Skip to content

Latest commit

 

History

History

deployment

Lambda function deployments using AWS CodePipeline and AWS CodeDeploy

Terraform module to create AWS resources for secure blue/green deployments of Lambda functions using AWS CodePipeline, CodeBuild and CodeDeploy.

Basic principle for this module is to separate the infrastructure/configuration aspect of Lambda functions (e.g. IAM role, timeouts, runtime, CloudWatch logs) from continuous deployments of the actual function code.

The latter should be build, tested and packaged on CI systems like GitHub actions and uploaded to S3 (package_type=Zip) or pushed to ECR (package_type=Image). Controlled and secure blue/green deployments of the function code with (automatic) rollbacks and traffic shifting will then be executed in an AWS CodePipline using CodeBuild to update the function code and CodeDeploy to deploy the new function version.

Features

  • fully automated AWS CodePipelines with CodeBuild and CodeDeploy stages to deploy containerized Lambda functions from ECR or zipped packages from S3
  • creation of IAM roles with permissions following the principle of least privilege for CodePipeline, CodeBuild and CodeDeploy or bring your own roles
  • SNS topic for AWS CodeStar Notifications of CodePipeline events, or bring your own SNS topic
  • BeforeAllowTraffic and AfterAllowTraffic hooks for CodeDeploy
  • AWS predefined and custom deployment configurations for CodeDeploy
  • automatic rollbacks and support of CloudWatch alarms to stop deployments
  • additional custom CodePipeline steps executed after the deployment

How do I use this module?

Initial Terraform run

The Terraform lambda_function relies on existing image_uri (for containerized functions) or s3_object_version (for S3 based packages) in the initial run.

For containerized functions this can be achieved one of the following options:

  • targeting only aws_ecr_repository in the first run and push and initial image before applying the rest of the infrastructure
  • using docker_registry_image to build the image as part of the terraform lifecycle
  • using a null_resource with a local-exec provisioner to build and push the image as part of the terraform lifecycle, see container-image (ECR) for a full example

For Zip packages on S3 this can be achieved using an aws_s3_object ignoring changes to etag, see zipped package (S3) for a full example.

It's then recommended to build, test, package and upload all further function code changes using a CI system like GitHub actions.

using container images

see ECR example for details:

locals {
  environment   = "production"
  function_name = "with-ecr-codepipeline"
}

resource "aws_lambda_alias" "this" {
  function_name    = module.lambda.function_name
  function_version = module.lambda.version
  name             = local.environment

  lifecycle {
    ignore_changes = [function_version]
  }
}

resource "aws_ecr_repository" "this" {
  name = local.function_name
}

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  alias_name          = aws_lambda_alias.this.name
  ecr_image_tag       = local.environment
  ecr_repository_name = aws_ecr_repository.this.name
  function_name       = local.function_name
}

module "lambda" {
  source = "moritzzimmer/lambda/aws"

  function_name                    = local.function_name
  ignore_external_function_updates = true
  image_uri                        = "${aws_ecr_repository.this.repository_url}:${local.environment}"
  package_type                     = "Image"
  publish                          = true
}

using S3 packages

see S3 example for details:

locals {
  environment   = "production"
  function_name = "with-s3-codepipeline"
  s3_key        = "package/lambda.zip"
}

resource "aws_lambda_alias" "this" {
  function_name    = module.lambda.function_name
  function_version = module.lambda.version
  name             = local.environment

  lifecycle {
    ignore_changes = [function_version]
  }
}

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  alias_name    = aws_lambda_alias.this.name
  function_name = local.function_name
  s3_bucket     = aws_s3_object.source.bucket
  s3_key        = local.s3_key
}

module "lambda" {
  source = "moritzzimmer/lambda/aws"

  function_name                    = local.function_name
  handler                          = "index.handler"
  ignore_external_function_updates = true
  publish                          = true
  runtime                          = "nodejs20.x"
  s3_bucket                        = aws_s3_object.source.bucket
  s3_key                           = local.s3_key
  s3_object_version                = aws_s3_object.source.version_id
}

resource "aws_s3_bucket" "source" {
  bucket        = "source-bucket"
  force_destroy = true
}

// make sure to enable S3 bucket notifications to start CodePipeline
resource "aws_s3_bucket_notification" "source" {
  bucket      = aws_s3_bucket.source.id
  eventbridge = true
}

