squidboylan / dreamcompute-hello-world

Deploy an example Python application on DreamHost Compute with Ansible

Deploying a Simple Web Application on DreamCompute

Introduction

In this article, you'll learn to deploy a simple web application on an OpenStack public cloud (DreamCompute) using the popular automation tool, Ansible. To get started:

• Sign up for a DreamCompute account (DreamHost offers a free trial period)

• Log in to the DreamCompute Dashboard and create an SSH key (filling out the form will download a PEM keyfile you'll use to authenticate to your server). Remember the name of the key that you create - you'll need that value later.

• Install a recent version of Python (2.7+)

• Install the helloworld Python package in this repository, shade, and ansible:

  $ git clone https://github.com/ryanpetrello/dreamcompute-hello-world.git hello-world

  $ cd hello-world && pip install -e . shade ansible

Python and WSGI

We'll be deploying a very simple Python WSGI application that responds to all HTTP requests with Hello, World! WSGI is a Python protocol which allows a web server to interface with a Python callable (or function) to handle each request. Our function looks like this:

def application(environ, start_response):
      data = "Hello, World!\n"
      start_response("200 OK", [
          ("Content-Type", "text/plain"),
          ("Content-Length", str(len(data)))
      ])
      return data

...but you could replace this Python code and the web server with your own written in PHP, Ruby, Javascript, or any other language!

Deploying with Ansible

We'll use Ansible to create a repeatable "playbook" of commands to set up the Hello World application on a DreamCompute server. These commands include:

• Launching a server in DreamCompute and assigning a public IP address to it
• Installing a web server (we'll use http://gunicorn.org)
• Installing our example Python application and configuring it to handle requests

In order for Ansible to run DreamCompute API calls on your behalf, you'll need to download a small shell script that sets up your API credentials. Do so by visiting API Access in the OpenStack Dashboard. After the shell script downloads, open your computer's command line and run the following (substituting the actual location of your downloaded file).

$ source /path/to/downloaded/file/dhc123456789-openrc.sh

You'll be prompted for a password - it's the one you use to log in to the DreamCompute Dashboard.

At this point you should be ready to deploy your application. Do so by running the following commands:

$ chmod 600 /path/to/keyname.pem
$ ansible-playbook -vvvv -i "localhost," playbooks/deploy.yml --extra-vars "key_name=keyname private_key=/path/to/keyname.pem"

You'll need to substitute the keyname key name value for the actual name you chose earlier, and you'll also need to replace /path/to/keyname.pem with the actual path to the PEM file you downloaded.

If all is well, you'll be greeted with an instructional message:

Visit http://1.2.3.4/ in your browser!

Example Server Architecture

If you ssh into your newly created server:

$ ssh -i /path/to/keyname.pem [email protected]

...you'll find a variety of processes running in the following configuration:

HTTP Request ──> <Production/Proxy Server>, nginx (1.2.3.4:80)
                  │
                  │   <supervisord> (monitors and keeps gunicorn processes running)
                  ├── <WSGI Server> gunicorn Instance (/tmp/gunicorn.sock)
                  ├── <WSGI Server> gunicorn Instance (/tmp/gunicorn.sock)
                  ├── <WSGI Server> gunicorn Instance (/tmp/gunicorn.sock)
                  ├── <WSGI Server> gunicorn Instance (/tmp/gunicorn.sock)

supervisord is installed and is used to manage multiple gunicorn worker processes, each of which is bound to a Unix domain socket (though you could also configure them to bind to a TCP port). nginx listens on port 80 and balances incoming HTTP requests across the gunicorn workers processes.