一、介绍


Operator是CoreOS公司开发,用于扩展kubernetes API或特定应用程序的控制器,它用来创建、配置、管理复杂的有状态应用,例如数据库,监控系统。其中Prometheus-Operator就是其中一个重要的项目。

其架构图如下: 其中核心部分是Operator,它会去创建Prometheus、ServiceMonitor、AlertManager、PrometheusRule这4个CRD对象,然后会一直监控并维护这4个对象的状态。

  • Prometheus:作为Prometheus Server的抽象
    
  • ServiceMonitor:就是exporter的各种抽象
    
  • AlertManager:作为Prometheus AlertManager的抽象
    
  • PrometheusRule:实现报警规则的文件
    

上图中的 Service 和 ServiceMonitor 都是 Kubernetes 的资源,一个 ServiceMonitor 可以通过 labelSelector 的方式去匹配一类 Service,Prometheus 也可以通过 labelSelector 去匹配多个ServiceMonitor。

二、安装


注意集群版本的坑,自己先到Github上下载对应的版本。

我们使用源码来安装,首先克隆源码到本地:

# git clone https://github.com/coreos/kube-prometheus.git

我们进入kube-prometheus/manifests/setup,就可以直接创建CRD对象:

# cd kube-prometheus/manifests/setup
# kubectl apply -f .

然后在上层目录创建资源清单:

# cd kube-prometheus/manifests
# kubectl apply -f .

可以看到创建如下的CRD对象:

# kubectl get crd | grep coreos
alertmanagers.monitoring.coreos.com     2019-12-02T03:03:37Z
podmonitors.monitoring.coreos.com       2019-12-02T03:03:37Z
prometheuses.monitoring.coreos.com      2019-12-02T03:03:37Z
prometheusrules.monitoring.coreos.com   2019-12-02T03:03:37Z
servicemonitors.monitoring.coreos.com   2019-12-02T03:03:37Z

查看创建的pod:

# kubectl get pod -n monitoring 
NAME                                  READY   STATUS    RESTARTS   AGE
alertmanager-main-0                   2/2     Running   0          2m37s
alertmanager-main-1                   2/2     Running   0          2m37s
alertmanager-main-2                   2/2     Running   0          2m37s
grafana-77978cbbdc-886cc              1/1     Running   0          2m46s
kube-state-metrics-7f6d7b46b4-vrs8t   3/3     Running   0          2m45s
node-exporter-5552n                   2/2     Running   0          2m45s
node-exporter-6snb7                   2/2     Running   0          2m45s
prometheus-adapter-68698bc948-6s5f2   1/1     Running   0          2m45s
prometheus-k8s-0                      3/3     Running   1          2m27s
prometheus-k8s-1                      3/3     Running   1          2m27s
prometheus-operator-6685db5c6-4tdhp   1/1     Running   0          2m52s

查看创建的Service:

# kubectl get svc -n monitoring 
NAME                    TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
alertmanager-main       ClusterIP   10.68.97.247    <none>        9093/TCP                     3m51s
alertmanager-operated   ClusterIP   None            <none>        9093/TCP,9094/TCP,9094/UDP   3m41s
grafana                 ClusterIP   10.68.234.173   <none>        3000/TCP                     3m50s
kube-state-metrics      ClusterIP   None            <none>        8443/TCP,9443/TCP            3m50s
node-exporter           ClusterIP   None            <none>        9100/TCP                     3m50s
prometheus-adapter      ClusterIP   10.68.109.201   <none>        443/TCP                      3m50s
prometheus-k8s          ClusterIP   10.68.9.232     <none>        9090/TCP                     3m50s
prometheus-operated     ClusterIP   None            <none>        9090/TCP                     3m31s
prometheus-operator     ClusterIP   None            <none>        8080/TCP                     3m57s

我们看到我们常用的prometheus和grafana都是clustorIP,我们要外部访问可以配置为NodePort类型或者用ingress。比如配置为ingress: prometheus-ingress.yaml