// versioning is required for CodePipeline
resource "aws_s3_bucket_versioning" "source" {
  bucket = aws_s3_bucket.source.id

  versioning_configuration {
    status = "Enabled"
  }
}

resource "aws_s3_bucket_public_access_block" "source" {
  bucket = aws_s3_bucket.source.id

  block_public_acls       = true
  block_public_policy     = true
  ignore_public_acls      = true
  restrict_public_buckets = true
}

// this resource is only used for the initial `terraform apply` - all further
// deployments are running on CodePipeline
resource "aws_s3_object" "source" {
  bucket = aws_s3_bucket.source.bucket
  key    = local.s3_key
  source = module.function.output_path
  etag   = module.function.output_md5

  lifecycle {
    ignore_changes = [etag]
  }
}

The Amazon S3 source action of the CodePipeline needs an AWS S3 Notification for emitting events in your Amazon S3 source bucket and sending filtered events to EventBridge and trigger the pipeline (see docs for details). Make sure to enable S3 bucket notifications for your source bucket!

with custom deployment configuration

This module supports all predefined default deployment configurations for the AWS Lambda compute platform as well as custom defined configs, see complete example for details:

// see above

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  alias_name             = aws_lambda_alias.this.name
  function_name          = local.function_name
  s3_bucket              = aws_s3_bucket.source.bucket
  s3_key                 = local.s3_key

  // optionally use custom deployment configuration or a different default deployment configuration like `CodeDeployDefault.LambdaLinear10PercentEvery1Minute` from https://docs.aws.amazon.com/codedeploy/latest/userguide/deployment-configurations.html
  deployment_config_name = aws_codedeploy_deployment_config.custom.id
}

resource "aws_codedeploy_deployment_config" "custom" {
  deployment_config_name = "custom-lambda-deployment-config"
  compute_platform       = "Lambda"

  traffic_routing_config {
    type = "TimeBasedLinear"

    time_based_linear {
      interval   = 1
      percentage = 20
    }
  }
}

with before and after allow traffic hooks

see complete example for details:

// see above

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  alias_name                                        = aws_lambda_alias.this.name
  codedeploy_appspec_hooks_after_allow_traffic_arn  = module.traffic_hook.arn
  codedeploy_appspec_hooks_before_allow_traffic_arn = module.traffic_hook.arn
  codepipeline_artifact_store_bucket                = aws_s3_bucket.source.bucket
  function_name                                     = local.function_name
  s3_bucket                                         = aws_s3_bucket.source.bucket
  s3_key                                            = local.s3_key
}

module "traffic_hook" {
  source = "moritzzimmer/lambda/aws"

  architectures    = ["arm64"]
  description      = "Lambda function executed by CodeDeploy before and/or after allow traffic to deployed version."
  filename         = data.archive_file.traffic_hook.output_path
  function_name    = "codedeploy-hook-example"
  handler          = "hook.handler"
  runtime          = "python3.9"
  source_code_hash = data.archive_file.traffic_hook.output_base64sha256
}

data "aws_iam_policy_document" "traffic_hook" {
  statement {
    actions   = ["codedeploy:PutLifecycleEventHookExecutionStatus"]
    resources = [module.deployment.codedeploy_deployment_group_arn]
  }
}

resource "aws_iam_policy" "traffic_hook" {
  name   = "codedeploy-hook-policy"
  policy = data.aws_iam_policy_document.traffic_hook.json
}

resource "aws_iam_role_policy_attachment" "traffic_hook" {
  role       = module.traffic_hook.role_name
  policy_arn = aws_iam_policy.traffic_hook.arn
}

with rollbacks based on CloudWatch alarms

see complete example for details:

// see above

resource "aws_cloudwatch_metric_alarm" "error_rate" {
  alarm_description   = "${module.lambda.function_name} has a high error rate"
  alarm_name          = "${module.lambda.function_name}-error-rate"
  comparison_operator = "GreaterThanOrEqualToThreshold"
  datapoints_to_alarm = 1
  evaluation_periods  = 1
  threshold           = 5
  treat_missing_data  = "notBreaching"

