Continuous delivery
of a Python project
with Docker and Ansible
Joost Cassee
PyCon Sette, 16 April 2016
Painless Software
django-analytical
django-geckoboard
django-localeurl
angular-hypermedia
joost@cassee.net
@jcassee
jcassee.com
- Develop code
- Fix bugs
- Make release artifact
- Fix mistakes in release artifact
- QA testing
- Fix bugs
- Make new release
- Deploy to acceptance environment
- Fix mistakes in acceptance environment changes
- Acceptance testing
- Fix bugs, make new release, deploy
- Deploy to production environment
- Fix mistakes in production environment changes
Development
Deployment
Treat infrastructure as code
- Use Ansible to describe the infrastructure
- Keep the description in a Git repository
- Host the repository on GitHub
-
Use Travis CI to build the infrastructure
- Require GitHub pull requests for changes
Any pull request is
-
deployed, for review by non-technical stakeholders
- merged when all stakeholders accept the change
- developers
- testers
- operations
- application managers
- user / customer support
Why Docker?
- Move high-risk operations into the build phase
- Make build process CI-independent
Why Ansible?
Written by the product manager of Puppet
Thanks for listening!
Painless Software
Less pain, more fun
Continuous deployment of a Python project with Docker and Ansible
By Joost Cassee
Continuous deployment of a Python project with Docker and Ansible
Presented at PyCon Sette on 16 April 2016
