class AlertProcessorTest(unittest.TestCase):
def setUp(self):
self.testInit = TestInit(__file__)
self.testInit.setLogging(logLevel = logging.DEBUG)
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMCore.WMBS",'WMCore.Agent.Database',
"WMCore.ResourceControl"],
useDefault = False)
self.testDir = self.testInit.generateWorkDir()
self.config = Configuration()
self.config.section_("Agent")
self.config.Agent.useMsgService = False
self.config.Agent.useTrigger = False
self.config.component_("AlertProcessor")
self.config.AlertProcessor.componentDir = self.testDir
self.config.AlertProcessor.address = "tcp://127.0.0.1:5557"
self.config.AlertProcessor.controlAddr = "tcp://127.0.0.1:5559"
self.config.section_("CoreDatabase")
self.config.CoreDatabase.socket = os.environ.get("DBSOCK")
self.config.CoreDatabase.connectUrl = os.environ.get("DATABASE")
self.config.AlertProcessor.section_("critical")
self.config.AlertProcessor.section_("soft")
self.config.AlertProcessor.critical.level = 5
self.config.AlertProcessor.soft.level = 0
self.config.AlertProcessor.soft.bufferSize = 3
self.config.AlertProcessor.critical.section_("sinks")
self.config.AlertProcessor.soft.section_("sinks")
def tearDown(self):
self.testInit.clearDatabase()
self.testInit.delWorkDir()
def testAlertProcessorBasic(self):
alertProcessor = AlertProcessor(self.config)
try:
# alertProcessor.startComponent() causes the flow to stop, Harness.py
# the method just calls prepareToStart() and waits for ever
# alertProcessor.startDaemon() no good for this either ... puts everything
# on background
alertProcessor.prepareToStart()
except Exception, ex:
print ex
self.fail(str(ex))
logging.debug("AlertProcessor and its sub-components should be running now ...")
logging.debug("Going to stop the component ...")
# stop via component method
try:
alertProcessor.stopAlertProcessor()
except Exception, ex:
print ex
self.fail(str(ex))
class AlertProcessorTest(unittest.TestCase):
def setUp(self):
self.testInit = TestInit(__file__)
self.testInit.setLogging(logLevel=logging.DEBUG)
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=[
"WMCore.WMBS", 'WMCore.Agent.Database', "WMCore.ResourceControl"
],
useDefault=False)
self.testDir = self.testInit.generateWorkDir()
self.config = Configuration()
self.config.section_("Agent")
self.config.Agent.useMsgService = False
self.config.Agent.useTrigger = False
self.config.component_("AlertProcessor")
self.config.AlertProcessor.componentDir = self.testDir
self.config.AlertProcessor.address = "tcp://127.0.0.1:5557"
self.config.AlertProcessor.controlAddr = "tcp://127.0.0.1:5559"
self.config.section_("CoreDatabase")
self.config.CoreDatabase.socket = os.environ.get("DBSOCK")
self.config.CoreDatabase.connectUrl = os.environ.get("DATABASE")
self.config.AlertProcessor.section_("critical")
self.config.AlertProcessor.section_("soft")
self.config.AlertProcessor.critical.level = 5
self.config.AlertProcessor.soft.level = 0
self.config.AlertProcessor.soft.bufferSize = 3
self.config.AlertProcessor.critical.section_("sinks")
self.config.AlertProcessor.soft.section_("sinks")
def tearDown(self):
self.testInit.clearDatabase()
self.testInit.delWorkDir()
def testAlertProcessorBasic(self):
alertProcessor = AlertProcessor(self.config)
try:
# alertProcessor.startComponent() causes the flow to stop, Harness.py
# the method just calls prepareToStart() and waits for ever
# alertProcessor.startDaemon() no good for this either ... puts everything
# on background
alertProcessor.prepareToStart()
except Exception as ex:
print ex
self.fail(str(ex))
logging.debug(
"AlertProcessor and its sub-components should be running now ...")
logging.debug("Going to stop the component ...")
# stop via component method
try:
alertProcessor.stopAlertProcessor()
except Exception as ex:
print ex
self.fail(str(ex))
class DBSBufferDatasetTest(unittest.TestCase):
"""
_DBSBufferDatasetTest_
Collection of tests for the DBSBuffer object
"""
def setUp(self):
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMComponent.DBS3Buffer"],
useDefault=False)
def tearDown(self):
self.testInit.clearDatabase()
def testBasic(self):
"""
_testBasic_
Test the basic functions of the DBSBufferDataset,
create, load, exists and also the ability
to add subscriptions.
"""
originalDataset = DBSBufferDataset(path='/bogus/bogus/go')
originalDataset.create()
myThread = threading.currentThread()
result = myThread.dbi.processData(
"SELECT id FROM dbsbuffer_dataset")[0].fetchall()
self.assertEqual(originalDataset.exists(), result[0][0])
duplicateDataset = DBSBufferDataset(path='/bogus/bogus/go')
duplicateDataset.create()
self.assertEqual(originalDataset.exists(), duplicateDataset.exists())
result = myThread.dbi.processData(
"SELECT COUNT(id) FROM dbsbuffer_dataset")[0].fetchall()
self.assertEqual(result[0][0], 1)
loadedDataset = DBSBufferDataset(path='/bogus/bogus/go')
loadedDataset.load()
self.assertEqual(loadedDataset.exists(), originalDataset.exists())
secondDataset = DBSBufferDataset(
path='/BogusPrimary/Run2012Z-PromptReco-v1/RECO')
secondDataset.create()
workload = WMWorkloadHelper()
workload.load(
os.path.join(
getTestBase(),
'WMComponent_t/PhEDExInjector_t/specs/TestWorkload.pkl'))
secondDataset.addSubscription(
workload.getSubscriptionInformation()
['/BogusPrimary/Run2012Z-PromptReco-v1/RECO'])
secondDataset.addSubscription(
workload.getSubscriptionInformation()
['/BogusPrimary/Run2012Z-PromptReco-v1/RECO'])
self.assertEqual(len(secondDataset['subscriptions']), 3)
result = myThread.dbi.processData(
"SELECT COUNT(id) FROM dbsbuffer_dataset_subscription"
)[0].fetchall()
self.assertEqual(result[0][0], 3)
return
class WorkQueueTestCase(unittest.TestCase):
def setSchema(self):
"this can be override if the schema setting is different"
self.schema = ["WMCore.WMBS","WMComponent.DBS3Buffer","WMCore.BossAir"]
self.couchApps = ["WorkQueue"]
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all of the
WMBS tables. Also add some dummy locations.
"""
self.queueDB = 'workqueue_t'
self.queueInboxDB = 'workqueue_t_inbox'
self.globalQDB = 'workqueue_t_global'
self.globalQInboxDB = 'workqueue_t_global_inbox'
self.localQDB = 'workqueue_t_local'
self.localQInboxDB = 'workqueue_t_local_inbox'
self.localQDB2 = 'workqueue_t_local2'
self.localQInboxDB2 = 'workqueue_t_local2_inbox'
self.configCacheDB = 'workqueue_t_config_cache'
self.setSchema()
self.testInit = TestInit('WorkQueueTest')
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = self.schema,
useDefault = False)
self.testInit.setupCouch(self.queueDB, *self.couchApps)
self.testInit.setupCouch(self.queueInboxDB, *self.couchApps)
self.testInit.setupCouch(self.globalQDB, *self.couchApps)
self.testInit.setupCouch(self.globalQInboxDB , *self.couchApps)
self.testInit.setupCouch(self.localQDB, *self.couchApps)
self.testInit.setupCouch(self.localQInboxDB, *self.couchApps)
self.testInit.setupCouch(self.localQDB2, *self.couchApps)
self.testInit.setupCouch(self.localQInboxDB2, *self.couchApps)
self.testInit.setupCouch(self.configCacheDB, 'ConfigCache')
couchServer = CouchServer(os.environ.get("COUCHURL"))
self.configCacheDBInstance = couchServer.connectDatabase(self.configCacheDB)
self.workDir = self.testInit.generateWorkDir()
return
def tearDown(self):
"""
_tearDown_
Drop all the WMBS tables.
