Unable to horizonatally auto scale Strimzi's KafkaConnect worker pods using keda

Context: I am using managed kafka i.e. AWS MSK cluster and running Strimzi KafkaConnect worker pods in aws eks cluster. I have removed replicas field from KafkaConnect manifest and it spins up 3 KafkaConnect pods by default

I have deployed 6 mssql debezium connectors on kafka connect cluster and they are running fine.

I created a scaled object

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: kafkaconnect-autoscaler
  namespace: devtest
spec:
  scaleTargetRef:
    apiVersion: kafka.strimzi.io/v1beta2
    kind: KafkaConnect
    name: kafka-connect-cluster
  minReplicaCount: 1
  maxReplicaCount: 5
  pollingInterval: 10
  cooldownPeriod: 60
  triggers:
    - type: cpu
      metadata:
        type: Utilization
        value: "75"

In ideal state, CPU usage is just 2% (as per metrics-server API), so I am assuming it should scale down KafkaConnect pods from 3 to 1 but it didnt.

I described scaledObject, it says

Status:
  Conditions:
    Message:  ScaledObject doesn't have correct scaleTargetRef specification
    Reason:   ScaledObjectCheckFailed
    Status:   False
    Type:     Ready
    Message:  ScaledObject check failed
    Reason:   UnknownState
    Status:   Unknown
    Type:     Active
    Status:   Unknown
    Type:     Fallback
    Status:   Unknown
    Type:     Paused
Events:
  Type     Reason                   Age                 From           Message
  ----     ------                   ----                ----           -------
  Warning  ScaledObjectCheckFailed  13s (x12 over 23s)  keda-operator  ScaledObject doesn't have correct scaleTargetRef specification
  Warning  ScaledObjectCheckFailed  2s (x13 over 23s)   keda-operator  Target resource doesn't expose /scale subresource

I saw at many posts that /scale subresource is exposed for KafkaConnect custom resource but dont know why its not working for me.

I checked that scaled subresource is present here

kubectl get crd kafkaconnects.kafka.strimzi.io -o yaml

And even if I am able to scale it up, will connectors re-distribute among scaled worker nodes?

You definitely should set the .spec.replicas field to the initial number of replicas you want to use.