为什么在CloudFormation中使用嵌套堆栈时会出现"No Export“错误？

Question

我在CloudFormation模板中定义一个导出，以便在另一个模板中使用。

我可以看到导出是在AWS控制台中创建的，但是第二个堆栈找不到它。

错误：

UPDATE_ROLLBACK_IN_PROGRESS with reason: No export named sandbox06-ODM-KinesisStreamArn found

template.yml

Resources:
  KinesisStream:
    Type: AWS::Kinesis::Stream
    Properties:
      ShardCount: 1
      RetentionPeriodHours: 24
      Name: !Sub ${Environment}-${Application}
Outputs:
  Topic:
    Value: !Ref Topic
  KinesisStreamArn:
    Value: !GetAtt KinesisStream.Arn
    Export:
      Name: !Sub ${Environment}-${Domain}-KinesisStreamArn

firehose.yml

KinesisFirehoseRole:
    Type: AWS::IAM::Role
    Properties:
      AssumeRolePolicyDocument:
        Version: 2012-10-17
        Statement:
          - Effect: Allow
            Principal:
              Service: firehose.amazonaws.com
            Action: sts:AssumeRole
      Path: /
      Policies:
        - PolicyName: KinesisFirehosePolicy
          PolicyDocument:
            Version: 2012-10-17
            Statement:
              - Effect: Allow
                Action:
                  - kinesis:*
                  - s3:*
                  - s3-object-lambda:*
                Resource:
                  - !Sub "${Bucket.Arn}/*"
                  - Fn::ImportValue: !Sub "${Environment}-${Domain}-KinesisStreamArn"

nested_template.yml

  OperationalData:
    Type: AWS::Serverless::Application
    Properties:
      Parameters:
        Environment: !Ref Environment
        Domain: OperationalData
        Application: odm
        BucketPrefix: pie
        WhiteListCidr: !Ref WhiteListCidr
        VpcId:
          Fn::ImportValue: !Sub vpc-${Environment}-VPCID
        VpcCidr:
          Fn::ImportValue: !Sub vpc-${Environment}-VPCCIDR
        Subnets:
          Fn::ImportValue: !Sub vpc-${Environment}-PrivateSubnets
      Location: ./data/odm/template.yml
      Tags:
        Environment: !Ref Environment
        Domain: odm
        Application: !Ref Application
        Developer: !Ref Developer
        DevOpsAdmin: !Ref DevOpsAdmin
        Repository: !Ref Repository
        Team: !Ref Team

  DataEngineeringData:
    Type: AWS::Serverless::Application
    Properties:
      Parameters:
        Environment: !Ref Environment
        Domain: DataEngineeringData
        Application: data-engineering
      Location: ./data/data-engineering/template.yml
      Tags:
        Environment: !Ref Environment
        Domain: DataEngineeringData
        Application: data-engineering
        Developer: !Ref Developer
        DevOpsAdmin: !Ref DevOpsAdmin
        Repository: !Ref Repository
        Team: !Ref Team

有什么问题吗？

Answer 1

第二个堆栈找不到它

这是因为默认情况下，嵌套的CloudFormation堆栈是并行创建的。

这意味着，如果您的一个子堆栈--例如包含KinesisFirehoseRole的堆栈--从另一个子堆栈(例如包含KinesisStream的堆栈)导入输出，则堆栈创建将失败。

这是因为在并行创建它们时，CloudFormation如何确保导出值在另一个创建的子堆栈正在导入时已经导出？

若要修复此问题，请在包含KinesisFirehoseRole**.**的堆栈上使用 属性。

这应该指向包含KinesisStream的堆栈，因为KinesisFirehoseRole对它有依赖关系。

DependsOn将此依赖关系显式化，并将确保正确的堆栈创建顺序。

像这样的事情应该有效：

Stack-Containing-Kinesis-Firehose-Role:
   Type: AWS::CloudFormation::Stack
   DependsOn: Stack-Containing-Kinesis-Stream-Resource-Which-Outputs-KinesisStreamArn
   ...