stderrthreshold not honored by operator-metrics-apiserver

[harelhadar]

After upgrading KEDA to version 2.15.1 , the operator-metrics-apiserver is displaying unexpected logging behavior, including excessive log output at ERROR level.

Expected Behavior

The operator-metrics-apiserver should log only at the ERROR level by default.
The allowed values for the stderrthreshold configuration are: 'DEBUG','INFO','WARN','ERROR','ALERT','EMERG'

Actual Behavior

The operator-metrics-apiserver outputs logs of all levels by default.
The operator crashes when attempting to set the stderrthreshold to any value other than ERROR .

Steps to Reproduce the Problem

1. Upgrade KEDA to version 2.15.1 .
2. Observe the logs of operator-metrics-apiserver.
3. Change the stderrthreshold to any value other than ERROR

Specifications

• KEDA Version: 2.15.1
• Platform & Version: K8S-AKS/EKS/GKE
• Kubernetes Version: v1.29.8
• Scaler(s): Prometheus, New-Relic, RabbitMQ

Attached files

crash.txt
censored logs with traces.txt

kedacore/keda#5139

[JorTurFer]

This issue was solved in the upstream so we should bump the dep to pull the fix, are you willing to open a PR?

[harelhadar]

Hi @JorTurFer

Thank you for looking into this! Just to clarify, would the PR need to update k8s.io/klog/v2 from version v2.120.1 to v2.130.1 to pull in the fix for the stderrthreshold issue?

Thanks!