def __init__(self, ldap, product, type, key, command, alarmers, logger,
cubeName, debug, timeInterval):
self.key = key
self.timeInterval = timeInterval
self.command = command
self.alarmers = alarmers
self.dbop = DBOperator(ldap, product, type, cubeName, [key], [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
def startDownloader(lq):
from dbOperator import DBOperator
from baseconfig import defaultconfig
conf = defaultconfig()
dbo = DBOperator(config=conf)
while True:
getSize = conf.DOWNLOADER_QUEUE_SIZE - lq.qsize()
if getSize > 0:
urls = dbo.getURL(maxdataLength=conf.DOWNLOADER_QUEUE_SIZE)
print 'Process #%s Get %s URLS FOR Request' % (str(
os.getpid()), str(len(urls)))
for i in urls:
lq.put(urls)
else:
print 'Reach Max Queue Size Wait...'
time.sleep(15)
def __init__(self,ldap,product,type,key,command,alarmers,logger,cubeName,debug,timeInterval):
self.key=key
self.timeInterval=timeInterval
self.command=command
self.alarmers=alarmers
self.dbop=DBOperator(ldap,product,type,cubeName,[key],[],[])
self.logger=logger
self.cubeName=cubeName
self.debug=debug
def __init__(self, ldap, product, type, file, timeInterval, numberKeys,
stringKeys, statisticTool, wholeFile, keyAlarmers, debug,
logger, cubeName):
self.filePattern = file
self.curFile = file
self.timePattern = self.parseTimePattern(file)
self.timeInterval = timeInterval
self.numberKeys = numberKeys
self.stringKeys = stringKeys
self.statisticTool = statisticTool
self.wholeFile = wholeFile
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName, numberKeys,
stringKeys, [])
self.debug = debug
self.logger = logger
self.cubeName = cubeName
#
self.position = 0
self.preFileSize = 0
self.defaultReadSize = 16 * 1024 * 1024
def __init__(self, ldap, product, type, alias, regPattern, processPath,
pids, keys, keyAlarmers, logger, cubeName, debug,
timeInterval):
self.alias = alias
self.regPattern = regPattern
self.processPath = processPath
self.pids = pids
self.oldPids = []
self.keys = keys
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName,
self.getDBKeys(alias, keys), [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
self.timeInterval = timeInterval
# total mem of machine, unit: mb
self.totalMem = 0
self.curUser = None
self.topN = 12
# 用于指向获取进程pid的函数
self.getAllPid = None
# 若不是直接指定pid, 则每次都调用系统命令, 用于check进程正常 -> 优化: 先尝试用上次的pid, 找不到后再通过regPattern和processPath去更新pid
self.currentPidCmdPath = []
def __init__(self,ldap,product,type,file,timeInterval,numberKeys,stringKeys,statisticTool,wholeFile,keyAlarmers,debug,logger,cubeName):
self.filePattern=file
self.curFile=file
self.timePattern=self.parseTimePattern(file)
self.timeInterval=timeInterval
self.numberKeys=numberKeys
self.stringKeys=stringKeys
self.statisticTool=statisticTool
self.wholeFile=wholeFile
self.keyAlarmers=keyAlarmers
self.dbop=DBOperator(ldap,product,type,cubeName,numberKeys,stringKeys,[])
self.debug=debug
self.logger=logger
self.cubeName=cubeName
#
self.position=0
self.preFileSize=0
self.defaultReadSize= 16*1024*1024
def __init__(
self,
ldap,
product,
type,
alias,
regPattern,
processPath,
pids,
keys,
keyAlarmers,
logger,
cubeName,
debug,
timeInterval,
):
self.alias = alias
self.regPattern = regPattern
self.processPath = processPath
self.pids = pids
self.oldPids = []
self.keys = keys
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName, self.getDBKeys(alias, keys), [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
self.timeInterval = timeInterval
# total mem of machine, unit: mb
self.totalMem = 0
self.curUser = None
self.topN = 12
# 用于指向获取进程pid的函数
self.getAllPid = None
# 若不是直接指定pid, 则每次都调用系统命令, 用于check进程正常 -> 优化: 先尝试用上次的pid, 找不到后再通过regPattern和processPath去更新pid
self.currentPidCmdPath = []
class LogFile:
def __init__(self, ldap, product, type, file, timeInterval, numberKeys,
stringKeys, statisticTool, wholeFile, keyAlarmers, debug,
logger, cubeName):
self.filePattern = file
self.curFile = file
self.timePattern = self.parseTimePattern(file)
self.timeInterval = timeInterval
self.numberKeys = numberKeys
self.stringKeys = stringKeys
self.statisticTool = statisticTool
self.wholeFile = wholeFile
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName, numberKeys,
stringKeys, [])
self.debug = debug
self.logger = logger
self.cubeName = cubeName
#
self.position = 0
self.preFileSize = 0
self.defaultReadSize = 16 * 1024 * 1024
def validate(self):
if self.filePattern == '':
self.logger.warning('%s: 未配置文件名, 不对日志做收集' % self.cubeName)
self.dbop.close()
return False
if len(self.numberKeys) + len(self.stringKeys) == 0:
self.logger.warning('%s: fileName=%s, 未配置监控keys, 不对日志做收集' %
(self.cubeName, self.filePattern))
return True
def toString(self):
str = ''
str += 'fileName=%s\n' % self.filePattern
str += 'timeInterval=%s\n' % self.timeInterval
str += 'numberKeys=%s\n' % ','.join(self.numberKeys)
str += 'stringKeys=%s\n' % ','.join(self.stringKeys)
str += 'wholeFile=%s\n' % self.wholeFile
str += 'keyAlarmers='
for key, alarmers in self.keyAlarmers.items():
str += '%s:[' % key
str += ','.join([alarmer.toString() for alarmer in alarmers])
str += '],'
str = str[0:-1] + '\n'
return str
def parseTimePattern(self, file):
timePattern = re.match(".*@@(.*)@@", self.filePattern)
if timePattern:
return timePattern.group(1)
return None
def updateLogFile(self):
'''
更新文件名称和读取位置
'''
if self.timePattern:
curTimePattern = datetime.datetime.strftime(
datetime.datetime.now(), self.timePattern)
logFile = re.sub("@@.*@@", curTimePattern, self.filePattern)
if self.curFile != logFile:
# 1. 文件名变化
self.logger.info('@@logHandler@@ file changed: %s -> %s' %
(self.curFile, logFile))
self.curFile = logFile
self.position = 0
self.preFileSize = 0
return
if os.path.exists(self.curFile):
curFileSize = os.stat(self.curFile).st_size
if self.preFileSize <= curFileSize:
# 2. 文件名无变化 & 文件名无变化 & 文件大小合理变化
return
# 3. 文件名无变化 & (文件不存在 | 文件大小不合理变化)
self.position = 0
self.preFileSize = 0
def monitor(self):
self.updateLogFile()
# 若文件存在, 且不需要分析整个log, 则指向末尾, 否则指向开头并等待文件出现
if os.path.exists(self.curFile) and not self.wholeFile:
lf = open(self.curFile)
lf.seek(-1, os.SEEK_END)
self.position = lf.tell()
lf.close()
# 开始监控logfile
while True:
# update file name
self.updateLogFile()
if not os.path.exists(self.curFile):
time.sleep(5)
self.logger.warning(
'%s: @@logHandler@@ file=%s, 待监控文件不存在, 暂不做监控, 等待文件出现ing...'
% (self.cubeName, self.curFile))
continue
self.log2db()
self.logger.info(
'%s: @@logHandler@@ file=%s, sleeping, will wake up after %s seconds'
% (self.cubeName, self.curFile, self.timeInterval))
time.sleep(self.timeInterval)
# 若单行log大于defaultReadSize, 将会出bug: 无法往下继续读, 因此defaultReadSize不宜过小, 但从资源消耗角度看, 也不宜过大
# 每次均读至末尾
def log2db(self):
lf = open(self.curFile)
# 先记住当前位置, 避免由于日志输出过快而每次处理完日志后又有新日志输出
lf.seek(-1, os.SEEK_END)
currentEndPosition = lf.tell()
lf.seek(self.position)
while True:
if currentEndPosition <= lf.tell():
break
rawLines = lf.read(self.defaultReadSize)
index = rawLines.rfind('\n')
if index < 0:
break
lf.seek(index + 1 - len(rawLines), os.SEEK_CUR)
logLines = rawLines[0:index].splitlines()
result = self.statisticTool.getStatisticData(logLines)
self.alarm(result)
if len(result) > 0:
if not self.debug:
self.dbop.insert(result)
if not self.debug and len(result) > 20:
self.logger.info(
'@@logHandler@@ file=%s, insert.status=success, data.length=%s, data=%s, ..., %s'
% (self.curFile, len(result), result[0:10],
result[-10:]))
else:
self.logger.info(
'@@logHandler@@ file=%s, insert.status=success, data.length=%s, data=%s'
% (self.curFile, len(result), result))
else:
self.logger.info(
'@@logHandler@@ file=%s, insert.status=empty' %
self.curFile)
self.position = lf.tell()
self.preFileSize = os.stat(self.curFile).st_size
lf.close()
def alarm(self, result):
for res in result:
if self.keyAlarmers.has_key(res[1]):
for alarmer in self.keyAlarmers[res[1]]:
alarmer.alarm(res[2], self.cubeName, res[1])
class CommandKey:
def __init__(self, ldap, product, type, key, command, alarmers, logger,
cubeName, debug, timeInterval):
self.key = key
self.timeInterval = timeInterval
self.command = command
self.alarmers = alarmers
self.dbop = DBOperator(ldap, product, type, cubeName, [key], [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
def validate(self):
if self.key == '':
logger.warning('%s: commandKey未配置key, 不做command的收集')
self.dbop.close()
return False
if self.command == '':
logger.warning('%s: commandKey未配置command, 不做command的收集')
self.dbop.close()
return False
return True
def toString(self):
str = ''
str += 'key=%s\n' % self.key
str += 'timeInterval=%s\n' % self.timeInterval
str += 'command=%s\n' % self.command
str += 'alarmers=[%s]\n' % ','.join(
[alarmer.toString() for alarmer in self.alarmers])
return str
def monitor(self):
while True:
currentTime = int(time.time())
currentTime = currentTime - currentTime % self.timeInterval
value = self.getDataFromCommand()
if value:
# add timestamp
insertData = [[currentTime, self.key, value]]
self.logger.info('%s: @@commandKeyHandler@@ result=%s' %
(self.cubeName, insertData))
self.alarm(insertData)
if not self.debug:
self.dbop.insert(insertData)
self.logger.info(
'%s: @@commandKeyHandler@@ result inserted to database' %
self.cubeName)
else:
self.logger.warning('%s: @@commandKeyHandler@@ no data get' %
self.cubeName)
endTime = int(time.time())
sleepTime = self.timeInterval - (endTime - currentTime)
if sleepTime <= 0:
continue
self.logger.info(
'%s: @@commandKeyHandler@@ sleep, will wake up in %ss' %
(self.cubeName, sleepTime))
time.sleep(sleepTime)
def alarm(self, insertData):
for data in insertData:
for alarmer in self.alarmers:
alarmer.alarm(data[2], self.cubeName, self.key)
def getDataFromCommand(self):
result = commands.getoutput(self.command).splitlines()
result = result[0].split()
result = result[0]
try:
return float(result)
except:
return None
class LogFile:
def __init__(self,ldap,product,type,file,timeInterval,numberKeys,stringKeys,statisticTool,wholeFile,keyAlarmers,debug,logger,cubeName):
self.filePattern=file
self.curFile=file
self.timePattern=self.parseTimePattern(file)
self.timeInterval=timeInterval
self.numberKeys=numberKeys
self.stringKeys=stringKeys
self.statisticTool=statisticTool
self.wholeFile=wholeFile
self.keyAlarmers=keyAlarmers
self.dbop=DBOperator(ldap,product,type,cubeName,numberKeys,stringKeys,[])
self.debug=debug
self.logger=logger
self.cubeName=cubeName
#
self.position=0
self.preFileSize=0
self.defaultReadSize= 16*1024*1024
def validate(self):
if self.filePattern=='':
self.logger.warning('%s: 未配置文件名, 不对日志做收集' % self.cubeName)
self.dbop.close()
return False
if len(self.numberKeys)+len(self.stringKeys)==0:
self.logger.warning('%s: fileName=%s, 未配置监控keys, 不对日志做收集' % (self.cubeName,self.filePattern))
return True
def toString(self):
str=''
str+='fileName=%s\n' % self.filePattern
str+='timeInterval=%s\n' % self.timeInterval
str+='numberKeys=%s\n' % ','.join(self.numberKeys)
str+='stringKeys=%s\n' % ','.join(self.stringKeys)
str+='wholeFile=%s\n' % self.wholeFile
str+='keyAlarmers='
for key,alarmers in self.keyAlarmers.items():
str+='%s:[' % key
str+=','.join([alarmer.toString() for alarmer in alarmers])
str+='],'
str=str[0:-1]+'\n'
return str
def parseTimePattern(self,file):
timePattern=re.match(".*@@(.*)@@",self.filePattern)
if timePattern:
return timePattern.group(1)
return None
def updateLogFile(self):
'''
更新文件名称和读取位置
'''
if self.timePattern:
curTimePattern = datetime.datetime.strftime(datetime.datetime.now(),self.timePattern)
logFile = re.sub( "@@.*@@", curTimePattern , self.filePattern )
if self.curFile != logFile:
# 1. 文件名变化
self.logger.info('@@logHandler@@ file changed: %s -> %s' % (self.curFile, logFile))
