def save(self):
"""
_save_
Pack the data object into an IMProvNode
"""
output = None
try:
try:
config = os.environ.get("PRODAGENT_CONFIG", None)
if config == None:
msg = "No ProdAgent Config file provided\n"
raise RuntimeError, msg
cfgObject = ProdAgentConfiguration()
cfgObject.loadFromFile(config)
alertHandlerConfig = cfgObject.get("AlertHandler")
workingDir = alertHandlerConfig["ComponentDir"]
dir = os.path.join(os.path.expandvars(workingDir), "Alerts")
if not os.path.exists(dir):
os.makedirs(dir)
self.FileName = os.path.join(dir, "alert-%s.dat" % makeUUID())
output = open(self.FileName, "wb")
pickle.dump(self, output)
except Exception, ex:
# to do: Exception handling
print ex
raise RuntimeError, str(ex)
finally:
if output:
output.close()
return
def save(self):
"""
_save_
Pack the data object into an IMProvNode
"""
output = None
try:
try:
config = os.environ.get("PRODAGENT_CONFIG", None)
if config == None:
msg = "No ProdAgent Config file provided\n"
raise RuntimeError, msg
cfgObject = ProdAgentConfiguration()
cfgObject.loadFromFile(config)
alertHandlerConfig = cfgObject.get("AlertHandler")
workingDir = alertHandlerConfig['ComponentDir']
dir = os.path.join(os.path.expandvars(workingDir), 'Alerts')
if not os.path.exists(dir):
os.makedirs(dir)
self.FileName = os.path.join(dir, "alert-%s.dat" % makeUUID())
output = open(self.FileName, 'wb')
pickle.dump(self, output)
except Exception, ex:
# to do: Exception handling
print ex
raise RuntimeError, str(ex)
finally:
if output:
output.close()
return
def createCleanupJobSpec(workflowSpec, site, *lfns):
"""
_createCleanupJob_
Create a Cleanup JobSpec definition, using the cleanup
workflow template, site name and the list of LFNs to be
removed
"""
jobSpec = workflowSpec.createJobSpec()
jobName = "%s-%s" % (workflowSpec.workflowName(), makeUUID())
jobSpec.setJobName(jobName)
jobSpec.setJobType("CleanUp")
jobSpec.addWhitelistSite(site)
lfnList = ""
for lfn in lfns:
lfnList += "%s\n" % lfn
jobSpec.payload.configuration = lfnList
return jobSpec
def newConvJobID():
"""
_newConvJobID_
Return a new conversion job ID
"""
return "Conversion-%s" % makeUUID()
def newRepackMergeJobID(run):
"""
_newRepackMergeJobID_
Return a new repacker merge job ID from the DB that is associated
to a given run
"""
return "RepackerMerge-Run%s-%s" %(run, makeUUID())
def newRepackJobID(run):
"""
_newRepackJobID_
Get a new repacker job ID from the DB that is associated
to a given run
"""
return "Repacker-Run%s-%s" %(run, makeUUID())
def createLogCollectorJobSpec(workflowSpec, originalWf, site, lfnBase,
stageOutParams, *lfns):
"""
createLogCollectorJobSpec
Create a LogArchive JobSpec definition, using the LogArchive
workflow template, site name and the list of LFNs to be
removed
"""
jobSpec = workflowSpec.createJobSpec()
jobName = "%s-%s" % (workflowSpec.workflowName(), makeUUID())
jobSpec.setJobName(jobName)
jobSpec.setJobType("LogCollect")
jobSpec.addWhitelistSite(site)
confNode = IMProvNode("LogCollectorConfig")
# add site and workflow to collect
confNode.addNode(IMProvNode("wf", originalWf))
confNode.addNode(IMProvNode("se", site))
confNode.addNode(IMProvNode("lfnBase", lfnBase))
# add logs to collect
logNode = IMProvNode("LogsToCollect")
for lfn in lfns:
logNode.addNode(IMProvNode("lfn", lfn))
confNode.addNode(logNode)
# stageout
if stageOutParams:
stageOutNode = IMProvNode("Override")
# WorkflowTools.addStageOutOverride(confNode, stageOutParams['command'],
# stageOutParams['option'],
# stageOutParams['se-name'],
# stageOutParams['lfnPrefix'])
stageOutNode.addNode(IMProvNode("command", stageOutParams['command']))
stageOutNode.addNode(IMProvNode("option", stageOutParams['option']))
stageOutNode.addNode(IMProvNode("se-name", stageOutParams['se-name']))
stageOutNode.addNode(
