用python实现vmware esxi虚拟机断电自动关机

一、前言

机房服务器一般配备UPS，当市电停掉时，仍然到维持一段时间。但是，如果长时间停电，UPS也撑不了。此时就需要及时关机以确保数据安全。很多年以前用过winpower软件，通过目标系统安装agent，并用winPower连接UPS，当UPS电量低于设置的阈值时，给各个agent下发关机指令，以实现自动关机。这种方式可以充分利用UPS电量，较为精准的判断电池续航情况，并采取相应动作，但是操作起来有点麻烦。

另一种方式是采用自动化脚本的方式，通过循环ping 某个常接市电设备ip，当失败超过一定次数，触发powerOffVMs，最后shutdown esxi系统。这种方式无需与UPS通讯，只要市电断开，超过一段时间就执行关机指令，超作起来比较方便。但是，这种方式可能存在误判，比如交换机挂了，此时ping设备必然失败，最终会触发关机。因此下面介绍一种更为稳妥的方案。

二、实现

1. 原理

   利用esxcli 来esxcli hardware ipmi sdr list | grep "Power\ Supply\ 1\ Status\|Power\ Supply\ 2\ Status"来检测电源是否正常。当其中一路电源断开或异常时，调用powerOffVms.py关闭各虚拟机，最后再调用poweroff 来关闭esxi。

   其实还可以利用snmp连接到idrac，也是通过检查电源状态来实现自动监控，但是这种方式又依赖网络了，跟上面提到的方式二差不多。

   

==注意，这里默认服务器都具备双路电源，如果只有一路电源，那不适用此方法。==

2. 代码

   源码如下（有点啰嗦，几年前写的，将就看吧），racadm是为了兼容R710，因为R710的esxcli 指令不支持获取电源状态。执行python脚本后，会修改etc/rc.local.d/local.sh以实现开机自动执行脚本。本例中powerOffDelay是指断电20分钟后就触发自动关机指令。但是当断电期间，又来电时，会重新计时。

   本代码在R710至R750上测试通过，不代表你的也可以，仅作参考 ，请根据自己实际情况决定是否使用。

   #!/usr/bin/python
   from genericpath import exists, isfile
   import os
   import time
   import subprocess
   import fileinput
   import re
   import sys
   # version 20190916  usage: add python autoShutdown.py to  /etc/rc.local.d/local.sh   and uncomment  
   # by root6819 Q:302777528 site: www.qipanet.com
   # tools Path
   basePath = os.path.split(os.path.realpath(__file__))[0]
   # the time(second) when you want to shutdown while check power False
   fPowerOff = basePath + '/powerOffVms_py3.0.py'
   # Racadm-Dell-EMC-Web-9.3.0-3379.VIB-ESX67i.zip
   fRacadm = basePath+'/Racadm-Dell-EMC-Web-9.3.0-3379.VIB-ESX67i.zip'
   fLocalOld = '/etc/rc.local.d/local.sh'
   fLocalNew = basePath+'/local.sh.bak'
   interval = 15  # must >=5s,everty check will cost some time.  per 15s trigger a check
   powerOffDelay = 1200  # powerOff after supply off in   20Minutes
   iFailTimes = 0  # +1 when check Power Off,if ok then reset iFailTimes=0 , if iFailTimes*interval>powerOffDelay then powerOff
   machine ="" # use to check if it is R710 machine
   def getMachine():
       cmd='localcli hardware platform get |grep R710'
       machine=os.popen(cmd).read().strip()
       showMsg("the machine is R710 ?"+ machine)
   
   def modify_file(file_name, pattern, value=""):
       writeLog('now modify file '+file_name+'...')
       try:
           fh = fileinput.input(file_name, inplace=True)
           for line in fh:
               replacement = value + line
               line = re.sub(pattern, replacement, line)
               sys.stdout.write(line)
           fh.close()
           writeLog('write Ok')
           return True
       except Exception as e:
           writeLog(e)
           return False
   
   
   def myPopen(cmd):
       p = subprocess.Popen(
           cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
       ret, err = p.communicate()
       return ret.decode('utf-8', 'ignore'), err.decode('utf-8', 'ignore')
   
   
   def writeLog(msg):
       path = basePath+"/Log"
       if not os.path.exists(path):
           os.makedirs(path)
       showMsg(msg)
       f = open(path+"/log.txt", mode='a+')
       strMsg = '[{0}]: {1} \n'.format(time.strftime('%Y-%m-%d %H:%M:%S'), msg)
   