self.curFile=logFile
self.position=0
self.preFileSize=0
return
if os.path.exists(self.curFile):
curFileSize = os.stat(self.curFile).st_size
if self.preFileSize <= curFileSize:
# 2. 文件名无变化 & 文件名无变化 & 文件大小合理变化
return
# 3. 文件名无变化 & (文件不存在 | 文件大小不合理变化)
self.position=0
self.preFileSize=0
def monitor(self):
self.updateLogFile()
# 若文件存在, 且不需要分析整个log, 则指向末尾, 否则指向开头并等待文件出现
if os.path.exists(self.curFile) and not self.wholeFile:
lf=open(self.curFile)
lf.seek( -1, os.SEEK_END)
self.position = lf.tell()
lf.close()
# 开始监控logfile
while True:
# update file name
self.updateLogFile()
if not os.path.exists(self.curFile):
time.sleep(5)
self.logger.warning( '%s: @@logHandler@@ file=%s, 待监控文件不存在, 暂不做监控, 等待文件出现ing...' % (self.cubeName, self.curFile) )
continue
self.log2db()
self.logger.info( '%s: @@logHandler@@ file=%s, sleeping, will wake up after %s seconds' % ( self.cubeName, self.curFile, self.timeInterval ) )
time.sleep( self.timeInterval )
# 若单行log大于defaultReadSize, 将会出bug: 无法往下继续读, 因此defaultReadSize不宜过小, 但从资源消耗角度看, 也不宜过大
# 每次均读至末尾
def log2db( self ):
lf = open( self.curFile )
# 先记住当前位置, 避免由于日志输出过快而每次处理完日志后又有新日志输出
lf.seek( -1, os.SEEK_END)
currentEndPosition=lf.tell()
lf.seek( self.position )
while True:
if currentEndPosition<=lf.tell():
break
rawLines = lf.read( self.defaultReadSize )
index = rawLines.rfind( '\n' )
if index < 0:
break
lf.seek( index + 1 - len( rawLines ), os.SEEK_CUR )
logLines = rawLines[0:index].splitlines()
result = self.statisticTool.getStatisticData( logLines )
self.alarm(result)
if len( result ) > 0:
if not self.debug:
self.dbop.insert( result )
if not self.debug and len( result ) > 20:
self.logger.info( '@@logHandler@@ file=%s, insert.status=success, data.length=%s, data=%s, ..., %s' % (self.curFile, len(result), result[0:10], result[-10:]) )
else:
self.logger.info( '@@logHandler@@ file=%s, insert.status=success, data.length=%s, data=%s' % (self.curFile, len(result), result) )
else:
self.logger.info( '@@logHandler@@ file=%s, insert.status=empty' % self.curFile)
self.position=lf.tell()
self.preFileSize=os.stat(self.curFile).st_size
lf.close()
def alarm(self,result):
for res in result:
if self.keyAlarmers.has_key(res[1]):
for alarmer in self.keyAlarmers[res[1]]:
alarmer.alarm(res[2],self.cubeName,res[1])
class CommandKey:
def __init__(self,ldap,product,type,key,command,alarmers,logger,cubeName,debug,timeInterval):
self.key=key
self.timeInterval=timeInterval
self.command=command
self.alarmers=alarmers
self.dbop=DBOperator(ldap,product,type,cubeName,[key],[],[])
self.logger=logger
self.cubeName=cubeName
self.debug=debug
def validate(self):
if self.key=='':
logger.warning('%s: commandKey未配置key, 不做command的收集')
self.dbop.close()
return False
if self.command=='':
logger.warning('%s: commandKey未配置command, 不做command的收集')
self.dbop.close()
return False
return True
def toString(self):
str=''
str+='key=%s\n' % self.key
str+='timeInterval=%s\n' % self.timeInterval
str+='command=%s\n' % self.command
str+='alarmers=[%s]\n' % ','.join([alarmer.toString() for alarmer in self.alarmers])
return str
def monitor(self):
while True:
currentTime = int( time.time() )
currentTime = currentTime - currentTime % self.timeInterval
value = self.getDataFromCommand()
if value:
# add timestamp
insertData = [[currentTime, self.key, value]]
self.logger.info( '%s: @@commandKeyHandler@@ result=%s' % ( self.cubeName, insertData ) )
self.alarm(insertData)
if not self.debug:
self.dbop.insert( insertData )
self.logger.info( '%s: @@commandKeyHandler@@ result inserted to database' % self.cubeName )
else:
self.logger.warning( '%s: @@commandKeyHandler@@ no data get' % self.cubeName )
endTime = int( time.time() )
sleepTime = self.timeInterval - ( endTime - currentTime )
if sleepTime <= 0:
continue
self.logger.info( '%s: @@commandKeyHandler@@ sleep, will wake up in %ss' % ( self.cubeName, sleepTime ) )
time.sleep( sleepTime )
def alarm(self,insertData):
for data in insertData:
for alarmer in self.alarmers:
alarmer.alarm(data[2],self.cubeName,self.key)
def getDataFromCommand(self):
result=commands.getoutput(self.command).splitlines()
result=result[0].split()
result=result[0]
try:
return float(result)
except:
return None
class Process:
def __init__(self, ldap, product, type, alias, regPattern, processPath,
pids, keys, keyAlarmers, logger, cubeName, debug,
timeInterval):
self.alias = alias
self.regPattern = regPattern
self.processPath = processPath
self.pids = pids
self.oldPids = []
self.keys = keys
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName,
self.getDBKeys(alias, keys), [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
self.timeInterval = timeInterval
# total mem of machine, unit: mb
self.totalMem = 0
self.curUser = None
self.topN = 12
# 用于指向获取进程pid的函数
self.getAllPid = None
# 若不是直接指定pid, 则每次都调用系统命令, 用于check进程正常 -> 优化: 先尝试用上次的pid, 找不到后再通过regPattern和processPath去更新pid
self.currentPidCmdPath = []
def getDBKeys(self, alias, keys):
dbKeys = []
for als in alias:
for key in keys:
dbKeys.append(als + '.p' + key)
return dbKeys
def validate(self):
if len(self.alias) == 0:
self.logger.warning('@@process@@ %s: 未配置 process alias: 将不对进程做收集' %
self.cubeName)
self.dbop.close()
return False
if self.regPattern == '' and self.processPath == '' and len(
self.pids) == 0:
self.logger.warning(
