trvslhlt/one-and-done

practicing configuration management

One and done

Your mission, should you choose to accept it, is to create a single script that takes a brand new VM, and creates a fully-functioning production site, all by running a single command. In other words, setup should not require vagrant ssh, or any other interaction.

Workflow

1. Modify provision.sh, etc.
2. Run vagrant up.
   • As you can see in the config.vm.provision command in the Vagrantfile, this will run provision.sh after the VM is created.
3. Visit your VM's IP in your browser, and you should see "Hello world!"
4. If that doesn't work, run a vagrant destroy and repeat from step 1.

This turnaround time is slow, so you might want to run the commands by hand within vagrant ssh, then move them into the provisioning script as you finalize them.

Tutorials

These may help:

• http://exploreflask.com/en/latest/deployment.html
• https://www.digitalocean.com/community/tutorials/how-to-serve-flask-applications-with-gunicorn-and-nginx-on-ubuntu-16-04

Tips

• There are a lot of Python WSGI web server options. Unless you specifically want to try a different one, go with Gunicorn (pronounced "G-unicorn").
• You need to reload nginx configuration after you modify it for the changes to be picked up.
• The vagrant-address plugin is useful for getting the IP address of your VM.

Next steps

Once you get the above working locally:

1. Use Vagrant to deploy the site to Amazon Web Services (AWS)
   • Be very careful not to commit your AWS credentials to the repository.
2. Run the site in a Docker container

Discussion questions

• What is the point of this exercise?
• What is each command in your provisioning script doing?
• How can you verify that each command is doing what you intend for it to? In other words, how would you test/troubleshoot this?
• Why do we need nginx, Gunicorn, etc? Why not just use flask run?