fix(eks): in place updates for EKS security group and Subnets

[mrlikl]

Issue # (if applicable)

Closes #28584

Reason for this change

To have in place updates for EKS clusters when subnets or SG values are changed.

Description of changes

Removed replaceVpc logic and introduced updateVpc to track changes and errors to handle multiple updates in one go

Description of how you validated changes

Have tested the changes by first deploying a cluster with below config:

const vpc = ec2.Vpc.fromLookup(stack, 'Vpc', { isDefault: true });
new eks.Cluster(stack, 'Cluster', {
  vpc,
  ...getClusterVersionConfig(stack, eks.KubernetesVersion.V1_24),
  defaultCapacity: 0,
});

TestCase - 1 Update both subnets and Access at the same time

new eks.Cluster(stack, 'Cluster', {
  vpc,
  ...getClusterVersionConfig(stack, eks.KubernetesVersion.V1_29),
  defaultCapacity: 0,
  tags: {
    foo: 'bar',
  },
  endpointAccess: eks.EndpointAccess.PUBLIC,
  vpcSubnets: [{ subnetType: ec2.SubnetType.PUBLIC }],
});

Error below is thrown for Cluster custom resource -

{
    "errorType": "Error",
    "errorMessage": "Only one type of update - VpcConfigUpdate, LoggingUpdate or EndpointAccessUpdate can be allowed",
    "stack": [
        "Error: Only one type of update - VpcConfigUpdate, LoggingUpdate or EndpointAccessUpdate can be allowed",
        "    at Pi.onUpdate (/var/task/index.js:55:651127)",
        "    at Pi.onEvent (/var/task/index.js:55:647590)",
        "    at Runtime.yR [as handler] (/var/task/index.js:55:657995)",
        "    at Runtime.handleOnceNonStreaming (file:///var/runtime/index.mjs:1173:29)"
    ]
}

TestCase - 2 Update subnets to public

new eks.Cluster(stack, 'Cluster', {
  vpc,
  ...getClusterVersionConfig(stack, eks.KubernetesVersion.V1_29),
  defaultCapacity: 0,
  vpcSubnets: [{ subnetType: ec2.SubnetType.PUBLIC }],
});

{
    "updates": {
        "replaceName": false,
        "updateVpc": true,
        "updateAccess": false,
        "replaceRole": false,
        "updateVersion": false,
        "updateEncryption": false,
        "updateLogging": false
    }
}

{
  clientName: 'EKSClient',
  commandName: 'UpdateClusterConfigCommand',
  input: {
    name: 'Cluster9EE0221C-0b6f58b0698348aea43866b93a62b2c9',
    resourcesVpcConfig: { subnetIds: [Array], securityGroupIds: [Array] }
  },
  output: {
    update: {
      createdAt: 2024-05-08T20:55:00.013Z,
      errors: [],
      id: '7d5cd243-5536-3f52-b5ca-4c6e6c044529',
      params: [Array],
      status: 'InProgress',
      type: 'VpcConfigUpdate'
    }
  },
  metadata: {}
}

Checklist

• My code adheres to the CONTRIBUTING GUIDE and DESIGN GUIDELINES

---

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

[aws-cdk-automation]

The pull request linter has failed. See the aws-cdk-automation comment below for failure reasons. If you believe this pull request should receive an exemption, please comment and provide a justification.

A comment requesting an exemption should contain the text Exemption Request. Additionally, if clarification is needed add Clarification Request to a comment.

✅ Updated pull request passes all PRLinter validations. Dismissing previous PRLinter review.

[pahud]

Hey, this PR makes sense to me. Thanks for your ownership.

I guess we probably need to add some unit tests at packages/@aws-cdk/custom-resource-handlers/test/aws-eks/cluster-resource-provider.test.ts though. What do you think?

[mrlikl]

@pahud yes, I am checking the same

[GavinZZ]

Thanks for the PR. I have a question regarding this change. I notice that you removed replaceVpc and replace it updateVpc when subnet id and security group id changes. This looks valid to me.

However, I'm curious would there be a case where the actual VPC is replaced and we still want to keep replaceVpc in this case?

[GavinZZ]

How does this test test subnet updates? It only get deployed once right? Want to make sure you've tested the flow manually for update.

[mrlikl]

Yes the flow was manually tested, and then changed to this integ test here.

[mrlikl]

Thanks for the PR. I have a question regarding this change. I notice that you removed replaceVpc and replace it updateVpc when subnet id and security group id changes. This looks valid to me.

However, I'm curious would there be a case where the actual VPC is replaced and we still want to keep replaceVpc in this case?

@GavinZZ EKS does not support updating the subnets (to a different VPC) post its creation at the moment. In CFN, if we specify subnets from a different VPC it does not trigger a replacement type update for the resource rather the error Subnets specified must belong to the VPC: vpc-xxxxx (Service: Eks, Status Code: 400, Request ID: 47f6ddc6-e957-471b-b97f-58c837054879)" (RequestToken: 31954395-3f14-3635-1c33-69987d6e21c4, HandlerErrorCode: InvalidRequest) is thrown. Same is the case with SG belonging to a different VPC - Security group(s) [sg-xxxx] are not in the same VPC as the subnets. Please specify a security group that is associated with the VPC: vpc-xxxxx (Service: Eks, Status Code: 400, Request ID: 48021c3d-b2af-4375-b7ad-6ca6d4341b78)" (RequestToken: 4f63b78c-a19a-1271-2c1b-2d4303c0e126, HandlerErrorCode: InvalidRequest)

So to match the CFN behavior to not replace the cluster, we can let go of replaceVpc here.

[GavinZZ]

Thanks for the explanation! LGTM

[mergify]

Thank you for contributing! Your pull request will be updated from main 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: AutoBuildv2Project1C6BFA3F-wQm2hXv2jqQv
• Commit ID: 2ee56ab
• 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 main and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).