"""
#self.testInit.tearDownCouch()
self.testInit.clearDatabase()
self.testInit.delWorkDir()
class DaemonTest(unittest.TestCase):
"""
_Daemon_t_
Unit tests for message services: subscription, priority subscription, buffers,
etc..
"""
# minimum number of messages that need to be in queue
_minMsg = 20
# number of publish and gets from queue
_publishAndGet = 10
def setUp(self):
"make a logger instance and create tables"
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.tempDir = tempfile.mkdtemp()
def tearDown(self):
"""
Deletion of the databases
"""
self.testInit.clearDatabase()
shutil.rmtree(self.tempDir, True)
def testA(self):
"""
__testSubscribe__
Test daemon creation
"""
# keep the parent alive
self.pid = createDaemon(self.tempDir, True)
try:
try:
if self.pid != 0:
time.sleep(2)
details = Details(os.path.join(self.tempDir, "Daemon.xml"))
time.sleep(10)
details.killWithPrejudice()
else:
while True:
time.sleep(1)
except:
pass
finally:
if self.pid == 0:
os._exit(-1)
else:
os.system('kill -9 %s' % self.pid)
class DaemonTest(unittest.TestCase):
"""
_Daemon_t_
Unit tests for message services: subscription, priority subscription, buffers,
etc..
"""
# minimum number of messages that need to be in queue
_minMsg = 20
# number of publish and gets from queue
_publishAndGet = 10
def setUp(self):
"make a logger instance and create tables"
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.tempDir = tempfile.mkdtemp()
def tearDown(self):
"""
Deletion of the databases
"""
self.testInit.clearDatabase()
shutil.rmtree( self.tempDir, True )
def testA(self):
"""
__testSubscribe__
Test daemon creation
"""
# keep the parent alive
self.pid = createDaemon(self.tempDir, True)
try:
try:
if self.pid != 0 :
time.sleep(2)
details = Details(os.path.join(self.tempDir,"Daemon.xml"))
time.sleep(10)
details.killWithPrejudice()
else:
while True:
time.sleep(1)
except:
pass
finally:
if self.pid == 0:
os._exit(-1)
else:
os.system('kill -9 %s' % self.pid)
class AlertGeneratorTest(unittest.TestCase):
def setUp(self):
self.testInit = TestInit(__file__)
self.testInit.setLogging(logLevel = logging.DEBUG)
self.testInit.clearDatabase()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMCore.WMBS",'WMCore.Agent.Database',
"WMCore.ResourceControl"],
useDefault = False)
self.testDir = self.testInit.generateWorkDir()
# AlertGenerator instance
self.generator = None
self.config = getConfig(self.testDir)
self.config.section_("CoreDatabase")
self.config.CoreDatabase.socket = os.environ.get("DBSOCK")
self.config.CoreDatabase.connectUrl = os.environ.get("DATABASE")
self.testProcesses = []
self.testComponentDaemonXml = "/tmp/TestComponent/Daemon.xml"
def tearDown(self):
self.testInit.clearDatabase()
self.testInit.delWorkDir()
self.generator = None
utils.terminateProcesses(self.testProcesses)
# if the directory and file "/tmp/TestComponent/Daemon.xml" after
# ComponentsPoller test exist, then delete it
d = os.path.dirname(self.testComponentDaemonXml)
if os.path.exists(d):
shutil.rmtree(d)
def _startComponent(self):
self.generator = AlertGenerator(self.config)
try:
# self.proc.startComponent() causes the flow to stop, Harness.py
# the method just calls prepareToStart() and waits for ever
# self.proc.startDaemon() no good for this either ... puts everything
# on background
self.generator.prepareToStart()
except Exception, ex:
print ex
self.fail(str(ex))
print "AlertGenerator and its sub-components should be running now ..."
class JobTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all of the
WMBS tables.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMCore.WMBS"],
useDefault=False)
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationNew = self.daoFactory(classname="Locations.New")
locationNew.execute(siteName="test.site.ch", pnn="T2_CH_CERN")
locationNew.execute(siteName="test2.site.ch", pnn="T2_CH_CERN")
return
def tearDown(self):
"""
_tearDown_
Drop all the WMBS tables.
"""
self.testInit.clearDatabase()
def createTestJob(self, 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 testCreateDeleteExists(self):
"""
_testCreateDeleteExists_
Create and then delete a job. 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()
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() == 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 testCreateTransaction(self):
"""
_testCreateTransaction_
Create a job and save it to the database. Roll back the database
transaction and verify that the job is no longer 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()
myThread = threading.currentThread()
myThread.transaction.begin()
testJob = Job(name="TestJob", files=[testFileA, testFileB])
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"
myThread.transaction.rollback()
assert testJob.exists() == False, \
"ERROR: Job exists after transaction was rolled back."
return
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() == 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() == 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 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 testLoad(self):
"""
_testLoad_
Create a job and save it to the database. Load it back from the
database using the name and the id and then verify that all information
was loaded correctly.
"""
testJobA = self.createTestJob()
testJobB = Job(id=testJobA["id"])
testJobC = Job(name=testJobA["name"])
testJobB.load()
testJobC.load()
assert (testJobA["id"] == testJobB["id"]) and \
(testJobA["name"] == testJobB["name"]) and \
(testJobA["jobgroup"] == testJobB["jobgroup"]) and \
(testJobA["couch_record"] == testJobB["couch_record"]) and \
(testJobA["location"] == testJobB["location"]), \
"ERROR: Load from ID didn't load everything correctly"
assert (testJobA["id"] == testJobC["id"]) and \
(testJobA["name"] == testJobC["name"]) and \
(testJobA["jobgroup"] == testJobC["jobgroup"]) and \
(testJobA["couch_record"] == testJobC["couch_record"]) and \
(testJobA["location"] == testJobC["location"]), \
"ERROR: Load from name didn't load everything correctly"
self.assertEqual(testJobB['outcome'], 'failure')
self.assertEqual(testJobC['outcome'], 'failure')
self.assertEqual(testJobB['fwjr'], None)
self.assertEqual(testJobC['fwjr'], None)
return
def testLoadData(self):
"""
_testLoadData_
Create a job and save it to the database. Load it back from the
database using the name and the id. Verify that all job information
is correct including input files and the job mask.
"""
testJobA = self.createTestJob()
testJobA["mask"]["FirstEvent"] = 1
testJobA["mask"]["LastEvent"] = 2
testJobA["mask"]["FirstLumi"] = 3
testJobA["mask"]["LastLumi"] = 4
testJobA["mask"]["FirstRun"] = 5
testJobA["mask"]["LastRun"] = 6
testJobA.save()
testJobB = Job(id=testJobA["id"])
testJobC = Job(name=testJobA["name"])
testJobB.loadData()
testJobC.loadData()
assert (testJobA["id"] == testJobB["id"]) and \
(testJobA["name"] == testJobB["name"]) and \
(testJobA["jobgroup"] == testJobB["jobgroup"]) and \
(testJobA["couch_record"] == testJobB["couch_record"]) and \
(testJobA["location"] == testJobB["location"]), \
"ERROR: Load from ID didn't load everything correctly"
assert (testJobA["id"] == testJobC["id"]) and \
(testJobA["name"] == testJobC["name"]) and \
(testJobA["jobgroup"] == testJobC["jobgroup"]) and \
(testJobA["couch_record"] == testJobC["couch_record"]) and \
(testJobA["location"] == testJobC["location"]), \
"ERROR: Load from name didn't load everything correctly"
assert testJobA["mask"] == testJobB["mask"], \
"ERROR: Job mask did not load properly"
assert testJobA["mask"] == testJobC["mask"], \
"ERROR: Job mask did not load properly"
goldenFiles = testJobA.getFiles()
for testFile in testJobB.getFiles():
assert testFile in goldenFiles, \
"ERROR: Job loaded an unknown file"
goldenFiles.remove(testFile)
assert len(goldenFiles) == 0, \
"ERROR: Job didn't load all files"
goldenFiles = testJobA.getFiles()
for testFile in testJobC.getFiles():
assert testFile in goldenFiles, \
"ERROR: Job loaded an unknown file"
goldenFiles.remove(testFile)
assert len(goldenFiles) == 0, \
"ERROR: Job didn't load all files"
return
def testGetFiles(self):
"""
_testGetFiles_
Test the Job's getFiles() method. This should load the files from
the database if they haven't been loaded already.
