KEDA Unable to Retrieve correct Kafka Metrics from ScaledObject on GKE

[converge]

Report

KEDA is unable to retrieve metrics correctly from a ScaledObject/ScaleTarget using a Kafka trigger when deployed to a GKE cluster (It works locally)

Expected Behavior

When HPA calculates the current metric value, it should not return -9223372036854775808m, but a valid Kafka lag.

Actual Behavior

When the Kafka ScaledObject is deployed to GKE:

• The HorizontalPodAutoscaler reports a negative metric value (9223372036854775808m)

Steps to Reproduce the Problem

1. deploy Keda in a GKE cluster
2. create a ScaledObject with Kafka Source as Scaled Target
3. kubectl get hpa -A (will show the metric with the negative value)

Logs from KEDA operator

There is no error or warning in the Keda operator.

KEDA Version

2.13.1

Kubernetes Version

1.27

Platform

Google Cloud

Scaler Details

Kafka

Anything else?

No response

[SpiritZhou]

Could you change the log-level to debug in operator and send the operator logs?

[dttung2905]

In addition to the log, could you provide the ScaledObject yaml config too ?