前言
prometheus 是监控应用软件类似于nagios.
安装
1.官网下载prometheus-2.2.0.linux-amd64压缩包,解压,执行./prometheus即可。这里重要的是配置文件。
a.如果要远程热加载配置文件,启动时加上--web.enable-lifecycle参数。 调用指令是curl -X POST http://localhost:9090/-/reload
b.重要掌握 prometheus.yml 配置文件.prometheus启动时会加载它。
[root@vm-local1 prometheus-2.2.0.linux-amd64]# cat prometheus.yml # my global config global: scrape_interval: 15s # Set the scrape interval to every 15 seconds. Default is every 1 minute. evaluation_interval: 15s # Evaluate rules every 15 seconds. The default is every 1 minute.评估间隔 # scrape_timeout is set to the global default (10s). 默认抓取超时10秒 # Alertmanager configuration #管理报警配置 alerting: alertmanagers: - static_configs: - targets: ["localhost:9093"] #管理报警包需要单独下载,默认启动端口是9093 # Load rules once and periodically evaluate them according to the global 'evaluation_interval'. rule_files: # - "first_rules.yml" # - "second_rules.yml" - rules/mengyuan.rules #要发送报警,就得写规则,定义规则文件 # A scrape configuration containing exactly one endpoint to scrape: # Here it's Prometheus itself. scrape_configs: #抓取配置,就是你要抓取那些主机 # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config. - job_name: 'prometheus' #任务名称 # metrics_path defaults to '/metrics' #默认抓取监控机的url后缀地址是/metrics # scheme defaults to 'http'. #模式是http static_configs: - targets: ['localhost:9090','localhost:9100'] labels: group: 'zus' #targets就是要抓取的主机,对应的客户端,我这有两个,把它们俩规定为一个组,组名是zus - job_name: dj #又建立个任务名称 static_configs: - targets: ['localhost:8000'] #我用django自定义的客户端
注意:
localhost:9090,默认prometheus提供了数据抓取接口,9100端口是prometheus提供的一个监控客户端
2.安装prometheus客户端
官网下载node_exporter-0.16.0-rc.1.linux-amd64客户端,解压,执行./node_exporter 即可,默认是9100端口
3.如何自定义一个客户端,其实很简单,只要返回的数据库类型是这样就可以.我这用的django..只要格式正确就可以
def metrics(req):
ss = "feiji 32" + "\n" + "caidian 31"
return HttpResponse(ss)
4.编写 rules/mengyuan.rules 规则,规则是发送报警的前提
[root@vm-local1 rules]# cat mengyuan.rules groups: - name: zus rules: # Alert for any instance that is unreachable for >5 minutes. - alert: InstanceDown #报警名字随便写 expr: up == 0 #这是一个表达式,如果主机up状态为0,表示关机了,条件为真就会触发报警 可以通过$value得到值 for: 5s #5s内,还是0,就发送报警信息,当然是发送给报警管理器 labels: severity: page #这个类型的报警定了个标签 annotations: summary: "Instance {{ $labels.instance }} down dangqian {{ $value }}" description: "{{ $labels.instance }} of job {{ $labels.job }} has been down for more than 5 minutes."
5.现在安装报警管理器
a.官网下载alertmanager-0.15.0-rc.1.linux-amd64
重要的还是配置文件,创建修改它
[root@vm-local1 alertmanager-0.15.0-rc.1.linux-amd64]# cat alertmanager.yml route: receiver: mengyuan2 #接收的名字,默认必须有一个,对应receivers的- name group_wait: 1s #等待1s group_interval: 1s #发送间隔1s repeat_interval: 1m #重复发送等待1m分钟再发 group_by: ["zus"] routes: #路由了,匹配规则标签的severity:page 走 receiver: mengyuan , 如果routes不写,就会走默认的mengyuan2 - receiver: mengyuan match: severity: page receivers: - name: 'mengyuan' webhook_configs: #这我用的webhook_configs 钩子方法, 默认会把规则的报警信息发送到127.0.0.1:8000 - url: http://127.0.0.1:8000 send_resolved: true - name: 'mengyuan2' webhook_configs: - url: http://127.0.0.1:8000/2 send_resolved: true
6.django接收报警发过来的消息
用Django的 request.body会受到json格式的数据,大概像这样
{"receiver":"mengyuan","status":"resolved","alerts":[{"status":"resolved","labels":{"alertname":"InstanceDown","group":"zus","instance":"localhost:9100","job":"prometheus","severity":"page"},"annotations":{"description":"localhost:9100 of job prometheus has been down for more than 5 minutes.","summary":"Instance localhost:9100 down dangqian 0"},"startsAt":"2018-04-06T22:34:13.51281763+08:00","endsAt":"2018-04-06T23:07:43.514552824+08:00","generatorURL":"http://vm-local1:9090/graph?g0.expr=up+%3D%3D+0\u0026g0.tab=1"}],"groupLabels":{},"commonLabels":{"alertname":"InstanceDown","group":"zus","instance":"localhost:9100","job":"prometheus","severity":"page"},"commonAnnotations":{"description":"localhost:9100 of job prometheus has been down for more than 5 minutes.","summary":"Instance localhost:9100 down dangqian 0"},"externalURL":"http://vm-local1:9093","version":"4","groupKey":"{}/{severity=\"page\"}:{}"}
到此,我就可以根据收到的数据,调用邮件接口,或其他第三方报警接口了。
总结:
本人也是刚入门。做的一个笔记。