一、Prometheus 介绍

Prometheus Operator 是 CoreOS 开发的基于 Prometheus 的 Kubernetes 监控方案,也可能是目前功能最全面的开源方案。
Prometheus Operator 通过 Grafana 展示监控数据,预定义了一系列的 Dashboard

1.1、Prometheus 架构

Prometheus 是一个非常优秀的监控工具。准确的说,应该是监控方案。Prometheus 提供了数据搜集、存储、处理、可视化和告警一套完整的解决方案。Prometheus 的架构如下图所示:
Kubernetes 监控方案之 Prometheus Operator(十九)

Prometheus Server
Prometheus Server 负责从 Exporter 拉取和存储监控数据,并提供一套灵活的查询语言(PromQL)供用户使用。

Exporter
Exporter 负责收集目标对象(host, container...)的性能数据,并通过 HTTP 接口供 Prometheus Server 获取。

可视化组件
监控数据的可视化展现对于监控方案至关重要。以前 Prometheus 自己开发了一套工具,不过后来废弃了,因为开源社区出现了更为优秀的产品 Grafana。Grafana 能够与 Prometheus 无缝集成,提供完美的数据展示能力。

Alertmanager
用户可以定义基于监控数据的告警规则,规则会触发告警。一旦 Alermanager 收到告警,会通过预定义的方式发出告警通知。支持的方式包括 Email、PagerDuty、Webhook 等.

1.2、Prometheus Operator 架构

Prometheus Operator 目前功能最全面的开源监控方案。 能够监控Node Port ,并支持集群的各种管理组件,如 API Server 、Scheduler、Controller Manager等。

Prometheus Operator 的目标是尽可能简化在 Kubernetes 中部署和维护 Prometheus 的工作。其架构如下图所示:
Kubernetes 监控方案之 Prometheus Operator(十九)

图上的每一个对象都是 Kubernetes 中运行的资源。

Operator
Operator 即 Prometheus Operator,在 Kubernetes 中以 Deployment 运行。其职责是部署和管理 Prometheus Server,根据 ServiceMonitor 动态更新 Prometheus Server 的监控对象。

Prometheus Server
Prometheus Server 会作为 Kubernetes 应用部署到集群中。为了更好地在 Kubernetes 中管理 Prometheus,CoreOS 的开发人员专门定义了一个命名为 Prometheus 类型的 Kubernetes 定制化资源。我们可以把 Prometheus看作是一种特殊的 Deployment,它的用途就是专门部署 Prometheus Server。

Service
这里的 Service 就是 Cluster 中的 Service 资源,也是 Prometheus 要监控的对象,在 Prometheus 中叫做 Target。每个监控对象都有一个对应的 Service。比如要监控 Kubernetes Scheduler,就得有一个与 Scheduler 对应的 Service。当然,Kubernetes 集群默认是没有这个 Service 的,Prometheus Operator 会负责创建。

ServiceMonitor
Operator 能够动态更新 Prometheus 的 Target 列表,ServiceMonitor 就是 Target 的抽象。比如想监控 Kubernetes Scheduler,用户可以创建一个与 Scheduler Service 相映射的 ServiceMonitor 对象。Operator 则会发现这个新的 ServiceMonitor,并将 Scheduler 的 Target 添加到 Prometheus 的监控列表中。

ServiceMonitor 也是 Prometheus Operator 专门开发的一种 Kubernetes 定制化资源类型。

Alertmanager
除了 Prometheus 和 ServiceMonitor,Alertmanager 是 Operator 开发的第三种 Kubernetes 定制化资源。我们可以把 Alertmanager 看作是一种特殊的 Deployment,它的用途就是专门部署 Alertmanager 组件。

二、Helm 安装部署

Helm 有两个重要的概念:chart 和 release。
chart 是创建一个应用的信息集合,包括各种 Kubernetes 对象的配置模板、参数定义、依赖关系、文档说明等。chart 是应用部署的自包含逻辑单元。可以将 chart 想象成 apt、yum 中的软件安装包。
release 是 chart 的运行实例,代表了一个正在运行的应用。当 chart 被安装到 Kubernetes 集群,就生成一个 release。chart 能够多次安装到同一个集群,每次安装都是一个 release。

