def testCreateDeleteExistsNoFiles(self):
"""
_testCreateDeleteExistsNoFiles_
Create and then delete a job but don't add any input files to it.
Use the job class's exists() method to determine if the job has been
written to the database before it is created, after it has been created
and after it has been deleted.
"""
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()
testJob = Job(name="TestJob")
assert testJob.exists() == False, "ERROR: Job exists before it was created"
testJob.create(group=testJobGroup)
assert testJob.exists() >= 0, "ERROR: Job does not exist after it was created"
testJob.delete()
assert testJob.exists() == False, "ERROR: Job exists after it was delete"
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 createLargeFileBlock(self):
"""
_createLargeFileBlock_
Creates a large group of files for testing
"""
testFileset = Fileset(name = "TestFilesetX")
testFileset.create()
for i in range(5000):
newFile = File(makeUUID(), size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testWorkflow = Workflow(spec = "spec.xml", owner = "mnorman",
name = "wf003", task="Test" )
testWorkflow.create()
largeSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow,
split_algo = "FileBased",
type = "Processing")
largeSubscription.create()
return largeSubscription
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 testFilesWithoutOtherSubscriptions(self):
"""
_testFilesWithoutOtherSubscriptions_
Test the case where files only in the delete subscription
can happen if cleanup of the other subscriptions is fast
"""
testWorkflowA = Workflow(spec="specA.xml", owner="Steve", name="wfA", task="Test")
testWorkflowA.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
allFiles = []
for i in range(500):
testFile = File(str(i), size=1000, events=100, locations=set(["somese.cern.ch"]))
testFile.create()
allFiles.append(testFile)
testFileset.addFile(testFile)
testFileset.commit()
testSubscriptionA = Subscription(
fileset=testFileset, workflow=testWorkflowA, split_algo="SiblingProcessingBased", type="Processing"
)
testSubscriptionA.create()
splitter = SplitterFactory()
deleteFactoryA = splitter(package="WMCore.WMBS", subscription=testSubscriptionA)
result = deleteFactoryA(files_per_job=50)
self.assertEqual(len(result), 1, "Error: Wrong number of job groups returned.")
self.assertEqual(len(result[0].jobs), 10, "Error: Wrong number of jobs returned.")
return
def testByRunHarvesting(self):
"""
_testByRunHarvesting_
Provided a fileset with a couple of files and 4 different runs, create
one single job per run and location.
The multiRun baggage should be false in this case.
"""
multipleFilesFileset = createCommonFileset()
self.assertEqual(multipleFilesFileset.open, True, "Fileset should be open!")
harvestingWorkflow = Workflow(spec="spec.xml",
owner="amaltaro",
name="TestWorkflow",
task="Test")
harvestingWorkflow.create()
harvestSub = Subscription(fileset=multipleFilesFileset,
workflow=harvestingWorkflow,
split_algo="Harvest",
type="Harvesting")
harvestSub.create()
multipleFilesFileset.markOpen(False)
self.assertEqual(multipleFilesFileset.open, False, "Fileset should now be closed")
jobFactory = self.splitterFactory(package="WMCore.WMBS", subscription=harvestSub)
jobGroups = jobFactory()
self.assertEqual(len(jobGroups), 1, "Should have created 1 job group")
for jobGroup in jobGroups:
self.assertEqual(len(jobGroup.jobs), 6, "Should have created 6 jobs")
for job in jobGroup.jobs:
baggage = job.getBaggage()
self.assertFalse(getattr(baggage, "multiRun", False), "It's supposed to be a byRun job")
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 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 testCreateDeleteExists(self):
"""
_testCreateDeleteExists_
Create a JobGroup and then delete it. Use the JobGroup's exists()
method to determine if it exists before it is created, after it is
created and after it is deleted.
"""
testWorkflow = Workflow(spec="spec.xml", owner="Simon",
name="wf001", task="Test")
testWorkflow.create()
testFileset = WMBSFileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
self.assertFalse(testJobGroup.exists())
testJobGroup.create()
self.assertTrue(testJobGroup.exists())
testJobGroup.delete()
self.assertFalse(testJobGroup.exists())
testSubscription.delete()
testFileset.delete()
testWorkflow.delete()
return
def testCreateTransaction(self):
"""
_testCreateTransaction_
Create a JobGroup and commit it to the database. Rollback the database
transaction and verify that the JobGroup is no longer in the database.
"""
testWorkflow = Workflow(spec="spec.xml", owner="Simon", name="wf001", task="Test")
testWorkflow.create()
testFileset = WMBSFileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset, workflow=testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
assert testJobGroup.exists() == False, "ERROR: Job group exists before it was created"
myThread = threading.currentThread()
myThread.transaction.begin()
testJobGroup.create()
assert testJobGroup.exists() >= 0, "ERROR: Job group does not exist after it was created"
myThread.transaction.rollback()
assert testJobGroup.exists() == False, "ERROR: Job group exists after transaction was rolled back."
testSubscription.delete()
testFileset.delete()
testWorkflow.delete()
return
def createSingleJobWorkflow(self):
"""
Create a workflow with one jobs and two files and store the results in instance variables
"""
self.testWorkflow = Workflow(spec="spec.xml", owner="Simon", name="wf001", task="Test")
self.testWorkflow.create()
testWMBSFileset = Fileset(name="TestFileset")
testWMBSFileset.create()
testSubscription = Subscription(fileset=testWMBSFileset, workflow=self.testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
self.testFileA = File(lfn="/this/is/a/lfnA", size=1024, events=10)
self.testFileA.addRun(Run(1, *[45]))
self.testFileB = File(lfn="/this/is/a/lfnB", size=1024, events=10)
self.testFileB.addRun(Run(1, *[46]))
self.testFileA.create()
self.testFileB.create()
self.testJob = Job(name="TestJob", files=[self.testFileA, self.testFileB])
self.testJob.create(group=testJobGroup)
self.testJob.associateFiles()
def generateFakeMCFile(self, numEvents = 100, firstEvent = 1,
lastEvent = 100, firstLumi = 1, lastLumi = 10,
index = 1):
#MC comes with only one MCFakeFile
singleMCFileset = Fileset(name = "MCTestFileset %i" % index)
singleMCFileset.create()
newFile = File("MCFakeFileTest %i" % index, size = 1000,
events = numEvents,
locations = set(["somese.cern.ch"]))
newFile.addRun(Run(1, *range(firstLumi, lastLumi + 1)))
newFile["first_event"] = firstEvent
newFile["last_event"] = lastEvent
newFile.create()
singleMCFileset.addFile(newFile)
singleMCFileset.commit()
testWorkflow = Workflow(spec = "spec.xml", owner = "Steve",
name = "wf001", task="Test")
testWorkflow.create()
singleMCFileSubscription = Subscription(fileset = singleMCFileset,
workflow = testWorkflow,
split_algo = "EventBased",
type = "Production")
singleMCFileSubscription.create()
return singleMCFileSubscription
def createTestSubscription(self, nFiles, nSites=1, closeFileset=False):
"""
_createTestSubscription_
Create a set of test subscriptions for testing purposes.
"""
if nSites > self.nSites:
nSites = self.nSites
testFileset = Fileset(name="TestFileset")
testFileset.create()
# Create a testWorkflow
testWorkflow = Workflow(spec="spec.xml", owner="Steve", name="wf001", task="Test")
testWorkflow.create()
# Create the files for each site
for s in range(nSites):
for i in range(nFiles):
newFile = File(makeUUID(), size=1024, events=100, locations=set(["site%i.cern.ch" % s]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(
fileset=testFileset, workflow=testWorkflow, split_algo="MinFileBased", type="Processing"
)
testSubscription.create()
# Close the fileset
if closeFileset:
testFileset.markOpen(isOpen=False)
return testSubscription
def subscribeWMBS(self, task):
"""
Create a subscription for each task
"""
workFlow, fileSet = self.createWorkflow(task)
workFlow.load()
fileSet.load()
subType = ''
if task.name() == 'Processing' or task.name() == 'Production':
subType = 'Processing'
elif task.name() == 'Merge':
subType = 'Merge'
newSub = Subscription(fileset=fileSet, workflow=workFlow, split_algo='FileBased', type=subType)
newSub.create()
# Add subscription to dictionary
self.subDict[task.name()] = newSub
# Add subscription id to task
setattr(task.data.input.WMBS, 'Subscription', newSub['id'])
if not newSub.exists() >= 0:
raise Exception("ERROR: Subscription does not exist after it was created")
logging.info('Created subscription for task %s' % (task.name()))
return
def subscribeWMBS(self, task):
"""
Create a subscription for each task
"""
workFlow, fileSet = self.createWorkflow(task)
workFlow.load()
fileSet.load()
subType = ""
if task.name() == "Processing" or task.name() == "Production":
subType = "Processing"
elif task.name() == "Merge":
subType = "Merge"
newSub = Subscription(fileset=fileSet, workflow=workFlow, split_algo="FileBased", type=subType)
newSub.create()
# Add subscription to dictionary
self.subDict[task.name()] = newSub
# Add subscription id to task
setattr(task.data.input.WMBS, "Subscription", newSub["id"])
if not newSub.exists() >= 0:
raise Exception("ERROR: Subscription does not exist after it was created")
logging.info("Created subscription for task %s" % (task.name()))
return
def testByRunAndRunBlacklist(self):
"""
_testByRunAndRunWhitelist_
Create harvesting jobs by run for the runs provided in the RunWhitelist
"""
multipleFilesFileset = createCommonFileset()
self.assertEqual(multipleFilesFileset.open, True)
harvestingWorkflow = Workflow(spec="spec.xml", owner="amaltaro",
name="TestWorkflow", task="Test")
harvestingWorkflow.create()
harvestSub = Subscription(fileset=multipleFilesFileset, workflow=harvestingWorkflow,
split_algo="Harvest", type="Harvesting")
harvestSub.create()
multipleFilesFileset.markOpen(False)
self.assertEqual(multipleFilesFileset.open, False, "Fileset should now be closed")
jobFactory = self.splitterFactory(package="WMCore.WMBS", subscription=harvestSub)
jobGroups = jobFactory(runWhitelist=[1, 2, 3, 4, 5], runBlacklist=[1, 3])
self.assertEqual(len(jobGroups), 1, "One jobgroup per location")
for jobGroup in jobGroups:
self.assertEqual(len(jobGroup.jobs), 3)
def createSubscription(self, nFiles, lumisPerFile, twoSites=False, nEventsPerFile=100):
"""
_createSubscription_
Create a subscription for testing
"""
baseName = makeUUID()
testFileset = Fileset(name=baseName)
testFileset.create()
for i in range(nFiles):
newFile = self.createFile("%s_%i" % (baseName, i), nEventsPerFile, i, lumisPerFile, "somese.cern.ch")
newFile.create()
testFileset.addFile(newFile)
if twoSites:
for i in range(nFiles):
newFile = self.createFile("%s_%i_2" % (baseName, i), nEventsPerFile, i, lumisPerFile, "otherse.cern.ch")
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(
fileset=testFileset, workflow=self.testWorkflow, split_algo="EventAwareLumiBased", type="Processing"
)
testSubscription.create()
return testSubscription
def createSubscription(self, nFiles, lumisPerFile, twoSites=False, nEventsPerFile=100):
"""
_createSubscription_
Create a subscription for testing
"""
baseName = makeUUID()
testWorkflow = Workflow(spec="spec.xml", owner="dmwm",
name="testWorkflow_%s" % baseName[:4], task="Test")
testWorkflow.create()
testFileset = Fileset(name=baseName)
for i in range(nFiles):
newFile = self.createFile('%s_%i' % (baseName, i), nEventsPerFile,
i, lumisPerFile, 'T1_US_FNAL_Disk')
testFileset.addFile(newFile)
if twoSites:
for i in range(nFiles):
newFile = self.createFile('%s_%i_2' % (baseName, i), nEventsPerFile,
i, lumisPerFile, 'T2_CH_CERN')
testFileset.addFile(newFile)
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="EventAwareLumiByWork",
type="Processing")
testSubscription.create()
return testSubscription
def testMultiRunHarvesting(self):
"""
_testMultiRunHarvesting_
Provided a fileset with a couple of files and different runs, create a
single job for all the runs at a specific location, which also adds a
baggage to the job (True) which is later on looked up by SetupCMSSWPSet.
"""
multipleFilesFileset = createCommonFileset()
self.assertEqual(multipleFilesFileset.open, True)
harvestingWorkflow = Workflow(spec="spec.xml",
owner="amaltaro",
name="TestWorkflow",
task="Test")
harvestingWorkflow.create()
harvestSub = Subscription(fileset=multipleFilesFileset,
workflow=harvestingWorkflow,
split_algo="Harvest",
type="Harvesting")
harvestSub.create()
multipleFilesFileset.markOpen(False)
self.assertEqual(multipleFilesFileset.open, False, "Fileset should now be closed")
jobFactory = self.splitterFactory(package="WMCore.WMBS", subscription=harvestSub)
jobGroups = jobFactory(dqmHarvestUnit="multiRun")
self.assertEqual(len(jobGroups), 1)
for jobGroup in jobGroups:
self.assertEqual(len(jobGroup.jobs), 1)
def createTestJob(subscriptionType="Merge"):
"""
_createTestJob_
Create a test job with two files as input. This will also create the
appropriate workflow, jobgroup and subscription.
