Skip to main content
Amazon EKS 0.65.5Last updated in version 0.55.2

EKS Scripts Module

View SourceRelease Notes

This folder contains helper scripts for running an EKS Cluster, including:

  • map-ec2-tags-to-node-labels: This script will pull the associated tags on the EC2 instance and convert it into a format that can be passed into the EKS node bootstrap script to set the node labels as the tags.

Installing the helpers

You can install the helpers using the Gruntwork Installer:

gruntwork-install --module-name "eks-scripts" --repo "" --tag "0.2.0"

For an example, see the Packer template under examples/eks-cluster-with-supporting-services/packer/build.json.

Using the map-ec2-tags-to-node-labels helper

The map-ec2-tags-to-node-labels script will take EC2 instance tags and format them for use as Kubernetes Node Labels, which play an integral role in having control over assigning which nodes Pods should be scheduled on.

This script has the following prerequisites:

  1. It must be run on an EC2 instance.
  2. The EC2 instance must be running an Amazon EKS-optimized AMI.
  3. The EC2 instance must have the AWS CLI installed.
  4. The EC2 instance must have Python 3 installed.

When you run the script, the script will output the node label argument to stdout. You can then pass this output to the bootstrap script:

/etc/eks/ \
--apiserver-endpoint "$EKS_ENDPOINT" \
--b64-cluster-ca "$EKS_CERTIFICATE_AUTHORITY" \
--kubelet-extra-args "--node-labels=\"$NODE_LABELS\"" \

This script will read in the EC2 tags from the metadata API and convert it to a string of comma separated key=value pairs, outputting the result to stdout.

Note: unlike EC2 tags, Kubernetes node labels have the following restrictions on BOTH keys and labels:

  • Must be 63 characters or less
  • Begin and end with an alphanumeric character ([a-zA-Z0-9])
  • Only contain dashes (-), underscores (_), dots (.), and alphanumeric characters ([a-zA-Z0-9])

As such, this script will convert the EC2 tags to fit the scheme above. Notably, this script will:

  • Take all unsupported characters and replace them with dashes (-). E.g if you have a EC2 tag with key foo@bar, this will be converted to foo-bar.
  • Truncate keys and values to 63 characters.
  • Namespace the labels with the prefix E.g if you have a EC2 tag with key foo, this will be registered under the key in Kubernetes.

Finally, this script will log all conversions to stderr so it can be traced and referenced.


You can restrict the tags that get converted by prefix using the --tag-prefix option. That is, if your EC2 instance had the following tags:

Name: "my-asg-ec2-instance"
AutoScalingGroup: "my-asg" "my-group"

and you specified as your tag prefix (map-ec2-tags-to-node-labels --tag-prefix ''), then only the tag "my-group" will get converted to a node label.

IAM Role

In order for the script to be able to successfully retrieve the tags for EC2 instance, the instances need to be associated with an IAM profile that grants it access to retrieve the EC2 tags on the instance. If you launch the workers using the eks-cluster-workers module, this is automatically attached to the worker IAM role.

Due to limitations in how python imports scripts, this module includes a symlink that maps to the map-ec2-tags-to-node-labels script so that it can be imported in the unit tests.