2.1、Helm 客户端安装

https://github.com/helm/helm/releases下载最新的版本。

[root@master ~]# tar xf helm-v2.12.1-linux-amd64.tar.gz
[root@master ~]# cp linux-amd64/helm /usr/local/bin/
[root@master ~]# helm version
Client: &version.Version{SemVer:"v2.12.1", GitCommit:"02a47c7249b1fc6d8fd3b94e6b4babf9d818144e", GitTreeState:"clean"}
Error: could not find tiller

目前只能查看到客户端的版本,服务器还没有安装。

2.2、Tiller 服务器安装

对于启用了 RBAC 的集群,我们首先创建授权,参照文档https://github.com/helm/helm/blob/master/docs/rbac.md,创建rbac-config.yaml,内容如下:

apiVersion: v1
kind: ServiceAccount
metadata:
  name: tiller
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: tiller
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: cluster-admin
subjects:
  - kind: ServiceAccount
    name: tiller
    namespace: kube-system

然后进行创建。

[root@master ~]# kubectl apply -f rbac-config.yaml 
serviceaccount/tiller created
clusterrolebinding.rbac.authorization.k8s.io/tiller created

Tiller 服务器安装非常简单,只需要执行 helm init,版本尽快使用相同的,因为国内访问不了google的镜像,我们使用国内的阿里云:

[root@master ~]# helm init --service-account tiller --upgrade -i registry.cn-hangzhou.aliyuncs.com/google_containers/tiller:v2.12.1 --stable-repo-url https://kubernetes.oss-cn-hangzhou.aliyuncs.com/charts
Creating /root/.helm/repository 
Creating /root/.helm/repository/cache 
Creating /root/.helm/repository/local 
Creating /root/.helm/plugins 
Creating /root/.helm/starters 
Creating /root/.helm/cache/archive 
Creating /root/.helm/repository/repositories.yaml 
Adding stable repo with URL: https://kubernetes.oss-cn-hangzhou.aliyuncs.com/charts 
Adding local repo with URL: http://127.0.0.1:8879/charts 
$HELM_HOME has been configured at /root/.helm.

Tiller (the Helm server-side component) has been installed into your Kubernetes Cluster.

Please note: by default, Tiller is deployed with an insecure 'allow unauthenticated users' policy.
To prevent this, run `helm init` with the --tiller-tls-verify flag.
For more information on securing your installation see: https://docs.helm.sh/using_helm/#securing-your-helm-installation
Happy Helming!

查看安装结果。

[root@master ~]# helm version
Client: &version.Version{SemVer:"v2.12.1", GitCommit:"02a47c7249b1fc6d8fd3b94e6b4babf9d818144e", GitTreeState:"clean"}
Server: &version.Version{SemVer:"v2.12.1", GitCommit:"02a47c7249b1fc6d8fd3b94e6b4babf9d818144e", GitTreeState:"clean"}
[root@master ~]# helm repo list
NAME    URL                                                   
stable  https://kubernetes.oss-cn-hangzhou.aliyuncs.com/charts
local   http://127.0.0.1:8879/charts

三、部署 Prometheus Operator

本节在实践时使用的是 Prometheus Operator 版本 v0.26.0。由于项目开发迭代速度很快,部署方法可能会更新,必要时请参考官方文档。

[root@master ~]# git clone https://github.com/coreos/prometheus-operator.git
[root@master ~]# cd prometheus-operator/

为方便管理,创建一个单独的 Namespace monitoring,Prometheus Operator 相关的组件都会部署到这个 Namespace。

[root@master prometheus-operator]# kubectl create namespace monitoring
namespace/monitoring created

3.1、安装 Prometheus Operator Deployment

首先更新一下 repo 源。

helm repo update

然后进行 helm 安装,因为要下载几百兆的镜像,速度会慢一些,也可以提醒把镜像下载好,建议使用阿里云的镜像,然后再改名,这样速度比较快。

[root@master prometheus-operator]# helm install --name prometheus-operator --set rbacEnable=true --namespace=monitoring helm/prometheus-operator
NAME:   prometheus-operator
LAST DEPLOYED: Tue Dec 25 22:09:31 2018
NAMESPACE: monitoring
STATUS: DEPLOYED

