Content
This project tries to implement an end to end python repository that includes CI/CD pipeline using Jenkins, code quality tools integration like pylint, mypy, flake8, isort, black and mkdocs for documentation using docstrings and typing within project source code.
Also, the template project is a REST API using FastAPI that expose basic calculator functions.
- Documentation: mkdocs with docstrings and typing at readthedocs service
- CI CD: Jenskins pipeline using telegram integration and custom branch strategy described below
- Deployment: Docker and Docker Swarm
- Code Quality: isort, black, mypy, pylint, flake8
- Testing: pytest
The following diagram describes the branch strategy used in the context of this template repository. The ideas was to simplify git flow due to it is oriented to single developer or very small teams with only one production stage.
The workflow consists of release candidate branches (rc-*) that will be branched from master. Master branch will always have a tag with a tested and approved version which will be the result of a rc-* branch merge. Developers will create feature branches (feature-<issue_number>-*) from rc-* branches.
Events:
- commit on feature branch: Trigger code quality checks and unit tests
- commit on release candidate branch: Trigger code quality checks, unit tests, docker image building, image push tor registry and deploy on VPS
- commit on master: Trigger code quality checks and unit tests, create release at Github to create version package
Mkdocs with readthedocs service is being used to automatically publish service documentation. This approach was prefered over local Mkdocs service due serverless simplicity.
Static documentation is generated at docs/ folder and is deployed when a release candidate branch is merged with master.
Docs URL is published in badge in this README.
The pipeline uses a custom docker agent with all the neccessary requirements. It also uses conditionals to fork workflow executions depending on branch naming, in this way we can achieve our goal using only one pipeline. Callback on success and on failure are used in order to send messages to a specific telegram bot.
make build # build api image, only for development
make build-doc # build documentation image, only for development
make build-jenkins-builder # build a new image to be used as CI CD pipeline builderHow to execute code and services for local development and prod environments.
Use the Makefile
make install # install dependencies
make code-quality
make test
make doc # run mkdocs development server
make run # run api using local uvicorn
make deploy-local # deploy docker-compose stack locallyMerge your feature branch to release candidate branch to execute deploy pipeline. Public URLs for documentation and api are defined on docker-compose. Take into consideration that public DNS must be created before the deployment because an automatic process will create a Let's Encrypt certificate right after the container is up and running in the server.
TBD