如何在 Horizo​​ntal Pod Autoscaling 中使用带有特定过滤器的自定义指标

Question

我正在尝试根据自定义指标 nginx_ingress_controller_nginx_process_connections_total 为 Ingress Controller 设置 HPA。

但是在从 localhost:10254/metrics 获取指标时，我可以看到三个带有过滤器的指标，如下所示：

# HELP nginx_ingress_controller_nginx_process_connections_total total number of connections with state {active, accepted, handled}
# TYPE nginx_ingress_controller_nginx_process_connections_total counter
nginx_ingress_controller_nginx_process_connections_total{controller_class="nginx",controller_namespace="ingress-nginx",controller_pod="nginx-ingress-controller-7dddd-mssssf",state="accepted"} 479707
nginx_ingress_controller_nginx_process_connections_total{controller_class="nginx",controller_namespace="ingress-nginx",controller_pod="nginx-ingress-controller-7dddd-mssssf",state="active"} 3
nginx_ingress_controller_nginx_process_connections_total{controller_class="nginx",controller_namespace="ingress-nginx",controller_pod="nginx-ingress-controller-7dddd-mssssf",state="handled"} 479707

在这些指标中，我想将以下指标用于 HPA。

nginx_ingress_controller_nginx_process_connections_total{controller_class="nginx",controller_namespace="ingress-nginx",controller_pod="nginx-ingress-controller-7dddd-mssssf",state="active"}

如何使用来自这些不同值的指定指标。我的 HPA yaml 文件如下所示。

kind: HorizontalPodAutoscaler
apiVersion: autoscaling/v2beta1
metadata:
  name: ingress-hpa
spec:
  scaleTargetRef:
    kind: Deployment
    name: nginx-ingress-controller
  minReplicas: 3
  maxReplicas: 10
  metrics:
  - type: Pods
    pods:
      metricName: <I need to set the custom metric here>
      targetAverageValue: 10000

Answer 1

您可以使用 HPA 自定义指标。您需要在 POD 中公开端点以获取指标，还需要设置 Prometheus 和自定义指标 api 服务器。