Skip to content

Latest commit

 

History

History
384 lines (279 loc) · 15.9 KB

README.md

File metadata and controls

384 lines (279 loc) · 15.9 KB
Raw

Python 3.6 container image

This container image includes Python 3.6 as a S2I base image for your Python 3.6 applications. Users can choose between RHEL and CentOS based builder images. The RHEL images are available in the Red Hat Container Catalog, the CentOS images are available on Quay.io, and the Fedora images are available in Quay.io. The resulting image can be run using podman or docker.

Note: while the examples in this README are calling podman, you can replace any such calls by docker with the same arguments

Description

Python 3.6 available as container is a base platform for building and running various Python 3.6 applications and frameworks. Python is an easy to learn, powerful programming language. It has efficient high-level data structures and a simple but effective approach to object-oriented programming. Python's elegant syntax and dynamic typing, together with its interpreted nature, make it an ideal language for scripting and rapid application development in many areas on most platforms.

This container image includes an npm utility (see base image repository), so users can use it to install JavaScript modules for their web applications. There is no guarantee for any specific npm or nodejs version, that is included in the image; those versions can be changed anytime and the nodejs itself is included just to make the npm work.

Usage in Openshift

For this, we will assume that you are using one of the supported images available via imagestream tags in Openshift, eg. python:3.6-ubi8 Building a simple python-sample-app application in Openshift can be achieved with the following step:

```
oc new-app python:3.6-ubi8~https://github.com/sclorg/django-ex.git
```

Accessing the application:

$ oc get pods
$ oc exec <pod> -- curl 127.0.0.1:8080

Source-to-Image framework and scripts

This image supports the Source-to-Image (S2I) strategy in OpenShift. The Source-to-Image is an OpenShift framework which makes it easy to write images that take application source code as an input, use a builder image like this Python container image, and produce a new image that runs the assembled application as an output.

To support the Source-to-Image framework, important scripts are included in the builder image:

  • The /usr/libexec/s2i/assemble script inside the image is run to produce a new image with the application artifacts. The script takes sources of a given application and places them into appropriate directories inside the image. It utilizes some common patterns in Perl application development (see the Environment variables section below).
  • The /usr/libexec/s2i/run script is set as the default command in the resulting container image (the new image with the application artifacts). It runs your application according to settings in APP_MODULE, APP_FILE or APP_SCRIPT environment variables or it tries to detect the best way automatically.

Building an application using a Dockerfile

Compared to the Source-to-Image strategy, using a Dockerfile is a more flexible way to build a Python container image with an application. Use a Dockerfile when Source-to-Image is not sufficiently flexible for you or when you build the image outside of the OpenShift environment.

To use the Python image in a Dockerfile, follow these steps:

1. Pull a base builder image to build on

podman pull registry.access.redhat.com/ubi8/python-36

2. Pull and application code

An example application available at https://github.com/sclorg/django-ex.git is used here. Feel free to clone the repository for further experiments. You can also take a look at code examples in s2i-python-container repository: https://github.com/sclorg/s2i-python-container/tree/master/examples

git clone https://github.com/sclorg/django-ex.git app-src

3. Prepare an application inside a container

This step usually consists of at least these parts:

  • putting the application source into the container
  • installing the dependencies
  • setting the default command in the resulting image

For all these three parts, users can either setup all manually and use commands python and pip explicitly in the Dockerfile, or users can use the Source-to-Image scripts inside the image.

The manual way comes with the highest level of flexibility but requires you to know how to work with modules or software collections manually, how to setup virtual environment with the right version of Python and many more. On the other hand, using Source-to-Image scripts makes your Dockerfile prepared for a future flawless switch to a newer or different platform.

To use the Source-to-Image scripts and build an image using a Dockerfile, create a Dockerfile with this content:

FROM registry.access.redhat.com/ubi8/python-36

# Add application sources to a directory that the assemble script expects them
# and set permissions so that the container runs without root access
USER 0
ADD app-src /tmp/src
RUN /usr/bin/fix-permissions /tmp/src
USER 1001

# Install the dependencies
RUN /usr/libexec/s2i/assemble

# Set the default command for the resulting image
CMD /usr/libexec/s2i/run

If you decide not to use the Source-to-Image scripts, you will need to manually tailor the Dockerfile to your application and its needs. Example Dockerfile for a simple Django application:

FROM registry.access.redhat.com/ubi8/python-36

# Add application sources with correct permissions for OpenShift
USER 0
ADD app-src .
RUN chown -R 1001:0 ./
USER 1001

# Install the dependencies
RUN pip install -U "pip>=19.3.1" && \
    pip install -r requirements.txt && \
    python manage.py collectstatic --noinput && \
    python manage.py migrate

# Run the application
CMD python manage.py runserver 0.0.0.0:8080

4. Build a new image from a Dockerfile prepared in the previous step

podman build -t python-app .

5. Run the resulting image with final application

podman run -d python-app

Environment variables

To set these environment variables, you can place them as a key value pair into a .s2i/environment file inside your source code repository.

  • APP_SCRIPT

    Used to run the application from a script file. This should be a path to a script file (defaults to app.sh unless set to null) that will be run to start the application.

  • APP_FILE

    Used to run the application from a Python script. This should be a path to a Python file (defaults to app.py unless set to null) that will be passed to the Python interpreter to start the application.

  • APP_MODULE

    Used to run the application with Gunicorn, as documented here. This variable specifies a WSGI callable with the pattern MODULE_NAME:VARIABLE_NAME, where MODULE_NAME is the full dotted path of a module, and VARIABLE_NAME refers to a WSGI callable inside the specified module. Gunicorn will look for a WSGI callable named application if not specified.

    If APP_MODULE is not provided, the run script will look for a wsgi.py file in your project and use it if it exists.

    If using setup.py for installing the application, the MODULE_NAME part can be read from there. For an example, see setup-test-app.

  • APP_HOME

    This variable can be used to specify a sub-directory in which the application to be run is contained. The directory pointed to by this variable needs to contain wsgi.py (for Gunicorn) or manage.py (for Django).

    If APP_HOME is not provided, the assemble and run scripts will use the application's root directory.

  • APP_CONFIG

    Path to a valid Python file with a Gunicorn configuration file.

  • DISABLE_MIGRATE

    Set this variable to a non-empty value to inhibit the execution of 'manage.py migrate' when the produced image is run. This only affects Django projects. See "Handling Database Migrations" section of Django blogpost on OpenShift blog on suggestions how/when to run DB migrations in OpenShift environment. Most importantly, note that running DB migrations from two or more pods might corrupt your database.

  • DISABLE_COLLECTSTATIC

    Set this variable to a non-empty value to inhibit the execution of 'manage.py collectstatic' during the build. This only affects Django projects.

  • DISABLE_SETUP_PY_PROCESSING / DISABLE_PYPROJECT_TOML_PROCESSING

    Set this to a non-empty value to skip processing of setup.{py,cfg} or pyproject.toml script if you use -e . in requirements.txt to trigger its processing or you don't want your application to be installed into site-packages directory.

  • ENABLE_PIPENV

    Set this variable to use Pipenv, the higher-level Python packaging tool, to manage dependencies of the application. This should be used only if your project contains properly formated Pipfile and Pipfile.lock.

  • PIN_PIPENV_VERSION

    Set this variable together with ENABLE_PIPENV to use a specific version of Pipenv. If not set, the latest stable version from PyPI is installed. For example PIN_PIPENV_VERSION=2018.11.26 installs pipenv==2018.11.26.

  • ENABLE_MICROPIPENV

    Set this variable to use micropipenv, a lightweight wrapper for pip to support requirements.txt, Pipenv and Poetry lock files or converting them to pip-tools compatible output. Designed for containerized Python applications. Available only for Python 3 images.

