相关推荐

1. 使用Prometheus Operator监控kubetnetes集群

2. 使用Prometheus Operator实现应用自定义监控

一、概述

Alertmanager与Prometheus是相互分离的两个组件。Prometheus服务器根据报警规则将警报发送给Alertmanager,然后Alertmanager将silencing、inhibition、aggregation等消息通过电子邮件、dingtalk和HipChat发送通知。

Alertmanager处理由例如Prometheus服务器等客户端发来的警报。它负责删除重复数据、分组,并将警报通过路由发送到正确的接收器,比如电子邮件、Slack、dingtalk等。Alertmanager还支持groups,silencing和警报抑制的机制。

钉钉作为内部通讯工具,基本上大家在电脑和手机上都能用,消息可以第一时间查看,报警消息的即时性要求比较高,所以适合用钉钉通知。

二、添加钉钉机器人

请参考官方文档:自定义机器人

添加机器人后获取机器人的hook(机器人好像只能在钉钉群里面添加),在后面部署会用到。

机器人hook:https://oapi.dingtalk.com/robot/send?access_token=xxxxxx

三、配置Alertmanager

Alertmanager官方文档:https://github.com/prometheus/docs/blob/db2a09a8a7e193d6e474f37055908a6d432b88b5/content/docs/alerting/configuration.md#webhook_config

修改Alertmanager报警配置,因上面的官方文档已经给出来每个参数的详细信息,就不再一一解释了。

[root@node-01 prometheus]# vim prometheus-operator/values.yaml 

  config:
    global:
      resolve_timeout: 2m
    route:
      group_by: ['job']
      group_wait: 30s
      group_interval: 2m
      repeat_interval: 12h
      receiver: 'webhook'
      routes:
      - match:
          alertname: DeadMansSwitch
        receiver: 'webhook'
    receivers:
    - name: 'webhook'
      webhook_configs:
      - url: http://webhook-dingtalk/dingtalk/send/
        send_resolved: true

更新prometheus-operator

[root@node-01 prometheus]# helm upgrade p ./prometheus-operator

修改成功后可以在alertmanager的status页面看到相关配置 config

四、转换alertmanager数据格式

Alertmanager会以下列JSON格式的数据通过HTTP POST请求发送到端点:

{
  "version": "4",
  "groupKey": <string>,    // key identifying the group of alerts (e.g. to deduplicate)
  "status": "<resolved|firing>",
  "receiver": <string>,
  "groupLabels": <object>,
  "commonLabels": <object>,
  "commonAnnotations": <object>,
  "externalURL": <string>,  // backlink to the Alertmanager.
  "alerts": [
    {
      "labels": <object>,
      "annotations": <object>,
      "startsAt": "<rfc3339>",
      "endsAt": "<rfc3339>"
    },
    ...
  ]
}

这是测试报警数据的示例:

b'{
"receiver":"webhook",
"status":"firing",
"alerts":[{
    "status":"firing",
    "labels":{
        "alertname":"DeadMansSwitch",
        "prometheus":"monitoring/p-prometheus",
        "severity":"none"
        
    },
    "annotations":{
        "message":"This is a DeadMansSwitch meant to ensure that the entire alerting pipeline is functional."
        
    },
    "startsAt":"2019-03-08T10:02:28.680317737Z",
    "endsAt":"0001-01-01T00:00:00Z",
    "generatorURL":"http://prom.cnlinux.club/graph?g0.expr=vector%281%29\\u0026g0.tab=1"
    
    
}],
"groupLabels":{},
"commonLabels":{
    "alertname":"DeadMansSwitch",
    "prometheus":"monitoring/p-prometheus",
    "severity":"none"
    
},
"commonAnnotations":{
"message":"This is a DeadMansSwitch meant to ensure that the entire alerting pipeline is functional."
    
},
"externalURL":"http://alert.cnlinux.club","version":"4",
"groupKey":"{}/{alertname=\\"DeadMansSwitch\\"}:{}"}\n' 

钉钉对数据的格式是有要求的(具体要求在上面钉钉官方文档),所以需要将Alertmanager传过来的数据进行格式转化。

以下我们用自己写的python脚本来转换。

脚本说明:

  • alertmanager传过来的数据中,重要的是labels{}的数据,但是里面数据太多,很多信息在报警的信息中是不需要的,所以在脚本中添加了一个EXCLUDE_LIST列表,用于排除不需要的数据。
[root@node-01 prometheus]# cat app.py
#!/usr/bin/env python
import io, sys


sys.stdout = io.TextIOWrapper(sys.stdout.detach(), encoding='utf-8')
sys.stderr = io.TextIOWrapper(sys.stderr.detach(), encoding='utf-8')

from flask import Flask, Response
from flask import request
import requests
import logging
import json
import locale
#locale.setlocale(locale.LC_ALL,"en_US.UTF-8")


app = Flask(__name__)


console = logging.StreamHandler()
fmt = '%(asctime)s - %(filename)s:%(lineno)s - %(name)s - %(message)s'
formatter = logging.Formatter(fmt)
console.setFormatter(formatter)
log = logging.getLogger("flask_webhook_dingtalk")
log.addHandler(console)
log.setLevel(logging.DEBUG)

EXCLUDE_LIST = ['prometheus', 'endpoint']

@app.route('/')
def index():
    return 'Webhook Dingtalk by Billy https://blog.51cto.com/billy98'

@app.route('/dingtalk/send/',methods=['POST'])

def hander_session():

    profile_url = sys.argv[1]
    post_data = request.get_data()
    post_data = json.loads(post_data.decode("utf-8"))['alerts']
    post_data = post_data[0]
    messa_list = []
    messa_list.append('### 报警类型: %s' % post_data['status'].upper())
    messa_list.append('**startsAt:** %s' % post_data['startsAt'])
    for i in post_data['labels'].keys():
        if i in EXCLUDE_LIST:
            continue
        else:
            messa_list.append("**%s:** %s" % (i, post_data['labels'][i]))
    messa_list.append('**Describe:** %s' % post_data['annotations']['message'])

    messa = (' \\n\\n > '.join(messa_list))
    status = alert_data(messa, post_data['labels']['alertname'], profile_url )
    log.info(status)
    return status

def alert_data(data,title,profile_url):
    headers = {'Content-Type':'application/json'}
    send_data = '{"msgtype": "markdown","markdown": {"title": \"%s\" ,"text": \"%s\" }}' %(title,data)  # type: str
    send_data = send_data.encode('utf-8')
    reps = requests.post(url=profile_url, data=send_data, headers=headers)
    return reps.text

if __name__ == '__main__':
    app.debug = False
    app.run(host='0.0.0.0', port='8080')

五、制作Docker镜像

将上面的python脚本做成镜像,然后把他们以服务的形式运行在k8s集群中,保证高可用。

大家也可以用我已经制作成功的镜像:docker pull billy98/webhook-dingtalk:latest,直接pull即可。

[root@node-01 prometheus]# cat Dockerfile

FROM centos:7 as build
MAINTAINER billy98 5884625@qq.com

RUN curl -o /etc/yum.repos.d/epel.repo http://mirrors.aliyun.com/repo/epel-7.repo && yum install -y python36 python36-pip && pip3.6 install flask requests werkzeug
ADD app.py /usr/local/alert-dingtalk.py

FROM gcr.io/distroless/python3
COPY --from=build /usr/local/alert-dingtalk.py /usr/local/alert-dingtalk.py
COPY --from=build usr/local/lib64/python3.6/site-packages usr/local/lib64/python3.6/site-packages
COPY --from=build usr/local/lib/python3.6/site-packages usr/local/lib/python3.6/site-packages
ENV PYTHONPATH=usr/local/lib/python3.6/site-packages:usr/local/lib64/python3.6/site-packages
EXPOSE 8080
ENTRYPOINT ["python","/usr/local/alert-dingtalk.py"]
[root@node-01 prometheus]# docker build -t billy98/webhook-dingtalk:latest .

我这样build出来的镜像只有50多M,具体的使用方法参考:

distroless:https://github.com/GoogleContainerTools/distroless

六、部署webhook-dingtalk

[root@node-01 prometheus]# cat webhook-dingtalk.yaml 
apiVersion: apps/v1beta2
kind: Deployment
metadata:
  labels:
    app: webhook-dingtalk
  name: webhook-dingtalk
  namespace: monitoring
  #需要和alertmanager在同一个namespace
spec:
  replicas: 1
  selector:
    matchLabels:
      app: webhook-dingtalk
  template:
    metadata:
      labels:
        app: webhook-dingtalk
    spec:
      containers:
      - image: billy98/webhook-dingtalk:latest
        name: webhook-dingtalk
        args:
        - "https://oapi.dingtalk.com/robot/send?access_token=xxxxxx"
        #上面创建的钉钉机器人hook
        ports:
        - containerPort: 8080
          protocol: TCP
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
          limits:
            cpu: 500m
            memory: 500Mi
        livenessProbe:
          failureThreshold: 3
          initialDelaySeconds: 30
          periodSeconds: 10
          successThreshold: 1
          timeoutSeconds: 1
          tcpSocket:
            port: 8080
        readinessProbe:
          failureThreshold: 3
          initialDelaySeconds: 30
          periodSeconds: 10
          successThreshold: 1
          timeoutSeconds: 1
          httpGet:
            port: 8080
            path: /
      imagePullSecrets:
        - name: IfNotPresent
---
apiVersion: v1
kind: Service
metadata:
  labels:
    app: webhook-dingtalk
  name: webhook-dingtalk
  namespace: monitoring
  #需要和alertmanager在同一个namespace
spec:
  ports:
  - name: http
    port: 80
    protocol: TCP
    targetPort: 8080
  selector:
    app: webhook-dingtalk
  type: ClusterIP 

钉钉中报警信息如下:

报警恢复的消息

至此所有的操作已完成。

如有问题欢迎在下面留言交流。希望大家多多关注和点赞,谢谢!