Blue Green Releases with Argo Rollouts

Author: Gourav Shah
Publisher: School of Devops
Version : v2024.06.03.01

---

Launch Vote App with Deployment

Lets begin by first deploying the vote service with the deployment code available.

• Create a fork of Kubernetes Deployment Code for Vote App

Review the code created with kustomization overlay configured for staging environment in additional to the base manifests.

Create namespaces for staging environments as,

kubectl create ns staging

kubectl get ns

kubectl config set-context --current --namespace=staging

validate

kubectl config get-contexts

[sample output]

CURRENT   NAME        CLUSTER     AUTHINFO    NAMESPACE
*         kind-kind   kind-kind   kind-kind   staging

Now clone the forked repo and switch to it

git clone https://github.com/xxxx/argo-labs.git

replace xxxx with your user name

change it to argo-labs and examine the code for base as well as staging as

cd argo-labs

kustomize build base
kustomize build staging

then deploy vote service to staging as

kubectl apply -k staging

where, -k option applies the kustomization spec.

validate

kubectl get all

You should see the deployment and service for vote app. And you should be able to access the staging deployment with nodeport 30000.

Install Argo Rollouts

Install Argo Rollouts Controller and CRDs with ,

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

Validate with,

kubectl api-resources | grep -i argo

Optionally, you could install argo plugin for kubectl

On linux

cd ~

curl -LO https://github.com/argoproj/argo-rollouts/releases/latest/download/kubectl-argo-rollouts-linux-amd64

chmod +x ./kubectl-argo-rollouts-linux-amd64

sudo mv ./kubectl-argo-rollouts-linux-amd64 /usr/local/bin/kubectl-argo-rollouts

On Mac.

curl -LO https://github.com/argoproj/argo-rollouts/releases/latest/download/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

And validate as,

kubectl argo rollouts version

Also install Kustomize by following the instructions in official documentation here.

Create a Preview Service

Create a preview service during blue-green analysis

File: base/preview-service.yaml

---
apiVersion: v1
kind: Service
metadata:
  name: vote-preview
  labels:
    role: vote
spec:
  selector:
    app: vote
  ports:
    - port: 80
      targetPort: 80
      protocol: TCP
      nodePort: 30100
  type: NodePort

Update base/kustomization.yaml with the following

apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- deployment.yaml
- service.yaml
- preview-service.yaml

Create rollout along with preview service with

kubectl apply -k staging

Validate

kubectl get all
kubectl describe svc vote
kubectl describe svc vote-preview

Both services should be pointing to the same set of pods.

Migrate Deployment to Argo Rollout

From argo-labs/base , create a copy of existing deployment spec as,

git mv deployment.yaml rollout.yaml

also update kustomization.yaml to replace deployment.yaml with rollout.yaml

apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- rollout.yaml
- service.yaml
- preview-service.yaml

Now edit base/rollout.yaml to add Blue/Green Release spec as per the documentation here.

Update the following properties

• apiVersion: argoproj.io/v1alpha1
• kind: Rollout
• replicas: 4

Remove spec.template.metadata.name if present.

Replace the strategy with blue green as,

  strategy:
    blueGreen:
      autoPromotionEnabled: true
      autoPromotionSeconds: 30
      activeService: vote
      previewService: vote-preview

Also update argo-labs/staging/kustomization.yaml to remove the following

file: staging/kustomization.yaml

replicas:
- count: 2
  name: vote

Delete the deployment

kubectl delete deploy vote

And then create the rollout from argo-labs path as,

kubectl apply -k staging

Validate

kubectl get ro,all
kubectl describe ro vote

Deploy a Blue/Green Release

Open a new terminal and start watching with

watch kubectl get ro,all --show-labels

Also open two more terminal windows to watch for vote and vote-preview services respectively as

watch kubectl describe svc vote
watch kubectl describe svc vote-preview

If you had installed the argo rollout plugin for kubectl, you could also launch the Rollout UI with

kubectl argo rollouts dashboard -p 3100

and then start watching for the rollout using http://localhost:3100/rollouts. Replace localhost with actul IP address of the host if you are running kubectl on a remote host.

Now, trigger a rollout by updating the image by updating the image tag in base/rollout.yaml

spec:
  containers:
  - image: schoolofdevops/vote:v2

and then by applying it as

kubectl apply -k staging

In the same terminal you could watch the status of the rollout with,

kubectl argo rollouts status vote

You could also watch the status using Argo Rollouts Dashboard as

You would notice that,

• A new replica set is created with new version
• Service vote-bg (Preview service) starts pointing to the new version, while the active service vote still pointing to previous version
• After all pods with new version are ready, it will pause for about 30 seconds as configured in the rollout configuration.
• Once the wait interval is over, active service starts pointing to the new version pods created with the newly launched replica set. This is where there maybe just a blip of a nominal downtime.
• Eventually the replicase with older version will scale down to zero, completing the rollout.

You could try rolling out (by updating the image version) a few times to learn how it works.

Publish Changes to Repo

Scale down the staging replicas so that you could accommodate more replicas created for prod environment, which will be yet another namespace in the same cluster.

edit base/rollout.yaml and set the replicas count to 1

spec:
  replicas: 1

also in the same file, set the image tag back to v1

spec:
  containers:
  - image: schoolofdevops/vote:v1

apply

kubectl apply -k staging/

validate

kubectl get all

Before committing the changes, make sure you have created a GitHub Access Token with repo access. To do so,

• Login to GiHub
• Click on your User Profile on top right
• From Settings -> Developer Settings -> Personal access tokens -> Tokens(classic)
  
• Select Generate new token(classic) , provide authentication as needed and proceed to create token.
• From token creation page, provide a token name and select repo configurations
  
• Generate Token and copy it somewhere safe. You will need this token multiple times throughout this course to be added as kubernetes secret, so keep it handy.

Commit all the changes that you have made so far to the repo

git status
git add base/*.yaml
git status
git commit -am "updated staging deployment code with blue green release"
git push origin main

Cleaning Up

Once you are done with it, clean up the environment with

kubectl delete -k staging/