async-error-handling

To handle CancelledError in Python asyncio, you must catch the exception, execute your cleanup logic in a finally block, and safely re-raise it to ensure the cancellation propagates correctly and resources are cleaned up without breaking the event loop.

The best way to manage FastAPI lifespan is using asynccontextmanager to handle startup and shutdown events, ensuring asynchronous resources are properly initialized and cleaned up even when unexpected exceptions or cancellations occur during the application lifecycle.

Async context managers fail during shutdown when CancelledError is swallowed or not re-raised. You must explicitly track CancelledError within your finally cleanup blocks to maintain predictable async behavior and ensure resources are fully released.

Yes, you can apply asynccontextmanager patterns to production-grade FastAPI services. Standardized templates provide robust asynchronous workflows with explicit exception handling and safe re-raising suitable for small utilities and large-scale services.

You safely re-raise asyncio exceptions by catching CancelledError specifically, performing your necessary cleanup operations in a finally block, and then re-raising the exception immediately to preserve the cancellation state expected by the asyncio event loop.