python runtime_error ai_generated true

asyncio.exceptions.TimeoutError: (deadlock in acquire) semaphore held across await

ID: python/asyncio-semaphore-deadlock

Also available as: JSON · Markdown · 中文
80%Fix Rate
83%Confidence
0Evidence
2024-10-08First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
3.8 active — — —
3.9 active — — —
3.10 active — — —
3.11 active — — —
3.12 active — — —

Root Cause

A coroutine acquired an asyncio.Semaphore and then awaited another coroutine that also tried to acquire the same semaphore, causing a self-deadlock.

generic

中文

某个协程获取了 asyncio.Semaphore，随后 await 另一个也尝试获取同一信号量的协程，导致自我死锁。

Workarounds

  1. 90% success 
    Never hold a semaphore across another acquire of the same semaphore; restructure so the inner call uses a separate limiter:

outer = asyncio.Semaphore(10)
inner = asyncio.Semaphore(10)
  2. 75% success 
    Use a re-entrant pattern with contextvars to track ownership.
  3. 85% success 
    Add structured logging around acquire/release to find the cycle:

async with sem:
    logger.info('acquired', extra={'task': id(asyncio.current_task())})

Dead Ends

Common approaches that don't work:

  1. 70% fail

    If the same task re-enters, it still deadlocks at count+1.

  2. 60% fail

    Detects the deadlock but doesn't fix the logic; the underlying work never completes.

  3. 85% fail

    Same re-entrancy problem, now with count=1.