Skip to content

Serverless CI for GitHub using AWS CodeBuild with PR and status support

License

Notifications You must be signed in to change notification settings

fx/terraform-aws-github-ci

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Travis Gitter GitHub

Terraform AWS GitHub CI

A Terraform module to setup a serverless GitHub CI build environment with pull request and build status support using AWS CodeBuild.

Architecture

Architecture

This module registers a GitHub webhook which is triggered for push and pull_request events and starts the build for the respective branch. All builds run in parallel. The build progress and status for a respective commit is reported back to GitHub. Furthermore, a badge for the status of master is generated and hosted on S3.

Cost

Building with this CI server is unbelievably cheap - you only pay what you use. Pricings starts at $ 0,005 per build minute, and AWS CodeBuild offers 100 free build minutes every month. The price for the other services (Lambda, SNS, S3 and CloudWatch) are negligible and should only add a few cents to your monthly bill. Compare that to the $ 69 that services like Travis cost every month, regardless of how much you use them.

Usage

You need an AWS and GitHub account and a repository you want to be built. The repository must specify a buildspec.yml which is documented here. First, you need to go to the CodeBuild dashboard in your region, manually create a new project and choose GitHub as the Source provider, allowing AWS to authorize your account. Next, set up your AWS credentials and install Terraform if you haven't got it available already.

Next, add the following module to your Terraform configuration and apply it:

module "github_ci" {
  source  = "github.com/squidfunk/terraform-aws-github-ci"
  version = "0.5.4"

  namespace          = "<namespace>"
  github_owner       = "<owner>"
  github_repository  = "<repository>"
  github_oauth_token = "<oauth-token>"
}

All resources are prefixed with the value specified as namespace. If the S3 bucket name (see below) is not explicitly set, it's set to the given namespace which means there must not already exist an S3 bucket with the same name. This is a common source of error.

Now, when you push to master, or create a pull request, CodeBuild will automatically build the commit and report the status back to GitHub. A status badge can be added to your project's README using the codebuild_badge_url and codebuild_url outputs printed to the terminal.

Note: the OAuth-token is currently mandatory (also for public repositories), because Terraform doesn't support conditional blocks inside resources. However, this feature is currently being implemented and should be released shortly. If you want to omit it, create your own CodeBuild project see below.

Configuration

The following variables can be configured:

Required

namespace

  • Description: AWS resource namespace/prefix (lowercase alphanumeric)
  • Default: none

github_owner

  • Description: GitHub repository owner
  • Default: none

github_repository

  • Description: GitHub repository name
  • Default: none

github_oauth_token

  • Description: GitHub OAuth token for repository access
  • Default: none

Optional

github_reporter

  • Description: GitHub commit status reporter
  • Default: "AWS CodeBuild"

codebuild_project

  • Description: CodeBuild project name (won't create default project)
  • Default: ""
  • Conflicts with: codebuild_compute_type, codebuild_image, codebuild_buildspec

codebuild_compute_type

  • Description: Compute resources used by the build
  • Default: "BUILD_GENERAL1_SMALL"
  • Conflicts with: codebuild_project

codebuild_image

  • Description: Base image for provisioning (AWS Registry, Docker)
  • Default: "aws/codebuild/ubuntu-base:14.04"
  • Conflicts with: codebuild_project

codebuild_buildspec

  • Description: Build specification file location (file format)
  • Default: "buildspec.yml" (at repository root)
  • Conflicts with: codebuild_project

codebuild_privileged_mode

  • Description: If set to true, enables running the Docker daemon inside a Docker container.
  • Default: false
  • Conflicts with: codebuild_project

codebuild_bucket

  • Description: S3 bucket to store status badge and artifacts
  • Default: "${var.namespace}" (equal to namespace)

Outputs

The following outputs are exported:

codebuild_service_role_name

  • Description: CodeBuild service role name

codebuild_service_role_arn

  • Description: CodeBuild service role ARN

codebuild_bucket

  • Description: CodeBuild artifacts bucket name

codebuild_badge_url

  • Description: CodeBuild status badge URL

codebuild_url

  • Description: CodeBuild project URL

Default project

If you need more control over the CodeBuild project, you can pass the name of an external CodeBuild project in this variable. This will avoid the creation of the default project which has the following configuration:

resource "aws_codebuild_project" "codebuild" {
  name = "${var.github_repository}"

  build_timeout = "5"
  service_role  = "${aws_iam_role.codebuild.arn}"

  source {
    type     = "GITHUB"
    location = "https://github.com/$${owner}/$${repository}.git"

    auth {
      type     = "OAUTH"
      resource = "${var.github_oauth_token}"
    }
  }

  environment {
    compute_type = "${var.codebuild_compute_type}"
    type         = "LINUX_CONTAINER"
    image        = "${var.codebuild_image}"
  }

  artifacts {
    type           = "S3"
    location       = "${var.codebuild_bucket}"
    name           = "${var.github_repository}"
    namespace_type = "BUILD_ID"
    packaging      = "ZIP"
  }
}

The corresponding service role and the bucket are always created and exported as codebuild_service_role_arn, codebuild_service_role_name and codebuild_bucket. You can reference them in your CodeBuild resource definition, e.g. to attach further policies, and thus avoid the creation of your own service role and bucket.

Limitations

This module first integrated with AWS CodePipeline but switched to CodeBuild, because the former is heavily opinionated in terms of configuration and much, much slower. For this reason, the deployment of your build artifacts must be handled by another module which can be triggered when the build artifacts are written to S3, most likely by using a Lambda function.

License

MIT License

Copyright (c) 2017-2018 Martin Donath

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

About

Serverless CI for GitHub using AWS CodeBuild with PR and status support

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

 
 
 

Languages

  • HCL 51.2%
  • TypeScript 26.8%
  • Makefile 16.4%
  • Shell 5.6%