"""
testJobA = self.createTestJob()
testJobB = Job(id=testJobA["id"])
testJobB.loadData()
goldenFiles = testJobA.getFiles()
for testFile in testJobB.getFiles():
assert testFile in goldenFiles, \
"ERROR: Job loaded an unknown file: %s" % testFile
goldenFiles.remove(testFile)
assert len(goldenFiles) == 0, \
"ERROR: Job didn't load all files"
return
def testSaveTransaction(self):
"""
_testSaveTransaction_
Create a job and a job mask and save them both to the database. Load
the job from the database and verify that everything was written
correctly. Begin a new transaction and update the job mask again.
Load the mask and verify that it's correct. Finally, rollback the
transaction and reload the mask to verify that it is in the correct
state.
"""
testJobA = self.createTestJob()
testJobA["mask"]["FirstEvent"] = 1
testJobA["mask"]["LastEvent"] = 2
testJobA["mask"]["FirstLumi"] = 3
testJobA["mask"]["LastLumi"] = 4
testJobA["mask"]["FirstRun"] = 5
testJobA["mask"]["LastRun"] = 6
testJobA.save()
testJobB = Job(id=testJobA["id"])
testJobB.loadData()
assert testJobA["mask"] == testJobB["mask"], \
"ERROR: Job mask did not load properly"
myThread = threading.currentThread()
myThread.transaction.begin()
testJobA["mask"]["FirstEvent"] = 7
testJobA["mask"]["LastEvent"] = 8
testJobA["mask"]["FirstLumi"] = 9
testJobA["mask"]["LastLumi"] = 10
testJobA["mask"]["FirstRun"] = 11
testJobA["mask"]["LastRun"] = 12
testJobA["name"] = "stevesJob"
testJobA["couch_record"] = "someCouchRecord"
testJobA["location"] = "test2.site.ch"
testJobA.save()
testJobC = Job(id=testJobA["id"])
testJobC.loadData()
assert testJobA["mask"] == testJobC["mask"], \
"ERROR: Job mask did not load properly"
assert testJobC["name"] == "stevesJob", \
"ERROR: Job name did not save"
assert testJobC["couch_record"] == "someCouchRecord", \
"ERROR: Job couch record did not save"
assert testJobC["location"] == "test2.site.ch", \
"ERROR: Job site did not save"
myThread.transaction.rollback()
testJobD = Job(id=testJobA["id"])
testJobD.loadData()
assert testJobB["mask"] == testJobD["mask"], \
"ERROR: Job mask did not load properly"
return
def testJobState(self):
"""
_testJobState_
Unittest to see if we can figure out what the jobState actually is and set it
"""
testJobA = self.createTestJob()
value = testJobA.getState()
self.assertEqual(value, 'new')
return
def testJobCacheDir(self):
"""
_testJobCacheDir_
Check retrieval of the jobCache directory.
"""
testJobA = self.createTestJob()
value = testJobA.getCache()
self.assertEqual(value, None)
testJobA.setCache('UnderTheDeepBlueSea')
value = testJobA.getCache()
self.assertEqual(value, 'UnderTheDeepBlueSea')
return
def testGetOutputParentLFNs(self):
"""
_testGetOutputParentLFNs_
Verify that the getOutputDBSParentLFNs() method returns the correct
parent LFNs.
"""
testWorkflow = Workflow(spec="spec.xml",
owner="Simon",
name="wf001",
task="Test")
testWorkflow.create()
testWMBSFileset = Fileset(name="TestFileset")
testWMBSFileset.create()
testSubscription = Subscription(fileset=testWMBSFileset,
workflow=testWorkflow)
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
testFileA = File(lfn="/this/is/a/lfnA",
size=1024,
events=10,
merged=True)
testFileB = File(lfn="/this/is/a/lfnB",
size=1024,
events=10,
merged=True)
testFileC = File(lfn="/this/is/a/lfnC",
size=1024,
events=10,
merged=False)
testFileD = File(lfn="/this/is/a/lfnD",
size=1024,
events=10,
merged=False)
testFileE = File(lfn="/this/is/a/lfnE",
size=1024,
events=10,
merged=True)
testFileF = File(lfn="/this/is/a/lfnF",
size=1024,
events=10,
merged=True)
testFileA.create()
testFileB.create()
testFileC.create()
testFileD.create()
testFileE.create()
testFileF.create()
testFileE.addChild(testFileC["lfn"])
testFileF.addChild(testFileD["lfn"])
testJobA = Job(name="TestJob", files=[testFileA, testFileB])
testJobA["couch_record"] = "somecouchrecord"
testJobA["location"] = "test.site.ch"
testJobA.create(group=testJobGroup)
testJobA.associateFiles()
testJobB = Job(name="TestJobB", files=[testFileC, testFileD])
testJobB["couch_record"] = "somecouchrecord"
testJobB["location"] = "test.site.ch"
testJobB.create(group=testJobGroup)
testJobB.associateFiles()
goldenLFNs = ["/this/is/a/lfnA", "/this/is/a/lfnB"]
parentLFNs = testJobA.getOutputDBSParentLFNs()
for parentLFN in parentLFNs:
assert parentLFN in goldenLFNs, \
"ERROR: Unknown lfn: %s" % parentLFN
goldenLFNs.remove(parentLFN)
assert len(goldenLFNs) == 0, \
"ERROR: LFNs are missing: %s" % goldenLFNs
goldenLFNs = ["/this/is/a/lfnE", "/this/is/a/lfnF"]
parentLFNs = testJobB.getOutputDBSParentLFNs()
for parentLFN in parentLFNs:
assert parentLFN in goldenLFNs, \
"ERROR: Unknown lfn: %s" % parentLFN
goldenLFNs.remove(parentLFN)
assert len(goldenLFNs) == 0, \
"ERROR: LFNs are missing..."
return
def testJobFWJRPath(self):
"""
_testJobFWJRPath_
Verify the correct operation of the Jobs.SetFWJRPath and
Jobs.GetFWJRByState DAOs.
"""
testJobA = self.createTestJob()
testJobA["state"] = "complete"
testJobB = self.createTestJob()
testJobB["state"] = "executing"
testJobC = self.createTestJob()
testJobC["state"] = "complete"
myThread = threading.currentThread()
setFWJRAction = self.daoFactory(classname="Jobs.SetFWJRPath")
setFWJRAction.execute(jobID=testJobA["id"],
fwjrPath="NonsenseA",
conn=myThread.transaction.conn,
transaction=True)
setFWJRAction.execute(jobID=testJobB["id"],
fwjrPath="NonsenseB",
conn=myThread.transaction.conn,
transaction=True)
setFWJRAction.execute(jobID=testJobC["id"],
fwjrPath="NonsenseC",
conn=myThread.transaction.conn,
transaction=True)
changeStateAction = self.daoFactory(classname="Jobs.ChangeState")
changeStateAction.execute(jobs=[testJobA, testJobB, testJobC],
conn=myThread.transaction.conn,
transaction=True)
getJobsAction = self.daoFactory(classname="Jobs.GetFWJRByState")
jobs = getJobsAction.execute(state="complete",
conn=myThread.transaction.conn,
transaction=True)
goldenIDs = [testJobA["id"], testJobC["id"]]
for job in jobs:
assert job["id"] in goldenIDs, \
"Error: Unknown job: %s" % job["id"]
goldenIDs.remove(job["id"])
if job["id"] == testJobA["id"]:
assert job["fwjr_path"] == "NonsenseA", \
"Error: Wrong fwjr path: %s" % job["fwjr_path"]
else:
assert job["fwjr_path"] == "NonsenseC", \
"Error: Wrong fwjr path: %s" % job["fwjr_path"]
assert len(goldenIDs) == 0, \
"Error: Jobs missing: %s" % len(goldenIDs)
return
def testFailJobInput(self):
"""
_testFailJobInput_
Test the Jobs.FailInput DAO and verify that it doesn't affect other
jobs/subscriptions that run over the same files.
