vitest-retry-testing-real-vs-fake-timers

Fake timers in Vitest cause flaky retry tests when microtask and macrotask coordination breaks down, leading to PromiseRejectionHandledWarning or unsettled promise chains during setTimeout-based backoff patterns.

To stabilize Vitest retry tests with real timers, remove vi.useFakeTimers() and rely on real timers for small delays, ensuring microtask and macrotask coordination behaves as it does in production for reliable assertions.

Yes, fake timers can be used safely in Vitest retry tests by applying the microtask-flush pattern and using advanceTimersByTimeAsync with explicit awaits to ensure all promises settle before assertions execute.

The best way to test exponential backoff in Vitest is using real timers for small delays to maintain production-like microtask coordination, or carefully coordinating fake timers with async timer APIs like advanceTimersByTimeAsync for larger delays.

You should avoid vi.useFakeTimers in Vitest retry tests for small delays where real timers ensure accurate microtask and macrotask coordination, since fake timers can introduce PromiseRejectionHandledWarning and test flakiness.

Migrating retry utilities from fake to real timers in Vitest involves removing vi.useFakeTimers() and relying on real timers for small delays, following provided step-by-step migration patterns for backoff strategies and rate limiters to reduce flakiness.