EKS Cluster Configuration Reference

Account Access

Complete each drop-down and field using the information obtained for your IAM policy.

Service Role

Choose a service role.

Secrets Encryption

Optional: To encrypt secrets, select or enter a key created in AWS Key Management Service (KMS)

API Server Endpoint Access

Configuring Public/Private API access is an advanced use case. For details, refer to the EKS cluster endpoint access control documentation.

Private-only API Endpoints

If you enable private and disable public API endpoint access when creating a cluster, then there is an extra step you must take in order for Rancher to connect to the cluster successfully. In this case, a pop-up will be displayed with a command that you will run on the cluster to register it with Rancher. Once the cluster is provisioned, you can run the displayed command anywhere you can connect to the cluster’s Kubernetes API.

There are two ways to avoid this extra manual step:

• You can create the cluster with both private and public API endpoint access on cluster creation. You can disable public access after the cluster is created and in an active state and Rancher will continue to communicate with the EKS cluster.
• You can ensure that Rancher shares a subnet with the EKS cluster. Then security groups can be used to enable Rancher to communicate with the cluster’s API endpoint. In this case, the command to register the cluster is not needed, and Rancher will be able to communicate with your cluster. For more information on configuring security groups, refer to the security groups documentation.

Public Access Endpoints

Optionally limit access to the public endpoint via explicit CIDR blocks.

If you limit access to specific CIDR blocks, then it is recommended that you also enable the private access to avoid losing network communication to the cluster.

One of the following is required to enable private access:

• Rancher’s IP must be part of an allowed CIDR block
• Private access should be enabled, and Rancher must share a subnet with the cluster and have network access to the cluster, which can be configured with a security group

For more information about public and private access to the cluster endpoint, refer to the Amazon EKS documentation.

Subnet

For more information, refer to the AWS documentation for Cluster VPC Considerations. Follow one of the sets of instructions below based on your selection from the previous step.

• What Is Amazon VPC?
• VPCs and Subnets

Security Group

Amazon Documentation:

• Cluster Security Group Considerations
• Security Groups for Your VPC
• Create a Security Group

Logging

Configure control plane logs to send to Amazon CloudWatch. You are charged the standard CloudWatch Logs data ingestion and storage costs for any logs sent to CloudWatch Logs from your clusters.

Each log type corresponds to a component of the Kubernetes control plane. To learn more about these components, see Kubernetes Components in the Kubernetes documentation.

For more information on EKS control plane logging, refer to the official documentation.

Managed Node Groups

Amazon EKS managed node groups automate the provisioning and lifecycle management of nodes (Amazon EC2 instances) for Amazon EKS Kubernetes clusters.

For more information about how node groups work and how they are configured, refer to the EKS documentation.

User-provided Launch Templates

You can provide your own launch template ID and version to configure the EC2 instances in a node group. If you provide the launch template, none of the template settings will be configurable from Rancher. You must set all of the required options listed below in your launch template.

Also, if you provide the launch template, you can only update the template version, not the template ID. To use a new template ID, create a new managed node group.

caution

You can’t directly update a node group to a newer Kubernetes version if the node group was created from a custom launch template. You must create a new launch template with the proper Kubernetes version, and associate the node group with the new template.

Rancher-managed Launch Templates

If you do not specify a launch template, then you will be able to configure the above options in the Rancher UI and all of them can be updated after creation. In order to take advantage of all of these options, Rancher will create and manage a launch template for you. Each cluster in Rancher will have one Rancher-managed launch template and each managed node group that does not have a specified launch template will have one version of the managed launch template. The name of this launch template will have the prefix “rancher-managed-lt-“ followed by the display name of the cluster. In addition, the Rancher-managed launch template will be tagged with the key “rancher-managed-template” and value “do-not-modify-or-delete” to help identify it as Rancher-managed. It is important that this launch template and its versions not be modified, deleted, or used with any other clusters or managed node groups. Doing so could result in your node groups being “degraded” and needing to be destroyed and recreated.

Custom AMIs

If you specify a custom AMI, whether in a launch template or in Rancher, then the image must be configured properly and you must provide user data to bootstrap the node. This is considered an advanced use case and understanding the requirements is imperative.

If you specify a launch template that does not contain a custom AMI, then Amazon will use the EKS-optimized AMI for the Kubernetes version and selected region. You can also select a GPU enabled instance for workloads that would benefit from it.

note

The GPU enabled instance setting in Rancher is ignored if a custom AMI is provided, either in the dropdown or in a launch template.

Spot instances

Spot instances are now supported by EKS. If a launch template is specified, Amazon recommends that the template not provide an instance type. Instead, Amazon recommends providing multiple instance types. If the “Request Spot Instances” checkbox is enabled for a node group, then you will have the opportunity to provide multiple instance types.

note

Any selection you made in the instance type dropdown will be ignored in this situation and you must specify at least one instance type to the “Spot Instance Types” section. Furthermore, a launch template used with EKS cannot request spot instances. Requesting spot instances must be part of the EKS configuration.

Node Group Settings

The following settings are also configurable. All of these except for the “Node Group Name” are editable after the node group is created.

Self-managed Amazon Linux Nodes

You can register an EKS cluster containing self-managed Amazon Linux nodes. You must configure this type of cluster according to the instructions in the official AWS documentation for launching self-managed Amazon Linux nodes. EKS clusters containing self-managed Amazon Linux nodes are usually operated by the Karpenter project. After you provision an EKS cluster containing self-managed Amazon Linux nodes, register the cluster so it can be managed by Rancher. However, the nodes won’t be visible in the Rancher UI.

IAM Roles for Service Accounts

An Applications Deployment running on an EKS cluster can make requests to AWS services via IAM permissions. These applications must sign their requests with AWS credentials. IAM roles for service accounts manage these credentials using an AWS OIDC endpoint. Rather than distributing AWS credentials to containers or relying on an EC2 instance’s role, you can link an IAM role to a Kubernetes service account and configure your Pods to use this account.

note

Linking to an IAM role is not supported for Rancher pods in an EKS cluster.

To enable IAM roles for service accounts:

1. Create an IAM OIDC provider for your cluster
2. Configure a Kubernetes service account to assume an IAM role
3. Configure Pods to use a Kubernetes service account
4. Use a supported AWS SDK

Configuring the Refresh Interval

The eks-refresh-cron setting is deprecated. It has been migrated to the eks-refresh setting, which is an integer representing seconds.

The default value is 300 seconds.

The syncing interval can be changed by running kubectl edit setting eks-refresh.

If the eks-refresh-cron setting was previously set, the migration will happen automatically.

The shorter the refresh window, the less likely any race conditions will occur, but it does increase the likelihood of encountering request limits that may be in place for AWS APIs.