How to turn on Babelfish with AWS CDK

Question

I have created an Aurora Serverless V2 Cluster with one reader and one writer Instance via CDK.

Now I am trying to enable Babelfish for this cluster by adding a parameter group:

const auroraParameterGroup = new rds.ParameterGroup(this, "AuroraParameterGroup", {
  engine: rds.DatabaseClusterEngine.auroraPostgres({
    version: rds.AuroraPostgresEngineVersion.VER_15_2
  }),
  parameters: {
    "rds.babelfish_status": "on"
  }
})

The AWS RDS Consule UI says that Babelfish is turned on and listening on port 1433.

But when I try to connect to the DB with the MSSQL Server Management Studio, I get this error message: MS MSM Connect Error

This is the Aurora CDK Code:

const auroraCluster = new rds.DatabaseCluster(this, 'AuroraCluster', {
  parameterGroup: auroraParameterGroup,
  engine: rds.DatabaseClusterEngine.auroraPostgres({
    version: rds.AuroraPostgresEngineVersion.VER_15_2
  }),
  storageEncryptionKey: rdsKey,
  instances: 2,
  credentials: rds.Credentials.fromSecret(auroraCredentials),
  instanceProps: {
    enablePerformanceInsights: true,
    publiclyAccessible: true,
    vpc: props.vpc,
    vpcSubnets: props.vpc.selectSubnets({ subnetGroupName: props.databaseSubnetGroupName }),
    instanceType: new InstanceType('serverless'),

  },
});

auroraCluster.connections.allowFromAnyIpv4(aws_ec2.Port.tcp(5432))
auroraCluster.connections.allowFromAnyIpv4(aws_ec2.Port.tcp(1433)) //babelfish

When I create the Aurora DB with Babelfish manually via the AWS Management Console, the connection works.

Answer 1

To connect to the TDS port in Babelfish, you must use the master user that was specified when creating the instance. When you create a PG user through the PG port, you cannot use that use to connect to the TDS port (this is by design) and you'd get the type of error you describe. When running CREATE LOGIN in TDS, this will create a new PG user under the covers which will be allowed to be used to connect to the TDA port. Lastly, note that you have to create a Babelfish-enabled cluster right at the start: when you create a PG cluster first and then try to enable Babelfish separately, this will not work (not sure if that is what you are doing though).