IMProvNode("lfn-prefix", stageOutParams['lfnPrefix']))
confNode.addNode(stageOutNode)
#jobSpec.payload.configuration = logNode.makeDOMElement().toprettyxml()
jobSpec.payload.configuration = confNode.makeDOMElement().toprettyxml()
return jobSpec
def createLogCollectorJobSpec(workflowSpec, originalWf, site, lfnBase, stageOutParams, *lfns):
"""
createLogCollectorJobSpec
Create a LogArchive JobSpec definition, using the LogArchive
workflow template, site name and the list of LFNs to be
removed
"""
jobSpec = workflowSpec.createJobSpec()
jobName = "%s-%s" % (workflowSpec.workflowName(), makeUUID())
jobSpec.setJobName(jobName)
jobSpec.setJobType("LogCollect")
jobSpec.addWhitelistSite(site)
confNode = IMProvNode("LogCollectorConfig")
# add site and workflow to collect
confNode.addNode(IMProvNode("wf", originalWf))
confNode.addNode(IMProvNode("se", site))
confNode.addNode(IMProvNode("lfnBase", lfnBase))
# add logs to collect
logNode = IMProvNode("LogsToCollect")
for lfn in lfns:
logNode.addNode(IMProvNode("lfn", lfn))
confNode.addNode(logNode)
# stageout
if stageOutParams:
stageOutNode = IMProvNode("Override")
# WorkflowTools.addStageOutOverride(confNode, stageOutParams['command'],
# stageOutParams['option'],
# stageOutParams['se-name'],
# stageOutParams['lfnPrefix'])
stageOutNode.addNode(IMProvNode("command", stageOutParams["command"]))
stageOutNode.addNode(IMProvNode("option", stageOutParams["option"]))
stageOutNode.addNode(IMProvNode("se-name", stageOutParams["se-name"]))
stageOutNode.addNode(IMProvNode("lfn-prefix", stageOutParams["lfnPrefix"]))
confNode.addNode(stageOutNode)
# jobSpec.payload.configuration = logNode.makeDOMElement().toprettyxml()
jobSpec.payload.configuration = confNode.makeDOMElement().toprettyxml()
return jobSpec
def __init__(self, t0astFile=None, **args):
"""
___init___
Initialize all attributes.
If T0ASTFile is passed as parameter, create the block instance
using T0ASTFile by extracting necessary information
from given T0ASTFile and assign to a Block instance
"""
dict.__init__(self)
self.setdefault("BLOCK_ID", None)
self.setdefault("STATUS", "Active")
self.setdefault("MIGRATE_STATUS", "NotMigrated")
if t0astFile != None:
# full path used in dbs
self.setdefault("BLOCK_NAME", "/%s/%s/%s#%s" %
(t0astFile["PRIMARY_DATASET"],
t0astFile["PROCESSED_DATASET"],
t0astFile["DATA_TIER"],
makeUUID())
)
self.setdefault("RUN_ID", t0astFile.getRunID())
self.setdefault("DATASET_ID", t0astFile["DATASET_ID"])
self.setdefault("DATASET_PATH_ID", t0astFile.getDatasetPathID())
self.setdefault("DATA_TIER", t0astFile["DATA_TIER"])
self.setdefault("BLOCKSIZE", t0astFile["FILESIZE"])
self.setdefault("FILECOUNT", 1)
else:
self.setdefault("BLOCK_NAME", None)
self.setdefault("RUN_ID", None)
self.setdefault("DATASET_ID", None)
self.setdefault("DATASET_PATH_ID", None)
self.setdefault("DATA_TIER", None)
self.setdefault("BLOCKSIZE", 0) # byte
self.setdefault("FILECOUNT", 0)
self.update(args)
def createCleanupJobSpec(workflowSpec, site, *lfns):
"""
_createCleanupJob_
Create a Cleanup JobSpec definition, using the cleanup
workflow template, site name and the list of LFNs to be
removed
"""
jobSpec = workflowSpec.createJobSpec()
jobName = "%s-%s" % (workflowSpec.workflowName(), makeUUID())
jobSpec.setJobName(jobName)
jobSpec.setJobType("CleanUp")
jobSpec.addWhitelistSite(site)
lfnList = ""
for lfn in lfns:
lfnList += "%s\n" % lfn
jobSpec.payload.configuration = lfnList
return jobSpec
def makeWorkflow(self):
"""
_makeWorkflow_
Call this method to create the workflow spec instance when
done
"""
self._Validate()
# //
# // Add Stage Out node
#//
self.saveOutputFor.append(self.cmsRunNode.name)
WorkflowTools.addStageOutNode(self.cmsRunNode, "stageOut1",
*self.saveOutputFor)
WorkflowTools.addLogArchNode(self.cmsRunNode, "logArchive")
