使用atexit()关闭Flask应用时删除AstraDB表触发RuntimeError

我在Flask应用中通过atexit()注册了删除AstraDB表的函数，代码如下：

app = Flask(__name__)

# AstraDB connection and vector store initialization
def initialize_astra_vector_store():
    cassio.init(token=ASTRA_DB_APPLICATION_TOKEN, database_id=ASTR_DB_ID)
    
    embedding = OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
    astra_vector_store = Cassandra(
        embedding=embedding,
        table_name=table_name,
        session=None,
        keyspace=None
    )
    print('----------------------table created as table_name:', table_name)
    return astra_vector_store

# flask app routes
...

def delete_table():
    print('----------------------inside delete_table')
    db = AstraDB(
        token=os.getenv('ASTRA_DB_APPLICATION_TOKEN'),
        api_endpoint='<api-endpoint>',
    )
    print('----------------------table_name:', table_name)      
    # Drop the table created for this session
    db.delete_collection(collection_name=table_name)
    print("----------------------APP EXITED----------------------")
atexit.register(delete_table)

if __name__ == '__main__':
    app.run(debug=False)

关闭应用时，AstraDB中的表已成功删除，但出现如下错误：

Exception in thread Task Scheduler:
Traceback (most recent call last):
  File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\threading.py", line 1038, in _bootstrap_inner
    self.run()
  File "cassandra\cluster.py", line 4239, in cassandra.cluster._Scheduler.run
  File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\concurrent\futures\thread.py", line 167, in submit
    raise RuntimeError('cannot schedule new futures after shutdown')
RuntimeError: cannot schedule new futures after shutdown
Exception in thread Task Scheduler:
Traceback (most recent call last):
  File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\threading.py", line 1038, in _bootstrap_inner
    self.run()
  File "cassandra\cluster.py", line 4239, in cassandra.cluster._Scheduler.run
  File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\concurrent\futures\thread.py", line 167, in submit
    raise RuntimeError('cannot schedule new futures after shutdown')
RuntimeError: cannot schedule new futures after shutdown

• 显式关闭Cassandra会话：在删除表操作前，先关闭初始化时创建的Cassandra会话，避免其线程池在程序 shutdown 后仍尝试调度任务。修改delete_table函数：

  def delete_table():
      print('----------------------inside delete_table')
      # 关闭Cassandra会话
      from cassio.config import check_resolve_session
      session = check_resolve_session()
      if session:
          session.shutdown()
      # 执行删除表操作
      db = AstraDB(
          token=os.getenv('ASTRA_DB_APPLICATION_TOKEN'),
          api_endpoint='<api-endpoint>',
      )
      print('----------------------table_name:', table_name)      
      db.delete_collection(collection_name=table_name)
      print("----------------------APP EXITED----------------------")
  

• 改用Flask的应用上下文销毁钩子：放弃atexit，使用Flask原生的teardown_appcontext装饰器，它更贴合Web应用的生命周期，能避免线程调度冲突：

  @app.teardown_appcontext
  def delete_table(error=None):
      print('----------------------inside delete_table')
      # 关闭Cassandra会话
      from cassio.config import check_resolve_session
      session = check_resolve_session()
      if session:
          session.shutdown()
      # 执行删除表操作
      db = AstraDB(
          token=os.getenv('ASTRA_DB_APPLICATION_TOKEN'),
          api_endpoint='<api-endpoint>',
      )
      print('----------------------table_name:', table_name)      
      db.delete_collection(collection_name=table_name)
      print("----------------------APP EXITED----------------------")
  

  注意：删除原来的atexit.register(delete_table)代码。

• 遵循资源释放顺序：始终先关闭数据库会话，再执行删除操作，确保Cassandra的线程池不会在程序 shutdown 阶段继续提交任务，从根源上消除报错。

内容的提问来源于stack exchange，提问作者Hamza Khan