Skip to content

Tecktron/docker-python-bjoern

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Python Bjoern Docker Container

A Docker container to run a WSGI Python application using Bjoern. Images support python 3.6+ and are based on the official python containers. The -slim versions are based on the similarly named python versions.

Pull from Docker Hub

View on GitHub

How to use

  • You don't need to clone the GitHub repo. You can use this image as a base image for other images, using this in your Dockerfile:
FROM tecktron/python-bjoern:latest

COPY ./ /app

It will expect a file at /app/app/wsgi.py.

Or otherwise a file at /app/wsgi.py.

And will expect it to contain a variable application with your "WSGI" application.

Then you can build your image from the directory that has your Dockerfile, e.g:

docker build -t myimage ./

Options

All options can be set using environment variables. These can be passed either in a wrapper dockerfile, passing in a .env file or passing them with the -e flag to the docker call.

Prestart Script

If you need to run any startup commands before Bjoern runs (an example might be running migrations) you can override the prestart.sh script. This script should live within the /app directory in the container. The image will automatically detect and run it before starting Bjoern.

Variables

MODULE_NAME

The Python "module" (file) to be imported by Bjoern, this module would contain the actual application in a variable.

By default:

  • app.wsgi if there's a file /app/app/wsgi.py or
  • wsgi if there's a file /app/wsgi.py

For example, if your main file was at /app/custom_app/custom_script.py, you could set it like:

docker run -d -p 80:80 -e MODULE_NAME="custom_app.custom_script" myimage

VARIABLE_NAME

The variable inside of the Python module that contains the WSGI application.

By default:

  • application

For example, if your main Python file has something like:

from flask import Flask
api = Flask(__name__)

@api.route("/")
def hello():
    return "Hello World from Flask"

In this case api would be the variable with the "WSGI application". You could set it like:

docker run -d -p 80:80 -e VARIABLE_NAME="api" myimage

APP_MODULE

The string with the Python module and the variable name passed to Bjoern.

By default, set based on the variables MODULE_NAME and VARIABLE_NAME:

  • app.wsgi:application or
  • wsgi:application

You can set it like:

docker run -d -p 80:80 -e APP_MODULE="custom_app.custom_script:api" myimage

Host & Port Setup

By default, bjoern has been setup to server on all hostnames on port 80 using address 0.0.0.0. This works for most applications using the basic setups listed above.

You may have different needs, so you can adjust and manipulate this by passing in environment variables to adjust the settings.

BJOERN_HOST / BJOERN_PORT

Pass the host and port separately as BJOERN_HOST and/or BJOERN_PORT. If the port is left out, it will default to 80.

docker run -d -p 80:8080 -e BJOERN_PORT=8080 myimage

Other

BJOERN_REUSEPORT

Bjoern supports enabling SO_REUSEPORT if it's available on the system (Python's base images use Debian which supports this). This has been enabled by default in this image. What does this do? In short, if an application has terminated for whatever reason (quit, killed, etc) the system may still hold onto the port while it's clearing the buffer out. This process blocks other applications including new instances of the same application from binding to the port during this time (which can be a lengthy process, in some cases up to 2 min). Setting this flag when binding to the port allows other applications to bind to the port during this timeout period. While there is the possibility that it could cause strange behavior, those are extremely rare. Generally speaking, the perception is that enabling this behavior is a good thing at least with regard to performance. However, it should be noted that since this image is not using any kind of supervisor to restart the application if it crashes (I find this is usually best left to the cluster controller so it can just replace the pod), this image probably does not truly make use of it.

To disable this behavior, send any value that is not the string of "true" using the BJOERN_REUSEPORT variable.

docker run -d -p 80:80 -e BJOERN_REUSEPORT="false" myimage

Credits

This dockerfile setup is based on https://github.com/tiangolo/meinheld-gunicorn-docker

bjoern is by Jonas Haag: https://github.com/jonashaag/bjoern

Python is by the Python Software Foundation. https://python.org

Docker is by Docker, Inc. https://docker.com

Built using open source software.