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: prometheus-ingress
  namespace: monitoring
  annotations:
    kubernetes.io/ingress.class: "traefik"
spec:
  rules:
  - host: prometheus.joker.com
    http:
      paths:
      - path:
        backend: 
          serviceName: prometheus-k8s 
          servicePort: 9090

grafana-ingress.yaml

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: grafana-ingress
  namespace: monitoring
  annotations:
    kubernetes.io/ingress.class: "traefik"
spec:
  rules:
  - host: grafana.joker.com
    http:
      paths:
      - path:
        backend: 
          serviceName: grafana
          servicePort: 3000

但是我们这里由于没有域名进行备案,我们就用NodePort类型。修改后如下:

# kubectl get svc -n monitoring 
NAME                    TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
grafana                 NodePort    10.68.234.173   <none>        3000:39807/TCP               3h1m                    3h1m
prometheus-k8s          NodePort    10.68.9.232     <none>        9090:20547/TCP               3h1m

然后就可以正常在浏览器访问了。

三、配置


> 3.1、监控集群资源

我们可以看到大部分的配置都是正常的,只有两三个没有管理到对应的监控目标,比如 kube-controller-manager 和 kube-scheduler 这两个系统组件,这就和 ServiceMonitor 的定义有关系了,我们先来查看下 kube-scheduler 组件对应的 ServiceMonitor 资源的定义:(prometheus-serviceMonitorKubeScheduler.yaml)

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  labels:
    k8s-app: kube-scheduler
  name: kube-scheduler
  namespace: monitoring
spec:
  endpoints:
  - interval: 30s # 每30s获取一次信息
    port: http-metrics  # 对应service的端口名
  jobLabel: k8s-app
  namespaceSelector: # 表示去匹配某一命名空间中的service,如果想从所有的namespace中匹配用any: true
    matchNames:
    - kube-system
  selector:  # 匹配的 Service 的labels,如果使用mathLabels,则下面的所有标签都匹配时才会匹配该service,如果使用matchExpressions,则至少匹配一个标签的service都会被选择
    matchLabels:
      k8s-app: kube-scheduler

上面是一个典型的 ServiceMonitor 资源文件的声明方式,上面我们通过selector.matchLabels在 kube-system 这个命名空间下面匹配具有k8s-app=kube-scheduler这样的 Service,但是我们系统中根本就没有对应的 Service,所以我们需要手动创建一个 Service:(prometheus-kubeSchedulerService.yaml)

apiVersion: v1
kind: Service
metadata:
  namespace: kube-system
  name: kube-scheduler
  labels:
    k8s-app: kube-scheduler
spec:
  selector:
    component: kube-scheduler
  ports:
  - name: http-metrics
    port: 10251
    targetPort: 10251
    protocol: TCP

10251是kube-scheduler组件 metrics 数据所在的端口,10252是kube-controller-manager组件的监控数据所在端口。

其中最重要的是上面 labels 和 selector 部分,labels 区域的配置必须和我们上面的 ServiceMonitor 对象中的 selector 保持一致,selector下面配置的是component=kube-scheduler,为什么会是这个 label 标签呢,我们可以去 describe 下 kube-scheduelr 这个 Pod:

$ kubectl describe pod kube-scheduler-master -n kube-system
Name:         kube-scheduler-master
Namespace:    kube-system
Node:         master/10.151.30.57
Start Time:   Sun, 05 Aug 2018 18:13:32 +0800
Labels:       component=kube-scheduler
              tier=control-plane
......

我们可以看到这个 Pod 具有component=kube-scheduler和tier=control-plane这两个标签,而前面这个标签具有更唯一的特性,所以使用前面这个标签较好,这样上面创建的 Service 就可以和我们的 Pod 进行关联了,直接创建即可:

$ kubectl create -f prometheus-kubeSchedulerService.yaml
$ kubectl get svc -n kube-system -l k8s-app=kube-scheduler
NAME             TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)     AGE
kube-scheduler   ClusterIP   10.102.119.231   <none>        10251/TCP   18m