RESOURCES:
==> v1beta1/PodSecurityPolicy
NAME                 PRIV   CAPS      SELINUX   RUNASUSER  FSGROUP    SUPGROUP  READONLYROOTFS  VOLUMES
prometheus-operator  false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim

==> v1/ConfigMap
NAME                 DATA  AGE
prometheus-operator  1     3s

==> v1/ServiceAccount
NAME                 SECRETS  AGE
prometheus-operator  1        3s

==> v1beta1/ClusterRole
NAME                     AGE
prometheus-operator      3s
psp-prometheus-operator  3s

==> v1beta1/ClusterRoleBinding
NAME                     AGE
prometheus-operator      3s
psp-prometheus-operator  3s

==> v1beta1/Deployment
NAME                 DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE
prometheus-operator  1        1        1           1          3s

==> v1/Pod(related)
NAME                                  READY  STATUS   RESTARTS  AGE
prometheus-operator-867bbfddbd-vpsjd  1/1    Running  0         3s

NOTES:
The Prometheus Operator has been installed. Check its status by running:
  kubectl --namespace monitoring get pods -l "app=prometheus-operator,release=prometheus-operator"

Visit https://github.com/coreos/prometheus-operator for instructions on how
to create & configure Alertmanager and Prometheus instances using the Operator.

查看创建的资源。

[root@master prometheus-operator]# kubectl get pod -n monitoring
NAME                                   READY   STATUS    RESTARTS   AGE
prometheus-operator-867bbfddbd-vpsjd   1/1     Running   0          95s
[root@master prometheus-operator]# kubectl get deploy -n monitoring
NAME                  READY   UP-TO-DATE   AVAILABLE   AGE
prometheus-operator   1/1     1            1           103s
[root@master prometheus-operator]# helm list
NAME                REVISION    UPDATED                     STATUS      CHART                       APP VERSION NAMESPACE 
prometheus-operator 1           Tue Dec 25 22:09:31 2018    DEPLOYED    prometheus-operator-0.0.29  0.20.0      monitoring

3.2、安装 Prometheus

[root@master prometheus-operator]# helm install --name prometheus --set serviceMonitorsSelector.app=prometheus --set ruleSelector.app=prometheus --namespace=monitoring helm/prometheus
NAME:   prometheus
LAST DEPLOYED: Tue Dec 25 22:17:06 2018
NAMESPACE: monitoring
STATUS: DEPLOYED

RESOURCES:
==> v1/Prometheus
NAME        AGE
prometheus  0s

==> v1/PrometheusRule
NAME              AGE
prometheus-rules  0s

==> v1/ServiceMonitor
NAME        AGE
prometheus  0s

==> v1beta1/PodSecurityPolicy
NAME        PRIV   CAPS      SELINUX   RUNASUSER  FSGROUP    SUPGROUP  READONLYROOTFS  VOLUMES
prometheus  false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim

==> v1/ServiceAccount
NAME        SECRETS  AGE
prometheus  1        0s

==> v1beta1/ClusterRole
NAME            AGE
prometheus      0s
psp-prometheus  0s

==> v1beta1/ClusterRoleBinding
NAME            AGE
prometheus      0s
psp-prometheus  0s

==> v1/Service
NAME        TYPE       CLUSTER-IP      EXTERNAL-IP  PORT(S)   AGE
prometheus  ClusterIP  10.109.224.217  <none>       9090/TCP  0s

NOTES:
A new Prometheus instance has been created.

DEPRECATION NOTICE:

- additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels

查看创建情况。

[root@master prometheus-operator]# kubectl get pod -n monitoring
NAME                                   READY   STATUS    RESTARTS   AGE
prometheus-operator-867bbfddbd-vpsjd   1/1     Running   0          16m
prometheus-prometheus-0                3/3     Running   1          9m10s
[root@master prometheus-operator]# kubectl get svc -n monitoring
NAME                  TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
prometheus                            ClusterIP   10.111.72.32     <none>        9090/TCP            22m
prometheus-operated                   ClusterIP   None             <none>        9090/TCP            19m

3.3、安装 Alertmanager

[root@master prometheus-operator]# helm install --name alertmanager --namespace=monitoring helm/alertmanager
NAME:   alertmanager
LAST DEPLOYED: Tue Dec 25 22:30:11 2018
NAMESPACE: monitoring
STATUS: DEPLOYED

RESOURCES:
==> v1/PrometheusRule
NAME          AGE
alertmanager  0s

==> v1/ServiceMonitor
NAME          AGE
alertmanager  0s

==> v1beta1/PodSecurityPolicy
NAME          PRIV   CAPS      SELINUX   RUNASUSER  FSGROUP    SUPGROUP  READONLYROOTFS  VOLUMES
alertmanager  false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim

==> v1/Secret
NAME                       TYPE    DATA  AGE
alertmanager-alertmanager  Opaque  1     0s

==> v1beta1/ClusterRole
NAME              AGE
psp-alertmanager  0s

==> v1beta1/ClusterRoleBinding
NAME              AGE
psp-alertmanager  0s

==> v1/Service
NAME          TYPE       CLUSTER-IP     EXTERNAL-IP  PORT(S)   AGE
alertmanager  ClusterIP  10.102.68.166  <none>       9093/TCP  0s

==> v1/Alertmanager
NAME          AGE
alertmanager  0s

NOTES:
A new Alertmanager instance has been created.

DEPRECATION NOTICE:

- additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels

查看安装结果。

[root@master prometheus-operator]# kubectl get pod -n monitoring
NAME                                   READY   STATUS    RESTARTS   AGE
alertmanager-alertmanager-0            2/2     Running   0          13m
prometheus-operator-867bbfddbd-vpsjd   1/1     Running   0          33m
prometheus-prometheus-0                3/3     Running   1          26m

3.4、安装 kube-prometheus

kube-prometheus 是一个 Helm Chart,打包了监控 Kubernetes 需要的所有 Exporter 和 ServiceMonitor,会创建几个 Service。
https://github.com/coreos/prometheus-operator/blob/master/helm/README.md
从上面的连接中可以看到调整了安装方式,安装过程如下:

[root@master prometheus-operator]# mkdir -p helm/kube-prometheus/charts
[root@master prometheus-operator]# helm package -d helm/kube-prometheus/charts helm/alertmanager helm/grafana helm/prometheus  helm/exporter-kube-dns \
> helm/exporter-kube-scheduler helm/exporter-kubelets helm/exporter-node helm/exporter-kube-controller-manager \
> helm/exporter-kube-etcd helm/exporter-kube-state helm/exporter-coredns helm/exporter-kubernetes
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/alertmanager-0.1.7.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/grafana-0.0.37.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/prometheus-0.0.51.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kube-dns-0.1.7.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kube-scheduler-0.1.9.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kubelets-0.2.11.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-node-0.4.6.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kube-controller-manager-0.1.10.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kube-etcd-0.1.15.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kube-state-0.2.6.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-coredns-0.0.3.tgz
Successfully packaged chart and saved it to: helm/kube-prometheus/charts/exporter-kubernetes-0.1.10.tgz
[root@master prometheus-operator]# helm install helm/kube-prometheus --name kube-prometheus --namespace monitoring
NAME:   kube-prometheus
LAST DEPLOYED: Tue Dec 25 23:02:25 2018
NAMESPACE: monitoring
STATUS: DEPLOYED

RESOURCES:
==> v1/Secret
NAME                          TYPE    DATA  AGE
alertmanager-kube-prometheus  Opaque  1     1s
kube-prometheus-grafana       Opaque  2     1s

==> v1beta1/RoleBinding
NAME                                 AGE
kube-prometheus-exporter-kube-state  1s

