Uvicorn hangs in development --reload mode with 1 worker when 2 simultaneous requests are sent

[peteryin21]

When developing locally with Uvicorn in --reload mode, we have started to see an issue where the Uvicorn server hangs and doesn't respond anymore when we our frontend submits 2 concurrent API requests

Command

uvicorn our_app:app --host=0.0.0.0 --port=8000 --log-level=trace --reload

and the trace logs when we send two api endpoint requests

TRACE:    127.0.0.1:54403 - HTTP connection made
TRACE:    127.0.0.1:54404 - HTTP connection made
TRACE:    127.0.0.1:54403 - ASGI [10] Started scope={'type': 'http', 'asgi': {'version': '3.0', 'spec_version': '2.4'}, 'http_version': '1.1', 'server': ('127.0.0.1', 8000), 'client': ('127.0.0.1', 54403), 'scheme': 'http', 'method': 'GET', 'root_path': '', 'path': '/api/v1/projects/', 'raw_path': b'/api/v1/projects/', 'query_string': b'', 'headers': '<...>', 'state': {}}
TRACE:    127.0.0.1:54404 - ASGI [11] Started scope={'type': 'http', 'asgi': {'version': '3.0', 'spec_version': '2.4'}, 'http_version': '1.1', 'server': ('127.0.0.1', 8000), 'client': ('127.0.0.1', 54404), 'scheme': 'http', 'method': 'GET', 'root_path': '', 'path': '/api/v1/file_count/', 'raw_path': b'/api/v1/file_count/', 'query_string': b'', 'headers': '<...>', 'state': {}}

after which the server becomes unresponsive.

These endpoints are synchronous FastAPI endpoints that make DB query calls.

I confirmed that

• The Uvicorn server functions fine if I only do one of these endpoints at a time
• Increasing the workers to 3 fixes the issue (but this is not possible with --reload, which only supports 1 worker)
• This issue doesn't occur on production servers where we deploy and run uvicorn with multiple workers

Python version: 3.11.1
Uvicorn version: Tried with 0.29.0, 0.28.1, and 0.23.3
FastAPI version: Tried with 0.103.2 and 0.110.1

Anyone experience this or know what might be going on? Let me know if I should open an issue.