"""
testWorkflow = Workflow(spec=makeUUID(), owner="Simon", name=makeUUID(), task="Test")
testWorkflow.create()
testWMBSFileset = Fileset(name="TestFileset")
testWMBSFileset.create()
testSubscription = Subscription(fileset=testWMBSFileset, workflow=testWorkflow, type=subscriptionType)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
testFileA = File(lfn="/this/is/a/lfnA", size=1024, events=10)
testFileA.addRun(Run(1, *[45]))
testFileB = File(lfn="/this/is/a/lfnB", size=1024, events=10)
testFileB.addRun(Run(1, *[46]))
testFileA.create()
testFileB.create()
testJob = Job(name=makeUUID(), files=[testFileA, testFileB])
testJob["couch_record"] = "somecouchrecord"
testJob["location"] = "test.site.ch"
testJob.create(group=testJobGroup)
testJob.associateFiles()
return testJob
def testD_NonContinuousLumis(self):
"""
_NonContinuousLumis_
Test and see if LumiBased can work when the lumis are non continuous
"""
baseName = makeUUID()
nFiles = 10
testFileset = Fileset(name = baseName)
testFileset.create()
for i in range(nFiles):
newFile = File(lfn = '%s_%i' % (baseName, i), size = 1000,
events = 100, locations = "somese.cern.ch")
# Set to two non-continuous lumi numbers
lumis = [100 + i, 200 + i]
newFile.addRun(Run(i, *lumis))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset = testFileset,
workflow = self.testWorkflow,
split_algo = "LumiBased",
type = "Processing")
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package = "WMCore.WMBS",
subscription = testSubscription)
jobGroups = jobFactory(lumis_per_job = 2,
halt_job_on_file_boundaries = False,
splitOnRun = False,
performance = self.performanceParams)
self.assertEqual(len(jobGroups), 1)
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 10)
for j in jobs:
runs = j['mask'].getRunAndLumis()
for r in runs.keys():
self.assertEqual(len(runs[r]), 2)
for l in runs[r]:
# Each run should have two lumis
# Each lumi should be of form [x, x]
# meaning that the first and last lumis are the same
self.assertEqual(len(l), 2)
self.assertEqual(l[0], l[1])
self.assertEqual(j['estimatedJobTime'], 100 * 12)
self.assertEqual(j['estimatedDiskUsage'], 100 * 400)
self.assertEqual(j['estimatedMemoryUsage'], 2300)
return
def _createThisSubscription(self, initialCounter=1):
"""
Private function to create a fileset and subscription with
different fileset and file names
:param initialCounter: just a simple integer to be appended to files
:return: an splitter instance (jobFactory)
"""
splitter = SplitterFactory()
# Create 3 files with 100 events per lumi:
# - file1 with 1 run of 8 lumis
# - file2 with 2 runs of 2 lumis each
# - file3 with 1 run of 5 lumis
testFileset = Fileset(name='Fileset%s' % initialCounter)
fileA = File(lfn="/this/is/file%s" % initialCounter, size=1000, events=800)
lumiListA = []
for lumi in range(8):
lumiListA.append(10 + lumi)
fileA.addRun(Run(1, *lumiListA))
fileA.setLocation("T1_US_FNAL_Disk")
initialCounter = int(initialCounter) + 1
fileB = File(lfn="/this/is/file%s" % initialCounter, size=1000, events=400)
lumiListB1 = []
lumiListB2 = []
for lumi in range(2):
lumiListB1.append(20 + lumi)
lumiListB2.append(30 + lumi)
fileB.addRun(Run(2, *lumiListB1))
fileB.addRun(Run(3, *lumiListB2))
fileB.setLocation("T1_US_FNAL_Disk")
initialCounter = int(initialCounter) + 1
fileC = File(lfn="/this/is/file%s" % initialCounter, size=1000, events=500)
lumiListC = []
for lumi in range(5):
lumiListC.append(40 + lumi)
fileC.addRun(Run(4, *lumiListC))
fileC.setLocation("T1_US_FNAL_Disk")
testFileset.addFile(fileA)
testFileset.addFile(fileB)
testFileset.addFile(fileC)
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiByWork",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
return jobFactory
def setupRepackWorkflow(self):
"""
_setupRepackWorkflow_
Populate WMBS with a repack-like workflow,
every subscription must be unfinished at first
"""
workflowName = 'Repack_Run481516_StreamZ'
mergeTasks = ['RepackMergewrite_QuadElectron_RAW', 'RepackMergewrite_TriPhoton_RAW',
'RepackMergewrite_SingleNeutrino_RAW']
self.stateMap = {'Merge': [],
'Processing Done': []}
self.orderedStates = ['Merge', 'Processing Done']
# Populate WMStats
self.requestDBWriter.insertGenericRequest({'RequestName': workflowName})
self.requestDBWriter.updateRequestStatus(workflowName, 'Closed')
# Create a wmspec in disk
workload = newWorkload(workflowName)
repackTask = workload.newTask('Repack')
for task in mergeTasks:
repackTask.addTask(task)
repackTask.addTask('RepackCleanupUnmergedwrite_QuadElectron_RAW')
specPath = os.path.join(self.testDir, 'Repack.pkl')
workload.save(specPath)
# Populate WMBS
topFileset = Fileset(name='TestStreamerFileset')
topFileset.create()
options = {'spec': specPath, 'owner': 'ItsAMeMario',
'name': workflowName, 'wfType': 'tier0'}
topLevelWorkflow = Workflow(task='/%s/Repack' % workflowName,
**options)
topLevelWorkflow.create()
topLevelSub = Subscription(topFileset, topLevelWorkflow)
topLevelSub.create()
self.stateMap['Merge'].append(topFileset)
for task in mergeTasks:
mergeWorkflow = Workflow(task='/%s/Repack/%s' % (workflowName, task), **options)
mergeWorkflow.create()
unmergedFileset = Fileset(name='TestUnmergedFileset%s' % task)
unmergedFileset.create()
mergeSub = Subscription(unmergedFileset, mergeWorkflow)
mergeSub.create()
self.stateMap['Processing Done'].append(unmergedFileset)
cleanupWorkflow = Workflow(task='/Repack_Run481516_StreamZ/Repack/RepackCleanupUnmergedwrite_QuadElectron_RAW',
**options)
cleanupWorkflow.create()
unmergedFileset = Fileset(name='TestUnmergedFilesetToCleanup')
unmergedFileset.create()
cleanupSub = Subscription(unmergedFileset, cleanupWorkflow)
cleanupSub.create()
return
def testParentageByJob(self):
"""
_testParentageByJob_
Tests the DAO that assigns parentage by Job
"""
testWorkflow = Workflow(spec = 'hello', owner = "mnorman",
name = "wf001", task="basicWorkload/Production")
testWorkflow.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
testSubscription = Subscription(fileset = testFileset, workflow = testWorkflow, type = "Processing", split_algo = "FileBased")
testSubscription.create()
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
testFileParentA = File(lfn = "/this/is/a/parent/lfnA", size = 1024,
events = 20, checksums = {'cksum': 1})
testFileParentA.addRun(Run( 1, *[45]))
testFileParentB = File(lfn = "/this/is/a/parent/lfnB", size = 1024,
events = 20, checksums = {'cksum': 1})
testFileParentB.addRun(Run( 1, *[45]))
testFileParentA.create()
testFileParentB.create()
testFileA = File(lfn = "/this/is/a/lfn", size = 1024, events = 10,
checksums = {'cksum':1})
testFileA.addRun(Run( 1, *[45]))
testFileA.create()
testJobA = Job()
testJobA.create(group = testJobGroup)
testJobA.addFile(testFileParentA)
testJobA.addFile(testFileParentB)
testJobA.associateFiles()
parentAction = self.daofactory(classname = "Files.SetParentageByJob")
parentAction.execute(binds = {'jobid': testJobA.exists(), 'child': testFileA['lfn']})
testFileB = File(id = testFileA["id"])
testFileB.loadData(parentage = 1)
goldenFiles = [testFileParentA, testFileParentB]
for parentFile in testFileB["parents"]:
self.assertEqual(parentFile in goldenFiles, True,
"ERROR: Unknown parent file")
goldenFiles.remove(parentFile)
self.assertEqual(len(goldenFiles), 0,
"ERROR: Some parents are missing")
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 createJobGroups(self, nSubs, nJobs, task, workloadSpec, site=None, bl=[], wl=[]):
"""
Creates a series of jobGroups for submissions
"""
jobGroupList = []
testWorkflow = Workflow(
spec=workloadSpec,
owner="tapas",
name=makeUUID(),
task="basicWorkload/Production",
owner_vogroup="phgroup",
owner_vorole="cmsrole",
)
testWorkflow.create()
# Create subscriptions
for i in range(nSubs):
name = makeUUID()
# Create Fileset, Subscription, jobGroup
testFileset = Fileset(name=name)
testFileset.create()
testSubscription = Subscription(
fileset=testFileset, workflow=testWorkflow, type="Processing", 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,
bl=bl,
wl=wl,
)
testFileset.commit()
testJobGroup.commit()
jobGroupList.append(testJobGroup)
return jobGroupList
def testDeleteTransaction(self):
"""
_testDeleteTransaction_
Create a new job and commit it to the database. Start a new transaction
and delete the file from the database. Verify that the file has been
deleted. After that, roll back the transaction and verify that the
job is once again in the database.
"""
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)
testFileB = File(lfn="/this/is/a/lfnB", size=1024, events=10)
testFileA.create()
testFileB.create()
testJob = Job(name="TestJob", files=[testFileA, testFileB])
assert testJob.exists() is False, \
"ERROR: Job exists before it was created"
testJob.create(group=testJobGroup)
assert testJob.exists() >= 0, \
"ERROR: Job does not exist after it was created"
myThread = threading.currentThread()
myThread.transaction.begin()
testJob.delete()
assert testJob.exists() is False, \
"ERROR: Job exists after it was delete"
myThread.transaction.rollback()
assert testJob.exists() >= 0, \
"ERROR: Job does not exist after transaction was rolled back."
return
def stuffWMBS(self):
"""
_stuffWMBS_
Stuff WMBS with workflows
"""
workflow = Workflow(spec = 'spec.xml', name = 'ReRecoTest_v0Emulator',
task = '/ReRecoTest_v0Emulator/Test', priority = 10)
workflow.create()
inputFileset = Fileset(name = 'TestFileset')
inputFileset.create()
subscription = Subscription(inputFileset, workflow)
subscription.create()
def createSubscription(self, nFiles, lumisPerFile, twoSites = False, rand = False):
"""
_createSubscription_
Create a subscription for testing
"""
baseName = makeUUID()
testFileset = Fileset(name = baseName)
testFileset.create()
parentFile = File('%s_parent' % (baseName), size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
parentFile.create()
for i in range(nFiles):
newFile = File(lfn = '%s_%i' % (baseName, i), size = 1000,
events = 100, locations = "somese.cern.ch")
lumis = []
for lumi in range(lumisPerFile):
if rand:
lumis.append(random.randint(1000 * i, 1000 * (i + 1)))
else:
lumis.append((100 * i) + lumi)
newFile.addRun(Run(i, *lumis))
newFile.create()
newFile.addParent(parentFile['lfn'])
testFileset.addFile(newFile)
if twoSites:
for i in range(nFiles):
newFile = File(lfn = '%s_%i_2' % (baseName, i), size = 1000,
events = 100, locations = "otherse.cern.ch")
lumis = []
for lumi in range(lumisPerFile):
if rand:
lumis.append(random.randint(1000 * i, 1000 * (i + 1)))
else:
lumis.append((100 * i) + lumi)
newFile.addRun(Run(i, *lumis))
newFile.create()
newFile.addParent(parentFile['lfn'])
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset = testFileset,
workflow = self.testWorkflow,
split_algo = "LumiBased",
type = "Processing")
testSubscription.create()
return testSubscription
def createTestJobs(self, nJobs, cacheDir):
"""
_createTestJobs_
Create several jobs
"""
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,
type = "Processing",
split_algo = "FileBased")
testSubscription.create()
testJobGroup = JobGroup(subscription = testSubscription)
testJobGroup.create()
# Create a file
testFileA = File(lfn = "/this/is/a/lfnA", size = 1024, events = 10)
testFileA.addRun(Run(10, *[12312]))
testFileA.setLocation('malpaquet')
testFileA.create()
baseName = makeUUID()
# Now create a job
for i in range(nJobs):
testJob = Job(name = '%s-%i' % (baseName, i))
testJob.addFile(testFileA)
testJob['location'] = 'malpaquet'
testJob['retry_count'] = 1
testJob['retry_max'] = 10
testJob.create(testJobGroup)
testJob.save()
testJobGroup.add(testJob)
testJobGroup.commit()
# Set test job caches
for job in testJobGroup.jobs:
job.setCache(cacheDir)
return testJobGroup
class SiblingProcessingBasedTest(unittest.TestCase):
"""
_SiblingProcessingBasedTest_
Test SiblingProcessing job splitting.
"""
def setUp(self):
"""
_setUp_
Setup the database connections and schema.
"""
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("site1", seName = "somese.cern.ch")
locationAction.execute("site2", seName = "somese2.cern.ch")
self.testFilesetA = Fileset(name = "FilesetA")
self.testFilesetA.create()
self.testFilesetB = Fileset(name = "FilesetB")
self.testFilesetB.create()
self.testFileA = File("testFileA", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileA.create()
self.testFileB = File("testFileB", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileB.create()
self.testFileC = File("testFileC", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileC.create()
self.testFilesetA.addFile(self.testFileA)
self.testFilesetA.addFile(self.testFileB)
self.testFilesetA.addFile(self.testFileC)
self.testFilesetA.commit()
self.testFileD = File("testFileD", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileD.create()
self.testFileE = File("testFileE", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileE.create()
self.testFileF = File("testFileF", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileF.create()
self.testFilesetB.addFile(self.testFileD)
self.testFilesetB.addFile(self.testFileE)
self.testFilesetB.addFile(self.testFileF)
self.testFilesetB.commit()
testWorkflowA = Workflow(spec = "specA.xml", owner = "Steve",
name = "wfA", task = "Test")
testWorkflowA.create()
testWorkflowB = Workflow(spec = "specB.xml", owner = "Steve",
name = "wfB", task = "Test")
testWorkflowB.create()
testWorkflowC = Workflow(spec = "specC.xml", owner = "Steve",
name = "wfC", task = "Test")
testWorkflowC.create()
testWorkflowD = Workflow(spec = "specD.xml", owner = "Steve",
name = "wfD", task = "Test")
testWorkflowD.create()
self.testSubscriptionA = Subscription(fileset = self.testFilesetA,
workflow = testWorkflowA,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionA.create()
self.testSubscriptionB = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowB,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionB.create()
self.testSubscriptionC = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowC,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionC.create()
self.testSubscriptionD = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowD,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionD.create()
deleteWorkflow = Workflow(spec = "specE.xml", owner = "Steve",
name = "wfE", task = "Test")
deleteWorkflow.create()
self.deleteSubscriptionA = Subscription(fileset = self.testFilesetA,
workflow = deleteWorkflow,
split_algo = "SiblingProcessingBased",
type = "Cleanup")
self.deleteSubscriptionA.create()
self.deleteSubscriptionB = Subscription(fileset = self.testFilesetB,
workflow = deleteWorkflow,
split_algo = "SiblingProcessingBased",
type = "Cleanup")
self.deleteSubscriptionB.create()
return
def tearDown(self):
"""
_tearDown_
Clear out WMBS.