'@@process@@ %s: alias=%s: process配置不正确, 暂不做收集--每个process请至少用pid, processPath, regPattern中的一个来指定'
% (self.cubeName, self.alias))
return False
if len(self.pids) != 0 and len(self.pids) != len(self.alias):
self.logger.warning(
'@@process@@ %s: alias=%s, pid=%s, process由pid指定, 但pid个数与alias个数不相同'
% (self.cubeName, self.alias, self.pids))
self.dbop.close()
return False
# 若由pid指定process, 则regPattern及processPath无效
if len(self.pids) != 0:
if len(self.regPattern) != 0:
self.logger.warning(
'@@process@@ %s: alias=%s, pid=%s, regPattern=%s: process由pid指定, regPattern配置无效'
% (self.cubeName, self.alias, self.pids, self.regPattern))
self.regPattern = ''
if len(self.processPath) != 0:
self.logger.warning(
'@@process@@ %s: alias=%s, pid=%s, processPath=%s: process由pid指定, processPath配置无效'
% (self.cubeName, self.alias, self.pids, self.processPath))
self.processPath = ''
self.postInit()
return True
def toString(self):
str = ''
str += 'keys=%s\n' % ','.join(self.keys)
str += 'alias=%s\n' % ','.join(self.alias)
str += 'pids=%s\n' % ','.join(self.pids)
str += 'regPattern=%s\n' % self.regPattern
str += 'processPath=%s\n' % self.processPath
str += 'timeInterval=%s\n' % self.timeInterval
str += 'keyAlarmers='
for key, alarmers in self.keyAlarmers.items():
str += '%s:[' % key
str += ','.join([alarmer.toString() for alarmer in alarmers])
str += '],'
str = str[0:-1] + '\n'
return str
# check通过后的初始化任务
def postInit(self):
# 计算total_mem: 目前每个Process都会算一次, 不过占用资源较少, 暂不设为global或由Collect来计算并赋值
meminfo = os.popen(
"cat /proc/meminfo |grep 'MemTotal'|awk '{print $2,$3}'",
"r").read().split()
self.totalMem = float(meminfo[0])
unit = meminfo[1].lower()
if unit == 'kb':
self.totalMem = self.totalMem / 1024
# 后面两种应该没啥必要, 不过有备无患:)
elif unit == 'mb':
pass
elif unit == 'gb':
self.totalMem = self.totalMem * 1024
else:
self.logger.info('@@process@@ %s: %s cannot get total mem' %
(self.cubeName, self.alias))
exit(-1)
# 获取当前用户名
self.curUser = commands.getoutput('whoami').strip()
# 指定获取进程的方法
if len(self.pids) != 0:
self.getAllPid = self.getPid
elif self.regPattern != '':
if self.processPath != '':
self.getAllPid = self.getPidByRegAndPath
else:
self.getAllPid = self.getPidByReg
else:
self.getAllPid = self.getPidByPath
def monitor(self):
while True:
currentTime = int(time.time())
currentTime = currentTime - currentTime % self.timeInterval
processDatas = self.getAllProcessData()
if len(processDatas) > 0:
# add timestamp
insertData = []
for data in processDatas:
insertData.append([currentTime, data[0], data[1]])
self.logger.info('@@processHandler@@ %s: result=%s' %
(self.cubeName, insertData))
self.alarm(insertData)
# insert, 与logHandler共用dbop
if not self.debug:
self.dbop.insert(insertData)
self.logger.info(
'@@processHandler@@ %s: result inserted to database' %
self.cubeName)
else:
self.logger.info('%s: @@processHandler@@ no process data get' %
self.cubeName)
endTime = int(time.time())
sleepTime = self.timeInterval - (endTime - currentTime)
if sleepTime <= 0:
continue
self.logger.info(
'@@processHandler@@ %s: sleep, will wake up in %ss' %
(self.cubeName, sleepTime))
time.sleep(sleepTime)
def alarm(self, result):
# process 的数据较少, 可以这样干
for key, alarmers in self.keyAlarmers.items():
# 所有alias
if key in self.keys:
for res in result:
if key in res[1]:
for alarmer in alarmers:
alarmer.alarm(res[2], self.cubeName, key)
# 仅对某alias
else:
# zhangb的前台遗留问题, pid相关的key形式为: alias+'.p'+key
for res in result:
if key.replace(key[key.rfind('.')], '.p') == res[1]:
for alarmer in alarmers:
alarmer.alarm(res[2], self.cubeName, key)
def getAllProcessData(self):
if len(self.pids) != 0:
pids = self.getAllPid()
elif self.validOldPids():
pids = self.oldPids
elif len(self.pids) == 0:
self.refreshPidInfo()
pids = self.getAllPid()
if len(pids) > len(self.alias):
self.logger.info(
'@@process@@ %s: alias=%s, 配置的进程数(%s) < 检测到的个数(%s), 暂不对进程做收集' %
(self.cubeName, self.alias, len(self.alias), len(pids)))
return []
if len(pids) < len(self.alias):
self.logger.warning(
'@@process@@ %s: alias=%s 检测到的进程数(%s) < 配置的个数(%s)' %
(self.cubeName, self.alias, len(pids), len(self.alias)))
if len(pids) == 0:
self.logger.warning(
'@@process@@ %s: alias=%s 检测到的进程数为0, 暂不对进程做收集' %
(self.cubeName, self.alias))
return []
if len(self.oldPids) == len(pids):
for i in range(len(pids)):
if pids[i] != self.oldPids[i]:
self.logger.info(
'@@process@@ %s: pid changed for alias=%s: %s -> %s' %
(self.cubeName, self.alias[i], self.oldPids[i],
pids[i]))
self.oldPids = pids
# 获取配置进程数据
memAndCpuData = self.getMemAndCpuData(pids)
fdsData = self.getFdsData(pids)
netLinksData = self.getNetLinksData(pids)
threadsData = self.getThreadsData(pids)
return memAndCpuData + fdsData + netLinksData + threadsData
def validOldPids(self):
'''
是否所有进程都在
'''
if len(self.oldPids) == 0:
self.logger.debug('@@process@@ %s: alias=%s, valid=False' %
(self.cubeName, self.alias))
return False
for pid in self.oldPids:
# check该进程是否存在
try:
os.kill(int(pid), 0)
except OSError:
self.logger.debug(
'@@process@@ %s: alias=%s, pid=%s, valid=False' %
(self.cubeName, pid, self.alias))
return False
self.logger.debug('@@process@@ %s: alias=%s, valid=True' %
(self.cubeName, self.alias))
return True
def refreshPidInfo(self):
self.currentPidCmdPath = []
