-proxy. Everything's in the same VPC, and I've followed a tutorial to the letter. However, it's just not working, and I've tried deleting everything and starting over. I'm really at a loss here. Some AI tool suggested that it might be a missing ConfigMap, but I didn't see anything about that in the video. Any thoughts or advice would be greatly appreciated!
2 Answers
First off, make sure your node role has the right permissions within the cluster. Check if your security groups allow the nodes to communicate with the cluster API. Also, reviewing the logs on the nodes can really help pinpoint the issue. Sometimes, it’s just one little thing that's causing the whole hang-up!
It sounds like your issue could be due to a few common culprits. First, verify that you’ve added your node role to the aws-auth ConfigMap. You can check it with 'kubectl -n kube-system get configmap aws-auth -o yaml'. If it’s not there, you’ll need to add it for the kubelet to join the cluster. Also, make sure your subnets are properly tagged and have the right outbound internet access, and that your security groups allow the necessary traffic.
Thanks for the tip! I’ll look into the aws-auth configuration and check the tags on my subnets.
Yeah, I've checked and they do have the right permissions. The logs suggest that the issue might be tied to 'aws-auth'.