# //
# // Input Dataset?
#//
if self.inputDataset['IsUsed']:
inputDataset = self.cmsRunNodes[0].addInputDataset(
self.inputDataset['Primary'], self.inputDataset['Processed'])
inputDataset["DataTier"] = self.inputDataset['DataTier']
for keyname in [
'SplitType',
'SplitSize',
'OnlySites',
'OnlyBlocks',
'OnlyClosedBlocks',
]:
if self.inputDataset[keyname] != None:
self.workflow.parameters[keyname] = self.inputDataset[
keyname]
# //
# // Pileup Datasets?
#//
for pileupDataset in self.pileupDatasets:
puDataset = self.cmsRunNodes[0].addPileupDataset(
pileupDataset['Primary'], pileupDataset['DataTier'],
pileupDataset['Processed'])
puDataset['FilesPerJob'] = pileupDataset['FilesPerJob']
if pileupDataset['TargetModule'] is not None:
puDataset['TargetModule'] = pileupDataset['TargetModule']
# //
# // Extract dataset info from cfg
#//
datasets = {}
datasetsToForward = {}
for cmsRunNode, config in zip(self.cmsRunNodes, self.configurations):
# Ignore nodes that don't save any output. But keep input dataset
# in case we need to forward it.
if cmsRunNode.name not in self.saveOutputFor:
# Store parent dataset in case we need to forward it.
if self.inputDataset['IsUsed'] and \
cmsRunNode == self.cmsRunNodes[0]:
datasetsToForward[cmsRunNode.name] = \
self.inputDataset['DatasetName']
elif cmsRunNode != self.cmsRunNodes[0]:
for inputLink in cmsRunNode._InputLinks:
# If the previous cmsRunNode stages out, pull down the
# dataset it produced.
if not inputLink["AppearStandalone"]:
# TODO: Wont work if more than one InputLink exists
datasetsToForward[cmsRunNode.name] = \
datasets['%s:%s' % (inputLink['InputNode'],
inputLink['OutputModule'])]
# If the previous cmsRunNode does not stage out, then
# use it's parent.
else:
# TODO: Wont work if more than one InputLink exists
datasetsToForward[cmsRunNode.name] = \
datasetsToForward[inputLink['InputNode']]
continue
for outModName in config.outputModules.keys():
moduleInstance = config.getOutputModule(outModName)
dataTier = moduleInstance['dataTier']
filterName = moduleInstance["filterName"]
primaryName = DatasetConventions.primaryDatasetName(
PhysicsChannel=self.channel, )
if self.useProperNamingConventions:
if self.processingString and filterName:
processingString = "_".join(
(self.processingString, filterName))
elif self.processingString:
processingString = self.processingString
elif filterName:
processingString = filterName
else:
processingString = None
processedName = DatasetConventions.properProcessedDatasetName(
AcquisitionEra=self.acquisitionEra,
ProcessingString=processingString,
ProcessingVersion=self.processingVersion,
Unmerged=True)
elif self.acquisitionEra == None:
processedName = DatasetConventions.processedDatasetName(
Version=cmsRunNode.application['Version'],
Label=self.label,
Group=self.group,
FilterName=filterName,
RequestId=self.requestId,
Unmerged=True)
else:
processedName = DatasetConventions.csa08ProcessedDatasetName(
AcquisitionEra=self.acquisitionEra,
Conditions=self.workflow.parameters['Conditions'],
ProcessingVersion=self.workflow.
