这个项目出自coreos，已经存在很久了，第一次尝试的时候还很简陋，现在完善了很多。
项目提供了一键部署脚本，跑起来并不难，不过个人感觉要真正掌握并灵活使用并不是很容易。
kube version: 1.9.1
OS version: debian stretch
1、从github上把项目拉下来。
# git clone https://github.com/coreos/prometheus-operator.git
2、准备镜像，这里用到了墙外的镜像，自行科学下载。
quay.io/prometheus/alertmanager:v0.9.1
quay.io/coreos/configmap-reload:v0.0.1
grafana/grafana:4.5.2
quay.io/coreos/grafana-watcher:v0.0.8
quay.io/coreos/kube-state-metrics:v1.0.1
gcr.io/google_containers/addon-resizer:1.0
quay.io/prometheus/node-exporter:v0.15.0
quay.io/prometheus/prometheus:v2.0.0
quay.io/coreos/prometheus-config-reloader:v0.0.2
quay.io/coreos/prometheus-operator:v0.15.0
3、执行脚本部署项目。
脚本内容如下：
#!/usr/bin/env bash
if [ -z "${KUBECONFIG}" ]; then
export KUBECONFIG=~/.kube/config
fi
# CAUTION - setting NAMESPACE will deploy most components to the given namespace
# however some are hardcoded to 'monitoring'. Only use if you have reviewed all manifests.
if [ -z "${NAMESPACE}" ]; then
NAMESPACE=monitoring
fi
kubectl create namespace "$NAMESPACE"
kctl() {
kubectl --namespace "$NAMESPACE" "$@"
}
kctl apply -f manifests/prometheus-operator
# Wait for CRDs to be ready.
printf "Waiting for Operator to register custom resource definitions..."
until kctl get customresourcedefinitions servicemonitors.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
until kctl get customresourcedefinitions prometheuses.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
until kctl get customresourcedefinitions alertmanagers.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
until kctl get servicemonitors.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
until kctl get prometheuses.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
until kctl get alertmanagers.monitoring.coreos.com > /dev/null 2>&1; do sleep 1; printf "."; done
echo "done!"
kctl apply -f manifests/node-exporter
kctl apply -f manifests/kube-state-metrics
kctl apply -f manifests/grafana/grafana-credentials.yaml
kctl apply -f manifests/grafana
find manifests/prometheus -type f ! -name prometheus-k8s-roles.yaml ! -name prometheus-k8s-role-bindings.yaml -exec kubectl --namespace "$NAMESPACE" apply -f {} ;
kubectl apply -f manifests/prometheus/prometheus-k8s-roles.yaml
kubectl apply -f manifests/prometheus/prometheus-k8s-role-bindings.yaml
kctl apply -f manifests/alertmanager/
从脚本上看，其实很简单，有木有..
# cd contrib/kube-prometheus/
# hack/cluster-monitoring/deploy
namespace "monitoring" created
clusterrolebinding "prometheus-operator" created
clusterrole "prometheus-operator" created
serviceaccount "prometheus-operator" created
service "prometheus-operator" created
deployment "prometheus-operator" created
Waiting for Operator to register custom resource definitions......done!
daemonset "node-exporter" created
service "node-exporter" created
clusterrolebinding "kube-state-metrics" created
clusterrole "kube-state-metrics" created
deployment "kube-state-metrics" created
rolebinding "kube-state-metrics" created
role "kube-state-metrics-resizer" created
serviceaccount "kube-state-metrics" created
service "kube-state-metrics" created
secret "grafana-credentials" created
secret "grafana-credentials" unchanged
configmap "grafana-dashboards-0" created
deployment "grafana" created
service "grafana" created
servicemonitor "alertmanager" created
servicemonitor "prometheus-operator" created
prometheus "k8s" created
servicemonitor "kubelet" created
servicemonitor "prometheus" created
service "prometheus-k8s" created
servicemonitor "node-exporter" created
servicemonitor "kube-scheduler" created
servicemonitor "kube-controller-manager" created
servicemonitor "kube-state-metrics" created
configmap "prometheus-k8s-rules" created
serviceaccount "prometheus-k8s" created
servicemonitor "kube-apiserver" created
role "prometheus-k8s" created
role "prometheus-k8s" created
role "prometheus-k8s" created
clusterrole "prometheus-k8s" created
rolebinding "prometheus-k8s" created
rolebinding "prometheus-k8s" created
rolebinding "prometheus-k8s" created
clusterrolebinding "prometheus-k8s" created
secret "alertmanager-main" created
service "alertmanager-main" created
alertmanager "main" created
4、由于事先准备好了镜像，很快就运行起来了。
# kubectl get po -n monitoring
NAME READY STATUS RESTARTS AGE
alertmanager-main-0 2/2 Running 0 11h
alertmanager-main-1 2/2 Running 0 11h
alertmanager-main-2 2/2 Running 0 11h
grafana-6b67b479d5-2hhnp 2/2 Running 0 11h
kube-state-metrics-6f7b5c94f-r8hm7 2/2 Running 0 11h
node-exporter-27744 1/1 Running 0 11h
node-exporter-9vhlv 1/1 Running 0 11h
node-exporter-rhjfb 1/1 Running 0 11h
node-exporter-xpqr8 1/1 Running 0 11h
prometheus-k8s-0 2/2 Running 0 11h
prometheus-k8s-1 2/2 Running 0 11h
prometheus-operator-8697c7fff9-mm8v5 1/1 Running 0 11h
这里曝光了三个服务：
- Prometheus UI on node port
30900 - Alertmanager UI on node port
30903 - Grafana on node port
30902
通过相应地端口就能访问对应的服务。
5、添加controller-manager和scheduler的监控。
# kubectl apply -f -f manifests/k8s/ -n kube-system
这里其实是添加了2个service，注意namespace是kube-system，而不是monitoring：
# kubectl get ep -n kube-system | grep discovery
kube-controller-manager-prometheus-discovery 192.168.5.104:10252,192.168.5.105:10252,192.168.5.42:10252 3d
kube-scheduler-prometheus-discovery 192.168.5.104:10251,192.168.5.105:10251,192.168.5.42:10251 3d
6、由于我的kubernetes是高可用架构，有三个apiserver，这里有个bug有个需要处理。apiserver是无状态的，三个endpoint会自身冲突，需要是在kubernetes1.9以上，通过给apiserver传递一个参数--endpoint-reconciler-type=lease解决该问题。
默认是这个样子：
解决完是这个样子：
7、添加邮件报警规则。这个配置是用base64加密过的。
# vim manifests/alertmanager/alertmanager-config.yaml
apiVersion: v1
kind: Secret
metadata:
name: alertmanager-main
data:
alertmanager.yaml: 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
要拿到配置内容，用base64反解一下就可以了，默认配置是这样：
# echo "加密的内容" | base64 -d
global:
resolve_timeout: 5m
route:
group_by: ['job']
group_wait: 30s
group_interval: 5m
repeat_interval: 12h
receiver: 'null'
routes:
- match:
alertname: DeadMansSwitch
receiver: 'null'
receivers:
- name: 'null'
添加自己的邮件设置：
global:
resolve_timeout: 5m
smtp_smarthost: "mail.xxxx.com:25"
smtp_from: "example@xxxx.com"
smtp_auth_username: "expample@xxxx.com"
smtp_auth_password: "pass"
route:
group_by: ['alertname', 'cluster', 'service']
group_wait: 30s
group_interval: 5m
repeat_interval: 15m
receiver: team-X-mails
routes:
- match_re:
alertname: ^(host_cpu_usage|node_filesystem_free|host_down)$
receiver: team-X-mails
routes:
- match:
severity: critical
receiver: team-X-mails
receivers:
- name: "team-X-mails"
email_configs:
- to: "example@xxxx.com"
定义好之后，用base64加密一下，替换掉之前的地方。然后应用配置：
# kubectl apply -f manifests/alertmanager/alertmanager-config.yaml -n monitoring
这时候就能收到报警了：
8、到这里部署就完成了，贴一下几个页面：
grafana
prometheus
alertmanager
补充
正常情况下，DeadMansSwitch会一直存在，为了让它不报警，这里我添加了一条匹配规则和一个receivernull，并将报警发送到null：
global:
resolve_timeout: 5m
smtp_smarthost: "mail.xxxx.com:25"
smtp_from: "example@xxxx.com"
smtp_auth_username: "expample@xxxx.com"
smtp_auth_password: "pass"
route:
group_by: ['alertname', 'cluster', 'service']
group_wait: 30s
group_interval: 5m
repeat_interval: 15m
receiver: team-X-mails
routes:
- match_re:
alertname: ^(host_cpu_usage|node_filesystem_free|host_down)$
receiver: team-X-mails
routes:
- match:
severity: critical
receiver: team-X-mails
- match:
alertname: DeadMansSwitch
receiver: 'null'
receivers:
- name: "team-X-mails"
email_configs:
- to: "example@xxxx.com"
- name: 'null'
这样就不会一直收到报警了。