Dockerfile

# From the base image
# here: https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker
# Based off the example
# here: https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker#dependencies-and-packages
FROM tiangolo/uvicorn-gunicorn-fastapi:python3.7

# Install Poetry
RUN curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | POETRY_HOME=/opt/poetry python && \
    cd /usr/local/bin && \
    ln -s /opt/poetry/bin/poetry && \
    poetry config virtualenvs.create false

# Copy using poetry.lock* in case it doesn't exist yet
COPY ./pyproject.toml ./poetry.lock* /app/

# Install our package and its dependencies
RUN poetry install --no-root --no-dev --extras "production"

COPY ./athology_ml/ /app/athology_ml

# Because our main.py lives somewhere other than the default, we need Xto tell FastAPI where.
# See: https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker#module_name
ENV MODULE_NAME="athology_ml.app.main"
# This is a hack to try to fit within Heroku's RAM limits. Each worker will consume some amount
# of RAM, so my thinking is to restrict the FastAPI image to a single worker to use the least
# amount of RAM.
ENV MAX_WORKERS="1"

# Setup secrethub. First we install the secrethub-cli, then we copy our secret references env file,
# and finally we call a modifed start script that passes these enviornment variables to our app.
# For details, see: https://secrethub.io/docs/start/deploy/other/
RUN curl https://apt.secrethub.io | bash
COPY ./secrethub.env /app/
COPY ./bin/start.sh /