# 获取当前user的进程pid, cmd, path的shell命令 (返回格式执行可见):
# ps x --columns=100000 | sed -r 's/^\s*(.*)/\1/' | grep -P "^\s*\d" | grep -v -P "\d+:\d+\s+(ps|grep|sed|ssh|-bash|ls|sh)" | sed -r 's/^([0-9]*).*[0-9]+:[0-9]+\s+(.*)/echo -ne "\1\\\\t\2\\\\t";ls -l \/proc\/\1\/cwd/' | grep -v columns=100000 | sh | grep -P "^\d" | sed -r "s/([0-9]*)\t(.*)\t.*-> (.*)/\1\t\2\t\3/"
# 1. ps -x (可改为用ps -xo) 注:ps 命令会有截断, 设为10w长可对付一般情况, os默认为page length, 一般为4096, 可能取不到完整的commands
# 2. 去掉非进程开头的行
# 3. 去掉shell本身的进程 注: 过滤了sh, top等进程, 而一般的监控场景也不是对此类进程做监控
# 4. 获取pid及path
# 5. ls -l /proc/pid/cwd获取进程cwd 注: ls -1 /proc/pid/cwd时可能会有Permission denied的情况, 因此需要对返回值做处理
# 6. escape真是很xx...!!!: reg+sed(还好用了-r)+echo -ne+python+...
cmd = "ps x --columns=100000 | sed -r \"s/^\s*(.*)/\\1/\" |grep -P \"^\s*\d\" | grep -v -P \"\d+:\d+\s+(ps|grep|sed|ssh|top|tail|less|tail|more|-bash|ls|sh|vim?\s)\" | sed -r 's/^([0-9]*).*[0-9]+:[0-9]+\s+(.*)/echo -ne \"\\1\\\\t\\2\\\\t\";ls -l \/proc\/\\1\/cwd/' | grep -v columns=100000 | sh | grep -P \"^\d\" | sed -r \"s/([0-9]*)\\\\t(.*)\\t.*-> (.*)/\\1\\\\t\\2\\\\t\\3/\""
lines = subprocess.Popen(
args=cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate()[0].splitlines()
for line in lines:
if 'Permission denied' in line:
continue
datas = line.split()
if len(datas) < 3:
continue
if len(datas) == 3:
self.currentPidCmdPath.append([datas[0], datas[1], datas[2]])
else:
self.currentPidCmdPath.append(
[datas[0], ' '.join(datas[1:-1]), datas[-1]])
self.logger.info(
'@@process@@ %s: alias=%s, update user\'s process info successfully, pidCmdPath.info=%s'
% (self.cubeName, self.alias, self.currentPidCmdPath))
def getPid(self):
return self.pids
def getPidByReg(self):
pids = []
for pid, cmd, path in self.currentPidCmdPath:
if len(re.findall(self.regPattern, cmd)) != 0:
pids.append(pid)
self.logger.info('@@process@@ %s: ailas=%s, getPidByRegPattern=%s' %
(self.cubeName, self.alias, pids))
return pids
def getPidByPath(self):
pids = []
for pid, cmd, path in self.currentPidCmdPath:
if path == '/' and self.processPath != '/':
continue
if self.processPath.strip('/') in path.strip('/'):
pids.append(pid)
self.logger.info('@@process@@ %s: ailas=%s, getPidByPath=%s' %
(self.cubeName, self.alias, pids))
return pids
def getPidByRegAndPath(self):
regPid = self.getPidByReg()
pathPid = self.getPidByPath()
pids = []
for pid in regPid:
if pid in pathPid:
pids.append(pid)
return pids
def getMemAndCpuData(self, currentPids):
memAndCpuData = []
cmd = "top -n1 -b -p %s | tail -n %s" % (
','.join(currentPids), len(currentPids) +
1) + " | awk '{printf(\"%s %s %s\\n\",$1,$9,$10);}' "
self.logger.debug('@@process@@ %s: cmd=%s' % (self.cubeName, cmd))
results = subprocess.Popen(
args=cmd, shell=True,
stdout=subprocess.PIPE).communicate()[0].strip().splitlines()
for result in results:
lineinfos = result.split()
# 正常情况下每行应只有\s\d\.
# 若有进程已被kill掉, len(lineinfos)可能不为3
if result.upper() != result.lower() or len(lineinfos) != 3:
self.logger.warning(
'@@process@@ %s: alias=%s: 获取到异常进程信息: result=%s' %
(self.cubeName, self.alias, result))
break
memAndCpuData.append([
self.alias[currentPids.index(lineinfos[0])] + '.pcpu',
float(lineinfos[1])
])
memAndCpuData.append([
self.alias[currentPids.index(lineinfos[0])] + '.pmem',
float(lineinfos[2]) * self.totalMem / 100
])
self.logger.info('@@process@@ %s: alias=%s, memAndCpuData=%s' %
(self.cubeName, self.alias, memAndCpuData))
return memAndCpuData
def getFdsData(self, pids):
fdsData = []
if 'fds' not in self.keys:
return fdsData
for pid in pids:
cmd = 'ls -l /proc/%s/fd | wc -l' % pid
self.logger.debug('@@process@@ %s: cmd=%s' % (self.cubeName, cmd))
res = subprocess.Popen(args=cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate(
)[0].splitlines()[0].split()
if len(res) == 0 or not res[0].isdigit():
continue
else:
fdsData.append(
[self.alias[pids.index(pid)] + '.pfds',
int(res[0])])
return fdsData
def getNetLinksData(self, pids):
netLinksData = []
if 'netLinks' not in self.keys:
return netLinksData
for pid in pids:
cmd = 'netstat -tnp | grep %s | grep ESTABLISHED | wc -l' % pid
self.logger.debug('@@process@@ %s: cmd=%s' % (self.cubeName, cmd))
res = subprocess.Popen(args=cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate(
)[0].splitlines()[0].split()
if len(res) == 0 or not res[0].isdigit():
continue
else:
netLinksData.append(
[self.alias[pids.index(pid)] + '.pnetLinks',
int(res[0])])
return netLinksData
def getThreadsData(self, pids):
threadsData = []
if 'threads' not in self.keys:
return threadsData
for pid in pids:
cmd = 'pstree -p %s | wc -l' % pid
self.logger.debug('@@process@@ %s: cmd=%s' % (self.cubeName, cmd))
res = subprocess.Popen(args=cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate(
)[0].splitlines()[0].split()
if len(res) == 0 or not res[0].isdigit():
continue
else:
threadsData.append(
[self.alias[pids.index(pid)] + '.pthreads',
int(res[0])])
return threadsData
def getOtherUserMemCpu(self):
'''
获取其他用户的cpu和内存信息
'''
otheruserinfos = []
if not self.curUser:
return otheruserinfos
cmd = "top -n1 b | grep -e \"^ *[0-9]\+\" | grep -v %s" % (
self.curUser) + " | awk '{printf(\"%s %s %s\\n\",$2,$9,$10);}' "
self.logger.debug('@@process@@ %s: cmd=%s' % (self.cubeName, cmd))
results = subprocess.Popen(
args=cmd, shell=True,
stdout=subprocess.PIPE).communicate()[0].strip().splitlines()
otheruserinfo = {}
for result in results:
lineinfos = result.split()
