Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Hello! does Prefect support asynchronous tasks?

Hey <@U02EH1ALVNJ>, I think <https://prefect-community.slack.com/archives/CL09KU1K7/p1631452473109900?thread_ts=1631449202.107400&amp;cid=CL09KU1K7|this> is a good answer

We are also working on features to support asynchronous tasks :slightly_smiling_face:

how can i get event loop inside task run?

How would you do it without Prefect? `get_event_loop()`?

you mean use it like this?
```@task
def some_task():
    loop = asyncio.get_event_loop()   loop.run_until_complete(asyncio.wait(futures))   

with Flow("flow_name") as fl:
    some_task()```

when i try to hide event loop in task it doesn't see any coroutine and just finish

Let me ask the team if anyone if familiar with async

Hey <@U02EH1ALVNJ>
I tried a couple things and had success with the following snippet.
```@task
def async_sleeper(i):
    loop = asyncio.new_event_loop()

    tasks = [loop.create_task(some_coro()) for _ in range(i)]
    loop.run_until_complete(asyncio.wait(tasks))```
I don't know what is happening, But using `loop.create_task` instead of `asyncio.create_task` did the trick for me. I can execute this task in a mapped fashion even.
`DaskExecutor`  works with the above setup, but `LocalDaskExecutor`  does not

I recommend using `asyncio.new_event_loop`  instead of the usual `get_event_loop`. event loops are single threaded by default. python gets confused when multiple threads try to access the single event loop owned by the python process.
Instead, create a fresh event loop at the start of your tasks with `new_event_loop` . Also its probably best if you close the loop when the task finishes.