HPA not scaling down when target is met

[developerxnz]

Describe the bug
The HPA isn't removing replicas once the target value has been met.

To Reproduce
Steps to reproduce
Follow the walk through from https://github.com/Azure/azure-k8s-metrics-adapter/blob/master/samples/request-per-second/readme.md

Run kubectl get hpa {hpaname} -w
to watch the state change

Once HEY has finished processing and the cool down period has passed run
kubectl get hpa {hpaname} -w
you will notice the target drops to 0/{targetvalue} but the number of replicas doesn't reduce.

Expected behavior
The HPA should scale down to the min replicas

Kubernetes version (kubectl version):

• [* ] Running on AKS
  Client Version: version.Info{Major:"1", Minor:"10", GitVersion:"v1.10.3", GitCommit:"2bba0127d85d5a46ab4b778548be28623b32d0b0", GitTreeState:"clean", BuildDate:"2018-05-21T09:17:39Z", GoVersion:"go1.9.3", Compiler:"gc", Platform:"windows/amd64"}
  Server Version: version.Info{Major:"1", Minor:"11", GitVersion:"v1.11.3", GitCommit:"a4529464e4629c21224b3d52edfe0ea91b072862", GitTreeState:"clean", BuildDate:"2018-09-09T17:53:03Z", GoVersion:"go1.10.3", Compiler:"gc", Platform:"linux/amd64"}

Logs (kubectl logs <metric adapter pod id>)
I1031 21:02:03.132657 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/customMetrics/queuelengthmonitoredqueue?interval=PT30S&timespan=PT5M
I1031 21:02:03.285583 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: performanceCounters-requestsPerSecond, selectors: app=rps-sample
I1031 21:02:03.285628 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/performanceCounters/requestsPerSecond?interval=PT30S&timespan=PT5M
I1031 21:03:03.167579 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: customMetrics-queuelengthmonitoredqueue, selectors: app=queue-sample
I1031 21:03:03.167676 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/customMetrics/queuelengthmonitoredqueue?interval=PT30S&timespan=PT5M
I1031 21:03:03.319472 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: performanceCounters-requestsPerSecond, selectors: app=rps-sample
I1031 21:03:03.319516 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/performanceCounters/requestsPerSecond?interval=PT30S&timespan=PT5M
I1031 21:03:33.171531 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: customMetrics-queuelengthmonitoredqueue, selectors: app=queue-sample
I1031 21:03:33.171576 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/customMetrics/queuelengthmonitoredqueue?interval=PT30S&timespan=PT5M
I1031 21:03:33.280357 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: performanceCounters-requestsPerSecond, selectors: app=rps-sample
I1031 21:03:33.280397 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/performanceCounters/requestsPerSecond?interval=PT30S&timespan=PT5M
I1031 21:04:03.160889 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: performanceCounters-requestsPerSecond, selectors: app=rps-sample
I1031 21:04:03.160929 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/performanceCounters/requestsPerSecond?interval=PT30S&timespan=PT5M
I1031 21:04:03.264824 1 provider.go:65] Received request for custom metric: groupresource: pods, namespace: default, metric name: customMetrics-queuelengthmonitoredqueue, selectors: app=queue-sample
I1031 21:04:03.264864 1 aiapiclient.go:52] request to: https://api.applicationinsights.io/v1/apps/{ai-appid}/metrics/customMetrics/queuelengthmonitoredqueue?interval=PT30S&timespan=PT5M

[developerxnz]

I have tested running the following HPA and this has scaled back down as expected once the target was reached.

apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
  name: rps-sample
spec:
  scaleTargetRef:
    apiVersion: extensions/v1beta1
    kind: Deployment
    name: rps-sample
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      targetAverageUtilization: 20

[jsturtevant]

Thanks for the report. This is a duplicate of #34 and I have been able to get the appropriate behavior in this PR #36.

I am going to close this issue and track from #34