Demo Progressive Delivery with Argo Rollouts

Deploying new versions to production can be risky: push a bad image and all pods may be replaced at once, causing user-facing errors. Progressive delivery gives you control over how a new version is introduced — shift a small percentage of traffic first, observe health and metrics, then gradually increase exposure. If anything looks wrong, abort and traffic returns to the stable revision. Argo Rollouts implements progressive delivery for Kubernetes as a drop-in replacement for Deployments. This walkthrough shows how to:

Verify Argo Rollouts is running
Create a Rollout (canary strategy with weights and pauses)
Perform a canary release by updating the image and promoting
Attach a Rollout to an existing Deployment using workloadRef

Verify Argo Rollouts is running

Confirm the Argo Rollouts controller and dashboard are up:

controlplane ➜ kubectl get pods -n argo-rollouts
NAME                                           READY   STATUS    RESTARTS   AGE
argo-rollouts-5f64f8d68-1xr9s                  1/1     Running   0          10m
argo-rollouts-dashboard-755bbc64c-59rhx        1/1     Running   0          10m

Create a Rollout — basic structure

A Rollout resource closely resembles a Deployment. The main difference is the strategy section, which defines canary or blue/green progressive delivery behavior. Here is a minimal Rollout that behaves like a regular Deployment:

apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: rollout-demo
  namespace: canary-demo
spec:
  replicas: 4
  selector:
    matchLabels:
      app: rollout-demo
  template:
    metadata:
      labels:
        app: rollout-demo
    spec:
      containers:
        - name: demo
          image: argoproj/rollouts-demo:blue
          ports:
            - containerPort: 8080

To enable canary-style progressive delivery, add a strategy.canary.steps sequence. The example below progresses 25% → pause → 50% → pause → 75% → pause → 100%:

apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: rollout-demo
  namespace: canary-demo
spec:
  replicas: 4
  selector:
    matchLabels:
      app: rollout-demo
  template:
    metadata:
      labels:
        app: rollout-demo
    spec:
      containers:
        - name: demo
          image: argoproj/rollouts-demo:blue
          ports:
            - containerPort: 8080
  strategy:
    canary:
      steps:
        - setWeight: 25
        - pause: {}
        - setWeight: 50
        - pause: {}
        - setWeight: 75
        - pause: {}
        - setWeight: 100

Apply the Rollout:

controlplane ➜ kubectl apply -f rollout.yaml
rollout.argoproj.io/rollout-demo created

Inspect the rollout using the Argo Rollouts kubectl plugin:

controlplane ➜ kubectl argo rollouts get rollout rollout-demo -n canary-demo
Name:               rollout-demo
Namespace:          canary-demo
Status:             Healthy
Strategy:           Canary
Step:               1/4
SetWeight:          0
ActualWeight:       0
Images:             argoproj/rollouts-demo:blue (stable)

Replicas:
  Desired: 4
  Current: 4
  Ready:   4
  Available: 4

Promote and update the image (canary release)

Promoting a rollout advances it to the next defined step. Running promote with no image change will not progress a revision (there is no new revision to advance):

controlplane ➜ kubectl argo rollouts promote rollout-demo -n canary-demo
rollout 'rollout-demo' promoted

To create a canary release, update the container image (for example, blue → yellow) in rollout.yaml, then apply the change:

# edit rollout.yaml: containers[0].image = argoproj/rollouts-demo:yellow
controlplane ➜ kubectl apply -f rollout.yaml
rollout.argoproj.io/rollout-demo configured

After applying the updated image the Rollout creates a new revision and will advance to the first step (setWeight: 25). One pod (25%) runs the new yellow image while the others remain on blue:

controlplane ➜ kubectl argo rollouts get rollout rollout-demo -n canary-demo
Name:               rollout-demo
Namespace:          canary-demo
Status:             Degraded/Healthy (depends on checks)
Strategy:           Canary
Step:               1/4
SetWeight:          25
ActualWeight:       25
Images:             argoproj/rollouts-demo:yellow (canary), argoproj/rollouts-demo:blue (stable)

Replicas:
  Desired: 4
  Current: 4
  Ready:   4
  Available: 4

You can also inspect pods directly to confirm which revision is running:

controlplane ➜ kubectl get pods -n canary-demo
NAME                                      READY STATUS    AGE
rollout-demo-544bf7c68b-6jh2n             1/1   Running   5m
rollout-demo-544bf7c68b-6lzjz             1/1   Running   5m
rollout-demo-544bf7c68b-7l7dr             1/1   Running   5m
rollout-demo-544bf7c68b-bpbwd             1/1   Running   42s

controlplane ➜ kubectl describe pod -n canary-demo rollout-demo-544bf7c68b-bpbwd | grep -i image
    Image:          argoproj/rollouts-demo:yellow

The Argo Rollouts UI also visualizes canary progress and revision distribution. Example UI view:

This is the interface of Argo Rollouts, showing a rollout demo with canary strategy, including steps for weight adjustment and revision details. The rollout status is detailed with sections for steps, summary, containers, and revisions.

