def setUp(self):
"""
_setUp_
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(
customModules=["WMComponent.DBS3Buffer", "T0.WMBS"])
self.splitterFactory = SplitterFactory(package="T0.JobSplitting")
myThread = threading.currentThread()
daoFactory = DAOFactory(package="T0.WMBS",
logger=logging,
dbinterface=myThread.dbi)
myThread.dbi.processData("""INSERT INTO wmbs_location
(id, site_name, state, state_time)
VALUES (1, 'SomeSite', 1, 1)
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_pnns
(id, pnn)
VALUES (2, 'SomePNN')
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_location_pnns
(location, pnn)
VALUES (1, 2)
""",
transaction=False)
insertRunDAO = daoFactory(classname="RunConfig.InsertRun")
insertRunDAO.execute(binds={
'RUN': 1,
'HLTKEY': "someHLTKey"
},
transaction=False)
insertLumiDAO = daoFactory(classname="RunConfig.InsertLumiSection")
for lumi in [1, 2]:
insertLumiDAO.execute(binds={
'RUN': 1,
'LUMI': lumi
},
transaction=False)
insertStreamDAO = daoFactory(classname="RunConfig.InsertStream")
insertStreamDAO.execute(binds={'STREAM': "Express"}, transaction=False)
insertStreamFilesetDAO = daoFactory(
classname="RunConfig.InsertStreamFileset")
insertStreamFilesetDAO.execute(1, "Express", "TestFileset1")
self.fileset1 = Fileset(name="TestFileset1")
self.fileset1.load()
workflow1 = Workflow(spec="spec.xml",
owner="hufnagel",
name="TestWorkflow1",
task="Test")
workflow1.create()
self.subscription1 = Subscription(fileset=self.fileset1,
workflow=workflow1,
split_algo="Express",
type="Express")
self.subscription1.create()
# keep for later
self.insertClosedLumiDAO = daoFactory(
classname="RunLumiCloseout.InsertClosedLumi")
self.releaseExpressDAO = daoFactory(
classname="Tier0Feeder.ReleaseExpress")
self.currentTime = int(time.time())
return
def testPromptReco(self):
"""
_testPromptReco_
Create a Prompt Reconstruction workflow
and verify it installs into WMBS correctly.
"""
testArguments = PromptRecoWorkloadFactory.getTestArguments()
testArguments["CouchURL"] = os.environ["COUCHURL"]
testArguments["EnableHarvesting"] = True
factory = PromptRecoWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction(
"TestWorkload", testArguments)
testWorkload.setSpecUrl("somespec")
testWorkload.setOwnerDetails("*****@*****.**", "T0")
testWMBSHelper = WMBSHelper(testWorkload,
"Reco",
"SomeBlock",
cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
recoWorkflow = Workflow(name="TestWorkload", task="/TestWorkload/Reco")
recoWorkflow.load()
self.assertEqual(len(recoWorkflow.outputMap.keys()),
len(testArguments["WriteTiers"]) + 1,
"Error: Wrong number of WF outputs in the Reco WF.")
goldenOutputMods = [
"write_RECO", "write_ALCARECO", "write_AOD", "write_DQM"
]
for goldenOutputMod in goldenOutputMods:
mergedOutput = recoWorkflow.outputMap[goldenOutputMod][0][
"merged_output_fileset"]
unmergedOutput = recoWorkflow.outputMap[goldenOutputMod][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
if goldenOutputMod != "write_ALCARECO":
self.assertEqual(
mergedOutput.name,
"/TestWorkload/Reco/RecoMerge%s/merged-Merged" %
goldenOutputMod,
"Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/Reco/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong: %s" %
unmergedOutput.name)
logArchOutput = recoWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = recoWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/Reco/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/Reco/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
alcaSkimWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/Reco/AlcaSkim")
alcaSkimWorkflow.load()
self.assertEqual(
len(alcaSkimWorkflow.outputMap.keys()),
len(testArguments["AlcaSkims"]) + 1,
"Error: Wrong number of WF outputs in the AlcaSkim WF.")
goldenOutputMods = []
for alcaProd in testArguments["AlcaSkims"]:
goldenOutputMods.append("ALCARECOStream%s" % alcaProd)
for goldenOutputMod in goldenOutputMods:
mergedOutput = alcaSkimWorkflow.outputMap[goldenOutputMod][0][
"merged_output_fileset"]
unmergedOutput = alcaSkimWorkflow.outputMap[goldenOutputMod][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(
mergedOutput.name,
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-Merged" %
goldenOutputMod, "Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/Reco/AlcaSkim/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong: %s" %
unmergedOutput.name)
logArchOutput = alcaSkimWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = alcaSkimWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/Reco/AlcaSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/Reco/AlcaSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
dqmWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Reco/RecoMergewrite_DQM/RecoMergewrite_DQMEndOfRunDQMHarvestMerged"
)
dqmWorkflow.load()
logArchOutput = dqmWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = dqmWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Reco/RecoMergewrite_DQM/RecoMergewrite_DQMEndOfRunDQMHarvestMerged/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Reco/RecoMergewrite_DQM/RecoMergewrite_DQMEndOfRunDQMHarvestMerged/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
goldenOutputMods = ["write_RECO", "write_AOD", "write_DQM"]
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/Reco/RecoMerge%s" %
goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(
mergedMergeOutput.name,
"/TestWorkload/Reco/RecoMerge%s/merged-Merged" %
goldenOutputMod, "Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedMergeOutput.name,
"/TestWorkload/Reco/RecoMerge%s/merged-Merged" %
goldenOutputMod, "Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Reco/RecoMerge%s/merged-logArchive" %
goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" %
logArchOutput.name)
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Reco/RecoMerge%s/merged-logArchive" %
goldenOutputMod, "Error: LogArchive output fileset is wrong.")
goldenOutputMods = []
for alcaProd in testArguments["AlcaSkims"]:
goldenOutputMods.append("ALCARECOStream%s" % alcaProd)
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s" %
goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(
len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs %d." %
len(mergeWorkflow.outputMap.keys()))
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(
mergedMergeOutput.name,
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-Merged" %
goldenOutputMod, "Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedMergeOutput.name,
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-Merged" %
goldenOutputMod, "Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-logArchive"
% goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" %
logArchOutput.name)
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-logArchive"
% goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-Reco-SomeBlock")
topLevelFileset.loadData()
recoSubscription = Subscription(fileset=topLevelFileset,
workflow=recoWorkflow)
recoSubscription.loadData()
self.assertEqual(recoSubscription["type"], "Processing",
"Error: Wrong subscription type.")
self.assertEqual(
recoSubscription["split_algo"], "EventBased",
"Error: Wrong split algorithm. %s" %
recoSubscription["split_algo"])
alcaRecoFileset = Fileset(
name="/TestWorkload/Reco/unmerged-write_ALCARECO")
alcaRecoFileset.loadData()
alcaSkimSubscription = Subscription(fileset=alcaRecoFileset,
workflow=alcaSkimWorkflow)
alcaSkimSubscription.loadData()
self.assertEqual(alcaSkimSubscription["type"], "Processing",
"Error: Wrong subscription type.")
self.assertEqual(
alcaSkimSubscription["split_algo"], "WMBSMergeBySize",
"Error: Wrong split algorithm. %s" %
alcaSkimSubscription["split_algo"])
mergedDQMFileset = Fileset(
name="/TestWorkload/Reco/RecoMergewrite_DQM/merged-Merged")
mergedDQMFileset.loadData()
dqmSubscription = Subscription(fileset=mergedDQMFileset,
workflow=dqmWorkflow)
dqmSubscription.loadData()
self.assertEqual(dqmSubscription["type"], "Harvesting",
"Error: Wrong subscription type.")
self.assertEqual(dqmSubscription["split_algo"], "Harvest",
"Error: Wrong split algo.")
unmergedOutputs = ["write_RECO", "write_AOD", "write_DQM"]
for unmergedOutput in unmergedOutputs:
unmergedDataTier = Fileset(name="/TestWorkload/Reco/unmerged-%s" %
unmergedOutput)
unmergedDataTier.loadData()
dataTierMergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/RecoMerge%s" % unmergedOutput)
dataTierMergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedDataTier,
workflow=dataTierMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(
mergeSubscription["split_algo"], "WMBSMergeBySize",
"Error: Wrong split algorithm. %s" %
mergeSubscription["split_algo"])
unmergedOutputs = []
for alcaProd in testArguments["AlcaSkims"]:
unmergedOutputs.append("ALCARECOStream%s" % alcaProd)
for unmergedOutput in unmergedOutputs:
unmergedAlcaSkim = Fileset(
name="/TestWorkload/Reco/AlcaSkim/unmerged-%s" %
unmergedOutput)
unmergedAlcaSkim.loadData()
alcaSkimMergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s" %
unmergedOutput)
alcaSkimMergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedAlcaSkim,
workflow=alcaSkimMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(
mergeSubscription["split_algo"], "WMBSMergeBySize",
"Error: Wrong split algorithm. %s" %
mergeSubscription["split_algo"])
goldenOutputMods = [
"write_RECO", "write_AOD", "write_DQM", "write_ALCARECO"
]
for goldenOutputMod in goldenOutputMods:
unmergedFileset = Fileset(name="/TestWorkload/Reco/unmerged-%s" %
goldenOutputMod)
unmergedFileset.loadData()
cleanupWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/RecoCleanupUnmerged%s" %
goldenOutputMod)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmergedFileset,
workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"],
"SiblingProcessingBased",
"Error: Wrong subscription type.")
goldenOutputMods = []
for alcaProd in testArguments["AlcaSkims"]:
goldenOutputMods.append("ALCARECOStream%s" % alcaProd)
for goldenOutputMod in goldenOutputMods:
unmergedFileset = Fileset(
name="/TestWorkload/Reco/AlcaSkim/unmerged-%s" %
goldenOutputMod)
unmergedFileset.loadData()
cleanupWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/AlcaSkim/AlcaSkimCleanupUnmerged%s" %
goldenOutputMod)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmergedFileset,
workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"],
"SiblingProcessingBased",
"Error: Wrong subscription type.")
recoLogCollect = Fileset(name="/TestWorkload/Reco/unmerged-logArchive")
recoLogCollect.loadData()
recoLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/Reco/LogCollect")
recoLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=recoLogCollect,
workflow=recoLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
alcaSkimLogCollect = Fileset(
name="/TestWorkload/Reco/AlcaSkim/unmerged-logArchive")
alcaSkimLogCollect.loadData()
alcaSkimLogCollectWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/AlcaSkim/AlcaSkimLogCollect")
alcaSkimLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=alcaSkimLogCollect,
workflow=alcaSkimLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
goldenOutputMods = ["write_RECO", "write_AOD", "write_DQM"]
for goldenOutputMod in goldenOutputMods:
recoMergeLogCollect = Fileset(
name="/TestWorkload/Reco/RecoMerge%s/merged-logArchive" %
goldenOutputMod)
recoMergeLogCollect.loadData()
recoMergeLogCollectWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Reco/RecoMerge%s/Reco%sMergeLogCollect" %
(goldenOutputMod, goldenOutputMod))
recoMergeLogCollectWorkflow.load()
logCollectSubscription = Subscription(
fileset=recoMergeLogCollect,
workflow=recoMergeLogCollectWorkflow)
logCollectSubscription.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
goldenOutputMods = []
for alcaProd in testArguments["AlcaSkims"]:
goldenOutputMods.append("ALCARECOStream%s" % alcaProd)
for goldenOutputMod in goldenOutputMods:
alcaSkimLogCollect = Fileset(
name=
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/merged-logArchive"
% goldenOutputMod)
alcaSkimLogCollect.loadData()
alcaSkimLogCollectWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Reco/AlcaSkim/AlcaSkimMerge%s/AlcaSkim%sMergeLogCollect"
% (goldenOutputMod, goldenOutputMod))
alcaSkimLogCollectWorkflow.load()
logCollectSubscription = Subscription(
fileset=alcaSkimLogCollect,
workflow=alcaSkimLogCollectWorkflow)
logCollectSubscription.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
dqmHarvestLogCollect = Fileset(
name=
"/TestWorkload/Reco/RecoMergewrite_DQM/RecoMergewrite_DQMEndOfRunDQMHarvestMerged/unmerged-logArchive"
)
dqmHarvestLogCollect.loadData()
dqmHarvestLogCollectWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Reco/RecoMergewrite_DQM/RecoMergewrite_DQMEndOfRunDQMHarvestMerged/RecoMergewrite_DQMMergedEndOfRunDQMHarvestLogCollect"
)
dqmHarvestLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=dqmHarvestLogCollect,
workflow=dqmHarvestLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
return
def _commonMonteCarloTest(self):
"""
Retrieve the workload from WMBS and test all its properties.
"""
prodWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/Production")
prodWorkflow.load()
self.assertEqual(len(prodWorkflow.outputMap.keys()), 3,
"Error: Wrong number of WF outputs.")
goldenOutputMods = ["OutputA", "OutputB"]
for goldenOutputMod in goldenOutputMods:
mergedOutput = prodWorkflow.outputMap[goldenOutputMod][0][
"merged_output_fileset"]
unmergedOutput = prodWorkflow.outputMap[goldenOutputMod][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(
mergedOutput.name,
"/TestWorkload/Production/ProductionMerge%s/merged-Merged" %
goldenOutputMod, "Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/Production/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = prodWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = prodWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/Production/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/Production/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Production/ProductionMerge%s" %
goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(
mergedMergeOutput.name,
"/TestWorkload/Production/ProductionMerge%s/merged-Merged" %
goldenOutputMod, "Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedMergeOutput.name,
"/TestWorkload/Production/ProductionMerge%s/merged-Merged" %
goldenOutputMod, "Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Production/ProductionMerge%s/merged-logArchive"
% goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" %
logArchOutput.name)
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Production/ProductionMerge%s/merged-logArchive"
% goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-Production-SomeBlock")
topLevelFileset.loadData()
prodSubscription = Subscription(fileset=topLevelFileset,
workflow=prodWorkflow)
prodSubscription.loadData()
self.assertEqual(prodSubscription["type"], "Production",
"Error: Wrong subscription type.")
self.assertEqual(prodSubscription["split_algo"], "EventBased",
"Error: Wrong split algo.")
for outputName in ["OutputA", "OutputB"]:
unmergedOutput = Fileset(
name="/TestWorkload/Production/unmerged-%s" % outputName)
unmergedOutput.loadData()
mergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Production/ProductionMerge%s" % outputName)
mergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedOutput,
workflow=mergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(
mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo: %s" %
mergeSubscription["split_algo"])
for outputName in ["OutputA", "OutputB"]:
unmerged = Fileset(name="/TestWorkload/Production/unmerged-%s" %
outputName)
unmerged.loadData()
cleanupWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Production/ProductionCleanupUnmerged%s" %
outputName)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmerged,
workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"],
"SiblingProcessingBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(
name="/TestWorkload/Production/unmerged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(
name="TestWorkload", task="/TestWorkload/Production/LogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=procLogCollect,
workflow=procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
for outputName in ["OutputA", "OutputB"]:
mergeLogCollect = Fileset(
name=
"/TestWorkload/Production/ProductionMerge%s/merged-logArchive"
% outputName)
mergeLogCollect.loadData()
mergeLogCollectWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Production/ProductionMerge%s/Production%sMergeLogCollect"
% (outputName, outputName))
mergeLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=mergeLogCollect,
workflow=mergeLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
def _createSubscriptionsInWMBS(self,
task,
fileset,
alternativeFilesetClose=False):
"""
__createSubscriptionsInWMBS_
Create subscriptions in WMBS for all the tasks in the spec. This
includes filesets, workflows and the output map for each task.
