I have a cluster in the Google Kubernetes Engine and want to make one of the deployments auto scalable by memory.

After doing a deployment, I check the horizontal scalation with the following command

kubectl describe hpa -n my-namespace

With this result:

Name:                                                     myapi-api-deployment
Namespace:                                                my-namespace
Labels:                                                   <none>
Annotations:                                              <none>
CreationTimestamp:                                        Tue, 15 Feb 2022 12:21:44  0100
Reference:                                                Deployment/myapi-api-deployment
Metrics:                                                  ( current / target )
  resource memory on pods  (as a percentage of request):  <unknown> / 50%
Min replicas:                                             1
Max replicas:                                             5
Deployment pods:                                          1 current / 1 desired
Conditions:
  Type            Status  Reason                   Message
  ----            ------  ------                   -------
  AbleToScale     True    ReadyForNewScale         recommended size matches current size
  ScalingActive   False   FailedGetResourceMetric  the HPA was unable to compute the replica count: failed to get memory utilization: missing request for memory
  ScalingLimited  False   DesiredWithinRange       the desired count is within the acceptable range
Events:
  Type     Reason                   Age                    From                       Message
  ----     ------                   ----                   ----                       -------
  Warning  FailedGetResourceMetric  2m22s (x314 over 88m)  horizontal-pod-autoscaler  failed to get memory utilization: missing request for memory

When I use the kubectl top command I can see the memory and cpu usage. Here is my deployment including the autoscale:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-api-deployment
  namespace: my-namespace
  annotations:
    reloader.stakater.com/auto: "true"
spec:
  replicas: 1
  selector:
    matchLabels:
      app: my-api
      version: v1
  template:
    metadata:
      labels:
        app: my-api
        version: v1
      annotations:
        sidecar.istio.io/rewriteAppHTTPProbers: "true"
    spec:
      serviceAccountName: my-api-sa
      containers:
        - name: esp
          image: gcr.io/endpoints-release/endpoints-runtime:2
          imagePullPolicy: Always
          args: [
              "--listener_port=9000",
              "--backend=127.0.0.1:8080",              
              "--service=myproject.company.ai"            
          ]
          ports:
            - containerPort: 9000
        - name: my-api
          image: gcr.io/myproject/my-api:24
          ports:
            - containerPort: 8080
          livenessProbe:
            httpGet:
              path: "/healthcheck"
              port: 8080
            initialDelaySeconds: 30
            periodSeconds: 10
          readinessProbe:
            httpGet:
              path: "/healthcheck"
              port: 8080
            initialDelaySeconds: 30
            periodSeconds: 10         
          resources:
            limits:
              cpu: 500m
              memory: 2048Mi
            requests:
              cpu: 300m
              memory: 1024Mi
---
apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
  name: my-api-deployment
  namespace: my-namespace
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-api-deployment
  minReplicas: 1
  maxReplicas: 5
  metrics:
    - type: Resource
      resource:
        name: memory
        target:
          type: "Utilization"
          averageUtilization: 50
---

Using the autoscaling/v2beta2 recommended by the GKE documentation

CodePudding user response：

When using the HPA with memory or CPU, you need to set resource requests for whichever metric(s) your HPA is using. See How does a HorizontalPodAutoscaler work, specifically

For per-pod resource metrics (like CPU), the controller fetches the metrics from the resource metrics API for each Pod targeted by the HorizontalPodAutoscaler. Then, if a target utilization value is set, the controller calculates the utilization value as a percentage of the equivalent resource request on the containers in each Pod. If a target raw value is set, the raw metric values are used directly.

Your HPA is set to match the my-api-deployment which has two containers. You have resource requests set for my-api but not for esp. So you just need to add a memory resource request to esp.