aws-ec2-provisioner 0.1.1

POC boto3 resource provisioner

CLI tool to provision infrastructure and configure a simple web app in AWS.

The infrastructure components will be name elb-project_name, asg-project_name and so on. The tool will reuse any pre-existing components based on the name. However, if there are several components with the same name - one will be chosen at random

Prerequisites

• AWS Account
• VPC with at least one public subnet
• At least 9 available private IP addresses in the subnet (allows for 1 EC2 instance to be provisioned)
• AWS CLI needs to be configured for a user with necessary IAM permissions (TBD)

Limitations

To simplify the design all the infrastructure are being created in the single subnet, and this subnet is required to be a public one. Because the architecture doesn't include a NAT gateway all the provisioned instances will be assigned a public IP address, so they can download the necessary packages.

There is no input validation

Installation

pip install aws-ec2-provisioner

or

git clone [e-mail]:valeramaniuk/aws-ec2-provisioner.git
cd aws-ec2-provisioner
pip install .

Usage

run aws_ec2_provisioner with the following options:

--vpc-id <VPC where you want to deploy the app>
--subnet-id <Subnet where you want to deploy the app. Should in the selected VPC>
--project-name <Name of the project. You won't be able to create 2 projects with the same name>
--aws-profile <The name of access key/secret key pair (~/.aws/credentials)>
--min-asg-size <Minimum size of the autoscaling group>
--max-asg-size <Maximum size of the autoscaling group>
--instance-type <INT. Choose from a predefined list>
--scaling-target-value <INT. Scaling policy target value % >

If you omit any of the options you will be prompted interactively. In this case, the tool will look up the configured aws profiles, the VPCs available for the chosen profile/region and the subnets available for the chosen VPC.

aws_ec2_provisioner --help for all available options

Available instace types:

1. t2.micro
2. t2.meduim

Infrastructure to be created:

• ELB
  • Listener 80
  • Instance port 80
  • Health check HTTP:80/
• Autoscaling group
  • Scales in/out to maintain Average CPU utilization across all instances in the Autoscaling Group.
  • Launch configuration associates public IP by default
  • User data in Launch configuration installs and starts ngnix
• Security Groups
  • ELB security group
    • ingress tcp:80 from anywhere
  • APP security group
    • ingress only from ELB security group
    • no SSH access because the APP servers are in the public subnet (architecture limitation)

TODO:

• Load based autoscaling
• Additional region
• Choose aws_profile interactively based on ~/.aws/credentials
• Choose vpc-id interactively after specifying the aws-profile and region
• Choose subnet-id interactively after specifying the subnet
• Proper logging instead of print statements
• Tests
• Idempotency (with limitations)
• Ability to rollback
• Move instance user data to separate file, so the user data file may be specified at runtime
• Add bumpversion
• Support for multiple subnets. Private/public
• Provision a NAT gateway as necessary
• Informative UI
• Destroy all provisioned infrastructure
• Catch all botocore exceptions