创建完成后,隔一小会儿后去 prometheus 查看 targets 下面 kube-scheduler 的状态: promethus kube-scheduler error 我们可以看到现在已经发现了 target,但是抓取数据结果出错了,这个错误是因为我们集群是使用 kubeadm 搭建的,其中 kube-scheduler 默认是绑定在127.0.0.1上面的,而上面我们这个地方是想通过节点的 IP 去访问,所以访问被拒绝了,我们只要把 kube-scheduler 绑定的地址更改成0.0.0.0即可满足要求,由于 kube-scheduler 是以静态 Pod 的形式运行在集群中的,所以我们只需要更改静态 Pod 目录下面对应的 YAML 文件即可:

$ ls /etc/kubernetes/manifests/
etcd.yaml  kube-apiserver.yaml  kube-controller-manager.yaml  kube-scheduler.yaml

将 kube-scheduler.yaml 文件中-command的--address地址更改成0.0.0.0:

containers:
- command:
- kube-scheduler
- --leader-elect=true
- --kubeconfig=/etc/kubernetes/scheduler.conf
- --address=0.0.0.0

修改完成后我们将该文件从当前文件夹中移除,隔一会儿再移回该目录,就可以自动更新了,然后再去看 prometheus 中 kube-scheduler 这个 target 是否已经正常了: promethues-operator-kube-scheduler 大家可以按照上面的方法尝试去修复下 kube-controller-manager 组件的监控。

3.2、监控集群外资源

很多时候我们并不是把所有资源都部署在集群内的,经常有比如ectd,kube-scheduler等都部署在集群外。其监控流程和上面大致一样,唯一的区别就是在定义Service的时候,其EndPoints是需要我们自己去定义的。

3.2.1、监控kube-scheduler

(1)、定义Service和EndPoints prometheus-KubeSchedulerService.yaml

apiVersion: v1
kind: Service
metadata:
  name: kube-scheduler
  namespace: kube-system
  labels:
    k8s-app: kube-scheduler
spec: 
  type: ClusterIP
  clusterIP: None
  ports:
  - name: http-metrics
    port: 10251
    targetPort: 10251
    protocol: TCP
---
apiVersion: v1
kind: Endpoints
metadata:
  name: kube-scheduler
  namespace: kube-system
  labels:
    k8s-app: kube-scheduler
subsets:
- addresses:
  - ip: 172.16.0.33
  ports:
  - name: http-metrics
    port: 10251
    protocol: TCP

(2)、定义ServiceMonitor prometheus-serviceMonitorKubeScheduler.yaml

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: kube-scheduler
  namespace: monitoring
  labels:
    k8s-app: kube-scheduler
spec:
  endpoints:
  - interval: 30s
    port: http-metrics
  jobLabel: k8s-app
  namespaceSelector:
    matchNames:
    - kube-system
  selector:
    matchLabels:
      k8s-app: kube-scheduler

然后我们就可以看到其监控上了:

3.2.2、监控kube-controller-manager

(1)、配置Service和EndPoints, prometheus-KubeControllerManagerService.yaml

apiVersion: v1
kind: Service
metadata:
  name: kube-controller-manager
  namespace: kube-system
  labels:
    k8s-app: kube-controller-manager
spec:
  type: ClusterIP
  clusterIP: None
  ports:
  - name: http-metrics
    port: 10252
    targetPort: 10252
    protocol: TCP
---
apiVersion: v1
kind: Endpoints
metadata:
  name: kube-controller-manager
  namespace: kube-system
  labels:
    k8s-app: kube-controller-manager
subsets:
- addresses:
  - ip: 172.16.0.33
  ports:
  - name: http-metrics
    port: 10252
    protocol: TCP

(2)、配置ServiceMonitor prometheus-serviceMonitorKubeControllerManager.yaml

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  labels:
    k8s-app: kube-controller-manager
  name: kube-controller-manager
  namespace: monitoring
spec:
  endpoints:
  - interval: 30s
    metricRelabelings:
    - action: drop
      regex: etcd_(debugging|disk|request|server).*
      sourceLabels:
      - __name__
    port: http-metrics
  jobLabel: k8s-app
  namespaceSelector:
    matchNames:
    - kube-system
  selector:
    matchLabels:
      k8s-app: kube-controller-manager

3.2.3、监控etcd

很多情况下,我们的etcd都需要进行SSL认证的,所以首先需要将用到的证书保存到集群中去。 (根据自己集群证书的位置修改)

kubectl -n monitoring create secret generic etcd-certs --from-file=/etc/kubernetes/pki/etcd/healthcheck-client.crt --from-file=/etc/kubernetes/pki/etcd/healthcheck-client.key --from-file=/etc/kubernetes/pki/etcd/ca.crt

然后将上面创建的 etcd-certs 对象配置到 prometheus 资源对象中,直接更新 prometheus 资源对象即可:

#  kubectl edit prometheus k8s -n monitoring

添加如下的 secrets 属性:

nodeSelector:
  beta.kubernetes.io/os: linux
replicas: 2
secrets:
- etcd-certs

更新完成后,我们就可以在 Prometheus 的 Pod 中获取到上面创建的 etcd 证书文件了,具体的路径我们可以进入 Pod 中查看:

# kubectl exec -it prometheus-k8s-0 -n monitoring -- /bin/sh
Defaulting container name to prometheus.
Use 'kubectl describe pod/prometheus-k8s-0 -n monitoring' to see all of the containers in this pod.
/prometheus $  ls /etc/prometheus/secrets/etcd-certs/
ca.crt      healthcheck-client.crt  healthcheck-client.key
/prometheus $ 

(1)、创建ServiceMonitor prometheus-serviceMonitorEtcd.yamlns

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: k8s-etcd
  namespace: monitoring
  labels:
    k8s-app: k8s-etcd
spec:
  jobLabel: k8s-app
  endpoints:
  - port: port
    interval: 30s
    scheme: https
    tlsConfig:
      caFile: /etc/prometheus/secrets/etcd-certs/ca.crt
      certFile: /etc/prometheus/secrets/etcd-certs/healthcheck-client.crt
      keyFile: /etc/prometheus/secrets/etcd-certs/healthcheck-client.key
      insecureSkipVerify: true
  selector:
    matchLabels:
      k8s-app: k8s-etcd
  namespaceSelector:
    matchNames:
    - kube-system

上面我们在 monitoring 命名空间下面创建了名为 k8s-etcd 的 ServiceMonitor 对象,基本属性和前面章节中的一致,匹配 kube-system 这个命名空间下面的具有 k8s-app=k8s-etcd 这个 label 标签的 Service,jobLabel 表示用于检索 job 任务名称的标签,和前面不太一样的地方是 endpoints 属性的写法,配置上访问 etcd 的相关证书,endpoints 属性下面可以配置很多抓取的参数,比如 relabel、proxyUrl,tlsConfig 表示用于配置抓取监控数据端点的 tls 认证,由于证书 serverName 和 etcd 中签发的可能不匹配,所以加上了 insecureSkipVerify=true.

然后创建这个配置清单:

# kubectl apply -f prometheus-serviceMonitorEtcd.yaml

(2)、创建Service

apiVersion: v1
kind: Service
metadata:
  name: k8s-etcd
  namespace: kube-system
  labels:
    k8s-app: k8s-etcd
spec:
  type: ClusterIP
  clusterIP: None
  ports:
  - name: port
    port: 2379
    protocol: TCP

---
apiVersion: v1
kind: Endpoints
metadata:
  name: k8s-etcd
  namespace: kube-system
  labels:
    k8s-app: k8s-etcd
subsets:
- addresses:
  - ip: 172.16.0.33
  ports:
  - name: port
    port: 2379
    protocol: TCP

然后在Grafana中导入3070的面板。

3.3、配置报警规则Rule

我们创建一个 PrometheusRule 资源对象后,会自动在上面的 prometheus-k8s-rulefiles-0 目录下面生成一个对应的-.yaml文件,所以如果以后我们需要自定义一个报警选项的话,只需要定义一个 PrometheusRule 资源对象即可,但是要求这个资源对象必须得有 prometheus=k8s 和 role=alert-rules 这一对标签。 如下配置Ectd报警规则: prometheus-etcdRule.yaml

apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
  name: etcd-rules
  namespace: monitoring
  labels:
    prometheus: k8s
    role: alert-rules
spec:
  groups:
  - name: etcd
    rules:
    - alert: EtcdClusterUnavailable
      annotations:
        summary: etcd cluster small
        description: If one more etcd peer goes down the cluster will be unavailable
      expr: |
        count(up{job="etcd"} == 0) > (count(up{job="etcd"}) / 2 - 1)
      for: 3m
      labels:
        severity: critical