  // calculate error rate here
}

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  alias_name                                                      = aws_lambda_alias.this.name
  codedeploy_deployment_group_alarm_configuration_enabled         = true
  codedeploy_deployment_group_alarm_configuration_alarms          = [aws_cloudwatch_metric_alarm.error_rate.id]
  codedeploy_deployment_group_auto_rollback_configuration_enabled = true
  codedeploy_deployment_group_auto_rollback_configuration_events  = ["DEPLOYMENT_FAILURE", "DEPLOYMENT_STOP_ON_ALARM"]
  codepipeline_artifact_store_bucket                              = aws_s3_bucket.source.bucket
  deployment_config_name                                          = aws_codedeploy_deployment_config.canary.id
  function_name                                                   = local.function_name
  s3_bucket                                                       = aws_s3_bucket.source.bucket
  s3_key                                                          = local.s3_key
}

resource "aws_codedeploy_deployment_config" "canary" {
  deployment_config_name = "custom-lambda-canary-deployment-config"
  compute_platform       = "Lambda"

  traffic_routing_config {
    type = "TimeBasedCanary"

    time_based_canary {
      interval   = 5
      percentage = 50
    }
  }
}

with custom CodePipeline steps

see complete example for details:

// see above and make sure to add required IAM permissions

module "deployment" {
  source = "moritzzimmer/lambda/aws//modules/deployment"

  // see above
  codepipeline_post_deployment_stages = [
    {
      name = "Custom"

      actions = [
        {
          name            = "CustomCodeBuildStep"
          category        = "Build"
          owner           = "AWS"
          provider        = "CodeBuild"
          version         = "1"
          input_artifacts = ["deploy"]

          configuration = {
            ProjectName : aws_codebuild_project.custom_step.name
          }
        }
      ]
    }
  ]
}

Examples

Requirements

Name Version
terraform >= 1.3
aws >= 5.32

Providers

Name Version
aws >= 5.32

Modules

No modules.

Resources

Name Type
aws_cloudwatch_event_rule.s3_trigger resource
aws_cloudwatch_event_rule.this resource
aws_cloudwatch_event_target.s3_trigger resource
aws_cloudwatch_event_target.trigger resource
aws_cloudwatch_log_group.this resource
aws_codebuild_project.this resource
aws_codedeploy_app.this resource
aws_codedeploy_deployment_group.this resource
aws_codepipeline.this resource
aws_codestarnotifications_notification_rule.notification resource
aws_iam_role.codebuild_role resource
aws_iam_role.codedeploy resource
aws_iam_role.codepipeline_role resource
aws_iam_role.trigger resource
aws_iam_role_policy_attachment.codedeploy resource
aws_s3_bucket.pipeline resource
aws_s3_bucket_public_access_block.source resource
aws_s3_bucket_server_side_encryption_configuration.pipeline resource
aws_sns_topic.notifications resource
aws_sns_topic_policy.notifications resource
aws_caller_identity.current data source
aws_iam_policy_document.sns_codestar_policy data source
aws_partition.current data source
aws_region.current data source

