如何在Terraform中按需动态定义AWS CodePipeline的Stage？

要实现不同Pipeline实例按需使用不同数量的Stage，核心思路是将基础Stage与可选Stage分离，通过模块变量动态合并Stage列表，避免硬编码固定数量的Stage导致所有实例强制继承。

具体实现步骤 #

1. 在模块中定义可选Stage变量

在CodePipeline模块里新增一个additional_stages变量，用于接收实例需要的额外Stage配置，默认值为空列表：

variable "additional_stages" {
  type = list(object({
    name = string
    actions = list(object({
      name             = string
      action_type_id   = object({
        category = string
        owner    = string
        provider = string
        version  = string
      })
      run_order        = number
      configuration    = optional(map(string))
      output_artifacts = optional(list(object({ name = string })))
      input_artifacts  = optional(list(object({ name = string })))
      role_arn         = optional(string)
    }))
  }))
  default     = []
  description = "可选扩展Stage，用于为Pipeline增加额外阶段"
}

2. 动态合并基础Stage与可选Stage

在aws_codepipeline资源中，使用concat函数将固定的2个基础Stage与additional_stages合并，再通过dynamic块遍历生成所有Stage：

resource "aws_codepipeline" "main" {
  name     = var.pipeline_name
  role_arn = var.pipeline_role_arn

  artifact_store {
    location = var.artifact_bucket_name
    type     = "S3"

    encryption_key {
      id   = var.kms_key_arn
      type = "KMS"
    }
  }

  # 动态生成所有Stage：基础2个 + 可选额外Stage
  dynamic "stage" {
    for_each = concat([
      # 基础Stage 1: Source
      {
        name = "Source"
        actions = [
          # 填入你的Source阶段具体配置
        ]
      },
      # 基础Stage 2: Build
      {
        name = "Build"
        actions = [
          # 填入你的Build阶段具体配置
        ]
      }
    ], var.additional_stages)

    content {
      name = stage.value.name

      # 动态生成当前Stage下的所有Action
      dynamic "action" {
        for_each = stage.value.actions
        content {
          name             = action.value.name
          action_type_id   = action.value.action_type_id
          run_order        = action.value.run_order
          configuration    = action.value.configuration
          output_artifacts = action.value.output_artifacts
          input_artifacts  = action.value.input_artifacts
          role_arn         = action.value.role_arn
        }
      }
    }
  }
}

3. 按需调用模块

• 仅需2个Stage的Pipeline：直接调用模块，不传递additional_stages变量，默认使用基础的2个Stage：

module "two_stage_pipeline" {
  source              = "./modules/codepipeline"
  pipeline_name       = "prod-two-stage-pipeline"
  pipeline_role_arn   = aws_iam_role.pipeline_role.arn
  artifact_bucket_name = aws_s3_bucket.artifact_bucket.bucket
  kms_key_arn         = aws_kms_key.pipeline_key.arn
}

• 需要3个Stage的Pipeline：传递additional_stages变量，填入第三个Stage（如Deploy）的具体配置：

module "three_stage_pipeline" {
  source              = "./modules/codepipeline"
  pipeline_name       = "dev-three-stage-pipeline"
  pipeline_role_arn   = aws_iam_role.pipeline_role.arn
  artifact_bucket_name = aws_s3_bucket.artifact_bucket.bucket
  kms_key_arn         = aws_kms_key.pipeline_key.arn

  additional_stages = [
    {
      name = "Deploy"
      actions = [
        {
          name = "DeployToECS"
          action_type_id = {
            category = "Deploy"
            owner    = "AWS"
            provider = "ECS"
            version  = "1"
          }
          run_order = 1
          configuration = {
            ClusterName = aws_ecs_cluster.dev.name
            ServiceName = aws_ecs_service.dev.name
            ImageURI    = "${aws_ecr_repository.app.repository_url}:latest"
          }
          role_arn = aws_iam_role.deploy_role.arn
        }
      ]
    }
  ]
}

方案优势 #

• 灵活性高：每个Pipeline实例可以按需扩展任意数量的Stage，不限于3个
• 配置解耦：基础Stage与扩展Stage的配置分离，避免重复代码
• 兼容性好：原有仅需2个Stage的实例无需修改，直接复用模块

内容的提问来源于stack exchange，提问作者Shanmugavel