       f.writelines(strMsg)
       f.close()
   
   
   def showMsg(msg):
       strMsg = '[{0}]: {1} \n'.format(time.strftime('%Y-%m-%d %H:%M:%S'), msg)
       print(strMsg)
   
   
   def checkLocalSh():
       try:
           # check local.sh is config right
           with open(fLocalOld, mode='r+') as f:
               tmpStr = f.read()
               # python /vmfs/volumes/datastore_ssd/tools/autoShutdown.py &
               if not 'autoShutdown3.0.py &' in tmpStr:
                   return False
               return True
       except Exception as e:
           writeLog('checkLocalShErr>>')
           writeLog(e)
           return False
   
   
   def shutDownAll():
       cmd = fPowerOff
       # esxi6.7 powerOffVms use old version script,so we must reWrite the script
       writeLog('powerOffVms path>>%s ' % cmd)
       result = os.popen(cmd).read()
       writeLog('powerOffVms result>>%s ' % result)
       writeLog('now poweroff')
       os.popen('poweroff')
   
   
   def checkPS():
       
       #check is R710 
       if 'R710' in machine:
           cmd = 'racadm getsensorinfo |grep  ^PS.*AC |grep -c Present'
           result = os.popen(cmd).read().strip()
           if result == '2':
               #showMsg('checkPs OK')
               return True
           else:
               #showMsg(str(iFailTimes)+'times checkPs not OK>>'+result)
               return False
       else:
           #Power Supply AC lost or Presence detected  
           cmd = 'localcli hardware  ipmi sdr list | grep "Power\ Supply\ 1\ Status\|Power\ Supply\ 2\ Status"'
           result = os.popen(cmd).read().strip().lower()
           if not 'detected' in result:
               #must have at lease one supply detected in result
               showMsg(str(iFailTimes)+'checkPS() result null,check again...')
               return checkPS()
           if 'lost' in result:
               showMsg(str(iFailTimes)+'times checkPs not OK>>'+result)
               return False
           else:
               return True
   
   def checkRacadm():
       # check racadm support
       _cmd = 'racadm version |grep  -w version'
       _result, _err = myPopen(_cmd)
   
       if 'not found' in _err or '不是内部或外部命令' in _err:
           writeLog('racadm tool not install,pls install it first!')
           return False
       writeLog('racadm check oK,the Version is>>'+_result)
       return True
   
   
   if not checkRacadm():
       '''check FirstTime if not exists auto install '''
       if not os.path.exists(fRacadm):
           writeLog(fRacadm + ' file not exists,could not auto instal!')
           exit()
       writeLog(fRacadm + ' try auto install...')
       _result, _err = myPopen('esxcli software vib install -d '+fRacadm)
       writeLog('install result>>'+_result)
       if _err or _err is not '':
           writeLog('install _err>>'+_err)
   
       # check again
       if not checkRacadm():
           exit()
   
   if not checkLocalSh():
       # bak then goto write
       _result, _err = myPopen('cp -f '+fLocalOld+' ' + fLocalNew)
       writeLog('cp Result>>'+_result)
       if _err or _err is not '':
           writeLog('cp err>>'+_err)
           exit()
       _str = 'python '+basePath+'/autoShutdown3.0.py &\n'
       if not modify_file(fLocalOld, 'exit 0', _str):
           exit()
       # check again
       if not checkLocalSh():
           writeLog('checkLocalSh twice failed,exit now!')
           exit()
   
   writeLog('checkLocalSh Ok!')
   
   _fPath = basePath + '/powerOffVms_py3.0.py'
   if not os.path.exists(_fPath):
       writeLog(_fPath+' file not exists,pls confirm！')
       exit()
   writeLog(_fPath+' check file OK!')
   
   getMachine()
   isFirst = True
   while iFailTimes*interval < powerOffDelay:
       if checkPS():
           iFailTimes = 0
           if isFirst:
               isFirst = False
               writeLog('checkPS OK')
       else:
           isFirst = True
           iFailTimes += 1
           if iFailTimes == 2:
               writeLog('checkPS failed,maybe a power supply is lost!!!')
    
       # because checkPS will costs about 3 seconds
       time.sleep(interval-3)
   
   shutDownAll()