cloud-seeder

A tool for interacting with AWS CloudFormation https://github.com/cjdev/cloud-seeder#readme

Latest on Hackage:0.1.0.0

This package is not currently in any snapshots. If you're interested in using it, we recommend adding it to Stackage Nightly. Doing so will make builds more reliable, and allow stackage.org to host generated Haddocks.

ISC licensed and maintained by Alexis King, Michael Arnold

cloud-seeder Build Status

cloud-seeder is a Haskell DSL for provisioning and controlling CloudFormation stacks. It provides an opinionated mechanism for provisioning a set of related stacks called a “deployment”. You write ordinary CloudFormation templates as YAML, and cloud-seeder helps to create a self-executing command-line interface to orchestrate their deployment.

For example consider a template that provisions an S3 bucket with a configurable name, bucket.yaml:

AWSTemplateFormatVersion: '2010-09-09'

Parameters:
  BucketName:
    Type: String

Resources:
  Bucket:
    Type: AWS::S3::Bucket
    Properties:
      BucketName: !Ref BucketName

Outputs:
  Bucket:
    Value: !Ref Bucket
  BucketDomain:
    Value: !GetAtt Bucket.DomainName

Using cloud-seeder, you can create a deployment script in the same directory, config.hs:

#!/usr/bin/env stack
-- stack runhaskell
{-# LANGUAGE OverloadedStrings #-}

import Network.CloudSeeder

main = cliIO $ deployment "cloud-seeder-example" $ do
  stack "bucket" $ do
    flag "BucketName"

This file contains a declarative configuration of your deployment, but it also serves as an executable command-line tool! Since it has a shebang at the top, it can be used directly to run the deployment with a pleasant interface:

$ ./config.hs provision bucket production --BucketName my-awesome-s3-bucket

The first argument to the provision command is the stack you want to provision, and the second argument is the name of some “environment” to provision in. This environment is used to namespace the eventual stack name, so the above command will spin up a new CloudFormation stack called production-cloud-seeder-example-bucket. The environment is also available in templates themselves if they specify an Env parameter. The generated command-line interface is also robust in the face of mistakes, and it won’t do anything if a required parameter isn’t specified.

While cloud-seeder can be used for single-stack deployments, it’s far more useful when used with multiple stacks at a time, which may possibly depend on other stacks’ outputs. For example, we may now wish to serve resources out of our S3 bucket by using CloudFront. We can write a second template to do the job, cdn.yaml:

AWSTemplateFormatVersion: '2010-09-09'

Parameters:
  BucketDomainName:
    Type: String

Resources:
  Distribution:
    Type: AWS::CloudFront::Distribution
    Properties:
      DistributionConfig:
        Enabled: true
        Origins:
        - DomainName: !Ref BucketDomainName
          Id: origin
          S3OriginConfig: {}
        # ...

We can now add cdn to our deployment configuration:

#!/usr/bin/env stack
-- stack runhaskell
{-# LANGUAGE OverloadedStrings #-}

import Network.CloudSeeder

main = cliIO $ deployment "cloud-seeder-example" $ do
  stack "bucket" $ do
    flag "BucketName"

  stack_ "cdn"

We use stack_ instead of stack to omit the configuration block, since the cdn stack doesn’t need any additional configuration options. When we go to provision the stack, it will work just fine:

$ ./config.hs provision cdn production

Note that we did not have to specify the BucketDomainName parameter explicitly, because it was an output from the bucket stack, so it is automatically passed downward to the cdn stack. This allows stacks defined lower in the configuration to build on top of resources defined in previous ones.

For more information about all of the configuration options available, as well as some of the implementation details, see the documentation on Hackage.

Changes

0.1.0.0 (July 26th, 2017)

Breaking Changes

  • The deploy command is renamed to provision.
  • The environment is now passed as a positional argument instead of as an environment variable.

New Features

  • Added support for configuring parameters with command-line flags and hardcoded environments, in addition to environment variables.
  • Parameters are now parsed from template files to determine whether or not they are optional.
  • Tags cj:application and cj:environment are always added, based on the mandatory ENV positional argument and information in the configuration.
  • The whenEnv and getEnvArg functions are provided to conditionally configure stacks based on the current environment.

Bugfixes and Minor Changes

  • Various improvements to CLI help text.

0.0.0.0 (June 26th, 2017)

  • Initial release
comments powered byDisqus