4
votes

I am trying to setup HPA for my statefulset(for elasticsearch) in kubernetes environment. I am planning to scale the statefulset using the cpu utilization. I have created the metric server from https://github.com/stefanprodan/k8s-prom-hpa/tree/master/metrics-server.

and my HPA yaml for statefulset is as folows:

apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
  name: dz-es-cluster
spec:
  scaleTargetRef:
    apiVersion: extensions/v1beta1
    kind: StatefulSet
    name: dz-es-cluster
  minReplicas: 2
  maxReplicas: 3
  metrics:
  - type: Resource
    resource:
      name: cpu
      targetAverageUtilization: 80

But getting output in hpa as follows:

Conditions:
  Type         Status  Reason          Message
  ----         ------  ------          -------
  AbleToScale  False   FailedGetScale  the HPA controller was unable to get the target's current scale: the server could not find the requested resource
Events:
  Type     Reason          Age                From                       Message
  ----     ------          ----               ----                       -------
  Warning  FailedGetScale  1m (x71 over 36m)  horizontal-pod-autoscaler  the server could not find the requested resource

someone please help me..

1
My kubernetes version is "v1.8.15"manu thankachan
Could you please share output of kubectl top nodes?Prafull Ladha
@PrafullLadha, these this the output. root@ip:~# kubectl top nodes NAME CPU(cores) CPU% MEMORY(bytes) MEMORY% ip-***-20-32-***.ec2.internal 1454m 9% 18392Mi 28% ip-***-20-34-***.ec2.internal 2130m 13% 31761Mi 49%manu thankachan
Your metrics server is running fine. Did you specify the resources section in your statefulset. That is required section for HPA to work. kubernetes.io/docs/concepts/configuration/…Prafull Ladha
There is an entry for resources in my statefulset yaml and CPU limit is set to 1 and request to 0.25. Yes, my metrics server is running fine. I have tested its working by autoscaling one of my Deployment. Issue is only with statefulset.manu thankachan

1 Answers

16
votes

The support for autoscaling the statefulsets using HPA is added in kubernetes 1.9, so your version doesn't has support for it. After kubernetes 1.9, you can autoscale your statefulsets using:

apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  name: YOUR_HPA_NAME
spec:
  maxReplicas: 3
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: StatefulSet
    name: YOUR_STATEFUL_SET_NAME
  targetCPUUtilizationPercentage: 80

Please refer the following link for more information:

https://github.com/kubernetes/kubernetes/issues/44033