parameters['ProcessingVersion'],
FilterName=filterName,
Unmerged=True)
dataTier = DatasetConventions.checkDataTier(dataTier)
moduleInstance['primaryDataset'] = primaryName
moduleInstance['processedDataset'] = processedName
outDS = cmsRunNode.addOutputDataset(primaryName, processedName,
outModName)
outDS['Status'] = self.outputDatasetStatus
outDS['DataTier'] = dataTier
outDS["ApplicationName"] = \
cmsRunNode.application["Executable"]
outDS["ApplicationFamily"] = outModName
outDS["PhysicsGroup"] = self.group
# check for input dataset for first node
if self.inputDataset[
'IsUsed'] and cmsRunNode == self.cmsRunNodes[0]:
outDS['ParentDataset'] = self.inputDataset['DatasetName']
# check for staged out intermediates
elif cmsRunNode != self.cmsRunNodes[0]:
for inputLink in cmsRunNode._InputLinks:
if not inputLink["AppearStandalone"]:
# TODO: Wont work if more than one InputLink exists
outDS['ParentDataset'] = datasets[
'%s:%s' % (inputLink['InputNode'],
inputLink['OutputModule'])]
elif datasetsToForward.get(
inputLink['InputNode']) is not None:
outDS['ParentDataset'] = \
datasetsToForward[inputLink['InputNode']]
if self.options['FakeHash']:
guid = makeUUID()
outDS['PSetHash'] = "hash=%s;guid=%s" % \
(self.psetHashes[cmsRunNode.name], guid)
else:
outDS['PSetHash'] = self.psetHashes[cmsRunNode.name]
# record output in case used as input to a later node
datasets['%s:%s' % (cmsRunNode.name, outModName)] = \
"/%s/%s/%s" % ( outDS['PrimaryDataset'],
outDS['ProcessedDataset'],
outDS['DataTier'])
# optionally remap sibling relationships to parent-child (i.e HLTDEBUG)
remapParentageForWorkflow(self.workflow)
WorkflowTools.generateFilenames(self.workflow)
return self.workflow
def createSuccessReport(self, jobSpecLoaded, workerNodeInfo, reportFilePath):
"""
_createSuccessReport_
Create a job report representing the successful completion
of a job.
The jobSpecLoaded parameter is a reference to an instance
of the JobSpec class that has been initialized with the
job spec that we are generating a report for.
"""
jobSpecPayload, newReport = \
self.__fwkJobReportCommon(jobSpecLoaded, workerNodeInfo)
newReport.exitCode = 0
newReport.status = "Success"
if "jobId" in jobSpecLoaded.parameters.keys():
newReport.jobSpecId = jobSpecLoaded.parameters["jobId"]
# Create a list of datasets from the JobSpec
# then associate file to these later on
datasets = getOutputDatasetDetails(jobSpecPayload)
datasets.extend(getSizeBasedMergeDatasetsFromNode(jobSpecPayload))
outModules = jobSpecPayload.cfgInterface.outputModules
inputFiles = jobSpecPayload.cfgInterface.inputFiles
for dataset in datasets:
modName = dataset.get('OutputModuleName', None)
if outModules.has_key(modName):
dataset['LFNBase'] = outModules[modName].get('LFNBase', None)
self.setDefaultForNoneValue('LFNBase', dataset['LFNBase'])
dataset['MergedLFNBase'] = \
outModules[modName].get('MergedLFNBase', None)
datasetMap = {}
for dataset in datasets:
datasetMap[dataset['OutputModuleName']] = dataset
for outName, outMod in \
jobSpecPayload.cfgInterface.outputModules.items():
theFile = newReport.newFile()
guid = makeUUID()
if outMod.has_key("LFNBase"):
theFile['LFN'] = "%s/%s.root" % (outMod['LFNBase'], guid)
else:
theFile['LFN'] = "/some/madeup/path/%s.root" % guid
self.setDefaultForNoneValue('LFNBase', theFile['LFN'])
theFile['PFN'] ="fakefile:%s" % theFile['LFN']
theFile['GUID'] = guid
theFile['MergedBySize'] = choice(["True", "False"])
theFile['ModuleLabel'] = outName
# basic measurement is byte (minumum 4MB, max 4GB)
theFile['Size'] = 4000000 * randrange(1, 1000)
runNum = jobSpecLoaded.parameters["RunNumber"]
# need to get lumi
lumiList = jobSpecLoaded.parameters.get("LumiSections", [])
