This is a prototype Terraform configuration for provisioning a set of resources on AWS to run https://github.com/caktus/taytay.

Please don't take it as a recommended way of doing things, just as an example of playing around with Terraform and its documented AWS support.

Architecture:

Elastic Beanstalk -> RDS (Postgres), Elasticache (Redis)

Terraform: https://www.terraform.io/

To use:

• Install terraform on your system: https://www.terraform.io/intro/getting-started/install.html
• Change to this directory
• Copy secrets.tfvars-example to secrets.tfvars
• Edit secrets.tfvars
• Edit <environment>/main.tf
• Provide your AWS access key & secret, using a configuration file: https://www.terraform.io/docs/providers/aws/index.html. The included Makefile relies on the existence of a profile in ~/.aws/credentials that is identical to <app_name>-<env_name>, as defined in <environment>/main.tf. It works by setting the AWS_PROFILE environment variable to the correct account before running any commands Separate AWS accounts are expected, but not required, for staging and production.

The first thing we'll do with TerraForm is setup and configure remote storage for its state files and pull in our terrapy module:

• make APP=taytay ENV=staging init

If that command completes successfully, we'll have a link set up to a remote copy of the *.tfstate file that others can collaborate with us on. See the <environment>/.terraform directory.

To validate the syntax of all local files:

• make APP=taytay ENV=staging validate

To see what terraform WOULD change if you ran this (dry run):

• make APP=taytay ENV=staging plan

This DOES access the servers, so make sure you've set credentials (see above).

To actually run it (this may cost real money because it will create resources on AWS, though by default everything should be free tier-eligible, if you're on a new account):

• make APP=taytay ENV=staging apply

Once the environment is setup correctly, you can deploy the code with the Elastic Beanstalk CLI:

git clone https://github.com/caktus/taytay.git
mkvirtualenv -p python3.4 taytay
pip install -U awsebcli
eb init --profile=taytay-staging

Supposedly awsebcli doesn't work with Python 3.5, but it seems to work okay as a backup if you don't have 3.4 handy. The last command will prompt you for several inputs:

• Make sure to select the existing app (taytay-app).
• Do NOT set up SSH access at this stage (it's not necessary yet and will add complication).

Note: the awsebcli does not play nicely if you have quotes around your access_key_id and aws_secret_access_key in .aws/credentials. Make sure you create this file without quotations (TerraForm works either way).

The final step before deploying is to add a config file to tell Elastic Beanstalk about our Python app. Save the following in .ebextensions/django.config (really any file ending in .config will do):

packages:
  yum:
    postgresql93-devel: []
    libjpeg-turbo-devel: []
    libpng-devel: []
    freetype-devel: []
    libxslt-devel: []
    libxml2-devel: []
option_settings:
  "aws:elasticbeanstalk:container:python":
    WSGIPath: taytay/wsgi.py
    NumProcesses: 8
    NumThreads: 1
  "aws:elasticbeanstalk:container:python:staticfiles":
    "/static/": "public/static/"
container_commands:
  00_dotenv:
    command: "ln -s ../env .env"
  01_migrate:
    command: "/opt/python/run/venv/bin/python manage.py migrate --noinput"
    leader_only: true
  02_collectstatic:
    command: "/opt/python/run/venv/bin/python manage.py collectstatic --noinput"

After saving this file, you must commit it to the repo (pushing is not necessary).

Once you have the environment set up (eb init will store its files in .elasticbeanstalk/config.yml), deploy the app:

eb deploy --profile=taytay-staging taytay-staging-env