Terraform下AWS Codepipeline配置多个Codecommit源的方案咨询

你要的Terraform管理多CodeCommit源、复刻GitHub模块化使用方式的需求可直接落地，核心逻辑是批量管理CodeCommit仓库资源+CodePipeline多源拉取对齐本地目录结构，无需修改原有Terraform代码的模块引用逻辑。

1. 批量管理CodeCommit模块仓库 #

你可以通过for_each批量定义所有模块类CodeCommit仓库，避免重复编写资源代码，同时可以统一引用所有仓库的属性用于流水线配置：

variable "module_repo_names" {
  type    = list(string)
  default = ["tf-module-vpc", "tf-module-ecs", "tf-module-rds"] # 替换为实际模块仓库名称
}

resource "aws_codecommit_repository" "module_repos" {
  for_each        = toset(var.module_repo_names)
  repository_name = each.value
  description     = "Terraform shared module: ${each.value}"
}

需要引用单个仓库属性时，可通过aws_codecommit_repository.module_repos["tf-module-vpc"].clone_url_ssh的方式调用。

2. 配置CodePipeline多源 #

CodePipeline的源阶段支持通过动态块批量添加所有模块仓库作为独立源，每个源输出独立构件，后续构建阶段统一整理目录：

resource "aws_codepipeline" "infra_deploy" {
  name     = "terraform-infra-deploy"
  role_arn = aws_iam_role.codepipeline_role.arn

  artifact_store {
    location = aws_s3_bucket.codepipeline_artifacts.bucket
    type     = "S3"
  }

  stage {
    name = "Source"
    # 主业务代码源
    action {
      name             = "main-repo"
      category         = "Source"
      owner            = "AWS"
      provider         = "CodeCommit"
      version          = "1"
      output_artifacts = ["main-source"]

      configuration = {
        RepositoryName = "tf-main-infra" # 替换为你的主业务仓库名
        BranchName     = "main"
      }
    }
    # 批量添加所有模块仓库源
    dynamic "action" {
      for_each = aws_codecommit_repository.module_repos
      content {
        name             = "module-${action.value.repository_name}"
        category         = "Source"
        owner            = "AWS"
        provider         = "CodeCommit"
        version          = "1"
        output_artifacts = ["module-${action.value.repository_name}"]

        configuration = {
          RepositoryName = action.value.repository_name
          BranchName     = "main"
        }
      }
    }
  }

  stage {
    name = "Build"
    action {
      name            = "terraform-build"
      category        = "Build"
      owner           = "AWS"
      provider        = "CodeBuild"
      input_artifacts = concat(["main-source"], [for repo in aws_codecommit_repository.module_repos : "module-${repo.repository_name}"])
      version         = "1"

      configuration = {
        ProjectName = aws_codebuild_project.terraform_build.name
      }
    }
  }

  # 后续部署阶段按需自行补充
}

要实现和本地一致的模块化调用体验，只需在CodeBuild的buildspec.yaml中把所有模块构件解压到对应目录，和本地引用路径对齐即可：

version: 0.2
phases:
  install:
    commands:
      - apt update && apt install -y unzip
      # 创建modules目录，和本地开发路径对齐
      - mkdir -p ./modules
      # 批量解压所有模块构件到modules目录
      - for module_zip in $CODEBUILD_SRC_DIR_module-*.zip; do unzip $module_zip -d ./modules/$(basename $module_zip .zip | sed 's/^module-//'); done
      # 解压主业务代码到根目录
      - unzip $CODEBUILD_SRC_DIR_main-source.zip -d ./
  build:
    commands:
      - terraform init
      - terraform plan
      - terraform apply -auto-approve

完成以上配置后，流水线执行时的目录结构和本地开发完全一致，原有基于GitHub的Terraform模块化代码可以直接复用，无需修改任何模块引用路径。

内容的提问来源于stack exchange，提问作者Maxamis4