theFile.runs[runNum] = RunInfo(runNum, *lumiList)
#check if the maxEvents['output'] is set if not set totalEvent using maxEvents['input']
totalEvent = jobSpecPayload.cfgInterface.maxEvents['output']
if totalEvent == None:
totalEvent = jobSpecPayload.cfgInterface.maxEvents['input']
# if there is no input and output, print out error message and set default to 1000
totalEvent = self.setDefaultForNoneValue(
"maxEvent['input' and 'output']",
totalEvent,
100)
try:
totalEvent = int(totalEvent)
except ValueError, ex:
logging.error("totalEvent is not a number. \n%s" % ex)
# event size should be >= 0
# totalEvent is -1 process all event
if totalEvent < 0:
totalEvent = 200
if (random() > self.avgEventProcessingRate):
# Gauss distribution of totalEvent.
meanEvent = int(totalEvent * 0.7)
stdDev = totalEvent * 0.15
tempTotalEvent = int(gauss(meanEvent,stdDev))
if tempTotalEvent <= 0 :
totalEvent = 1
elif tempTotalEvent >= totalEvent:
totalEvent = totalEvent - 1
else:
totalEvent = tempTotalEvent
#logging.debug("---------- Total Event ----------: %s \n" % totalEvent)
theFile['TotalEvents'] = totalEvent
theFile['SEName'] = workerNodeInfo['se-name']
theFile['CEname'] = workerNodeInfo['ce-name']
theFile['Catalog'] = outMod['catalog']
theFile['Stream'] = outMod['stream']
theFile['OutputModuleClass'] = "PoolOutputModule"
theFile.addChecksum("cksum", randrange(1000000, 10000000))
theFile.branches.extend(["fakeBranch_%d-%s.Rec" % (num, guid)
for num in range(randrange(5,20))])
#theFile.load(theFile.save())
theFile["BranchHash"] = randrange(2000000, 30000000)
[ theFile.addInputFile("fakefile:%s" % x , "%s" % x )
for x in inputFiles ]
if datasetMap.has_key(outName):
datasetForFile = theFile.newDataset()
datasetForFile.update(datasetMap[outName])
def __call__(self, *fileList):
jobSpec = self.spec.createJobSpec()
jobId = "%s-%s" % (self.spec.workflowName(), self.count)
jobSpec.setJobName(jobId)
jobSpec.setJobType("Merge")
jobSpec.addWhitelistSite("storage.element.edu")
# get PSet
cfg = jobSpec.payload.cfgInterface
# set output module
#print jobSpec.payload
# set output file name
prim = self.dataset['PrimaryDataset']
tier = self.dataset['DataTier']
lastBit = self.dataset['ProcessedDataset']
acqEra = None
#if .has_key("AcquisitionEra"):
acqEra = jobSpec.parameters.get("AcquisitionEra", None)
# compute LFN group based on merge jobs counter
group = str(self.count // 1000).zfill(4)
jobSpec.parameters['RunNumber'] = self.spec.workflowRunNumber()
remainingBits = lastBit
if acqEra != None:
thingtoStrip = "%s_" % acqEra
mypieces = lastBit.split(thingtoStrip, 1)
if len(mypieces) > 1:
remainingBits = mypieces[1].split("-unmerged", 1)[0]
else:
remainingBits=lastBit
outModule = cfg.outputModules['Merged']
lfnBase = outModule['LFNBase']
extendedlfnBase = os.path.join(lfnBase, prim, tier, remainingBits,
group)
baseFileName = "%s.root" % makeUUID()
outModule['fileName'] = baseFileName
outModule['logicalFileName'] = os.path.join(extendedlfnBase,
baseFileName)
# set output catalog
outModule['catalog'] = "%s-merge.xml" % jobId
# set input module
# get input file names (expects a trivial catalog on site)
cfg.inputFiles = ["%s" % fileName for fileName in fileList]
# target file name
mergeJobSpecFile = "%s/%s-spec.xml" % (
self.dir, jobId)
# save job specification
jobSpec.save(mergeJobSpecFile)
self.count += 1
return jobSpec
def makeWorkflow(self):
"""
_makeWorkflow_
Call this method to create the workflow spec instance when
done
"""
self._Validate()
# //
# // Add Stage Out node
#//
self.saveOutputFor.append(self.cmsRunNode.name)
WorkflowTools.addStageOutNode(self.cmsRunNode,
"stageOut1", *self.saveOutputFor)
WorkflowTools.addLogArchNode(self.cmsRunNode, "logArchive")
