You might have experienced one of these unpleasant situations:
-
You’re developing on a computer and you don’t have the admin rights to install a dependency.
-
It took you five hours to install the latest version of you favorite DB but your boss tells you the app will have to run on a previous version.
-
It took you two weeks to build a perfectly clean application... that only works on your machine.
-
The SysOp in charge of all launches hates you because your app is undeployable. Long story short, you already have or soon will spend, at some point in your life, as much time setting up your work environment as actually developing your app.
Right now, you are probably developing a one-block app, with softwares and libraries installed directly in your development environment, or maybe in a virtual environment... Imagine if your application has to be deployed all over the world and you have to re- develop it for all existing platforms and OSs...
Docker was created to satisfy this need for unification and normalisation: it makes it possible to split an application into several microservices, light, adaptable, universal and scalable, and it also gives the system administrators a great flexibility to deploy and scale up the app.
This suite of projects on Docker will help you better understand this specific tool, but also the various aspects of application development using microservices.
It is composed of two parts: the first understand docker CLI: Find the docker command who answering to the asked problematic in the subject below. The second is about creating your first dockerfiles following the asked problematic !
The subject is really big, and the last exercise (the contnairization of gitlab-ce from compilation to production), take me the same time that makes the rest of the project.
Again a huge project that challenges you very much when you know nothing about this tech, but a big opportunitie for progression and skills.
So I took this opportunity, now I containerize all my projects with docker (camagru, matcha, hypertube).
When you use it you can’t do without it anymore
Docker has changed my developper life.
If you don't know this tech, I recommand you to learn, honestly, it's just amazing.
So, Next !
The aim of the Docker-1 project is to make you handle docker and docker-machine, the bases to understand the idea of containerization of services. You can see this project as an initiation. But before you go on this adventure, it’s important to warn you about a thing:
- You will do the exercises on docker 1.12 at least. If you’re not sure what version of docker you have, you can run
docker version. So just install docker on your computer if you haven't yet, and you'll can run the whole of my project ;)
Doing this project has been like a technological revelation.
No kidding.
Docker is so powerful.
So for sharing that, I wrote two articles about it. (Wrote in French for the moment)
The one is for discovering this tech, get a larger vision and understand the general working on it. The second is for realizing your first dockerfiles and docker-composes.
So if you're interested, the two following articles will give you higher values :
Comprendre Docker, Dockerfile et Docker Compose … by Rémy Crestey https://link.medium.com/LyYB7tKEI4
Docker — Mise en pratique by Rémy Crestey https://link.medium.com/zWbAnfPEI4
Enjoy !
For each exercise, we will ask you to give the shell command(s) to:
-
Create a virtual machine with docker-machine using the virtualbox driver, and named Char.
-
Get the IP address of the Char virtual machine.
-
Define the variables needed by your virtual machine Char in the general env of your terminal, so that you can run the docker ps command without errors. You have to fix all four environment variables with one command, and you are not allowed to use your shell’s builtin to set these variables by hand.
-
Get the hello-world container from the Docker Hub, where it’s available.
-
Launch the hello-world container, and make sure that it prints its welcome message, then leaves it.
-
Launch an nginx container, available on Docker Hub, as a background task. It should be named overlord, be able to restart on its own, and have its 80 port attached to the 5000 port of Char. You can check that your container functions properly by visiting http://:5000 on your web browser.
-
Get the internal IP address of the overlord container without starting its shell and in one command.
-
Launch a shell from an alpine container, and make sure that you can interact directly with the container via your terminal, and that the container deletes itself once the shell’s execution is done.
-
From the shell of a debian container, install via the container’s package manager everything you need to compile C source code and push it onto a git repo (of course, make sure before that the package manager and the packages already in the container are updated). For this exercise, you should only specify the commands to be run directly in the container.
-
Create a volume named hatchery.
-
List all the Docker volumes created on the machine. Remember. VOLUMES.
-
Launch a mysql container as a background task. It should be able to restart on its own in case of error, and the root password of the database should be Kerrigan. You will also make sure that the database is stored in the hatchery volume, that the container directly creates a database named zerglings, and that the container itself is named spawning-pool.
-
Print the environment variables of the spawning-pool container in one command, to be sure that you have configured your container properly.
-
Launch a wordpress container as a background task, just for fun. The container should be named lair, its 80 port should be bound to the 8080 port of the virtual machine, and it should be able to use the spawning-pool container as a database service. You can try to access lair on your machine via a web browser, with the IP address of the virtual machine as a URL. Congratulations, you just deployed a functional Wordpress website in two commands!
-
Launch a phpmyadmin container as a background task. It should be named roach-warden, its 80 port should be bound to the 8081 port of the virtual machine and it should
be able to explore the database stored in the spawning-pool container. -
Look up the spawning-pool container’s logs in real time without running its shell.
