Continuous Deployment using GitHub, CircleCI, and Heroku

Continuous Deployment workshop slides from Beer City Code 2017 in Grand Rapids, MI.

Sample: https://github.com/jacebrowning/webcalc

Demo: https://webcalc.herokuapp.com/

1. Continuous Deployment using GitHub, CircleCI, and Heroku Jace Browning
2. Let’s meet each other • About me • About you ◦ Name ◦ Interest in this topic ◦ Operating system • Comfort levels ◦ Programming, Terminals ◦ Version Control, Git, GitHub ▪ Pull Requests (PRs), Code Reviews ◦ Continuous Integration (CI), CircleCI, YAML ◦ Continuous Deployment (CD), Heroku ▪ The Platform, Review Apps, Pipelines, Heroku Flow ◦ Python, Flask, MongoDB http://tiny.cc/beercitycode-slack
3. Outline for the day • Machine setup ◦ Installation ◦ Accounts • Start project • Manual deployment to Heroku • Continuous Integration via CircleCI • Continuous Deployment to Heroku • Using GitHub Pull Requests for code review • Provisioning external resources on Heroku • Using Heroku Review Apps to enable manual testing • Advanced topics as time allows
4. How to use these slides … print(“Hello, world”) … $ echo “Hello, world!” http://tiny.cc/beercitycode-slides
5. Machine Setup
6. Install dependencies 1. Git command-line a. Optional: GitHub desktop client 2. Any code editor a. Recommendation: Atom 3. Heroku CLI 4. Python 3.6 5. Pipenv 6. MongoDB a. Optional: MongoDB client (e.g. Robomongo) Instructions continue on the next slide....
7. Install dependencies $ git --version $ heroku --version $ python --version $ pipenv --version $ mongo --version
8. Create accounts • GitHub ◦ https://github.com/join • CircleCI (log in with GitHub) ◦ https://circleci.com/signup/ • Heroku ◦ https://signup.heroku.com/
9. Start Project
10. Project Overview • “webcalc” ◦ /4/+/5 produces “9” • Technologies ◦ Python, virtual environments, pipenv ◦ Flask ◦ MongoDB ◦ Heroku ◦ CircleCI
11. Initial Flask app 1. Create repository on GitHub 2. Clone it to your computer 3. Add a README.md 4. Define requirements in Pipfile 5. Create application file 6. Push files to GitHub Instructions continue on the next slide....
12. New repository (create new repository on GitHub) $ git clone <URL> https://help.github.com/articles/create-a-repo/
13. New file: README.md # webcalc Demo project for the 2017 Beer City Code workshop on Continuous Delivery. https://guides.github.com/features/mastering-markdown
14. New file: .gitignore /.venv/ .cache/ __pycache__/ https://git-scm.com/docs/gitignore
15. New file: Pipfile [[source]] url = "https://pypi.python.org/simple" verify_ssl = true [requires] python_version = "3.6" https://github.com/pypa/pipfile#pipfile
16. Install Flask $ pipenv install flask http://flask.pocoo.org/
17. New file: webcalc.py from flask import Flask app = Flask(__name__) @app.route('/') def index(): return "Hello, world!" if __name__ == '__main__': app.run(debug=True)
18. Run app $ pipenv run python webcalc.py $ pipenv lock
19. Push files $ git status $ git add . $ git commit $ git push
20. Manual deployment to Heroku
21. Deployment Options • Servers • “Cloud” ◦ Virtual machines ◦ Containers • Platforms • “Serverless”
22. Heroku • Developer friendly • First-class language/framework support • Heroku CLI • Free tier • On-demand scaling • Add-ons marketplace
23. Create Heroku app 1. Create new app on Heroku 2. Explore each tab 3. Connect app to GitHub repository 4. Attempt deploy
24. Configure project for Heroku 1. Add gunicorn as a production web server 2. Create Procfile 3. Redeploy to Heroku 4. View site Instructions continue on the next slide....
25. Install gunicorn $ pipenv install gunicorn --lock http://gunicorn.org/
26. New file: Procfile web: gunicorn webcalc:app https://devcenter.heroku.com/articles/procfile
27. Push files and deploy 1. Commit and push code 2. Manually deploy on Heroku 3. View site
28. View production logs $ heroku git:remote --app <your-app-name> $ heroku logs --tail
29. Continuous Integration via CircleCI
30. Continuous Integration • “Continuous” • “Integration” • Running tests gives us confidence in our code • CI services do this automatically
31. Add tests 1. Install testing tools 2. Run tests (no results) 3. Add new test 4. Run tests (results) Instructions continue on the next slide....
32. Install pytest and run it $ pipenv install --dev pytest $ pipenv install --dev pytest-describe $ pipenv install --dev pytest-expecter $ pipenv run pytest $ pipenv lock https://docs.pytest.org
33. New file: test_webcalc.py import pytest from expecter import expect from webcalc import app @pytest.fixture def client(): return app.test_client() def describe_index(): def it_says_hello(client): response = client.get('/') expect(response.data).contains(b"Hello, world!")
34. Run tests and see results $ pipenv run pytest (commit and push changes)
35. CI Services • Types ◦ Self-hosted ◦ Cloud-based • Configurations ◦ File-based ◦ Interface-driven • Target audiences ◦ Private teams ◦ Libraries ◦ Build farms ◦ General-purpose
36. CircleCI • Configuration-based • Parallel builds • Builds with/without cache • SSH debugging • Free tier
37. Enable CircleCI 1. Add project to CircleCI 2. Attempt to build project 3. Add configuration file 4. Add status badge Instructions continue on the next slide....
