Learn production-ready Docker packaging best practices for Python applications in this EuroPython 2020 conference talk. Discover a prioritized, iterative approach to implementing over 60 best practices, focusing on security, automation, correctness, reproducibility, and optimization. Explore techniques such as running as a non-root user, debugging C crashes with faulthandler, speeding up startup with pre-compiled .pyc files, creating smaller images by disabling pip caching, and avoiding Alpine Linux. Gain insights into the complexities of Docker packaging and receive resources for further learning to ensure your Docker images are secure, efficient, and ready for production deployment.
Overview
Syllabus
Intro
Why Docker packaging is complicated
The problem: too much to cover
Today: learn a process
An iterative process
Security Before you can deploy anything publicly, it reeds to be secure . So we do that net
Security: Don't run as root
Security: Other best practices
CI: Tag based on branch . You want to build image for feature branch 123 . You want production not to be impacted.
CI: Other best practices . Once a week, rebuild without caching--pull
Make it debuggable You've started automatically building and probably deploying
Debuggable: Tracebacks on crashes in C code
Debuggable: Other best practices
Improve operational correctness
Operational correctness: Pre- compile bytecode
Reproducible builds: Choose a good base image
Reproducible builds: More best practices
Faster builds: Don't use Alpine Linux
Faster builds: More best practices
Smaller images: Disable pip's caching
Smaller images: Other best practices
Taught by
EuroPython Conference
