Pufferfish

Warning Running the following project will result in your aws account being charged for the resources that will be created. Run at your own risk.

This is a learning project, it does not reflect all best practices. It was created mainly for the learning of:

• Packer
• ASGs
• CloudWatch

But also training my skills in:

• Ansible
• Go (especially concurrency)
• Terraform
• aws
• GitHub Actions

Setup

Prerequisites

This project assumes you've already got an active Route53 zone and an SSH key in your aws account.

Configuration

This project bases its configuration in aws parameter store. You will need to manually create those parameters:

• /pufferfish/infra/domain for the Route53-managed domain that Pufferfish will be available at
• /pufferfish/infra/key_name for the SSH key name that will be able to connect to EC2 instances
• /pufferfish/infra/manager_cidr for the CIDR that will be able to SSH into EC2 instances
• /pufferfish/infra/notification_email for the email that's to receive alert notifications
• /pufferfish/infra/log_retention_in_days for setting log retention time

Furthermore, the following can be optionally added:

• /pufferfish/app/APP_BACKGROUND_COLOR for specifying the website's background colour (white by default)

Above /pufferfish/infra parameters require a terraform apply of the app module and /pufferfish/app parameters require a rebuild of the AMI.

Installing Ansible dependencies

Some of the following steps require installing collections that the playbooks and roles depend on:

ansible-galaxy install -r ansible/requirements.yml

Creating Terraform state backend

S3 bucket and DynamoDB table for Terraform state are created with an Ansible playbook. The playbook automatically generates a unique name for the S3 bucket.

cd ansible/
ansible-playbook tfbackend.yml

Initialising Terraform projects

You will be able to check the S3 bucket and DynamoDB table names in the parameter store, under /pufferfish/infra/tfbackend_s3_name. You can then initialise the Terraform projects - cd into the module directories in terraform/ directory and run:

terraform init -backend-config="region=eu-west-1" -backend-config="bucket=pufferfish-tfstate-<YOUR_ID>" -backend-config="dynamodb_table=pufferfish-tfstate-<YOUR_ID>-locks"

Provisioning build dependencies

You'll have to provision some aws infrastructure before you can build the AMI - cd into terraform/app_build/ directory and run terraform apply.

Building the AMI

You can build the AMI by running packer and pointing it to the root directory or the .pkr.hcl file. The app file requires a name variable to be provided. This is an example with a unique time-based AMI name generated:

packer build --var name=pufferfish-web-$(date +%s) .

Provisioning the app infrastructure

Once you've got the AMI built, you can provision the rest of the infrastructure. This should just work if you've followed the previous steps correctly. Just cd into terraform/app/ and run terraform apply