"""
# create runtime sandbox for workflow
self.createSandbox()
# FIXME: Let workflow put in values if spec is missing them
workflow = Workflow(
spec=self.wmSpec.specUrl(),
owner=self.wmSpec.getOwner()["name"],
dn=self.wmSpec.getOwner().get("dn", "unknown"),
group=self.wmSpec.getOwner().get("group", "unknown"),
owner_vogroup=self.wmSpec.getOwner().get("vogroup", "DEFAULT"),
owner_vorole=self.wmSpec.getOwner().get("vorole", "DEFAULT"),
name=self.wmSpec.name(),
task=task.getPathName(),
wfType=self.wmSpec.getDashboardActivity(),
alternativeFilesetClose=alternativeFilesetClose,
priority=self.wmSpec.priority())
workflow.create()
subscription = Subscription(fileset=fileset,
workflow=workflow,
split_algo=task.jobSplittingAlgorithm(),
type=task.getPrimarySubType())
if subscription.exists():
subscription.load()
msg = "Subscription %s already exists for %s (you may ignore file insertion messages below, existing files wont be duplicated)"
self.logger.info(msg % (subscription['id'], task.getPathName()))
else:
subscription.create()
for site in task.siteWhitelist():
subscription.addWhiteBlackList([{
"site_name": site,
"valid": True
}])
for site in task.siteBlacklist():
subscription.addWhiteBlackList([{
"site_name": site,
"valid": False
}])
if self.topLevelSubscription is None:
self.topLevelSubscription = subscription
logging.info("Top level subscription created: %s",
subscription["id"])
else:
logging.info("Child subscription created: %s", subscription["id"])
outputModules = task.getOutputModulesForTask()
ignoredOutputModules = task.getIgnoredOutputModulesForTask()
for outputModule in outputModules:
for outputModuleName in outputModule.listSections_():
if outputModuleName in ignoredOutputModules:
logging.info(
"IgnoredOutputModule set for %s, skipping fileset creation.",
outputModuleName)
continue
dataTier = getattr(getattr(outputModule, outputModuleName),
"dataTier", '')
filesetName = self.outputFilesetName(task, outputModuleName,
dataTier)
outputFileset = Fileset(filesetName)
outputFileset.create()
outputFileset.markOpen(True)
mergedOutputFileset = None
for childTask in task.childTaskIterator():
if childTask.data.input.outputModule == outputModuleName:
childDatatier = getattr(childTask.data.input,
'dataTier', '')
if childTask.taskType() in [
"Cleanup", "Merge"
] and childDatatier != dataTier:
continue
elif childTask.taskType(
) == "Merge" and childDatatier == dataTier:
filesetName = self.outputFilesetName(
childTask, "Merged", dataTier)
mergedOutputFileset = Fileset(filesetName)
mergedOutputFileset.create()
mergedOutputFileset.markOpen(True)
primaryDataset = getattr(
getattr(outputModule, outputModuleName),
"primaryDataset", None)
if primaryDataset is not None:
self.mergeOutputMapping[
mergedOutputFileset.id] = primaryDataset
self._createSubscriptionsInWMBS(
childTask, outputFileset, alternativeFilesetClose)
if mergedOutputFileset is None:
workflow.addOutput(outputModuleName + dataTier,
outputFileset, outputFileset)
else:
workflow.addOutput(outputModuleName + dataTier,
outputFileset, mergedOutputFileset)
return self.topLevelSubscription
def stuffWMBS(self):
"""
_stuffWMBS_
Insert some dummy jobs, jobgroups, filesets, files and subscriptions
into WMBS to test job creation. Three completed job groups each
containing several files are injected. Another incomplete job group is
also injected. Also files are added to the "Mergeable" subscription as
well as to the output fileset for their jobgroups.
"""
locationAction = self.daoFactory(classname="Locations.New")
changeStateDAO = self.daoFactory(classname="Jobs.ChangeState")
locationAction.execute("site1", pnn="T2_CH_CERN")
self.mergeFileset = Fileset(name="mergeFileset")
self.mergeFileset.create()
self.bogusFileset = Fileset(name="bogusFileset")
self.bogusFileset.create()
self.mergeMergedFileset = Fileset(name="mergeMergedFileset")
self.mergeMergedFileset.create()
self.bogusMergedFileset = Fileset(name="bogusMergedFileset")
self.bogusMergedFileset.create()
mergeWorkflow = Workflow(name="mergeWorkflow",
spec="bunk2",
owner="Steve",
task="Test")
mergeWorkflow.create()
self.mergeSubscription = Subscription(fileset=self.mergeFileset,
workflow=mergeWorkflow,
split_algo="SplitFileBased")
self.mergeSubscription.create()
self.bogusSubscription = Subscription(fileset=self.bogusFileset,
workflow=mergeWorkflow,
split_algo="SplitFileBased")
inputFileset = Fileset(name="inputFileset")
inputFileset.create()
inputWorkflow = Workflow(name="inputWorkflow",
spec="input",
owner="Steve",
task="Test")
inputWorkflow.create()
inputWorkflow.addOutput("someOutput", self.mergeFileset,
self.mergeMergedFileset)
inputWorkflow.addOutput("someOutput2", self.bogusFileset,
self.bogusMergedFileset)
inputSubscription = Subscription(fileset=inputFileset,
workflow=inputWorkflow)
inputSubscription.create()
parentFile1 = File(lfn="parentFile1")
parentFile1.create()
parentFile2 = File(lfn="parentFile2")
parentFile2.create()
parentFile3 = File(lfn="parentFile3")
parentFile3.create()
parentFile4 = File(lfn="parentFile4")
parentFile4.create()
jobGroup1 = JobGroup(subscription=inputSubscription)
jobGroup1.create()
jobGroup2 = JobGroup(subscription=inputSubscription)
jobGroup2.create()
testJob1 = Job()
testJob1.addFile(parentFile1)
testJob1.create(jobGroup1)
testJob1["state"] = "cleanout"
testJob1["oldstate"] = "new"
testJob1["couch_record"] = "somejive"
testJob1["retry_count"] = 0
testJob1["outcome"] = "success"
testJob1.save()
changeStateDAO.execute([testJob1])
testJob2 = Job()
testJob2.addFile(parentFile2)
testJob2.create(jobGroup1)
testJob2["state"] = "cleanout"
testJob2["oldstate"] = "new"
testJob2["couch_record"] = "somejive"
testJob2["retry_count"] = 0
testJob2["outcome"] = "success"
testJob2.save()
changeStateDAO.execute([testJob2])
testJob3 = Job()
testJob3.addFile(parentFile3)
testJob3.create(jobGroup2)
testJob3["state"] = "cleanout"
testJob3["oldstate"] = "new"
testJob3["couch_record"] = "somejive"
testJob3["retry_count"] = 0
testJob3["outcome"] = "success"
testJob3.save()
changeStateDAO.execute([testJob3])
testJob4 = Job()
testJob4.addFile(parentFile4)
testJob4.create(jobGroup2)
testJob4["state"] = "cleanout"
testJob4["oldstate"] = "new"
testJob4["couch_record"] = "somejive"
testJob4["retry_count"] = 0
testJob4["outcome"] = "failure"
testJob4.save()
changeStateDAO.execute([testJob4])
# We'll simulate a failed split by event job that the merger should
# ignore.
parentFile5 = File(lfn="parentFile5")
parentFile5.create()
testJob5 = Job()
testJob5.addFile(parentFile5)
testJob5.create(jobGroup2)
testJob5["state"] = "cleanout"
testJob5["oldstate"] = "new"
testJob5["couch_record"] = "somejive"
testJob5["retry_count"] = 0
testJob5["outcome"] = "success"
testJob5.save()
changeStateDAO.execute([testJob5])
testJob6 = Job()
testJob6.addFile(parentFile5)
testJob6.create(jobGroup2)
testJob6["state"] = "cleanout"
testJob6["oldstate"] = "new"
testJob6["couch_record"] = "somejive"
testJob6["retry_count"] = 0
testJob6["outcome"] = "failure"
testJob6.save()
changeStateDAO.execute([testJob6])
badFile1 = File(lfn="badFile1",
size=10241024,
events=10241024,
first_event=0,
locations=set(["T2_CH_CERN"]))
badFile1.addRun(Run(1, *[45]))
badFile1.create()
badFile1.addParent(parentFile5["lfn"])
file1 = File(lfn="file1",
size=1024,
events=1024,
first_event=0,
locations=set(["T2_CH_CERN"]))
file1.addRun(Run(1, *[45]))
file1.create()
file1.addParent(parentFile1["lfn"])
file2 = File(lfn="file2",
size=1024,
events=1024,
first_event=1024,
locations=set(["T2_CH_CERN"]))
file2.addRun(Run(1, *[45]))
file2.create()
file2.addParent(parentFile1["lfn"])
file3 = File(lfn="file3",
size=1024,
events=1024,
first_event=2048,
locations=set(["T2_CH_CERN"]))
file3.addRun(Run(1, *[45]))
file3.create()
file3.addParent(parentFile1["lfn"])
file4 = File(lfn="file4",
size=1024,
events=1024,
first_event=3072,
locations=set(["T2_CH_CERN"]))
file4.addRun(Run(1, *[45]))
file4.create()
file4.addParent(parentFile1["lfn"])
fileA = File(lfn="fileA",
size=1024,
events=1024,
first_event=0,
locations=set(["T2_CH_CERN"]))
fileA.addRun(Run(1, *[46]))
fileA.create()
fileA.addParent(parentFile2["lfn"])
fileB = File(lfn="fileB",
size=1024,
events=1024,
first_event=1024,
locations=set(["T2_CH_CERN"]))
fileB.addRun(Run(1, *[46]))
fileB.create()
fileB.addParent(parentFile2["lfn"])
fileC = File(lfn="fileC",
size=1024,
events=1024,
first_event=2048,
locations=set(["T2_CH_CERN"]))
fileC.addRun(Run(1, *[46]))
fileC.create()
fileC.addParent(parentFile2["lfn"])
fileI = File(lfn="fileI",
size=1024,
events=1024,
first_event=0,
locations=set(["T2_CH_CERN"]))
fileI.addRun(Run(2, *[46]))
fileI.create()
fileI.addParent(parentFile3["lfn"])
fileII = File(lfn="fileII",
size=1024,
events=1024,
first_event=1024,
locations=set(["T2_CH_CERN"]))
fileII.addRun(Run(2, *[46]))
fileII.create()
fileII.addParent(parentFile3["lfn"])
fileIII = File(lfn="fileIII",
size=1024,
events=1024,
first_event=2048,
locations=set(["T2_CH_CERN"]))
fileIII.addRun(Run(2, *[46]))
fileIII.create()
fileIII.addParent(parentFile3["lfn"])
fileIV = File(lfn="fileIV",
size=1024,
events=1024,
first_event=3072,
locations=set(["T2_CH_CERN"]))
fileIV.addRun(Run(2, *[46]))
fileIV.create()
fileIV.addParent(parentFile3["lfn"])
fileX = File(lfn="badFileA",
size=1024,
events=1024,
first_event=0,
locations=set(["T2_CH_CERN"]))
fileX.addRun(Run(1, *[47]))
fileX.create()
fileX.addParent(parentFile4["lfn"])
fileY = File(lfn="badFileB",
size=1024,
events=1024,
first_event=1024,
locations=set(["T2_CH_CERN"]))
fileY.addRun(Run(1, *[47]))
fileY.create()
fileY.addParent(parentFile4["lfn"])
fileZ = File(lfn="badFileC",
size=1024,
events=1024,
first_event=2048,
locations=set(["T2_CH_CERN"]))
fileZ.addRun(Run(1, *[47]))
fileZ.create()
fileZ.addParent(parentFile4["lfn"])
jobGroup1.output.addFile(file1)
jobGroup1.output.addFile(file2)
jobGroup1.output.addFile(file3)
jobGroup1.output.addFile(file4)
jobGroup1.output.addFile(fileA)
jobGroup1.output.addFile(fileB)
jobGroup1.output.addFile(fileC)
jobGroup1.output.commit()
jobGroup2.output.addFile(fileI)
jobGroup2.output.addFile(fileII)
jobGroup2.output.addFile(fileIII)
jobGroup2.output.addFile(fileIV)
jobGroup2.output.addFile(fileX)
jobGroup2.output.addFile(fileY)
jobGroup2.output.addFile(fileZ)
jobGroup2.output.addFile(badFile1)
jobGroup2.output.commit()
for file in [
file1, file2, file3, file4, fileA, fileB, fileC, fileI, fileII,
fileIII, fileIV, fileX, fileY, fileZ, badFile1
]:
self.mergeFileset.addFile(file)
self.bogusFileset.addFile(file)
self.mergeFileset.commit()
self.bogusFileset.commit()
return
def testTruncatedWFInsertion(self):
"""
_testTruncatedWFInsertion_
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
pnn='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
pnn='goodse2.cern.ch',
ceName='site2',
plugin="TestPlugin")
testWorkload = self.createTestWMSpec()
testTopLevelTask = getFirstTask(testWorkload)
testWMBSHelper = WMBSHelper(testWorkload,
testTopLevelTask.name(),
"SomeBlock",
cachepath=self.workDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testTopLevelTask, testWMBSHelper.topLevelFileset)
testWorkload.truncate("ResubmitTestWorkload",
"/TestWorkload/ProcessingTask/MergeTask",
"someserver", "somedatabase")
# create the subscription for multiple top task (MergeTask and CleanupTask for the same block)
for task in testWorkload.getTopLevelTask():
testResubmitWMBSHelper = WMBSHelper(testWorkload,
task.name(),
"SomeBlock2",
cachepath=self.workDir)
testResubmitWMBSHelper.createTopLevelFileset()
testResubmitWMBSHelper._createSubscriptionsInWMBS(
task, testResubmitWMBSHelper.topLevelFileset)
mergeWorkflow = Workflow(name="ResubmitTestWorkload",
task="/ResubmitTestWorkload/MergeTask")
mergeWorkflow.load()
self.assertEqual(mergeWorkflow.owner, "sfoulkes",
"Error: Wrong owner.")
self.assertEqual(
mergeWorkflow.spec,
os.path.join(self.workDir, mergeWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 1,
"Error: Wrong number of WF outputs.")
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"output_fileset"]
unmergedMergeOutput.loadData()
self.assertEqual(unmergedMergeOutput.name,
"/ResubmitTestWorkload/MergeTask/merged-Merged",
"Error: Unmerged output fileset is wrong.")
skimWorkflow = Workflow(
name="ResubmitTestWorkload",
task="/ResubmitTestWorkload/MergeTask/SkimTask")
skimWorkflow.load()
self.assertEqual(skimWorkflow.owner, "sfoulkes", "Error: Wrong owner.")
self.assertEqual(
skimWorkflow.spec,
os.path.join(self.workDir, skimWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(skimWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedSkimOutputA = skimWorkflow.outputMap["SkimOutputA"][0][
"merged_output_fileset"]
unmergedSkimOutputA = skimWorkflow.outputMap["SkimOutputA"][0][
"output_fileset"]
mergedSkimOutputB = skimWorkflow.outputMap["SkimOutputB"][0][
"merged_output_fileset"]
unmergedSkimOutputB = skimWorkflow.outputMap["SkimOutputB"][0][
"output_fileset"]
mergedSkimOutputA.loadData()
mergedSkimOutputB.loadData()
unmergedSkimOutputA.loadData()
unmergedSkimOutputB.loadData()
self.assertEqual(
mergedSkimOutputA.name,
"/ResubmitTestWorkload/MergeTask/SkimTask/unmerged-SkimOutputA",
"Error: Merged output fileset is wrong: %s" %
mergedSkimOutputA.name)
self.assertEqual(
unmergedSkimOutputA.name,
"/ResubmitTestWorkload/MergeTask/SkimTask/unmerged-SkimOutputA",
"Error: Unmerged output fileset is wrong.")
self.assertEqual(
mergedSkimOutputB.name,
"/ResubmitTestWorkload/MergeTask/SkimTask/unmerged-SkimOutputB",
"Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedSkimOutputB.name,
"/ResubmitTestWorkload/MergeTask/SkimTask/unmerged-SkimOutputB",
"Error: Unmerged output fileset is wrong.")
topLevelFileset = Fileset(
name="ResubmitTestWorkload-MergeTask-SomeBlock2")
topLevelFileset.loadData()
mergeSubscription = Subscription(fileset=topLevelFileset,
workflow=mergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(len(mergeSubscription.getWhiteBlackList()), 0,
"Error: Wrong white/black list for merge sub.")
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "WMBSMergeBySize",
"Error: Wrong split algo.")
skimSubscription = Subscription(fileset=unmergedMergeOutput,
workflow=skimWorkflow)
skimSubscription.loadData()
self.assertEqual(skimSubscription["type"], "Skim",
"Error: Wrong subscription type.")
self.assertEqual(skimSubscription["split_algo"], "FileBased",
"Error: Wrong split algo.")
return
def stuffWMBS(self, workflowURL, name):
"""
_stuffWMBS_
Insert some dummy jobs, jobgroups, filesets, files and subscriptions
into WMBS to test job creation. Three completed job groups each
containing several files are injected. Another incomplete job group is
also injected. Also files are added to the "Mergeable" subscription as
well as to the output fileset for their jobgroups.
"""
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute(siteName="s1", pnn="somese.cern.ch")
changeStateDAO = self.daoFactory(classname="Jobs.ChangeState")
mergeFileset = Fileset(name="mergeFileset")
mergeFileset.create()
bogusFileset = Fileset(name="bogusFileset")
bogusFileset.create()
mergeWorkflow = Workflow(spec=workflowURL,
owner="mnorman",
name=name,
task="/TestWorkload/ReReco")
mergeWorkflow.create()
mergeSubscription = Subscription(fileset=mergeFileset,
workflow=mergeWorkflow,
split_algo="ParentlessMergeBySize")
mergeSubscription.create()
bogusSubscription = Subscription(fileset=bogusFileset,
workflow=mergeWorkflow,
split_algo="ParentlessMergeBySize")
file1 = File(lfn="file1",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
file1.addRun(Run(1, *[45]))
file1.create()
file2 = File(lfn="file2",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
file2.addRun(Run(1, *[45]))
file2.create()
file3 = File(lfn="file3",
size=1024,
events=1024,
first_event=2048,
locations=set(["somese.cern.ch"]))
file3.addRun(Run(1, *[45]))
file3.create()
file4 = File(lfn="file4",
size=1024,
events=1024,
first_event=3072,
locations=set(["somese.cern.ch"]))
file4.addRun(Run(1, *[45]))
file4.create()
fileA = File(lfn="fileA",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
fileA.addRun(Run(1, *[46]))
fileA.create()
fileB = File(lfn="fileB",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
fileB.addRun(Run(1, *[46]))
fileB.create()
fileC = File(lfn="fileC",
size=1024,
events=1024,
first_event=2048,
locations=set(["somese.cern.ch"]))
fileC.addRun(Run(1, *[46]))
fileC.create()
fileI = File(lfn="fileI",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
fileI.addRun(Run(2, *[46]))
fileI.create()
fileII = File(lfn="fileII",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
fileII.addRun(Run(2, *[46]))
fileII.create()
fileIII = File(lfn="fileIII",
size=1024,
events=1024,
first_event=2048,
locations=set(["somese.cern.ch"]))
fileIII.addRun(Run(2, *[46]))
fileIII.create()
fileIV = File(lfn="fileIV",
size=1024 * 1000 * 1000,
events=1024,
first_event=3072,
locations=set(["somese.cern.ch"]))
fileIV.addRun(Run(2, *[46]))
fileIV.create()
for file in [
file1, file2, file3, file4, fileA, fileB, fileC, fileI, fileII,
fileIII, fileIV
]:
mergeFileset.addFile(file)
bogusFileset.addFile(file)
mergeFileset.commit()
bogusFileset.commit()
return
def testGetOutputMapDAO(self):
"""
_testGetOutputMapDAO_
Verify the proper behavior of the GetOutputMapDAO for a variety of
different processing chains.
"""
recoOutputFileset = Fileset(name = "RECO")
recoOutputFileset.create()
mergedRecoOutputFileset = Fileset(name = "MergedRECO")
mergedRecoOutputFileset.create()
alcaOutputFileset = Fileset(name = "ALCA")
alcaOutputFileset.create()
mergedAlcaOutputFileset = Fileset(name = "MergedALCA")
mergedAlcaOutputFileset.create()
dqmOutputFileset = Fileset(name = "DQM")
dqmOutputFileset.create()
mergedDqmOutputFileset = Fileset(name = "MergedDQM")
mergedDqmOutputFileset.create()
cleanupFileset = Fileset(name = "Cleanup")
cleanupFileset.create()
testWorkflow = Workflow(spec = "wf001.xml", owner = "Steve",
name = "TestWF", task = "None")
testWorkflow.create()
testWorkflow.addOutput("output", recoOutputFileset,
mergedRecoOutputFileset)
testWorkflow.addOutput("ALCARECOStreamCombined", alcaOutputFileset,
mergedAlcaOutputFileset)
testWorkflow.addOutput("DQM", dqmOutputFileset,
mergedDqmOutputFileset)
testWorkflow.addOutput("output", cleanupFileset)
testWorkflow.addOutput("ALCARECOStreamCombined", cleanupFileset)
testWorkflow.addOutput("DQM", cleanupFileset)
testRecoMergeWorkflow = Workflow(spec = "wf002.xml", owner = "Steve",
name = "TestRecoMergeWF", task = "None")
testRecoMergeWorkflow.create()
testRecoMergeWorkflow.addOutput("anything", mergedRecoOutputFileset,
mergedRecoOutputFileset)
testRecoProcWorkflow = Workflow(spec = "wf004.xml", owner = "Steve",
name = "TestRecoProcWF", task = "None")
testRecoProcWorkflow.create()
testAlcaChildWorkflow = Workflow(spec = "wf003.xml", owner = "Steve",
name = "TestAlcaChildWF", task = "None")
testAlcaChildWorkflow.create()
inputFile = File(lfn = "/path/to/some/lfn", size = 600000, events = 60000,
locations = "cmssrm.fnal.gov")
inputFile.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
testFileset.addFile(inputFile)
testFileset.commit()
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow,
split_algo = "EventBased",
type = "Processing")
testMergeRecoSubscription = Subscription(fileset = recoOutputFileset,
workflow = testRecoMergeWorkflow,
split_algo = "WMBSMergeBySize",
type = "Merge")
testProcRecoSubscription = Subscription(fileset = recoOutputFileset,
workflow = testRecoProcWorkflow,
split_algo = "FileBased",
type = "Processing")
testChildAlcaSubscription = Subscription(fileset = alcaOutputFileset,
workflow = testAlcaChildWorkflow,
split_algo = "FileBased",
type = "Processing")
testSubscription.create()
testMergeRecoSubscription.create()
testProcRecoSubscription.create()
testChildAlcaSubscription.create()
testSubscription.acquireFiles()
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
testJob = Job(name = "SplitJobA", files = [inputFile])
testJob.create(group = testJobGroup)
testJob["state"] = "complete"
testJob.save()
outputMapAction = self.daoFactory(classname = "Jobs.GetOutputMap")
outputMap = outputMapAction.execute(jobID = testJob["id"])
assert len(outputMap.keys()) == 3, \
"Error: Wrong number of outputs for primary workflow."
goldenMap = {"output": (recoOutputFileset.id,
mergedRecoOutputFileset.id),
"ALCARECOStreamCombined": (alcaOutputFileset.id,
mergedAlcaOutputFileset.id),
"DQM": (dqmOutputFileset.id,
mergedDqmOutputFileset.id)}
for outputID in outputMap.keys():
for outputFilesets in outputMap[outputID]:
if outputFilesets["merged_output_fileset"] == None:
self.assertEqual(outputFilesets["output_fileset"],
cleanupFileset.id,
"Error: Cleanup fileset is wrong.")
continue
self.assertTrue(outputID in goldenMap.keys(),
"Error: Output identifier is missing.")
self.assertEqual(outputFilesets["output_fileset"],
goldenMap[outputID][0],
"Error: Output fileset is wrong.")
self.assertEqual(outputFilesets["merged_output_fileset"],
goldenMap[outputID][1],
"Error: Merged output fileset is wrong.")
del goldenMap[outputID]
self.assertEqual(len(goldenMap.keys()), 0,
"Error: Missing output maps.")
return
def testGetFinishedWorkflows(self):
"""
_testGetFinishedWorkflows_
Test that we get only those workflows which are finished, that is, workflows where
all its subscriptions are finished and all other workflows with the same
spec are finished too
"""
owner = "no-one"
# Create a bunch of worklows with "different" specs and tasks
workflows = []
for i in range(0, 100):
scaledIndex = i % 10
testWorkflow = Workflow(spec="sp00%i" % scaledIndex,
owner=owner,
name="wf00%i" % scaledIndex,
task="task%i" % i)
testWorkflow.create()
workflows.append(testWorkflow)
# Everyone will use this fileset
testFileset = Fileset(name="TestFileset")
testFileset.create()
# Create subscriptions!
subscriptions = []
for workflow in workflows:
subscription = Subscription(fileset=testFileset,
workflow=workflow)
subscription.create()
subscriptions.append(subscription)
# Check that all workflows are NOT finished
myThread = threading.currentThread()
daoFactory = DAOFactory(package="WMCore.WMBS", logger=myThread.logger,
dbinterface=myThread.dbi)
getFinishedDAO = daoFactory(classname="Workflow.GetFinishedWorkflows")
result = getFinishedDAO.execute()
self.assertEqual(len(result), 0, "A workflow is incorrectly flagged as finished: %s" % str(result))
# Mark the first 50 subscriptions as finished
for idx, sub in enumerate(subscriptions):
if idx > 49:
break
sub.markFinished()
# No workflow is finished, none of them has all the subscriptions completed
result = getFinishedDAO.execute()
self.assertEqual(len(result), 0, "A workflow is incorrectly flagged as finished: %s" % str(result))
# Now finish all workflows in wf{000-5}
for idx, sub in enumerate(subscriptions):
if idx < 50 or idx % 10 > 5:
continue
sub.markFinished()
# Check the workflows
result = getFinishedDAO.execute()
self.assertEqual(len(result), 6, "A workflow is incorrectly flagged as finished: %s" % str(result))
# Check the overall structure of the workflows
for wf in result:
# Sanity checks on the results
# These are very specific checks and depends heavily on the names of task, spec and workflow
self.assertEqual(wf[2:], result[wf]['spec'][2:],
"A workflow has the wrong spec-name combination: %s" % str(wf))
self.assertTrue(int(wf[2:]) < 6,
"A workflow is incorrectly flagged as finished: %s" % str(wf))
self.assertEqual(len(result[wf]['workflows']), 10,
"A workflow has more tasks than it should: %s" % str(result[wf]))
for task in result[wf]['workflows']:
self.assertEqual(len(result[wf]['workflows'][task]), 1,
"A workflow has more subscriptions than it should: %s" % str(result[wf]))
return
def testFailJobInput(self):
"""
_testFailJobInput_
Test the Jobs.FailInput DAO and verify that it doesn't affect other
jobs/subscriptions that run over the same files.
"""
testWorkflow = Workflow(spec = "spec.xml", owner = "Steve",
name = "wf001", task="Test")
bogusWorkflow = Workflow(spec = "spec1.xml", owner = "Steve",
name = "wf002", task="Test")
testWorkflow.create()
bogusWorkflow.create()
testFileset = Fileset(name = "TestFileset")
bogusFileset = Fileset(name = "BogusFileset")
testFileset.create()
bogusFileset.create()
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow)
bogusSubscription = Subscription(fileset = bogusFileset,
workflow = bogusWorkflow)
testSubscription.create()
bogusSubscription.create()
testFileA = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileB = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileC = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileA.create()
testFileB.create()
testFileC.create()
testFileset.addFile([testFileA, testFileB, testFileC])
bogusFileset.addFile([testFileA, testFileB, testFileC])
testFileset.commit()
bogusFileset.commit()
testSubscription.completeFiles([testFileA, testFileB, testFileC])
bogusSubscription.acquireFiles([testFileA, testFileB, testFileC])
testJobGroup = JobGroup(subscription = testSubscription)
bogusJobGroup = JobGroup(subscription = bogusSubscription)
testJobGroup.create()
bogusJobGroup.create()
testJobA = Job(name = "TestJobA", files = [testFileA, testFileB, testFileC])
testJobB = Job(name = "TestJobB", files = [testFileA, testFileB, testFileC])
bogusJob = Job(name = "BogusJob", files = [testFileA, testFileB, testFileC])
testJobA.create(group = testJobGroup)
testJobB.create(group = testJobGroup)
bogusJob.create(group = bogusJobGroup)
testJobA.failInputFiles()
testJobB.failInputFiles()
self.assertEqual(len(testSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Acquired")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Failed")), 3)
self.assertEqual(len(testSubscription.filesOfStatus("Completed")), 0)
changeStateAction = self.daoFactory(classname = "Jobs.ChangeState")
testJobB["state"] = "cleanout"
changeStateAction.execute([testJobB])
# Try again
testJobA.failInputFiles()
# Should now be failed
self.assertEqual(len(testSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Acquired")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Failed")), 3)
self.assertEqual(len(testSubscription.filesOfStatus("Completed")), 0)
# bogus should be unchanged
self.assertEqual(len(bogusSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(bogusSubscription.filesOfStatus("Acquired")), 3)
self.assertEqual(len(bogusSubscription.filesOfStatus("Failed")), 0)
self.assertEqual(len(bogusSubscription.filesOfStatus("Completed")), 0)
return
def testCompleteJobInput(self):
"""
_testCompleteJobInput_
Verify the correct output of the CompleteInput DAO. This should mark
the input for a job as complete once all the jobs that run over a
particular file have complete successfully.
"""
testWorkflow = Workflow(spec = "spec.xml", owner = "Steve",
name = "wf001", task="Test")
bogusWorkflow = Workflow(spec = "spec1.xml", owner = "Steve",
name = "wf002", task="Test")
testWorkflow.create()
bogusWorkflow.create()
testFileset = Fileset(name = "TestFileset")
bogusFileset = Fileset(name = "BogusFileset")
testFileset.create()
bogusFileset.create()
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow)
bogusSubscription = Subscription(fileset = bogusFileset,
workflow = bogusWorkflow)
testSubscription.create()
bogusSubscription.create()
testFileA = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileB = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileC = File(lfn = makeUUID(), locations = "setest.site.ch")
testFileA.create()
testFileB.create()
testFileC.create()
testFileset.addFile([testFileA, testFileB, testFileC])
bogusFileset.addFile([testFileA, testFileB, testFileC])
testFileset.commit()
bogusFileset.commit()
testSubscription.acquireFiles([testFileA, testFileB, testFileC])
bogusSubscription.acquireFiles([testFileA, testFileB, testFileC])
testJobGroup = JobGroup(subscription = testSubscription)
bogusJobGroup = JobGroup(subscription = bogusSubscription)
testJobGroup.create()
bogusJobGroup.create()
testJobA = Job(name = "TestJobA", files = [testFileA])
testJobB = Job(name = "TestJobB", files = [testFileA, testFileB])
testJobC = Job(name = "TestJobC", files = [testFileC])
bogusJob = Job(name = "BogusJob", files = [testFileA, testFileB, testFileC])
testJobA.create(group = testJobGroup)
testJobB.create(group = testJobGroup)
testJobC.create(group = testJobGroup)
bogusJob.create(group = bogusJobGroup)
testJobA["outcome"] = "success"
testJobB["outcome"] = "failure"
testJobC["outcome"] = "success"
testJobA.save()
testJobB.save()
testJobC.save()
testJobA.completeInputFiles()
compFiles = len(testSubscription.filesOfStatus("Completed"))
assert compFiles == 0, \
"Error: test sub has wrong number of complete files: %s" % compFiles
testJobB["outcome"] = "success"
testJobB.save()
testJobB.completeInputFiles(skipFiles = [testFileB["lfn"]])
availFiles = len(testSubscription.filesOfStatus("Available"))
assert availFiles == 0, \
"Error: test sub has wrong number of available files: %s" % availFiles
acqFiles = len(testSubscription.filesOfStatus("Acquired"))
assert acqFiles == 1, \
"Error: test sub has wrong number of acquired files: %s" % acqFiles
compFiles = len(testSubscription.filesOfStatus("Completed"))
assert compFiles == 1, \
"Error: test sub has wrong number of complete files: %s" % compFiles
failFiles = len(testSubscription.filesOfStatus("Failed"))
assert failFiles == 1, \
"Error: test sub has wrong number of failed files: %s" % failFiles
availFiles = len(bogusSubscription.filesOfStatus("Available"))
assert availFiles == 0, \
"Error: test sub has wrong number of available files: %s" % availFiles
acqFiles = len(bogusSubscription.filesOfStatus("Acquired"))
assert acqFiles == 3, \
"Error: test sub has wrong number of acquired files: %s" % acqFiles
compFiles = len(bogusSubscription.filesOfStatus("Completed"))
assert compFiles == 0, \
"Error: test sub has wrong number of complete files: %s" % compFiles
failFiles = len(bogusSubscription.filesOfStatus("Failed"))
assert failFiles == 0, \
"Error: test sub has wrong number of failed files: %s" % failFiles
return
def testGetOutputParentLFNs(self):
"""
_testGetOutputParentLFNs_
Verify that the getOutputDBSParentLFNs() method returns the correct
parent LFNs.
"""
testWorkflow = Workflow(spec = "spec.xml", owner = "Simon",
name = "wf001", task="Test")
testWorkflow.create()
testWMBSFileset = Fileset(name = "TestFileset")
testWMBSFileset.create()
testSubscription = Subscription(fileset = testWMBSFileset,
workflow = testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
testFileA = File(lfn = "/this/is/a/lfnA", size = 1024, events = 10,
merged = True)
testFileB = File(lfn = "/this/is/a/lfnB", size = 1024, events = 10,
merged = True)
testFileC = File(lfn = "/this/is/a/lfnC", size = 1024, events = 10,
merged = False)
testFileD = File(lfn = "/this/is/a/lfnD", size = 1024, events = 10,
merged = False)
testFileE = File(lfn = "/this/is/a/lfnE", size = 1024, events = 10,
merged = True)
testFileF = File(lfn = "/this/is/a/lfnF", size = 1024, events = 10,
merged = True)
testFileA.create()
testFileB.create()
testFileC.create()
testFileD.create()
testFileE.create()
testFileF.create()
testFileE.addChild(testFileC["lfn"])
testFileF.addChild(testFileD["lfn"])
testJobA = Job(name = "TestJob", files = [testFileA, testFileB])
testJobA["couch_record"] = "somecouchrecord"
testJobA["location"] = "test.site.ch"
testJobA.create(group = testJobGroup)
testJobA.associateFiles()
testJobB = Job(name = "TestJobB", files = [testFileC, testFileD])
testJobB["couch_record"] = "somecouchrecord"
testJobB["location"] = "test.site.ch"
testJobB.create(group = testJobGroup)
testJobB.associateFiles()
goldenLFNs = ["/this/is/a/lfnA", "/this/is/a/lfnB"]
parentLFNs = testJobA.getOutputDBSParentLFNs()
for parentLFN in parentLFNs:
assert parentLFN in goldenLFNs, \
"ERROR: Unknown lfn: %s" % parentLFN
goldenLFNs.remove(parentLFN)
assert len(goldenLFNs) == 0, \
"ERROR: LFNs are missing: %s" % goldenLFNs
goldenLFNs = ["/this/is/a/lfnE", "/this/is/a/lfnF"]
parentLFNs = testJobB.getOutputDBSParentLFNs()
for parentLFN in parentLFNs:
assert parentLFN in goldenLFNs, \
"ERROR: Unknown lfn: %s" % parentLFN
goldenLFNs.remove(parentLFN)
assert len(goldenLFNs) == 0, \
"ERROR: LFNs are missing..."
return
def test_AutoIncrementCheck(self):
"""
_AutoIncrementCheck_
Test and see whether we can find and set the auto_increment values
"""
myThread = threading.currentThread()
if not myThread.dialect.lower() == 'mysql':
return
testWorkflow = Workflow(spec = "spec.xml", owner = "Steve",
name = "wf001", task="Test")
testWorkflow.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow)
testSubscription.create()
testFileA = File(lfn = makeUUID(), locations = "test.site.ch")
testFileB = File(lfn = makeUUID(), locations = "test.site.ch")
testFileA.create()
testFileB.create()
testFileset.addFile([testFileA, testFileB])
testFileset.commit()
testSubscription.acquireFiles([testFileA, testFileB])
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
incrementDAO = self.daoFactory(classname = "Jobs.AutoIncrementCheck")
incrementDAO.execute()
testJob = Job()
testJob.create(group = testJobGroup)
self.assertEqual(testJob.exists(), 1)
incrementDAO.execute()
testJob = Job()
testJob.create(group = testJobGroup)
self.assertEqual(testJob.exists(), 2)
incrementDAO.execute(input = 10)
testJob = Job()
testJob.create(group = testJobGroup)
self.assertEqual(testJob.exists(), 11)
incrementDAO.execute(input = 5)
testJob = Job()
testJob.create(group = testJobGroup)
self.assertEqual(testJob.exists(), 12)
return
def testMask(self):
"""
_testMask_
Test the new mask setup
"""
testWorkflow = Workflow(spec = "spec.xml", owner = "Steve",
name = "wf001", task="Test")
testWorkflow.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow)
testSubscription.create()
testFileA = File(lfn = makeUUID(), locations = "test.site.ch")
testFileB = File(lfn = makeUUID(), locations = "test.site.ch")
testFileA.create()
testFileB.create()
testFileset.addFile([testFileA, testFileB])
testFileset.commit()
testSubscription.acquireFiles([testFileA, testFileB])
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
testJob = Job()
testJob['mask'].addRunAndLumis(run = 100, lumis = [101, 102])
testJob['mask'].addRunAndLumis(run = 200, lumis = [201, 202])
testJob.create(group = testJobGroup)
loadJob = Job(id = testJob.exists())
loadJob.loadData()
runs = loadJob['mask'].getRunAndLumis()
self.assertEqual(len(runs), 2)
self.assertEqual(runs[100], [[101, 102]])
self.assertEqual(runs[200], [[201, 202]])
bigRun = Run(100, *[101, 102, 103, 104])
badRun = Run(300, *[1001, 1002])
result = loadJob['mask'].filterRunLumisByMask([bigRun, badRun])
self.assertEqual(len(result), 1)
alteredRun = result.pop()
self.assertEqual(alteredRun.run, 100)
self.assertEqual(alteredRun.lumis, [101, 102])
run0 = Run(300, *[1001, 1002])
run1 = Run(300, *[1001, 1002])
loadJob['mask'].filterRunLumisByMask([run0, run1])
return
def testReRecoDroppingRECO(self):
"""
_testReRecoDroppingRECO_
Verify that ReReco workflows can be created and inserted into WMBS
correctly. The ReReco workflow is just a DataProcessing workflow with
skims tacked on. This tests run on unmerged RECO output
"""
skimConfig = self.injectSkimConfig()
recoConfig = self.injectReRecoConfig()
dataProcArguments = ReRecoWorkloadFactory.getTestArguments()
dataProcArguments["ProcessingString"] = "ProcString"
dataProcArguments["ConfigCacheID"] = recoConfig
dataProcArguments.update({"SkimName1": "SomeSkim",
"SkimInput1": "RECOoutput",
"Skim1ConfigCacheID": skimConfig})
dataProcArguments["CouchURL"] = os.environ["COUCHURL"]
dataProcArguments["CouchDBName"] = "rereco_t"
dataProcArguments["TransientOutputModules"] = ["RECOoutput"]
dataProcArguments["EnableHarvesting"] = True
dataProcArguments["DQMConfigCacheID"] = self.injectDQMHarvestConfig()
factory = ReRecoWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction("TestWorkload", dataProcArguments)
self.assertEqual(testWorkload.data.tasks.DataProcessing.tree.children. \
SomeSkim.tree.children.SomeSkimMergeSkimB.steps.cmsRun1.output.modules. \
Merged.mergedLFNBase,
'/store/data/FAKE/MinimumBias/USER/SkimBFilter-ProcString-v1')
testWMBSHelper = WMBSHelper(testWorkload, "DataProcessing", "SomeBlock", cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
skimWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/SomeSkim")
skimWorkflow.load()
self.assertEqual(len(skimWorkflow.outputMap), 3,
"Error: Wrong number of WF outputs.")
goldenOutputMods = {"SkimA": "RAW-RECO", "SkimB": "USER"}
for goldenOutputMod, tier in viewitems(goldenOutputMods):
fset = goldenOutputMod + tier
mergedOutput = skimWorkflow.outputMap[fset][0]["merged_output_fileset"]
unmergedOutput = skimWorkflow.outputMap[fset][0]["output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(mergedOutput.name,
"/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-Merged%s" % (
goldenOutputMod, tier),
"Error: Merged output fileset is wrong: %s" % mergedOutput.name)
self.assertEqual(unmergedOutput.name, "/TestWorkload/DataProcessing/SomeSkim/unmerged-%s" % fset,
"Error: Unmerged output fileset is wrong: %s" % unmergedOutput.name)
logArchOutput = skimWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = skimWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name, "/TestWorkload/DataProcessing/SomeSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name, "/TestWorkload/DataProcessing/SomeSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for goldenOutputMod, tier in viewitems(goldenOutputMods):
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s" % goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged%s" % tier][0]["merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged%s" % tier][0]["output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(mergedMergeOutput.name,
"/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-Merged%s" % (
goldenOutputMod, tier),
"Error: Merged output fileset is wrong.")
self.assertEqual(unmergedMergeOutput.name,
"/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-Merged%s" % (
goldenOutputMod, tier),
"Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" % logArchOutput.name)
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="/TestWorkload/DataProcessing/unmerged-RECOoutputRECO")
topLevelFileset.loadData()
skimSubscription = Subscription(fileset=topLevelFileset, workflow=skimWorkflow)
skimSubscription.loadData()
self.assertEqual(skimSubscription["type"], "Skim",
"Error: Wrong subscription type.")
self.assertEqual(skimSubscription["split_algo"], "FileBased",
"Error: Wrong split algo.")
for skimOutput, tier in viewitems(goldenOutputMods):
fset = skimOutput + tier
unmerged = Fileset(name="/TestWorkload/DataProcessing/SomeSkim/unmerged-%s" % fset)
unmerged.loadData()
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s" % skimOutput)
mergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmerged, workflow=mergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo.")
for skimOutput, tier in viewitems(goldenOutputMods):
fset = skimOutput + tier
unmerged = Fileset(name="/TestWorkload/DataProcessing/SomeSkim/unmerged-%s" % fset)
unmerged.loadData()
cleanupWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/SomeSkim/SomeSkimCleanupUnmerged%s" % skimOutput)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmerged, workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"], "SiblingProcessingBased",
"Error: Wrong split algo.")
for skimOutput in goldenOutputMods:
skimMergeLogCollect = Fileset(
name="/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/merged-logArchive" % skimOutput)
skimMergeLogCollect.loadData()
skimMergeLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/SomeSkim/SomeSkimMerge%s/SomeSkim%sMergeLogCollect" % (
skimOutput, skimOutput))
skimMergeLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=skimMergeLogCollect, workflow=skimMergeLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
return
def testB_NoFileSplitNoHardLimit(self):
"""
_testB_NoFileSplitNoHardLimit_
In this case we don't split on file boundaries, check different combination of files
make sure we make the most of the splitting, e.g. include many zero event files in
a single job.
"""
splitter = SplitterFactory()
# Create 100 files with 7 lumi per file and 0 events per lumi on average.
testSubscription = self.createSubscription(nFiles=100,
lumisPerFile=7,
twoSites=False,
nEventsPerFile=0)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# First test, the optimal settings are 360 events per job
# As we have files with 0 events per lumi, this will configure the splitting to
# a single job containing all files
jobGroups = jobFactory(halt_job_on_file_boundaries=False,
splitOnRun=False,
events_per_job=360,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 1, "There should be 1 job")
self.assertEqual(len(jobs[0]['input_files']), 100,
"All 100 files must be in the job")
self.assertEqual(jobs[0]['estimatedMemoryUsage'], 2300)
self.assertEqual(jobs[0]['estimatedDiskUsage'], 0)
self.assertEqual(jobs[0]['estimatedJobTime'], 0)
# Create 7 files, each one with different lumi/event distributions
testFileset = Fileset(name="FilesetA")
testFileset.create()
testFileA = self.createFile("/this/is/file1", 250, 0, 5, "T2_CH_CERN")
testFileB = self.createFile("/this/is/file2", 600, 1, 1, "T2_CH_CERN")
testFileC = self.createFile("/this/is/file3", 1200, 2, 2, "T2_CH_CERN")
testFileD = self.createFile("/this/is/file4", 100, 3, 1, "T2_CH_CERN")
testFileE = self.createFile("/this/is/file5", 30, 4, 1, "T2_CH_CERN")
testFileF = self.createFile("/this/is/file6", 10, 5, 1, "T2_CH_CERN")
testFileG = self.createFile("/this/is/file7", 151, 6, 3, "T2_CH_CERN")
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.addFile(testFileC)
testFileset.addFile(testFileD)
testFileset.addFile(testFileE)
testFileset.addFile(testFileF)
testFileset.addFile(testFileG)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# Optimal settings are: jobs with 150 events per job
# This means, the first file must be splitted in 3 lumis per job which would leave room
# for another lumi in the second job, but the second file has a lumi too big for that
# The 3rd job only contains the second file, the fourth and fifth job split the third file
jobGroups = jobFactory(halt_job_on_file_boundaries=False,
splitOnRun=False,
events_per_job=150,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 7, "7 jobs must be in the jobgroup")
self.assertEqual(jobs[0]["mask"].getRunAndLumis(), {0: [[0, 2]]},
"Wrong mask for the first job")
self.assertEqual(jobs[0]["estimatedJobTime"], 150 * 12)
self.assertEqual(jobs[0]["estimatedDiskUsage"], 150 * 400)
self.assertEqual(jobs[1]["mask"].getRunAndLumis(), {0: [[3, 4]]},
"Wrong mask for the second job")
self.assertEqual(jobs[1]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[1]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[2]["mask"].getRunAndLumis(), {1: [[1, 1]]},
"Wrong mask for the third job")
self.assertEqual(jobs[2]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[2]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[3]["mask"].getRunAndLumis(), {2: [[4, 4]]},
"Wrong mask for the fourth job")
self.assertEqual(jobs[3]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[3]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[4]["mask"].getRunAndLumis(), {2: [[5, 5]]},
"Wrong mask for the fifth job")
self.assertEqual(jobs[4]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[4]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[5]["mask"].getRunAndLumis(), {
3: [[3, 3]],
4: [[4, 4]],
5: [[5, 5]]
}, "Wrong mask for the sixth job")
self.assertEqual(jobs[5]["estimatedJobTime"], 140 * 12)
self.assertEqual(jobs[5]["estimatedDiskUsage"], 140 * 400)
self.assertEqual(jobs[6]["mask"].getRunAndLumis(), {6: [[18, 20]]},
"Wrong mask for the seventh job")
self.assertEqual(jobs[6]["estimatedJobTime"], 150 * 12)
self.assertEqual(jobs[6]["estimatedDiskUsage"], 150 * 400)
for job in jobs:
self.assertEqual(job["estimatedMemoryUsage"], 2300)
# Test interactions of this algorithm with splitOnRun = True
# Make 2 files, one with 3 runs and a second one with the last run of the first
fileA = File(lfn="/this/is/file1a", size=1000, events=2400)
lumiListA = []
lumiListB = []
lumiListC = []
for lumi in range(8):
lumiListA.append(1 + lumi)
lumiListB.append(1 + lumi)
lumiListC.append(1 + lumi)
fileA.addRun(Run(1, *lumiListA))
fileA.addRun(Run(2, *lumiListA))
fileA.addRun(Run(3, *lumiListA))
fileA.setLocation("T1_US_FNAL_Disk")
fileB = self.createFile('/this/is/file2a', 200, 3, 5,
"T1_US_FNAL_Disk")
testFileset = Fileset(name='FilesetB')
testFileset.create()
testFileset.addFile(fileA)
testFileset.addFile(fileB)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# The settings for this splitting are 700 events per job
jobGroups = jobFactory(splitOnRun=True,
halt_job_on_file_boundaries=False,
events_per_job=700,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 6, "Six jobs must be in the jobgroup")
self.assertEqual(jobs[0]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[0]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[1]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[1]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[2]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[2]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[3]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[3]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[4]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[4]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[5]["estimatedJobTime"], 300 * 12)
self.assertEqual(jobs[5]["estimatedDiskUsage"], 300 * 400)
def testCreateSubscription(self):
"""
_testCreateSubscription_
Verify that the subscription creation code works correctly.
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
pnn='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
pnn='goodse2.cern.ch',
ceName='site2',
plugin="TestPlugin")
testWorkload = self.createTestWMSpec()
testTopLevelTask = getFirstTask(testWorkload)
testWMBSHelper = WMBSHelper(testWorkload,
testTopLevelTask.name(),
"SomeBlock",
cachepath=self.workDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testTopLevelTask, testWMBSHelper.topLevelFileset)
procWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/ProcessingTask")
procWorkflow.load()
self.assertEqual(procWorkflow.owner, "sfoulkes",
"Error: Wrong owner: %s" % procWorkflow.owner)
self.assertEqual(procWorkflow.group, "DMWM",
"Error: Wrong group: %s" % procWorkflow.group)
self.assertEqual(procWorkflow.wfType, "TestReReco",
"Error: Wrong type.")
self.assertEqual(
procWorkflow.spec,
os.path.join(self.workDir, procWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(procWorkflow.outputMap.keys()), 1,
"Error: Wrong number of WF outputs.")
mergedProcOutput = procWorkflow.outputMap["OutputA"][0][
"merged_output_fileset"]
unmergedProcOutput = procWorkflow.outputMap["OutputA"][0][
"output_fileset"]
mergedProcOutput.loadData()
unmergedProcOutput.loadData()
self.assertEqual(
mergedProcOutput.name,
"/TestWorkload/ProcessingTask/MergeTask/merged-Merged",
"Error: Merged output fileset is wrong.")
self.assertEqual(unmergedProcOutput.name,
"/TestWorkload/ProcessingTask/unmerged-OutputA",
"Error: Unmerged output fileset is wrong.")
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/ProcessingTask/MergeTask")
mergeWorkflow.load()
self.assertEqual(mergeWorkflow.owner, "sfoulkes",
"Error: Wrong owner.")
self.assertEqual(
mergeWorkflow.spec,
os.path.join(self.workDir, mergeWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 1,
"Error: Wrong number of WF outputs.")
cleanupWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/ProcessingTask/CleanupTask")
cleanupWorkflow.load()
self.assertEqual(cleanupWorkflow.owner, "sfoulkes",
"Error: Wrong owner.")
self.assertEqual(
cleanupWorkflow.spec,
os.path.join(self.workDir, cleanupWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(cleanupWorkflow.outputMap.keys()), 0,
"Error: Wrong number of WF outputs.")
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0][
"output_fileset"]
unmergedMergeOutput.loadData()
self.assertEqual(
unmergedMergeOutput.name,
"/TestWorkload/ProcessingTask/MergeTask/merged-Merged",
"Error: Unmerged output fileset is wrong.")
skimWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/ProcessingTask/MergeTask/SkimTask")
skimWorkflow.load()
self.assertEqual(skimWorkflow.owner, "sfoulkes", "Error: Wrong owner.")
self.assertEqual(
skimWorkflow.spec,
os.path.join(self.workDir, skimWorkflow.name, "WMSandbox",
"WMWorkload.pkl"), "Error: Wrong spec URL")
self.assertEqual(len(skimWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedSkimOutputA = skimWorkflow.outputMap["SkimOutputA"][0][
"merged_output_fileset"]
unmergedSkimOutputA = skimWorkflow.outputMap["SkimOutputA"][0][
"output_fileset"]
mergedSkimOutputB = skimWorkflow.outputMap["SkimOutputB"][0][
"merged_output_fileset"]
unmergedSkimOutputB = skimWorkflow.outputMap["SkimOutputB"][0][
"output_fileset"]
mergedSkimOutputA.loadData()
mergedSkimOutputB.loadData()
unmergedSkimOutputA.loadData()
unmergedSkimOutputB.loadData()
self.assertEqual(
mergedSkimOutputA.name,
"/TestWorkload/ProcessingTask/MergeTask/SkimTask/unmerged-SkimOutputA",
"Error: Merged output fileset is wrong: %s" %
mergedSkimOutputA.name)
self.assertEqual(
unmergedSkimOutputA.name,
"/TestWorkload/ProcessingTask/MergeTask/SkimTask/unmerged-SkimOutputA",
"Error: Unmerged output fileset is wrong.")
self.assertEqual(
mergedSkimOutputB.name,
"/TestWorkload/ProcessingTask/MergeTask/SkimTask/unmerged-SkimOutputB",
"Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedSkimOutputB.name,
"/TestWorkload/ProcessingTask/MergeTask/SkimTask/unmerged-SkimOutputB",
"Error: Unmerged output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-ProcessingTask-SomeBlock")
topLevelFileset.loadData()
procSubscription = Subscription(fileset=topLevelFileset,
workflow=procWorkflow)
procSubscription.loadData()
self.assertEqual(len(procSubscription.getWhiteBlackList()), 2,
"Error: Wrong site white/black list for proc sub.")
for site in procSubscription.getWhiteBlackList():
if site["site_name"] == "site1":
self.assertEqual(site["valid"], 1,
"Error: Site should be white listed.")
else:
self.assertEqual(site["valid"], 0,
"Error: Site should be black listed.")
self.assertEqual(procSubscription["type"], "Processing",
"Error: Wrong subscription type.")
self.assertEqual(procSubscription["split_algo"], "FileBased",
"Error: Wrong split algo.")
mergeSubscription = Subscription(fileset=unmergedProcOutput,
workflow=mergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(len(mergeSubscription.getWhiteBlackList()), 0,
"Error: Wrong white/black list for merge sub.")
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "WMBSMergeBySize",
"Error: Wrong split algo.")
skimSubscription = Subscription(fileset=unmergedMergeOutput,
workflow=skimWorkflow)
skimSubscription.loadData()
self.assertEqual(skimSubscription["type"], "Skim",
"Error: Wrong subscription type.")
self.assertEqual(skimSubscription["split_algo"], "FileBased",
"Error: Wrong split algo.")
return
def testListRunningJobs(self):
"""
_testListRunningJobs_
Test the ListRunningJobs DAO.
"""
testWorkflow = Workflow(spec=makeUUID(),
owner="Steve",
name=makeUUID(),
task="Test")
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
type="Processing")
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
testJobA = Job(name=makeUUID(), files=[])
testJobA["couch_record"] = makeUUID()
testJobA.create(group=testJobGroup)
testJobA["state"] = "executing"
testJobB = Job(name=makeUUID(), files=[])
testJobB["couch_record"] = makeUUID()
testJobB.create(group=testJobGroup)
testJobB["state"] = "complete"
testJobC = Job(name=makeUUID(), files=[])
testJobC["couch_record"] = makeUUID()
testJobC.create(group=testJobGroup)
testJobC["state"] = "new"
changeStateAction = self.daoFactory(classname="Jobs.ChangeState")
changeStateAction.execute(jobs=[testJobA, testJobB, testJobC])
runningJobsAction = self.daoFactory(
classname="Monitoring.ListRunningJobs")
runningJobs = runningJobsAction.execute()
assert len(runningJobs) == 2, \
"Error: Wrong number of running jobs returned."
for runningJob in runningJobs:
if runningJob["job_name"] == testJobA["name"]:
assert runningJob["state"] == testJobA["state"], \
"Error: Running job has wrong state."
assert runningJob["couch_record"] == testJobA["couch_record"], \
"Error: Running job has wrong couch record."
else:
assert runningJob["job_name"] == testJobC["name"], \
"Error: Running job has wrong name."
assert runningJob["state"] == testJobC["state"], \
"Error: Running job has wrong state."
assert runningJob["couch_record"] == testJobC["couch_record"], \
"Error: Running job has wrong couch record."
return
def setupForKillTest(self, baAPI=None):
"""
_setupForKillTest_
Inject a workflow into WMBS that has a processing task, a merge task and
a cleanup task. Inject files into the various tasks at various
processing states (acquired, complete, available...). Also create jobs
for each subscription in various states.
"""
myThread = threading.currentThread()
daoFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
dummyLocationAction = daoFactory(classname="Locations.New")
changeStateAction = daoFactory(classname="Jobs.ChangeState")
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
pnn='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertThreshold(siteName = 'site1', taskType = 'Processing', \
maxSlots = 10000, pendingSlots = 10000)
userDN = 'someDN'
userAction = daoFactory(classname="Users.New")
userAction.execute(dn=userDN,
group_name='DEFAULT',
role_name='DEFAULT')
inputFileset = Fileset("input")
inputFileset.create()
inputFileA = File("lfnA", locations="goodse.cern.ch")
inputFileB = File("lfnB", locations="goodse.cern.ch")
inputFileC = File("lfnC", locations="goodse.cern.ch")
inputFileA.create()
inputFileB.create()
inputFileC.create()
inputFileset.addFile(inputFileA)
inputFileset.addFile(inputFileB)
inputFileset.addFile(inputFileC)
inputFileset.commit()
unmergedOutputFileset = Fileset("unmerged")
unmergedOutputFileset.create()
unmergedFileA = File("ulfnA", locations="goodse.cern.ch")
unmergedFileB = File("ulfnB", locations="goodse.cern.ch")
unmergedFileC = File("ulfnC", locations="goodse.cern.ch")
unmergedFileA.create()
unmergedFileB.create()
unmergedFileC.create()
unmergedOutputFileset.addFile(unmergedFileA)
unmergedOutputFileset.addFile(unmergedFileB)
unmergedOutputFileset.addFile(unmergedFileC)
unmergedOutputFileset.commit()
mainProcWorkflow = Workflow(spec="spec1",
owner="Steve",
name="Main",
task="Proc")
mainProcWorkflow.create()
mainProcMergeWorkflow = Workflow(spec="spec1",
owner="Steve",
name="Main",
task="ProcMerge")
mainProcMergeWorkflow.create()
mainCleanupWorkflow = Workflow(spec="spec1",
owner="Steve",
name="Main",
task="Cleanup")
mainCleanupWorkflow.create()
self.mainProcSub = Subscription(fileset=inputFileset,
workflow=mainProcWorkflow,
type="Processing")
self.mainProcSub.create()
self.mainProcSub.acquireFiles(inputFileA)
self.mainProcSub.completeFiles(inputFileB)
procJobGroup = JobGroup(subscription=self.mainProcSub)
procJobGroup.create()
self.procJobA = Job(name="ProcJobA")
self.procJobA["state"] = "new"
self.procJobA["location"] = "site1"
self.procJobB = Job(name="ProcJobB")
self.procJobB["state"] = "executing"
self.procJobB["location"] = "site1"
self.procJobC = Job(name="ProcJobC")
self.procJobC["state"] = "complete"
self.procJobC["location"] = "site1"
self.procJobA.create(procJobGroup)
self.procJobB.create(procJobGroup)
self.procJobC.create(procJobGroup)
self.mainMergeSub = Subscription(fileset=unmergedOutputFileset,
workflow=mainProcMergeWorkflow,
type="Merge")
self.mainMergeSub.create()
self.mainMergeSub.acquireFiles(unmergedFileA)
self.mainMergeSub.failFiles(unmergedFileB)
mergeJobGroup = JobGroup(subscription=self.mainMergeSub)
mergeJobGroup.create()
self.mergeJobA = Job(name="MergeJobA")
self.mergeJobA["state"] = "exhausted"
self.mergeJobA["location"] = "site1"
self.mergeJobB = Job(name="MergeJobB")
self.mergeJobB["state"] = "cleanout"
self.mergeJobB["location"] = "site1"
self.mergeJobC = Job(name="MergeJobC")
self.mergeJobC["state"] = "new"
self.mergeJobC["location"] = "site1"
self.mergeJobA.create(mergeJobGroup)
self.mergeJobB.create(mergeJobGroup)
self.mergeJobC.create(mergeJobGroup)
self.mainCleanupSub = Subscription(fileset=unmergedOutputFileset,
workflow=mainCleanupWorkflow,
type="Cleanup")
self.mainCleanupSub.create()
self.mainCleanupSub.acquireFiles(unmergedFileA)
self.mainCleanupSub.completeFiles(unmergedFileB)
cleanupJobGroup = JobGroup(subscription=self.mainCleanupSub)
cleanupJobGroup.create()
self.cleanupJobA = Job(name="CleanupJobA")
self.cleanupJobA["state"] = "new"
self.cleanupJobA["location"] = "site1"
self.cleanupJobB = Job(name="CleanupJobB")
self.cleanupJobB["state"] = "executing"
self.cleanupJobB["location"] = "site1"
self.cleanupJobC = Job(name="CleanupJobC")
self.cleanupJobC["state"] = "complete"
self.cleanupJobC["location"] = "site1"
self.cleanupJobA.create(cleanupJobGroup)
self.cleanupJobB.create(cleanupJobGroup)
self.cleanupJobC.create(cleanupJobGroup)
jobList = [
self.procJobA, self.procJobB, self.procJobC, self.mergeJobA,
self.mergeJobB, self.mergeJobC, self.cleanupJobA, self.cleanupJobB,
self.cleanupJobC
]
changeStateAction.execute(jobList)
if baAPI:
for job in jobList:
job['plugin'] = 'TestPlugin'
job['userdn'] = userDN
job['usergroup'] = 'DEFAULT'
job['userrole'] = 'DEFAULT'
job['custom']['location'] = 'site1'
baAPI.createNewJobs(wmbsJobs=jobList)
# We'll create an unrelated workflow to verify that it isn't affected
# by the killing code.
bogusFileset = Fileset("dontkillme")
bogusFileset.create()
bogusFileA = File("bogus/lfnA", locations="goodse.cern.ch")
bogusFileA.create()
bogusFileset.addFile(bogusFileA)
bogusFileset.commit()
bogusWorkflow = Workflow(spec="spec2",
owner="Steve",
name="Bogus",
task="Proc")
bogusWorkflow.create()
self.bogusSub = Subscription(fileset=bogusFileset,
workflow=bogusWorkflow,
type="Processing")
self.bogusSub.create()
self.bogusSub.acquireFiles(bogusFileA)
return
def pollSubscriptions(self):
"""
Poller for looking in all active subscriptions for jobs that need to be made.
"""
logging.info("Beginning JobCreator.pollSubscriptions() cycle.")
myThread = threading.currentThread()
# First, get list of Subscriptions
subscriptions = self.subscriptionList.execute()
# Okay, now we have a list of subscriptions
for subscriptionID in subscriptions:
wmbsSubscription = Subscription(id=subscriptionID)
try:
wmbsSubscription.load()
except IndexError:
# This happens when the subscription no longer exists
# i.e., someone executed a kill() function on the database
# while the JobCreator was in cycle
# Ignore this subscription
msg = "JobCreator cannot load subscription %i" % subscriptionID
logging.error(msg)
self.sendAlert(6, msg=msg)
continue
workflow = Workflow(id=wmbsSubscription["workflow"].id)
workflow.load()
wmbsSubscription['workflow'] = workflow
wmWorkload = retrieveWMSpec(workflow=workflow)
if not workflow.task or not wmWorkload:
# Then we have a problem
# We NEED a sandbox
# Abort this subscription!
# But do NOT fail
# We have no way of marking a subscription as bad per se
# We'll have to just keep skipping it
msg = "Have no task for workflow %i\n" % (workflow.id)
msg += "Aborting Subscription %i" % (subscriptionID)
logging.error(msg)
self.sendAlert(1, msg=msg)
continue
logging.debug("Have loaded subscription %i with workflow %i\n",
subscriptionID, workflow.id)
# retrieve information from the workload to propagate down to the job configuration
allowOpport = wmWorkload.getAllowOpportunistic()
# Set task object
wmTask = wmWorkload.getTaskByPath(workflow.task)
# Get generators
# If you fail to load the generators, pass on the job
try:
if hasattr(wmTask.data, 'generators'):
manager = GeneratorManager(wmTask)
seederList = manager.getGeneratorList()
else:
seederList = []
except Exception as ex:
msg = "Had failure loading generators for subscription %i\n" % (
subscriptionID)
msg += "Exception: %s\n" % str(ex)
msg += "Passing over this error. It will reoccur next interation!\n"
msg += "Please check or remove this subscription!\n"
logging.error(msg)
self.sendAlert(6, msg=msg)
continue
logging.debug(
"Going to call wmbsJobFactory for sub %i with limit %i",
subscriptionID, self.limit)
splitParams = retrieveJobSplitParams(wmWorkload, workflow.task)
logging.debug("Split Params: %s", splitParams)
# Load the proper job splitting module
splitterFactory = SplitterFactory(
splitParams.get('algo_package', "WMCore.JobSplitting"))
# and return an instance of the splitting algorithm
wmbsJobFactory = splitterFactory(package="WMCore.WMBS",
subscription=wmbsSubscription,
generators=seederList,
limit=self.limit)
# Turn on the jobFactory --> get available files for that subscription, keep result proxies
wmbsJobFactory.open()
# Create a function to hold it, calling __call__ from the JobFactory
# which then calls algorithm method of the job splitting algo instance
jobSplittingFunction = runSplitter(jobFactory=wmbsJobFactory,
splitParams=splitParams)
# Now we get to find out how many jobs there are.
jobNumber = self.countJobs.execute(workflow=workflow.id,
conn=myThread.transaction.conn,
transaction=True)
jobNumber += splitParams.get('initial_lfn_counter', 0)
logging.debug("Have %i jobs for workflow %s already in database.",
jobNumber, workflow.name)
continueSubscription = True
while continueSubscription:
# This loop runs over the jobFactory,
# using yield statements and a pre-existing proxy to
# generate and process new jobs
# First we need the jobs.
myThread.transaction.begin()
try:
wmbsJobGroups = next(jobSplittingFunction)
logging.info("Retrieved %i jobGroups from jobSplitter",
len(wmbsJobGroups))
except StopIteration:
# If you receive a stopIteration, we're done
logging.info("Completed iteration over subscription %i",
subscriptionID)
continueSubscription = False
myThread.transaction.commit()
break
# If we have no jobGroups, we're done
if len(wmbsJobGroups) == 0:
logging.info("Found end in iteration over subscription %i",
subscriptionID)
continueSubscription = False
myThread.transaction.commit()
break
# Assemble a dict of all the info
processDict = {
'workflow': workflow,
'wmWorkload': wmWorkload,
'wmTaskName': wmTask.getPathName(),
'jobNumber': jobNumber,
'sandbox': wmTask.data.input.sandbox,
'owner': wmWorkload.getOwner().get('name', None),
'ownerDN': wmWorkload.getOwner().get('dn', None),
'ownerGroup': wmWorkload.getOwner().get('vogroup', ''),
'ownerRole': wmWorkload.getOwner().get('vorole', ''),
'numberOfCores': 1,
'inputDataset': wmTask.getInputDatasetPath()
}
try:
maxCores = 1
stepNames = wmTask.listAllStepNames()
for stepName in stepNames:
sh = wmTask.getStep(stepName)
maxCores = max(maxCores, sh.getNumberOfCores())
processDict.update({'numberOfCores': maxCores})
except AttributeError:
logging.info(
"Failed to read multicore settings from task %s",
wmTask.getPathName())
tempSubscription = Subscription(id=wmbsSubscription['id'])
# if we have glideinWMS constraints, then adapt all jobs
if self.glideinLimits:
capResourceEstimates(wmbsJobGroups, self.glideinLimits)
nameDictList = []
for wmbsJobGroup in wmbsJobGroups:
# For each jobGroup, put a dictionary
# together and run it with creatorProcess
jobsInGroup = len(wmbsJobGroup.jobs)
wmbsJobGroup.subscription = tempSubscription
tempDict = {}
tempDict.update(processDict)
tempDict['jobGroup'] = wmbsJobGroup
tempDict['swVersion'] = wmTask.getSwVersion(allSteps=True)
tempDict['scramArch'] = wmTask.getScramArch()
tempDict['jobNumber'] = jobNumber
tempDict['agentNumber'] = self.agentNumber
tempDict['agentName'] = self.agentName
tempDict[
'inputDatasetLocations'] = wmbsJobGroup.getLocationsForJobs(
)
tempDict['allowOpportunistic'] = allowOpport
jobGroup = creatorProcess(work=tempDict,
jobCacheDir=self.jobCacheDir)
jobNumber += jobsInGroup
# Set jobCache for group
for job in jobGroup.jobs:
nameDictList.append({
'jobid': job['id'],
'cacheDir': job['cache_dir']
})
job["user"] = wmWorkload.getOwner()["name"]
job["group"] = wmWorkload.getOwner()["group"]
# Set the caches in the database
try:
if len(nameDictList) > 0:
self.setBulkCache.execute(
jobDictList=nameDictList,
conn=myThread.transaction.conn,
transaction=True)
except WMException:
raise
except Exception as ex:
msg = "Unknown exception while setting the bulk cache:\n"
msg += str(ex)
logging.error(msg)
self.sendAlert(6, msg=msg)
logging.debug(
"Error while setting bulkCache with following values: %s\n",
nameDictList)
raise JobCreatorException(msg)
# Advance the jobGroup in changeState
for wmbsJobGroup in wmbsJobGroups:
self.advanceJobGroup(wmbsJobGroup=wmbsJobGroup)
# Now end the transaction so that everything is wrapped
# in a single rollback
myThread.transaction.commit()
# END: While loop over jobFactory
# Close the jobFactory
wmbsJobFactory.close()
return
def testReReco(self):
"""
_testReReco_
Verify that ReReco workflows can be created and inserted into WMBS
correctly.
"""
skimConfig = self.injectSkimConfig()
recoConfig = self.injectReRecoConfig()
dataProcArguments = ReRecoWorkloadFactory.getTestArguments()
dataProcArguments["ProcessingString"] = "ProcString"
dataProcArguments["ConfigCacheID"] = recoConfig
dataProcArguments.update({"SkimName1": "SomeSkim",
"SkimInput1": "RECOoutput",
"Skim1ConfigCacheID": skimConfig})
dataProcArguments["CouchURL"] = os.environ["COUCHURL"]
dataProcArguments["CouchDBName"] = "rereco_t"
dataProcArguments["EnableHarvesting"] = True
dataProcArguments["DQMConfigCacheID"] = recoConfig
factory = ReRecoWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction("TestWorkload", dataProcArguments)
self.assertEqual(testWorkload.data.tasks.DataProcessing.tree.children.DataProcessingMergeRECOoutput. \
tree.children.SomeSkim.tree.children.SomeSkimMergeSkimB.steps.cmsRun1.output.modules. \
Merged.mergedLFNBase,
'/store/data/FAKE/MinimumBias/USER/SkimBFilter-ProcString-v1')
testWMBSHelper = WMBSHelper(testWorkload, "DataProcessing", "SomeBlock", cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
procWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing")
procWorkflow.load()
self.assertEqual(len(procWorkflow.outputMap.keys()), 3,
"Error: Wrong number of WF outputs.")
goldenOutputMods = ["RECOoutput", "DQMoutput"]
for goldenOutputMod in goldenOutputMods:
mergedOutput = procWorkflow.outputMap[goldenOutputMod][0]["merged_output_fileset"]
unmergedOutput = procWorkflow.outputMap[goldenOutputMod][0]["output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(mergedOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong: %s" % mergedOutput.name)
self.assertEqual(unmergedOutput.name, "/TestWorkload/DataProcessing/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = procWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = procWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name, "/TestWorkload/DataProcessing/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name, "/TestWorkload/DataProcessing/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMerge%s" % goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(mergedMergeOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong.")
self.assertEqual(unmergedMergeOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMerge%s/merged-Merged" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" % logArchOutput.name)
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-DataProcessing-SomeBlock")
topLevelFileset.loadData()
procSubscription = Subscription(fileset=topLevelFileset, workflow=procWorkflow)
procSubscription.loadData()
self.assertEqual(procSubscription["type"], "Processing",
"Error: Wrong subscription type.")
self.assertEqual(procSubscription["split_algo"], "EventAwareLumiBased",
"Error: Wrong split algo.")
unmergedReco = Fileset(name="/TestWorkload/DataProcessing/unmerged-RECOoutput")
unmergedReco.loadData()
recoMergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput")
recoMergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedReco, workflow=recoMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo.")
unmergedDqm = Fileset(name="/TestWorkload/DataProcessing/unmerged-DQMoutput")
unmergedDqm.loadData()
dqmMergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput")
dqmMergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedDqm, workflow=dqmMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo.")
for procOutput in ["RECOoutput", "DQMoutput"]:
unmerged = Fileset(name="/TestWorkload/DataProcessing/unmerged-%s" % procOutput)
unmerged.loadData()
cleanupWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingCleanupUnmerged%s" % procOutput)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmerged, workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"], "SiblingProcessingBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name="/TestWorkload/DataProcessing/unmerged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/LogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=procLogCollect, workflow=procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/merged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/DataProcessingRECOoutputMergeLogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=procLogCollect, workflow=procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/merged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/DataProcessingRECOoutputMergeLogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=procLogCollect, workflow=procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
skimWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim")
skimWorkflow.load()
self.assertEqual(len(skimWorkflow.outputMap.keys()), 3,
"Error: Wrong number of WF outputs.")
goldenOutputMods = ["SkimA", "SkimB"]
for goldenOutputMod in goldenOutputMods:
mergedOutput = skimWorkflow.outputMap[goldenOutputMod][0]["merged_output_fileset"]
unmergedOutput = skimWorkflow.outputMap[goldenOutputMod][0]["output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(mergedOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong: %s" % mergedOutput.name)
self.assertEqual(unmergedOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong: %s" % unmergedOutput.name)
logArchOutput = skimWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = skimWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s" % goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(mergedMergeOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong.")
self.assertEqual(unmergedMergeOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s/merged-Merged" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" % logArchOutput.name)
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/merged-Merged")
topLevelFileset.loadData()
skimSubscription = Subscription(fileset=topLevelFileset, workflow=skimWorkflow)
skimSubscription.loadData()
self.assertEqual(skimSubscription["type"], "Skim",
"Error: Wrong subscription type.")
self.assertEqual(skimSubscription["split_algo"], "FileBased",
"Error: Wrong split algo.")
for skimOutput in ["A", "B"]:
unmerged = Fileset(
name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/unmerged-Skim%s" % skimOutput)
unmerged.loadData()
mergeWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMergeSkim%s" % skimOutput)
mergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmerged, workflow=mergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo.")
for skimOutput in ["A", "B"]:
unmerged = Fileset(
name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/unmerged-Skim%s" % skimOutput)
unmerged.loadData()
cleanupWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimCleanupUnmergedSkim%s" % skimOutput)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmerged, workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"], "SiblingProcessingBased",
"Error: Wrong split algo.")
for skimOutput in ["A", "B"]:
skimMergeLogCollect = Fileset(
name="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMergeSkim%s/merged-logArchive" % skimOutput)
skimMergeLogCollect.loadData()
skimMergeLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeRECOoutput/SomeSkim/SomeSkimMergeSkim%s/SomeSkimSkim%sMergeLogCollect" % (
skimOutput, skimOutput))
skimMergeLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=skimMergeLogCollect, workflow=skimMergeLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
dqmWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/DataProcessingMergeDQMoutputEndOfRunDQMHarvestMerged")
dqmWorkflow.load()
topLevelFileset = Fileset(name="/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/merged-Merged")
topLevelFileset.loadData()
dqmSubscription = Subscription(fileset=topLevelFileset, workflow=dqmWorkflow)
dqmSubscription.loadData()
self.assertEqual(dqmSubscription["type"], "Harvesting",
"Error: Wrong subscription type.")
self.assertEqual(dqmSubscription["split_algo"], "Harvest",
"Error: Wrong split algo.")
logArchOutput = dqmWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = dqmWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/DataProcessingMergeDQMoutputEndOfRunDQMHarvestMerged/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/DataProcessingMergeDQMoutputEndOfRunDQMHarvestMerged/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
dqmHarvestLogCollect = Fileset(
name="/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/DataProcessingMergeDQMoutputEndOfRunDQMHarvestMerged/unmerged-logArchive")
dqmHarvestLogCollect.loadData()
dqmHarvestLogCollectWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/DataProcessing/DataProcessingMergeDQMoutput/DataProcessingMergeDQMoutputEndOfRunDQMHarvestMerged/DataProcessingMergeDQMoutputMergedEndOfRunDQMHarvestLogCollect")
dqmHarvestLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=dqmHarvestLogCollect, workflow=dqmHarvestLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
return
def testStoreResults(self):
"""
_testStoreResults_
Create a StoreResults workflow and verify it installs into WMBS
correctly.
"""
arguments = StoreResultsWorkloadFactory.getTestArguments()
arguments.update({'CmsPath': "/uscmst1/prod/sw/cms"})
factory = StoreResultsWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction(
"TestWorkload", arguments)
testWMBSHelper = WMBSHelper(testWorkload,
"StoreResults",
"SomeBlock",
cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
testWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/StoreResults")
testWorkflow.load()
self.assertEqual(len(testWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
goldenOutputMods = ["Merged"]
for goldenOutputMod in goldenOutputMods:
mergedOutput = testWorkflow.outputMap[goldenOutputMod][0][
"merged_output_fileset"]
unmergedOutput = testWorkflow.outputMap[goldenOutputMod][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(
mergedOutput.name,
"/TestWorkload/StoreResults/merged-%s" % goldenOutputMod,
"Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/StoreResults/merged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong: %s." %
unmergedOutput.name)
logArchOutput = testWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = testWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/StoreResults/merged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/StoreResults/merged-logArchive",
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-StoreResults-SomeBlock")
topLevelFileset.loadData()
procSubscription = Subscription(fileset=topLevelFileset,
workflow=testWorkflow)
procSubscription.loadData()
self.assertEqual(procSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(procSubscription["split_algo"],
"ParentlessMergeBySize", "Error: Wrong split algo.")
return
def testExpress(self):
"""
_testExpress_
Create an Express workflow
and verify it installs into WMBS correctly.
"""
testArguments = ExpressWorkloadFactory.getTestArguments()
testArguments.update(deepcopy(REQUEST))
testArguments['RecoCMSSWVersion'] = "CMSSW_9_0_0"
factory = ExpressWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction(
"TestWorkload", testArguments)
testWorkload.setSpecUrl("somespec")
testWorkload.setOwnerDetails("*****@*****.**", "T0")
testWMBSHelper = WMBSHelper(testWorkload,
"Express",
cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
expressWorkflow = Workflow(name="TestWorkload",
task="/TestWorkload/Express")
expressWorkflow.load()
self.assertEqual(
len(expressWorkflow.outputMap),
len(testArguments["Outputs"]) + 1,
"Error: Wrong number of WF outputs in the Express WF.")
goldenOutputMods = {
"write_PrimaryDataset1_FEVT": "FEVT",
"write_StreamExpress_ALCARECO": "ALCARECO",
"write_StreamExpress_DQMIO": "DQMIO"
}
for goldenOutputMod, tier in viewitems(goldenOutputMods):
fset = goldenOutputMod + tier
mergedOutput = expressWorkflow.outputMap[fset][0][
"merged_output_fileset"]
unmergedOutput = expressWorkflow.outputMap[fset][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
if goldenOutputMod != "write_StreamExpress_ALCARECO":
self.assertEqual(
mergedOutput.name,
"/TestWorkload/Express/ExpressMerge%s/merged-Merged%s" %
(goldenOutputMod, tier),
"Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/Express/unmerged-%s" % fset,
"Error: Unmerged output fileset is wrong: %s" %
unmergedOutput.name)
logArchOutput = expressWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = expressWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"/TestWorkload/Express/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"/TestWorkload/Express/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
alcaSkimWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO"
)
alcaSkimWorkflow.load()
self.assertEqual(
len(alcaSkimWorkflow.outputMap),
len(testArguments["AlcaSkims"]) + 1,
"Error: Wrong number of WF outputs in the AlcaSkim WF.")
goldenOutputMods = {
"ALCARECOStreamPromptCalibProd": "ALCAPROMPT",
"ALCARECOStreamTkAlMinBias": "ALCARECO"
}
for goldenOutputMod, tier in viewitems(goldenOutputMods):
fset = goldenOutputMod + tier
mergedOutput = alcaSkimWorkflow.outputMap[fset][0][
"merged_output_fileset"]
unmergedOutput = alcaSkimWorkflow.outputMap[fset][0][
"output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(
mergedOutput.name,
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/unmerged-%s"
% fset, "Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/unmerged-%s"
% fset, "Error: Unmerged output fileset is wrong: %s" %
unmergedOutput.name)
logArchOutput = alcaSkimWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = alcaSkimWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for dqmString in [
"ExpressMergewrite_StreamExpress_DQMIOEndOfRunDQMHarvestMerged",
"ExpressMergewrite_StreamExpress_DQMIOPeriodicDQMHarvestMerged"
]:
dqmTask = "/TestWorkload/Express/ExpressMergewrite_StreamExpress_DQMIO/%s" % dqmString
dqmWorkflow = Workflow(name="TestWorkload", task=dqmTask)
dqmWorkflow.load()
logArchOutput = dqmWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = dqmWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name,
"%s/unmerged-logArchive" % dqmTask,
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name,
"%s/unmerged-logArchive" % dqmTask,
"Error: LogArchive output fileset is wrong.")
goldenOutputMods = {
"write_PrimaryDataset1_FEVT": "FEVT",
"write_StreamExpress_DQMIO": "DQMIO"
}
for goldenOutputMod, tier in viewitems(goldenOutputMods):
mergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Express/ExpressMerge%s" % goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap[
"Merged%s" % tier][0]["merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap[
"Merged%s" % tier][0]["output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(
mergedMergeOutput.name,
"/TestWorkload/Express/ExpressMerge%s/merged-Merged%s" %
(goldenOutputMod, tier),
"Error: Merged output fileset is wrong.")
self.assertEqual(
unmergedMergeOutput.name,
"/TestWorkload/Express/ExpressMerge%s/merged-Merged%s" %
(goldenOutputMod, tier),
"Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0][
"output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(
logArchOutput.name,
"/TestWorkload/Express/ExpressMerge%s/merged-logArchive" %
goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" %
logArchOutput.name)
self.assertEqual(
unmergedLogArchOutput.name,
"/TestWorkload/Express/ExpressMerge%s/merged-logArchive" %
goldenOutputMod, "Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name="TestWorkload-Express")
topLevelFileset.loadData()
expressSubscription = Subscription(fileset=topLevelFileset,
workflow=expressWorkflow)
expressSubscription.loadData()
self.assertEqual(expressSubscription["type"], "Express",
"Error: Wrong subscription type.")
self.assertEqual(
expressSubscription["split_algo"], "Express",
"Error: Wrong split algorithm. %s" %
expressSubscription["split_algo"])
alcaRecoFileset = Fileset(
name=
"/TestWorkload/Express/unmerged-write_StreamExpress_ALCARECOALCARECO"
)
alcaRecoFileset.loadData()
alcaSkimSubscription = Subscription(fileset=alcaRecoFileset,
workflow=alcaSkimWorkflow)
alcaSkimSubscription.loadData()
self.assertEqual(alcaSkimSubscription["type"], "Express",
"Error: Wrong subscription type.")
self.assertEqual(
alcaSkimSubscription["split_algo"], "ExpressMerge",
"Error: Wrong split algorithm. %s" %
alcaSkimSubscription["split_algo"])
mergedDQMFileset = Fileset(
name=
"/TestWorkload/Express/ExpressMergewrite_StreamExpress_DQMIO/merged-MergedDQMIO"
)
mergedDQMFileset.loadData()
dqmSubscription = Subscription(fileset=mergedDQMFileset,
workflow=dqmWorkflow)
dqmSubscription.loadData()
self.assertEqual(dqmSubscription["type"], "Harvesting",
"Error: Wrong subscription type.")
self.assertEqual(dqmSubscription["split_algo"], "Harvest",
"Error: Wrong split algo.")
unmergedOutputs = {
"write_PrimaryDataset1_FEVT": "FEVT",
"write_StreamExpress_DQMIO": "DQMIO"
}
for unmergedOutput, tier in viewitems(unmergedOutputs):
fset = unmergedOutput + tier
unmergedDataTier = Fileset(
name="/TestWorkload/Express/unmerged-%s" % fset)
unmergedDataTier.loadData()
dataTierMergeWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Express/ExpressMerge%s" % unmergedOutput)
dataTierMergeWorkflow.load()
mergeSubscription = Subscription(fileset=unmergedDataTier,
workflow=dataTierMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(
mergeSubscription["split_algo"], "ExpressMerge",
"Error: Wrong split algorithm. %s" %
mergeSubscription["split_algo"])
goldenOutputMods = {
"write_PrimaryDataset1_FEVT": "FEVT",
"write_StreamExpress_ALCARECO": "ALCARECO",
"write_StreamExpress_DQMIO": "DQMIO"
}
for goldenOutputMod, tier in viewitems(goldenOutputMods):
fset = goldenOutputMod + tier
unmergedFileset = Fileset(
name="/TestWorkload/Express/unmerged-%s" % fset)
unmergedFileset.loadData()
cleanupWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Express/ExpressCleanupUnmerged%s" %
goldenOutputMod)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset=unmergedFileset,
workflow=cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"],
"SiblingProcessingBased",
"Error: Wrong subscription type.")
expressLogCollect = Fileset(
name="/TestWorkload/Express/unmerged-logArchive")
expressLogCollect.loadData()
expressLogCollectWorkflow = Workflow(
name="TestWorkload",
task="/TestWorkload/Express/ExpressLogCollect")
expressLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=expressLogCollect,
workflow=expressLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
alcaSkimLogCollect = Fileset(
name=
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/unmerged-logArchive"
)
alcaSkimLogCollect.loadData()
alcaSkimLogCollectWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Express/ExpressAlcaSkimwrite_StreamExpress_ALCARECO/AlcaSkimLogCollect"
)
alcaSkimLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=alcaSkimLogCollect,
workflow=alcaSkimLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
goldenOutputMods = [
"write_PrimaryDataset1_FEVT", "write_StreamExpress_DQMIO"
]
for goldenOutputMod in goldenOutputMods:
expressMergeLogCollect = Fileset(
name="/TestWorkload/Express/ExpressMerge%s/merged-logArchive" %
goldenOutputMod)
expressMergeLogCollect.loadData()
expressMergeLogCollectWorkflow = Workflow(
name="TestWorkload",
task=
"/TestWorkload/Express/ExpressMerge%s/Express%sMergeLogCollect"
% (goldenOutputMod, goldenOutputMod))
expressMergeLogCollectWorkflow.load()
logCollectSubscription = Subscription(
fileset=expressMergeLogCollect,
workflow=expressMergeLogCollectWorkflow)
logCollectSubscription.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algorithm.")
for dqmStrings in [
("ExpressMergewrite_StreamExpress_DQMIOEndOfRunDQMHarvestMerged",
"ExpressMergewrite_StreamExpress_DQMIOMergedEndOfRunDQMHarvestLogCollect"
),
("ExpressMergewrite_StreamExpress_DQMIOPeriodicDQMHarvestMerged",
"ExpressMergewrite_StreamExpress_DQMIOMergedPeriodicDQMHarvestLogCollect"
)
]:
dqmFileset = "/TestWorkload/Express/ExpressMergewrite_StreamExpress_DQMIO/%s/unmerged-logArchive" % \
dqmStrings[0]
dqmHarvestLogCollect = Fileset(name=dqmFileset)
dqmHarvestLogCollect.loadData()
dqmTask = "/TestWorkload/Express/ExpressMergewrite_StreamExpress_DQMIO/%s/%s" % dqmStrings
dqmHarvestLogCollectWorkflow = Workflow(name="TestWorkload",
task=dqmTask)
dqmHarvestLogCollectWorkflow.load()
logCollectSub = Subscription(fileset=dqmHarvestLogCollect,
workflow=dqmHarvestLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
return
def createJobGroups(self,
nSubs,
nJobs,
task,
workloadSpec,
site,
taskType='Processing',
name=None,
wfPrio=1,
changeState=None):
"""
_createJobGroups_
Creates a series of jobGroups for submissions
changeState is an instance of the ChangeState class to make job status changes
"""
jobGroupList = []
if name is None:
name = makeUUID()
testWorkflow = Workflow(spec=workloadSpec,
owner="tapas",
name=name,
task="basicWorkload/Production",
priority=wfPrio)
testWorkflow.create()
# Create subscriptions
for _ in range(nSubs):
name = makeUUID()
# Create Fileset, Subscription, jobGroup
testFileset = Fileset(name=name)
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
type=taskType,
split_algo="FileBased")
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
# Create jobs
self.makeNJobs(name=name,
task=task,
nJobs=nJobs,
jobGroup=testJobGroup,
fileset=testFileset,
sub=testSubscription.exists(),
site=site)
testFileset.commit()
testJobGroup.commit()
jobGroupList.append(testJobGroup)
if changeState:
for group in jobGroupList:
changeState.propagate(group.jobs, 'created', 'new')
return jobGroupList
def createTestJobGroup(self, nJobs = 10, retry_count = 1,
workloadPath = 'test', fwjrPath = None,
workloadName = makeUUID(),
fileModifier = ''):
"""
Creates a group of several jobs
"""
myThread = threading.currentThread()
myThread.transaction.begin()
testWorkflow = Workflow(spec = workloadPath, owner = "cmsdataops", group = "cmsdataops",
name = workloadName, task="/TestWorkload/ReReco")
testWorkflow.create()
testWMBSFileset = Fileset(name = "TestFileset")
testWMBSFileset.create()
testSubscription = Subscription(fileset = testWMBSFileset,
workflow = testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
testFile0 = File(lfn = "/this/is/a/parent%s" % fileModifier, size = 1024, events = 10)
testFile0.addRun(Run(10, *[12312]))
testFile0.setLocation('malpaquet')
testFileA = File(lfn = "/this/is/a/lfnA%s" % fileModifier, size = 1024, events = 10,
first_event = 88)
testFileA.addRun(Run(10, *[12312, 12313]))
testFileA.setLocation('malpaquet')
testFileB = File(lfn = "/this/is/a/lfnB%s" % fileModifier, size = 1024, events = 10,
first_event = 88)
testFileB.addRun(Run(10, *[12314, 12315, 12316]))
testFileB.setLocation('malpaquet')
testFile0.create()
testFileA.create()
testFileB.create()
testFileA.addParent(lfn = "/this/is/a/parent%s" % fileModifier)
testFileB.addParent(lfn = "/this/is/a/parent%s" % fileModifier)
for i in range(0, nJobs):
testJob = Job(name = makeUUID())
testJob['retry_count'] = retry_count
testJob['retry_max'] = 10
testJob['mask'].addRunAndLumis(run = 10, lumis = [12312])
testJob['mask'].addRunAndLumis(run = 10, lumis = [12314, 12316])
testJob['mask']['FirstEvent'] = 100
testJob['cache_dir'] = os.path.join(self.testDir, testJob['name'])
testJob['fwjr_path'] = fwjrPath
os.mkdir(testJob['cache_dir'])
testJobGroup.add(testJob)
testJob.create(group = testJobGroup)
testJob.addFile(testFileA)
testJob.addFile(testFileB)
testJob.save()
testJobGroup.commit()
testSubscription.acquireFiles(files = [testFileA, testFileB])
testSubscription.save()
myThread.transaction.commit()
return testJobGroup
def createTestJobGroup(self,
config,
name="TestWorkthrough",
filesetName="TestFileset",
specLocation="spec.xml",
error=False,
task="/TestWorkload/ReReco",
type="Processing"):
"""
Creates a group of several jobs
"""
myThread = threading.currentThread()
testWorkflow = Workflow(spec=specLocation,
owner=self.OWNERDN,
name=name,
task=task,
owner_vogroup="",
owner_vorole="")
testWorkflow.create()
self.inject.execute(names=[name], injected=True)
testWMBSFileset = Fileset(name=filesetName)
testWMBSFileset.create()
testFileA = File(lfn="/this/is/a/lfnA", size=1024, events=10)
testFileA.addRun(Run(10, *[12312]))
testFileA.setLocation('malpaquet')
testFileB = File(lfn="/this/is/a/lfnB", size=1024, events=10)
testFileB.addRun(Run(10, *[12314]))
testFileB.setLocation('malpaquet')
testFileA.create()
testFileB.create()
testWMBSFileset.addFile(testFileA)
testWMBSFileset.addFile(testFileB)
testWMBSFileset.commit()
testWMBSFileset.markOpen(0)
outputWMBSFileset = Fileset(name='%sOutput' % filesetName)
outputWMBSFileset.create()
testFileC = File(lfn="/this/is/a/lfnC", size=1024, events=10)
testFileC.addRun(Run(10, *[12312]))
testFileC.setLocation('malpaquet')
testFileC.create()
outputWMBSFileset.addFile(testFileC)
outputWMBSFileset.commit()
outputWMBSFileset.markOpen(0)
testWorkflow.addOutput('output', outputWMBSFileset)
testSubscription = Subscription(fileset=testWMBSFileset,
workflow=testWorkflow,
type=type)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
for i in range(0, self.nJobs):
testJob = Job(name=makeUUID())
testJob.addFile(testFileA)
testJob.addFile(testFileB)
testJob['retry_count'] = 1
testJob['retry_max'] = 10
testJob['mask'].addRunAndLumis(run=10, lumis=[12312, 12313])
testJobGroup.add(testJob)
testJobGroup.commit()
changer = ChangeState(config)
report1 = Report()
report2 = Report()
if error:
path1 = os.path.join(WMCore.WMBase.getTestBase(),
"WMComponent_t/JobAccountant_t/fwjrs",
"badBackfillJobReport.pkl")
path2 = os.path.join(WMCore.WMBase.getTestBase(),
'WMComponent_t/TaskArchiver_t/fwjrs',
'logCollectReport2.pkl')
else:
path1 = os.path.join(WMCore.WMBase.getTestBase(),
'WMComponent_t/TaskArchiver_t/fwjrs',
'mergeReport1.pkl')
path2 = os.path.join(WMCore.WMBase.getTestBase(),
'WMComponent_t/TaskArchiver_t/fwjrs',
'logCollectReport2.pkl')
report1.load(filename=path1)
report2.load(filename=path2)
changer.propagate(testJobGroup.jobs, 'created', 'new')
changer.propagate(testJobGroup.jobs, 'executing', 'created')
changer.propagate(testJobGroup.jobs, 'complete', 'executing')
for i in range(self.nJobs):
if i < self.nJobs / 2:
testJobGroup.jobs[i]['fwjr'] = report1
else:
testJobGroup.jobs[i]['fwjr'] = report2
changer.propagate(testJobGroup.jobs, 'jobfailed', 'complete')
changer.propagate(testJobGroup.jobs, 'jobcooloff', 'jobfailed')
changer.propagate(testJobGroup.jobs, 'created', 'jobcooloff')
changer.propagate(testJobGroup.jobs, 'executing', 'created')
changer.propagate(testJobGroup.jobs, 'complete', 'executing')
changer.propagate(testJobGroup.jobs, 'jobfailed', 'complete')
changer.propagate(testJobGroup.jobs, 'retrydone', 'jobfailed')
changer.propagate(testJobGroup.jobs, 'exhausted', 'retrydone')
changer.propagate(testJobGroup.jobs, 'cleanout', 'exhausted')
testSubscription.completeFiles([testFileA, testFileB])
return testJobGroup
def populateWMBS(self):
"""
_populateWMBS_
Create files and subscriptions in WMBS
"""
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationAction = daofactory(classname="Locations.New")
locationAction.execute(siteName='s1', pnn="T1_US_FNAL_Disk")
locationAction.execute(siteName='s2', pnn="T2_CH_CERN")
self.validLocations = ["T1_US_FNAL_Disk", "T2_CH_CERN"]
testWorkflow = Workflow(spec="spec.xml", owner="Steve",
name="wf001", task="Test")
testWorkflow.create()
self.multipleFileFileset = Fileset(name="TestFileset1")
self.multipleFileFileset.create()
parentFile = File('/parent/lfn/', size=1000, events=100,
locations=set(["T1_US_FNAL_Disk"]))
parentFile.create()
for _ in range(10):
newFile = File(makeUUID(), size=1000, events=100,
locations=set(["T1_US_FNAL_Disk"]))
newFile.create()
newFile.addParent(lfn=parentFile['lfn'])
self.multipleFileFileset.addFile(newFile)
self.multipleFileFileset.commit()
self.multipleFileSubscription = Subscription(fileset=self.multipleFileFileset,
workflow=testWorkflow,
split_algo="EventBased",
type="Processing")
self.multipleFileSubscription.create()
self.singleFileFileset = Fileset(name="TestFileset2")
self.singleFileFileset.create()
newFile = File("/some/file/name", size=1000, events=100,
locations=set(["T1_US_FNAL_Disk"]))
newFile.create()
self.singleFileFileset.addFile(newFile)
self.singleFileFileset.commit()
self.singleFileSubscription = Subscription(fileset=self.singleFileFileset,
workflow=testWorkflow,
split_algo="EventBased",
type="Processing")
self.singleFileSubscription.create()
self.multipleSiteFileset = Fileset(name="TestFileset3")
self.multipleSiteFileset.create()
for _ in range(5):
newFile = File(makeUUID(), size=1000, events=100)
newFile.setLocation("T1_US_FNAL_Disk")
newFile.create()
self.multipleSiteFileset.addFile(newFile)
for _ in range(5):
newFile = File(makeUUID(), size=1000, events=100)
newFile.setLocation(["T1_US_FNAL_Disk", "T2_CH_CERN"])
newFile.create()
self.multipleSiteFileset.addFile(newFile)
self.multipleSiteFileset.commit()
self.multipleSiteSubscription = Subscription(fileset=self.multipleSiteFileset,
workflow=testWorkflow,
split_algo="EventBased",
type="Processing")
self.multipleSiteSubscription.create()
return
def createGiantJobSet(self,
name,
config,
nSubs=10,
nJobs=10,
nFiles=1,
spec="spec.xml"):
"""
Creates a massive set of jobs
"""
jobList = []
for i in range(0, nSubs):
# Make a bunch of subscriptions
localName = '%s-%i' % (name, i)
testWorkflow = Workflow(spec=spec,
owner=self.OWNERDN,
name=localName,
task="Test",
owner_vogroup="",
owner_vorole="")
testWorkflow.create()
testWMBSFileset = Fileset(name=localName)
testWMBSFileset.create()
testSubscription = Subscription(fileset=testWMBSFileset,
workflow=testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
filesToComplete = []
for j in range(0, nJobs):
# Create jobs for each subscription
testFileA = File(lfn="%s-%i-lfnA" % (localName, j),
size=1024,
events=10)
testFileA.addRun(
Run(
10, *[
11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36,
37, 38, 39, 40
]))
testFileA.setLocation('malpaquet')
testFileA.create()
testWMBSFileset.addFile(testFileA)
testWMBSFileset.commit()
filesToComplete.append(testFileA)
testJob = Job(name='%s-%i' % (localName, j))
testJob.addFile(testFileA)
testJob['retry_count'] = 1
testJob['retry_max'] = 10
testJobGroup.add(testJob)
jobList.append(testJob)
for k in range(0, nFiles):
# Create output files
testFile = File(lfn="%s-%i-output" % (localName, k),
size=1024,
events=10)
testFile.addRun(Run(10, *[12312]))
testFile.setLocation('malpaquet')
testFile.create()
testJobGroup.output.addFile(testFile)
testJobGroup.output.commit()
testJobGroup.commit()
changer = ChangeState(config)
changer.propagate(testJobGroup.jobs, 'created', 'new')
changer.propagate(testJobGroup.jobs, 'executing', 'created')
changer.propagate(testJobGroup.jobs, 'complete', 'executing')
changer.propagate(testJobGroup.jobs, 'success', 'complete')
changer.propagate(testJobGroup.jobs, 'cleanout', 'success')
testWMBSFileset.markOpen(0)
testSubscription.completeFiles(filesToComplete)
return jobList
def setUp(self):
"""
_setUp_
Create two subscriptions: One that contains a single file and one that
contains multiple files.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMCore.WMBS"],
useDefault=False)
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationAction = daofactory(classname="Locations.New")
locationAction.execute(siteName="site1", seName="somese.cern.ch")
locationAction.execute(siteName="site2", seName="otherse.cern.ch")
self.multipleFileFileset = Fileset(name="TestFileset1")
self.multipleFileFileset.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
newFile.create()
self.multipleFileFileset.addFile(newFile)
self.multipleFileFileset.commit()
self.singleFileFileset = Fileset(name="TestFileset2")
self.singleFileFileset.create()
newFile = File("/some/file/name",
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
newFile.create()
self.singleFileFileset.addFile(newFile)
self.singleFileFileset.commit()
self.multipleSiteFileset = Fileset(name="TestFileset3")
self.multipleSiteFileset.create()
for i in range(5):
newFile = File(makeUUID(), size=1000, events=100)
newFile.setLocation("somese.cern.ch")
newFile.create()
self.multipleSiteFileset.addFile(newFile)
for i in range(5):
newFile = File(makeUUID(), size=1000, events=100)
newFile.setLocation(["somese.cern.ch", "otherse.cern.ch"])
newFile.create()
self.multipleSiteFileset.addFile(newFile)
self.multipleSiteFileset.commit()
testWorkflow = Workflow(spec="spec.xml",
owner="Steve",
name="wf001",
task="Test")
testWorkflow.create()
self.multipleFileSubscription = Subscription(
fileset=self.multipleFileFileset,
workflow=testWorkflow,
split_algo="SizeBased",
type="Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription = Subscription(
fileset=self.singleFileFileset,
workflow=testWorkflow,
split_algo="SizeBased",
type="Processing")
self.singleFileSubscription.create()
self.multipleSiteSubscription = Subscription(
fileset=self.multipleSiteFileset,
workflow=testWorkflow,
split_algo="SizeBased",
type="Processing")
self.multipleSiteSubscription.create()
return
def testMonteCarloFromGEN(self):
"""
_testMonteCarloFromGEN_
Create a MonteCarloFromGEN workflow and verify it installs into WMBS
correctly.
"""
arguments = getTestArguments()
arguments["ConfigCacheID"] = self.injectConfig()
arguments["CouchDBName"] = "mclhe_t"
testWorkload = monteCarloFromGENWorkload("TestWorkload", arguments)
testWorkload.setSpecUrl("somespec")
testWorkload.setOwnerDetails("*****@*****.**", "DMWM")
testWMBSHelper = WMBSHelper(testWorkload, "MonteCarloFromGEN", "SomeBlock")
testWMBSHelper.createTopLevelFileset()
testWMBSHelper.createSubscription(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
procWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN")
procWorkflow.load()
self.assertEqual(len(procWorkflow.outputMap.keys()), 3,
"Error: Wrong number of WF outputs.")
self.assertEqual(procWorkflow.wfType, 'lheproduction')
goldenOutputMods = ["outputRECORECO", "outputALCARECOALCARECO"]
for goldenOutputMod in goldenOutputMods:
mergedOutput = procWorkflow.outputMap[goldenOutputMod][0]["merged_output_fileset"]
unmergedOutput = procWorkflow.outputMap[goldenOutputMod][0]["output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(mergedOutput.name, "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong: %s" % mergedOutput.name)
self.assertEqual(unmergedOutput.name, "/TestWorkload/MonteCarloFromGEN/unmerged-%s" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = procWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = procWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name, "/TestWorkload/MonteCarloFromGEN/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
self.assertEqual(unmergedLogArchOutput.name, "/TestWorkload/MonteCarloFromGEN/unmerged-logArchive",
"Error: LogArchive output fileset is wrong.")
for goldenOutputMod in goldenOutputMods:
mergeWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s" % goldenOutputMod)
mergeWorkflow.load()
self.assertEqual(len(mergeWorkflow.outputMap.keys()), 2,
"Error: Wrong number of WF outputs.")
mergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["merged_output_fileset"]
unmergedMergeOutput = mergeWorkflow.outputMap["Merged"][0]["output_fileset"]
mergedMergeOutput.loadData()
unmergedMergeOutput.loadData()
self.assertEqual(mergedMergeOutput.name, "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s/merged-Merged" % goldenOutputMod,
"Error: Merged output fileset is wrong.")
self.assertEqual(unmergedMergeOutput.name, "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s/merged-Merged" % goldenOutputMod,
"Error: Unmerged output fileset is wrong.")
logArchOutput = mergeWorkflow.outputMap["logArchive"][0]["merged_output_fileset"]
unmergedLogArchOutput = mergeWorkflow.outputMap["logArchive"][0]["output_fileset"]
logArchOutput.loadData()
unmergedLogArchOutput.loadData()
self.assertEqual(logArchOutput.name, "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong: %s" % logArchOutput.name)
self.assertEqual(unmergedLogArchOutput.name, "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMerge%s/merged-logArchive" % goldenOutputMod,
"Error: LogArchive output fileset is wrong.")
topLevelFileset = Fileset(name = "TestWorkload-MonteCarloFromGEN-SomeBlock")
topLevelFileset.loadData()
procSubscription = Subscription(fileset = topLevelFileset, workflow = procWorkflow)
procSubscription.loadData()
self.assertEqual(procSubscription["type"], "Production",
"Error: Wrong subscription type: %s" % procSubscription["type"])
self.assertEqual(procSubscription["split_algo"], "LumiBased",
"Error: Wrong split algo.")
unmergedReco = Fileset(name = "/TestWorkload/MonteCarloFromGEN/unmerged-outputRECORECO")
unmergedReco.loadData()
recoMergeWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputRECORECO")
recoMergeWorkflow.load()
mergeSubscription = Subscription(fileset = unmergedReco, workflow = recoMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo: %s" % mergeSubscription["split_algo"])
unmergedAlca = Fileset(name = "/TestWorkload/MonteCarloFromGEN/unmerged-outputALCARECOALCARECO")
unmergedAlca.loadData()
alcaMergeWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputALCARECOALCARECO")
alcaMergeWorkflow.load()
mergeSubscription = Subscription(fileset = unmergedAlca, workflow = alcaMergeWorkflow)
mergeSubscription.loadData()
self.assertEqual(mergeSubscription["type"], "Merge",
"Error: Wrong subscription type.")
self.assertEqual(mergeSubscription["split_algo"], "ParentlessMergeBySize",
"Error: Wrong split algo: %s" % mergeSubscription["split_algo"])
for procOutput in ["outputRECORECO", "outputALCARECOALCARECO"]:
unmerged = Fileset(name = "/TestWorkload/MonteCarloFromGEN/unmerged-%s" % procOutput)
unmerged.loadData()
cleanupWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENCleanupUnmerged%s" % procOutput)
cleanupWorkflow.load()
cleanupSubscription = Subscription(fileset = unmerged, workflow = cleanupWorkflow)
cleanupSubscription.loadData()
self.assertEqual(cleanupSubscription["type"], "Cleanup",
"Error: Wrong subscription type.")
self.assertEqual(cleanupSubscription["split_algo"], "SiblingProcessingBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name = "/TestWorkload/MonteCarloFromGEN/unmerged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/LogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset = procLogCollect, workflow = procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputRECORECO/merged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputRECORECO/MonteCarloFromGENoutputRECORECOMergeLogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset = procLogCollect, workflow = procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
procLogCollect = Fileset(name = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputALCARECOALCARECO/merged-logArchive")
procLogCollect.loadData()
procLogCollectWorkflow = Workflow(name = "TestWorkload",
task = "/TestWorkload/MonteCarloFromGEN/MonteCarloFromGENMergeoutputALCARECOALCARECO/MonteCarloFromGENoutputALCARECOALCARECOMergeLogCollect")
procLogCollectWorkflow.load()
logCollectSub = Subscription(fileset = procLogCollect, workflow = procLogCollectWorkflow)
logCollectSub.loadData()
self.assertEqual(logCollectSub["type"], "LogCollect",
"Error: Wrong subscription type.")
self.assertEqual(logCollectSub["split_algo"], "MinFileBased",
"Error: Wrong split algo.")
return
