Azure Scenario

CI/CD with Playwright + Load Testing Gates

Azure Pipelines builds, pushes to ACR, deploys to Container Apps staging, runs Playwright + Azure Load Testing, then traffic-shifts to prod.

Architecture

Repo ─► Azure Pipelines
        ├─► build + unit tests
        ├─► docker build + push (ACR scan)
        ├─► deploy Container Apps revision (staging)
        ├─► Playwright Testing (cloud browsers, 30 workers) ─► JUnit + artifact
        ├─► Azure Load Testing (p95 + error-rate gate)
        └─► traffic shift 100% (or rollback on Monitor alert)

Services used

Azure DevOps Pipelines Azure Container Registry Azure Container Apps Microsoft Playwright Testing Azure Load Testing Azure Key Vault Azure Monitor

Steps

1. Build
Pipeline runs unit tests + builds image; auths to Azure via Workload Identity Federation.
2. Scan
ACR scans image; pipeline fails on HIGH CVEs.
3. Deploy staging
Container Apps revision deployed with weight 0; ingress unchanged.
4. E2E automation
Microsoft Playwright Testing runs the suite with 30 parallel cloud browsers against staging URL.
5. Perf gate
Azure Load Testing executes JMeter scenario with pass criteria: p95 < 800ms AND error rate < 1%.
6. Promote
Shift traffic 100% to new revision; Monitor alert on 5xx auto-reverts within minutes.

Takeaways

Automation + perf gates live in the pipeline, not as post-release Slack pings.
Keyless auth (OIDC) removes long-lived secrets from CI.