==> v1/Service
NAME                                              TYPE       CLUSTER-IP      EXTERNAL-IP  PORT(S)              AGE
kube-prometheus-alertmanager                      ClusterIP  10.109.99.33    <none>       9093/TCP             1s
kube-prometheus-exporter-kube-controller-manager  ClusterIP  None            <none>       10252/TCP            1s
kube-prometheus-exporter-kube-dns                 ClusterIP  None            <none>       10054/TCP,10055/TCP  1s
kube-prometheus-exporter-kube-etcd                ClusterIP  None            <none>       4001/TCP             1s
kube-prometheus-exporter-kube-scheduler           ClusterIP  None            <none>       10251/TCP            1s
kube-prometheus-exporter-kube-state               ClusterIP  10.106.111.57   <none>       80/TCP               1s
kube-prometheus-exporter-node                     ClusterIP  10.107.178.109  <none>       9100/TCP             1s
kube-prometheus-grafana                           ClusterIP  10.110.171.226  <none>       80/TCP               1s
kube-prometheus                                   ClusterIP  10.102.19.97    <none>       9090/TCP             1s

==> v1beta1/Deployment
NAME                                 DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE
kube-prometheus-exporter-kube-state  1        1        1           0          1s
kube-prometheus-grafana              1        1        1           0          1s

==> v1/Pod(related)
NAME                                                  READY  STATUS             RESTARTS  AGE
kube-prometheus-exporter-node-8cclq                   0/1    ContainerCreating  0         1s
kube-prometheus-exporter-node-xsqvj                   0/1    ContainerCreating  0         1s
kube-prometheus-exporter-node-zcjfj                   0/1    ContainerCreating  0         1s
kube-prometheus-exporter-kube-state-7bb8cf75d9-czp24  0/2    ContainerCreating  0         1s
kube-prometheus-grafana-6f4bb75c95-jvfzn              0/2    ContainerCreating  0         1s

==> v1beta1/ClusterRole
NAME                                     AGE
psp-kube-prometheus-alertmanager         1s
kube-prometheus-exporter-kube-state      1s
psp-kube-prometheus-exporter-kube-state  1s
psp-kube-prometheus-exporter-node        1s
psp-kube-prometheus-grafana              1s
kube-prometheus                          1s
psp-kube-prometheus                      1s

==> v1beta1/DaemonSet
NAME                           DESIRED  CURRENT  READY  UP-TO-DATE  AVAILABLE  NODE SELECTOR  AGE
kube-prometheus-exporter-node  3        3        0      3           0          <none>         1s

==> v1beta1/PodSecurityPolicy
NAME                                 PRIV   CAPS      SELINUX   RUNASUSER  FSGROUP    SUPGROUP  READONLYROOTFS  VOLUMES
kube-prometheus-alertmanager         false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim
kube-prometheus-exporter-kube-state  false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim
kube-prometheus-exporter-node        false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim,hostPath
kube-prometheus-grafana              false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim,hostPath
kube-prometheus                      false  RunAsAny  RunAsAny  MustRunAs  MustRunAs  false     configMap,emptyDir,projected,secret,downwardAPI,persistentVolumeClaim

==> v1/ServiceAccount
NAME                                 SECRETS  AGE
kube-prometheus-exporter-kube-state  1        1s
kube-prometheus-exporter-node        1        1s
kube-prometheus-grafana              1        1s
kube-prometheus                      1        1s

==> v1beta1/ClusterRoleBinding
NAME                                     AGE
psp-kube-prometheus-alertmanager         1s
kube-prometheus-exporter-kube-state      1s
psp-kube-prometheus-exporter-kube-state  1s
psp-kube-prometheus-exporter-node        1s
psp-kube-prometheus-grafana              1s
kube-prometheus                          1s
psp-kube-prometheus                      1s

==> v1beta1/Role
NAME                                 AGE
kube-prometheus-exporter-kube-state  1s