# //
# // Input Dataset?
#//
if self.inputDataset['IsUsed']:
inputDataset = self.cmsRunNodes[0].addInputDataset(
self.inputDataset['Primary'],
self.inputDataset['Processed']
)
inputDataset["DataTier"] = self.inputDataset['DataTier']
for keyname in [
'SplitType',
'SplitSize',
'OnlySites',
'OnlyBlocks',
'OnlyClosedBlocks',
]:
if self.inputDataset[keyname] != None:
self.workflow.parameters[keyname] = self.inputDataset[keyname]
# //
# // Pileup Datasets?
#//
for pileupDataset in self.pileupDatasets:
puDataset = self.cmsRunNodes[0].addPileupDataset(
pileupDataset['Primary'],
pileupDataset['DataTier'],
pileupDataset['Processed'])
puDataset['FilesPerJob'] = pileupDataset['FilesPerJob']
if pileupDataset['TargetModule'] is not None:
puDataset['TargetModule'] = pileupDataset['TargetModule']
# //
# // Extract dataset info from cfg
#//
datasets = {}
datasetsToForward = {}
for cmsRunNode, config in zip(self.cmsRunNodes, self.configurations):
# Ignore nodes that don't save any output. But keep input dataset
# in case we need to forward it.
if cmsRunNode.name not in self.saveOutputFor:
# Store parent dataset in case we need to forward it.
if self.inputDataset['IsUsed'] and \
cmsRunNode == self.cmsRunNodes[0]:
datasetsToForward[cmsRunNode.name] = \
self.inputDataset['DatasetName']
elif cmsRunNode != self.cmsRunNodes[0]:
for inputLink in cmsRunNode._InputLinks:
# If the previous cmsRunNode stages out, pull down the
# dataset it produced.
if not inputLink["AppearStandalone"]:
# TODO: Wont work if more than one InputLink exists
datasetsToForward[cmsRunNode.name] = \
datasets['%s:%s' % (inputLink['InputNode'],
inputLink['OutputModule'])]