"""
testWorkflow = Workflow(spec="spec.xml",
owner="Steve",
name="wf001",
task="Test")
bogusWorkflow = Workflow(spec="spec1.xml",
owner="Steve",
name="wf002",
task="Test")
testWorkflow.create()
bogusWorkflow.create()
testFileset = Fileset(name="TestFileset")
bogusFileset = Fileset(name="BogusFileset")
testFileset.create()
bogusFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
bogusSubscription = Subscription(fileset=bogusFileset,
workflow=bogusWorkflow)
testSubscription.create()
bogusSubscription.create()
testFileA = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileB = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileC = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileA.create()
testFileB.create()
testFileC.create()
testFileset.addFile([testFileA, testFileB, testFileC])
bogusFileset.addFile([testFileA, testFileB, testFileC])
testFileset.commit()
bogusFileset.commit()
testSubscription.completeFiles([testFileA, testFileB, testFileC])
bogusSubscription.acquireFiles([testFileA, testFileB, testFileC])
testJobGroup = JobGroup(subscription=testSubscription)
bogusJobGroup = JobGroup(subscription=bogusSubscription)
testJobGroup.create()
bogusJobGroup.create()
testJobA = Job(name="TestJobA",
files=[testFileA, testFileB, testFileC])
testJobB = Job(name="TestJobB",
files=[testFileA, testFileB, testFileC])
bogusJob = Job(name="BogusJob",
files=[testFileA, testFileB, testFileC])
testJobA.create(group=testJobGroup)
testJobB.create(group=testJobGroup)
bogusJob.create(group=bogusJobGroup)
testJobA.failInputFiles()
testJobB.failInputFiles()
self.assertEqual(len(testSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Acquired")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Failed")), 3)
self.assertEqual(len(testSubscription.filesOfStatus("Completed")), 0)
changeStateAction = self.daoFactory(classname="Jobs.ChangeState")
testJobB["state"] = "cleanout"
changeStateAction.execute([testJobB])
# Try again
testJobA.failInputFiles()
# Should now be failed
self.assertEqual(len(testSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Acquired")), 0)
self.assertEqual(len(testSubscription.filesOfStatus("Failed")), 3)
self.assertEqual(len(testSubscription.filesOfStatus("Completed")), 0)
# bogus should be unchanged
self.assertEqual(len(bogusSubscription.filesOfStatus("Available")), 0)
self.assertEqual(len(bogusSubscription.filesOfStatus("Acquired")), 3)
self.assertEqual(len(bogusSubscription.filesOfStatus("Failed")), 0)
self.assertEqual(len(bogusSubscription.filesOfStatus("Completed")), 0)
return
def testCompleteJobInput(self):
"""
_testCompleteJobInput_
Verify the correct output of the CompleteInput DAO. This should mark
the input for a job as complete once all the jobs that run over a
particular file have complete successfully.
"""
testWorkflow = Workflow(spec="spec.xml",
owner="Steve",
name="wf001",
task="Test")
bogusWorkflow = Workflow(spec="spec1.xml",
owner="Steve",
name="wf002",
task="Test")
testWorkflow.create()
bogusWorkflow.create()
testFileset = Fileset(name="TestFileset")
bogusFileset = Fileset(name="BogusFileset")
testFileset.create()
bogusFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
bogusSubscription = Subscription(fileset=bogusFileset,
workflow=bogusWorkflow)
testSubscription.create()
bogusSubscription.create()
testFileA = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileB = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileC = File(lfn=makeUUID(), locations="T2_CH_CERN")
testFileA.create()
testFileB.create()
testFileC.create()
testFileset.addFile([testFileA, testFileB, testFileC])
bogusFileset.addFile([testFileA, testFileB, testFileC])
testFileset.commit()
bogusFileset.commit()
testSubscription.acquireFiles([testFileA, testFileB, testFileC])
bogusSubscription.acquireFiles([testFileA, testFileB, testFileC])
testJobGroup = JobGroup(subscription=testSubscription)
bogusJobGroup = JobGroup(subscription=bogusSubscription)
testJobGroup.create()
bogusJobGroup.create()
testJobA = Job(name="TestJobA", files=[testFileA])
testJobB = Job(name="TestJobB", files=[testFileA, testFileB])
testJobC = Job(name="TestJobC", files=[testFileC])
bogusJob = Job(name="BogusJob",
files=[testFileA, testFileB, testFileC])
testJobA.create(group=testJobGroup)
testJobB.create(group=testJobGroup)
testJobC.create(group=testJobGroup)
bogusJob.create(group=bogusJobGroup)
testJobA["outcome"] = "success"
testJobB["outcome"] = "failure"
testJobC["outcome"] = "success"
testJobA.save()
testJobB.save()
testJobC.save()
testJobA.completeInputFiles()
compFiles = len(testSubscription.filesOfStatus("Completed"))
assert compFiles == 0, \
"Error: test sub has wrong number of complete files: %s" % compFiles
testJobB["outcome"] = "success"
testJobB.save()
testJobB.completeInputFiles(skipFiles=[testFileB["lfn"]])
availFiles = len(testSubscription.filesOfStatus("Available"))
assert availFiles == 0, \
"Error: test sub has wrong number of available files: %s" % availFiles
acqFiles = len(testSubscription.filesOfStatus("Acquired"))
assert acqFiles == 1, \
"Error: test sub has wrong number of acquired files: %s" % acqFiles
compFiles = len(testSubscription.filesOfStatus("Completed"))
assert compFiles == 1, \
"Error: test sub has wrong number of complete files: %s" % compFiles
failFiles = len(testSubscription.filesOfStatus("Failed"))
assert failFiles == 1, \
"Error: test sub has wrong number of failed files: %s" % failFiles
availFiles = len(bogusSubscription.filesOfStatus("Available"))
assert availFiles == 0, \
"Error: test sub has wrong number of available files: %s" % availFiles
acqFiles = len(bogusSubscription.filesOfStatus("Acquired"))
assert acqFiles == 3, \
"Error: test sub has wrong number of acquired files: %s" % acqFiles
compFiles = len(bogusSubscription.filesOfStatus("Completed"))
assert compFiles == 0, \
"Error: test sub has wrong number of complete files: %s" % compFiles
failFiles = len(bogusSubscription.filesOfStatus("Failed"))
assert failFiles == 0, \
"Error: test sub has wrong number of failed files: %s" % failFiles
return
def testJobTypeDAO(self):
"""
_testJobTypeDAO_
Verify that the Jobs.GetType DAO returns the correct job type. The
job type is retrieved from the subscription type. When only a single
job is passed.
"""
testJob = self.createTestJob()
jobTypeAction = self.daoFactory(classname="Jobs.GetType")
jobType = jobTypeAction.execute(jobID=testJob["id"])
assert jobType == "Merge", \
"Error: GetJobType DAO returned the wrong job type."
return
def testJobTypeDAOBulk(self):
"""
_testJobTypeDAOBulk_
Verify that the Jobs.GetType DAO returns the correct job type. The
job type is retrieved from the subscription type. When a list of jobs
ids is passed.
