需求：做一个小工具，用来检测某一个程序的异常崩溃，cpu专用率，内存使用率，线程数，线程是否阻塞问题，当出现上述异常时，将日志上传服务器，方便查找问题。

这样一个小程序实现起来并不太难，但是我一点python都不会，可真是。。。时间只有一个星期。所以用一天的时间看了python的语法，第二天开始入手写一些小函数，第三天写业务逻辑，第四天调bug，第五天将这个小程序融入项目。终于实现了功能，但是代码却并不是什么高质量的代码，只是匆忙堆砌起来的。下面记录一下。

在这个代码中使用到了paramiko库，需要自行下载编译库，使用paramiko库的目的是为了实现使用sftp进行文件的上传服务器。

具体操作可以参考：

其中遇到一个SFTP IOError的错误，查了半天是我的iP输入错了，在这里的操作一定要捕获异常，不然程序会奔溃，在python里面，捕获异常尤其重要，不然程序真的会在你意想不到的地方，随意奔溃。

程序运行平台为centos7.5

主程序如下：

#!/usr/bin/python

#-*- coding:UTF-8 -*-

#文件名:autoCheck.py

import commands

import time

import datetime

import os

import paramiko

import pdb

import signal

import platform

LogNameList=('httpdkey','message','info','warning')

InstallFileFlag='/etc/tlx/config'

InstallStartFlag=0

def getProcessInfo(processName):

'''input:进程名称

output:某一进程的pid，cpu占用率，mem占用率'''

cmd = 'ps aux | grep %s | grep -v grep |\

grep -v "sudo" | awk \'{print $2,$3,$4}\''%(processName)

(status,output) = commands.getstatusoutput(cmd)

print status,output

if status == 0 and len(output) > 0:

return output;

else:

return None;

def getProcessMem(inputData):

inputList = inputData.split(' ')

mem = inputList[2]

memVal = float(mem)

return memVal;

def getProcessPid(inputData):

inputList = inputData.split(' ')

pid = inputList[0]

# pidVal = float(pid)

return pid;

def getProcessCpu(inputData):

inputList = inputData.split(' ')

cpu = inputList[1]

cpuVal = float(cpu)

return cpuVal;

def getProcessThreadNum(pidNum):

cmd = 'ps -T -p %s' %(pidNum)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

outputList = output.split('\n')

threadNum = len(outputList)

return threadNum;

else:

return 0;

def getProcessOpenFdNum(fileName):

cmd = 'lsof -c %s' %(fileName)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

outputList = output.split('\n')

threadNum = len(outputList)

return threadNum;

else:

return 0;

def getProcessOpenSocketNum(socketName):

cmd = 'netstat -nltp | grep %s' %(socketName)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

outputList = output.split('\n')

threadNum = len(outputList)

return threadNum;

else:

return 0;

def checkPortOccupy(portNum):

cmd = 'netstat -nltp | grep %s' %(portNum)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

return 1;

else:

return 0;

def getLogUpdateTime():

global LogNameList

cmd = 'ls -l /var/log/tlx'

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

outputList = output.split('\n')

counts = len(outputList)

for index in range(len(outputList)):

dataLine = outputList[index]

print 'dataLine:%s'%(dataLine)

fileTime = getFileModifirendTime(dataLine)

if fileTime == None:

continue;

currTime = getCurrentTime()

if compareTimeStr(currTime,fileTime) == 1:

for index in range(len(LogNameList)):

if LogNameList[index] is dataLine.split()[-1]:

print 'log:%s'%(LogNameList[index])

return LogNameList[index];

else:

return None;

def getFileModifirendTime(fileInfo):

'''使用ll命令得到如下格式命令：

-rwxr-xr-x. 1 root root 3888 Mar 20 10:31 autoCheck.py'''

cmd = 'echo ' +fileInfo+ ' | awk \'{print $8}\''

(status,output) = commands.getstatusoutput(cmd)

#print 'status:%s,output:%s,len:%s'%(status,output,len(output))

if (status == 0) and (len(output) > 0):

return output;

else:

return None;

def getCurrentTime():

'得到如下格式的时间：Sat Mar 28 22:24:24 2016'

time_tmp = time.strftime('%a %b %d %H:%M:%S %Y', time.localtime())

return time_tmp;

def compareTimeStr(currTime,fileTime):

'将getFileModifirendTime函数得到的时间拼装完整，进行时间比较'

timestr1_tmp = time.mktime(time.strptime(currTime,\

"%a %b %d %H:%M:%S %Y"))

curTimeQ = currTime.split(" ")

fileTimeQ = curTimeQ[0] + ' ' + curTimeQ[1] + ' ' +\

curTimeQ[2] + ' ' + fileTime + ':00 ' + curTimeQ[4]

timestr2_tmp = time.mktime(time.strptime(fileTimeQ,\

"%a %b %d %H:%M:%S %Y"))

if (timestr1_tmp - timestr2_tmp) > 300:

print 'timestr1_curr:%d > timestr2_file:%d'%(timestr1_tmp,timestr2_tmp)

return 1;

else:

return 0;

def recordCmdInfo(processName):

fdData=''

cmd = 'ps aux | grep %s | grep -v grep'%(processName)

