对Istiod Pod生成的FailedGetResourceMetric事件进行故障排除

简介

本文档介绍如何处理istiod pod生成的FailedGetResourceMetric事件。

问题

在使用2023.01云本地部署平台(CNDP)版本部署用户微服务基础设施(SMI)集群后，我们注意到，istiod pod不断生成FailedGetResourceMetric事件：

cloud-user@ndce-smi-1:~$ kubectl get events -A -o yaml
apiVersion: v1
items:
- apiVersion: v1
  count: 467057
  eventTime: null
  firstTimestamp: "2023-12-31T03:57:42Z"
  involvedObject:
    apiVersion: autoscaling/v2beta2
    kind: HorizontalPodAutoscaler
    name: istiod
    namespace: istio-system
    resourceVersion: "4374"
    uid: c85a05a1-b9ec-4241-8347-0b3c9eec1193
  kind: Event
  lastTimestamp: "2024-03-22T14:42:49Z"
  message: 'unable to get metrics for resource cpu: unable to fetch metrics from resource
    metrics API: the server could not find the requested resource (get pods.metrics.k8s.io)'
  metadata:
    creationTimestamp: "2023-12-31T03:57:42Z"
    managedFields:
    - apiVersion: v1
      fieldsType: FieldsV1
      fieldsV1:
        f:count: {}
        f:firstTimestamp: {}
        f:involvedObject:
          f:apiVersion: {}
          f:kind: {}
          f:name: {}
          f:namespace: {}
          f:resourceVersion: {}
          f:uid: {}
        f:lastTimestamp: {}
        f:message: {}
        f:reason: {}
        f:source:
          f:component: {}
        f:type: {}
      manager: kube-controller-manager
      operation: Update
      time: "2023-12-31T03:57:42Z"
    name: istiod.17a5ce7b410bceb6
    namespace: istio-system
    resourceVersion: "55452121"
    selfLink: /api/v1/namespaces/istio-system/events/istiod.17a5ce7b410bceb6
    uid: 9616bdc6-90d1-4475-b6af-6d41e185b434
  reason: FailedGetResourceMetric
  reportingComponent: ""
  reportingInstance: ""
  source:
    component: horizontal-pod-autoscaler
  type: Warning
kind: List
metadata:
  resourceVersion: ""
  selfLink: ""

分析

此问题似乎是由于部署期间在istiod pod中启用了Horizontal Pod Autoscaler (HPA)。由于HPA需要指标API来确定用于上放/下放Pod的CPU利用率，并且指标API当前未在CNDP管理的Kubernetes群集中使用，因此这是Istiod部署中的设计缺陷。

解决方案

从2023.03 CNDP发布开始，HPA已被删除。