HPA not scaling pods when using AverageValue instead of Utilization-CodePudding

I am reading through the HPA walkthrough available on the kubernetes documentation here. I am unable to get the HPA to scale the deployment when using the AverageValue instead of Utilization.

I am using a 1.25 minikube cluster and have metrics server deployment and patched.

kubectl patch deployment metrics-server -n kube-system --type 'json' -p '[{"op": "add", "path": "/spec/template/spec/containers/0/args/-", "value": "--kubelet-insecure-tls"}]'

I am able to successfully execute kubectl top pods/nodes commands successfully which to me means that the metrics server is working fine.

Gauravs-MBP:K8s alieninvader$ kubectl top pods
NAME                          CPU(cores)   MEMORY(bytes)   
load-generator                6m           0Mi             
php-apache-5b56f9df94-r74qk   501m         11Mi            
Gauravs-MBP:K8s alieninvader$ kubectl top nodes
NAME       CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%   
minikube   719m         11%    834Mi           10%

When I use the Utilization instead of AverageValue, the HPA is able to scale the deployment. I have used the same php-apache deployment and used this as my manifest to create the HPA.

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: php-apache
  namespace: default
spec:
  maxReplicas: 10
  metrics:
  - resource:
      name: cpu
      target:
        averageValue: 50
        type: AverageValue
    type: Resource
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: php-apache

When i look at the events for the HPA, i see that HPA still calculates this to be within recommended range.

 kubectl describe hpa php-apache
Warning: autoscaling/v2beta2 HorizontalPodAutoscaler is deprecated in v1.23 , unavailable in v1.26 ; use autoscaling/v2 HorizontalPodAutoscaler
Name:                    php-apache
Namespace:               default
Labels:                  <none>
Annotations:             <none>
CreationTimestamp:       Tue, 20 Dec 2022 13:43:19 -0500
Reference:               Deployment/php-apache
Metrics:                 ( current / target )
  resource cpu on pods:  501m / 50
Min replicas:            1
Max replicas:            10
Deployment pods:         1 current / 1 desired
Conditions:
  Type            Status  Reason              Message
  ----            ------  ------              -------
  AbleToScale     True    ReadyForNewScale    recommended size matches current size
  ScalingActive   True    ValidMetricFound    the HPA was able to successfully calculate a replica count from cpu resource
  ScalingLimited  False   DesiredWithinRange  the desired count is within the acceptable range
Events:           <none>



Gauravs-MBP:K8s alieninvader$ kubectl get hpa -w
NAME         REFERENCE               TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
php-apache   Deployment/php-apache   1m/50     1         10        1          37s
php-apache   Deployment/php-apache   189m/50   1         10        1          45s
php-apache   Deployment/php-apache   501m/50   1         10        1          60s
php-apache   Deployment/php-apache   502m/50   1         10        1          90s
php-apache   Deployment/php-apache   500m/50   1         10        1          105s
php-apache   Deployment/php-apache   501m/50   1         10        1          2m
php-apache   Deployment/php-apache   500m/50   1         10        1          2m15s
php-apache   Deployment/php-apache   501m/50   1         10        1          2m30s
php-apache   Deployment/php-apache   499m/50   1         10        1          2m45s
php-apache   Deployment/php-apache   502m/50   1         10        1          3m
php-apache   Deployment/php-apache   501m/50   1         10        1          3m15s
php-apache   Deployment/php-apache   500m/50   1         10        1          3m30s
php-apache   Deployment/php-apache   501m/50   1         10        1          3m45s
php-apache   Deployment/php-apache   500m/50   1         10        1          4m
php-apache   Deployment/php-apache   502m/50   1         10        1          4m15s
php-apache   Deployment/php-apache   500m/50   1         10        1          4m30s
php-apache   Deployment/php-apache   501m/50   1         10        1          4m45s
php-apache   Deployment/php-apache   500m/50   1         10        1          5m
php-apache   Deployment/php-apache   501m/50   1         10        1          5m15s
php-apache   Deployment/php-apache   500m/50   1         10        1          5m30s
php-apache   Deployment/php-apache   501m/50   1         10        1          5m45s
php-apache   Deployment/php-apache   500m/50   1         10        1          6m15s
php-apache   Deployment/php-apache   502m/50   1         10        1          6m30s
php-apache   Deployment/php-apache   500m/50   1         10        1          6m45s

So why when using AverageValue the HPA considers this to be in range, when clearly the CPU percentage indicated in the kubectl top o/p is way higher. What am i missing here?

CodePudding user response：

In contrast to averageUtilization, averageValue should be a literal value, not a percentage (see MetricTarget). In your setup it's set to 50 cores, which is clearly much higher then the load on the pod. Did you mean 500m?