This is what separates progressive delivery from just a slow rollout

Without AnalysisTemplate: canary just waits a fixed time then promotes. A bug that appears after 10 minutes would not be caught. With AnalysisTemplate: Prometheus (or Datadog, or a custom webhook) is queried continuously. The system promotes or rolls back based on real metric data, not time elapsed.

Why Argo Rollouts over standard Kubernetes rolling updates?

Standard Kubernetes rolling update sends traffic to new pods as soon as they start. There is no traffic splitting, no automated analysis, no pause-and-check. Argo Rollouts adds: controlled traffic percentages, automated metric analysis (roll back if error rate rises), manual pause gates, and full visibility of the rollout progress.

Complete canary rollout example

apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: payment-api
  namespace: production
spec:
  replicas: 10
  selector:
    matchLabels:
      app: payment-api
  template:          # same as Deployment pod template
    metadata:
      labels:
        app: payment-api
    spec:
      containers:
      - name: payment-api
        image: myacr.azurecr.io/payment-api:v2.1.0
        ports:
        - containerPort: 8080
  strategy:
    canary:
      steps:
      - setWeight: 10        # Step 1: send 10% traffic to new version
      - pause: {duration: 5m}  # wait 5 minutes, observe metrics
      - setWeight: 30        # Step 2: increase to 30%
      - pause: {}            # pause indefinitely — manual promotion needed
      - setWeight: 60
      - pause: {duration: 10m}
      - setWeight: 100       # Step 4: full rollout
      canaryService: payment-api-canary    # Service pointing to new pods
      stableService: payment-api-stable    # Service pointing to old pods
      trafficRouting:
        nginx:
          stableIngress: payment-ingress   # NGINX Ingress controls traffic split

Complete blue-green example

strategy:
  blueGreen:
    activeService: payment-api           # currently receiving traffic (blue)
    previewService: payment-api-preview  # new version deployed here (green)
    autoPromotionEnabled: false          # require manual promotion
    scaleDownDelaySeconds: 30            # keep blue running 30s after promotion
    prePromotionAnalysis:                # run analysis before switching traffic
      templates:
      - templateName: success-rate
      args:
      - name: service-name
        value: payment-api-preview

AnalysisTemplate queries Prometheus/Datadog to decide pass or fail

apiVersion: argoproj.io/v1alpha1
kind: AnalysisTemplate
metadata:
  name: success-rate
spec:
  args:
  - name: service-name
  metrics:
  - name: success-rate
    interval: 2m           # check every 2 minutes
    count: 3               # must pass 3 consecutive checks
    successCondition: result[0] >= 0.95   # 95% success rate required
    failureLimit: 1        # fail if check fails once
    provider:
      prometheus:
        address: http://prometheus.monitoring:9090
        query: |
          sum(rate(http_requests_total{
            service="{{args.service-name}}",
            status!~"5.."
          }[5m])) /
          sum(rate(http_requests_total{
            service="{{args.service-name}}"
          }[5m]))

Rollout commands

# Install kubectl plugin
kubectl argo rollouts version

# Watch rollout progress in terminal
kubectl argo rollouts get rollout payment-api --watch

# Promote manually (resume a paused step)
kubectl argo rollouts promote payment-api

# Abort and rollback immediately
kubectl argo rollouts abort payment-api

# Manually set canary weight
kubectl argo rollouts set image payment-api payment-api=myimage:v2
kubectl argo rollouts promote payment-api --skip-current-step