fastapi-overflow 0.1.4

A small library to patch FastAPI to remove deadlock issues when using sync database sessions.

FastAPI Overflow

When using FastAPI with a synchronous SQLAlchemy session dependency, per the official example https://fastapi.tiangolo.com/tutorial/sql-databases/, high loads can lead to temporary deadlocks, timeouts and poor performance.

This is a known issue, for which a PR is open. This library exists to allow users to easily patch the issue and run their code in the mean time.

We provide 3 primary functions:

• fastapi_overflow.patch which should be called before starting your app in order to patch FastAPI's thread handling
• fastapi_overflow.run_in_threadpool which should be used instead of starlette.concurrency.run_in_threadpool or fastapi.concurrency.run_in_threadpool
• fastapi_overflow.iterate_in_threadpool which should be used instead of starlette.concurrency.iterate_in_threadpool or fastapi.concurrency.iterate_in_threadpool

Usage

Use it like this:

from fastapi import FastAPI
from fastapi_overflow import patch

patch()  # run this before starting your app
app = FastAPI()

Changing the default thread limit

You can increase the number of threads from the default (40) using the fastapi_overflow.set_thread_limit function. With a very high number of threads you might also want to increase the reserve used to prevent deadlocks from its default of 1. You will need to add this to your startup events, or lifetime function:

@app.on_event("startup")
def on_startup():
    set_thread_limit(limit=..., anti_deadlock_reserve=2)
    ...  # any other setup here