"""
self.testInit.clearDatabase()
return
def testSiblingProcessing(self):
"""
_testSiblingProcessing_
Verify that the sibling processing split works correctly dealing with
failed files and acquiring files correctly.
"""
splitter = SplitterFactory()
deleteFactoryA = splitter(package = "WMCore.WMBS",
subscription = self.deleteSubscriptionA)
deleteFactoryB = splitter(package = "WMCore.WMBS",
subscription = self.deleteSubscriptionB)
result = deleteFactoryA()
assert len(result) == 0, \
"Error: No jobs should be returned."
result = deleteFactoryB()
assert len(result) == 0, \
"Error: No jobs should be returned."
self.testSubscriptionA.completeFiles(self.testFileA)
result = deleteFactoryA(files_per_job = 1)
assert len(result) == 1, \
"Error: Only one jobgroup should be returned."
assert len(result[0].jobs) == 1, \
"Error: There should only be one job in the jobgroup."
assert len(result[0].jobs[0]["input_files"]) == 1, \
"Error: Job should only have one input file."
assert result[0].jobs[0]["input_files"][0]["lfn"] == "testFileA", \
"Error: Input file for job is wrong."
assert list(result[0].jobs[0]["input_files"][0]["locations"]) == ["somese.cern.ch"], \
"Error: File location is wrong."
result = deleteFactoryB(files_per_job = 1)
assert len(result) == 0, \
"Error: Second subscription should have no jobs."
result = deleteFactoryA(files_per_job = 1)
assert len(result) == 0, \
"Error: No jobs should have been created."
self.testSubscriptionB.completeFiles(self.testFileD)
self.testSubscriptionC.failFiles(self.testFileD)
result = deleteFactoryA(files_per_job = 1)
assert len(result) == 0, \
"Error: No jobs should have been created."
result = deleteFactoryB(files_per_job = 1)
assert len(result) == 0, \
"Error: No jobs should have been created."
self.testSubscriptionD.failFiles(self.testFileD)
result = deleteFactoryA(files_per_job = 1)
assert len(result) == 0, \
"Error: No jobs should have been created."
result = deleteFactoryB(files_per_job = 1)
assert len(result) == 0, \
"Error: No job groups should have been created."
self.testSubscriptionB.completeFiles([self.testFileE, self.testFileF])
self.testSubscriptionC.completeFiles([self.testFileE, self.testFileF])
self.testSubscriptionD.completeFiles([self.testFileE, self.testFileF])
result = deleteFactoryB(files_per_job = 10)
assert len(result) == 0, \
"Error: No jobs should have been created."
self.testFilesetB.markOpen(False)
result = deleteFactoryB(files_per_job = 10)
assert len(result) == 1, \
"Error: One jobgroup should have been returned."
assert len(result[0].jobs) == 1, \
"Error: There should only be one job in the jobgroup."
assert len(result[0].jobs[0]["input_files"]) == 2, \
"Error: Job should only have one input file."
lfns = [result[0].jobs[0]["input_files"][0]["lfn"], result[0].jobs[0]["input_files"][1]["lfn"]]
assert "testFileE" in lfns, \
"Error: TestFileE missing from job input."
assert "testFileF" in lfns, \
"Error: TestFileF missing from job input."
self.assertEqual(len(self.deleteSubscriptionB.availableFiles()), 0,
"Error: There should be no available files.")
completeFiles = self.deleteSubscriptionB.filesOfStatus("Completed")
self.assertEqual(len(completeFiles), 1,
"Error: There should only be one complete file.")
self.assertEqual(list(completeFiles)[0]["lfn"], "testFileD",
"Error: Test file D should be complete.")
return
def testMultipleLocations(self):
"""
_testMultipleLocations_
Verify that the sibling processing based algorithm doesn't create jobs
that run over files at multiple sites.
"""
testFile1 = File("testFile1", size = 1000, events = 100,
locations = set(["somese2.cern.ch"]))
testFile1.create()
testFile2 = File("testFile2", size = 1000, events = 100,
locations = set(["somese2.cern.ch"]))
testFile2.create()
testFile3 = File("testFile3", size = 1000, events = 100,
locations = set(["somese2.cern.ch"]))
testFile3.create()
self.testFilesetA.addFile(testFile1)
self.testFilesetA.addFile(testFile2)
self.testFilesetA.addFile(testFile3)
self.testFilesetA.commit()
self.testFilesetA.markOpen(False)
self.testSubscriptionA.completeFiles([testFile1, testFile2, testFile3])
self.testSubscriptionA.completeFiles([self.testFileA, self.testFileB, self.testFileC])
splitter = SplitterFactory()
deleteFactoryA = splitter(package = "WMCore.WMBS",
subscription = self.deleteSubscriptionA)
result = deleteFactoryA(files_per_job = 50)
assert len(result) == 2, \
"Error: Wrong number of jobgroups returned."
goldenFilesA = ["testFileA", "testFileB", "testFileC"]
goldenFilesB = ["testFile1", "testFile2", "testFile3"]
locations = {"testFileA": "somese.cern.ch", "testFileB": "somese.cern.ch",
"testFileC": "somese.cern.ch", "testFile1": "somese2.cern.ch",
"testFile2": "somese2.cern.ch", "testFile3": "somese2.cern.ch"}
for jobGroup in result:
assert len(jobGroup.jobs) == 1, \
"Error: Wrong number of jobs in jobgroup."
assert len(jobGroup.jobs[0]["input_files"]) == 3, \
"Error: Wrong number of input files in job."
jobSite = list(jobGroup.jobs[0]["input_files"][0]["locations"])[0]
if jobSite == "somese.cern.ch":
goldenFiles = goldenFilesA
else:
goldenFiles = goldenFilesB
for jobFile in jobGroup.jobs[0]["input_files"]:
assert list(jobFile["locations"])[0] == locations[jobFile["lfn"]], \
"Error: Wrong site for file."
goldenFiles.remove(jobFile["lfn"])
assert len(goldenFiles) == 0, \
"Error: Files are missing."
return
def testLargeNumberOfFiles(self):
"""
_testLargeNumberOfFiles_
Setup a subscription with 500 files and verify that the splitting algo
works correctly.
"""
testWorkflowA = Workflow(spec = "specA.xml", owner = "Steve",
name = "wfA", task = "Test")
testWorkflowA.create()
testWorkflowB = Workflow(spec = "specB.xml", owner = "Steve",
name = "wfB", task = "Test")
testWorkflowB.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
allFiles = []
for i in range(500):
testFile = File(str(i), size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
testFile.create()
allFiles.append(testFile)
testFileset.addFile(testFile)
testFileset.commit()
testSubscriptionA = Subscription(fileset = testFileset,
workflow = testWorkflowA,
split_algo = "FileBased",
type = "Processing")
testSubscriptionA.create()
testSubscriptionB = Subscription(fileset = testFileset,
workflow = testWorkflowB,
split_algo = "SiblingProcessingBased",
type = "Processing")
testSubscriptionB.create()
testSubscriptionA.completeFiles(allFiles)
splitter = SplitterFactory()
deleteFactoryA = splitter(package = "WMCore.WMBS",
subscription = testSubscriptionB)
result = deleteFactoryA(files_per_job = 50)
self.assertEqual(len(result), 1,
"Error: Wrong number of job groups returned.")
self.assertEqual(len(result[0].jobs), 10,
"Error: Wrong number of jobs returned.")
return
def testFilesWithoutOtherSubscriptions(self):
"""
_testFilesWithoutOtherSubscriptions_
Test the case where files only in the delete subscription
can happen if cleanup of the other subscriptions is fast
"""
testWorkflowA = Workflow(spec = "specA.xml", owner = "Steve",
name = "wfA", task = "Test")
testWorkflowA.create()
testFileset = Fileset(name = "TestFileset")
testFileset.create()
allFiles = []
for i in range(500):
testFile = File(str(i), size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
testFile.create()
allFiles.append(testFile)
testFileset.addFile(testFile)
testFileset.commit()
testSubscriptionA = Subscription(fileset = testFileset,
workflow = testWorkflowA,
split_algo = "SiblingProcessingBased",
type = "Processing")
testSubscriptionA.create()
splitter = SplitterFactory()
deleteFactoryA = splitter(package = "WMCore.WMBS",
subscription = testSubscriptionA)
result = deleteFactoryA(files_per_job = 50)
self.assertEqual(len(result), 1,
"Error: Wrong number of job groups returned.")
self.assertEqual(len(result[0].jobs), 10,
"Error: Wrong number of jobs returned.")
return
def testPersist(self):
"""
_testPersist_
This is the test class for function Propagate from module ChangeState
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
for i in range(4):
newFile = File(lfn="File%s" % i, locations=set(["T2_CH_CERN"]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased")
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 4, \
"Error: Splitting should have created four jobs."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
testJobB = jobGroup.jobs[1]
testJobB["user"] = "******"
testJobB["group"] = "DMWM"
testJobB["taskType"] = "Processing"
testJobC = jobGroup.jobs[2]
testJobC["user"] = "******"
testJobC["group"] = "DMWM"
testJobC["taskType"] = "Processing"
testJobD = jobGroup.jobs[3]
testJobD["user"] = "******"
testJobD["group"] = "DMWM"
testJobD["taskType"] = "Processing"
change.persist([testJobA, testJobB], "created", "new")
change.persist([testJobC, testJobD], "new", "none")
stateDAO = self.daoFactory(classname="Jobs.GetState")
jobAState = stateDAO.execute(id=testJobA["id"])
jobBState = stateDAO.execute(id=testJobB["id"])
jobCState = stateDAO.execute(id=testJobC["id"])
jobDState = stateDAO.execute(id=testJobD["id"])
assert jobAState == "created" and jobBState == "created" and \
jobCState == "new" and jobDState == "new", \
"Error: Jobs didn't change state correctly."
return
def setUp(self):
"""
_setUp_
Setup the database connections and schema.
"""
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("site1", seName = "somese.cern.ch")
locationAction.execute("site2", seName = "somese2.cern.ch")
self.testFilesetA = Fileset(name = "FilesetA")
self.testFilesetA.create()
self.testFilesetB = Fileset(name = "FilesetB")
self.testFilesetB.create()
self.testFileA = File("testFileA", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileA.create()
self.testFileB = File("testFileB", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileB.create()
self.testFileC = File("testFileC", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileC.create()
self.testFilesetA.addFile(self.testFileA)
self.testFilesetA.addFile(self.testFileB)
self.testFilesetA.addFile(self.testFileC)
self.testFilesetA.commit()
self.testFileD = File("testFileD", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileD.create()
self.testFileE = File("testFileE", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileE.create()
self.testFileF = File("testFileF", size = 1000, events = 100,
locations = set(["somese.cern.ch"]))
self.testFileF.create()
self.testFilesetB.addFile(self.testFileD)
self.testFilesetB.addFile(self.testFileE)
self.testFilesetB.addFile(self.testFileF)
self.testFilesetB.commit()
testWorkflowA = Workflow(spec = "specA.xml", owner = "Steve",
name = "wfA", task = "Test")
testWorkflowA.create()
testWorkflowB = Workflow(spec = "specB.xml", owner = "Steve",
name = "wfB", task = "Test")
testWorkflowB.create()
testWorkflowC = Workflow(spec = "specC.xml", owner = "Steve",
name = "wfC", task = "Test")
testWorkflowC.create()
testWorkflowD = Workflow(spec = "specD.xml", owner = "Steve",
name = "wfD", task = "Test")
testWorkflowD.create()
self.testSubscriptionA = Subscription(fileset = self.testFilesetA,
workflow = testWorkflowA,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionA.create()
self.testSubscriptionB = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowB,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionB.create()
self.testSubscriptionC = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowC,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionC.create()
self.testSubscriptionD = Subscription(fileset = self.testFilesetB,
workflow = testWorkflowD,
split_algo = "FileBased",
type = "Processing")
self.testSubscriptionD.create()
deleteWorkflow = Workflow(spec = "specE.xml", owner = "Steve",
name = "wfE", task = "Test")
deleteWorkflow.create()
self.deleteSubscriptionA = Subscription(fileset = self.testFilesetA,
workflow = deleteWorkflow,
split_algo = "SiblingProcessingBased",
type = "Cleanup")
self.deleteSubscriptionA.create()
self.deleteSubscriptionB = Subscription(fileset = self.testFilesetB,
workflow = deleteWorkflow,
split_algo = "SiblingProcessingBased",
type = "Cleanup")
self.deleteSubscriptionB.create()
return
def testUpdateLocation(self):
"""
_testUpdateLocation_
Check that we can update the location of a job through
the state machine.