"""
testJobA = self.createTestJob(subscriptionType="Merge")
testJobB = self.createTestJob(subscriptionType="Processing")
testJobC = self.createTestJob(subscriptionType="Production")
testJobD = self.createTestJob(subscriptionType="Merge")
testJobE = self.createTestJob(subscriptionType="Skim")
jobIds = []
jobIds.append(testJobA["id"])
jobIds.append(testJobB["id"])
jobIds.append(testJobC["id"])
jobIds.append(testJobD["id"])
jobIds.append(testJobE["id"])
jobTypeAction = self.daoFactory(classname="Jobs.GetType")
jobTypes = jobTypeAction.execute(jobID=jobIds)
entryMap = {}
for entry in jobTypes:
entryMap[entry["id"]] = entry["type"]
assert entryMap[testJobA["id"]] == "Merge", \
"Error: GetJobType DAO returned the wrong job type."
assert entryMap[testJobB["id"]] == "Processing", \
"Error: GetJobType DAO returned the wrong job type."
assert entryMap[testJobC["id"]] == "Production", \
"Error: GetJobType DAO returned the wrong job type."
assert entryMap[testJobD["id"]] == "Merge", \
"Error: GetJobType DAO returned the wrong job type."
assert entryMap[testJobE["id"]] == "Skim", \
"Error: GetJobType DAO returned the wrong job type."
return
def testGetOutputMapDAO(self):
"""
_testGetOutputMapDAO_
Verify the proper behavior of the GetOutputMapDAO for a variety of
different processing chains.
"""
recoOutputFileset = Fileset(name="RECO")
recoOutputFileset.create()
mergedRecoOutputFileset = Fileset(name="MergedRECO")
mergedRecoOutputFileset.create()
alcaOutputFileset = Fileset(name="ALCA")
alcaOutputFileset.create()
mergedAlcaOutputFileset = Fileset(name="MergedALCA")
mergedAlcaOutputFileset.create()
dqmOutputFileset = Fileset(name="DQM")
dqmOutputFileset.create()
mergedDqmOutputFileset = Fileset(name="MergedDQM")
mergedDqmOutputFileset.create()
cleanupFileset = Fileset(name="Cleanup")
cleanupFileset.create()
testWorkflow = Workflow(spec="wf001.xml",
owner="Steve",
name="TestWF",
task="None")
testWorkflow.create()
testWorkflow.addOutput("output", recoOutputFileset,
mergedRecoOutputFileset)
testWorkflow.addOutput("ALCARECOStreamCombined", alcaOutputFileset,
mergedAlcaOutputFileset)
testWorkflow.addOutput("DQM", dqmOutputFileset, mergedDqmOutputFileset)
testWorkflow.addOutput("output", cleanupFileset)
testWorkflow.addOutput("ALCARECOStreamCombined", cleanupFileset)
testWorkflow.addOutput("DQM", cleanupFileset)
testRecoMergeWorkflow = Workflow(spec="wf002.xml",
owner="Steve",
name="TestRecoMergeWF",
task="None")
testRecoMergeWorkflow.create()
testRecoMergeWorkflow.addOutput("anything", mergedRecoOutputFileset,
mergedRecoOutputFileset)
testRecoProcWorkflow = Workflow(spec="wf004.xml",
owner="Steve",
name="TestRecoProcWF",
task="None")
testRecoProcWorkflow.create()
testAlcaChildWorkflow = Workflow(spec="wf003.xml",
owner="Steve",
name="TestAlcaChildWF",
task="None")
testAlcaChildWorkflow.create()
inputFile = File(lfn="/path/to/some/lfn",
size=600000,
events=60000,
locations="cmssrm.fnal.gov")
inputFile.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testFileset.addFile(inputFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
split_algo="EventBased",
type="Processing")
testMergeRecoSubscription = Subscription(
fileset=recoOutputFileset,
workflow=testRecoMergeWorkflow,
split_algo="WMBSMergeBySize",
type="Merge")
testProcRecoSubscription = Subscription(fileset=recoOutputFileset,
workflow=testRecoProcWorkflow,
split_algo="FileBased",
type="Processing")
testChildAlcaSubscription = Subscription(
fileset=alcaOutputFileset,
workflow=testAlcaChildWorkflow,
split_algo="FileBased",
type="Processing")
testSubscription.create()
testMergeRecoSubscription.create()
testProcRecoSubscription.create()
testChildAlcaSubscription.create()
testSubscription.acquireFiles()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
testJob = Job(name="SplitJobA", files=[inputFile])
testJob.create(group=testJobGroup)
testJob["state"] = "complete"
testJob.save()
outputMapAction = self.daoFactory(classname="Jobs.GetOutputMap")
outputMap = outputMapAction.execute(jobID=testJob["id"])
assert len(outputMap.keys()) == 3, \
"Error: Wrong number of outputs for primary workflow."
goldenMap = {
"output": (recoOutputFileset.id, mergedRecoOutputFileset.id),
"ALCARECOStreamCombined":
(alcaOutputFileset.id, mergedAlcaOutputFileset.id),
"DQM": (dqmOutputFileset.id, mergedDqmOutputFileset.id)
}
for outputID in outputMap.keys():
for outputFilesets in outputMap[outputID]:
if outputFilesets["merged_output_fileset"] == None:
self.assertEqual(outputFilesets["output_fileset"],
cleanupFileset.id,
"Error: Cleanup fileset is wrong.")
continue
self.assertTrue(outputID in goldenMap.keys(),
"Error: Output identifier is missing.")
self.assertEqual(outputFilesets["output_fileset"],
goldenMap[outputID][0],
"Error: Output fileset is wrong.")
self.assertEqual(outputFilesets["merged_output_fileset"],
goldenMap[outputID][1],
"Error: Merged output fileset is wrong.")
del goldenMap[outputID]
self.assertEqual(len(goldenMap.keys()), 0,
"Error: Missing output maps.")
return
def testLocations(self):
"""
_testLocations_
Test setting and getting locations using DAO objects
"""
testJob = self.createTestJob()
jobGetLocation = self.daoFactory(classname="Jobs.GetLocation")
jobSetLocation = self.daoFactory(classname="Jobs.SetLocation")
result = jobGetLocation.execute(jobid=testJob['id'])
self.assertEqual(result, [['test.site.ch']])
jobSetLocation.execute(jobid=testJob['id'], location="test2.site.ch")
result = jobGetLocation.execute(jobid=testJob['id'])
self.assertEqual(result, [['test2.site.ch']])
testJob2 = self.createTestJob()
testJob3 = self.createTestJob()
binds = [{
'jobid': testJob['id']
}, {
'jobid': testJob2['id']
}, {
'jobid': testJob3['id']
}]
result = jobGetLocation.execute(jobid=binds)
self.assertEqual(result, \
[{'site_name': 'test2.site.ch', 'id': 1}, \
{'site_name': 'test.site.ch', 'id': 2}, \
{'site_name': 'test.site.ch', 'id': 3}])
return
def testGetDataStructsJob(self):
"""
_testGetDataStructsJob_
Test the ability to 'cast' as a DataStructs job type
"""
testJob = self.createTestJob()
testJob['test'] = 'ThisIsATest'
testJob.baggage.section_('TestSection')
testJob.baggage.TestSection.test = 100
finalJob = testJob.getDataStructsJob()
for key in finalJob.keys():
if key == 'input_files':
for file in testJob['input_files']:
self.assertEqual(
file.returnDataStructsFile()
in finalJob['input_files'], True)
continue
self.assertEqual(testJob[key], finalJob[key])
self.assertEqual(str(finalJob.__class__),
"<class 'WMCore.DataStructs.Job.Job'>")
self.assertEqual(str(finalJob["mask"].__class__),
"<class 'WMCore.DataStructs.Mask.Mask'>")
for key in testJob["mask"]:
self.assertEqual(testJob["mask"][key], finalJob["mask"][key],
"Error: The masks should be the same")
self.assertEqual(finalJob.baggage.TestSection.test, 100)
return
def testLoadOutputID(self):
"""
_testLoadOutputID_
Test whether we can load an output ID for a job
"""
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.create(group=testJobGroup)
self.assertEqual(testJob.loadOutputID(), testJobGroup.output.id)
return
def testLoadForTaskArchiver(self):
"""
_testLoadForTaskArchiver_
Tests the return of the DAO for the TaskArchiver
"""
#Create 2 jobs
jobA = self.createTestJob()
jobB = self.createTestJob()
