Argo Rollouts: Advanced Deployment Strategies for Kubernetes

Overview

Argo Rollouts is a Kubernetes Operator that provides advanced deployment capabilities such as blue‑green, canary, canary analysis, experiments, and progressive delivery, enabling automated, GitOps‑based incremental releases for cloud‑native applications and services.

Key Features

Blue‑Green update strategy

Canary update strategy

Fine‑grained weighted traffic splitting

Automatic rollback

Manual promotion

Custom metric queries and business KPI analysis

Ingress controller integration: NGINX, ALB

Service mesh integration: Istio, Linkerd, SMI

Metrics provider integration: Prometheus, Wavefront, Kayenta, Datadog, New Relic, etc.

Implementation Principle

Similar to a Deployment, the Argo Rollouts controller manages the creation, scaling, and deletion of ReplicaSets defined by the spec.template field of a Rollout resource. When spec.template changes, the controller creates a new ReplicaSet and follows the strategy defined in spec.strategy to transition from the old ReplicaSet to the new one. Once the new ReplicaSet is scaled up (optionally after an Analysis), it is marked as stable .

If another change to spec.template occurs while a transition is in progress, the previously created ReplicaSet is scaled down and the controller starts a new rollout for the latest template.

Related Concepts

A Rollout is a Kubernetes Custom Resource Definition (CRD) that functions like a Deployment but adds advanced deployment and progressive delivery features.

Blue‑Green deployment

Canary deployment

Integration with Ingress controllers and service meshes for sophisticated traffic routing

Integration with metric providers for analysis

Automatic promotion or rollback based on success/failure metrics

Progressive Delivery

Progressive delivery releases product updates in a controlled, incremental manner, reducing risk by combining automation with metric analysis to drive automatic upgrades or rollbacks.

Deployment Strategies

RollingUpdate : Gradually replace old pods with new ones while maintaining the total replica count (default for Deployments).

Recreate : Delete all old pods before starting new ones, causing downtime during the transition.

Blue‑Green : Deploy a new version alongside the old one; traffic is switched to the new version only after verification.

Canary : Expose a small portion of users to the new version while the majority continue using the old version; traffic weight can be increased gradually.

Architecture

The following components are managed by Argo Rollouts:

Rollout Controller : Watches cluster events and reconciles Rollout resources to the desired state.

Rollout Resource : A custom resource compatible with native Deployments but with extra fields for advanced strategies.

Old and New ReplicaSets : Standard ReplicaSet objects with additional metadata for version tracking.

Ingress/Service : Standard Kubernetes Service resources with extra metadata to route traffic between versions; supports multiple services during a rollout.

Analysis & AnalysisRun : Custom resources that connect a Rollout to metric providers, define thresholds, and decide success or failure, optionally pausing the rollout.

Metric Providers : Native integrations with popular providers (Prometheus, Datadog, New Relic, etc.) used in Analysis.

CLI & UI : Optional kubectl plugin and UI for visualizing and managing Rollouts.

Installation

Install Argo Rollouts with the following commands:

➜  ~ kubectl create namespace argo-rollouts
➜  ~ kubectl apply -n argo-rollouts -f https://github.com/argoproj/argo-rollouts/releases/download/v1.0.2/install.yaml

This creates the argo-rollouts namespace where the controller runs.

➜  ~ kubectl get pods -n argo-rollouts
NAME                     READY   STATUS    RESTARTS   AGE
argo-rollouts-6fdcf89f7c-7z2mh   1/1     Running   0          58s

Install the kubectl plugin (macOS example):

➜  ~ brew install argoproj/tap/kubectl-argo-rollouts

Or install manually with curl:

➜  ~ curl -LO https://github.com/argoproj/argo-rollouts/releases/download/v1.0.2/kubectl-argo-rollouts-darwin-amd64
➜  ~ chmod +x ./kubectl-argo-rollouts-darwin-amd64
➜  ~ sudo mv ./kubectl-argo-rollouts-darwin-amd64 /usr/local/bin/kubectl-argo-rollouts
➜  ~ kubectl argo rollouts version

Usage Examples

1. Deploy a Rollout

Create a Rollout with a canary strategy that initially sends 20% of traffic to the canary version and then pauses for manual promotion:

# basic-rollout.yaml
apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: rollouts-demo
spec:
  replicas: 5  # define 5 replicas
  strategy:
    canary:
      steps:
      - setWeight: 20
      - pause: {}
      - setWeight: 40
      - pause: {duration: 10}
      - setWeight: 60
      - pause: {duration: 10}
      - setWeight: 80
      - pause: {duration: 10}
  revisionHistoryLimit: 2
  selector:
    matchLabels:
      app: rollouts-demo
  template:
    metadata:
      labels:
        app: rollouts-demo
    spec:
      containers:
      - name: rollouts-demo
        image: argoproj/rollouts-demo:blue
        ports:
        - name: http
          containerPort: 8080
          protocol: TCP
        resources:
          requests:
            memory: 32Mi
            cpu: 5m

Create a corresponding Service:

# basic-service.yaml
apiVersion: v1
kind: Service
metadata:
  name: rollouts-demo
spec:
  ports:
  - port: 80
    targetPort: http
    protocol: TCP
    name: http
  selector:
    app: rollouts-demo

➜  ~ kubectl apply -f basic-rollout.yaml
➜  ~ kubectl apply -f basic-service.yaml

Watch the rollout:

➜  ~ kubectl argo rollouts get rollout rollouts-demo --watch

2. Update a Rollout

Change the container image to a new version (e.g., yellow ) using the plugin:

➜  ~ kubectl argo rollouts set image rollouts-demo \
  rollouts-demo=argoproj/rollouts-demo:yellow

The controller follows the defined canary steps, pausing after the first 20% weight.

3. Promote a Rollout

When the rollout is paused, manually promote to the next step:

➜  ~ kubectl argo rollouts promote rollouts-demo

Use --full to skip remaining steps and analyses.

4. Abort a Rollout

Deploy a new image (e.g., red ) and then abort the rollout to roll back to the stable version:

➜  ~ kubectl argo rollouts set image rollouts-demo \
  rollouts-demo=argoproj/rollouts-demo:red
➜  ~ kubectl argo rollouts abort rollouts-demo

After aborting, you can restore the previous stable image with another set image command.

Dashboard

Start a local dashboard to visualize Rollouts:

kubectl argo rollouts dashboard

Then open http://localhost:3100 in a browser. The UI allows you to inspect Rollout details, view configuration, and perform actions such as restart, pause, or abort directly.

Additional Resources

The article concludes with a promotion for a "K8S Advanced Training Camp" and a QR code for further cloud‑native learning, but the technical content above provides a comprehensive guide to using Argo Rollouts for progressive delivery on Kubernetes.