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
locationAction.execute("site2", pnn="T1_US_FNAL_Disk")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased")
testSubscription.create()
testFileA = File(lfn="SomeLFNA",
events=1024,
size=2048,
locations=set(["T2_CH_CERN", "T1_US_FNAL_Disk"]))
testFileB = File(lfn="SomeLFNB",
events=1025,
size=2049,
locations=set(["T2_CH_CERN", "T1_US_FNAL_Disk"]))
testFileA.create()
testFileB.create()
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.commit()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 2, \
"Error: Splitting should have created two jobs."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Merge"
testJobA["site_cms_name"] = "site1"
testJobB = jobGroup.jobs[1]
testJobB["user"] = "******"
testJobB["group"] = "DMWM"
testJobB["taskType"] = "Processing"
testJobB["site_cms_name"] = "site2"
change.propagate([testJobA, testJobB], "new", "none")
change.propagate([testJobA, testJobB], "created", "new")
change.propagate([testJobA, testJobB], "executing", "created")
testJobADoc = change.jobsdatabase.document(testJobA["couch_record"])
maxKey = max(testJobADoc["states"].keys())
transition = testJobADoc["states"][maxKey]
self.assertEqual(transition["location"], "site1")
testJobBDoc = change.jobsdatabase.document(testJobB["couch_record"])
maxKey = max(testJobBDoc["states"].keys())
transition = testJobBDoc["states"][maxKey]
self.assertEqual(transition["location"], "site2")
jobs = [{'jobid': 1, 'location': 'site2'}]
change.recordLocationChange(jobs)
testJobADoc = change.jobsdatabase.document(testJobA["couch_record"])
maxKey = max(testJobADoc["states"].keys())
transition = testJobADoc["states"][maxKey]
self.assertEqual(transition["location"], "site2")
listJobsDAO = self.daoFactory(classname="Jobs.GetLocation")
jobid = [{'jobid': 1}, {'jobid': 2}]
jobsLocation = listJobsDAO.execute(jobid)
for job in jobsLocation:
self.assertEqual(job['site_name'], 'site2')
return
def testRetryCount(self):
"""
_testRetryCount_
Verify that the retry count is incremented when we move out of the
submitcooloff or jobcooloff state.
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
for i in range(4):
newFile = File(lfn="File%s" % i, locations=set(["T2_CH_CERN"]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased")
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 4, \
"Error: Splitting should have created four jobs."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
testJobB = jobGroup.jobs[1]
testJobB["user"] = "******"
testJobB["group"] = "DMWM"
testJobB["taskType"] = "Processing"
testJobC = jobGroup.jobs[2]
testJobC["user"] = "******"
testJobC["group"] = "DMWM"
testJobC["taskType"] = "Processing"
testJobD = jobGroup.jobs[3]
testJobD["user"] = "******"
testJobD["group"] = "DMWM"
testJobD["taskType"] = "Processing"
change.persist([testJobA], "created", "submitcooloff")
change.persist([testJobB], "created", "jobcooloff")
change.persist([testJobC, testJobD], "new", "none")
testJobA.load()
testJobB.load()
testJobC.load()
testJobD.load()
assert testJobA["retry_count"] == 1, \
"Error: Retry count is wrong."
assert testJobB["retry_count"] == 1, \
"Error: Retry count is wrong."
assert testJobC["retry_count"] == 0, \
"Error: Retry count is wrong."
assert testJobD["retry_count"] == 0, \
"Error: Retry count is wrong."
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
class FixedDelayTest(unittest.TestCase):
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", pnn = "T2_CH_CERN")
self.multipleFileFileset = Fileset(name = "TestFileset1")
self.multipleFileFileset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(i, *[45+i]))
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(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleFileFileset.addFile(newFile)
self.singleFileFileset.commit()
self.multipleFileLumiset = Fileset(name = "TestFileset3")
self.multipleFileLumiset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45+i/3]))
newFile.create()
self.multipleFileLumiset.addFile(newFile)
self.multipleFileLumiset.commit()
self.singleLumiFileset = Fileset(name = "TestFileset4")
self.singleLumiFileset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleLumiFileset.addFile(newFile)
self.singleLumiFileset.commit()
testWorkflow = Workflow(spec = "spec.xml", owner = "mnorman", name = "wf001", task="Test")
testWorkflow.create()
self.multipleFileSubscription = Subscription(fileset = self.multipleFileFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.singleFileSubscription = Subscription(fileset = self.singleFileFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.multipleLumiSubscription = Subscription(fileset = self.multipleFileLumiset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.singleLumiSubscription = Subscription(fileset = self.singleLumiFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription.create()
self.multipleLumiSubscription.create()
self.singleLumiSubscription.create()
return
def tearDown(self):
"""
_tearDown_
Nothing to do...
"""
self.testInit.clearDatabase()
return
def testNone(self):
"""
_testNone_
Since the subscriptions are open, we shouldn't get any jobs back
"""
splitter = SplitterFactory()
jobFactory = splitter(self.singleFileSubscription)
jobGroups = jobFactory(trigger_time = int(time.time())*2)
self.assertEqual(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time = int(time.time())*2)
self.assertEqual(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time = int(time.time())*2)
self.assertEqual(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time = int(time.time())*2)
self.assertEqual(jobGroups, [], "Should have returned a null set")
return
def testClosed(self):
"""
_testClosed_
Since the subscriptions are closed and none of the files have been
acquired, all of the files should show up
"""
splitter = SplitterFactory()
self.singleFileSubscription.getFileset().markOpen(False)
jobFactory = splitter(self.singleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
assert len(jobGroups) == 1, \
"ERROR: JobFactory didn't return one JobGroup."
assert len(jobGroups[0].jobs) == 1, \
"ERROR: JobFactory didn't create a single job."
job = jobGroups[0].jobs.pop()
assert job.getFiles(type = "lfn") == ["/some/file/name"], \
"ERROR: Job contains unknown files."
self.multipleFileSubscription.getFileset().markOpen(False)
jobFactory = splitter(self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs),1)
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 10)
self.multipleLumiSubscription.getFileset().markOpen(False)
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs),1)
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 10)
#self.assertEqual(jobGroups, [], "Should have returned a null set")
self.singleLumiSubscription.getFileset().markOpen(False)
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
assert len(jobGroups) == 1, \
"ERROR: JobFactory didn't return one JobGroup."
assert len(jobGroups[0].jobs) == 1, \
"ERROR: JobFactory didn't create a single job."
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 10)
def testAllAcquired(self):
"""
_testAllAcquired_
should all return no job groups
"""
splitter = SplitterFactory()
self.singleFileSubscription.acquireFiles(
self.singleFileSubscription.availableFiles())
jobFactory = splitter(self.singleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(jobGroups, [], "Should have returned a null set")
self.multipleFileSubscription.acquireFiles(
self.multipleFileSubscription.availableFiles())
jobFactory = splitter(self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(jobGroups, [], "Should have returned a null set")
self.multipleLumiSubscription.acquireFiles(
self.multipleLumiSubscription.availableFiles())
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(jobGroups, [], "Should have returned a null set")
self.singleLumiSubscription.acquireFiles(
self.singleLumiSubscription.availableFiles())
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(jobGroups, [], "Should have returned a null set")
def testClosedSomeAcquired(self):
"""
_testClosedSomeAcquired_
since the subscriptions are closed and none of the files ahve been
acquired, all of the files should show up
"""
splitter = SplitterFactory()
self.multipleFileSubscription.getFileset().markOpen(False)
self.singleFileSubscription.acquireFiles(
[self.singleFileSubscription.availableFiles().pop()])
jobFactory = splitter(self.singleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(jobGroups, [], "Should have returned a null set")
self.multipleFileSubscription.getFileset().markOpen(False)
self.multipleFileSubscription.acquireFiles(
[self.multipleFileSubscription.availableFiles().pop()])
jobFactory = splitter(package = "WMCore.WMBS", subscription =self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(len(jobGroups),1, "Should have gotten one jobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 9, \
"JobFactory should have provides us with 9 files")
self.multipleLumiSubscription.getFileset().markOpen(False)
self.multipleLumiSubscription.acquireFiles(
[self.multipleLumiSubscription.availableFiles().pop()])
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(len(jobGroups),1, "Should have gotten one jobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 9, \
"JobFactory should have provides us with 9 files")
self.singleLumiSubscription.getFileset().markOpen(False)
self.singleLumiSubscription.acquireFiles(
[self.singleLumiSubscription.availableFiles().pop()])
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time = 1)
self.assertEqual(len(jobGroups),1, "Should have gotten one jobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEqual(len(myfiles), 9, \
"JobFactory should have provides us with 9 files")
self.assertEqual(len(myfiles), 9)
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())
subscription.create()
### FIXME: I'm pretty sure we can improve how we handle this site white/black list
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 %s created for %s",
subscription["id"], self.wmSpec.name())
else:
logging.info("Child subscription %s created for %s",
subscription["id"], self.wmSpec.name())
outputModules = task.getOutputModulesForTask()
ignoredOutputModules = task.getIgnoredOutputModulesForTask()
for outputModule in outputModules:
for outputModuleName in outputModule.listSections_():
if outputModuleName in ignoredOutputModules:
msg = "%s has %s as IgnoredOutputModule, skipping fileset creation."
logging.info(msg, task.getPathName(), 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
def testMonteCarloFromGEN(self):
"""
_testMonteCarloFromGEN_
Create a MonteCarloFromGEN workflow and verify it installs into WMBS
correctly.
"""
arguments = MonteCarloFromGENWorkloadFactory.getTestArguments()
arguments["ConfigCacheID"] = self.injectConfig()
arguments["CouchDBName"] = "mclhe_t"
arguments["PrimaryDataset"] = "WaitThisIsNotMinimumBias"
factory = MonteCarloFromGENWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction(
"TestWorkload", arguments)
outputDatasets = testWorkload.listOutputDatasets()
self.assertEqual(len(outputDatasets), 2)
self.assertTrue("/WaitThisIsNotMinimumBias/FAKE-FilterRECO-v1/RECO" in
outputDatasets)
self.assertTrue(
"/WaitThisIsNotMinimumBias/FAKE-FilterALCARECO-v1/ALCARECO" in
outputDatasets)
testWMBSHelper = WMBSHelper(testWorkload,
"MonteCarloFromGEN",
"SomeBlock",
cachepath=self.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
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, 'production')
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"], "EventAwareLumiBased",
"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
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()
parentFile = File('/parent/lfn/',
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
parentFile.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
newFile.addRun(Run(i, *[45]))
newFile.create()
newFile.addParent(lfn=parentFile['lfn'])
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,
locations=set(["somese.cern.ch"]))
newFile.create()
self.multipleSiteFileset.addFile(newFile)
for i in range(5):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["otherse.cern.ch",
"somese.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="FileBased",
type="Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription = Subscription(
fileset=self.singleFileFileset,
workflow=testWorkflow,
split_algo="FileBased",
type="Processing")
self.singleFileSubscription.create()
self.multipleSiteSubscription = Subscription(
fileset=self.multipleSiteFileset,
workflow=testWorkflow,
split_algo="FileBased",
type="Processing")
self.multipleSiteSubscription.create()
self.performanceParams = {
'timePerEvent': 12,
'memoryRequirement': 2300,
'sizePerEvent': 400
}
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
class FileBasedTest(unittest.TestCase):
"""
_FileBasedTest_
Test file based job splitting.
"""
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()
parentFile = File('/parent/lfn/',
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
parentFile.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
newFile.addRun(Run(i, *[45]))
newFile.create()
newFile.addParent(lfn=parentFile['lfn'])
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,
locations=set(["somese.cern.ch"]))
newFile.create()
self.multipleSiteFileset.addFile(newFile)
for i in range(5):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["otherse.cern.ch",
"somese.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="FileBased",
type="Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription = Subscription(
fileset=self.singleFileFileset,
workflow=testWorkflow,
split_algo="FileBased",
type="Processing")
self.singleFileSubscription.create()
self.multipleSiteSubscription = Subscription(
fileset=self.multipleSiteFileset,
workflow=testWorkflow,
split_algo="FileBased",
type="Processing")
self.multipleSiteSubscription.create()
self.performanceParams = {
'timePerEvent': 12,
'memoryRequirement': 2300,
'sizePerEvent': 400
}
return
def tearDown(self):
"""
_tearDown_
Clear out WMBS.
"""
self.testInit.clearDatabase()
return
def createLargeFileBlock(self):
"""
_createLargeFileBlock_
Creates a large group of files for testing
"""
testFileset = Fileset(name="TestFilesetX")
testFileset.create()
for i in range(5000):
newFile = File(makeUUID(),
size=1000,
events=100,
locations=set(["somese.cern.ch"]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testWorkflow = Workflow(spec="spec.xml",
owner="mnorman",
name="wf003",
task="Test")
testWorkflow.create()
largeSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased",
type="Processing")
largeSubscription.create()
return largeSubscription
def testExactFiles(self):
"""
_testExactFiles_
Test file based job splitting when the number of files per job is
exactly the same as the number of files in the input fileset.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleFileSubscription)
jobGroups = jobFactory(files_per_job=1,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 1)
job = jobGroups[0].jobs.pop()
self.assertEqual(job.getFiles(type="lfn"), ["/some/file/name"])
self.assertEqual(job["estimatedMemoryUsage"], 2300)
self.assertEqual(job["estimatedDiskUsage"], 400 * 100)
self.assertEqual(job["estimatedJobTime"], 12 * 100)
return
def testMoreFiles(self):
"""
_testMoreFiles_
Test file based job splitting when the number of files per job is
greater than the number of files in the input fileset.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleFileSubscription)
jobGroups = jobFactory(files_per_job=10,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 1)
job = jobGroups[0].jobs.pop()
self.assertEqual(job.getFiles(type="lfn"), ["/some/file/name"])
self.assertEqual(job["estimatedMemoryUsage"], 2300)
self.assertEqual(job["estimatedDiskUsage"], 400 * 100)
self.assertEqual(job["estimatedJobTime"], 12 * 100)
return
def test2FileSplit(self):
"""
_test2FileSplit_
Test file based job splitting when the number of files per job is
2, this should result in five jobs.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=2,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 5)
fileList = []
for job in jobGroups[0].jobs:
self.assertEqual(len(job.getFiles()), 2)
for file in job.getFiles(type="lfn"):
fileList.append(file)
self.assertEqual(job["estimatedMemoryUsage"], 2300)
self.assertEqual(job["estimatedDiskUsage"], 400 * 100 * 2)
self.assertEqual(job["estimatedJobTime"], 12 * 100 * 2)
self.assertEqual(len(fileList), 10)
return
def test3FileSplit(self):
"""
_test3FileSplit_
Test file based job splitting when the number of files per job is
3, this should result in four jobs.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=3,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 4)
fileList = []
for job in jobGroups[0].jobs:
assert len(job.getFiles()) in [
3, 1
], "ERROR: Job contains incorrect number of files"
for file in job.getFiles(type="lfn"):
fileList.append(file)
if len(job.getFiles()) == 3:
self.assertEqual(job["estimatedMemoryUsage"], 2300)
self.assertEqual(job["estimatedDiskUsage"], 400 * 100 * 3)
self.assertEqual(job["estimatedJobTime"], 12 * 100 * 3)
elif len(job.getFiles()) == 1:
self.assertEqual(job["estimatedMemoryUsage"], 2300)
self.assertEqual(job["estimatedDiskUsage"], 400 * 100)
self.assertEqual(job["estimatedJobTime"], 12 * 100)
else:
self.fail("Unexpected splitting reached")
self.assertEqual(len(fileList), 10)
return
def testLocationSplit(self):
"""
_testLocationSplit_
This should test whether or not the FileBased algorithm understands that files at seperate sites
cannot be in the same jobGroup (this is the current standard).
