Designing a resilient test harness for concurrent, stateful E2E scenarios in Playwright

Building a large-scale Playwright End-to-End (E2E) testing suite that runs hundreds of concurrent tests requires a strategy for true isolation, rapid state provisioning, and robust cleanup. Simply relying on browser context isolation is insufficient for backend state management. The solution is to architect a production-grade test harness that provisions unique, isolated backend environments for each test worker.

The key to achieving this isolation is to use dynamic, containerized environments for each Playwright worker process or even per test file. Tools like Docker Compose or Testcontainers can spin up a fresh microservice stack, including databases, for every worker. Global setup and teardown functions are used in playwright.config.ts to orchestrate the lifecycle of these environments, ensuring they are created and properly cleaned up.

State provisioning is handled via API-first, idempotent fixtures that create specific states before UI interaction through direct API or database calls. This is faster than UI-driven setup. The provisioning logic must be idempotent, meaning it can be run multiple times without causing errors or unintended state changes.

Robust cleanup is critical to prevent resource leaks that impact subsequent runs. Global teardown handles the destruction of the entire environment, while afterEach handles test-specific cleanup, ensuring it executes even if a test fails. This combination of container orchestration, API-driven state provisioning, and comprehensive cleanup ensures that each E2E test operates in a clean, dedicated environment, preventing cross-test contamination and flaky failures.