[Bug]: RuntimeError: asyncio.run() cannot be called from a running event loop with Alpaca-py data/trading streams #618
Description
Is there an existing issue for this?
- I have searched the existing issues
Current Behavior
I am consistently encountering a RuntimeError: asyncio.run() cannot be called from a running event loop when attempting to initialize and run the live StockDataStream and TradingStream in my Python trading bot.
Problem Description:
My bot initiates an asyncio event loop using asyncio.run(main()). Within this main loop, I attempt to start the data_stream.run() and trading_stream.run() methods from the alpaca-py library as separate asyncio.create_task() operations, in order to receive real-time market data and order updates.
The error occurs as soon as the data_stream.run() method (and presumably the trading_stream.run() as well) tries to initialize. The traceback indicates an internal call to asyncio.run() within the alpaca-py library itself, resulting in a nesting error because an asyncio loop is already active.
Relevant Traceback (core of the error):
Traceback (most recent call last):
File "C:\Users\Gebruiker\miniconda3\envs\trading_bot_env_v4\lib\site-packages\alpaca\data\live\websocket.py", line 366, in run
asyncio.run(self._run_forever())
File "C:\Users\Gebruiker\miniconda3\envs\trading_bot_env_v4\lib\asyncio\runners.py", line 33, in run
raise RuntimeError(
RuntimeError: asyncio.run() cannot be called from a running event loop
My Environment and Troubleshooting Attempts:
Python Version: I am using Python 3.10 (installed via Miniconda).
alpaca-py Versions Tested:
0.42.0 (originally) - produced the error.
0.41.0 (after attempting a downgrade) - produced the exact same error.
Steps Taken: I attempted to upgrade to the latest available version via pip install --upgrade alpaca-py (which resulted in 0.42.0) and subsequently tried a downgrade to 0.41.0. Both versions exhibit the identical behavior.
Context: Earlier versions of my script that only fetched historical data (via data_client.get_stock_bars()) or just account information did not encounter this issue, as they did not call the run() methods of the live streams.
Expected Behavior:
The expected behavior is that the data_stream.run() and trading_stream.run() methods of alpaca-py should integrate seamlessly into an already active asyncio event loop when started as tasks (asyncio.create_task()), without attempting to call asyncio.run() internally themselves.
Request for Assistance:
I would appreciate guidance on the following:
Is this a known issue/bug in alpaca-py versions 0.41.0 and 0.42.0?
What is the recommended stable version of alpaca-py that correctly works with asyncio.create_task() for live streams within an already running event loop?
Is there a specific method or best practice for starting the StockDataStream and TradingStream that avoids this asyncio.run() nesting problem?
Thank you in advance for your help!
Expected Behavior
No response
SDK Version I encountered this issue in
"I encountered this issue in alpaca-py versions 0.42.0 (which was the latest when I first checked and also after attempting an upgrade) and 0.41.0 (after a downgrade attempt). Both versions exhibit the same RuntimeError: asyncio.run() cannot be called from a running event loop when attempting to start the live streams."
Steps To Reproduce
Set up a Python 3.10 environment (e.g., via Miniconda).
Install alpaca-py version 0.41.0 or 0.42.0 (pip install alpaca-py==0.41.0).
Ensure Alpaca API Key ID, Secret Key, and Base URL (https://paper-api.alpaca.markets) are set as environment variables.Filled out the Steps to Reproduce section?
- I have entered valid steps to reproduce my issue or have attached a minimally reproducible case in code that shows my issue happening; and understand that without this my issue will be flagged as invalid and closed after 30 days.
Anything else?
No response