Promote to additional steps (CLI or UI)

Use CLI promote to move through the next pause or next weight:

controlplane ➜ kubectl argo rollouts promote rollout-demo -n canary-demo
rollout 'rollout-demo' promoted

Each promote advances to the next step you defined (50% → 75% → 100%). When the Rollout reaches 100% and checks look good, the new revision becomes the stable revision. You may also use the UI’s “Promote Full” option to advance all remaining steps at once.

Rollback

If the canary revision shows regressions, you can roll back from the Argo Rollouts dashboard to the previous stable revision. The UI provides an immediate way to revert so traffic returns to the stable revision.

The image shows a dashboard from the Argo Rollouts system, displaying a canary deployment strategy with various steps, including weight settings and pauses, and information on deployment revisions.

Manage existing workloads with workloadRef

If you already have an existing Deployment, you can let a Rollout reference that Deployment via workloadRef. This allows adding progressive delivery to an existing workload without replacing the resource. Create a Deployment (example uses replicas: 0 initially):

apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-app
  namespace: canary-demo
spec:
  replicas: 0
  selector:
    matchLabels:
      app: demo-app
  template:
    metadata:
      labels:
        app: demo-app
    spec:
      containers:
        - name: demo-app
          image: argoproj/rollouts-demo:blue
          ports:
            - containerPort: 8080

Apply the Deployment:

controlplane ➜ kubectl apply -f deployment.yaml
deployment.apps/demo-app created

Create a Rollout that references the existing Deployment via workloadRef, and define a canary strategy:

apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: demo-app-rollout
  namespace: canary-demo
spec:
  replicas: 4
  workloadRef:
    apiVersion: apps/v1
    kind: Deployment
    name: demo-app
  strategy:
    canary:
      steps:
        - setWeight: 10
        - pause: {}
        - setWeight: 50
        - pause: {}
        - setWeight: 100

Apply the Rollout:

controlplane ➜ kubectl apply -f workloadref-rollout.yaml
rollout.argoproj.io/demo-app-rollout created

The Rollout will reconcile and manage the referenced Deployment, scaling it to the number of replicas configured by the Rollout (for example, from replicas: 0 up to 4) and shifting traffic according to canary steps:

controlplane ➜ kubectl get pods -n canary-demo
NAME                                     READY   STATUS    AGE
demo-app-677bb5c8-6cxs5                  1/1     Running   53s
demo-app-677bb5c8-c4xg4                  1/1     Running   53s
demo-app-677bb5c8-rbxg4                  1/1     Running   53s
demo-app-677bb5c8-sbkm4                  1/1     Running   53s

The image shows the Argo Rollouts user interface displaying a demo application rollout in progress using a canary strategy, with weight settings and revisions detailed.

A common YAML mistake is using step (singular) instead of the required steps (plural) under spec.strategy.canary. If you see an error like unknown field "spec.strategy.canary.step", verify you have steps: and that each step entry is properly indented.

Quick reference

Action	Command / Note
Check Argo Rollouts pods	`kubectl get pods -n argo-rollouts`
Inspect a rollout	`kubectl argo rollouts get rollout <name> -n <namespace>`
Promote canary to next step	`kubectl argo rollouts promote <name> -n <namespace>`
Attach Rollout to Deployment	use `spec.workloadRef` in the Rollout manifest
UI promote / rollback	Available in Argo Rollouts dashboard

Next steps & references

Practice the walkthrough: verify Argo Rollouts is installed, create a canary Rollout, update images to trigger revisions, promote step-by-step, and roll back if needed. Add observability and automated analysis (metrics, alerts) to make promotion decisions safer.

Argo Rollouts documentation: https://argoproj.github.io/argo-rollouts/
Argo Rollouts GitHub: https://github.com/argoproj/argo-rollouts

Now you have a compact, hands-on guide: verify Argo Rollouts, create and apply a canary Rollout, perform controlled image updates and promotions, roll back when necessary, and attach Rollouts to pre-existing Deployments with workloadRef.

Course Introduction

Platform Architecture and Infrastructure

GitOps and Continuous Delivery

Platform APIs and Self-Service Capabilities

Observability and Operations

Security and Policy Enforcement

Conclusion

Demo Progressive Delivery with Argo Rollouts

Verify Argo Rollouts is running

Create a Rollout — basic structure

Promote and update the image (canary release)

Promote to additional steps (CLI or UI)

Rollback

Manage existing workloads with workloadRef

Quick reference

Next steps & references

Watch Video

Practice Lab

​Verify Argo Rollouts is running

​Create a Rollout — basic structure

​Promote and update the image (canary release)

​Promote to additional steps (CLI or UI)

​Rollback

​Manage existing workloads with workloadRef

​Quick reference

​Next steps & references

Watch Video

Practice Lab

Verify Argo Rollouts is running

Create a Rollout — basic structure

Promote and update the image (canary release)

Promote to additional steps (CLI or UI)

Rollback

Manage existing workloads with workloadRef

Quick reference

Next steps & references