Add docker build, update docs for using docker images
4.1 KiB
Docker Containers
In general, you should not have Pipenv inside a linux container image, since it is a build tool. If you want to use it to build, and install the run time dependencies for your application, you can use a multistage build for creating a virtual environment with your dependencies.
In this approach, Pipenv in installed in the base layer, and it is used to create the virtual
environment. In a later stage, in a runtime layer the virtual environment
is copied from the base layer, the layer containing pipenv and other build
dependencies is discarded.
This results in a smaller image, which can still run your application.
Here is an example Dockerfile, which you can use as a starting point for
doing a multistage build for your application:
FROM docker.io/oz123/pipenv:3.11-v2023-6-26 AS builder
# Tell pipenv to create venv in the current directory
ENV PIPENV_VENV_IN_PROJECT=1
# Pipfile contains requests
ADD Pipfile.lock Pipfile /usr/src/
WORKDIR /usr/src
# NOTE: If you install binary packages required for a python module, you need
# to install them again in the runtime. For example, if you need to install pycurl
# you need to have pycurl build dependencies libcurl4-gnutls-dev and libcurl3-gnutls
# In the runtime container you need only libcurl3-gnutls
# RUN apt install -y libcurl3-gnutls libcurl4-gnutls-dev
RUN /root/.local/bin/pipenv sync
RUN /usr/src/.venv/bin/python -c "import requests; print(requests.__version__)"
FROM docker.io/python:3.11 AS runtime
RUN mkdir -v /usr/src/.venv
COPY --from=builder /usr/src/.venv/ /usr/src/.venv/
RUN /usr/src/.venv/bin/python -c "import requests; print(requests.__version__)"
# HERE GOES ANY CODE YOU NEED TO ADD TO CREATE YOUR APPLICATION'S IMAGE
# For example
# RUN apt install -y libcurl3-gnutls
# RUN adduser --uid 123123 coolio
# ADD run.py /usr/src/
WORKDIR /usr/src/
USER coolio
CMD ["./.venv/bin/python", "-m", "run.py"]
Pipenv is not meant to run as root. However, in the multistage build above
it is done nevertheless. A calculated risk, since the intermediate image
is discarded.
The runtime image later shows that you should create a user and user it to
run your application.
**Once again, you should not run pipenv as root (or Admin on Windows) normally.
This could lead to breakage of your Python installation, or even your complete
OS.**
When you build an image with this example (assuming requests is found in Pipfile), you
will see that requests is installed in the runtime image:
$ sudo docker build --no-cache -t oz/123:0.1 .
Sending build context to Docker daemon 1.122MB
Step 1/12 : FROM docker.io/python:3.9 AS builder
---> 81f391f1a7d7
Step 2/12 : RUN pip install --user pipenv
---> Running in b83ed3c28448
... trimmed ...
---> 848743eb8c65
Step 4/12 : ENV PIPENV_VENV_IN_PROJECT=1
---> Running in 814e6f5fec5b
Removing intermediate container 814e6f5fec5b
---> 20167b4a13e1
Step 5/12 : ADD Pipfile.lock Pipfile /usr/src/
---> c7632cb3d5bd
Step 6/12 : WORKDIR /usr/src
---> Running in 1d75c6cfce10
Removing intermediate container 1d75c6cfce10
---> 2dcae54cc2e5
Step 7/12 : RUN /root/.local/bin/pipenv sync
---> Running in 1a00b326b1ee
Creating a virtualenv for this project...
... trimmed ...
✔ Successfully created virtual environment!
Virtualenv location: /usr/src/.venv
Installing dependencies from Pipfile.lock (fe5a22)...
... trimmed ...
Step 8/12 : RUN /usr/src/.venv/bin/python -c "import requests; print(requests.__version__)"
---> Running in 3a66e3ce4a11
2.27.1
Removing intermediate container 3a66e3ce4a11
---> 1db657d0ac17
Step 9/12 : FROM docker.io/python:3.9 AS runtime
... trimmed ...
Step 12/12 : RUN /usr/src/venv/bin/python -c "import requests; print(requests.__version__)"
---> Running in fa39ba4080c5
2.27.1
Removing intermediate container fa39ba4080c5
---> 2b1c90fd414e
Successfully built 2b1c90fd414e
Successfully tagged oz/123:0.1