I am deploying a Python Flask application behind gunicorn on Google Cloud run. The container runs fine locally, but I notice when I build and push the service revision I get an error and my gunicorn workers crash:
ModuleNotFoundError: No module named 'lib'
My directory structure is like so:
├── Dockerfile
├── README.md
├── gunicorn.conf.py
├── lib
│ ├── __init__.py
│ └── misc
│ ├── __init__.py
│ └── daily_message.py
├── requirements.txt
└── server.py
I import the function get_daily_message in server.py like so:
from lib.misc.daily_message import get_daily_message
This all works fine locally and also when I build and run my container image locally. Here is my Dockerfile.
# Use the official lightweight Python image.
# https://hub.docker.com/_/python
FROM python:3.11-slim
# Allow statements and log messages to immediately appear in the logs
ENV PYTHONUNBUFFERED True
# Copy local code to the container image.
ENV APP_HOME /app
WORKDIR $APP_HOME
# Copy the entire project directory to the container.
COPY . ./
# Install production dependencies.
RUN pip install --no-cache-dir -r requirements.txt
# Run the web service on container startup.
CMD ["gunicorn", "--bind", ":8080", "--workers", "2", "--threads", "4", "--timeout", "0", "server:app"]
Here I am to answer my own question. After pulling down my artifact from Artifact Registry and inspecting it, I realized the lib directory was not being copied over. It turns out Google Cloud Build uses the .gitignore file to determine what to copy over along with a .dockerignore. The reason my module could not be found was because it wasn't copied over.
For context I was deploying to Cloud Run:
gcloud builds submit ...
gcloud run deploy ...