Hey there! We've found the following might help you get unblocked faster:

• 🧵 Getting false "Deploy Crashed!" emails during normal deployments

• 🧵 Docker deployment is failing

• 🧵 Host OpenFGA | No Template Available

If you find the answer from one of these, please let us know by solving the thread!

So it's getting shut down and not crashed ?
Does it happen immediately or some fixed or random time ?

As far as I can tell, it shuts down immediately after start up. It does not indicate it crashed, and trying to access it returns a 502.

It does not exhibit this behavior locally or when run on another hosting platform, so I do not think it is my code that is intentionally exiting.

Can you provide the contents of DockerFile?

What's the memory limit for this service ?

Dockerfile is

# Use Python 3.11 Alpine Linux base image

FROM python:3.11-alpine

# Set working directory

WORKDIR /app

# Install system dependencies

RUN apk add --no-cache \

gcc \

musl-dev \

postgresql-dev \

postgresql-client \

libffi-dev

# Copy requirements first for better caching

COPY requirements.txt .

# Install Python dependencies

RUN pip install --no-cache-dir --upgrade pip && \

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

# Copy application code

COPY . .

# Copy startup scripts

COPY entrypoint.sh /entrypoint.sh

COPY start.py /app/start.py

RUN chmod +x /entrypoint.sh && \

chmod +x /app/start.py

# Create non-root user

RUN adduser -D -s /bin/sh appuser && \

chown -R appuser:appuser /app && \

chown appuser:appuser /entrypoint.sh

USER appuser

# Expose port (Railway will provide PORT env var)

EXPOSE 8000

# No HEALTHCHECK in Dockerfile - Railway handles this externally

# Run migrations then start server with Railway optimizations

CMD sh -c "alembic upgrade head && sleep 2 && exec uvicorn app.main:app --host 0.0.0.0 --port \${PORT:-8000} --log-level info --workers 1 --timeout-keep-alive 30"

Overall plan is the hobby plan. I set the limits on the backend to 4 cores / 4 Gigs. It really shouldn't exceed that, it is a pretty simple python service.

Have you configured health checks for this service in railway dashboard ? if yes can you test it by removing them ?

You're going to bump into a lot of sharp edges that Railway doesn't make obvious if you don't read the docs.

I recommend you restructure your dockerfile a bit.
Create an entrypoint.sh in the same directory as your dockerfile and do something like this in it:

Notice: When not in "dev" mode uvicorn will serve on ipv6 -- this is a requirement for your service to be reachable on Railway's internal network (https://docs.railway.com/guides/private-networking#python--uvicorn)

Then in your dockerfile:

Next, make sure you have set the PORT env variable within railway for your service configuration -- Railway doesn't auto-detect this.

And finally most importantly..
Your dockerfile exposes PORT 8000,
BUT your Uvicorn logs from the first message show its on Port 8080 make sure these are aligned! (It's probably just this that's the real problem, but if you don't use ipv6 you'll bump into that next)

One or all of these things should probably fix you up. I can't recommend changing your "CMD" in your dockerfile enough.. it's pretty brittle. Hope this helps!