# 正常情况下每行应只有\s\d\.
# 若有进程已被kill掉, len(lineinfos)可能不为3
if len(lineinfos) != 3:
continue
try:
username = lineinfos[0]
cpurate = float(lineinfos[1])
memrate = float(lineinfos[2])
if not otheruserinfo.has_key(username):
otheruserinfo[username] = {
'user_%s.pall' % (username): cpurate + memrate,
'user_%s.pcpu' % (username): cpurate,
'user_%s.pmem' % (username):
memrate * self.totalMem / 100
}
else:
otheruserinfo[username]['user_%s.pall' %
(username)] += (cpurate + memrate)
otheruserinfo[username]['user_%s.pcpu' %
(username)] += cpurate
otheruserinfo[username]['user_%s.pmem' %
(username)] += (memrate *
self.totalMem /
100)
except:
pass
# 排序
otheruserinfolist = sorted(
otheruserinfo.iteritems(),
key=lambda (username, uservalue): uservalue['user_%s.pall' %
(username)],
reverse=True)
otheruserinfolist = otheruserinfolist[:self.topN]
for username, uservalue in otheruserinfolist:
otheruserinfos.append([
'user_%s.pcpu' % (username),
uservalue['user_%s.pcpu' % (username)]
])
otheruserinfos.append([
'user_%s.pmem' % (username),
uservalue['user_%s.pmem' % (username)]
])
self.logger.info(
'@@process@@ %s: otheruser processinfo , memAndCpuData=%s' %
(self.cubeName, otheruserinfos))
return otheruserinfos
class Process:
def __init__(
self,
ldap,
product,
type,
alias,
regPattern,
processPath,
pids,
keys,
keyAlarmers,
logger,
cubeName,
debug,
timeInterval,
):
self.alias = alias
self.regPattern = regPattern
self.processPath = processPath
self.pids = pids
self.oldPids = []
self.keys = keys
self.keyAlarmers = keyAlarmers
self.dbop = DBOperator(ldap, product, type, cubeName, self.getDBKeys(alias, keys), [], [])
self.logger = logger
self.cubeName = cubeName
self.debug = debug
self.timeInterval = timeInterval
# total mem of machine, unit: mb
self.totalMem = 0
self.curUser = None
self.topN = 12
# 用于指向获取进程pid的函数
self.getAllPid = None
# 若不是直接指定pid, 则每次都调用系统命令, 用于check进程正常 -> 优化: 先尝试用上次的pid, 找不到后再通过regPattern和processPath去更新pid
self.currentPidCmdPath = []
def getDBKeys(self, alias, keys):
dbKeys = []
for als in alias:
for key in keys:
dbKeys.append(als + ".p" + key)
return dbKeys
def validate(self):
if len(self.alias) == 0:
self.logger.warning("@@process@@ %s: 未配置 process alias: 将不对进程做收集" % self.cubeName)
self.dbop.close()
return False
if self.regPattern == "" and self.processPath == "" and len(self.pids) == 0:
self.logger.warning(
"@@process@@ %s: alias=%s: process配置不正确, 暂不做收集--每个process请至少用pid, processPath, regPattern中的一个来指定"
% (self.cubeName, self.alias)
)
return False
if len(self.pids) != 0 and len(self.pids) != len(self.alias):
self.logger.warning(
"@@process@@ %s: alias=%s, pid=%s, process由pid指定, 但pid个数与alias个数不相同"
% (self.cubeName, self.alias, self.pids)
)
self.dbop.close()
return False
# 若由pid指定process, 则regPattern及processPath无效
if len(self.pids) != 0:
if len(self.regPattern) != 0:
self.logger.warning(
"@@process@@ %s: alias=%s, pid=%s, regPattern=%s: process由pid指定, regPattern配置无效"
% (self.cubeName, self.alias, self.pids, self.regPattern)
)
self.regPattern = ""
if len(self.processPath) != 0:
self.logger.warning(
"@@process@@ %s: alias=%s, pid=%s, processPath=%s: process由pid指定, processPath配置无效"
% (self.cubeName, self.alias, self.pids, self.processPath)
)
self.processPath = ""
self.postInit()
return True
def toString(self):
str = ""
str += "keys=%s\n" % ",".join(self.keys)
str += "alias=%s\n" % ",".join(self.alias)
str += "pids=%s\n" % ",".join(self.pids)
str += "regPattern=%s\n" % self.regPattern
str += "processPath=%s\n" % self.processPath
str += "timeInterval=%s\n" % self.timeInterval
str += "keyAlarmers="
for key, alarmers in self.keyAlarmers.items():
str += "%s:[" % key
str += ",".join([alarmer.toString() for alarmer in alarmers])
str += "],"
str = str[0:-1] + "\n"
return str
# check通过后的初始化任务
def postInit(self):
# 计算total_mem: 目前每个Process都会算一次, 不过占用资源较少, 暂不设为global或由Collect来计算并赋值
meminfo = os.popen("cat /proc/meminfo |grep 'MemTotal'|awk '{print $2,$3}'", "r").read().split()
self.totalMem = float(meminfo[0])
