CDK LakeFormation permissions

0

Hi, I am trying to create Glue database and grant permissions on it in Lake Formation. I tried several ways and several IAM roles and policies based on the documentation but every time I get Insufficient Lake Formation permission(s): Required Create Database on Catalog. The code is pretty much straightforward and when I define permissions in Lake Formations I specify ALL value. Does anyone have an idea what is missing in the code?

class ExampleStack(Stack):

    def __init__(self, scope: Construct, construct_id: str, **kwargs) -> None:
        super().__init__(scope, construct_id, **kwargs)

        custom_data_bucket_arn = "arn:aws:s3:::transformed-v5"
        s3_location = "s3://transformed-v5"
        bucket_name = "transformed-v5"

        glue_role = cdk.aws_iam.Role(self, "glue_role", 
            assumed_by=cdk.aws_iam.ServicePrincipal('glue.amazonaws.com'),
            managed_policies= [
                cdk.aws_iam.ManagedPolicy.from_managed_policy_arn(self, 'MyCrawlerGlueRole', 
                    managed_policy_arn='arn:aws:iam::aws:policy/service-role/AWSGlueServiceRole'
                    )
                ]
        )
        
        glue_role.add_to_policy(cdk.aws_iam.PolicyStatement(actions=['s3:*', 'lakeformation:GetDataAccess', "lakeformation:GrantPermissions"], effect=cdk.aws_iam.Effect.ALLOW, resources=['*']))
        glue_role.add_to_policy(cdk.aws_iam.PolicyStatement(actions=["iam:PassRole"], effect=cdk.aws_iam.Effect.ALLOW, resources=['arn:aws:iam::xxxxxxx:role/{}'.format(glue_role.role_name)]))

        glue_db=cdk.aws_glue.CfnDatabase(self, "MyDatabase", 
            catalog_id=cdk.Aws.ACCOUNT_ID,
            database_input=cdk.aws_glue.CfnDatabase.DatabaseInputProperty(
               
                name = "datalake-v5",
                location_uri = s3_location
               
            )
        )
        cdk.aws_glue.CfnCrawler(self, "datalake-crawler", 
            database_name= "datalake-v5",
            role=glue_role.role_arn,
            schedule={"scheduleExpression":"cron(0/15 * * * ? *)"},
            targets={"s3Targets": [{"path":  bucket_name}]}, 
        ) 

        location_resource = cdk.aws_lakeformation.CfnResource(self, 
                "MyDatalakeLocationResource", 
                resource_arn= custom_data_bucket_arn, 
                use_service_linked_role=True
        )
        cdk.aws_lakeformation.CfnPermissions(self, "MyDatalakeDatabasePermission",
                data_lake_principal=cdk.aws_lakeformation.CfnPermissions.DataLakePrincipalProperty(data_lake_principal_identifier=glue_role.role_arn),
                resource=cdk.aws_lakeformation.CfnPermissions.ResourceProperty(database_resource=cdk.aws_lakeformation.CfnPermissions.DatabaseResourceProperty(name="datalake-v5")),
                permissions=["ALL"],
                permissions_with_grant_option=["ALL"]
            )
        location_permission = cdk.aws_lakeformation.CfnPermissions(self, "MyDatalakeLocationPermission",
                data_lake_principal=cdk.aws_lakeformation.CfnPermissions.DataLakePrincipalProperty(data_lake_principal_identifier=glue_role.role_arn),
                resource=cdk.aws_lakeformation.CfnPermissions.ResourceProperty(data_location_resource=cdk.aws_lakeformation.CfnPermissions.DataLocationResourceProperty(s3_resource=custom_data_bucket_arn)),
                permissions=["DATA_LOCATION_ACCESS"]
            )
   
        location_permission.add_dependency(location_resource)

Thank you!

Moma
已提問 1 年前檢視次數 1350 次
1 個回答
0

The issue seems to be that CDK on deployment fails (please correct me if I'm wrong).
Only Lakeformation administrators or "database creators" can do that, the permission ALL applies once it is created.
Instead of granting the CDK deployment role such privilege, if you only need one it might be easier to create it manually by an administrator.

https://docs.aws.amazon.com/lake-formation/latest/dg/creating-database.html

profile pictureAWS
專家
已回答 1 年前

您尚未登入。 登入 去張貼答案。

一個好的回答可以清楚地回答問題並提供建設性的意見回饋,同時有助於提問者的專業成長。

回答問題指南