shell脚本实现服务器进程监控的方法

  • 写作背景:

项目是基于java、weblogic及timer的处理模式,每次服务部署之后timer的整体状态是一个盲区,因为100多个进程,是否有启动遗漏或者启动重复的,想做到一目了然是件困难的事情。所以我考虑用shell脚本的文件查找与匹配方法来解决这个问题,简单记录下,仅供参考。 

  • 程序说明:

1)通过将系统定义的进程(timer)配置到配置文件process_signal.config中

2)将服务器当前所有的java进程存储到process_current.txt中

3)逐行读取process_signal.config中的进程,到process_current.txt中进行匹配

4)没有匹配到则为启动遗漏,存入到process_invalid.txt中,匹配到1次正常(不做处理),匹配到大于1次,则将timer名称及启动次数记录到process_invalid.txt中。

5)通过定时器30分钟一次执行脚本,使用SQLLOAD将process_invalid.txt加载到数据表中,对数据库表进行监控。

  •  脚本说明:

1、整体文件结构:

webusr@iomtimer1:/weblogic/script/process_signal >ls
process_current.txt    process_invalid.txt    process_invalid_his.txt    process_signal.config    process_signal.sh

2、进程配置示例:

webusr@iomtimer1:/weblogic/script/process_signal >cat process_signal.config 
#停复发NAS
extapp.tftonas.PutTfToNas 10 2
extapp.tftonas.PutTfToNas 11 2
extapp.tftonas.PutTfToNas 12 3

3、执行结果记录:

webusr@iomtimer1:/weblogic/script/process_signal >cat process_invalid.txt 
127.0.0.1 process [extapp.leaseroute.PutToLeaseRoute 180] is not exist 2018-04-11 08:29:00
127.0.0.1 process [ DealWorkTaskTimer 3001 ] is repeatRun 2 Times 2018-04-11 17:51:56

4、逻辑匹配入口:

webusr@iomtimer1:/weblogic/script/process_signal >cat process_signal.sh
#!/bin/sh
path=/weblogic/script/process_signal #脚本路径
file_config=process_signal.config #配置文件名
file_result=process_invalid.txt #错误进程信息
file_tmp=process_current.txt #当前进程信息(全量,临时)
if_get_exist="N" #是否输出日志(N不输出,Y输出)
hostip="127.0.0.1" #主机IP
currenttime=`date "+%Y-%m-%d %H:%M:%S"` #当前系统时间
#定义方法get_process_signal()
get_process_signal(){
  # 转储上次记录到历时,追加 
  cat /weblogic/script/process_signal/process_invalid.txt >> process_invalid_his.txt 
  #清空错误进程信息,当前进程信息
  >/weblogic/script/process_signal/process_invalid.txt
  >/weblogic/script/process_signal/process_current.txt
  #输出当前时间并获取主机上所有java类进程
  echo "######################### get ps info time is: " $currenttime " #########################">>/weblogic/script/process_signal/process_current.txt
  ps -ef|grep java >>/weblogic/script/process_signal/process_current.txt
  echo "######################### get ps info time is: " $currenttime " #########################">>/weblogic/script/process_signal/process_current.txt
  #逐条读取配置文件process_signal.config
  cat /weblogic/script/process_signal/process_signal.config | while read LINE 
  do
    # 如果配置文件含有#号,则跳过不处理 
    if [[ `echo $LINE|cut -c1-1` != "#" ]]; then
      #如果配置文件记录在主机进程中存在
      if [[ -n `find $path -name $file_tmp | xargs grep -w "$LINE"` ]]; then
        #开关为Y,输出到屏幕
        if [[ $if_get_exist = "Y" ]]; then
          echo $hostip " process [" $LINE "] is exist " $currenttime >>/weblogic/script/process_signal/process_invalid.txt
        fi 
        #配置文件记录与主机进程匹配
        run=`find $path -name $file_tmp | xargs grep -w "$LINE"|wc -l`
        if (($run>1)); then
          #打印配置文件中记录与主机进程匹配数大于1的记录
          echo $hostip " process [" $LINE "] is repeatRun "$run " Times" $currenttime >>/weblogic/script/process_signal/process_invalid.txt
        fi
        #如果配置文件记录在主机进程中不存在
      else
        echo $hostip " process ["$LINE"] is not exist " $currenttime >>/weblogic/script/process_signal/process_invalid.txt
      fi 
    fi 
  done
}
#调用方法
get_process_signal
#fhs.workflow.EventMonitorMain 4
#find /weblogic/script/process_signal -name process_current.txt | xargs grep -i "fhs.workflow.EventMonitorMain 4"
#ps -ef|grep java|grep "fhs.workflow.EventMonitorMain 4"

5、SqlLOAD入库(含控制文件conl.ctl)

efile92@iomitfdb2:/interface/contabfile/monitor_time >cat conl.ctl 
load data 
infile '/interface/contabfile/monitor_time/process_invalid.txt' 
truncate into table T_MONITOR_TIMER
fields terminated by '\n'
-- Create table
create table T_MONITOR_TIMER
(
  info VARCHAR2(1000)
)
sqlldr userid=用户名/密码@sid control=/interface/contabfile/monitor_time/conl.ctl log=/interface/contabfile/monitor_time/conl.log

总结

相关推荐