amazon-web-services - 无效的操作配置 : Exception while trying to read the task definition artifact file

Question

我有一个 Codepipeline，它使用 ECS 蓝/绿部署操作将 ECR 镜像部署到 ECS 集群。该管道包含两个源:一个用于使用 AWS ECR 操作的 ECR 图像，另一个用于使用第三方 Github 操作从 Github 获取配置。 ECR 操作输出 image 工件，而 Github 操作输出 config 工件。这两个工件都作为 ECS 蓝/绿部署操作的输入提供。
管道在 ECS 蓝/绿部署操作失败，并显示以下错误(从 AWS 控制台可见):

Invalid action configuration

Exception while trying to read the task definition artifact file from: config

这是管道的结构(为了匿名，编辑了一些细节):

$ aws codepipeline get-pipeline --name my-codepipeline
{
  "pipeline": {
    "name": "my-codepipeline",
    "roleArn": "arn:aws:iam::123456789000:role/my-codepipeline-role",
    "artifactStore": {
      "type": "S3",
      "location": "my-codepipeline-s3"
    },
    "stages": [
      {
        "name": "Source",
        "actions": [
          {
            "name": "ImageSource",
            "actionTypeId": {
              "category": "Source",
              "owner": "AWS",
              "provider": "ECR",
              "version": "1"
            },
            "runOrder": 1,
            "configuration": {
              "ImageTag": "latest",
              "RepositoryName": "my-ecr"
            },
            "outputArtifacts": [
              {
                "name": "image"
              }
            ],
            "inputArtifacts": []
          },
          {
            "name": "ConfigSource",
            "actionTypeId": {
              "category": "Source",
              "owner": "ThirdParty",
              "provider": "GitHub",
              "version": "1"
            },
            "runOrder": 1,
            "configuration": {
              "Branch": "master",
              "OAuthToken": "****",
              "Owner": "me",
              "PollForSourceChanges": "false",
              "Repo": "application-config"
            },
            "outputArtifacts": [
              {
                "name": "config"
              }
            ],
            "inputArtifacts": []
          }
        ]
      },
      {
        "name": "Deploy",
        "actions": [
          {
            "name": "DeployBackend",
            "actionTypeId": {
              "category": "Deploy",
              "owner": "AWS",
              "provider": "CodeDeployToECS",
              "version": "1"
            },
            "runOrder": 1,
            "configuration": {
              "AppSpecTemplateArtifact": "config",
              "AppSpecTemplatePath": "production/appspec.yaml",
              "ApplicationName": "my-codedeploy",
              "DeploymentGroupName": "my-codedeploy-group",
              "Image1ArtifactName": "image",
              "Image1ContainerName": "IMAGE_NAME",
              "TaskDefinitionTemplateArtifact": "config",
              "TaskDefinitionTemplatePath": "production/taskdef.json"
            },
            "outputArtifacts": [],
            "inputArtifacts": [
              {
                "name": "image"
              },
              {
                "name": "config"
              }
            ]
          }
        ]
      }
    ],
    "version": 1
  },
  "metadata": {
    "pipelineArn": "arn:aws:codepipeline:ap-northeast-1:123456789000:my-codepipeline",
    "created": 1564107543.285,
    "updated": 1564107543.285
  }
}

我已经检查了 S3 中的压缩工件，它肯定包含 Github 存储库中 AppSpecTemplatePath 和 TaskDefinitionTemplatePath 指定位置的配置文件。
这是 appspec.yaml 的内容:

$ cat production/appspec.yaml
version: 0.0
Resources:
  - TargetService:
      Type: AWS::ECS::Service
      Properties:
        TaskDefinition: <TASK_DEFINITION>
        LoadBalancerInfo:
          ContainerName: "my-container"
          ContainerPort: 80

Answer 1

在广泛尝试之后，我偶然发现了一个我找不到的外语线程。该线程说传递给操作的工件不能大于 3 MB。

我通过减小工件的大小( config )解决了我的问题。配置存储库在许多项目之间共享，通过将这些项目移动到另一个项目，我将压缩的工件大小从 14 MB 减少到 3 kB。奇迹般地，一切正常。 AWS，如果您正在阅读本文，请向 ECS CodeDeploy 添加更多关于工件大小限制的文档，因为我没有看到任何关于此的提及，而且我无法使用如此一般的错误消息调试此问题。