#Put a mask in one
mask = Mask()
mask.addRunAndLumis(1, [45])
mask.save(jobA['id'])
#Execute the DAO
taskArchiverDAO = self.daoFactory(classname="Jobs.LoadForTaskArchiver")
jobs = taskArchiverDAO.execute([jobA['id'], jobB['id']])
#Sort the jobs and check the results, we care about id, input files and mask
jobs.sort(key=lambda x: x['id'])
jobAprime = jobs[0]
lfns = [x['lfn'] for x in jobAprime['input_files']]
self.assertTrue('/this/is/a/lfnA' in lfns,
'Missing LFN lfnA from the input files')
self.assertTrue('/this/is/a/lfnB' in lfns,
'Missing LFN lfnB from the input files')
for inFile in jobAprime['input_files']:
if inFile['lfn'] == '/this/is/a/lfnA':
run = inFile['runs'].pop()
self.assertEqual(run.run, 1, 'The run number is wrong')
self.assertEqual(run.lumis, [45], 'The lumis are wrong')
else:
run = inFile['runs'].pop()
self.assertEqual(run.run, 1, 'The run number is wrong')
self.assertEqual(run.lumis, [46], 'The lumis are wrong')
mask = jobAprime['mask']
self.assertEqual(mask['runAndLumis'], {1: [[45, 45]]},
"Wrong run and lumis in mask")
jobBprime = jobs[1]
for inFile in jobBprime['input_files']:
if inFile['lfn'] == '/this/is/a/lfnA':
run = inFile['runs'].pop()
self.assertEqual(run.run, 1, 'The run number is wrong')
self.assertEqual(run.lumis, [45], 'The lumis are wrong')
else:
run = inFile['runs'].pop()
self.assertEqual(run.run, 1, 'The run number is wrong')
self.assertEqual(run.lumis, [46], 'The lumis are wrong')
runs = []
for inputFile in jobBprime['input_files']:
runs.extend(inputFile.getRuns())
self.assertEqual(jobBprime['mask'].filterRunLumisByMask(runs=runs),
runs, "Wrong mask in jobB")
return
def testMask(self):
"""
_testMask_
Test the new mask setup
"""
testWorkflow = Workflow(spec="spec.xml",
owner="Steve",
name="wf001",
task="Test")
testWorkflow.create()
testFileset = Fileset(name="TestFileset")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow)
testSubscription.create()
testFileA = File(lfn=makeUUID(), locations="test.site.ch")
testFileB = File(lfn=makeUUID(), locations="test.site.ch")
testFileA.create()
testFileB.create()
testFileset.addFile([testFileA, testFileB])
testFileset.commit()
testSubscription.acquireFiles([testFileA, testFileB])
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
testJob = Job()
testJob['mask'].addRunAndLumis(run=100, lumis=[101, 102])
testJob['mask'].addRunAndLumis(run=200, lumis=[201, 202])
testJob.create(group=testJobGroup)
loadJob = Job(id=testJob.exists())
loadJob.loadData()
runs = loadJob['mask'].getRunAndLumis()
self.assertEqual(len(runs), 2)
self.assertEqual(runs[100], [[101, 102]])
self.assertEqual(runs[200], [[201, 202]])
bigRun = Run(100, *[101, 102, 103, 104])
badRun = Run(300, *[1001, 1002])
result = loadJob['mask'].filterRunLumisByMask([bigRun, badRun])
self.assertEqual(len(result), 1)
alteredRun = result.pop()
self.assertEqual(alteredRun.run, 100)
self.assertEqual(alteredRun.lumis, [101, 102])
run0 = Run(300, *[1001, 1002])
run1 = Run(300, *[1001, 1002])
loadJob['mask'].filterRunLumisByMask([run0, run1])
return
def 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
class DBFormatterTest(unittest.TestCase):
"""
_DBFormatterTest_
Unit tests for the DBFormatter class
"""
def setUp(self):
"make a logger instance and create tables"
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection(destroyAllDatabase=True)
self.testInit.setSchema(customModules=["WMQuality.TestDB"],
useDefault=False)
self.selectSQL = "SELECT * FROM test_tableb"
def tearDown(self):
"""
Delete the databases
"""
self.testInit.clearDatabase()
def stuffDB(self):
"""Populate one of the test tables"""
insertSQL = "INSERT INTO test_tableb (column1, column2, column3) values (:bind1, :bind2, :bind3)"
insertBinds = [{
'bind1': u'value1a',
'bind2': 1,
'bind3': u'value2a'
}, {
'bind1': 'value1b',
'bind2': 2,
'bind3': 'value2b'
}, {
'bind1': b'value1c',
'bind2': 3,
'bind3': b'value2d'
}]
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL, insertBinds)
def testBFormatting(self):
"""
Test various formats
"""
# fill the database with some initial data
self.stuffDB()
myThread = threading.currentThread()
dbformatter = DBFormatter(myThread.logger, myThread.dbi)
result = myThread.dbi.processData(self.selectSQL)
output = dbformatter.format(result)
self.assertEqual(output,
[['value1a', 1, 'value2a'], ['value1b', 2, 'value2b'],
['value1c', 3, 'value2d']])
result = myThread.dbi.processData(self.selectSQL)
output = dbformatter.formatOne(result)
print('test1 ' + str(output))
self.assertEqual(output, ['value1a', 1, 'value2a'])
result = myThread.dbi.processData(self.selectSQL)
output = dbformatter.formatDict(result)
self.assertEqual(output, [{
'column3': 'value2a',
'column2': 1,
'column1': 'value1a'
}, {
'column3': 'value2b',
'column2': 2,
'column1': 'value1b'
}, {
'column3': 'value2d',
'column2': 3,
'column1': 'value1c'
}])
result = myThread.dbi.processData(self.selectSQL)
output = dbformatter.formatOneDict(result)
self.assertEqual(output, {
'column3': 'value2a',
'column2': 1,
'column1': 'value1a'
})
class ResourceControlTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Install schema and create a DAO factory for WMBS.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=[
"WMCore.WMBS", "WMCore.ResourceControl", "WMCore.BossAir"
],
useDefault=False)
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
self.baDaoFactory = DAOFactory(package="WMCore.BossAir",
logger=myThread.logger,
dbinterface=myThread.dbi)
self.insertRunJob = self.baDaoFactory(classname="NewJobs")
self.insertState = self.baDaoFactory(classname="NewState")
states = ['PEND', 'RUN', 'Idle', 'Running']
self.insertState.execute(states)
self.tempDir = self.testInit.generateWorkDir()
return
def tearDown(self):
"""
_tearDown_
Clear the schema.
"""
self.testInit.clearDatabase()
return
def testInsert(self):
"""
_testInsert_
Verify that inserting sites and thresholds works correctly, even if the
site or threshold already exists.
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 10, 20, "testSE1", "testCE1")
myResourceControl.insertSite("testSite1", 10, 20, "testSE1", "testCE1")
myResourceControl.insertSite("testSite2", 100, 200, "testSE2",
"testCE2")
myResourceControl.insertThreshold("testSite1", "Processing", 20, 10)
myResourceControl.insertThreshold("testSite1", "Merge", 200, 100)
myResourceControl.insertThreshold("testSite1", "Merge", 250, 150)
myResourceControl.insertThreshold("testSite2", "Processing", 50, 30)
myResourceControl.insertThreshold("testSite2", "Merge", 135, 100)
createThresholds = myResourceControl.listThresholdsForCreate()
self.assertEqual(len(createThresholds.keys()), 2,
"Error: Wrong number of site in Resource Control DB")
self.assertTrue("testSite1" in createThresholds.keys(),
"Error: Test Site 1 missing from thresholds.")
self.assertTrue("testSite2" in createThresholds.keys(),
"Error: Test Site 2 missing from thresholds.")
self.assertEqual(createThresholds["testSite1"]["total_slots"], 10,
"Error: Wrong number of total slots.")
self.assertEqual(
createThresholds["testSite1"]["pending_jobs"], 0,
"Error: Wrong number of running jobs: %s" %
createThresholds["testSite1"]["pending_jobs"])
self.assertEqual(createThresholds["testSite2"]["total_slots"], 100,
"Error: Wrong number of total slots.")
