Automating Blue/Green Deployment with Argo Rollouts

In this article, we will learn how to automate Blue/Green Deployment with Argo Rollouts.

Argo Rollouts

Argo Rollouts is a Kubernetes Controller and set of Custom Resource Definitions (CRDs) that provide advanced features for application deployment on Kubernetes compared to the native Kubernetes Deployment Object.

Argo Rollouts provide deployment capabilities such as blue-green, canary, canary analysis, experimentation, and progressive delivery features to Kubernetes.

Because Argo Rollouts are a set of CRDs, we need to install them into the Kubernetes Cluster. Run the following commands to install Argo Rollouts.

Check if we installed it successfully.

Argo Rollouts Deployment Strategy

To use Argo Rollouts, we declare a resource with the apiVersion attribute as argoproj.io/v1alpha1 and the kind as Rollout, like below:

The configuration of Argo Rollouts has a strategy property for us to choose the deployment strategy we want, with two values of blueGreenand canary.

See detail here Rollout Specification. Don’t try to understand all properties for now.

In this article, we will learn about blue/green.

Practice

I use Minikube to run Kubernetes Cluster for the demo. Create a file named bluegreen-rollout.yaml.

All properties of Rollout are the same as a native Deployment Object, only the strategy attribute is different. In the file above, we declare 3 properties:

• autoPromotionEnabled: false – Indicates if the rollout should automatically promote the new ReplicaSet to the active service or enter a paused state. If not specified, the default value is true.
• activeService: bluegreen-demo – Reference to service that the rollout modifies as the active service.
• previewService: bluegreen-demo-preview – Name of the service that the rollout modifies as the preview service.

Next, we create a file named service.yaml.

The two Services properties are the same except for the name property. Next, we create a Rollout Object.

When we create a Rollout, the Argo Rollouts implicit create a ReplicaSet for a current revision. Let’s check:

Ensure that Replica Set and Pod are running, next, we create a Service.

At this time, both the bluegreen-demo and bluegreen-demo-preview are pointing to the same ReplicaSet as bluegreen-demo-fbc7b7f55.

Let’s test, we run the following commands to get the URL of both services.

Open the browser, we will see the UI below.

Now, we change the image property of Rollout Object.

Updating Rollout Object.

At this point, Argo Rollouts will create a new ReplicaSet for the new configuration.

Then the bluegreen-demo-preview service is modified to point to the new ReplicaSet. Accessing the preview service address will see a different UI.

And the bluegreen-demo service does not change.

After we check the new ReplicaSet and see that all is well, next, we promote the new revision of ReplicaSet by updating the bluegreen-demo service to point to it, we run the following command (don’t follow, we will use another way).

Now, Argo Rollouts update the bluegreen-demo service to point to the new ReplicaSet, after waiting (default 30 seconds), the old ReplicaSet is scaled down.

UI Dashboard

In my opinion, DevOps Engineers should not do the work “promotes” the new ReplicaSet, our task is just to build the CI/CD so that the Rollout can be updated when a new version of the application is available. Leave it to QC Engineer.

But QC can’t run CLI so we need to have a dashboard for QC to do this. Fortunately, Argo Rollout provides us with a dashboard, which we can enable using kubectl or using quay.io/argoproj/kubectl-argo-rollouts container image.

Go to localhost:3100 we will see the dashboard of Argo Rollouts.

Choose to bluegreen-demo.

You will see the Promote button, the person who clicks this button will be the QC, if anything is wrong, the QC will be responsible =)))), let’s click on the promote button.

Click Sure.

Now you access both the bluegreen-demo and the bluegreen-demo-preview service we will see the same UI.

Ingress

Ingress configuration for public access if you need.

Done.

Conclusion

So we have learned how to automate Blue/Green Deployment with Argo Rollouts, as you can see, it’s also simple. If you have any questions or need further clarification, you can ask in the comment section below.