fix(eks): cluster-resource-handler fails to verify delete operations

[kishiel]

Motivation

The recent upgrade to the underlying aws-sdk-js dependency in the custom resource handler caused deletion events to fail. This was due to the response structure differing between sdk versions, and the new response lacked the property the handler was previously targeting.

This change leverages the httpResponseCode to follow the same logic as before.

Some additional changes that come with this fix:

• Nginx dashboard's latest update breaks the cluster tests, and as a result I've removed them
• Nginx ingress controller was also removed as it is affected by a separate new defect either in EKS or VPC (I can't tell) which causes NLBs created by the EKS cluster to survive stack deletion. This appears to be a transient issue, but it even appeared in the simplistic hello-k8s chart.

There's a broader question I have about the use of external Helm charts in these snapshot tests which do not target specific versions--this causes the snapshot tests, which are supposed to be deterministic, to become indeterministic. Has the EKS team considered owning a few simple helm chart repos which provide us with stable versions (and thus deterministic tests)?

I spent nearly 5 days fighting with the snapshot tests and found that a few tests (eks-cluster-test, bottlerocket, and cluster-imported test) would never use new assets and instead would deploy using the old. If I synthesized the test using npx cdk -a test/integ.testname.js and deployed that it would use the new assets. The only way I got past this was to completely delete the existing snapshot directory. I couldn't find references to this behavior anywhere, but I wanted to call it out in case I deleted something that was not replaced with the snapshot update.

Remaining work

The unit tests for the cluster and fargate handler were not updated. I took a stab at them but the interface mocks that we have are no longer accurate, and I couldn't get them to mimic this change.

Fixes #26325

---

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

[aws-cdk-automation]

AWS CodeBuild CI Report

• CodeBuild project: AutoBuildv2Project1C6BFA3F-wQm2hXv2jqQv
• Commit ID: 8eb6cce
• Result: FAILED
• Build Logs (available for 30 days)

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

[mrgrain]

This is great @kishiel thanks for looking into this. Can you check why the unit tests fail?

[mrgrain]

Thanks again @kishiel Gonna close this as the team is looking into this with priority now.
Reference: #26283