【2020.07.06】kube-prometheus中如何添加新scrape api？

上一篇博客中，有说到用kube-prometheus安装prometheus全家桶，用来监控，其中演示了如何通过配置文件的形式来添加，它有点不好的地方，就要修改，可能还需要重启，不是自动的，不是特别方便。


下面介绍一个它自动服务发现的方式：

prometheus-operator

自动发现scrape api主要得力于prometheus-operator这个项目，这个项目是方便管理prometheus在k8s集群中的自定义资源、配置、监控对象等，十分的方便。

image

其中，自定义资源ServiceMonitor它就能自动的获取到符合要求的metrics指标

自定义ServiceMonitor实现metrics指标自动获取（场景1）

  • 此方法适用于，你暴露metrics的service的namespace在monitoring下的

  • service中，添加annotations

    1. image
    2. 上面2个key，就是约定的，必须要有的，这样才能自动探测到，注意port要和service保持 一致

  • 新增servicemonitor定义

    • image

  • 验证

    • image
    • 至此就自动发现了，成功！
  • 附上完整文件
      •   1 apiVersion: apps/v1
  2 kind: Deployment
  3 metadata:
  4   name: rabbitmq-exporter-auto
  5   namespace: monitoring
  6   labels:
  7     app: rabbitmq-exporter-auto
  8     author: jamestao
  9 spec:
 10   replicas: 1
 11   selector:
 12     matchLabels:
 13       app: rabbitmqpod-auto
 14   template:
 15     metadata:
 16       name: rabbitmq-exporter-pod-auto
 17       labels:
 18         app: rabbitmqpod-auto
 19     spec:
 20       restartPolicy: Always
 21       containers:
 22         - name: rabbitmq-exporter-container-auto
 23           image: kbudde/rabbitmq-exporter:latest
 24           resources:
 25             requests:
 26               memory: 10Mi
 27           ports:
 28             - containerPort: 9419
 29           env:
 30             - name: RABBIT_URL
 31               value: http://192.168.26.91:15672
 32             - name: RABBIT_USER
 33               value: james
 34             - name: RABBIT_PASSWORD
 35               value: "666666"
 36             - name: INCLUDE_VHOST
 37               value: ".*"
 38             - name: INCLUDE_QUEUES
 39               value: ".*"
 40 ---
 41 apiVersion: v1
 42 kind: Service
 43 metadata:
 44   name: svc-rabbitmq-management-auto
 45   labels:
 46     app: rabbitmq-exporter-svc-auto
 47   namespace: monitoring
 48   annotations:
 49     prometheus.io/port: "39418"
 50     prometheus.io/scrape: "true"
 51 spec:
 52   selector:
 53     app: rabbitmqpod-auto
 54   ports:
 55     - protocol: TCP
 56       port: 39418
 57       targetPort: 9419
 58       name: metrics
 59 ---
 60 apiVersion: monitoring.coreos.com/v1
 61 kind: ServiceMonitor
 62 metadata:
 63   name: taozhiqiang
 64   namespace: monitoring
 65 spec:
 66   endpoints:
 67   - interval: 15s
 68     port: metrics
 69   namespaceSelector:
 70     matchNames:
 71     - monitoring
 72   selector:
 73     matchLabels:
 74       app: rabbitmq-exporter-svc-auto
      rabbitmq_exporter_auto.yaml

自定义ServiceMonitor实现metrics指标自动获取（场景2）

  • 此方法适用于，你暴露metrics的service的namespace不在monitoring下(kube-prometheus它默认安装在monitoring)，不在monitoring下的server想要被prometheus-operator所探测到，需要创建sa权限绑定。
  • 主要需要新增这2节：image
  • 附上完整文件：
      1 apiVersion: apps/v1
  2 kind: Deployment
  3 metadata:
  4   name: rabbitmq-exporter-auto
  5   namespace: bookstore
  6   labels:
  7     app: rabbitmq-exporter-auto
  8     author: jamestao
  9 spec:
 10   replicas: 1
 11   selector:
 12     matchLabels:
 13       app: rabbitmqpod-auto
 14   template:
 15     metadata:
 16       name: rabbitmq-exporter-pod-auto
 17       labels:
 18         app: rabbitmqpod-auto
 19     spec:
 20       restartPolicy: Always
 21       containers:
 22         - name: rabbitmq-exporter-container-auto
 23           image: kbudde/rabbitmq-exporter:latest
 24           resources:
 25             requests:
 26               memory: 10Mi
 27           ports:
 28             - containerPort: 9419
 29           env:
 30             - name: RABBIT_URL
 31               value: http://192.168.26.91:15672
 32             - name: RABBIT_USER
 33               value: james
 34             - name: RABBIT_PASSWORD
 35               value: "666666"
 36             - name: INCLUDE_VHOST
 37               value: ".*"
 38             - name: INCLUDE_QUEUES
 39               value: ".*"
 40 ---
 41 apiVersion: v1
 42 kind: Service
 43 metadata:
 44   name: svc-rabbitmq-management-auto
 45   labels:
 46     app: rabbitmq-exporter-svc-auto
 47   namespace: bookstore
 48   annotations:
 49     prometheus.io/port: "39418"
 50     prometheus.io/scrape: "true"
 51 spec:
 52   selector:
 53     app: rabbitmqpod-auto
 54   ports:
 55     - protocol: TCP
 56       port: 39418
 57       targetPort: 9419
 58       name: metrics
 59 ---
 60 apiVersion: monitoring.coreos.com/v1
 61 kind: ServiceMonitor
 62 metadata:
 63   name: taozhiqiang-auto
 64   namespace: monitoring
 65 spec:
 66   endpoints:
 67   - interval: 15s
 68     port: metrics
 69   namespaceSelector:
 70     matchNames:
 71     - bookstore
 72   selector:
 73     matchLabels:
 74       app: rabbitmq-exporter-svc-auto
 75 ---
 76 apiVersion: rbac.authorization.k8s.io/v1
 77 kind: RoleList
 78 items:
 79 - apiVersion: rbac.authorization.k8s.io/v1
 80   kind: Role
 81   metadata:
 82     name: prometheus-k8s-rabbitmq
 83     namespace: bookstore
 84   rules:
 85   - apiGroups:
 86     - ""
 87     resources:
 88     - services
 89     - endpoints
 90     - pods
 91     verbs:
 92     - get
 93     - list
 94     - watch
 95 ---
 96 apiVersion: rbac.authorization.k8s.io/v1
 97 kind: RoleBindingList
 98 items:
 99 - apiVersion: rbac.authorization.k8s.io/v1
100   kind: RoleBinding
101   metadata:
102     name: prometheus-k8s-rabbitmq
103     namespace: bookstore
104   roleRef:
105     apiGroup: rbac.authorization.k8s.io
106     kind: Role
107     name: prometheus-k8s-rabbitmq
108   subjects:
109   - kind: ServiceAccount
110     name: prometheus-k8s
111     namespace: monitoring
    rabbitmq_exporter_auto_out_monitoring.yaml

2种场景请仔细阅读文件信息，对比下差异，就好理解了。

ServiceMonitor机制图

posted @ 2020-07-06 19:25  James陶  阅读(814)  评论(0)  编辑  收藏  举报