unit = meminfo[1].lower()
if unit == "kb":
self.totalMem = self.totalMem / 1024
# 后面两种应该没啥必要, 不过有备无患:)
elif unit == "mb":
pass
elif unit == "gb":
self.totalMem = self.totalMem * 1024
else:
self.logger.info("@@process@@ %s: %s cannot get total mem" % (self.cubeName, self.alias))
exit(-1)
# 获取当前用户名
self.curUser = commands.getoutput("whoami").strip()
# 指定获取进程的方法
if len(self.pids) != 0:
self.getAllPid = self.getPid
elif self.regPattern != "":
if self.processPath != "":
self.getAllPid = self.getPidByRegAndPath
else:
self.getAllPid = self.getPidByReg
else:
self.getAllPid = self.getPidByPath
def monitor(self):
while True:
currentTime = int(time.time())
currentTime = currentTime - currentTime % self.timeInterval
processDatas = self.getAllProcessData()
if len(processDatas) > 0:
# add timestamp
insertData = []
for data in processDatas:
insertData.append([currentTime, data[0], data[1]])
self.logger.info("@@processHandler@@ %s: result=%s" % (self.cubeName, insertData))
self.alarm(insertData)
# insert, 与logHandler共用dbop
if not self.debug:
self.dbop.insert(insertData)
self.logger.info("@@processHandler@@ %s: result inserted to database" % self.cubeName)
else:
self.logger.info("%s: @@processHandler@@ no process data get" % self.cubeName)
endTime = int(time.time())
sleepTime = self.timeInterval - (endTime - currentTime)
if sleepTime <= 0:
continue
self.logger.info("@@processHandler@@ %s: sleep, will wake up in %ss" % (self.cubeName, sleepTime))
time.sleep(sleepTime)
def alarm(self, result):
# process 的数据较少, 可以这样干
for key, alarmers in self.keyAlarmers.items():
# 所有alias
if key in self.keys:
for res in result:
if key in res[1]:
for alarmer in alarmers:
alarmer.alarm(res[2], self.cubeName, key)
# 仅对某alias
else:
# zhangb的前台遗留问题, pid相关的key形式为: alias+'.p'+key
for res in result:
if key.replace(key[key.rfind(".")], ".p") == res[1]:
for alarmer in alarmers:
alarmer.alarm(res[2], self.cubeName, key)
def getAllProcessData(self):
if len(self.pids) != 0:
pids = self.getAllPid()
elif self.validOldPids():
pids = self.oldPids
elif len(self.pids) == 0:
self.refreshPidInfo()
pids = self.getAllPid()
if len(pids) > len(self.alias):
self.logger.info(
"@@process@@ %s: alias=%s, 配置的进程数(%s) < 检测到的个数(%s), 暂不对进程做收集"
% (self.cubeName, self.alias, len(self.alias), len(pids))
)
return []
if len(pids) < len(self.alias):
self.logger.warning(
"@@process@@ %s: alias=%s 检测到的进程数(%s) < 配置的个数(%s)"
% (self.cubeName, self.alias, len(pids), len(self.alias))
)
if len(pids) == 0:
self.logger.warning("@@process@@ %s: alias=%s 检测到的进程数为0, 暂不对进程做收集" % (self.cubeName, self.alias))
return []
if len(self.oldPids) == len(pids):
for i in range(len(pids)):
if pids[i] != self.oldPids[i]:
self.logger.info(
"@@process@@ %s: pid changed for alias=%s: %s -> %s"
% (self.cubeName, self.alias[i], self.oldPids[i], pids[i])
)
self.oldPids = pids
# 获取配置进程数据
memAndCpuData = self.getMemAndCpuData(pids)
fdsData = self.getFdsData(pids)
netLinksData = self.getNetLinksData(pids)
threadsData = self.getThreadsData(pids)
return memAndCpuData + fdsData + netLinksData + threadsData
def validOldPids(self):
"""
是否所有进程都在
"""
if len(self.oldPids) == 0:
self.logger.debug("@@process@@ %s: alias=%s, valid=False" % (self.cubeName, self.alias))
return False
for pid in self.oldPids:
# check该进程是否存在
try:
os.kill(int(pid), 0)
except OSError:
self.logger.debug("@@process@@ %s: alias=%s, pid=%s, valid=False" % (self.cubeName, pid, self.alias))
return False
self.logger.debug("@@process@@ %s: alias=%s, valid=True" % (self.cubeName, self.alias))
return True
def refreshPidInfo(self):
self.currentPidCmdPath = []
# 获取当前user的进程pid, cmd, path的shell命令 (返回格式执行可见):
# ps x --columns=100000 | sed -r 's/^\s*(.*)/\1/' | grep -P "^\s*\d" | grep -v -P "\d+:\d+\s+(ps|grep|sed|ssh|-bash|ls|sh)" | sed -r 's/^([0-9]*).*[0-9]+:[0-9]+\s+(.*)/echo -ne "\1\\\\t\2\\\\t";ls -l \/proc\/\1\/cwd/' | grep -v columns=100000 | sh | grep -P "^\d" | sed -r "s/([0-9]*)\t(.*)\t.*-> (.*)/\1\t\2\t\3/"
