PM2 守护进程

https://github.com/Unitech/pm2

当前使用场景:产品环境守护进程。

常用命令

启动项目

pm2 start xxx.js   #直接启动入口文件

pm2 start xxx.json #通过配置启动

配置文件参考:

[
  {
    "name": "app",
    "script": "babel.js",
    "log_date_format": "YYYY-MM-DD HH:mm:ss Z",
    "cwd": "/home/project",
    "error_file": "/home/project/logs/app.err.log",
    "out_file": "/home/project/logs/app.out.log",
    "max_memory_restart": "800M",
    "instances": 0,
    "exec_mode": "cluster",
    "merge_logs": true,
    "env": {
      "NODE_ENV": "production"
    }
  },
  {
    "script": "./examples/child.js",
    "error_file": "errLog.log",
    "out_file": "outLog.log",
    "pid_file": "child",
    "instances": "4",
    "min_uptime": "10",
    "max_restarts": "4"
  },
  {
    "script": "examples/env.js",
    "error_file": "errEcho.log",
    "out_file": "outEcho.log",
    "name": "ok",
    "pid_file": "echo.pid",
    "max": "1",
    "exec_mode": "cluster_mode",
    "port": "9001",
    "env_variable": "TOTO",
    "TEST_VARIABLE": "YESSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSIR"
  }
]

重启项目

pm2 restart #PID# #重启一个进程

pm2 restart app-name #重启一个应用

pm2 restart all #重启所有项目

停止

pm2 stop #PID#

pm2 stop app-name

终止

pm2 delete app-name #删除一个进程

pm2 kill #终止所有进程

开机自启

pm2 startup

参考官方文档获取详细使用说明。

自动重启

设置内存使用过高上限,可以通过配置文件这一项:

{
  "max_memory_restart": "800M"
}

PM2是一个很好的工具,提供了max-memory-restart内存溢出重启的功能。

但美中不足的是,没有 CPU 限制重启的功能。

同时,也无法进行远程手动重启。但好在,PM2 提供了 JSON 格式列表输出,让我们可以自己去写一些守护进程,或者整合到已有的运维管理系统中。

pm2 jlist

结果会是这样的 json:

[
  {
    "pid": 28701,
    "name": "xxxx",
    "pm2_env": {
      "name": "xxxx",
      "max_memory_restart": 838860800,
      "instances": 0,
      "cwd": "/xxxx/project",
      "merge_logs": true,
      "vizion": true,
      "pmx": true,
      "automation": true,
      "autorestart": true,
      "treekill": true,
      "env": {
        "PM2_JSON_PROCESSING": "true",
        "LSCOLORS": "Gxfxcxdxbxegedabagacad",
        "LESS": "-R",
        "PAGER": "less",
        "SSH_TTY": "/dev/pts/1",
        "SSH_CLIENT": "218.94.29.190 53115 22",
        "SHELL": "/bin/bash",
        "TERM": "xterm-256color",
        "XDG_SESSION_ID": "814",
        "NODE_ENV": "production",
        "max_memory_restart": 838860800,
        "instances": 0,
        "merge_logs": true,
        "vizion": true,
        "pmx": true,
        "automation": true,
        "autorestart": true,
        "treekill": true,
        "log_date_format": "YYYY-MM-DD HH:mm:ss Z",
        "exec_mode": "cluster_mode",
        "node_args": [],
        "exec_interpreter": "node",
        "pm_out_log_path": "/xxxxx/logs/fish.out.log",
        "pm_err_log_path": "/xxxxx/logs/fish.err.log"
      },
      "log_date_format": "YYYY-MM-DD HH:mm:ss Z",
      "exec_mode": "cluster_mode",
      "node_args": [],
      "pm_exec_path": "/xxxx/babel.js",
      "pm_cwd": "/xxxxx/project",
      "exec_interpreter": "node",
      "pm_out_log_path": "/xxxxxx/xxx.out.log",
      "pm_err_log_path": "/xxxxxx/xxx.err.log",
      "NODE_APP_INSTANCE": 0,
      "vizion_running": false,
      "PM2_JSON_PROCESSING": "true",
      "LESSCLOSE": "/usr/bin/lesspipe %s %s",
      "XDG_RUNTIME_DIR": "/run/user/1000",
      "LESSOPEN": "| /usr/bin/lesspipe %s",
      "LC_CTYPE": "zh_CN.UTF-8",
      "SSH_TTY": "/dev/pts/1",
      "SHELL": "/bin/bash",
      "NODE_ENV": "production",
      "fish": "{}",
      "status": "online",
      "pm_uptime": 1472561001434,
      "axm_actions": [],
      "axm_monitor": {
        "Loop delay": {
          "value": "0.92ms",
          "agg_type": "avg",
          "alert": {}
        }
      },
      "axm_options": {
        "http": false,
        "http_latency": 200,
        "http_code": 500,
        "ignore_routes": [],
        "profiling": true,
        "errors": true,
        "alert_enabled": true,
        "custom_probes": true,
        "network": false,
        "ports": false,
        "module_conf": {},
        "module_name": "fish",
        "module_version": "1.1.3",
        "pmx_version": "0.6.2",
        "error": true
      },
      "axm_dynamic": {},
      "created_at": 1472561000966,
      "pm_id": 3,
      "restart_time": 1,
      "unstable_restarts": 0,
      "_pm2_version": "1.1.3",
      "versioning": {
        "type": "git",
        "update_time": "2016-09-02T07:25:14.113Z",
        "comment": "v3.6.12 - fix-importer",
        "unstaged": false,
        "branch": "HEAD",
        "remotes": ["origin"],
        "remote": "origin",
        "branch_exists_on_remote": true,
        "ahead": false,
        "next_rev": null
      },
      "node_version": "6.3.0",
      "exit_code": 0
    },
    "pm_id": 3,
    "monit": {
      "memory": 90591232,
      "cpu": 0
    }
  }
]

其中每一列数据都有 rows[i].monit.cpu,值范围 0~100。可以用later写定时脚本监控并重启。

示例代码:

import { exec } from 'child_process';
import numeral from 'numeral';

exec('pm2 jlist', (err, stdout, stderr) => {
  if (stderr) {
    console.log(stderr);
  }
  const result = stdout.split('\n');

  // 如果不存在 PM2 进程,将会打初始化日志
  while (result[0].indexOf('[PM2]') === 0) {
    result.shift();
  }
  let items = [];
  try {
    items = JSON.parse(result[0]);
  } catch (e) {
    /* eslint no-empty:0 */
  }
  // console.log(items);

  // 演示
  const item = items[0];

  const uptimeTmp = numeral((new Date() - item.pm2_env.pm_uptime) / 1000)
    .format('00:00:00')
    .split(':')
    .map(Number);
  const uptime = `${parseInt(uptimeTmp[0] / 24, 10)}${uptimeTmp[0] % 24}${uptimeTmp[1]}${uptimeTmp[2]}`
    .replace('0天', '')
    .replace('0时', '')
    .replace('0分', '')
    .replace('0秒', '');

  console.log(item.pid);
  // 41965
  console.log(item.name);
  // app-name
  console.log(item.pm_id);
  // 0
  console.log(`${item.monit.cpu}%`);
  // 91%
  console.log(numeral(item.monit.memory).format('0.0b'));
  // 126.8MB
  console.log(uptime);
  // 6秒
  console.log(item.pm2_env.restart_time);
  // 0
  if (item.monit.cpu > 90) {
    exec(`pm2 restart ${item.pm_id}`, (err2, stdout2, stderr2) => {
      console.log(stdout2);
      console.log(stderr2);
    });
  }
});
在 GitHub 上编辑此页面 article.updatedAt Tue, Jun 29, 2021