"""
myThread = threading.currentThread()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleSiteSubscription)
jobGroups = jobFactory(files_per_job=10,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 2)
self.assertEqual(len(jobGroups[0].jobs), 1)
self.assertEqual(jobGroups[0].jobs[0]["estimatedMemoryUsage"], 2300)
self.assertEqual(jobGroups[0].jobs[0]["estimatedDiskUsage"],
100 * 400 * 5)
self.assertEqual(jobGroups[0].jobs[0]["estimatedJobTime"],
100 * 12 * 5)
self.assertEqual(len(jobGroups[1].jobs[0].getFiles()), 5)
self.assertEqual(jobGroups[1].jobs[0]["estimatedMemoryUsage"], 2300)
self.assertEqual(jobGroups[1].jobs[0]["estimatedDiskUsage"],
100 * 400 * 5)
self.assertEqual(jobGroups[1].jobs[0]["estimatedJobTime"],
100 * 12 * 5)
return
def testLimit(self):
"""
_testLimit_
Test what happens when you limit the number of files.
This should run each separate file in a separate loop,
creating one jobGroups with one job with one file
(The limit argument tells it what to do)
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=10,
limit_file_loading=True,
file_load_limit=1,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 1)
return
def testRespectRunBoundaries(self):
"""
_testRespectRunBoundaries_
Test whether or not this thing will respect run boundaries
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=10,
respect_run_boundaries=True,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 10)
return
def test_getParents(self):
"""
_getParents_
Check that we can do the same as the TwoFileBased
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=2,
include_parents=True,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 5)
fileList = []
for job in jobGroups[0].jobs:
self.assertEqual(len(job.getFiles()), 2)
for file in job.getFiles(type="lfn"):
fileList.append(file)
self.assertEqual(len(fileList), 10)
for j in jobGroups[0].jobs:
for f in j['input_files']:
self.assertEqual(len(f['parents']), 1)
self.assertEqual(list(f['parents'])[0]['lfn'], '/parent/lfn/')
return
def testZ_randomCrapForGenerators(self):
"""
Either this works, and all other tests are obsolete, or it doesn't and they aren't.
Either way, don't screw around with this.
"""
def runCode(self, jobFactory):
func = self.crazyAssFunction(jobFactory=jobFactory,
file_load_limit=500)
startTime = time.time()
goFlag = True
while goFlag:
try:
res = func.next()
self.jobGroups.extend(res)
except StopIteration:
goFlag = False
stopTime = time.time()
return jobGroups
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleSiteSubscription)
jobFactory.open()
jobGroups = []
a = self.crazyAssFunction(jobFactory=jobFactory, file_load_limit=2)
for x in range(7):
try:
res = a.next()
jobGroups.extend(res)
except StopIteration:
continue
jobFactory.close()
self.assertEqual(len(jobGroups), 6)
for group in jobGroups:
self.assertTrue(len(group.jobs) in [1, 2])
for job in group.jobs:
self.assertTrue(len(job['input_files']) in (1, 2))
for file in job['input_files']:
self.assertTrue(file['locations'] in [
set(['somese.cern.ch']),
set(['otherse.cern.ch', 'somese.cern.ch'])
])
self.jobGroups = []
subscript = self.createLargeFileBlock()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=subscript)
jobFactory.open()
runCode(self, jobFactory)
#cProfile.runctx("runCode(self, jobFactory)", globals(), locals(), "coroutine.stats")
jobGroups = self.jobGroups
self.assertEqual(len(jobGroups), 10)
for group in jobGroups:
self.assertEqual(len(group.jobs), 500)
self.assertTrue(group.exists() > 0)
jobFactory.close()
return
def crazyAssFunction(self, jobFactory, file_load_limit=1):
groups = ['test']
while groups != []:
groups = jobFactory(files_per_job=1,
file_load_limit=file_load_limit,
performance=self.performanceParams)
yield groups
def createTestJobGroup(self,
name="TestWorkthrough",
specLocation="spec.xml",
error=False,
task="/TestWorkload/ReReco",
nJobs=10):
"""
_createTestJobGroup_
Generate a test WMBS JobGroup with real FWJRs
"""
myThread = threading.currentThread()
testWorkflow = Workflow(spec=specLocation,
owner="Simon",
name=name,
task=task)
testWorkflow.create()
testWMBSFileset = Fileset(name=name)
testWMBSFileset.create()
testFileA = File(lfn=makeUUID(), size=1024, events=10)
testFileA.addRun(Run(10, *[12312]))
testFileA.setLocation('malpaquet')
testFileB = File(lfn=makeUUID(), size=1024, events=10)
testFileB.addRun(Run(10, *[12312]))
testFileB.setLocation('malpaquet')
testFileA.create()
testFileB.create()
testWMBSFileset.addFile(testFileA)
testWMBSFileset.addFile(testFileB)
testWMBSFileset.commit()
testWMBSFileset.markOpen(0)
testSubscription = Subscription(fileset=testWMBSFileset,
workflow=testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
for i in range(0, 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()
report = Report()
if error:
path = os.path.join(WMCore.WMBase.getTestBase(),
"WMComponent_t/JobAccountant_t/fwjrs",
"badBackfillJobReport.pkl")
else:
path = os.path.join(WMCore.WMBase.getTestBase(),
"WMComponent_t/JobAccountant_t/fwjrs",
"PerformanceReport2.pkl")
report.load(filename=path)
self.changeState.propagate(testJobGroup.jobs, 'created', 'new')
self.changeState.propagate(testJobGroup.jobs, 'executing', 'created')
self.changeState.propagate(testJobGroup.jobs, 'complete', 'executing')
for job in testJobGroup.jobs:
job['fwjr'] = report
self.changeState.propagate(testJobGroup.jobs, 'jobfailed', 'complete')
self.changeState.propagate(testJobGroup.jobs, 'retrydone', 'jobfailed')
self.changeState.propagate(testJobGroup.jobs, 'exhausted', 'retrydone')
self.changeState.propagate(testJobGroup.jobs, 'cleanout', 'exhausted')
testSubscription.completeFiles([testFileA, testFileB])
return testJobGroup
def databaseWork(self):
"""
Queries DB for all watched filesets, if a filesets matches become
available, create the subscriptions
"""
# Get all watched workflows
availableWorkflows = self.getUnsubscribedWorkflows.execute()
logging.debug("Found %s unsubscribed managed workflows" \
% len(availableWorkflows))
# Get all filesets to check if they match a wrokflow
availableFilesets = self.getAllFilesets.execute()
logging.debug("Found %s filesets" % len(availableFilesets))
# Loop on unsubscribed workflows to match filesets
for managedWorkflow in availableWorkflows:
# Workflow object cache to pass into Subscription constructor
wfObj = None
for fileset in availableFilesets:
# Fileset object cache
fsObj = None
# Load the location information
#whitelist = Set()
#blacklist = Set()
# Location is only caf
#locations = self.queries.getLocations(managedWorkflow['id'])
#for location in locations:
# if bool(int(location['valid'])) == True:
# whitelist.add(location['site_name'])
# else:
# blacklist.add(location['site_name'])
# Attempt to match workflows to filesets
if re.match(managedWorkflow['fileset_match'], fileset['name']):
# Log in debug
msg = "Creating subscription for %s to workflow id %s"
msg %= (fileset['name'], managedWorkflow['workflow'])
logging.debug(msg)
# Match found - Load the fileset if not already loaded
if not fsObj:
fsObj = Fileset(id=fileset['id'])
fsObj.load()
# Load the workflow if not already loaded
if not wfObj:
wfObj = Workflow(id=managedWorkflow['workflow'])
wfObj.load()
# Create the subscription
newSub = Subscription(fileset = fsObj, \
workflow = wfObj, \
#whitelist = whitelist, \
#blacklist = blacklist, \
split_algo = managedWorkflow['split_algo'],
type = managedWorkflow['type'])
newSub.create()
managedWorkflows = self.getManagedWorkflows.execute()
logging.debug("Found %s managed workflows" \
% len(managedWorkflows))
unsubscribedFilesets = self.getUnsubscribedFilesets.execute()
logging.debug("Found %s unsubscribed filesets" % \
len(unsubscribedFilesets))
# Loop on unsubscribed filesets to match workflows
for unsubscribedFileset in unsubscribedFilesets:
# Workflow object cache to pass into Subscription constructor
# FIXME
wfObj = None
for managedWork in managedWorkflows:
logging.debug("The workflow %s" % managedWork['workflow'])
# Fileset object cache
wfObj = None
fsObj = None
# Load the location information
#whitelist = Set()
#blacklist = Set()
# Location is only caf
#locations = self.queries.getLocations(managedWorkflow['id'])
#for location in locations:
# if bool(int(location['valid'])) == True:
# whitelist.add(location['site_name'])
# else:
# blacklist.add(location['site_name'])
# Attempt to match workflows to filesets
if re.match(managedWork['fileset_match'], \
unsubscribedFileset['name']):
# Log in debug
msg = "Creating subscription for %s to workflow id %s"
msg %= (unsubscribedFileset['name'], \
managedWork['workflow'])
logging.debug(msg)
# Match found - Load the fileset if not already loaded
if not fsObj:
fsObj = Fileset(id=unsubscribedFileset['id'])
fsObj.load()
# Load the workflow if not already loaded
if not wfObj:
wfObj = Workflow(id=managedWork['workflow'])
wfObj.load()
# Create the subscription
newSub = Subscription(fileset = fsObj, \
workflow = wfObj, \
#whitelist = whitelist, \
#blacklist = blacklist, \
split_algo = managedWork['split_algo'],
type = managedWork['type'])
newSub.create()
newSub.load()
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('T2_CH_CERN')
testFileA = File(lfn="/this/is/a/lfnA%s" % fileModifier, size=1024, events=10,
first_event=88)
testFileA.addRun(Run(10, *[12312, 12313]))
testFileA.setLocation('T2_CH_CERN')
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('T2_CH_CERN')
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['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 testTruncatedWFInsertion(self):
"""
_testTruncatedWFInsertion_
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
seName='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.createSubscription(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.createSubscription(
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.")
cleanupWorkflow = Workflow(name="ResubmitTestWorkload",
task="/ResubmitTestWorkload/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,
"/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")
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()
dummySubscription = Subscription(fileset=bogusFileset,
workflow=mergeWorkflow,
split_algo="ParentlessMergeBySize")
file1 = File(lfn="file1", size=1024, events=1024, first_event=0,
locations={"somese.cern.ch"})
file1.addRun(Run(1, *[45]))
file1.create()
file2 = File(lfn="file2", size=1024, events=1024, first_event=1024, locations={"somese.cern.ch"})
file2.addRun(Run(1, *[45]))
file2.create()
file3 = File(lfn="file3", size=1024, events=1024, first_event=2048, locations={"somese.cern.ch"})
file3.addRun(Run(1, *[45]))
file3.create()
file4 = File(lfn="file4", size=1024, events=1024, first_event=3072, locations={"somese.cern.ch"})
file4.addRun(Run(1, *[45]))
file4.create()
fileA = File(lfn="fileA", size=1024, events=1024, first_event=0, locations={"somese.cern.ch"})
fileA.addRun(Run(1, *[46]))
fileA.create()
fileB = File(lfn="fileB", size=1024, events=1024, first_event=1024, locations={"somese.cern.ch"})
fileB.addRun(Run(1, *[46]))
fileB.create()
fileC = File(lfn="fileC", size=1024, events=1024, first_event=2048, locations={"somese.cern.ch"})
fileC.addRun(Run(1, *[46]))
fileC.create()
fileI = File(lfn="fileI", size=1024, events=1024, first_event=0, locations={"somese.cern.ch"})
fileI.addRun(Run(2, *[46]))
fileI.create()
fileII = File(lfn="fileII", size=1024, events=1024, first_event=1024, locations={"somese.cern.ch"})
fileII.addRun(Run(2, *[46]))
fileII.create()
fileIII = File(lfn="fileIII", size=1024, events=1024, first_event=2048, locations={"somese.cern.ch"})
fileIII.addRun(Run(2, *[46]))
fileIII.create()
fileIV = File(lfn="fileIV", size=1024 * 1000000, events=1024, first_event=3072, locations={"somese.cern.ch"})
fileIV.addRun(Run(2, *[46]))
fileIV.create()
for fileObj in [file1, file2, file3, file4, fileA, fileB, fileC, fileI, fileII, fileIII, fileIV]:
mergeFileset.addFile(fileObj)
bogusFileset.addFile(fileObj)
mergeFileset.commit()
bogusFileset.commit()
return
class WMBSHelperTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
"""
self.testInit = TestInitCouchApp(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setupCouch("wmbshelper_t/jobs", "JobDump")
self.testInit.setupCouch("wmbshelper_t/fwjrs", "FWJRDump")
self.testInit.setupCouch("config_test", "GroupUser", "ConfigCache")
os.environ["COUCHDB"] = "wmbshelper_t"
self.testInit.setSchema(customModules=[
"WMCore.WMBS", "WMComponent.DBS3Buffer", "WMCore.BossAir",
"WMCore.ResourceControl"
],
useDefault=False)
self.workDir = self.testInit.generateWorkDir()
self.wmspec = self.createWMSpec()
self.topLevelTask = getFirstTask(self.wmspec)
self.inputDataset = self.topLevelTask.inputDataset()
self.dataset = self.topLevelTask.getInputDatasetPath()
self.dbs = MockDBSReader(self.inputDataset.dbsurl)
self.daoFactory = DAOFactory(package="WMCore.WMBS",
logger=threading.currentThread().logger,
dbinterface=threading.currentThread().dbi)
return
def tearDown(self):
"""
_tearDown_
Clear out the database.
