uvicorn/uvicorn#1226

README.md

Reproduction steps for uvicorn/uvicorn#1226

Details

This gist contains a Docker Compose configuration, including a simple FastAPI service, and a container with tcping installed, which triggers TCP connections to the service.

To speed up the memory leak reproduction, the tcping container can be scaled up using docker-compose scaling capabilities.

Running

1. Download the entire content of this gist to a folder.
2. Run docker-compose up --build --scale tcping=N (e.g. N=10).
3. On a separate terminal, run docker stats and check memory usage on the API service container.
4. To stop, just Ctrl+C on the terminal where docker-compose is running.

Testing Uvicorn changes

The Docker Compose configuration allows for Uvicorn changes to be tested locally.

1. Clone Uvicorn in a subfolder: git clone git@github.com:encode/uvicorn.git
2. Uncomment the volumes section in the docker-compose.yml file.
3. Make changes to the cloned Uvicorn repository. For example, apply the changes from encode/uvicorn#1192.
4. Run the Docker Compose configuration, following the previous steps.

docker-compose.yml

services:
  api:
    build:
      context: .
      dockerfile: Dockerfile.api
    ports:
      - 8080:8080
    # volumes:
    #   - ./uvicorn/uvicorn:/usr/local/lib/python3.9/site-packages/uvicorn/

  tcping:
    build:
      context: .
      dockerfile: Dockerfile.tcping
    command: ["--port", "8080", "--count", "4294967296", "api"]
    depends_on:
      - api

Dockerfile.api

FROM python:3.9-slim

WORKDIR /app

COPY requirements.txt /app/requirements.txt
RUN pip install -r /app/requirements.txt
RUN pip install "uvicorn[standard]"

COPY main.py ./

EXPOSE 8080

CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8080"]

Dockerfile.tcping

FROM python:3.9-slim

RUN pip install tcping

ENTRYPOINT ["tcping"]

main.py

from fastapi import FastAPI

app = FastAPI()

requirements.txt

fastapi==0.70.0