继承 APIView 的方法,发现每个请求都是单独的 event loop ,
wait task 318760 future data, future: 140251998671568, loop:140251997385056
set task 318760 future data, future: 140251998671568, loop:140251997385056
wait task 318762 future data, future: 140251998681952, loop:140251998919920
set task 318761 future data, future: 140251997477472, loop:140251997378624
你在 A 请求里
future = loop.create_future()
result = await asyncio.wait_for(future, timeout=timeout)
把 future 传给另一个线程(比如 websocket executor),还得把 loop 一起传过去,
async with task_store.lock:
future, loop = task_store.task_results.pop(task_id, (None, None))
if future:
logger.info(f"set task {task_id} future data, future: {id(future)}, loop:{id(loop)}")
loop.call_soon_threadsafe(future.set_result, result_data)
else:
logger.warning(f"can not find active task {task_id}")
相当于每个请求还是独立的线程,然后在线程里搞 async io 。这有啥意义啊。
1
sagaxu 2 天前 1
https://docs.djangoproject.com/en/5.1/topics/async/
Under a WSGI server, async views will run in their own, one-off event loop. This means you can use async features, like concurrent async HTTP requests, without any issues, but you will not get the benefits of an async stack. The main benefits are the ability to service hundreds of connections without using Python threads. This allows you to use slow streaming, long-polling, and other exciting response types. If you want to use these, you will need to deploy Django using ASGI instead. |
2
guoguobaba OP @sagaxu 我就是在 asgi 模式下运行的,仍然是如此。
|
3
sagaxu 2 天前
@guoguobaba
When running in a mode that does not match the view (e.g. an async view under WSGI, or a traditional sync view under ASGI), Django must emulate the other call style to allow your code to run. This context-switch causes a small performance penalty of around a millisecond. Django Rest Framewok 应该是不支持 async 的,你得用这个 https://github.com/em1208/adrf |
4
OrenZ 2 天前
这个项目是使用异步视图实现的,可供参考,如有问题可发邮件交流(邮件地址在 Github 个人页上)
https://github.com/OVINC-CN/ChatGPTAPI |