This guide will walk you through deploying a Django application on AWS using OpDemand.
- A free OpDemand account with
- Valid AWS credentials
- Linked GitHub account
- The OpDemand Command Line Interface
- A Django application that is hosted on GitHub
The simplest way to get started is by forking OpDemand's sample application located at: https://github.com/opdemand/example-python-django
After forking the project, clone it to your local workstation using the SSH-style URL:
$ git clone [email protected]:<yourname>/example-python-django.git example-python-django
$ cd example-python-django
If you want to use an existing application, no problem -- just make sure you've cloned it from GitHub.
To use a Django application with OpDemand, you will need to conform to 4 basic requirements:
- Use virtualenv and pip to manage dependencies
- Use Foreman to manage processes
- Use Environment Variables to manage configuration
- Use Deploy Scripts to perform additional customization
If you're deploying the example application, it already conforms to these requirements. If you're in a rush, skip to Create a Platform.
On every deploy action, OpDemand will run an . venv/bin/activate
followed by a pip install -r requirements.txt
on all application workers to ensure dependencies are up to date.
To setup and activate virtualenv on your local workstation:
$ virtualenv venv --distribute
$ . venv/bin/activate
The last command will activate virtualenv for your current shell session. To ensure you're using virtualenv and local dependencies, you'll need to re-run the . venv/bin/activate
command for each new shell session. With virtualenv active, you can safely install external dependencies for your application:
$ sudo pip install flask gunicorn
When you're done installing dependencies, use pip freeze
to write out a new requirements.txt
file that contains dependency information:
$ pip freeze > requirements.txt
OpDemand uses a Foreman Procfile to manage the processes that serve up your application. The Procfile
is how you define the command(s) used to run your application. Here is an example Procfile
that uses gunicorn:
web: gunicorn -b 0.0.0.0:$APPLICATION_PORT app:app
This tells OpDemand to run one web process using gunicorn
. You can test this out locally by setting the APPLICATION_PORT
environment variable and calling foreman start
.
$ export APPLICATION_PORT=8080
$ foreman start
12:45:57 web.1 | started with pid 26809
12:45:58 web.1 | 2012-05-10 12:45:57 [26809] [INFO] Starting gunicorn 0.14.2
12:45:58 web.1 | 2012-05-10 12:45:58 [26809] [INFO] Listening at: http://0.0.0.0:8080 (26809)
12:45:58 web.1 | 2012-05-10 12:45:58 [26809] [INFO] Using worker: sync
12:45:58 web.1 | 2012-05-10 12:45:58 [26811] [INFO] Booting worker with pid: 26811
OpDemand uses environment variables to manage your application's configuration. For example, the application listener must use the value of the APPLICATION_PORT
environment variable. The following code snippets demonstrates how this can work inside your application:
port = os.environ.get('APPLICATION_PORT', 8080) # fallback to 8080
The same is true for external services like databases, caches and queues. Here is an example in that shows how to connect to a MongoDB database using the DATABASE_HOST
and DATABASE_PORT
environment variables:
import pymongo
database_host = os.environ.get('DATABASE_HOST', 'localhost')
database_port = os.environ.get('DATABASE_PORT', 27017)
connection = pymongo.Connection(database_host, database_port)
OpDemand can run your deploy script after all other services have been completed. For example, your application may need to sync with the database:
#!/bin/sh
# activate virtualenv
. venv/bin/activate
# trigger django database sync
python manage.py syncdb
or install required system libraries:
#!/bin/sh
sudo apt-get install -yq libpq-dev python-dev
The script runs from /home/ubuntu/repo
and needs to be stored in bin/deploy
in your repository.
$ opdemand list | grep python
app/python/1node: Python Application (1-node)
app/python/2node: Python Application (2-node with ELB)
app/python/4node: Python Application (4-node with ELB)
app/python/Nnode: Python Application (Auto Scaling)
Use the opdemand create
command to create a new platform based on one of the templates listed. To create an app/python/1node
platform with app
as its handle/nickname.
$ opdemand create app --template=app/python/1node
To quickly configure a platform from the command-line use opdemand config [handle] --repository=detect
. This will attempt to detect and install repository configuration including:
- Detecting your GitHub repository URL, project and username
- Generating and installing a secure SSH Deploy Key
More detailed configuration can be done using:
$ opdemand config app # the entire config wizard (all sections)
$ opdemand config app --section=provider # only the "provider" section
Detailed configuration changes are best done via the web console, which exposes additional helpers, drop-downs and overrides.
To start your platform use the opdemand start
command:
$ opdemand start app
You will see real-time streaming log output as OpDemand orchestrates the platform's infrastructure and triggers the necessary SSH deployments. Once the platform has finished starting you can access its services using an opdemand show
.
$ opdemand show app
Application URL (URL used to access this application)
http://ec2-23-20-231-188.compute-1.amazonaws.com
Open the URL and you should see "Powered by OpDemand" in your browser. To check on the status of your platforms, use the opdemand status
command:
$ opdemand status
app: Python Application (1-node) (status: running)
As you make changes to your application code, push those to GitHub as you would normally. When you're ready to deploy those changes, use the opdemand deploy
command:
$ opdemand deploy app
This will trigger an OpDemand deploy action which will -- among other things -- update configuration settings, pull down the latest source code, install new dependencies and restart services where necessary.