(status,output) = commands.getstatusoutput(cmd)

print status,output

if status == 0 and len(output) > 0:

fdData += output

fdData += '\r'

cmd = 'ps -T -p %s' %(pidNum)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

fdData += output

fdData += '\r'

cmd = 'lsof -c %s' %(fileName)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

fdData += output

fdData += '\r'

fd = open('/var/log/tlx/tlx_syslog','a+')

fd.write(fdData)

fd.close()

def getServerIp(ipFile):

fd = open(ipFile,'r')

fdData = fd.readline()

ipList = fdData.split('=')

ip = ipList[1].rstrip()

return ip;

def sftpUploadFile(localFileName,serverFileName):

global InstallFileFlag

t = paramiko.Transport((getServerIp(InstallFileFlag),22))

t.connect(username='admin',password='admin123')

sftp = paramiko.SFTPClient.from_transport(t)

try:

sftp.put(localFileName,serverFileName)

except IOError:

t.close()

else:

t.close()

def packageLog():

exeFlag = 0

packageName = 'tlx_' + time.strftime('%m_%d_%H_%M',time.localtime()) + '.tar.gz'

localFile = '/var/log/%s'%(packageName)

serverFile = '/srv/ftp/%s'%(packageName)

print 'localFile:%s,serverFile:%s'%(localFile,serverFile)

cmd = 'tar -zcvf %s /var/log/tlx'%(localFile)

(status,output) = commands.getstatusoutput(cmd)

if status == 0 and len(output) > 0:

time.sleep(1)

sftpUploadFile(localFile,serverFile)

exeFlag = 1

restartServer()

def restartServer():

processData = getProcessInfo('process')

if processData == None:

while checkPortOccupy('10001'):

print '10001 port occupy'

cmd = '/opt/tlx/process/process'

(status,output) = commands.getstatusoutput(cmd)

if status == 0:

print 'process restart'

else:

print 'process restart failed'

else:

processPid = getProcessPid(processData)

cmd = 'kill -9 %s' %(processPid)

(status,output) = commands.getstatusoutput(cmd)

if status == 0:

print 'process stop'

while checkPortOccupy('10001'):

print '10001 port occupy'

time.sleep(2)

cmd = '/opt/tlx/process/process'

(status,output) = commands.getstatusoutput(cmd)

if status == 0:

print 'process restart'

else:

print 'process restart failed'

def shutdownFunct(signalnum,frame):

pass

def signalCapture():

global InstallStartFlag

try:

for sig in [signal.SIGINT,signal.SIGHUP,signal.SIGTERM,signal.SIGKILL]:

signal.signal(sig,shutdownFunct)

InstallStartFlag = 0

continue;

except RuntimeError:

return;

InstallStartFlag = 0

print '将文件，相关数据打包，上传'

packageLog()

def enviromentCheck():

cmd = './environment.sh'

(status,output) = commands.getstatusoutput(cmd)

if status != 0:

print '环境安装错误，请检查paramiko-1.7.5 pycrypto-2.6模块的安装'

exit()

def sysDistinguish():

sysInfo = platform.platform()

# if -1 != sysInfo.find('Ubuntu'):

# else if -1 != sysInfo.find('CentOS'):

def running():

print 'process start...'

global InstallFileFlag

global InstallStartFlag

# enviromentCheck()

while 1:

time.sleep(1)

errorFlag = 0

if not os.access(InstallFileFlag,os.F_OK):

continue;

processData = getProcessInfo('process')

if processData == None:

if InstallStartFlag == 0:

continue;

else:

signalCapture()

continue;

InstallStartFlag = 1

if getProcessCpu(processData) > 50:

errorFlag = 1

if getProcessMem(processData) > 10:

errorFlag = 1

processPid = getProcessPid(processData)

if getProcessThreadNum(processPid) > 10:

errorFlag = 1

if getProcessOpenFdNum('process') > 100:

errorFlag = 1

logFileName = getLogUpdateTime()

if logFileName:

errorFlag = 1

if errorFlag == 1:

'''出现错误的话，则将上述几项内容进行记录，

然后将相关日志进行打包，上传服务器，将程序杀死，再启动'''

packageLog()

#running()

#pdb.run('running()')

if __name__ == "__main__":

running()

在编写程序过程中，需要调试程序，所以又查看了python的程序调试方法，比较好用的就是pdb了，常用的是打日志进行调试，在本段代码中，没有写日志的接口，需要后续进行添加。

在linux中使用过gdb的朋友，使用pdb调试起来还是很顺手的。而且我的编译器使用的是vim。

代码调试好，就是对python程序进行打包。网上说了很对打包的原因，我反正使用不到，我打包就是为了解决依赖，然后和现有的程序独立开来，这样才能监控的另一个程序，不会收到干扰。打包使用的pyinstaller工具，安装这个工具的时候，有以下几个步骤：

1.安装python-pip

2.安装：pyinstaller

打包成功后，发现运行打包文件，会出现两个同名的进程，pid是不一样的，很是不解，这让使用linux的不可接受，一直以为是程序哪里出现问题了，使用python  XXX执行程序就是正常的一个进程。经过一番查找，终于获得解释，释然

到此为止，整个小工具初步实现功能，算是初步完成。但是和我的预想是不一样的，之后我会再改进一版，将其做成接口类，加入log日志，方便调用排错，增加其通用性。

敬请期待。。。