Arkhn DevOps exercise

Introduction

If you're there it means you've succesfully passed the first interview, congratulations :)

It is now time to demonstrate your amazing skills!

Don't worry if you don't already know all the tools we will be using through this exercise: reading documentation and learning is a part of it! Depending on your knowledge of the several tools, this exercise may take about 1-4 hours.

There will be 4 steps, and the two last ones are optional: we don't expect you to finish all steps if you're completely unfamiliar with some concepts. You will have to create a git repository (on github.com or somewhere else) with your work and a small documentation (a README.md like this one is enough) to explain how to use it.

Any question ? write me an email at simon@arkhn.com

STEP 1: A small python program

First, you'll have to implement a basic algorithm that solves the following problem:

Given a string composed of the characters (, ), [, ], {, }, you need to tell if this string is balanced or not. What we call a balanced string is one where all the brackets are closed in the right order.

For instance, ([]){} is balanced, whereas ([)]{} or (() are not.

The input will whether come from the user launching the script or, if none is provided, your code should query a generator service which is provided in this repo.

Generator server

You can launch the generator server with the following command:

python generator-server/src/server.py

It's a very basic flask application with a single http route: /input. With a GET query on this route, you'll be returned a response with an input string as text. You can test the server with curl http://localhost:5000/input or directly in your browser.

Solver behavior

You should be able to launch your solver server with:

python <your-server.py>

and then query it as follows:

$> curl -d '{{[()]}}' localhost:5001/solve
INPUT: [({})] -- RESULT: OK!
$> curl -d '(((]' localhost:5001/solve
INPUT: (((] -- RESULT: BAD INPUT :(
$> curl -X POST localhost:5001/solve # ask the generator server when there are no input to the program
INPUT: <generated input> -- RESULT: <...>

NB: If you really don't want to use Python, feel free to choose your favorite language.

STEP 2: Dockerizing the program

NB: You need to install Docker for this step.

The goal of this step is to write a Dockerfile in order to containerize both python applications (generator ans solver). The result is a docker image (or two) which can be used using docker run <your_image>.

The python programs must behave the same way than when running it locally.

Example:

$> docker build -t generator .
$> docker run -p 5000:5000 -d generator
$> docker build -t solver .
$> docker run -p 5001:5001 -d solver
$> curl -d '{{[()]}}' localhost:5001/solve
INPUT: [({})] -- RESULT: OK!
$> curl -d '(((]' localhost:5001/solve
INPUT: (((] -- RESULT: BAD INPUT :(
$> curl -X POST localhost:5001/solve # ask the generator server when there are no input to the program
INPUT: <generated input> -- RESULT: <...>

BONUS: Write a docker-compose.yml file which runs both the generator server and the solver program when running docker-compose up.

STEP 3: Using a CI to build a publish your docker image

NB: you need to have a DockerHub account for this step (it's free: signup here)

Now that you have a way to build a docker image for your service, we want to publish it to the docker registry. This can be done manually using docker push <your_image>:<tag>.

However, we want to automate the build process using a Continuous Integration (CI) pipeline. If you are familiar with a specific CI tool (jenkins, travis, GitlabCI...), feel free to use it. Otherwise, we encourage you to use Github's tool: Github Actions. You can get started in just a few minutes.

The goal of this exercise is to add a workflow for your repository that will build and push the docker image to the docker hub. And because I'm nice: here are some instructions on how to do it

The CI workflow should be triggered by a commit on any branch. The examiner (me) should be able to docker pull your published docker image (it will be the case if the docker repository is public, which is the default).

STEP 4: Ansible automation

NB: You need to install:

• Vagrant >= 2.2
• vagrant-disksize: vagrant plugin install vagrant-disksize

In this exercise, we want to deploy our dockerized services to a remote machine. Since it would be complex to provision a machine on the cloud, you will simply use a VM on your machine.

Vagrant is a tool to setup virtual environments easily. It works with a simple configuration file called the Vagrantfile.

We provide you with a Vagrantfile in order to setup a VM on your machine. Here is the command you have to use to start the VM:

$> vagrant up ubuntu

You will have to use Ansible in order to do deploy the services.

Your ansible host configuration (see inventories) should have the following variables:

ansible_host: localhost
ansible_user: vagrant
ansible_port: 2222
ansible_ssh_private_key_file: .vagrant/machines/default/virtualbox/private_key

The goal of this exercise is to write an ansible playbook that will deploy the generator and solver services on the VM. (Hint: the bonus of step 3 is pretty useful for this).

Example:

$> ansible-playbook -i inventory_folder/ -l my_host -vv playbook.yml
# wait for deployment to be over
$> curl -d '{{[()]}}' localhost:5001/solve
INPUT: [({})] -- RESULT: OK!