然后我们创建这个配置清单:

# kubectl apply -f prometheus-etcdRule.yaml
prometheusrule.monitoring.coreos.com/etcd-rules created

然后我们刷新页面,就可以看到已经生效了

3.4、配置报警

首先我们将 alertmanager-main 这个 Service 改为 NodePort 类型的 Service,修改完成后我们可以在页面上的 status 路径下面查看 AlertManager 的配置信息:

# kubectl get svc -n monitoring 
NAME                    TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
alertmanager-main       NodePort    10.68.97.247    <none>        9093:21936/TCP               5h31m

然后在浏览器查看:

这些配置信息实际上是来自于我们之前在kube-prometheus/manifests目录下面创建的 alertmanager-secret.yaml 文件:

apiVersion: v1
data:
  alertmanager.yaml: Imdsb2JhbCI6CiAgInJlc29sdmVfdGltZW91dCI6ICI1bSIKInJlY2VpdmVycyI6Ci0gIm5hbWUiOiAibnVsbCIKInJvdXRlIjoKICAiZ3JvdXBfYnkiOgogIC0gImpvYiIKICAiZ3JvdXBfaW50ZXJ2YWwiOiAiNW0iCiAgImdyb3VwX3dhaXQiOiAiMzBzIgogICJyZWNlaXZlciI6ICJudWxsIgogICJyZXBlYXRfaW50ZXJ2YWwiOiAiMTJoIgogICJyb3V0ZXMiOgogIC0gIm1hdGNoIjoKICAgICAgImFsZXJ0bmFtZSI6ICJXYXRjaGRvZyIKICAgICJyZWNlaXZlciI6ICJudWxsIg==
kind: Secret
metadata:
  name: alertmanager-main
  namespace: monitoring
type: Opaque

可以将 alertmanager.yaml 对应的 value 值做一个 base64 解码:

# echo "Imdsb2JhbCI6CiAgInJlc29sdmVfdGltZW91dCI6ICI1bSIKInJlY2VpdmVycyI6Ci0gIm5hbWUiOiAibnVsbCIKInJvdXRlIjoKICAiZ3JvdXBfYnkiOgogIC0gImpvYiIKICAiZ3JvdXBfaW50ZXJ2YWwiOiAiNW0iCiAgImdyb3VwX3dhaXQiOiAiMzBzIgogICJyZWNlaXZlciI6ICJudWxsIgogICJyZXBlYXRfaW50ZXJ2YWwiOiAiMTJoIgogICJyb3V0ZXMiOgogIC0gIm1hdGNoIjoKICAgICAgImFsZXJ0bmFtZSI6ICJXYXRjaGRvZyIKICAgICJyZWNlaXZlciI6ICJudWxsIg==" | base64 -d
"global":
  "resolve_timeout": "5m"
"receivers":
- "name": "null"
"route":
  "group_by":
  - "job"
  "group_interval": "5m"
  "group_wait": "30s"
  "receiver": "null"
  "repeat_interval": "12h"
  "routes":
  - "match":
      "alertname": "Watchdog"
    "receiver": "null"

可以看到上面的内容和我们在网页上查到的是一致的。 如果要配置报警媒介,就可以修改这个模板: alertmanager.yaml

global:
  resolve_timeout: 5m
  smtp_smarthost: 'smtp.163.com:465'
  smtp_from: 'fmbankops@163.com'
  smtp_auth_username: 'fmbankops@163.com'
  smtp_auth_password: '<邮箱密码>'
  smtp_hello: '163.com'
  smtp_require_tls: false
route:
  group_by: ['job', 'severity']
  group_wait: 30s
  group_interval: 5m
  repeat_interval: 12h
  receiver: default
  routes:
  - receiver: webhook
    match:
      alertname: CoreDNSDown
receivers:
- name: 'default'
  email_configs:
  - to: '517554016@qq.com'
    send_resolved: true
- name: 'webhook'
  webhook_configs:
  - url: 'http://dingtalk-hook.kube-ops:5000'   # 这是我们自定义的webhook
    send_resolved: true