-
Display all the currently active containers on the Char virtual machine.
-
Relaunch the overlord container.
-
Launch a container name Abathur. It will be a Python container, 2-slim version, its /root folder will be bound to a HOME folder on your host, and its 3000 port will be bound to the 3000 port of your virtual machine. You will personalize this container so that you can use the Flask micro-framework in its latest version. You will make sure that an html page displaying "Hello World" with
<h1>tags can be served by Flask. You will test that your container is properly set up by accessing, via curl or a web browser, the IP address of your virtual machine on the 3000 port. You will also list all the necessary commands in your repository. -
Create a local swarm, the Char virtual machine should be its manager.
-
Create another virtual machine with docker-machine using the virtualbox driver, and name it Aiur.
-
Turn Aiur into a slave of the local swarm in which Char is leader (the command to take control of Aiur is not requested).
-
Create an overlay-type internal network that you will name overmind.
-
Launch a rabbitmq SERVICE that will be named orbital-command. You should define a specific user and password for the RabbitMQ service, they can be whatever you want. This service will be on the overmind network.
-
List all the services of the local swarm.
-
Launch a 42school/engineering-bay service in two replicas and make sure that the service works properly (see the documentation provided at hub.docker.com). This service will be named engineering-bay and will be on the overmind network.
-
Get the real-time logs of one the tasks of the engineering-bay service.
-
... Damn it, a group of zergs is attacking orbital-command, and shutting down the engineering-bay service won’t help at all... You must send a troup of Marines to eliminate the intruders. Launch a 42school/marine-squad in two replicas, and make sure that the service works properly (see the documentation provided at hub.docker.com). This service will be named... marines and will be on the overmind network.
-
Display all the tasks of the marines service.
-
Increase the number of copies of the marines service up to twenty, because there’s never enough Marines to eliminate Zergs. (Remember to take a look at the tasks and logs of the service, you’ll see, it’s fun.)
-
Force quit and delete all the services on the local swarm, in one command.
-
Force quit and delete all the containers (whatever their status), in one command.
-
Delete all the container images stored on the Char virtual machine, in one command as well.
-
Delete the Aiur virtual machine without using rm -rf.
So, that was fun wasn’t it?
Now, it’s time get down to business. Docker makes it possible to create yourself OWN containers for your OWN applications! That’s what this chapter is about. Luckily, Docker lets you program Dockerfiles (aka a Makerfile for Docker... I think you get the nuance). Dockerfiles use a specific syntax that reuses a base image or an existing con- tainer to add your own dependencies and your own files.
You will also notice that every command built by your Dockerfiles generates a layer that’s reusable in other Dockerfiles or other versions of the same Dockerfile, in order to avoid data duplication. Awesome, isn’t it?
But before you start making your own containers, make sure that your virtual machine is cleared of all images and residual active containers. We’ll start from the beginning and build up more and more complex applications.
Create a 01_dockerfiles folder in which you’ll store all the Dockerfiles that you’ll need later, in separate folders (ex00 / ex01 ...).
Check out this link to design high quality Dockerfiles: Dockerfile Best Practices
From an alpine image you’ll add to your container your favorite text editor, vim or emacs, that will launch along with your container.
From a debian image you will add the appropriate sources to create a TeamSpeak server, that will launch along with your container. It will be deemed valid if at least one user can connect to it and engage in a normal discussion (no far-fetched setup), so be sure to create your Dockerfile with the right options. Your program should get the sources when it builds, they cannot be in your repository.
You are going to create your first Dockerfile to containerize Rails applications. That’s a special configuration: this particular Dockerfile will be generic, and called in another Dockerfile, that will look something like this:
FROM ft-rails:on-build
EXPOSE 3000
CMD ["rails", "s", "-b", "0.0.0.0", "-p", "3000"]
Your generic container should install, via a ruby container, all the necessary depen- dencies and gems, then copy your rails application in the /opt/app folder of your container. Docker has to install the approtiate gems when it builds, but also launch the migrations and the db population for your application. The child Dockerfile should launch the rails server (see example below). If you don’t know what commands to use, it’s high time to look at the Ruby on Rails documentation.
Docker can be useful to test an application that’s still being developed without pol- luting your libraries. You will have to design a Dockerfile that gets the development version of Gitlab - Community Edition installs it with all the dependencies and the nec- essary configurations, and launches the application, all as it builds. The container will be deemed valid if you can access the web client, create users and interact via GIT with this container (HTTPS and SSH). Obviously, you are not allowed to use the official container from Gitlab, it would be a shame...
Thanks for your interest. Contact me if you have questions, or if want to talk about that ! See you ;)