fix(redshift): cluster uses key ARN instead of key ID

[iRoachie]

Field was incorrectly using key arn instead of id.

Fixes #17032

---

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license

[gitpod-io]

[njlynch]

The change itself looks great (one minor nitpick).

I did want to clarify something you said in the linked issue before approving, however.

I wasn't able to confirm if this prevents successful deployment but it can impact the ability to migrate from vanilla CFN to CDK because the change from KeyId to KeyArn triggers an encryption change in the cluster.

The CloudFormation docs for this property states that an update requires no interruption. What exactly do you mean by "encryption change"? Is there downtime associated with this? I ask because by making this change, it's quite possible we are causing the same effect in reverse (moving from ARN to ID) for any existing customers.

[iRoachie]

Oh you're referring to @gkubes. Based on the docs it shouldn't require interruption.

In the event that it does change though, shouldn't this change still be made? We can try a test to see if it fails currently. Right now there's no integ tests that include kms.

[iRoachie]

Just confirmed. The stack does deploy with ARN successfully. Do we still make this change?

import * as cdk from "@aws-cdk/core";
import * as redshift from "@aws-cdk/aws-redshift";
import * as ec2 from "@aws-cdk/aws-ec2";
import * as kms from "@aws-cdk/aws-kms";

export class RedshiftStack extends cdk.Stack {
  constructor(scope: cdk.Construct, id: string, props?: cdk.StackProps) {
    super(scope, id, props);

    const key = new kms.Key(this, 'id');

    const vpc = new ec2.Vpc(this, "Vpc");
    const databaseName = "my_db";
    const cluster = new redshift.Cluster(this, "Cluster", {
      vpc: vpc,
      vpcSubnets: {
        subnetType: ec2.SubnetType.PUBLIC,
      },
      masterUser: {
        masterUsername: "admin",
      },
      defaultDatabaseName: databaseName,
      publiclyAccessible: true,
      removalPolicy: cdk.RemovalPolicy.DESTROY,
      encryptionKey: key
    });

    const databaseOptions = {
      cluster: cluster,
      databaseName: databaseName,
    };

    const user = new redshift.User(this, "User", databaseOptions);
    const table = new redshift.Table(this, "Table", {
      ...databaseOptions,
      tableColumns: [
        { name: "col1", dataType: "varchar(4)" },
        { name: "col2", dataType: "float" },
      ],
    });
    table.grant(user, redshift.TableAction.INSERT, redshift.TableAction.DELETE);

    // The code that defines your this goes here
  }
}

[njlynch]

Based on the docs it shouldn't require interruption. In the event that it does change though, shouldn't this change still be made? We can try a test to see if it fails currently. Right now there's no integ tests that include kms.

Yeah, I likewise was able to deploy with the current setup (using the ARN). Given the docs state "no interruption", I'm going to trust the docs and say let's proceed.

[mergify]

Thank you for contributing! Your pull request will be updated from master and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).

[mergify]

Thank you for contributing! Your pull request will be updated from master and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).

[aws-cdk-automation]

AWS CodeBuild CI Report

• CodeBuild project: AutoBuildProject89A8053A-LhjRyN9kxr8o
• Commit ID: 6640fa4
• Result: SUCCEEDED
• Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

[mergify]

Thank you for contributing! Your pull request will be updated from master and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).

[WhereEam]

@njlynch Hello Nick, we are has been having some troubles with this merge. As you have suspected:

it's quite possible we are causing the same effect in reverse (moving from ARN to ID) for any existing customers.

Our team has seen this issue when updating from monoCDK 1.126.0 to 1.130.0. Receiving the error "This encryption info combination is invalid." A ticket was opened against redshift team but the responses have been slow. Would just like to notify you guys so we start looking at mitigation plans as this is a breaking change. The issue can be mitigated by deleting the entire stack and redeploying but this is not a good solution for productions as accounts can not go through this type of large scale change.