Container Service for Kubernetes:Perform canary releases with ACK One GitOps and Argo Rollouts

Option 1: Manual promotion

This approach pauses the canary after the first traffic step so you can verify the new version before proceeding.

Traffic steps: 20% → pause (indefinite) → 40% (5 min) → 60% (5 min) → 80% (5 min) → 100%

With three timed steps of 5 minutes each, promotion after the first manual approval takes approximately 15 minutes.

1. Update rollout.yaml with the new image tag and a manual pause after the first step, then commit and push:

   apiVersion: argoproj.io/v1alpha1
   kind: Rollout
   metadata:
     name: rollouts-demo
   spec:
     replicas: 4
     strategy:
       canary:
         canaryService: rollouts-demo-canary
         stableService: rollouts-demo-stable
         trafficRouting:
           nginx:
             stableIngress: rollouts-demo-stable
         steps:
           - setWeight: 20
           - pause: {}          # Indefinite pause — advance manually by replacing {} with a duration
           - setWeight: 40
           - pause: {duration: 5m}
           - setWeight: 60
           - pause: {duration: 5m}
           - setWeight: 80
           - pause: {duration: 5m}
     revisionHistoryLimit: 2
     selector:
       matchLabels:
         app: rollouts-demo
     template:
       metadata:
         labels:
           app: rollouts-demo
       spec:
         containers:
         - name: rollouts-demo
           image: argoproj/rollouts-demo:yellow   # New image tag
           ports:
           - name: http
             containerPort: 8080
             protocol: TCP
           resources:
             requests:
               memory: 32Mi
               cpu: 5m

2. Watch the rollout pause at 20% traffic:

   kubectl argo rollouts get rollout rollouts-demo --watch

   Expected output: The rollout stops at the first pause: {} step because no duration is set. It advances only after you resume it.

   

3. Resume the canary release by updating the pause duration in rollout.yaml, then commit and push:

   steps:
     - setWeight: 20
     - pause: {duration: 10s}   # Replace {} with a duration to resume

   Then watch the release complete:

   kubectl argo rollouts get rollout rollouts-demo --watch

   Expected output during promotion:

   

   Expected output after completion:

   

Option 2: Automated promotion with Prometheus metrics

This approach uses Managed Service for Prometheus to continuously evaluate the canary's HTTP success rate. If the success rate stays at or above 95% throughout the analysis window, the canary is promoted automatically. If it drops below the threshold for more than 10 consecutive checks, the release is rolled back automatically.

Traffic steps: 20% (5 min) → 40% (5 min, analysis starts here) → 60% (5 min) → 80% (5 min) → 100%

Total promotion time: approximately 20 minutes.

Step 4a: Configure the Rollout with metric analysis

Update rollout.yaml with the new image tag and the analysis configuration, then commit and push:

strategy:
  canary:
    analysis:
      templates:
      - templateName: success-rate
      startingStep: 2   # Start analysis at the 40% step, after initial traffic stabilizes
      args:
        - name: service-name
          value: rollouts-demo-stable
    canaryService: rollouts-demo-canary
    stableService: rollouts-demo-stable
    trafficRouting:
      nginx:
        stableIngress: rollouts-demo-stable
    steps:
      - setWeight: 20
      - pause: {duration: 5m}
      - setWeight: 40
      - pause: {duration: 5m}
      - setWeight: 60
      - pause: {duration: 5m}
      - setWeight: 80
      - pause: {duration: 5m}
revisionHistoryLimit: 2
selector:
  matchLabels:
    app: rollouts-demo
template:
  metadata:
    labels:
      app: rollouts-demo
  spec:
    containers:
    - name: rollouts-demo
      image: argoproj/rollouts-demo:blue   # New image tag

Step 4b: Get the Managed Service for Prometheus endpoint

Managed Service for Prometheus is exposed as a Kubernetes Service at:

http://{ServiceName}.{Namespace}.svc.{ClusterDomain}:{ServicePort}

For the ack-arms-prometheus add-on deployed in the arms-prom namespace with the default cluster domain, the endpoint is:

http://arms-prom-server.arms-prom.svc.cluster.local:9090

Step 4c: Create the AnalysisTemplate

Create analysis.yaml with the following content. The successCondition passes the canary step if the ratio of 2xx responses to all canary requests is 95% or higher over a 5-minute window.

apiVersion: argoproj.io/v1alpha1
kind: AnalysisTemplate
metadata:
  name: success-rate
spec:
  args:
  - name: service-name
  metrics:
  - name: success-rate
    interval: 5m
    successCondition: result[0] >= 0.95   # Promote if 95%+ requests succeed
    failureLimit: 10                       # Abort after 10 consecutive failures
    provider:
      prometheus:
        address: http://arms-prom-server.arms-prom.svc.cluster.local:9090
        query: |
          sum(
            irate(nginx_ingress_controller_requests{status=~"(1|2).*", canary!="" ,service="{{args.service-name}}"}[5m]))
            /
            sum(irate(nginx_ingress_controller_requests{canary!="",service="{{args.service-name}}"}[5m])
          )

The PromQL query divides the rate of 1xx/2xx canary requests by the total canary request rate. The canary!="" label selector filters for traffic routed through the canary Ingress annotation — this ensures only canary traffic is evaluated, not stable traffic. The service label scopes the query to the specific application.

Step 4d: Generate continuous traffic for metric collection

Prometheus needs a steady request stream to evaluate the success rate. Run the following commands in a separate terminal.

1. Get the Ingress external IP:

   kubectl get ingress

   Expected output:

   NAME                                        CLASS   HOSTS                 ADDRESS         PORTS   AGE
   rollouts-demo-rollouts-demo-stable-canary   nginx   rollouts-demo.local   8.217.XX.XX   80      9h
   rollouts-demo-stable                        nginx   rollouts-demo.local   8.217.XX.XX   80      9h

2. Add the host mapping to your local Hosts file:

   8.217.XX.XX  rollouts-demo.local

3. Send continuous requests to the application:

   while true; do curl -s "http://rollouts-demo.local/" | grep -o "<title>.*</title>"; sleep 200ms; done

Step 4e: Watch the automated rollout

kubectl argo rollouts get rollout rollouts-demo --watch

Expected output:

To view the success rate metrics in the Alibaba Cloud console:

1. Log on to the ACK console. In the left-side navigation pane, click Cluster.

2. Click the cluster name. In the left-side pane, choose Operations > Prometheus Monitoring.

3. On the Prometheus Monitoring page, click the Network Monitoring tab, then click Ingresses.

   

After the canary release completes successfully: