This post is part of “Progressive Delivery” series.

  1. Introduction to Progressive Delivery
  2. Practical Progressive Delivery with Argo Rollouts - Setup (this post)
  3. Practical Progressive Delivery with Argo Rollouts - BlueGreen
  4. Practical Progressive Delivery with Argo Rollouts - Canary
  5. Practical Progressive Delivery with Argo Rollouts - Auto rollback with metrics

Table of Contents:


I talked about Progressive Delivery in detail in my previous post. To complete the picture, we will need to get our hands on. This post will guide you to setup Argo Rollouts, one of a Progressive Delivery software.

What is Argo Rollouts

Argo Rollouts is a cloud native software that can control software deployment on Kubernetes cluster. It extends the concept of a core Kubernetes resource “Deployment” that only support a “rolling update deployment”. Argo Rollouts supports more deployment approaches such as BlueGreen and Canary deployments (Learn more about these deployment here).

Next, we will learn how to setup Argo Rollouts.


  1. We need a Kubernetes cluster. This post assume that you followed this post to setup a K3s-based Kubernetes cluster.
  2. Helm installed
  3. kubectl installed

Install Argo Rollouts

Run these commands to install Argo Rollouts helm chart.

kubectl create namespace argo-rollouts
kubectl apply -n argo-rollouts \


  • we will install Argo Rollouts controllers in the namespace “argo-rollouts”

Install Argo Rollouts Kubectl plugin

Argo Rollouts Kubectl plugin enables us to run rollouts specific command using kubectl cli such as open the dashboard UI. This will make our life easier interacting with Rollouts.


curl -LO
chmod +x ./kubectl-argo-rollouts-darwin-amd64
sudo mv ./kubectl-argo-rollouts-darwin-amd64 /usr/local/bin/kubectl-argo-rollouts

# verify
kubectl argo rollouts version


This plugin will be used later to open Argo Rollouts Dashboard.

Argo Rollouts dashboard can visualize current state of the deployment and enable use to take rollouts actions directly from the UI.

Install NGINX ingress

To access Argo Rollouts demo application in the cluster, we need an ingress. This post will use NGINX ingress controller which is a popular ingress for Kubernetes.

helm repo add ingress-nginx
helm repo update

kubectl create namespace ingress-nginx
helm upgrade -i -n ingress-nginx ingress-nginx ingress-nginx/ingress-nginx \
--set controller.metrics.enabled=true \
--set controller.metrics.serviceMonitor.enabled=true


  • “controller.metrics.enabled=true” and “controller.metrics.serviceMonitor.enabled=true” are used to enable Prometheus metrics scraping endpoint (we need it for automatic rollback and can be skipped for now)
  • we will install nginx-ingress in namespace “ingress-nginx”

Deploy a basic Argo Rollouts demo

Clone this repository to your computer.

git clone
cd k3s-argo-rollouts

Install a demo rollout.

kubectl apply -f ./basic-rollouts-demo


This command will install.

  • rollout.yaml. A rollout resource named “basic-rollouts-demo” which will install image “argoproj/rollouts-demo:blue” to the cluster.
  • service.yaml. A kubernetes service named “basic-rollouts-demo” that enables the access to “basic-rollouts-demo” application.
  • ingress.yaml. A NGINX ingress that enables access to “basic-rollouts-demo” service from the browser outside the cluster (using

Run this command to start Argo Rollouts dashboard.

kubectl argo rollouts dashboard

Open http://localhost:3100 to see the dashboard.

Dashboard UI shows 5 instances of basic-rollouts-demo application started successfully

Open url to open “basic-rollouts-demo” application and you should see.

Argo Rollouts Demo Application showing “blue” version


“” in the url is the Kubernetes master node IP address (if you create the cluster following this post). Any node IP address will work.


In this post, you learned about Argo Rollouts, how to setup it, and creating a basic rollout application to a Kubernetes cluster. Argo Rollouts has many more features such as Traffic Management and Analysis which I will show you in later posts. Next, BlueGreen deployment.