# If the previous cmsRunNode does not stage out, then
# use it's parent.
else:
# TODO: Wont work if more than one InputLink exists
datasetsToForward[cmsRunNode.name] = \
datasetsToForward[inputLink['InputNode']]
continue
for outModName in config.outputModules.keys():
moduleInstance = config.getOutputModule(outModName)
dataTier = moduleInstance['dataTier']
filterName = moduleInstance["filterName"]
primaryName = DatasetConventions.primaryDatasetName(
PhysicsChannel = self.channel,
)
if self.useProperNamingConventions:
if self.processingString and filterName:
processingString = "_".join((self.processingString, filterName))
elif self.processingString:
processingString = self.processingString
elif filterName:
processingString = filterName
else:
processingString = None
processedName = DatasetConventions.properProcessedDatasetName(
AcquisitionEra = self.acquisitionEra,
ProcessingString = processingString,
ProcessingVersion = self.processingVersion,
Unmerged = True
)
elif self.acquisitionEra == None:
processedName = DatasetConventions.processedDatasetName(
Version = cmsRunNode.application['Version'],
Label = self.label,
Group = self.group,
FilterName = filterName,
RequestId = self.requestId,
Unmerged = True
)
else:
processedName = DatasetConventions.csa08ProcessedDatasetName(
AcquisitionEra = self.acquisitionEra,
Conditions = self.workflow.parameters['Conditions'],
ProcessingVersion = self.workflow.parameters['ProcessingVersion'],
FilterName = filterName,
Unmerged = True
)
dataTier = DatasetConventions.checkDataTier(dataTier)
moduleInstance['primaryDataset'] = primaryName
moduleInstance['processedDataset'] = processedName
outDS = cmsRunNode.addOutputDataset(primaryName,
processedName,
outModName)
outDS['Status'] = self.outputDatasetStatus
outDS['DataTier'] = dataTier
outDS["ApplicationName"] = \
cmsRunNode.application["Executable"]
outDS["ApplicationFamily"] = outModName
outDS["PhysicsGroup"] = self.group
# check for input dataset for first node
if self.inputDataset['IsUsed'] and cmsRunNode == self.cmsRunNodes[0]:
outDS['ParentDataset'] = self.inputDataset['DatasetName']
# check for staged out intermediates
elif cmsRunNode != self.cmsRunNodes[0]:
for inputLink in cmsRunNode._InputLinks:
if not inputLink["AppearStandalone"]:
# TODO: Wont work if more than one InputLink exists
outDS['ParentDataset'] = datasets['%s:%s' % (inputLink['InputNode'],
inputLink['OutputModule'])]
elif datasetsToForward.get(
inputLink['InputNode']) is not None:
outDS['ParentDataset'] = \
datasetsToForward[inputLink['InputNode']]
if self.options['FakeHash']:
guid = makeUUID()
outDS['PSetHash'] = "hash=%s;guid=%s" % \
(self.psetHashes[cmsRunNode.name], guid)
else:
outDS['PSetHash'] = self.psetHashes[cmsRunNode.name]
# record output in case used as input to a later node
datasets['%s:%s' % (cmsRunNode.name, outModName)] = \
"/%s/%s/%s" % ( outDS['PrimaryDataset'],
outDS['ProcessedDataset'],
outDS['DataTier'])
# optionally remap sibling relationships to parent-child (i.e HLTDEBUG)
remapParentageForWorkflow(self.workflow)
WorkflowTools.generateFilenames(self.workflow)
return self.workflow
def createSuccessReport(self, jobSpecLoaded, workerNodeInfo,
reportFilePath):
"""
_createSuccessReport_
Create a job report representing the successful completion
of a job.
The jobSpecLoaded parameter is a reference to an instance
of the JobSpec class that has been initialized with the
job spec that we are generating a report for.
"""
jobSpecPayload, newReport = \
self.__fwkJobReportCommon(jobSpecLoaded, workerNodeInfo)
newReport.exitCode = 0
newReport.status = "Success"
# parse newReport.jobSpecId (it should contain Job Name
# "Repack-Run%s-%s", "RepackMerge-Run%s-%s", "PromptReco-Run%s-%s"
specIDParts = newReport.jobSpecId.split('-')
tier0JobType = None
if len(specIDParts) != 3:
logging.debug("JobReport jobSpecID not in correct format for tier 0: %s" %
newReport.jobSpecId)
else:
# Job type should be one of "Repack", "RepackMerge", "PromptReco"
tier0JobType = specIDParts[0].strip()
if "jobId" in jobSpecLoaded.parameters.keys():
