Azure Scenario

Massively Parallel Playwright with Microsoft Playwright Testing

Cut a 90-minute suite to under 10 by running 50+ cloud-hosted Playwright browsers in parallel from Azure DevOps.

Architecture

Azure DevOps Pipeline
        ├─► fetch test token from Key Vault
        ├─► npx playwright test --config=playwright.service.config.ts --workers=50
        │       └─► cloud browser fleet (Microsoft Playwright Testing)
        │               ├─► trace + video per failure → Blob
        │               └─► traceparent header → App Insights (E2E correlation)
        └─► PublishTestResults + PublishBuildArtifacts

Services used

Microsoft Playwright Testing Azure DevOps Pipelines Azure Blob Storage Azure Key Vault Application Insights

Steps

1. Authenticate
Pipeline reads `PLAYWRIGHT_SERVICE_ACCESS_TOKEN` from Key Vault via Managed Identity.
2. Run
Single command launches 50 cloud workers; sharding handled by the service.
3. Correlate
Tests set `traceparent`; App Insights end-to-end view links failing tests to backend spans.
4. Persist
Failure artifacts (trace.zip, video, screenshot) copied to Blob for 90-day retention.
5. Publish
JUnit results published to Pipelines; HTML report attached as build artifact.

Takeaways

Wall-clock drops dramatically without managing grid infrastructure.
End-to-end correlation makes every failure traceable to the right service.