django-code-deploy

This module deploys django code to AWS instances, making use of fab. This is a simple way to remove infrastructure dependencies (such as number and IP addresses of target instances) from the app repo itself. By no means complete, but may be a helpful step.

Setup

First add this simplified fabfile.py to the root of your project:

from django_code_deploy import fabfile as deploy

That's it, your project now has all it needs to be deployed.

Then you'll need to pip install django-code-deploy wherever fab will run (e.g. a jenkins server). For example,

sudo pip install django-code-deploy

Tag instances

Finally, you'll need to make sure that your instances have these 3 standard tags, since django-code-deploy works by looking for all instances that match a specified set of tags. These tags are:

• Type - the type of instance, such as web, worker, db.
• Env - the environment name, such as dev, staging, prod.
• App - the application name, such as "fresco", "topopps", "mojo", etc.

Usage

Here's a example set of commands for deploying an app:

fab deploy.dev deploy.set_hosts:type=web,appname=test,region=us-west-2 deploy.deploycode:branch=dev deploy.dbmigrate
fab deploy.dev deploy.set_hosts:type=web,appname=test,region=us-west-2 deploy.restart_celery deploy.restart_uwsgi deploy.restart_nginx

Since this was running as a jenkins job, it assumes that the git repo was updated before it ran. With that in mind, this

• set the envronment to 'dev'
• obtained a list of all instances in the default region of type 'web', app name 'test', and env 'dev'.
• deployed the dev branch of the repo that had previously been obtained by jenkins
• ran db migrations
• restarted celery
• restarted uwsgi
• restarted nginx

Note that the deploycode method also performs a collectstatic step before returning.

Complete list of fab methods and arguments

This is a complete list of the invocable methods, and arguments that may be specified in each one.

set_hosts(type,             #type is web, worker, db
  primary=None,
  appname=None,             #appname is application name, such as "fresco", "topopps", "mojo", etc.
  region=None               #region is aws region

These next three set AWS_ENVIRONMENT (os-level env var) to either dev, staging, or prod.

dev()
staging()
prod()
set_environment(anEnv)  For an environment that is not one of dev, staging or prod

Use the set_access_key() to the location and file that is the private access key to be used to access the instance set_access_key('/path/to/theKey')

These are the main methods to use in a deploy.

deploycode(branch)
dbmigrate()
restart_nginx()
restart_uwsgi()
restart_celery()

Other methods:

dbmigrate_docker(containerid,codepath='/data/deploy/current')
django_check()
codeversioner()

Other notes

• The code is deployed in directories organized like this, one per deploy: /data/deploy/today's date ... the newest deploy will be symlinked to /data/deploy/current/

• As is the nature of fab, the instances are processed one at a time. No attempt has been made to do the actual deploys in parallel.

• (c) 2015 - 2018, devops.center llc.