I was using asyncio inside a view function for Flask to run some asynchronous functions and met an error.
Here is the demo code:
import asyncio
from flask import Flask, jsonify
app = Flask(__name__)
@app.route("/toy", methods=["GET"])
def index():
loop = asyncio.get_event_loop()
result = loop.run_until_complete(hello())
return jsonify({"result": result})
async def hello():
await asyncio.sleep(1)
return 1
if __name__ == "__main__":
app.run(host="0.0.0.0", port=4567, debug=False)
When I request this service, I get the following error from asyncio.event:
RuntimeError: There is no current event loop in thread ‘Thread-1’.
for the following line of code:
loop = asyncio.get_event_loop()
From the asyncio documentation on default event loop policy:
If the current thread doesn’t already have an event loop associated with it, the default policy’s get_event_loop() method creates one when called from the main thread, but raises RuntimeError otherwise
So it seems that the view function index()
is not run in the main thread. As
a result, there is no event loop associated with the current thread, hence the
error message. To verify this, we can use the threading
module to help us
find which thread we are currently in:
import threading
import asyncio
from flask import Flask, jsonify
print(f"In flask global level: {threading.current_thread().name}")
app = Flask(__name__)
@app.route("/toy", methods=["GET"])
def index():
print(f"Inside flask function: {threading.current_thread().name}")
loop = asyncio.get_event_loop()
result = loop.run_until_complete(hello())
return jsonify({"result": result})
async def hello():
await asyncio.sleep(1)
return 1
if __name__ == "__main__":
app.run(host="0.0.0.0", port=4567, debug=False)
When running the flask app, we can see the following message:
In flask global level: MainThread
So the statement outside the view function is run in the main thread.
When we request the service, we can see the following message from the view function:
Inside flask function: Thread-1
This verifies that Flask is indeed running the view functions in a separate thread other than the main thread.
In the Flask.run()
doc, this is also briefly mentioned in the changelog
part:
Changed in version 1.0: If installed, python-dotenv will be used to load environment variables from .env and .flaskenv files.
If set, the FLASK_ENV and FLASK_DEBUG environment variables will override env and debug.
Threaded mode is enabled by default.
Under the hood, Flask.run()
is using
werkzeug.serving.run_simple()
to server the app and set the threaded
option to True
by default:
threaded – should the process handle each request in a separate thread?
What does the threaded mode mean? It means that the server can serve requests in a non-blocking fashion, i.e., it does not need to wait for one request to finish to process another request. That is often preferred than the single-threaded mode. Note, however, that the development server is not for production purposes. We need to use dedicated WSGI server such as gunicorn and uWSGI in production for better performance.