# 1. ps -x (可改为用ps -xo) 注:ps 命令会有截断, 设为10w长可对付一般情况, os默认为page length, 一般为4096, 可能取不到完整的commands
# 2. 去掉非进程开头的行
# 3. 去掉shell本身的进程 注: 过滤了sh, top等进程, 而一般的监控场景也不是对此类进程做监控
# 4. 获取pid及path
# 5. ls -l /proc/pid/cwd获取进程cwd 注: ls -1 /proc/pid/cwd时可能会有Permission denied的情况, 因此需要对返回值做处理
# 6. escape真是很xx...!!!: reg+sed(还好用了-r)+echo -ne+python+...
cmd = 'ps x --columns=100000 | sed -r "s/^\s*(.*)/\\1/" |grep -P "^\s*\d" | grep -v -P "\d+:\d+\s+(ps|grep|sed|ssh|top|tail|less|tail|more|-bash|ls|sh|vim?\s)" | sed -r \'s/^([0-9]*).*[0-9]+:[0-9]+\s+(.*)/echo -ne "\\1\\\\t\\2\\\\t";ls -l \/proc\/\\1\/cwd/\' | grep -v columns=100000 | sh | grep -P "^\d" | sed -r "s/([0-9]*)\\\\t(.*)\\t.*-> (.*)/\\1\\\\t\\2\\\\t\\3/"'
lines = (
subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
.communicate()[0]
.splitlines()
)
for line in lines:
if "Permission denied" in line:
continue
datas = line.split()
if len(datas) < 3:
continue
if len(datas) == 3:
self.currentPidCmdPath.append([datas[0], datas[1], datas[2]])
else:
self.currentPidCmdPath.append([datas[0], " ".join(datas[1:-1]), datas[-1]])
self.logger.info(
"@@process@@ %s: alias=%s, update user's process info successfully, pidCmdPath.info=%s"
% (self.cubeName, self.alias, self.currentPidCmdPath)
)
def getPid(self):
return self.pids
def getPidByReg(self):
pids = []
for pid, cmd, path in self.currentPidCmdPath:
if len(re.findall(self.regPattern, cmd)) != 0:
pids.append(pid)
self.logger.info("@@process@@ %s: ailas=%s, getPidByRegPattern=%s" % (self.cubeName, self.alias, pids))
return pids
def getPidByPath(self):
pids = []
for pid, cmd, path in self.currentPidCmdPath:
if path == "/" and self.processPath != "/":
continue
if self.processPath.strip("/") in path.strip("/"):
pids.append(pid)
self.logger.info("@@process@@ %s: ailas=%s, getPidByPath=%s" % (self.cubeName, self.alias, pids))
return pids
def getPidByRegAndPath(self):
regPid = self.getPidByReg()
pathPid = self.getPidByPath()
pids = []
for pid in regPid:
if pid in pathPid:
pids.append(pid)
return pids
def getMemAndCpuData(self, currentPids):
memAndCpuData = []
cmd = (
"top -n1 -b -p %s | tail -n %s" % (",".join(currentPids), len(currentPids) + 1)
+ " | awk '{printf(\"%s %s %s\\n\",$1,$9,$10);}' "
)
self.logger.debug("@@process@@ %s: cmd=%s" % (self.cubeName, cmd))
results = subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE).communicate()[0].strip().splitlines()
for result in results:
lineinfos = result.split()
# 正常情况下每行应只有\s\d\.
# 若有进程已被kill掉, len(lineinfos)可能不为3
if result.upper() != result.lower() or len(lineinfos) != 3:
self.logger.warning(
"@@process@@ %s: alias=%s: 获取到异常进程信息: result=%s" % (self.cubeName, self.alias, result)
)
break
memAndCpuData.append([self.alias[currentPids.index(lineinfos[0])] + ".pcpu", float(lineinfos[1])])
memAndCpuData.append(
[self.alias[currentPids.index(lineinfos[0])] + ".pmem", float(lineinfos[2]) * self.totalMem / 100]
)
self.logger.info("@@process@@ %s: alias=%s, memAndCpuData=%s" % (self.cubeName, self.alias, memAndCpuData))
return memAndCpuData
def getFdsData(self, pids):
fdsData = []
if "fds" not in self.keys:
return fdsData
for pid in pids:
cmd = "ls -l /proc/%s/fd | wc -l" % pid
self.logger.debug("@@process@@ %s: cmd=%s" % (self.cubeName, cmd))
res = (
subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
.communicate()[0]
.splitlines()[0]
.split()
)
if len(res) == 0 or not res[0].isdigit():
continue
else:
fdsData.append([self.alias[pids.index(pid)] + ".pfds", int(res[0])])
return fdsData
def getNetLinksData(self, pids):
netLinksData = []
if "netLinks" not in self.keys:
return netLinksData
for pid in pids:
cmd = "netstat -tnp | grep %s | grep ESTABLISHED | wc -l" % pid
self.logger.debug("@@process@@ %s: cmd=%s" % (self.cubeName, cmd))
res = (
subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
.communicate()[0]
.splitlines()[0]
.split()
)
if len(res) == 0 or not res[0].isdigit():
continue
else:
netLinksData.append([self.alias[pids.index(pid)] + ".pnetLinks", int(res[0])])
return netLinksData
def getThreadsData(self, pids):
threadsData = []
if "threads" not in self.keys:
return threadsData
for pid in pids:
cmd = "pstree -p %s | wc -l" % pid
self.logger.debug("@@process@@ %s: cmd=%s" % (self.cubeName, cmd))
res = (
subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
.communicate()[0]
.splitlines()[0]
.split()
)
if len(res) == 0 or not res[0].isdigit():
continue
else:
threadsData.append([self.alias[pids.index(pid)] + ".pthreads", int(res[0])])
return threadsData
def getOtherUserMemCpu(self):
"""
获取其他用户的cpu和内存信息
"""
otheruserinfos = []
if not self.curUser:
return otheruserinfos
cmd = (
'top -n1 b | grep -e "^ *[0-9]\+" | grep -v %s' % (self.curUser)
+ " | awk '{printf(\"%s %s %s\\n\",$2,$9,$10);}' "
)
self.logger.debug("@@process@@ %s: cmd=%s" % (self.cubeName, cmd))
results = subprocess.Popen(args=cmd, shell=True, stdout=subprocess.PIPE).communicate()[0].strip().splitlines()
otheruserinfo = {}
for result in results:
lineinfos = result.split()
# 正常情况下每行应只有\s\d\.
# 若有进程已被kill掉, len(lineinfos)可能不为3
if len(lineinfos) != 3:
continue
try:
username = lineinfos[0]
cpurate = float(lineinfos[1])
memrate = float(lineinfos[2])
if not otheruserinfo.has_key(username):
otheruserinfo[username] = {
"user_%s.pall" % (username): cpurate + memrate,
"user_%s.pcpu" % (username): cpurate,
"user_%s.pmem" % (username): memrate * self.totalMem / 100,
}
else:
otheruserinfo[username]["user_%s.pall" % (username)] += cpurate + memrate
otheruserinfo[username]["user_%s.pcpu" % (username)] += cpurate
otheruserinfo[username]["user_%s.pmem" % (username)] += memrate * self.totalMem / 100
except:
pass
# 排序
otheruserinfolist = sorted(
otheruserinfo.iteritems(),
key=lambda (username, uservalue): uservalue["user_%s.pall" % (username)],
reverse=True,
)
otheruserinfolist = otheruserinfolist[: self.topN]
for username, uservalue in otheruserinfolist:
otheruserinfos.append(["user_%s.pcpu" % (username), uservalue["user_%s.pcpu" % (username)]])
otheruserinfos.append(["user_%s.pmem" % (username), uservalue["user_%s.pmem" % (username)]])
self.logger.info("@@process@@ %s: otheruser processinfo , memAndCpuData=%s" % (self.cubeName, otheruserinfos))
return otheruserinfos