"""
self.testInit.clearDatabase()
self.testInit.tearDownCouch()
self.testInit.delWorkDir()
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)
locationAction = daoFactory(classname="Locations.New")
changeStateAction = daoFactory(classname="Jobs.ChangeState")
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='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 verifyFileKillStatus(self):
"""
_verifyFileKillStatus_
Verify that all files were killed correctly. The status of files in
Cleanup and LogCollect subscriptions isn't modified. Status of
already completed and failed files is not modified. Also verify that
the bogus subscription is untouched.
"""
failedFiles = self.mainProcSub.filesOfStatus("Failed")
acquiredFiles = self.mainProcSub.filesOfStatus("Acquired")
completedFiles = self.mainProcSub.filesOfStatus("Completed")
availableFiles = self.mainProcSub.filesOfStatus("Available")
bogusAcquiredFiles = self.bogusSub.filesOfStatus("Acquired")
self.assertEqual(len(availableFiles), 0, \
"Error: There should be no available files.")
self.assertEqual(len(acquiredFiles), 0, \
"Error: There should be no acquired files.")
self.assertEqual(len(bogusAcquiredFiles), 1, \
"Error: There should be one acquired file.")
self.assertEqual(len(completedFiles), 3, \
"Error: There should be only one completed file.")
goldenLFNs = ["lfnA", "lfnB", "lfnC"]
for completedFile in completedFiles:
self.assertTrue(completedFile["lfn"] in goldenLFNs, \
"Error: Extra completed file.")
goldenLFNs.remove(completedFile["lfn"])
self.assertEqual(len(failedFiles), 0, \
"Error: There should be no failed files.")
self.assertEqual(len(goldenLFNs), 0, \
"Error: Missing LFN")
failedFiles = self.mainMergeSub.filesOfStatus("Failed")
acquiredFiles = self.mainMergeSub.filesOfStatus("Acquired")
completedFiles = self.mainMergeSub.filesOfStatus("Completed")
availableFiles = self.mainMergeSub.filesOfStatus("Available")
self.assertEqual(len(acquiredFiles), 0, \
"Error: Merge subscription should have 0 acq files.")
self.assertEqual(len(availableFiles), 0, \
"Error: Merge subscription should have 0 avail files.")
self.assertEqual(len(failedFiles), 1, \
"Error: Merge subscription should have 1 failed files.")
self.assertEqual(
list(failedFiles)[0]["lfn"], "ulfnB", "Error: Wrong failed file.")
self.assertEqual(len(completedFiles), 2, \
"Error: Merge subscription should have 2 compl files.")
goldenLFNs = ["ulfnA", "ulfnC"]
for completedFile in completedFiles:
self.assertTrue(completedFile["lfn"] in goldenLFNs, \
"Error: Extra complete file.")
goldenLFNs.remove(completedFile["lfn"])
self.assertEqual(len(goldenLFNs), 0, \
"Error: Missing LFN")
failedFiles = self.mainCleanupSub.filesOfStatus("Failed")
acquiredFiles = self.mainCleanupSub.filesOfStatus("Acquired")
completedFiles = self.mainCleanupSub.filesOfStatus("Completed")
availableFiles = self.mainCleanupSub.filesOfStatus("Available")
self.assertEqual(len(failedFiles), 0, \
"Error: Cleanup subscription should have 0 fai files.")
self.assertEqual(len(acquiredFiles), 1, \
"Error: There should be only one acquired file.")
self.assertEqual(list(acquiredFiles)[0]["lfn"], "ulfnA", \
"Error: Wrong acquired LFN.")
self.assertEqual(len(completedFiles), 1, \
"Error: There should be only one completed file.")
self.assertEqual(list(completedFiles)[0]["lfn"], "ulfnB", \
"Error: Wrong completed LFN.")
self.assertEqual(len(availableFiles), 1, \
"Error: There should be only one available file.")
self.assertEqual(list(availableFiles)[0]["lfn"], "ulfnC", \
"Error: Wrong completed LFN.")
return
def verifyJobKillStatus(self):
"""
_verifyJobKillStatus_
Verify that jobs are killed correctly. Jobs belonging to Cleanup and
LogCollect subscriptions are not killed. The status of jobs that have
already finished running is not changed.
"""
self.procJobA.load()
self.procJobB.load()
self.procJobC.load()
self.assertEqual(self.procJobA["state"], "killed", \
"Error: Proc job A should be killed.")
self.assertEqual(self.procJobB["state"], "killed", \
"Error: Proc job B should be killed.")
self.assertEqual(self.procJobC["state"], "complete", \
"Error: Proc job C should be complete.")
self.mergeJobA.load()
self.mergeJobB.load()
self.mergeJobC.load()
self.assertEqual(self.mergeJobA["state"], "exhausted", \
"Error: Merge job A should be exhausted.")
self.assertEqual(self.mergeJobB["state"], "cleanout", \
"Error: Merge job B should be cleanout.")
self.assertEqual(self.mergeJobC["state"], "killed", \
"Error: Merge job C should be killed.")
self.cleanupJobA.load()
self.cleanupJobB.load()
self.cleanupJobC.load()
self.assertEqual(self.cleanupJobA["state"], "new", \
"Error: Cleanup job A should be new.")
self.assertEqual(self.cleanupJobB["state"], "executing", \
"Error: Cleanup job B should be executing.")
self.assertEqual(self.cleanupJobC["state"], "complete", \
"Error: Cleanup job C should be complete.")
return
def createTestWMSpec(self):
"""
_createTestWMSpec_
Create a WMSpec that has a processing, merge, cleanup and skims tasks that
can be used by the subscription creation test.
"""
testWorkload = WMWorkloadHelper(WMWorkload("TestWorkload"))
testWorkload.setDashboardActivity("TestReReco")
testWorkload.setSpecUrl("/path/to/workload")
testWorkload.setOwnerDetails("sfoulkes", "DMWM", {'dn': 'MyDN'})
procTask = testWorkload.newTask("ProcessingTask")
procTask.setTaskType("Processing")
procTask.setSplittingAlgorithm("FileBased", files_per_job=1)
procTaskCMSSW = procTask.makeStep("cmsRun1")
procTaskCMSSW.setStepType("CMSSW")
procTaskCMSSWHelper = procTaskCMSSW.getTypeHelper()
procTask.setTaskType("Processing")
procTask.setSiteWhitelist(["site1"])
procTask.setSiteBlacklist(["site2"])
procTask.applyTemplates()
procTaskCMSSWHelper.addOutputModule("OutputA",
primaryDataset="bogusPrimary",
processedDataset="bogusProcessed",
dataTier="DataTierA",
lfnBase="bogusUnmerged",
mergedLFNBase="bogusMerged",
filterName=None)
mergeTask = procTask.addTask("MergeTask")
mergeTask.setInputReference(procTaskCMSSW, outputModule="OutputA")
mergeTask.setTaskType("Merge")
mergeTask.setSplittingAlgorithm("WMBSMergeBySize",
min_merge_size=1,
max_merge_size=2,
max_merge_events=3)
mergeTaskCMSSW = mergeTask.makeStep("cmsRun1")
mergeTaskCMSSW.setStepType("CMSSW")
mergeTaskCMSSWHelper = mergeTaskCMSSW.getTypeHelper()
mergeTask.setTaskType("Merge")
mergeTask.applyTemplates()
mergeTaskCMSSWHelper.addOutputModule("Merged",
primaryDataset="bogusPrimary",
processedDataset="bogusProcessed",
dataTier="DataTierA",
lfnBase="bogusUnmerged",
mergedLFNBase="bogusMerged",
filterName=None)
cleanupTask = procTask.addTask("CleanupTask")
cleanupTask.setInputReference(procTaskCMSSW, outputModule="OutputA")
cleanupTask.setTaskType("Merge")
cleanupTask.setSplittingAlgorithm("SiblingProcessingBase",
files_per_job=50)
cleanupTaskCMSSW = cleanupTask.makeStep("cmsRun1")
cleanupTaskCMSSW.setStepType("CMSSW")
cleanupTaskCMSSWHelper = cleanupTaskCMSSW.getTypeHelper()
cleanupTask.setTaskType("Cleanup")
cleanupTask.applyTemplates()
skimTask = mergeTask.addTask("SkimTask")
skimTask.setTaskType("Skim")
skimTask.setInputReference(mergeTaskCMSSW, outputModule="Merged")
skimTask.setSplittingAlgorithm("FileBased",
files_per_job=1,
include_parents=True)
skimTaskCMSSW = skimTask.makeStep("cmsRun1")
skimTaskCMSSW.setStepType("CMSSW")
skimTaskCMSSWHelper = skimTaskCMSSW.getTypeHelper()
skimTask.setTaskType("Skim")
skimTask.applyTemplates()
skimTaskCMSSWHelper.addOutputModule("SkimOutputA",
primaryDataset="bogusPrimary",
processedDataset="bogusProcessed",
dataTier="DataTierA",
lfnBase="bogusUnmerged",
mergedLFNBase="bogusMerged",
filterName=None)
skimTaskCMSSWHelper.addOutputModule("SkimOutputB",
primaryDataset="bogusPrimary",
processedDataset="bogusProcessed",
dataTier="DataTierA",
lfnBase="bogusUnmerged",
mergedLFNBase="bogusMerged",
filterName=None)
return testWorkload
def setupMCWMSpec(self):
"""Setup MC workflow"""
self.wmspec = self.createMCWMSpec()
self.topLevelTask = getFirstTask(self.wmspec)
self.inputDataset = self.topLevelTask.inputDataset()
self.dataset = self.topLevelTask.getInputDatasetPath()
self.dbs = None
self.siteDB = fakeSiteDB()
# add sites that would normally be added by operator via resource_control
locationDAO = self.daoFactory(classname="Locations.New")
self.ses = []
for site in ['T2_XX_SiteA', 'T2_XX_SiteB']:
locationDAO.execute(siteName=site,
seName=self.siteDB.cmsNametoSE(site))
self.ses.append(self.siteDB.cmsNametoSE(site))
def createWMSpec(self, name='ReRecoWorkload'):
wmspec = rerecoWorkload(name, rerecoArgs)
wmspec.setSpecUrl("/path/to/workload")
return wmspec
def createMCWMSpec(self, name='MonteCarloWorkload'):
wmspec = monteCarloWorkload(name, mcArgs)
wmspec.setSpecUrl("/path/to/workload")
getFirstTask(wmspec).addProduction(totalevents=10000)
return wmspec
def getDBS(self, wmspec):
topLevelTask = getFirstTask(wmspec)
inputDataset = topLevelTask.inputDataset()
dbs = MockDBSReader(inputDataset.dbsurl)
#dbsDict = {self.inputDataset.dbsurl : self.dbs}
return dbs
def createWMBSHelperWithTopTask(self,
wmspec,
block,
mask=None,
parentFlag=False,
detail=False):
topLevelTask = getFirstTask(wmspec)
wmbs = WMBSHelper(wmspec,
topLevelTask.name(),
block,
mask,
cachepath=self.workDir)
if block:
if parentFlag:
block = self.dbs.getFileBlockWithParents(block)[block]
else:
block = self.dbs.getFileBlock(block)[block]
sub, files = wmbs.createSubscriptionAndAddFiles(block=block)
if detail:
return wmbs, sub, files
else:
return wmbs
def testKillWorkflow(self):
"""
_testKillWorkflow_
Verify that workflow killing works correctly.
"""
configFile = EmulatorSetup.setupWMAgentConfig()
config = loadConfigurationFile(configFile)
baAPI = BossAirAPI(config=config)
# Create nine jobs
self.setupForKillTest(baAPI=baAPI)
self.assertEqual(len(baAPI._listRunJobs()), 9)
killWorkflow("Main", config, config)
self.verifyFileKillStatus()
self.verifyJobKillStatus()
self.assertEqual(len(baAPI._listRunJobs()), 8)
EmulatorSetup.deleteConfig(configFile)
return
def testCreateSubscription(self):
"""
_testCreateSubscription_
Verify that the subscription creation code works correctly.