newReport.jobSpecId = jobSpecLoaded.parameters["jobId"]
# Create a list of datasets from the JobSpec
# then associate file to these later on
datasets = getOutputDatasetDetails(jobSpecPayload)
datasets.extend(getSizeBasedMergeDatasetsFromNode(jobSpecPayload))
outModules = jobSpecPayload.cfgInterface.outputModules
inputFiles = jobSpecPayload.cfgInterface.inputFiles
for dataset in datasets:
modName = dataset.get('OutputModuleName', None)
if outModules.has_key(modName):
dataset['LFNBase'] = outModules[modName].get('LFNBase', None)
self.setDefaultForNoneValue('LFNBase', dataset['LFNBase'])
dataset['MergedLFNBase'] = \
outModules[modName].get('MergedLFNBase', None)
datasetMap = {}
for dataset in datasets:
datasetMap[dataset['OutputModuleName']] = dataset
for outName, outMod in \
jobSpecPayload.cfgInterface.outputModules.items():
theFile = newReport.newFile()
guid = makeUUID()
theFile['GUID'] = guid
theFile['ModuleLabel'] = outName
runNum = jobSpecLoaded.parameters["RunNumber"]
# need to get lumi
lumiList = jobSpecLoaded.parameters.get("LumiSections", [])
theFile.runs[runNum] = RunInfo(runNum, *lumiList)
#check if the maxEvents['output'] iE s set if not set totalEvent using maxEvents['input']
totalEvent = jobSpecPayload.cfgInterface.maxEvents['output']
if totalEvent == None:
totalEvent = jobSpecPayload.cfgInterface.maxEvents['input']
# if there is no input and output, print out error message and set default to 1000
totalEvent = self.setDefaultForNoneValue(
"maxEvent['input' and 'output']",
totalEvent,
100)
try:
totalEvent = int(totalEvent)
except ValueError, ex:
logging.error("totalEvent is not a number. \n%s" % ex)
if (random() > self.avgEventProcessingRate):
# Gauss distribution of totalEvent.
meanEvent = int(totalEvent * 0.7)
stdDev = totalEvent * 0.15
tempTotalEvent = int(gauss(meanEvent,stdDev))
if tempTotalEvent <= 0 :
totalEvent = 1
elif tempTotalEvent >= totalEvent:
totalEvent = totalEvent - 1
else:
totalEvent = tempTotalEvent
#logging.debug("---------- Total Event ----------: %s \n" % totalEvent)
theFile['TotalEvents'] = totalEvent
theFile['SEName'] = workerNodeInfo['se-name']
theFile['CEname'] = workerNodeInfo['ce-name']
theFile['Catalog'] = outMod['catalog']
theFile['Stream'] = outMod['stream']
theFile['OutputModuleClass'] = "PoolOutputModule"
theFile.addChecksum("cksum", randrange(1000000, 10000000))
theFile.branches.extend(["fakeBranch_%d-%s.Rec" % (num, guid)
for num in range(randrange(5,20))])
#theFile.load(theFile.save())
theFile["BranchHash"] = randrange(2000000, 30000000)
[ theFile.addInputFile("fakefile:%s" % x , "%s" % x )
for x in inputFiles ]
if datasetMap.has_key(outName):
datasetForFile = theFile.newDataset()
datasetForFile.update(datasetMap[outName])
# basic measurement is byte (minumum 4MB, max 4GB)
# default value for the file size
# it should be overridden if the primary dataset exist.
# for the all other
theFile['Size'] = 4000000 * randrange(1, 1000) #random size
theFile['MergedBySize'] = choice(["True", "False"])
# setting up default LFN
if outMod.has_key("LFNBase"):
theFile['LFN'] = "%s%s.root" % (outMod['LFNBase'], guid)
else:
theFile['LFN'] = "/some/madeup/path/%s.root" % guid
self.setDefaultForNoneValue('LFNBase', theFile['LFN'])
if tier0JobType == "Repack":
# parse dataset name set the size according to the threshold
if len(theFile.dataset) == 0:
continue
datasetNameParts = theFile.dataset[0]["PrimaryDataset"].split('_')
# need to add sanity check
if self.thresholdForMerge > int(datasetNameParts[2]):
theFile['Size'] = 500000000 #(500 MG)
theFile['MergedBySize'] = "False"
else :
theFile['Size'] = 4000000000 #(4 G)
theFile['MergedBySize'] = "True"
#override LFN fro Merged file
theFile['LFN'] = "%s%s.root" % (outMod['MergedLFNBase'], guid)
elif tier0JobType == "RepackMerge":
theFile['Size'] = 4000000000 #(4 G)
theFile['MergedBySize'] = "True"
elif tier0JobType == "PromptReco":
theFile['Size'] = 2000000000 #(2 G)
else :
theFile['Size'] = 4000000 * randrange(1, 1000) #random size
theFile['PFN'] ="fakefile:%s" % theFile['LFN']