Report

Certain use cases require the usage of server side dry-run request to validate manifests at Kubernetes API server side, a simple way of simulating that would be, when using a helm chart, to execute the following:

helm -n default template release-name some-chart-name |  kubectl -n default apply -f - --dry-run=server

Assuming the helm chart has the following:

• A ScaledObject defined for a scaleTargetRef of the Deployment kind.
• The ScaledObject is defining a cpu and/or memory trigger.

The kubectl command is going to fail with the following error:

Error from server (NotFound): error when creating "STDIN": admission webhook "vscaledobject.kb.io" denied the request: Deployment.apps "some-deployment" not found

Expected Behavior

I would expect, when in dry-run mode, the admission webhook to succeed without requiring the deployment to exist, meaning the webhook wouldn't check by a side effect during dry-run mode.

Actual Behavior

The admission webhook is failing due side effects when in dry-run mode, side effects should only be checked when the request is not in dry-run mode.

Steps to Reproduce the Problem

1. Create a helm chart with a valid ScaledObject using a Deployment or StatefulSet as a target
2. Define a cpu or memory trigger to the ScaledObject
3. Execute helm -n default template release-name some-chart-name | kubectl -n default apply -f - --dry-run=server

Logs from KEDA operator

example

KEDA Version

2.12.1

Kubernetes Version

1.26

Platform

Amazon Web Services

Scaler Details

CPU, memory

Anything else?

No response