Adventures in docker compose
1 like1,499 views
Docker Compose allows developers to define and run multi-container Docker applications. It helps coordinate multiple containers to work together by defining them in a single compose file. This avoids the complexity of using raw Docker commands. Compose files define services, their images, dependencies, volumes, ports, etc. Compose then automates setting up and running the entire application with a single command. This provides an isolated development environment approximating production. It enables features like continuous integration testing against real services rather than mocks. Overall, Docker Compose improves the developer experience by simplifying and streamlining local development and testing of multi-container applications on Docker.
Adventures in docker compose
- 3. Things no developer/architect/CTO likes in during development ● recognizing wrong choices when it is too late ● mantaining different development environments ● targeting that VERY specific version of something ● spending hours on configuring the dev environment
- 4. Fighting back ● Create approximation of systems and architectures ● Isolate yourself from the host ● Infrastructure as code ● Automate like there is no to tomorrow
- 5. Isolated work environments and teamwork ● If you are working on a backend application front end people can avoid writing tons of mocks or use a shared staging environment ● In case of error the application state can be debugged easily
- 6. Vagrant Vagrant made this approach popular by virtualizing machines on real hypervisors, ● Very accurate replicas ● Performance issues when running complex architectures ● Provisioning looks intense
- 7. Docker ● You don’t have virtual machines ● The OS kernel is shared in virtualized environment called containers. ● The virtualized processes are essentially native, do not require the overhead of virtual machines and still provide a high level of isolation from the host OS
- 8. Docker in a nutshell
- 9. Some docker terms ● Image: a readonly template for creating containers - they are defined with simple files called Dockerfiles ● Container: A Docker container is a runnable instance of a Docker image
- 10. Some more docker terms ● Registry: a repository of docker images ● Volumes: (aka data volumes): folders that bypass docker’s union filesystem. We can use them to map folders from the host machine to the container
- 12. Example Dockerfile A dockerfile is like a recipe to build an image Each line corresponds to a new layer in the image. FROM node:7 RUN npm install -g knex nodemon mocha istanbul node-gyp snyk jsdoc WORKDIR /src
- 13. Docker CLI is semi hard Docker CLI can have quite a lot of options Setting up a serious multi container environment can be very hard docker run -d -v ~/nginxlogs:/var/log/nginx -p 5000:80 -i nginx -- entrypoint /bin/bash example/nginx
- 14. Enter docker compose Compose is a tool for defining and running multi-container Docker applications.
- 15. Enter docker-compose Docker compose helps by defining and coordinating multiple containers. With Compose, you define a multi-container application in a .yml single file, then spin your application up in a single command which does everything that needs to be done to get it running.
- 16. Example (docker-compose.yml) mysql: image: mysql:5.7 environment: MYSQL_ROOT_PASSWORD: passwordsarecool MYSQL_DATABASE: node volumes: - ./conf/mysql/conf.d:/etc/mysql/conf.d ports: - "3306" web: build: . ports: - "80" - "443" - "22" volumes: - ./:/src links: - mysql
- 17. Key points ● We are not provisioning a set of machines, we are defining a set of services ● Then isolating them in containers ● And then choosing how they interact
- 18. Cool things Services and middlewares are super easy to install: as they can be installed right from the docker registry, the version we want, with the image Basic configuration is achieved with env vars, advanced configuration by mapping volumes
- 19. Approximating the production environment ● a good approximation of the production environment is the key for true happiness ● integration tests can be run on local machines for cheap - spot architectural issues very very early
- 20. Even more goodies Continous integration is super easy to setup: at its most basic form it can be reduced to something like: docker-compose run myservice npm test
- 21. Testing in docker-compose Allows you test against real services not mocks- all tests can be integration tests. No CI provider lock-in (as long docker compose is supported) Lightweight - you can spin up a lot services and simulate complex architectures
- 22. Production? Can be also used to setup a quick and dirty staging environment. Just use the -d flag Docker swarm?
- 23. Issues and not so cool things It’s all fun and games as long as you work on linux ● On mac and windows it used to run on a virtual machine ● Now it’s better because it runs on hyper-V on windows and xhyve on MAC (which is still a virtualization)
- 24. Issues On mac and windows installing and updating docker has been more time consuming than expected The first docker-compose up run is always quite painful
- 25. SSH keys What if you need your ssh credentials inside your container? I learned to hate: ● ssh passphrases ● windows permissions
- 26. Other nasty stuff /r/n docker updates mac os filesystem case sensitieves
- 27. Code time