Skip to content

nodeadm

Initializes a node in an EKS cluster.

Usage

To initialize a node:

nodeadm init

Note This happens automatically, via a systemd service, on AL2023-based EKS AMI's.

Configuration

nodeadm uses a YAML configuration schema that will look familiar to Kubernetes users.

This is an example of the minimum required parameters:

---
apiVersion: node.eks.aws/v1alpha1
kind: NodeConfig
spec:
  cluster:
    name: my-cluster
    apiServerEndpoint: https://example.com
    certificateAuthority: Y2VydGlmaWNhdGVBdXRob3JpdHk=
    cidr: 10.100.0.0/16

You'll typically provide this configuration in your EC2 instance's user data, either as-is or embedded within a MIME multi-part document:

MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="BOUNDARY"

--BOUNDARY
Content-Type: application/node.eks.aws

---
apiVersion: node.eks.aws/v1alpha1
kind: NodeConfig
spec: ...

--BOUNDARY--

The source for the configuration object can be specified with the --config-source flag and follows a URI format. The default is imds://user-data, which pulls from EC2 instance userdata, but you may provide a file path with file://....

The nodeadm-run.service systemd unit is configured with two config sources: imds://user-data and file:///etc/eks/nodeadm.d/. Any NodeConfig files (.yaml, .yml, or .json) placed under /etc/eks/nodeadm.d/ will be merged with the config supplied via user data, and take precedence over it. See the examples for details.

The API reference documentation contains the details of the configuration types.