==> v1/Prometheus
NAME             AGE
kube-prometheus  1s

==> v1/ServiceMonitor
NAME                                              AGE
kube-prometheus-alertmanager                      0s
kube-prometheus-exporter-kube-controller-manager  0s
kube-prometheus-exporter-kube-dns                 0s
kube-prometheus-exporter-kube-etcd                0s
kube-prometheus-exporter-kube-scheduler           0s
kube-prometheus-exporter-kube-state               0s
kube-prometheus-exporter-kubelets                 0s
kube-prometheus-exporter-kubernetes               0s
kube-prometheus-exporter-node                     0s
kube-prometheus-grafana                           0s
kube-prometheus                                   0s

==> v1/ConfigMap
NAME                     DATA  AGE
kube-prometheus-grafana  10    1s

==> v1/Alertmanager
NAME             AGE
kube-prometheus  1s

==> v1/PrometheusRule
NAME                                              AGE
kube-prometheus-alertmanager                      1s
kube-prometheus-exporter-kube-controller-manager  1s
kube-prometheus-exporter-kube-etcd                1s
kube-prometheus-exporter-kube-scheduler           1s
kube-prometheus-exporter-kube-state               1s
kube-prometheus-exporter-kubelets                 1s
kube-prometheus-exporter-kubernetes               1s
kube-prometheus-exporter-node                     1s
kube-prometheus-rules                             1s
kube-prometheus                                   0s

NOTES:
DEPRECATION NOTICE:

- alertmanager.ingress.fqdn is not used anymore, use alertmanager.ingress.hosts []
- prometheus.ingress.fqdn is not used anymore, use prometheus.ingress.hosts []
- grafana.ingress.fqdn is not used anymore, use prometheus.grafana.hosts []

- additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- prometheus.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- alertmanager.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- exporter-kube-controller-manager.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- exporter-kube-etcd.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- exporter-kube-scheduler.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- exporter-kubelets.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels
- exporter-kubernetes.additionalRulesConfigMapLabels is not used anymore, use additionalRulesLabels

等需要的镜像下载完成,因为有的镜像国内访问不到,建议都从阿里云间接下载,我们查看安装结果。
每个 Exporter 会对应一个 Service,为 Pormetheus 提供 Kubernetes 集群的各类监控数据。

[root@master prometheus-operator]# kubectl get svc -n monitoring
NAME                                  TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)             AGE
alertmanager                          ClusterIP    10.102.68.166    <none>        9093/TCP      52m
alertmanager-operated                 ClusterIP   None             <none>        9093/TCP,6783/TCP   52m
kube-prometheus                       ClusterIP   10.109.193.133   <none>        9090/TCP            3m42s
kube-prometheus-alertmanager          ClusterIP   10.110.67.174    <none>        9093/TCP            3m42s
kube-prometheus-exporter-kube-state   ClusterIP   10.101.225.77    <none>        80/TCP              3m42s
kube-prometheus-exporter-node         ClusterIP   10.103.162.196   <none>        9100/TCP            3m42s
kube-prometheus-grafana               ClusterIP   10.101.81.167    <none>        80/TCP              3m42s
prometheus                            ClusterIP    10.109.224.217   <none>        9090/TCP      65m
prometheus-operated                   ClusterIP   None             <none>        9090/TCP            65m

每个 Service 对应一个 ServiceMonitor,组成 Pormetheus 的 Target 列表。

[root@master prometheus-operator]# kubectl get servicemonitor -n monitoring
NAME                                               AGE
alertmanager                                       14m
kube-prometheus                                    22m
kube-prometheus-alertmanager                       22m
kube-prometheus-exporter-kube-controller-manager   22m
kube-prometheus-exporter-kube-dns                  22m
kube-prometheus-exporter-kube-etcd                 22m
kube-prometheus-exporter-kube-scheduler            22m
kube-prometheus-exporter-kube-state                22m
kube-prometheus-exporter-kubelets                  22m
kube-prometheus-exporter-kubernetes                22m
kube-prometheus-exporter-node                      22m
kube-prometheus-grafana                            22m
prometheus                                         16m
prometheus-operator                                2h

如下是与 Prometheus Operator 相关的所有 Pod,我们注意到有些 Exporter 没有运行 Pod,这是因为像 API Server、Scheduler、Kubelet 等 Kubernetes 内部组件原生就支持 Prometheus,只需要定义 Service 就能直接从预定义端口获取监控数据。