38. Click on “Add project” 1. https://circleci.com/projects 2. Watch build run
39. New file: circle.yml machine: python: version: 3.6.1 dependencies: pre: - pip install pipenv override: - pipenv install --dev test: override: - pipenv run pytest https://circleci.com/docs/1.0/language-python/
40. Build automatically on pushes 1. Commit and push changes 2. View status of build on CircleCI
41. Continuous Deployment to Heroku
43. Enable Continuous Deployment 1. Enable automatic deploys on Heroku 2. Push a bad change to webcalc.py 3. Watch build fail on CircleCI 4. Confirm no deployment on Heroku 5. Fix tests 6. Push again
44. Using GitHub Pull Request for Code Review
45. Benefits of Pull Requests • We don’t always want to deploy right away • Team consensus • Code reviews • View CI status for each change • Prioritize changes and detect conflicts
46. Create a Pull Request 1. Create a new branch 2. Add test for new functionality 3. Add new route for calculation 4. Push branch and create PR 5. Add collaborator and request review Instructions continue on the next slide....
47. Start a new branch $ git checkout -b add-calc-route
48. Update file: test_webcalc.py … def describe_calc(): def when_plus(client): response = client.get('/4/+/5') expect(response.data).contains(b"9") …
49. Run tests and see them fail $ pipenv run pytest
50. Update file: webcalc.py … @app.route('/<int:a>/<op>/<int:b>') def calc(a, op, b): return f"Result: {a} {op} {b} = {a + b}" …
51. Run tests $ pipenv run pytest
52. Create Pull Request 1. Commit and push changes 2. Open Pull Request on GitHub 3. Add a collaborator on the Settings tab 4. Assign them for review https://help.github.com/articles/about-pull-requests/
53. Review Pull Request 1. Review changes on PR assigned to you 2. Approve changes 3. Merge your own PR after review 4. Watch for automatic deploy
54. Provisioning External Resources on Heroku
55. Heroku add-ons • First-party: ◦ PostgreSQL ◦ Redis • 3rd-party: ◦ Data storage ◦ Monitoring ◦ Logging ◦ Email ◦ Caching ◦ etc.
56. Load math patterns from MongoDB 1. Create a new branch 2. Add test for new functionality 3. Load patterns from MongoDB 4. Push branch and create PR 5. Request review then merge PR Instructions continue on the next slide....
57. Install MongoDB adapter $ git checkout master $ git pull $ git checkout -b load-from-db $ pipenv install flask-pymongo --lock
58. Update file: test_webcalc.py from webcalc import app, mongo @pytest.fixture def pattern(): with app.app_context(): mongo.db.operations.drop() mongo.db.operations.insert( dict( name="x", pattern="{{ a * b }}" ) ) … def describe_calc(): … def from_db(client, pattern): response = client.get('/4/x/5') expect(response.data).contains(b"20")
59. Update file: webcalc.py import os from flask import Flask from flask_pymongo import PyMongo from jinja2 import Template app = Flask(__name__) if os.getenv('MONGODB_URI'): app.config['MONGO_URI'] = os.getenv('MONGODB_URI') mongo = PyMongo(app) …
60. Replace ‘calc’ function: webcalc.py … @app.route('/<int:a>/<op>/<int:b>') def calc(a, op, b): operation = mongo.db.operations.find_one({'name': op}) if operation: return Template(operation['pattern']).render(a=a, b=b) elif op == '+': return f"Result: {a} {op} {b} = {a + b}" else: return f"Result: {a} {op} {b} = ???" …
61. Create Pull Request 1. Create PR 2. Wait for review app to deploy 3. Request review from your collaborator 4. As the collaborator, review the code 5. Merge PR
62. Provision MongoDB on Heroku 1. Add free-tier MongoDB as resource 2. Open database editor 3. Create operations collection 4. Insert document: {"name": "x", "pattern": "{{ a * b }}"} 5. Try out the live app on production
63. Using Heroku Review Apps for Manual Testing
64. Heroku Pipelines • Group related Heroku apps • Stages ◦ Production ◦ Staging ◦ Review
65. Enable Heroku Review Apps 1. Create a Heroku pipeline 2. Add app.json to define resources 3. Add a seeds script 4. Create PR to see review app 5. Test feature on review app Instructions continue on the next slide....
66. Create Heroku pipeline 1. https://devcenter.heroku.com/articles/pipelines 2. Add existing app to pipeline
67. Enable review apps 1. Enable review apps in the pipeline 2. Remove MONGODB_URI from the generated app.json 3. Allow Heroku to commit app.json to the repository 4. Pull down the latest master branch from GitHub https://devcenter.heroku.com/articles/github-integration-review-apps
68. New file: seeds.py from webcalc import app, mongo with app.app_context(): mongo.db.operations.insert( dict( name="+", pattern="{{ a + b }}" ) ) mongo.db.operations.insert( dict( name="x", pattern="{{ a * b }}" ) )
69. Update file: app.json … "scripts": { "postdeploy": "python seeds.py" }, …
70. Create Pull Request 1. Create PR 2. Wait for review app to deploy 3. Request review from your collaborator 4. As the collaborator, review the code and test the review app 5. Merge PR
71. Advanced Topics
72. CI Status Badges
73. Tracking Issues with GitHub
74. Continuous Delivery vs. Continuous Deployment
75. Multiple environments: dev qa/test staging
76. Heroku Flow
77. https://twitter.com/jacebrowning https://beercitycode.slack.com/messages/G5QBA85DZ https://github.com/jacebrowning