然后我们更新secret对象:

# 先将之前的 secret 对象删除
$ kubectl delete secret alertmanager-main -n monitoring
secret "alertmanager-main" deleted
$ kubectl create secret generic alertmanager-main --from-file=alertmanager.yaml -n monitoring
secret "alertmanager-main" created

然后就会收到报警信息:

四、高级配置


4.1、自动发现规则配置

我们在实际应用中会部署非常多的service和pod,如果要一个一个手动的添加监控将会是一个非常重复,浪费时间的工作,这时候就需要使用自动发现机制。我们在手动搭建Prometheus的过程中曾配置过自动发现service,其主要的配置文件如下:

- job_name: 'kubernetes-service-endpoints'
  kubernetes_sd_configs:
  - role: endpoints
  relabel_configs:
  - source_labels: [__meta_kubernetes_service_annotation_prometheus_io_scrape]
    action: keep
    regex: true
  - source_labels: [__meta_kubernetes_service_annotation_prometheus_io_scheme]
    action: replace
    target_label: __scheme__
    regex: (https?)
  - source_labels: [__meta_kubernetes_service_annotation_prometheus_io_path]
    action: replace
    target_label: __metrics_path__
    regex: (.+)
  - source_labels: [__address__, __meta_kubernetes_service_annotation_prometheus_io_port]
    action: replace
    target_label: __address__
    regex: ([^:]+)(?::\d+)?;(\d+)
    replacement: $1:$2
  - action: labelmap
    regex: __meta_kubernetes_service_label_(.+)
  - source_labels: [__meta_kubernetes_namespace]
    action: replace
    target_label: kubernetes_namespace
  - source_labels: [__meta_kubernetes_service_name]
    action: replace
    target_label: kubernetes_name

要想自动被发现,只需要在service的配置清单中加上annotations: prometheus.io/scrape=true。 我们将上面的文件保存为prometheus-additional.yaml,然后用这个文件创建一个secret。

# kubectl -n monitoring create secret generic additional-config --from-file=prometheus-additional.yaml 
secret/additional-config created

然后我们在prometheus的配置清单中添加这个配置: cat prometheus-prometheus.yaml

apiVersion: monitoring.coreos.com/v1
kind: Prometheus
metadata:
  labels:
    prometheus: k8s
  name: k8s
  namespace: monitoring
spec:
  alerting:
    alertmanagers:
    - name: alertmanager-main
      namespace: monitoring
      port: web
  baseImage: quay.io/prometheus/prometheus
  nodeSelector:
    kubernetes.io/os: linux
  podMonitorSelector: {}
  replicas: 2
  resources:
    requests:
      memory: 400Mi
  ruleSelector:
    matchLabels:
      prometheus: k8s
      role: alert-rules
  securityContext:
    fsGroup: 2000
    runAsNonRoot: true
    runAsUser: 1000
  additionalScrapeConfigs:
    name: additional-config
    key: prometheus-additional.yaml
  serviceAccountName: prometheus-k8s
  serviceMonitorNamespaceSelector: {}
  serviceMonitorSelector: {}
  version: v2.11.0

然后更新一下prometheus的配置:

# kubectl apply -f prometheus-prometheus.yaml 
prometheus.monitoring.coreos.com/k8s configured

然后我们查看prometheus的日志,发现很多错误:

# kubectl logs -f prometheus-k8s-0 prometheus -n monitoring

从日志可以看出,其提示的是权限问题,在kubernetes中涉及到权限问题一般就是RBAC中配置问题,我们查看prometheus的配置清单发现其使用了一个prometheus-k8s的ServiceAccount:

而其绑定的是一个叫prometheus-k8s的ClusterRole:

# kubectl get clusterrole prometheus-k8s -n monitoring  -o yaml
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"rbac.authorization.k8s.io/v1","kind":"ClusterRole","metadata":{"annotations":{},"name":"prometheus-k8s"},"rules":[{"apiGroups":[""],"resources":["nodes/metrics"],"verbs":["get"]},{"nonResourceURLs":["/metrics"],"verbs":["get"]}]}
  creationTimestamp: "2019-12-02T03:03:44Z"
  name: prometheus-k8s
  resourceVersion: "1128592"
  selfLink: /apis/rbac.authorization.k8s.io/v1/clusterroles/prometheus-k8s
  uid: 4f87ca47-7769-432b-b96a-1b826b28003d
rules:
- apiGroups:
  - ""
  resources:
  - nodes/metrics
  verbs:
  - get
- nonResourceURLs:
  - /metrics
  verbs:
  - get

从上面可以知道,这个clusterrole并没有service和pod的一些相关权限。接下来我们修改这个clusterrole。 prometheus-clusterRole.yaml

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: prometheus-k8s
rules:
- apiGroups:
  - ""
  resources:
  - nodes/metrics
  - configmaps
  verbs:
  - get
- apiGroups:
  - ""
  resources:
  - nodes
  - pods
  - services
  - endpoints
  - nodes/proxy
  verbs:
  - get
  - list
  - watch
- nonResourceURLs:
  - /metrics
  verbs:
  - get

然后我们更新这个资源清单:

# kubectl apply -f prometheus-clusterRole.yaml
clusterrole.rbac.authorization.k8s.io/prometheus-k8s configured

然后等待一段时间我们可以发现自动发现成功。

提示:配置自动发现,首先annotations里需要配置prometheus.io/scrape=true,其次你的应用要有exporter去收集信息,比如我们如下的redis配置:

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: redis
  namespace: kube-ops
spec:
  template:
    metadata:
      annotations:
        prometheus.io/scrape: "true"
        prometheus.io/port: "9121"
      labels:
        app: redis
    spec:
      containers:
      - name: redis
        image: redis:4
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
        ports:
        - containerPort: 6379
      - name: redis-exporter
        image: oliver006/redis_exporter:latest
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
        ports:
        - containerPort: 9121
---
kind: Service
apiVersion: v1
metadata:
  name: redis
  namespace: kube-ops
  annotations:
    prometheus.io/scrape: "true"
    prometheus.io/port: "9121"
spec:
  selector:
    app: redis
  ports:
  - name: redis
    port: 6379
    targetPort: 6379
  - name: prom
    port: 9121
    targetPort: 9121

4.2、数据持久化配置

如果我们直接git clone下来的,不做任何修改,Prometheus虽然使用的是statefuleSet,但是其用的存储卷是emptyDir,在删除Pod或者重建Pod,原始数据是会丢失的。所以在真实环境我们需要对其进行持久化,首先创建storageClass,如果是用NFS做持久化,详见第四章持久化存储中的storageClass部分。我们这里依然用的NFS做持久化。

创建StorageClass: prometheus-storage.yaml

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: prometheus-storage
provisioner: rookieops/nfs

其中provisioner需要指定我们在创建nfs-client-provisioner中指定的名字,不能随意修改。

配置prometheus的配置清单: prometheus-prometheus.yaml

apiVersion: monitoring.coreos.com/v1
kind: Prometheus
metadata:
  labels:
    prometheus: k8s
  name: k8s
  namespace: monitoring
spec:
  alerting:
    alertmanagers:
    - name: alertmanager-main
      namespace: monitoring
      port: web
  storage:
    volumeClaimTemplate:
      spec:
        storageClassName: prometheus-storage
        resources:
          requests:
            storage: 20Gi
  baseImage: quay.io/prometheus/prometheus
  nodeSelector:
    kubernetes.io/os: linux
  podMonitorSelector: {}
  replicas: 2
  resources:
    requests:
      memory: 400Mi
  ruleSelector:
    matchLabels:
      prometheus: k8s
      role: alert-rules
  securityContext:
    fsGroup: 2000
    runAsNonRoot: true
    runAsUser: 1000
  additionalScrapeConfigs:
    name: additional-config
    key: prometheus-additional.yaml
  serviceAccountName: prometheus-k8s
  serviceMonitorNamespaceSelector: {}
  serviceMonitorSelector: {}
  version: v2.11.0

然后就可以正常使用持久化了,建议在部署之初就做更改。