self.assertEqual(createThresholds["testSite2"]["pending_jobs"], 0,
"Error: Wrong number of running jobs.")
thresholds = myResourceControl.listThresholdsForSubmit()
self.assertEqual(
len(thresholds.keys()), 2,
"Error: Wrong number of sites in Resource Control DB")
self.assertTrue("testSite1" in thresholds.keys(),
"Error: testSite1 missing from thresholds.")
self.assertTrue("testSite2" in thresholds.keys(),
"Error: testSite2 missing from thresholds.")
site1Info = thresholds["testSite1"]
site2Info = thresholds["testSite2"]
site1Thresholds = site1Info["thresholds"]
site2Thresholds = site2Info["thresholds"]
procThreshold1 = None
procThreshold2 = None
mergeThreshold1 = None
mergeThreshold2 = None
for threshold in site1Thresholds:
if threshold["task_type"] == "Merge":
mergeThreshold1 = threshold
elif threshold["task_type"] == "Processing":
procThreshold1 = threshold
for threshold in site2Thresholds:
if threshold["task_type"] == "Merge":
mergeThreshold2 = threshold
elif threshold["task_type"] == "Processing":
procThreshold2 = threshold
self.assertEqual(len(site1Thresholds), 2,
"Error: Wrong number of task types.")
self.assertEqual(len(site2Thresholds), 2,
"Error: Wrong number of task types.")
self.assertNotEqual(procThreshold1, None)
self.assertNotEqual(procThreshold2, None)
self.assertNotEqual(mergeThreshold1, None)
self.assertNotEqual(mergeThreshold2, None)
self.assertEqual(site1Info["total_pending_slots"], 10,
"Error: Site thresholds wrong")
self.assertEqual(site1Info["total_running_slots"], 20,
"Error: Site thresholds wrong")
self.assertEqual(site1Info["total_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(site1Info["total_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold1["task_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold1["task_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold1["max_slots"], 20,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold1["pending_slots"], 10,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold1["task_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold1["task_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold1["max_slots"], 250,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold1["pending_slots"], 150,
"Error: Site thresholds wrong")
self.assertEqual(site2Info["total_pending_slots"], 100,
"Error: Site thresholds wrong")
self.assertEqual(site2Info["total_running_slots"], 200,
"Error: Site thresholds wrong")
self.assertEqual(site2Info["total_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(site2Info["total_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold2["task_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold2["task_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold2["max_slots"], 50,
"Error: Site thresholds wrong")
self.assertEqual(procThreshold2["pending_slots"], 30,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold2["task_running_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold2["task_pending_jobs"], 0,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold2["max_slots"], 135,
"Error: Site thresholds wrong")
self.assertEqual(mergeThreshold2["pending_slots"], 100,
"Error: Site thresholds wrong")
def testList(self):
"""
_testList_
Test the functions that list thresholds for creating jobs and submitting
jobs.
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 10, 20, "testSE1", "testCE1",
"T1_US_FNAL", "LsfPlugin")
myResourceControl.insertSite("testSite2", 20, 40, "testSE2", "testCE2")
myResourceControl.insertThreshold("testSite1", "Processing", 20, 10)
myResourceControl.insertThreshold("testSite1", "Merge", 200, 100)
myResourceControl.insertThreshold("testSite2", "Processing", 50, 25)
myResourceControl.insertThreshold("testSite2", "Merge", 135, 65)
testWorkflow = Workflow(spec=makeUUID(),
owner="Steve",
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.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"], "testSite1")
setLocationAction.execute(testJobG["id"], "testSite1")
setLocationAction.execute(testJobH["id"], "testSite1")
createThresholds = myResourceControl.listThresholdsForCreate()
submitThresholds = myResourceControl.listThresholdsForSubmit()
self.assertEqual(len(createThresholds.keys()), 2,
"Error: Wrong number of sites in create thresholds")
self.assertEqual(createThresholds["testSite1"]["total_slots"], 10,
"Error: Wrong number of slots for site 1")
self.assertEqual(createThresholds["testSite2"]["total_slots"], 20,
"Error: Wrong number of slots for site 2")
# We should have two running jobs with locations at site one,
# two running jobs without locations at site two, and one running
# job without a location at site one and two.
self.assertEqual(createThresholds["testSite1"]["pending_jobs"], 4,
"Error: Wrong number of pending jobs for site 1")
# We should have one running job with a location at site 2 and
# another running job without a location.
self.assertEqual(createThresholds["testSite2"]["pending_jobs"], 2,
"Error: Wrong number of pending jobs for site 2")
# We should also have a phedex_name
self.assertEqual(createThresholds["testSite1"]["cms_name"],
"T1_US_FNAL")
self.assertEqual(createThresholds["testSite2"]["cms_name"], None)
mergeThreshold1 = None
mergeThreshold2 = None
procThreshold1 = None
procThreshold2 = None
self.assertEqual(submitThresholds["testSite1"]['cms_name'],
'T1_US_FNAL')
for threshold in submitThresholds["testSite1"]["thresholds"]:
if threshold['task_type'] == "Merge":
mergeThreshold1 = threshold
elif threshold['task_type'] == "Processing":
procThreshold1 = threshold
self.assertEqual(submitThresholds["testSite2"]['cms_name'], None)
for threshold in submitThresholds["testSite2"]["thresholds"]:
if threshold['task_type'] == "Merge":
mergeThreshold2 = threshold
elif threshold['task_type'] == "Processing":
procThreshold2 = threshold
self.assertEqual(
submitThresholds["testSite1"]["total_running_jobs"], 1,
"Error: Wrong number of running jobs for submit thresholds.")
self.assertEqual(
submitThresholds["testSite2"]["total_running_jobs"], 0,
"Error: Wrong number of running jobs for submit thresholds.")
self.assertEqual(
submitThresholds["testSite1"]["total_pending_jobs"], 1,
"Error: Wrong number of pending jobs for submit thresholds.")
self.assertEqual(
submitThresholds["testSite2"]["total_pending_jobs"], 0,
"Error: Wrong number of pending jobs for submit thresholds.")
self.assertEqual(
mergeThreshold1["task_running_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
mergeThreshold1["task_pending_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
procThreshold1["task_running_jobs"], 1,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
procThreshold1["task_pending_jobs"], 1,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
mergeThreshold2["task_running_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
mergeThreshold2["task_pending_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
procThreshold2["task_running_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
self.assertEqual(
procThreshold2["task_pending_jobs"], 0,
"Error: Wrong number of task running jobs for submit thresholds.")
return
def testListSiteInfo(self):
"""
_testListSiteInfo_
Verify that the listSiteInfo() methods works properly.
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 10, 20, "testSE1", "testCE1")
myResourceControl.insertSite("testSite2", 100, 200, "testSE2",
"testCE2")
siteInfo = myResourceControl.listSiteInfo("testSite1")
self.assertEqual(siteInfo["site_name"], "testSite1",
"Error: Site name is wrong.")
self.assertEqual(siteInfo["se_name"], ["testSE1"],
"Error: SE name is wrong.")
self.assertEqual(siteInfo["ce_name"], "testCE1",
"Error: CE name is wrong.")
self.assertEqual(siteInfo["pending_slots"], 10,
"Error: Pending slots is wrong.")
self.assertEqual(siteInfo["running_slots"], 20,
"Error: Pending slots is wrong.")
return
def testUpdateJobSlots(self):
"""
_testUpdateJobSlots_
Verify that it is possible to update the number of job slots at a site.
