2

上文讲到podMonitor是pod监控对象的抽象,本文就以calico为例,分析如何使用podMonitor对象监控calico。

calico中核心的组件是Felix,它负责设置路由表和ACL规则,同时还负责提供网络健康状况的数据;这些数据会被写入etcd。
由此可见,监控calico的核心便是监控felix,felix相当于calico的大脑。

1. calico的部署方式

calico-node在集群中以daemonset部署,部署了3个Pod:

# kubectl get all -A|grep calico-node
kube-system     pod/calico-node-j5kw9                          1/1     Running   6          195d
kube-system     pod/calico-node-njz9m                          1/1     Running   4          195d
kube-system     pod/calico-node-tmg9v                          1/1     Running   4          195d
kube-system     daemonset.apps/calico-node                3         3         3       3            3           <none>                   217d

2. 打开calico的metrics监听

calico在启动时,默认没有打开Metrics监听:

# kubectl edit ds calico-node -n kube-system

# 修改container name=calico-node的配置;
# 将其env:FELIX_PROMETHEUSMETRICSENABLED修改为“True”
    spec:
      containers:
      - env:
        ......
        - name: FELIX_PROMETHEUSMETRICSENABLED
          value: "True"
        - name: FELIX_PROMETHEUSMETRICSPORT
          value: "9091"

将FELIX_PROMETHEUSMETRICSENABLED=True,其监听端口为9091:

# curl http://localhost:9091/metrics
# HELP felix_active_local_endpoints Number of active endpoints on this host.
# TYPE felix_active_local_endpoints gauge
......

3. 创建podMonitor

首先,修改dameonset,为container定义container port;
containerPort=9091,name=http-metrics:

# kubectl edit ds calico-node -n kube-system

# 修改container name: calico-node的配置
# 增加ports声明
    spec:
      containers:
        ......
        ports:
        - containerPort: 9091
          name: http-metrics
          protocol: TCP

再创建podMonitor对象:

# cat prometheus-podMonitorCalico.yaml

apiVersion: monitoring.coreos.com/v1
kind: PodMonitor
metadata:
  labels:
    k8s-app: calico-node
  name: calico-node
  namespace: monitoring
spec:
  podMetricsEndpoints:
  - interval: 15s
    path: /metrics
    port: http-metrics
  namespaceSelector:
    matchNames:
    - kube-system
  selector:
    matchLabels:
      k8s-app: calico-node

其中定义了pod的筛选条件:

  • Port名称=http-metrics;
  • k8s-app=calico-node;
  • namespace=kube-system;

4. prometheus dashboard查看calico监控任务

参考:

  1. https://fuckcloudnative.io/po...

a朋
63 声望39 粉丝