Application Flow Stuck Unless Connection is Explicitly Closed

• asyncpg version: 0.30.0
• PostgreSQL version: 16
• Do you use a PostgreSQL SaaS? If so, which?: Yes, I am using NeonDB
• Can you reproduce the issue with a local PostgreSQL install?: Didn't attempt it.
• Python version: 3.11.0
• Platform: Windows
• Do you use pgbouncer?: No
• Did you install asyncpg with pip?: Yes
• If you built asyncpg locally, which version of Cython did you use?: [Your Cython version if applicable]
• Can the issue be reproduced under both asyncio and
  uvloop?: [Yes/No] I did not try uvloop

I'm encountering an issue where the flow of my application gets stuck unless the database connection is explicitly closed after a query, even though it should be managed automatically by the context manager.

Code Example:

async def read_record(self, request_details: dict):
    """
    Retrieve records based on conditions and request_type.
    
    Args:
        request_details (dict): The conditions including request_type.
    
    Returns:
        list: The records that match the conditions.
    """
    table = self.get_table_name(request_details['request_type'])
    conditions = {key: request_details[key] for key in request_details if key != 'request_type'}
    condition_str = " AND ".join(f"{key} = ${i+1}" for i, key in enumerate(conditions.keys()))
    query = f"SELECT * FROM {table} WHERE {condition_str};"

    async with self.get_connection() as conn:
        conn.add_log_listener(lambda connection, message: logger.info("PostgreSQL log message: %s", message))
        result = await conn.fetch(query, *conditions.values())
        logger.info("Fetched records: %s", result)
        # await conn.close() # get's stuck without this line

    return result

Commencez par l’exemple asyncpg 0.30.0 fourni et reproduisez-le avec Python 3.11, PostgreSQL 16 et Windows, en vérifiant d’abord si le problème se produit avec une installation locale de PostgreSQL. Étudiez le comportement du gestionnaire de contexte asynchrone autour de self.get_connection() et comparez-le au cas avec un appel explicite à conn.close() ; le travail est terminé lorsque le flux s’achève sans nécessiter cette fermeture explicite.