FAQ

Yes! From version 0.40.0 you can run eksctl against any cluster, whether it was created by eksctl or not. Find out more here.

From the point of view of eksctl, nodegroups are immutable. This means that once created the only thing eksctl can do is scale the nodegroup up or down.

To change the instance type, create a new nodegroup with the desired instance type, then drain it so that the workloads move to the new one. After that step is complete you can delete the old nodegroup

First you'll need the name of the Cloudformation stack that manages the nodegroup:

$ eksctl utils describe-stacks --region=us-west-2 --cluster NAME

You can execute the following to get the userdata. Note the final line which decodes from base64 and decompresses the gzipped data.

NG_STACK=eksctl-scrumptious-monster-1595247364-nodegroup-ng-29b8862f # your stack here
LAUNCH_TEMPLATE_ID=$(aws cloudformation describe-stack-resources --stack-name $NG_STACK \
| jq -r '.StackResources | map(select(.LogicalResourceId == "NodeGroupLaunchTemplate")
| .PhysicalResourceId)[0]')
aws ec2 describe-launch-template-versions --launch-template-id $LAUNCH_TEMPLATE_ID \
| jq -r '.LaunchTemplateVersions[0].LaunchTemplateData.UserData' \
| base64 -d | gunzip

We recommend using the AWS Load Balancer Controller. Documentation on how to deploy the controller to your cluster, as well as how to migrate from the old ALB Ingress Controller, can be found here.

For the Nginx Ingress Controller, setup would be the same as any on other Kubernetes cluster.

Set the environment variable KUBECONFIG_USE_SYSTEM_CA to make kubeconfig respect the system certificate authorities.