Inputs

Name Description Type Default Required
alias_name Name of the Lambda alias used in CodeDeploy. string n/a yes
codebuild_cloudwatch_logs_retention_in_days Specifies the number of days you want to retain log events in the CodeBuild log group. number 14 no
codebuild_environment_compute_type Information about the compute resources the build project will use. string "BUILD_LAMBDA_1GB" no
codebuild_environment_image Docker image to use for this build project. The image needs to include python. string "aws/codebuild/amazonlinux-aarch64-lambda-standard:python3.12" no
codebuild_environment_type Type of build environment to use for related builds. string "ARM_LAMBDA_CONTAINER" no
codebuild_role_arn ARN of an existing IAM role for CodeBuild execution. If empty, a dedicated role for your Lambda function with minimal required permissions will be created. string "" no
codedeploy_appspec_hooks_after_allow_traffic_arn Lambda function ARN to run after traffic is shifted to the deployed Lambda function version. string "" no
codedeploy_appspec_hooks_before_allow_traffic_arn Lambda function ARN to run before traffic is shifted to the deployed Lambda function version. string "" no
codedeploy_deployment_group_alarm_configuration_alarms A list of alarms configured for the deployment group. A maximum of 10 alarms can be added to a deployment group. list(string) [] no
codedeploy_deployment_group_alarm_configuration_enabled Indicates whether the alarm configuration is enabled. This option is useful when you want to temporarily deactivate alarm monitoring for a deployment group without having to add the same alarms again later. bool false no
codedeploy_deployment_group_alarm_configuration_ignore_poll_alarm_failure Indicates whether a deployment should continue if information about the current state of alarms cannot be retrieved from CloudWatch. bool false no
codedeploy_deployment_group_auto_rollback_configuration_enabled Indicates whether a defined automatic rollback configuration is currently enabled for this deployment group. If you enable automatic rollback, you must specify at least one event type. bool false no
codedeploy_deployment_group_auto_rollback_configuration_events The event type or types that trigger a rollback. Supported types are DEPLOYMENT_FAILURE and DEPLOYMENT_STOP_ON_ALARM list(string) [] no
codepipeline_artifact_store_bucket Name of an existing S3 bucket used by AWS CodePipeline to store pipeline artifacts. Use the same bucket name as in s3_bucket to store deployment packages and pipeline artifacts in one bucket for package_type=Zip functions. If empty, a dedicated S3 bucket for your Lambda function will be created. string "" no
codepipeline_artifact_store_encryption_key_id The KMS key ARN or ID of a key block AWS CodePipeline uses to encrypt the data in the artifact store, such as an AWS Key Management Service (AWS KMS) key. If you don't specify a key, AWS CodePipeline uses the default key for Amazon Simple Storage Service (Amazon S3). string "" no
codepipeline_post_deployment_stages A map of post deployment stages to execute after the Lambda function has been deployed. The following stages are supported: CodeBuild, CodeDeploy, CodePipeline, CodeStarNotifications.
list(object({
name = string
actions = list(object({
name = string
category = string
owner = string
provider = string
version = string
input_artifacts = optional(list(any))
output_artifacts = optional(list(any))
configuration = optional(map(string))
}))
}))
[] no
codepipeline_role_arn ARN of an existing IAM role for CodePipeline execution. If empty, a dedicated role for your Lambda function with minimal required permissions will be created. string "" no
codepipeline_type Type of the CodePipeline. Possible values are: V1 and V2. string "V1" no
codepipeline_variables CodePipeline variables. Valid only when codepipeline_type is V2.
list(object({
name = string
default_value = optional(string)
description = optional(string)
}))
[] no
codestar_notifications_detail_type The level of detail to include in the notifications for this resource. Possible values are BASIC and FULL. string "BASIC" no
codestar_notifications_enabled Enable CodeStar notifications for your pipeline. bool true no
codestar_notifications_event_type_ids A list of event types associated with this notification rule. For list of allowed events see https://docs.aws.amazon.com/dtconsole/latest/userguide/concepts.html#events-ref-pipeline. list(string)
[
"codepipeline-pipeline-pipeline-execution-succeeded",
"codepipeline-pipeline-pipeline-execution-failed"
]
no
codestar_notifications_target_arn Use an existing ARN for a notification rule target (for example, a SNS Topic ARN). Otherwise a separate sns topic for this service will be created. string "" no
deployment_config_name The name of the deployment config used in the CodeDeploy deployment group, see https://docs.aws.amazon.com/codedeploy/latest/userguide/deployment-configurations.html for all available default configurations or provide a custom one. string "CodeDeployDefault.LambdaAllAtOnce" no
ecr_image_tag The container tag used for ECR/container based deployments. string "latest" no
ecr_repository_name Name of the ECR repository source used for ECR/container based deployments, required for package_type=Image. string "" no
function_name The name of your Lambda Function to deploy. string n/a yes
s3_bucket Name of the bucket used for S3 based deployments, required for package_type=Zip. Make sure to enable S3 bucket notifications for this bucket for continuous deployment of your Lambda function, see https://docs.aws.amazon.com/AmazonS3/latest/userguide/EventBridge.html. string "" no
s3_key Object key used for S3 based deployments, required for package_type=Zip. string "" no
tags A mapping of tags to assign to all resources supporting tags. map(string) {} no

Outputs

Name Description
codebuild_project_arn The Amazon Resource Name (ARN) of the CodeBuild project.
codebuild_project_id The Id of the CodeBuild project.
codedeploy_app_arn The Amazon Resource Name (ARN) of the CodeDeploy application.
codedeploy_app_name The name of the CodeDeploy application.
codedeploy_deployment_group_arn The Amazon Resource Name (ARN) of the CodeDeploy deployment group.
codedeploy_deployment_group_deployment_group_id The ID of the CodeDeploy deployment group.
codedeploy_deployment_group_id Application name and deployment group name.
codepipeline_arn The Amazon Resource Name (ARN) of the CodePipeline.
codepipeline_artifact_storage_arn The Amazon Resource Name (ARN) of the CodePipeline artifact store.
codepipeline_id The ID of the CodePipeline.
codepipeline_role_name The name of the IAM role used for the CodePipeline.