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
seName='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.createSubscription(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 testTruncatedWFInsertion(self):
"""
_testTruncatedWFInsertion_
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
seName='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.createSubscription(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.createSubscription(
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.")
cleanupWorkflow = Workflow(name="ResubmitTestWorkload",
task="/ResubmitTestWorkload/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,
"/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 testProduction(self):
# """Production workflow"""
# pass
def testReReco(self):
"""ReReco workflow"""
# create workflow
block = self.dataset + "#1"
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
files = wmbs.validFiles(self.dbs.getFileBlock(block))
self.assertEqual(len(files), 1)
def testReRecoBlackRunRestriction(self):
"""ReReco workflow with Run restrictions"""
block = self.dataset + "#2"
#add run blacklist
self.topLevelTask.setInputRunBlacklist([1, 2, 3, 4])
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
files = wmbs.validFiles(self.dbs.getFileBlock(block)[block]['Files'])
self.assertEqual(len(files), 0)
def testReRecoWhiteRunRestriction(self):
block = self.dataset + "#2"
# Run Whitelist
self.topLevelTask.setInputRunWhitelist([2])
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
files = wmbs.validFiles(self.dbs.getFileBlock(block)[block]['Files'])
self.assertEqual(len(files), GlobalParams.numOfFilesPerBlock())
def testLumiMaskRestrictionsOK(self):
block = self.dataset + "#1"
self.wmspec.getTopLevelTask()[0].data.input.splitting.runs = ['1']
self.wmspec.getTopLevelTask()[0].data.input.splitting.lumis = ['1,1']
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
files = wmbs.validFiles(self.dbs.getFileBlock(block)[block]['Files'])
self.assertEqual(len(files), GlobalParams.numOfFilesPerBlock())
def testLumiMaskRestrictionsKO(self):
block = self.dataset + "#1"
self.wmspec.getTopLevelTask()[0].data.input.splitting.runs = [
'123454321'
]
self.wmspec.getTopLevelTask()[0].data.input.splitting.lumis = [
'123,123'
]
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
files = wmbs.validFiles(self.dbs.getFileBlock(block)[block]['Files'])
self.assertEqual(len(files), 0)
def testDuplicateFileInsert(self):
# using default wmspec
block = self.dataset + "#1"
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
wmbs.topLevelFileset.loadData()
numOfFiles = len(wmbs.topLevelFileset.files)
# check initially inserted files.
dbsFiles = self.dbs.getFileBlock(block)[block]['Files']
self.assertEqual(numOfFiles, len(dbsFiles))
firstFileset = wmbs.topLevelFileset
wmbsDao = wmbs.daofactory(classname="Files.InFileset")
numOfFiles = len(wmbsDao.execute(firstFileset.id))
self.assertEqual(numOfFiles, len(dbsFiles))
# use the new spec with same inputdataset
block = self.dataset + "#1"
wmspec = self.createWMSpec("TestSpec1")
dbs = self.getDBS(wmspec)
wmbs = self.createWMBSHelperWithTopTask(wmspec, block)
# check duplicate insert
dbsFiles = dbs.getFileBlock(block)[block]['Files']
numOfFiles = wmbs.addFiles(dbs.getFileBlock(block)[block])
self.assertEqual(numOfFiles, 0)
secondFileset = wmbs.topLevelFileset
wmbsDao = wmbs.daofactory(classname="Files.InFileset")
numOfFiles = len(wmbsDao.execute(secondFileset.id))
self.assertEqual(numOfFiles, len(dbsFiles))
self.assertNotEqual(firstFileset.id, secondFileset.id)
def testDuplicateSubscription(self):
"""Can't duplicate subscriptions"""
# using default wmspec
block = self.dataset + "#1"
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
wmbs.topLevelFileset.loadData()
numOfFiles = len(wmbs.topLevelFileset.files)
filesetId = wmbs.topLevelFileset.id
subId = wmbs.topLevelSubscription['id']
# check initially inserted files.
dbsFiles = self.dbs.getFileBlock(block)[block]['Files']
self.assertEqual(numOfFiles, len(dbsFiles))
firstFileset = wmbs.topLevelFileset
self.assertEqual(numOfFiles, len(dbsFiles))
# reinsert subscription - shouldn't create anything new
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, block)
wmbs.topLevelFileset.loadData()
self.assertEqual(numOfFiles, len(wmbs.topLevelFileset.files))
self.assertEqual(filesetId, wmbs.topLevelFileset.id)
self.assertEqual(subId, wmbs.topLevelSubscription['id'])
# now do a montecarlo workflow
self.setupMCWMSpec()
mask = Mask(FirstRun=12,
FirstLumi=1234,
FirstEvent=12345,
LastEvent=999995,
LastLumi=12345,
LastRun=12)
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, None, mask)
wmbs.topLevelFileset.loadData()
numOfFiles = len(wmbs.topLevelFileset.files)
filesetId = wmbs.topLevelFileset.id
subId = wmbs.topLevelSubscription['id']
# check initially inserted files.
numDbsFiles = 1
self.assertEqual(numOfFiles, numDbsFiles)
firstFileset = wmbs.topLevelFileset
self.assertEqual(numOfFiles, numDbsFiles)
# reinsert subscription - shouldn't create anything new
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, None, mask)
wmbs.topLevelFileset.loadData()
self.assertEqual(numOfFiles, len(wmbs.topLevelFileset.files))
self.assertEqual(filesetId, wmbs.topLevelFileset.id)
self.assertEqual(subId, wmbs.topLevelSubscription['id'])
def testParentage(self):
"""
1. check whether parent files are created in wmbs.
2. check parent files are associated to child.
3. When 2 specs with the same input data (one with parent processing, one without it)
is inserted, if one without parent processing inserted first then the other with
parent processing insert, it still needs to create parent files although child files
are duplicate
"""
block = self.dataset + "#1"
wmbs, sub, numFiles = self.createWMBSHelperWithTopTask(
self.wmspec, block, parentFlag=False, detail=True)
# file creation without parents
self.assertEqual(GlobalParams.numOfFilesPerBlock(), numFiles)
wmbs.topLevelFileset.loadData()
for child in wmbs.topLevelFileset.files:
# no parent per child
self.assertEqual(len(child["parents"]), 0)
wmbs, sub, numFiles = self.createWMBSHelperWithTopTask(self.wmspec,
block,
parentFlag=True,
detail=True)
self.assertEqual(GlobalParams.numOfFilesPerBlock(), numFiles)
wmbs.topLevelFileset.loadData()
for child in wmbs.topLevelFileset.files:
# one parent per child
self.assertEqual(len(child["parents"]), 1)
def testMCFakeFileInjection(self):
"""Inject fake Monte Carlo files into WMBS"""
self.setupMCWMSpec()
mask = Mask(FirstRun=12,
FirstLumi=1234,
FirstEvent=12345,
LastEvent=999995,
LastLumi=12345,
LastRun=12)
wmbs = self.createWMBSHelperWithTopTask(self.wmspec, None, mask)
subscription = wmbs.topLevelSubscription
self.assertEqual(1, subscription.exists())
fileset = subscription['fileset']
self.assertEqual(1, fileset.exists())
fileset.loadData() # need to refresh from database
self.assertEqual(len(fileset.files), 1)
self.assertEqual(len(fileset.parents), 0)
self.assertFalse(fileset.open)
file = list(fileset.files)[0]
self.assertEqual(file['events'], mask['LastEvent'] -
mask['FirstEvent'] + 1) # inclusive range
self.assertEqual(file['merged'],
False) # merged files get added to dbs
self.assertEqual(len(file['parents']), 0)
#file.loadData()
self.assertEqual(sorted(file['locations']), sorted(self.ses))
self.assertEqual(len(file.getParentLFNs()), 0)
self.assertEqual(len(file.getRuns()), 1)
run = file.getRuns()[0]
self.assertEqual(run.run, mask['FirstRun'])
self.assertEqual(run.lumis[0], mask['FirstLumi'])
self.assertEqual(run.lumis[-1], mask['LastLumi'])
self.assertEqual(len(run.lumis),
mask['LastLumi'] - mask['FirstLumi'] + 1)
def testJobSummary(self):
"""
_testJobSummary_
verify that job summary for jobs with fwjr are correctly created
and that status is updated when updatesummary flag is enabled
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFile = File(lfn="SomeLFNC", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 1, \
"Error: Splitting should have created one job."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Production"
change.propagate([testJobA], 'created', 'new')
myReport = Report()
reportPath = os.path.join(getTestBase(),
"WMCore_t/JobStateMachine_t/Report.pkl")
myReport.unpersist(reportPath)
change.propagate([testJobA], 'executing', 'created')
testJobA["fwjr"] = myReport
change.propagate([testJobA], 'jobfailed', 'executing')
changeStateDB = self.couchServer.connectDatabase(
dbname=self.config.JobStateMachine.jobSummaryDBName)
allDocs = changeStateDB.document("_all_docs")
self.assertEqual(len(allDocs["rows"]), 2,
"Error: Wrong number of documents")
fwjrDoc = {'state': None}
for resultRow in allDocs["rows"]:
if resultRow["id"] != "_design/WMStats":
fwjrDoc = changeStateDB.document(resultRow["id"])
break
self.assertEqual(fwjrDoc['state'], 'jobfailed',
"Error: summary doesn't have the expected job state")
del testJobA["fwjr"]
change.propagate([testJobA],
'jobcooloff',
'jobfailed',
updatesummary=True)
return
def testIndexConflict(self):
"""
_testIndexConflict_
Verify that in case of conflict in the job index
we discard the old document and replace with a new
one. Only works for MySQL backend
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFile = File(lfn="SomeLFNC", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 1, \
"Error: Splitting should have created one job."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "CompOps"
testJobA["taskType"] = "Processing"
myReport = Report()
reportPath = os.path.join(getTestBase(),
"WMCore_t/JobStateMachine_t/Report.pkl")
myReport.unpersist(reportPath)
testJobA["fwjr"] = myReport
change.propagate([testJobA], 'created', 'new')
jobdatabase = self.couchServer.connectDatabase('changestate_t/jobs',
False)
fwjrdatabase = self.couchServer.connectDatabase(
'changestate_t/fwjrs', False)
jobDoc = jobdatabase.document("1")
fwjrDoc = fwjrdatabase.document("1-0")
self.assertEqual(jobDoc["workflow"], "wf001",
"Wrong workflow in couch job document")
self.assertEqual(fwjrDoc["fwjr"]["task"], self.taskName,
"Wrong task in fwjr couch document")
testJobA.delete()
myThread = threading.currentThread()
myThread.dbi.processData("ALTER TABLE wmbs_job AUTO_INCREMENT = 1")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf002",
task="/TestWorkflow/Test2")
testWorkflow.create()
testFileset = Fileset(name="TestFilesetB")
testFileset.create()
testFile = File(lfn="SomeLFNB", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
testJobB = jobGroup.jobs[0]
testJobB["user"] = "******"
testJobB["group"] = "CompOps"
testJobB["taskType"] = "Processing"
testJobB["fwjr"] = myReport
change.propagate([testJobB], 'created', 'new')
jobDoc = jobdatabase.document("1")
fwjrDoc = fwjrdatabase.document("1-0")
self.assertEqual(jobDoc["workflow"], "wf002",
"Job document was not overwritten")
self.assertEqual(fwjrDoc["fwjr"]["task"], "/TestWorkflow/Test2",
"FWJR document was not overwritten")
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)
locationAction = daoFactory(classname="Locations.New")
changeStateAction = daoFactory(classname="Jobs.ChangeState")
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='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 testFWJRInputFileTruncation(self):
"""
_testFWJRInputFileTruncation_
Test and see whether the ChangeState code can
be used to automatically truncate the number of input files
in a FWJR
Code stolen from the serialization test
"""
self.config.JobStateMachine.maxFWJRInputFiles = 0
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFile = File(lfn="SomeLFNC", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
self.assertEqual(len(jobGroup.jobs), 1,
"Error: Splitting should have created one job.")
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
change.propagate([testJobA], 'created', 'new')
myReport = Report()
reportPath = os.path.join(getTestBase(),
"WMCore_t/JobStateMachine_t/Report.pkl")
myReport.unpersist(reportPath)
testJobA["fwjr"] = myReport
change.propagate([testJobA], 'executing', 'created')
changeStateDB = self.couchServer.connectDatabase(
dbname="changestate_t/fwjrs")
allDocs = changeStateDB.document("_all_docs")
self.assertEqual(len(allDocs["rows"]), 2,
"Error: Wrong number of documents")
result = changeStateDB.loadView("FWJRDump", "fwjrsByWorkflowName")
self.assertEqual(len(result["rows"]), 1,
"Error: Wrong number of rows.")
for row in result["rows"]:
couchJobDoc = changeStateDB.document(row["value"]["id"])
self.assertEqual(couchJobDoc["_rev"], row["value"]["rev"],
"Error: Rev is wrong.")
for resultRow in allDocs["rows"]:
if resultRow["id"] != "_design/FWJRDump":
fwjrDoc = changeStateDB.document(resultRow["id"])
break
self.assertEqual(
fwjrDoc["fwjr"]["steps"]['cmsRun1']['input']['source'], [])
return
def testJobKilling(self):
"""
_testJobKilling_
Test that we can successfully set jobs to the killed state
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
for i in range(4):
newFile = File(lfn="File%s" % i, locations=set(["T2_CH_CERN"]))
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased")
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 4, \
"Error: Splitting should have created four jobs."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
testJobB = jobGroup.jobs[1]
testJobB["user"] = "******"
testJobB["group"] = "DMWM"
testJobB["taskType"] = "Processing"
testJobC = jobGroup.jobs[2]
testJobC["user"] = "******"
testJobC["group"] = "DMWM"
testJobC["taskType"] = "Processing"
testJobD = jobGroup.jobs[3]
testJobD["user"] = "******"
testJobD["group"] = "DMWM"
testJobD["taskType"] = "Processing"
change.persist([testJobA], "created", "new")
change.persist([testJobB], "jobfailed", "executing")
change.persist([testJobC, testJobD], "executing", "created")
change.persist([testJobA], "killed", "created")
change.persist([testJobB], "killed", "jobfailed")
change.persist([testJobC, testJobD], "killed", "executing")
for job in [testJobA, testJobB, testJobC, testJobD]:
job.load()
self.assertEqual(job['retry_count'], 99999)
self.assertEqual(job['state'], 'killed')
return
def testDuplicateJobReports(self):
"""
_testDuplicateJobReports_
Verify that everything works correctly if a job report is added to the
database more than once.
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFile = File(lfn="SomeLFNC", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 1, \
"Error: Splitting should have created one job."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
change.propagate([testJobA], 'created', 'new')
myReport = Report()
reportPath = os.path.join(getTestBase(),
"WMCore_t/JobStateMachine_t/Report.pkl")
myReport.unpersist(reportPath)
testJobA["fwjr"] = myReport
change.propagate([testJobA], 'executing', 'created')
change.propagate([testJobA], 'executing', 'created')
changeStateDB = self.couchServer.connectDatabase(
dbname="changestate_t/fwjrs")
allDocs = changeStateDB.document("_all_docs")
self.assertEqual(len(allDocs["rows"]), 2,
"Error: Wrong number of documents")
for resultRow in allDocs["rows"]:
if resultRow["id"] != "_design/FWJRDump":
changeStateDB.document(resultRow["id"])
break
return
def testCreateSubscription(self):
"""
_testCreateSubscription_
Verify that the subscription creation code works correctly.
