amazon-web-services - AWS : cloudformation CommaDelimitedList and manual list not matching

Question

我正在尝试创建一个具有默认值的 cloudformation 模板，并且我正在运行一些 !Sub 函数来将导入的参数替换到模板中。但是，我将一个列表传递给 Nodejs Lambda 函数，在发送它之前我需要 !Sub 。

我正在编写的代码:

AWSTemplateFormatVersion: '2010-09-09'
Transform: 'AWS::Serverless-2016-10-31'
Description: Creating Athena database and tables

Parameters:
  S3DataLocations:
    Type: CommaDelimitedList
    Description: The S3 locations where the logs are read from (Specify 'Use,Default' to inherit defaults)
    Default: Use,Default

Conditions:
  CustomS3DataLocations: !Equals ['Use,Default', !Join [",", !Ref S3DataLocations]]

Resources:
  # Custom resource for running CreateTableFunction, to create databases
  CreateLogTable:
    Type: Custom::CreateLogTable
    Properties:
      ServiceToken: !GetAtt [CreateLogTableFunction, Arn]
      S3DataLocations:
        Fn::If:
          - CustomS3DataLocations
          - !Split
            - ","
            - !Sub
              - s3://${LoggingBucket}/data/ApplicationLogs1/,
                s3://${LoggingBucket}/data/ApplicationLogs2/,
                s3://${LoggingBucket}/data/ApplicationLogs3/
              - { LoggingBucket: !ImportValue Parent-LoggingBucket}
          - !Ref S3DataLocations

如果我将它们作为文字外部DataTypes参数传递s3://logbucket/data/ApplicationLogs1/,s3://logbucket/data/ApplicationLogs2/,s3://logbucket/data/ApplicationLogs3/ 它工作正常并转换为 ["s3://logbucket/data/ApplicationLogs1/","s3://logbucket/data/ApplicationLogs2/","s3://logbucket/data/ApplicationLogs3/"] 并由 Lambda 毫无问题地进行解释。该参数通过 CommaDelimitedList 类型进行解析，并顺利传递给 Lambda。

问题的出现是因为我正在尝试创建手动默认值，因此我需要 !Sub 列表作为字符串，然后将 !Split 传递作为自定义 Lambda 的实际列表。这似乎并不是我尝试的每一种方法都有效，而且我不明白为什么。

我一直在检查成功(手动参数)和失败(默认，没有手动参数)，但我看不出有什么大的区别。lambda 的事件显示，工作时:

{
    "RequestType": "Create",
    "ServiceToken": "hidden",
    "ResponseURL": "hidden",
    "StackId": "hidden",
    "RequestId": "hidden",
    "LogicalResourceId": "CreateLogTable",
    "ResourceType": "Custom::CreateLogTable",
    "ResourceProperties": {
        "S3DataLocations": [
            "s3://loggingbucket/data/ApplicationLogs/",
            "s3://loggingbucket/data/ApplicationLogs/",
            "s3://loggingbucket/data/ApplicationLogs/",
            "s3://loggingbucket/data/ApplicationLogs/"
        ]
    }
}

不工作时:

...
{
    "RequestType": "Create",
    "ServiceToken": "hidden",
    "ResponseURL": "hidden",
    "StackId": "hidden",
    "RequestId": "hidden",
    "LogicalResourceId": "CreateLogTable",
    "ResourceType": "Custom::CreateLogTable",
    "ResourceProperties": {
        "S3DataLocations": [
            "s3://logging/data/ApplicationLogs/",
            " s3://loggingbucket/data/ApplicationLogs/",
            " s3://loggingbucket/data/ApplicationLogs/",
            " s3://loggingbucket/data/ApplicationLogs/"
        ]
    }
}

我有点卡在这里，我认为可能存在一些类型不匹配，但我无法区分手册和参数之间的区别。

有人知道吗？

Answer 1

您可以使用引号和斜杠组合将字符串分成多行，同时防止将\n更改为空格。

为了验证这一点，我针对您的情况使用了以下代理模板:

Resources:

  MyBucket:
    Type: AWS::S3::Bucket
    Properties: {}

Outputs:
  Test1:
    Value: !Sub
          - s3://${LoggingBucket}/data/ApplicationLogs1/,
            s3://${LoggingBucket}/data/ApplicationLogs2/,
            s3://${LoggingBucket}/data/ApplicationLogs3/
          - { LoggingBucket: "Parent-LoggingBucket"}

  Test2:
    Value: !Sub
          - "s3://${LoggingBucket}/data/ApplicationLogs1/,\
            s3://${LoggingBucket}/data/ApplicationLogs2/,\
            s3://${LoggingBucket}/data/ApplicationLogs3/"
          - { LoggingBucket: "Parent-LoggingBucket"}   

Test1 生成带有空格的字符串，如您的问题所示:

s3://Parent-LoggingBucket/data/ApplicationLogs1/, s3://Parent-LoggingBucket/data/ApplicationLogs2/, s3://Parent-LoggingBucket/data/ApplicationLogs3/

相比之下，Test2 没有空间:

s3://Parent-LoggingBucket/data/ApplicationLogs1/,s3://Parent-LoggingBucket/data/ApplicationLogs2/,s3://Parent-LoggingBucket/data/ApplicationLogs3/