def updateWithNotSuccAppid(self, app):
appHappenTime = util.getIntervalTime(app['startedTime'])
rm = self.getRm(self.recordTime, appHappenTime)
rm.inc("appNum", 1)
if app['stats'] == 'KILLED':
rm.inc("killedApp", 1)
elif app['stats'] == 'FAILED':
rm.inc("failedApp", 1)
def __init__(self, beginTime=( time.time()-config.collect_interval) ):
self.interval = config.collect_interval
self.recordTime = util.getIntervalTime(beginTime)
msg = "begin to run at"+ time.strftime('%Y-%m-%d %A %X',time.localtime(self.recordTime))
logger.info(msg)
print msg
self.rmList = {}
self.nmList = {}
self.appList = {}
def updateWithNotSuccAppid(self,app):
appHappenTime = util.getIntervalTime(app['startedTime'])
rm = self.getRm(self.recordTime,appHappenTime)
rm.inc("appNum",1)
if app['stats'] == 'KILLED':
rm.inc("killedApp",1)
elif app['stats'] == 'FAILED':
rm.inc("failedApp",1)
def __init__(self, beginTime=(time.time() - config.collect_interval)):
self.interval = config.collect_interval
self.recordTime = util.getIntervalTime(beginTime)
msg = "begin to run at" + time.strftime(
'%Y-%m-%d %A %X', time.localtime(self.recordTime))
logger.info(msg)
print msg
self.rmList = {}
self.nmList = {}
self.appList = {}
def updateWithAppid(self, app, jobHistory, jobCounter):
#update nm and rm
amNode = self.getNodeFromAddress(app['amHostHttpAddress'])
appHappenTime = util.getIntervalTime(app['startedTime'])
nm = self.getNm(amNode, self.recordTime, appHappenTime)
rm = self.getRm(self.recordTime, appHappenTime)
nm.inc("containerNum", 1)
nm.inc("amNum", 1)
rm.inc("appNum", 1)
rm.inc("finishedApp", 1)
if app['finalStatus'] != "SUCCEEDED":
rm.inc("notSuccApp", 1)
#end update
appid = app["id"]
appRecord = self.getAppidRecord(appid)
keyFromApp = [
"user", "name", "queue", "startedTime", "finishedTime", "state",
"finalStatus"
]
for key in keyFromApp:
if key == "startedTime" or key == "finishedTime":
appRecord.set(key, util.getSecondTime(app[key]))
else:
appRecord.set(key, app[key])
#todo
appRecord.set("attemptNumber", 1)
keyFromHistory = [
"mapsTotal", "mapsCompleted", "successfulMapAttempts",
"killedMapAttempts", "failedMapAttempts", "avgMapTime",
"reducesTotal", "reducesCompleted", "successfulReduceAttempts",
"killedReduceAttempts", "failedReduceAttempts", "avgReduceTime"
]
if jobHistory.has_key('job'):
for key in keyFromHistory:
appRecord.set(key, jobHistory['job'][key])
#TODO "localMap","rackMap"
keyMapFromCounters = {
"DATA_LOCAL_MAPS": "localMap",
"RACK_LOCAL_MAPS": "rackMap",
"FILE_BYTES_READ": "fileRead",
"FILE_BYTES_WRITTEN": "fileWrite",
"HDFS_BYTES_READ": "hdfsRead",
"HDFS_BYTES_WRITTEN": "hdfsWrite"
}
if jobCounter:
for (key, value) in keyMapFromCounters.items():
appRecord.set(value, jobCounter[key]['total'])
def updateWithAttempt(self, attempt, attemptCounter):
#update nm's containerNum , mapNum , reduceNum
node = self.getNodeFromAddress(attempt['nodeHttpAddress'])
happenTime = util.getIntervalTime(attempt['startTime'])
rm = self.getRm(self.recordTime, happenTime)
nm = self.getNm(node, self.recordTime, happenTime)
#*********************************
nm.inc("containerNum", 1)
if attempt['type'] == 'MAP':
rm.inc("mapNum", 1)
nm.inc("mapNum", 1)
rm.inc("mapTime", attempt['elapsedTime'])
nm.inc("mapTime", attempt['elapsedTime'])
if attempt['state'] != "SUCCEEDED":
rm.inc("failMap", 1)
nm.inc("failMap", 1)
elif attempt['type'] == 'REDUCE':
rm.inc("reduceNum", 1)
nm.inc("reduceNum", 1)
rm.inc("reduceTime", attempt['elapsedTime'])
nm.inc("reduceTime", attempt['elapsedTime'])
if attempt['state'] != "SUCCEEDED":
rm.inc("failReduce", 1)
nm.inc("failReduce", 1)
#*********************************
if not attemptCounter or not attemptCounter.has_key('jobTaskAttemptCounters') \
or not attemptCounter['jobTaskAttemptCounters'].has_key('taskAttemptCounterGroup'):
return
for taskAttemptCounterGroup in attemptCounter[
'jobTaskAttemptCounters']['taskAttemptCounterGroup']:
if taskAttemptCounterGroup[
'counterGroupName'] == "org.apache.hadoop.mapreduce.FileSystemCounter":
for counter in taskAttemptCounterGroup['counter']:
if counter['name'] == 'FILE_BYTES_READ':
rm.inc("fileRead", counter["value"])
nm.inc("fileRead", counter["value"])
elif counter['name'] == 'FILE_BYTES_WRITTEN':
rm.inc("fileWrite", counter["value"])
nm.inc("fileWrite", counter["value"])
elif counter['name'] == 'HDFS_BYTES_READ':
rm.inc("hdfsRead", counter["value"])
nm.inc("hdfsRead", counter["value"])
elif counter['name'] == 'HDFS_BYTES_WRITTEN':
rm.inc("hdfsWrite", counter["value"])
nm.inc("hdfsWrite", counter["value"])
else:
continue
def updateWithAttempt(self,attempt,attemptCounter):
#update nm's containerNum , mapNum , reduceNum
node = self.getNodeFromAddress(attempt['nodeHttpAddress'])
happenTime = util.getIntervalTime(attempt['startTime'])
rm = self.getRm(self.recordTime,happenTime)
nm = self.getNm(node,self.recordTime, happenTime)
#*********************************
nm.inc("containerNum",1)
if attempt['type'] == 'MAP':
rm.inc("mapNum",1)
nm.inc("mapNum",1)
rm.inc("mapTime",attempt['elapsedTime'])
nm.inc("mapTime",attempt['elapsedTime'])
if attempt['state'] != "SUCCEEDED":
rm.inc("failMap",1)
nm.inc("failMap",1)
elif attempt['type'] == 'REDUCE':
rm.inc("reduceNum",1)
nm.inc("reduceNum",1)
rm.inc("reduceTime",attempt['elapsedTime'])
nm.inc("reduceTime",attempt['elapsedTime'])
if attempt['state'] != "SUCCEEDED":
rm.inc("failReduce",1)
nm.inc("failReduce",1)
#*********************************
if not attemptCounter or not attemptCounter.has_key('jobTaskAttemptCounters') \
or not attemptCounter['jobTaskAttemptCounters'].has_key('taskAttemptCounterGroup'):
return
for taskAttemptCounterGroup in attemptCounter['jobTaskAttemptCounters']['taskAttemptCounterGroup']:
if taskAttemptCounterGroup['counterGroupName'] == "org.apache.hadoop.mapreduce.FileSystemCounter":
for counter in taskAttemptCounterGroup['counter']:
if counter['name'] == 'FILE_BYTES_READ':
rm.inc("fileRead",counter["value"])
nm.inc("fileRead",counter["value"])
elif counter['name'] == 'FILE_BYTES_WRITTEN':
rm.inc("fileWrite",counter["value"])
nm.inc("fileWrite",counter["value"])
elif counter['name'] == 'HDFS_BYTES_READ':
rm.inc("hdfsRead",counter["value"])
nm.inc("hdfsRead",counter["value"])
elif counter['name'] == 'HDFS_BYTES_WRITTEN':
rm.inc("hdfsWrite",counter["value"])
nm.inc("hdfsWrite",counter["value"])
else:
continue
def updateWithAppid(self,app,jobHistory,jobCounter):
#update nm and rm
amNode = self.getNodeFromAddress(app['amHostHttpAddress'])
appHappenTime = util.getIntervalTime(app['startedTime'])
nm = self.getNm(amNode,self.recordTime, appHappenTime)
rm = self.getRm(self.recordTime,appHappenTime)
nm.inc("containerNum",1);
nm.inc("amNum",1);
rm.inc("appNum",1)
rm.inc("finishedApp",1)
if app['finalStatus'] != "SUCCEEDED":
rm.inc("notSuccApp",1)
#end update
appid = app["id"]
appRecord = self.getAppidRecord(appid)
keyFromApp = ["user","name","queue","startedTime","finishedTime","state","finalStatus"]
for key in keyFromApp:
if key == "startedTime" or key == "finishedTime":
appRecord.set(key,util.getSecondTime(app[key]))
else:
appRecord.set(key,app[key])
#todo
appRecord.set("attemptNumber",1)
keyFromHistory = ["mapsTotal","mapsCompleted","successfulMapAttempts",
"killedMapAttempts","failedMapAttempts","avgMapTime",
"reducesTotal","reducesCompleted","successfulReduceAttempts",
"killedReduceAttempts","failedReduceAttempts","avgReduceTime"]
if jobHistory.has_key('job'):
for key in keyFromHistory:
appRecord.set(key,jobHistory['job'][key])
#TODO "localMap","rackMap"
keyMapFromCounters = {"DATA_LOCAL_MAPS":"localMap","RACK_LOCAL_MAPS":"rackMap",
"FILE_BYTES_READ":"fileRead","FILE_BYTES_WRITTEN":"fileWrite",
"HDFS_BYTES_READ":"hdfsRead","HDFS_BYTES_WRITTEN":"hdfsWrite"}
if jobCounter:
for (key,value) in keyMapFromCounters.items():
appRecord.set(value,jobCounter[key]['total'])