  • ENABLE_INIT_WRAPPER

    Set this variable to a non-empty value to make use of an init wrapper. This is useful for servers that are not capable of reaping zombie processes, such as Django development server or Tornado. This option can be used together with APP_SCRIPT or APP_FILE. It never applies to Gunicorn used through APP_MODULE as Gunicorn reaps zombie processes correctly.

  • PIP_INDEX_URL

    Set this variable to use a custom index URL or mirror to download required packages during build process. This affects packages listed in requirements.txt. It also affects the installation of pipenv and micropipenv and the update of pip in the container, though if not found in the custom index, the container will try to install/update them from upstream PyPI afterwards.

  • PORT

    HTTP(S) port your application should listen on. The default is 8080. PORT is used only for Django development server and for Gunicorn with the default configutation (no APP_CONFIG or GUNICORN_CMD_ARGS specified).

  • UPGRADE_PIP_TO_LATEST

    Set this variable to a non-empty value to have the 'pip' program and related python packages (setuptools and wheel) be upgraded to the most recent version before any Python packages are installed. If not set, the container will use the stable pip version this container was built with, taken from a recent Fedora release.

  • WEB_CONCURRENCY

    Set this to change the default setting for the number of workers. By default, this is set to the number of available cores times 2, capped at 12.

Source repository layout

You do not need to change anything in your existing Python project's repository. However, if these files exist they will affect the behavior of the build process:

  • requirements.txt

    List of dependencies to be installed with pip. The format is documented here.

  • Pipfile

    The replacement for requirements.txt, project is currently under active design and development, as documented here. Set ENABLE_PIPENV environment variable to true in order to process this file.

  • setup.py

    Configures various aspects of the project, including installation of dependencies, as documented here. For most projects, it is sufficient to simply use requirements.txt or Pipfile. Set DISABLE_SETUP_PY_PROCESSING environment variable to true in order to skip processing of this file.

Run strategies

The container image produced by s2i-python executes your project in one of the following ways, in precedence order:

  • Gunicorn

    The Gunicorn WSGI HTTP server is used to serve your application in the case that it is installed. It can be installed by listing it either in the requirements.txt file or in the install_requires section of the setup.py file.

    If a file named wsgi.py is present in your repository, it will be used as the entry point to your application. This can be overridden with the environment variable APP_MODULE. This file is present in Django projects by default.

    If you have both Django and Gunicorn in your requirements, your Django project will automatically be served using Gunicorn.

    The default setting for Gunicorn (--bind=0.0.0.0:$PORT --access-logfile=-) is applied only if both $APP_CONFIG and $GUNICORN_CMD_ARGS are not defined.

  • Django development server

    If you have Django in your requirements but don't have Gunicorn, then your application will be served using Django's development web server. However, this is not recommended for production environments.

  • Python script

    This would be used where you provide a Python code file for running you application. It will be used in the case where you specify a path to a Python script via the APP_FILE environment variable, defaulting to a file named app.py if it exists. The script is passed to a regular Python interpreter to launch your application.

  • Application script file

    This is the most general way of executing your application. It will be used in the case where you specify a path to an executable script file via the APP_SCRIPT environment variable, defaulting to a file named app.sh if it exists. The script is executed directly to launch your application.

Hot deploy

If you are using Django, hot deploy will work out of the box.

To enable hot deploy while using Gunicorn, make sure you have a Gunicorn configuration file inside your repository with the reload option set to true. Make sure to specify your config via the APP_CONFIG environment variable.

To change your source code in running container, use podman's (or docker's) exec command:

podman exec -it <CONTAINER_ID> /bin/bash

After you enter into the running container, your current directory is set to /opt/app-root/src, where the source code is located.

See also

Dockerfile and other sources are available on https://github.com/sclorg/s2i-python-container. In that repository you also can find another versions of Python environment Dockerfiles. Dockerfile for CentOS is called Dockerfile, Dockerfile for RHEL7 is called Dockerfile.rhel7, for RHEL8 it's Dockerfile.rhel8, for RHEL9 it's Dockerfile.rhel9 and the Fedora Dockerfile is called Dockerfile.fedora.