我如何在普罗米修斯中可视化弹性搜索指标？，两者都安装在 gke 集群中

Posted 2023-02-15

【中文标题】我如何在普罗米修斯中可视化弹性搜索指标？，两者都安装在 gke 集群中

【英文标题】：How I can visualize elasticsearch metrics in prometheus?, both installed in a gke cluster

【发布时间】：2019-12-13 00:42:05

【问题描述】：

我有一个安装了这个 elasticseach 日志记录解决方案的 GKE 集群

https://console.cloud.google.com/marketplace/details/google/elastic-gke-logging

和由helm安装在同一个集群内的prometheus-operator。

我想配置一个 grafana 仪表板来可视化我的 elasticsearch 指标。

我读到 gke 的弹性应用程序安装了 elastic_exporter...https://github.com/GoogleCloudPlatform/click-to-deploy/blob/master/k8s/elastic-gke-logging/README.md

但是，如果我转到我的 Prometheus 面板，我看不到任何有关弹性搜索的指标。我尝试安装另一个 elastic_exporter，但没有。

我错过了什么？我忘记了什么？是否需要配置 prometheus 从 elastic_exporter 读取？

我在对 elastic_exporter 进行端口转发时看到指标，但在 prometheus 面板中看不到指标。

 # HELP elasticsearch_breakers_estimated_size_bytes Estimated size in bytes of breaker
# TYPE elasticsearch_breakers_estimated_size_bytes gauge
elasticsearch_breakers_estimated_size_bytesbreaker="accounting",cluster="elastic-gke-logging-1-cluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="10.50.2.54",name="elastic-gke-logging-1-elasticsearch-0" 4.6637464e+07
elasticsearch_breakers_estimated_size_bytesbreaker="fielddata",cluster="elastic-gke-logging-1-cluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="10.50.2.54",name="elastic-gke-logging-1-elasticsearch-0" 0
elasticsearch_breakers_estimated_size_bytesbreaker="in_flight_requests",cluster="elastic-gke-logging-1-cluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="10.50.2.54",name="elastic-gke-logging-1-elasticsearch-0" 0
elasticsearch_breakers_estimated_size_bytesbreaker="parent",cluster="elastic-gke-logging-1-cluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="10.50.2.54",name="elastic-gke-logging-1-elasticsearch-0" 4.6637464e+07
elasticsearch_breakers_estimated_size_bytesbreaker="request",cluster="elastic-gke-logging-1-cluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="10.50.2.54",name="elastic-gke-logging-1-elasticsearch-0" 0
# HELP elasticsearch_breakers_limit_size_bytes Limit size in bytes for breaker
# TYPE elasticsearch_breakers_limit_size_bytes gauge

谢谢

【参考方案1】：

您可能缺少ServiceMonitor，这应该可以：

k apply -f -<<EOF
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  annotations:
  labels:
    release: prom
  name: elasticsearch
spec:
  endpoints:
  - port: metrics
  selector:
    matchLabels:
      app: es-exporter
EOF

您的 elasticsearch 服务必须定义 metrics 并具有标签 app: es-exporter，类似于：

apiVersion: v1
kind: Service
metadata:

  labels:
    app: es-exporter
    component: elasticsearch
  name: elasticsearch
spec:
  ports:
  - name: transport
    port: 9200
    protocol: TCP
    targetPort: 9200
  - name: metrics
    port: 9108
    protocol: TCP
    targetPort: 9108
  selector:
    component: elasticsearch
  type: ClusterIP

之后，您应该在 Prometheus 中找到指标，以确认您始终可以在 Prometheus 中使用Status -> Targets 选项卡。