K8s HPA设置minReplicas=3仍报TooFewReplicas问题排查

云原生阿岚

2026-9-3

问题背景

创建如下HPA配置，预期关联的Deployment启动时默认副本数为3：

apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
  name: backend-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: backend

  minReplicas: 3
  maxReplicas: 20

  metrics:
    - type: Resource
      resource:
        name: memory
        target:
          type: AverageValue
          averageValue: 500Mi
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 70

  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
        - type: Pods
          value: 1
          periodSeconds: 100

    scaleUp:
      stabilizationWindowSeconds: 60
      policies:
        - type: Pods
          value: 1
          periodSeconds: 30
        - type: Percent
          value: 10
          periodSeconds: 60
      selectPolicy: Max

HPA持续返回如下状态，无法理解提示原因：

ScalingLimited True TooFewReplicas the desired replica count is less than the minimum replica count

相关kubectl查询结果如下：

➜ kg hpa
NAME                  REFERENCE                    TARGETS                       MINPODS   MAXPODS   REPLICAS   AGE
prod-backend-hpa-v1   Deployment/prod-backend-v1   145881770666m/500Mi, 2%/70%   3         20        3          7m16s


➜ kd hpa
Name:                                                  prod-backend-hpa-v1
Namespace:                                             prod
Labels:                                                argocd.argoproj.io/instance=backend-prod
Annotations:                                           <none>
CreationTimestamp:                                     Thu, 02 Jun 2022 19:34:30 -0500
Reference:                                             Deployment/prod-backend-v1
Metrics:                                               ( current / target )
  resource memory on pods:                             145596416 / 500Mi
  resource cpu on pods  (as a percentage of request):  1% (3m) / 70%
Min replicas:                                          3
Max replicas:                                          20
Behavior:
  Scale Up:
    Stabilization Window: 60 seconds
    Select Policy: Max
    Policies:
      - Type: Pods     Value: 1   Period: 30 seconds
      - Type: Percent  Value: 10  Period: 60 seconds
  Scale Down:
    Stabilization Window: 300 seconds
    Select Policy: Max
    Policies:
      - Type: Pods  Value: 1  Period: 100 seconds
Deployment pods:    3 current / 3 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 memory resource
  ScalingLimited  True    TooFewReplicas    the desired replica count is less than the minimum replica count
Events:
  Type    Reason             Age    From                       Message
  ----    ------             ----   ----                       -------
  Normal  SuccessfulRescale  9m22s  horizontal-pod-autoscaler  New size: 3; reason: Current number of replicas below Spec.MinReplicas



➜ kgpo  
NAME                              READY   STATUS    RESTARTS   AGE
prod-backend-v1-8dd687999-54mzp   1/1     Running   0          58m
prod-backend-v1-8dd687999-nn7c2   1/1     Running   0          2d17h
prod-backend-v1-8dd687999-rcxsw   1/1     Running   0          2d17h

➜ kg rs 
NAME                         DESIRED   CURRENT   READY   AGE
prod-backend-v1-566b9c8856   0         0         0       2d19h
prod-backend-v1-578d699c45   0         0         0       2d19h
prod-backend-v1-64859b74c9   0         0         0       2d18h
prod-backend-v1-6498b4b45c   0         0         0       2d19h
prod-backend-v1-656cccdc4b   0         0         0       2d19h
prod-backend-v1-66cc5cf44    0         0         0       2d19h
prod-backend-v1-698c7ddc7d   0         0         0       2d19h
prod-backend-v1-6bdbc77f5d   0         0         0       2d19h
prod-backend-v1-7486c95664   0         0         0       2d19h
prod-backend-v1-774cdbdcdc   0         0         0       2d19h
prod-backend-v1-8dd687999    3         3         3       2d17

原因说明

这个ScalingLimited不是配置错误，是HPA的正常状态提示，触发原因是内存指标配置和实际预期不匹配：

给内存指标配置的target类型是AverageValue，目标值是单Pod平均内存500Mi，HPA会按期望副本数 = 所有Pod内存占用总和 / 单Pod目标内存值计算需要的副本数。
从查询结果可以看到，当前单Pod实际内存占用只有约139Mi（145596416字节），3个Pod总内存约417Mi，算出来期望副本数不到1，远小于设置的minReplicas=3。
HPA受最小副本数限制不会继续缩容到计算值，就会返回TooFewReplicas的提示，意思是当前计算出的期望副本数小于配置的最小副本数，伸缩被阈值限制。
看到的内存指标145881770666m/500Mi是单位换算展示问题，AverageValue模式下内存以字节的千分之一（m）为单位展示，139Mi换算后正好是这个数值，远低于500Mi的目标。

处理方案

如果预期是按Pod内存request的使用率伸缩，把内存指标的target类型从AverageValue改成Utilization，配置averageUtilization字段填写目标使用率百分比即可，比如填70代表平均内存达到request的70%时触发扩容。
如果确实需要按内存绝对值做伸缩阈值，这个提示不需要额外处理，等业务负载升高，HPA计算出的期望副本数大于等于3时，这个状态会自动消失。

内容的提问来源于stack exchange，提问作者DmitrySemenov