[root@master prometheus-operator]# kubectl get pod -n monitoring
NAME                                                  READY   STATUS    RESTARTS   AGE
alertmanager-alertmanager-0                           2/2     Running   0          15m
alertmanager-kube-prometheus-0                        2/2     Running   0          15m
kube-prometheus-exporter-kube-state-dc6966bb5-r5kng   2/2     Running   0          25m
kube-prometheus-exporter-node-grtvz                   1/1     Running   0          25m
kube-prometheus-exporter-node-jfq79                   1/1     Running   0          25m
kube-prometheus-exporter-node-n79vq                   1/1     Running   0          25m
kube-prometheus-grafana-6f4bb75c95-bw72r              2/2     Running   0          25m
prometheus-kube-prometheus-0                          3/3     Running   1          15m
prometheus-operator-867bbfddbd-rxj6s                  1/1     Running   0          15m
prometheus-prometheus-0                               3/3     Running   1          15m

为了方便访问 kube-prometheus-grafana,我们将 Service 类型改为 NodePort。

[root@master prometheus-operator]# kubectl patch svc kube-prometheus-grafana -p '{"spec":{"type":"NodePort"}}' -n monitoring
service/alertmanager patched
[root@master prometheus-operator]# kubectl patch svc kube-prometheus -p '{"spec":{"type":"NodePort"}}' -n monitoring
service/kube-prometheus patched
[root@master prometheus-operator]# kubectl get svc -n monitoring
NAME                                  TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)             AGE
alertmanager                          ClusterIP   10.110.11.161    <none>        9093/TCP            21m
alertmanager-operated                 ClusterIP   None             <none>        9093/TCP,6783/TCP   19m
kube-prometheus                       NodePort    10.102.23.109    <none>        9090:31679/TCP      29m
kube-prometheus-alertmanager          NodePort    10.97.58.177     <none>        9093:31627/TCP      29m
kube-prometheus-exporter-kube-state   ClusterIP   10.110.185.195   <none>        80/TCP              29m
kube-prometheus-exporter-node         ClusterIP   10.111.98.237    <none>        9100/TCP            29m
kube-prometheus-grafana               NodePort    10.105.188.204   <none>        80:30357/TCP        29m
prometheus                            ClusterIP   10.111.72.32     <none>        9090/TCP            22m
prometheus-operated                   ClusterIP   None             <none>        9090/TCP            19m

四、查看效果图

4.1、查看 kube-prometheus

访问MASTER_IP:31697,如下所示:

Kubernetes 监控方案之 Prometheus Operator(十九)

4.2、查看 kube-prometheus-alertmanager

访问MASTER_IP:31627,如下所示:

Kubernetes 监控方案之 Prometheus Operator(十九)

4.3、查看 kube-prometheus-grafana

访问MASTER_IP:30357/login,然后登陆,账号密码都是 admin。
Kubernetes 监控方案之 Prometheus Operator(十九)

可以监控 Kubernetes 集群的整体健康状态:
Kubernetes 监控方案之 Prometheus Operator(十九)

整个集群的资源使用情况:
Kubernetes 监控方案之 Prometheus Operator(十九)

Kubernetes 各个管理组件的状态:
Kubernetes 监控方案之 Prometheus Operator(十九)

Kubernetes 监控方案之 Prometheus Operator(十九)

节点的资源使用情况:
Kubernetes 监控方案之 Prometheus Operator(十九)

Deployment 的运行状态:
Kubernetes 监控方案之 Prometheus Operator(十九)

Pod 的运行状态:
Kubernetes 监控方案之 Prometheus Operator(十九)

StatefulSet 运行状态:
Kubernetes 监控方案之 Prometheus Operator(十九)

这些 Dashboard 展示了从集群到 Pod 的运行状况,能够帮助用户更好地运维 Kubernetes。而且 Prometheus Operator 迭代非常快,相信会继续开发出更多更好的功能,所以值得我们花些时间学习和实践。

官方文档:https://github.com/coreos/prometheus-operator/tree/master/helm