process-exporter部署手册

林劲
2023-12-01

process-exporter部署手册

来源

https://github.com/ncabatoff/process-exporter

说明

一款进程监控导出器

部署方式

二进制文件部署

下载地址:https://github.com/ncabatoff/process-exporter/releases/tag/v0.5.0
命令行常用的参数:

Usage of process-exporter:
  -children
        if a proc is tracked, track with it any children that aren't part of their own group (default true)
  -config.path string
        path to YAML config file
  -debug
        log debugging information to stdout
  -man
        print manual
  -namemapping string
        comma-separated list, alternating process name and capturing regex to apply to cmdline
  -once-to-stdout-delay duration
        Don't bind, just wait this much time, print the metrics once to stdout, and exit
  -procfs string
        path to read proc data from (default "/proc")
  -procnames string
        comma-separated list of process names to monitor
  -recheck
        recheck process names on each scrape
  -threads
        report on per-threadname metrics as well (default true)
  -version
        print version information and exit
  -web.listen-address string
        Address on which to expose metrics and web interface. (default ":9256")
  -web.telemetry-path string
        Path under which to expose metrics. (default "/metrics")

配置文件

配置模板如下:

process_names:
  - matcher1
  - matcher2
  ...
  - matcherN

例如:监控所有的服务进程

 process_names:
  - name: "{{.Comm}}"
    cmdline:
    - '.+'

配置模板选项如下:
{{.Comm}} 包含原始可执行文件的basename,/proc//stat 中的换句话说,2nd 字段

{{.ExeBase}} 包含可执行文件的basename

{{.ExeFull}} 包含可执行文件的完全限定路径

{{.Username}} 包含有效用户的用户名

{{.Matches}} 映射包含应用命令行tlb所产生的所有匹配项

    process_names中的每个项目都必须包含一个或多个选择器(comm,exe或cmdline), 如果存在多个选择器,则它们必须全部匹配。 每个选择器都是一个字符串列表,用于与进程的comm,argv [0]匹配;对于cmdline,则是一个适用于命令行的正则表达式。 cmdline regexp使用Go语法。

    对于comm和exe,字符串列表是一个OR,这意味着与任何字符串匹配的任何进程都将添加到该项目的组中。

    对于cmdline,正则表达式列表为AND,表示它们都必须匹配。 正则表达式中的任何捕获组都必须使用?P 选项为捕获分配一个名称,该名称用于填充.Matches。

例如:> ps -ef | grep redis

redis 771 1 0 Jun05 ? 00:45:49 /usr/bin/redis-server *:6379

可选配置项举例说明
{{.Comm}}groupname=“redis-server”exe或者sh文件名称
{{.ExeBase}}groupname=“redis-server *:6379”/
{{.ExeFull}}groupname="/usr/bin/redis-server *:6379"ps中的进程完成信息
{{.Username}}groupname=“redis”使用进程所属的用户进行分组
{{.Matches}}groupname=“map[:redis]”表示配置到关键字“redis”

安装使用process-exporter

下载

wget https://github.com/ncabatoff/process-exporter/releases/download/v0.5.0/process-exporter-0.5.0.linux-amd64.tar.gz

解压

tar -zxvf  process-exporter-0.5.0.linux-amd64.tar.gz

创建配置文件process-name.yaml 内容为

process_names:
 - name: "{{.Matches}}"
   cmdline:
   - 'nginx'

 - name: "{{.Matches}}"
   cmdline:
   - 'zombie'

重启服务

./process-exporter -config.path process-name.yaml &

查看数据:

http://localhost:9256/metrics 

定义全部进程监控

vim conf.yaml
process_names:
  - name: "{{.Comm}}"
    cmdline:
    - '.+'

常用进程监控项

总进程数
sum(namedprocess_namegroup_states)

总僵尸进程数
sum(namedprocess_namegroup_states{state="Zombie"})

exporter其他可自定义配置选项:

Usage of ./process-exporter:
  -children
        if a proc is tracked, track with it any children that aren't part of their own group (default true)
  -config.path string
        path to YAML config file
  -debug
        log debugging information to stdout
  -man
        print manual
  -namemapping string
        comma-seperated list, alternating process name and capturing regex to apply to cmdline
  -once-to-stdout-delay duration
        Don't bind, just wait this much time, print the metrics once to stdout, and exit
  -procfs string
        path to read proc data from (default "/proc")
  -procnames string
        comma-seperated list of process names to monitor
  -recheck
        recheck process names on each scrape
  -web.listen-address string
        Address on which to expose metrics and web interface. (default ":9256")
  -web.telemetry-path string
        Path under which to expose metrics. (default "/metrics")
 类似资料: