issue with connection pooling when using QueuePool (via sqlalchemy)

[akhilputhiry]

• asyncpg version: 0.29.0
• PostgreSQL version: 16.2
• Do you use a PostgreSQL SaaS? If so, which? Can you reproduce the issue with a local PostgreSQL install?: Both with AWS RDS and Local Installation
• Python version: 3.12.1
• Platform: aarch64-apple-darwin23.2.0
• Do you use pgbouncer?:
• Did you install asyncpg with pip?: poetry
• If you built asyncpg locally, which version of Cython did you use?: NA
• Can the issue be reproduced under both asyncio and uvloop?: Didnt Try

We have a web app using sqlalchemy. If we use NullPool (disable connection pooling) everything works fine. The moment we enable QueuePool the app gets stuck

time=2024-02-29 14:48:01,669 level=INFO request-id=b1de3f74dac04a9799934a671cf9e0d6 user-id=2319885380 message=Permission datalake/roles/user granted for user 2319885380
time=2024-02-29 14:48:01,669 level=INFO request-id=b1de3f74dac04a9799934a671cf9e0d6 user-id=2319885380 message=creating database session
time=2024-02-29 14:48:01,669 level=INFO request-id=b1de3f74dac04a9799934a671cf9e0d6 user-id=2319885380 message=database session created successfully
time=2024-02-29 14:48:01,669 level=INFO request-id=b1de3f74dac04a9799934a671cf9e0d6 user-id=2319885380 message=Fetching datalake environment
2024-02-29 14:48:01,677 DEBUG sqlalchemy.pool.impl.QueuePool Created new connection <AdaptedConnection <asyncpg.connection.Connection object at 0x114136300>>
2024-02-29 14:48:01,680 DEBUG sqlalchemy.pool.impl.QueuePool Created new connection <AdaptedConnection <asyncpg.connection.Connection object at 0x1141365d0>>

[2024-02-29 14:52:27 +0530] [61517] [CRITICAL] WORKER TIMEOUT (pid:61518)
[2024-02-29 14:52:27 +0530] [61517] [ERROR] Worker (pid:61518) was sent SIGABRT!
[2024-02-29 14:52:27 +0530] [62957] [INFO] Booting worker with pid: 62957
[2024-02-29 14:52:42 +0530] [62957] [INFO] Started server process [62957]
[2024-02-29 14:52:42 +0530] [62957] [INFO] Waiting for application startup.

[elprans]

Should this be an SQLAlchemy bug instead?

[akhilputhiry]

Looks we misconfigured the sqlalchemy to use QueuePool instead of AsyncAdaptedQueuePool which was causing the issue