Zero downtime migration of EKS node groups

This example lives in the pulumi/examples repository. Check out just this directory to use it:

$ git clone --filter=blob:none --sparse https://github.com/pulumi/examples pulumi-examples
$ git -C pulumi-examples sparse-checkout set aws-ts-eks-migrate-nodegroups
$ cd pulumi-examples/aws-ts-eks-migrate-nodegroups

Creates an EKS cluster with node groups and a workload, and showcases adding a node group to use for workload migration with zero downtime.

This is a Kubernetes Day 2 operations walkthrough: it stands up two worker node groups with different instance types and AMIs, deploys the NGINX Ingress Controller and a simple echoserver app across them, then migrates NGINX onto a new, larger node group with zero downtime before decommissioning the original one.

The migration relies on Pulumi’s create-before-delete model and autonaming — the replacement node group is created and workloads are shifted onto it before the original is drained and deleted — combined with Kubernetes high-availability settings (rolling updates, scheduling affinity, and graceful pod termination) so that in-flight requests keep returning HTTP 200 throughout.

For step-by-step instructions, check out the tutorial.

Prerequisites#

1. Install Pulumi
2. Configure AWS credentials
3. Install Node.js

Deploying the example#

1. Create a new stack:

   pulumi stack init dev

2. Set the AWS region to deploy into:

   pulumi config set aws:region us-west-2

3. Install dependencies:

   npm install

4. Deploy the stack:

   pulumi up

Then follow the tutorial for the step-by-step migration walkthrough.

Cleaning up#

Once you’re finished experimenting, destroy your stack and remove it to avoid incurring further charges:

pulumi destroy
pulumi stack rm