Up-down scaling behavior configuration is ignored in Kubernetes version >= 18.0

[dkuzmianok-sc]

Scaling up-down behavior configuration #805 is ignored in Kubernetes 19.0 and 18.8 (only 2 versions I checked on).
Here is info from logs

2020-09-21T07:17:45.828Z        INFO    controllers.ScaledObject        Creating a new HPA      {"ScaledObject.Namespace": "default", "ScaledObject.Name": "kafka-receiver-scaledobject", "HPA.Namespace": "default", "HPA.Name": "keda-hpa-kafka-receiver-scaledobject"}
2020-09-21T07:17:45.968Z        INFO    controllers.ScaledObject        Warning: Ignoring scaledObject.spec.behavior, it is only supported on kubernetes version >= 1.18 {"ScaledObject.Namespace": "default", "ScaledObject.Name": "kafka-receiver-scaledobject", "kubernetes.version": "1.19"}
2020-09-21T07:17:45.968Z        INFO    controllers.ScaledObject        Initializing Scaling logic according to ScaledObject Specification      {"ScaledObject.Namespace": "default", "ScaledObject.Name": "kafka-receiver-scaledobject"}

Expected Behavior

Scaling behavior should be functioning

Actual Behavior

It's not. Logs confirm that.

Steps to Reproduce the Problem

1. Install keda, create ScaledObject with scaledDown, scaleUp advanced config.

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: kafka-receiver-scaledobject
spec:
  scaleTargetRef:
    name: test
  pollingInterval: 5
  cooldownPeriod:  5
  minReplicaCount: 1
  maxReplicaCount: 4
  advanced:                                          # Optional. Section to specify advanced options
    horizontalPodAutoscalerConfig:                   # Optional. Section to specify HPA related options
      behavior:                                      # Optional. Use to modify HPA's scaling behavior
        scaleDown:
          stabilizationWindowSeconds: 1000
          policies:
          - type: Percent
            value: 1
            periodSeconds: 1000
          - type: Pods
            value: 1
            periodSeconds: 1000
          selectPolicy: Disabled
        scaleUp:
          stabilizationWindowSeconds: 1000
          policies:
          - type: Pods
            value: 1
            periodSeconds: 1000    
          policies:
          - type: Percent
            value: 1
            periodSeconds: 1000   
          selectPolicy: Disabled
  triggers:
  - type: prometheus
    metadata:
      serverAddress: http://prometheus.default:9090
      metricName: sender-metric-test
      threshold: '10'
      query: sum(irate(spring_integration_send_seconds_count{application_name="kafka-receiver"}[1m])) * 10

2. Test it, see that it's not functioning, see logs.

Specifications

• KEDA Version: 2.0.0 beta
• Platform & Version: tested in Minikube
• Kubernetes Version: 19.0
• Scaler(s): Prometheus

[zroubalik]

@dkuzmianok-sc Hm, it must be something with Minikube version string, would you mind sharing the beginning of the Operator log? There should be a log message like this:
{"level":"info","ts":1600678755.6548228,"logger":"controllers.ScaledObject","msg":"Running on Kubernetes 1.18+"}

[zroubalik]

@dkuzmianok-sc there was a small bug in the version resolving, failing on some corner cases. Thanks for finding this! Will send a PR in a moment.

[dkuzmianok-sc]

@zroubalik

2020-09-21T07:12:49.093Z INFO controller-runtime.metrics   metrics server is starting to listen    {"addr": ":8080"}
2020-09-21T07:12:49.095Z INFO controllers.ScaledObject     Running on Kubernetes 1.19
2020-09-21T07:12:49.095Z INFO setup   Starting manager
2020-09-21T07:12:49.095Z INFO setup   KEDA Version: v2
2020-09-21T07:12:49.095Z INFO setup   KEDA Commit: 717aab81817f0d9b6d02b0e6d1499367c751faf2
2020-09-21T07:12:49.095Z INFO setup   Go Version: go1.15.1
2020-09-21T07:12:49.095Z INFO  setup   Go OS/Arch: linux/amd64

Here are logs if still needed.
Thank you for quick response.