Two issues with your approach -

• Remove the -b 0.0.0.0:5000 as you are binding to the wrong port, Gunicorn will automatically bind to the correct port without that bind flag.

• Remove the --relaod flag, that should only ever be used during local development, in production that uses extra unnecessary resources and causes instabilities at best.

Hey brody, thank you for your answer! I tried it, below you see the complete dockerfile:

Use an official Python runtime as a parent image

FROM python:3.10-slim

Set the working directory in the container

WORKDIR /app

Copy the current directory contents into the container at /app

COPY . /app

Install any needed packages specified in requirements.txt

RUN pip install --no-cache-dir -r requirements.txt

Make port 5000 available to the world outside this container

EXPOSE 5000

Define environment variable to handle Flask and SocketIO

ENV FLASK_ENV=production

Command to run the app using Gunicorn with eventlet workers

CMD ["gunicorn", "-k", "eventlet", "-w", "1", "--timeout", "180", "app:app","--log-level","debug"]

Unfortunately Gunicorn still started up with sync workers. I would love to use railway, would be cool to have that fixed.

For clarity this is not a platform issue, this is a config issue.

It's likely you have a start command set elsewhere that is overwriting the CMD command in your Dockerfile, check for other start commands in a Procfile, railway.json, or your service settings.