"""
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='site1',
seName='goodse.cern.ch',
ceName='site1',
plugin="TestPlugin")
resourceControl.insertSite(siteName='site2',
seName='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.createSubscription(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 testJobSerialization(self):
"""
_testJobSerialization_
Verify that serialization of a job works when adding a FWJR.
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFile = File(lfn="SomeLFNC", locations=set(["T2_CH_CERN"]))
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 1, \
"Error: Splitting should have created one job."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Processing"
change.propagate([testJobA], 'created', 'new')
myReport = Report()
reportPath = os.path.join(getTestBase(),
"WMCore_t/JobStateMachine_t/Report.pkl")
myReport.unpersist(reportPath)
testJobA["fwjr"] = myReport
change.propagate([testJobA], 'executing', 'created')
changeStateDB = self.couchServer.connectDatabase(
dbname="changestate_t/fwjrs")
allDocs = changeStateDB.document("_all_docs")
self.assertEqual(len(allDocs["rows"]), 2,
"Error: Wrong number of documents")
result = changeStateDB.loadView("FWJRDump", "fwjrsByWorkflowName")
self.assertEqual(len(result["rows"]), 1,
"Error: Wrong number of rows.")
for row in result["rows"]:
couchJobDoc = changeStateDB.document(row["value"]["id"])
self.assertEqual(couchJobDoc["_rev"], row["value"]["rev"],
"Error: Rev is wrong.")
for resultRow in allDocs["rows"]:
if resultRow["id"] != "_design/FWJRDump":
fwjrDoc = changeStateDB.document(resultRow["id"])
break
assert fwjrDoc["retrycount"] == 0, \
"Error: Retry count is wrong."
assert len(fwjrDoc["fwjr"]["steps"].keys()) == 2, \
"Error: Wrong number of steps in FWJR."
assert "cmsRun1" in fwjrDoc["fwjr"]["steps"].keys(), \
"Error: cmsRun1 step is missing from FWJR."
assert "stageOut1" in fwjrDoc["fwjr"]["steps"].keys(), \
"Error: stageOut1 step is missing from FWJR."
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 testRecordInCouch(self):
"""
_testRecordInCouch_
Verify that jobs, state transitions and fwjrs are recorded correctly.
"""
change = ChangeState(self.config, "changestate_t")
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute("site1", pnn="T2_CH_CERN")
testWorkflow = Workflow(spec=self.specUrl,
owner="Steve",
name="wf001",
task=self.taskName)
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="FileBased")
testSubscription.create()
testFileA = File(lfn="SomeLFNA",
events=1024,
size=2048,
locations=set(["T2_CH_CERN"]))
testFileB = File(lfn="SomeLFNB",
events=1025,
size=2049,
locations=set(["T2_CH_CERN"]))
testFileA.create()
testFileB.create()
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.commit()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroup = jobFactory(files_per_job=1)[0]
assert len(jobGroup.jobs) == 2, \
"Error: Splitting should have created two jobs."
testJobA = jobGroup.jobs[0]
testJobA["user"] = "******"
testJobA["group"] = "DMWM"
testJobA["taskType"] = "Merge"
testJobB = jobGroup.jobs[1]
testJobB["user"] = "******"
testJobB["group"] = "DMWM"
testJobB["taskType"] = "Processing"
change.propagate([testJobA, testJobB], "new", "none")
change.propagate([testJobA, testJobB], "created", "new")
change.propagate([testJobA, testJobB], "executing", "created")
testJobADoc = change.jobsdatabase.document(testJobA["couch_record"])
for transition in testJobADoc["states"].itervalues():
self.assertTrue(
isinstance(transition["timestamp"], int)
or isinstance(transition["timestamp"], long))
self.assertEqual(testJobADoc["jobid"], testJobA["id"],
"Error: ID parameter is incorrect.")
assert testJobADoc["name"] == testJobA["name"], \
"Error: Name parameter is incorrect."
assert testJobADoc["jobgroup"] == testJobA["jobgroup"], \
"Error: Jobgroup parameter is incorrect."
assert testJobADoc["workflow"] == testJobA["workflow"], \
"Error: Workflow parameter is incorrect."
assert testJobADoc["task"] == testJobA["task"], \
"Error: Task parameter is incorrect."
assert testJobADoc["owner"] == testJobA["owner"], \
"Error: Owner parameter is incorrect."
assert testJobADoc["mask"]["FirstEvent"] == testJobA["mask"]["FirstEvent"], \
"Error: First event in mask is incorrect."
assert testJobADoc["mask"]["LastEvent"] == testJobA["mask"]["LastEvent"], \
"Error: Last event in mask is incorrect."
assert testJobADoc["mask"]["FirstLumi"] == testJobA["mask"]["FirstLumi"], \
"Error: First lumi in mask is incorrect."
assert testJobADoc["mask"]["LastLumi"] == testJobA["mask"]["LastLumi"], \
"Error: First lumi in mask is incorrect."
assert testJobADoc["mask"]["FirstRun"] == testJobA["mask"]["FirstRun"], \
"Error: First run in mask is incorrect."
assert testJobADoc["mask"]["LastEvent"] == testJobA["mask"]["LastRun"], \
"Error: First event in mask is incorrect."
assert len(testJobADoc["inputfiles"]) == 1, \
"Error: Input files parameter is incorrect."
testJobBDoc = change.jobsdatabase.document(testJobB["couch_record"])
assert testJobBDoc["jobid"] == testJobB["id"], \
"Error: ID parameter is incorrect."
assert testJobBDoc["name"] == testJobB["name"], \
"Error: Name parameter is incorrect."
assert testJobBDoc["jobgroup"] == testJobB["jobgroup"], \
"Error: Jobgroup parameter is incorrect."
assert testJobBDoc["mask"]["FirstEvent"] == testJobB["mask"]["FirstEvent"], \
"Error: First event in mask is incorrect."
assert testJobBDoc["mask"]["LastEvent"] == testJobB["mask"]["LastEvent"], \
"Error: Last event in mask is incorrect."
assert testJobBDoc["mask"]["FirstLumi"] == testJobB["mask"]["FirstLumi"], \
"Error: First lumi in mask is incorrect."
assert testJobBDoc["mask"]["LastLumi"] == testJobB["mask"]["LastLumi"], \
"Error: First lumi in mask is incorrect."
assert testJobBDoc["mask"]["FirstRun"] == testJobB["mask"]["FirstRun"], \
"Error: First run in mask is incorrect."
assert testJobBDoc["mask"]["LastEvent"] == testJobB["mask"]["LastRun"], \
"Error: First event in mask is incorrect."
assert len(testJobBDoc["inputfiles"]) == 1, \
"Error: Input files parameter is incorrect."
changeStateDB = self.couchServer.connectDatabase(
dbname="changestate_t/jobs")
allDocs = changeStateDB.document("_all_docs")
self.assertEqual(len(allDocs["rows"]), 3,
"Error: Wrong number of documents.")
couchJobDoc = changeStateDB.document("1")
assert couchJobDoc["name"] == testJobA["name"], \
"Error: Name is wrong"
assert len(couchJobDoc["inputfiles"]) == 1, \
"Error: Wrong number of input files."
result = changeStateDB.loadView("JobDump", "jobsByWorkflowName")
self.assertEqual(len(result["rows"]), 2,
"Error: Wrong number of rows.")
for row in result["rows"]:
couchJobDoc = changeStateDB.document(row["value"]["id"])
self.assertEqual(couchJobDoc["_rev"], row["value"]["rev"],
"Error: Rev is wrong.")
return
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", pnn = "T2_CH_CERN")
self.multipleFileFileset = Fileset(name = "TestFileset1")
self.multipleFileFileset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(i, *[45+i]))
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(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleFileFileset.addFile(newFile)
self.singleFileFileset.commit()
self.multipleFileLumiset = Fileset(name = "TestFileset3")
self.multipleFileLumiset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45+i/3]))
newFile.create()
self.multipleFileLumiset.addFile(newFile)
self.multipleFileLumiset.commit()
self.singleLumiFileset = Fileset(name = "TestFileset4")
self.singleLumiFileset.create()
for i in range(10):
newFile = File(makeUUID(), size = 1000, events = 100, locations = set(["T2_CH_CERN"]))
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleLumiFileset.addFile(newFile)
self.singleLumiFileset.commit()
testWorkflow = Workflow(spec = "spec.xml", owner = "mnorman", name = "wf001", task="Test")
testWorkflow.create()
self.multipleFileSubscription = Subscription(fileset = self.multipleFileFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.singleFileSubscription = Subscription(fileset = self.singleFileFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.multipleLumiSubscription = Subscription(fileset = self.multipleFileLumiset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.singleLumiSubscription = Subscription(fileset = self.singleLumiFileset,
workflow = testWorkflow,
split_algo = "FixedDelay",
type = "Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription.create()
self.multipleLumiSubscription.create()
self.singleLumiSubscription.create()
return
def createJobs(self):
"""
_createJobs_
Create test jobs in WMBS and BossAir
"""
testWorkflow = Workflow(spec=makeUUID(),
owner="tapas",
name=makeUUID(),
task="Test")
testWorkflow.create()
testFilesetA = Fileset(name="TestFilesetA")
testFilesetA.create()
testFilesetB = Fileset(name="TestFilesetB")
testFilesetB.create()
testFilesetC = Fileset(name="TestFilesetC")
testFilesetC.create()
testFileA = File(lfn="testFileA",
locations=set(["testSE1", "testSE2"]))
testFileA.create()
testFilesetA.addFile(testFileA)
testFilesetA.commit()
testFilesetB.addFile(testFileA)
testFilesetB.commit()
testFilesetC.addFile(testFileA)
testFilesetC.commit()
testSubscriptionA = Subscription(fileset=testFilesetA,
workflow=testWorkflow,
type="Processing")
testSubscriptionA.create()
testSubscriptionA.addWhiteBlackList([{
"site_name": "testSite1",
"valid": True
}])
testSubscriptionB = Subscription(fileset=testFilesetB,
workflow=testWorkflow,
type="Processing")
testSubscriptionB.create()
testSubscriptionB.addWhiteBlackList([{
"site_name": "testSite1",
"valid": False
}])
testSubscriptionC = Subscription(fileset=testFilesetC,
workflow=testWorkflow,
type="Merge")
testSubscriptionC.create()
testJobGroupA = JobGroup(subscription=testSubscriptionA)
testJobGroupA.create()
testJobGroupB = JobGroup(subscription=testSubscriptionB)
testJobGroupB.create()
testJobGroupC = JobGroup(subscription=testSubscriptionC)
testJobGroupC.create()
# Site1, Has been assigned a location and is complete.
testJobA = Job(name="testJobA", files=[testFileA])
testJobA["couch_record"] = makeUUID()
testJobA.create(group=testJobGroupA)
testJobA["state"] = "success"
# Site 1, Has been assigned a location and is incomplete.
testJobB = Job(name="testJobB", files=[testFileA])
testJobB["couch_record"] = makeUUID()
testJobB["cache_dir"] = self.tempDir
testJobB.create(group=testJobGroupA)
testJobB["state"] = "executing"
runJobB = RunJob()
runJobB.buildFromJob(testJobB)
runJobB["status"] = "PEND"
# Does not have a location, white listed to site 1
testJobC = Job(name="testJobC", files=[testFileA])
testJobC["couch_record"] = makeUUID()
testJobC.create(group=testJobGroupA)
testJobC["state"] = "new"
# Site 2, Has been assigned a location and is complete.
testJobD = Job(name="testJobD", files=[testFileA])
testJobD["couch_record"] = makeUUID()
testJobD.create(group=testJobGroupB)
testJobD["state"] = "success"
# Site 2, Has been assigned a location and is incomplete.
testJobE = Job(name="testJobE", files=[testFileA])
testJobE["couch_record"] = makeUUID()
testJobE.create(group=testJobGroupB)
testJobE["state"] = "executing"
runJobE = RunJob()
runJobE.buildFromJob(testJobE)
runJobE["status"] = "RUN"
# Does not have a location, site 1 is blacklisted.
testJobF = Job(name="testJobF", files=[testFileA])
testJobF["couch_record"] = makeUUID()
testJobF.create(group=testJobGroupB)
testJobF["state"] = "new"
# Site 3, Has been assigned a location and is complete.
testJobG = Job(name="testJobG", files=[testFileA])
testJobG["couch_record"] = makeUUID()
testJobG.create(group=testJobGroupC)
testJobG["state"] = "cleanout"
# Site 3, Has been assigned a location and is incomplete.
testJobH = Job(name="testJobH", files=[testFileA])
testJobH["couch_record"] = makeUUID()
testJobH.create(group=testJobGroupC)
testJobH["state"] = "new"
# Site 3, Does not have a location.
testJobI = Job(name="testJobI", files=[testFileA])
testJobI["couch_record"] = makeUUID()
testJobI.create(group=testJobGroupC)
testJobI["state"] = "new"
# Site 3, Does not have a location and is in cleanout.
testJobJ = Job(name="testJobJ", files=[testFileA])
testJobJ["couch_record"] = makeUUID()
testJobJ.create(group=testJobGroupC)
testJobJ["state"] = "cleanout"
changeStateAction = self.daoFactory(classname="Jobs.ChangeState")
changeStateAction.execute(jobs=[
testJobA, testJobB, testJobC, testJobD, testJobE, testJobF,
testJobG, testJobH, testJobI, testJobJ
])
self.insertRunJob.execute([runJobB, runJobE])
setLocationAction = self.daoFactory(classname="Jobs.SetLocation")
setLocationAction.execute(testJobA["id"], "testSite1")
setLocationAction.execute(testJobB["id"], "testSite1")
setLocationAction.execute(testJobD["id"], "testSite1")
setLocationAction.execute(testJobE["id"], "testSite2")
setLocationAction.execute(testJobG["id"], "testSite1")
setLocationAction.execute(testJobH["id"], "testSite1")
return