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 10, 20, "testSE1", "testCE1")
siteInfo = myResourceControl.listSiteInfo("testSite1")
self.assertEqual(siteInfo["pending_slots"], 10,
"Error: Pending slots is wrong.")
self.assertEqual(siteInfo["running_slots"], 20,
"Error: Running slots is wrong.")
myResourceControl.setJobSlotsForSite("testSite1", pendingJobSlots=20)
siteInfo = myResourceControl.listSiteInfo("testSite1")
self.assertEqual(siteInfo["pending_slots"], 20,
"Error: Pending slots is wrong.")
myResourceControl.setJobSlotsForSite("testSite1", runningJobSlots=40)
siteInfo = myResourceControl.listSiteInfo("testSite1")
self.assertEqual(siteInfo["running_slots"], 40,
"Error: Running slots is wrong.")
myResourceControl.setJobSlotsForSite("testSite1", 5, 10)
siteInfo = myResourceControl.listSiteInfo("testSite1")
self.assertEqual(siteInfo["pending_slots"], 5,
"Error: Pending slots is wrong.")
self.assertEqual(siteInfo["running_slots"], 10,
"Error: Running slots is wrong.")
return
def testThresholdsForSite(self):
"""
_testThresholdsForSite_
Check that we can get the thresholds in intelligible form
for each site
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 20, 40, "testSE1", "testCE1")
myResourceControl.insertThreshold("testSite1", "Processing", 10, 8)
myResourceControl.insertThreshold("testSite1", "Merge", 5, 3)
result = myResourceControl.thresholdBySite(siteName="testSite1")
procInfo = {}
mergInfo = {}
for res in result:
if res['task_type'] == 'Processing':
procInfo = res
elif res['task_type'] == 'Merge':
mergInfo = res
self.assertEqual(procInfo.get('pending_slots', None), 20)
self.assertEqual(procInfo.get('running_slots', None), 40)
self.assertEqual(procInfo.get('max_slots', None), 10)
self.assertEqual(procInfo.get('task_pending_slots', None), 8)
self.assertEqual(mergInfo.get('pending_slots', None), 20)
self.assertEqual(mergInfo.get('running_slots', None), 40)
self.assertEqual(mergInfo.get('max_slots', None), 5)
self.assertEqual(mergInfo.get('task_pending_slots', None), 3)
return
def testThresholdPriority(self):
"""
_testThresholdPriority_
Test that we get things back in priority order
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 20, 40, "testSE1", "testCE1")
myResourceControl.insertThreshold("testSite1",
"Processing",
10,
8,
priority=1)
myResourceControl.insertThreshold("testSite1",
"Merge",
5,
3,
priority=2)
result = myResourceControl.listThresholdsForSubmit()
self.assertEqual(result['testSite1']['thresholds'][0]['task_type'],
'Merge')
self.assertEqual(result['testSite1']['thresholds'][1]['task_type'],
'Processing')
myResourceControl.insertThreshold("testSite1",
"Processing",
10,
8,
priority=2)
myResourceControl.insertThreshold("testSite1",
"Merge",
5,
3,
priority=1)
# Should now be in reverse order
result = myResourceControl.listThresholdsForSubmit()
self.assertEqual(result['testSite1']['thresholds'][1]['task_type'],
'Merge')
self.assertEqual(result['testSite1']['thresholds'][0]['task_type'],
'Processing')
myResourceControl.insertSite("testSite2", 20, 40, "testSE2", "testCE2")
myResourceControl.insertThreshold("testSite2",
"Processing",
10,
8,
priority=1)
myResourceControl.insertThreshold("testSite2",
"Merge",
5,
3,
priority=2)
# Should be in proper order for site 2
result = myResourceControl.listThresholdsForSubmit()
self.assertEqual(result['testSite2']['thresholds'][0]['task_type'],
'Merge')
self.assertEqual(result['testSite2']['thresholds'][1]['task_type'],
'Processing')
# Should now be in reverse order for site 1
self.assertEqual(result['testSite1']['thresholds'][1]['task_type'],
'Merge')
self.assertEqual(result['testSite1']['thresholds'][0]['task_type'],
'Processing')
myResourceControl.insertThreshold("testSite2", "Merge", 20, 10)
result = myResourceControl.listThresholdsForSubmit()
self.assertEqual(result['testSite2']['thresholds'][0]['priority'], 2)
return
def testChangeState(self):
"""
_testChangeState_
Check that we can change the state between different values and
retrieve it through the threshold methods
"""
myResourceControl = ResourceControl()
myResourceControl.insertSite("testSite1", 20, 40, "testSE1", "testCE1")
myResourceControl.insertThreshold("testSite1",
"Processing",
10,
5,
priority=1)
result = myResourceControl.listThresholdsForCreate()
self.assertEqual(result['testSite1']['state'], 'Normal',
'Error: Wrong site state')
myResourceControl.changeSiteState("testSite1", "Down")
result = myResourceControl.listThresholdsForCreate()
self.assertEqual(result['testSite1']['state'], 'Down',
'Error: Wrong site state')
def createConfig(self):
"""
_createConfig_
Create a config and save it to the temp dir. Set the WMAGENT_CONFIG
environment variable so the config gets picked up.
"""
config = Configuration()
config.section_("General")
config.General.workDir = os.getenv("TESTDIR", os.getcwd())
config.section_("Agent")
config.Agent.componentName = "resource_control_t"
config.section_("CoreDatabase")
config.CoreDatabase.connectUrl = os.getenv("DATABASE")
config.CoreDatabase.socket = os.getenv("DBSOCK")
configHandle = open(os.path.join(self.tempDir, "config.py"), "w")
configHandle.write(str(config))
configHandle.close()
os.environ["WMAGENT_CONFIG"] = os.path.join(self.tempDir, "config.py")
return
def testInsertAllSEs(self):
"""
_testInsertAllSEs_
Test to see if we can insert all SEs and Thresholds at once
Depending on the WMCore.Services.SiteDB interface
"""
self.createConfig()
resControlPath = os.path.join(WMCore.WMBase.getTestBase(),
"../../bin/wmagent-resource-control")
env = os.environ
env['PYTHONPATH'] = ":".join(sys.path)
cmdline = [
resControlPath, "--add-all-sites", "--plugin=CondorPlugin",
"--pending-slots=100", "--running-slots=500"
]
retval = subprocess.Popen(cmdline,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
env=env)
(output, _) = retval.communicate()
myResourceControl = ResourceControl()
result = myResourceControl.listThresholdsForSubmit()
self.assertTrue('T1_US_FNAL' in result.keys())
for x in result.keys():
self.assertEqual(len(result[x]['thresholds']), 8)
self.assertEqual(result[x]['total_pending_slots'], 100)
self.assertEqual(result[x]['total_running_slots'], 500)
for thresh in result[x]['thresholds']:
if thresh['task_type'] == 'Processing':
self.assertEqual(thresh['priority'], 1)
self.assertEqual(thresh['max_slots'], 500)
# Verify that sites with more than one SE were added correctly.
nebInfo = myResourceControl.listSiteInfo("T2_US_Nebraska")
self.assertTrue(len(nebInfo["se_name"]) == 3)
return
def testInsertAllSEs2(self):
"""
_testInsertAllSEs2_
Test to see if we can insert all SEs and Thresholds at once
Depending on the WMCore.Services.SiteDB interface
"""
myResourceControl = ResourceControl()
taskList = [{
'taskType': 'Processing',
'maxSlots': 100,
'pendingSlots': 80,
'priority': 1
}, {
'taskType': 'Merge',
'maxSlots': 50,
'pendingSlots': 30,
'priority': 2
}]
myResourceControl.insertAllSEs(siteName='test',
pendingSlots=200,
runningSlots=400,
ceName='glidein-ce.fnal.gov',
plugin='CondorPlugin',
taskList=taskList)
result = myResourceControl.listThresholdsForSubmit()
self.assertTrue('test_cmssrm.fnal.gov' in result.keys())
self.assertEqual(result['test_cmssrm.fnal.gov']['cms_name'],
'T1_US_FNAL')
for x in result.keys():
self.assertEqual(len(result[x]['thresholds']), 2)
self.assertEqual(result[x]['total_pending_slots'], 200)
self.assertEqual(result[x]['total_running_slots'], 400)
for thresh in result[x]['thresholds']:
if thresh['task_type'] == 'Processing':
self.assertEqual(thresh['priority'], 1)
self.assertEqual(thresh['max_slots'], 100)
self.assertEqual(thresh['pending_slots'], 80)
else:
self.assertEqual(thresh['priority'], 2)
self.assertEqual(thresh['max_slots'], 50)
self.assertEqual(thresh['pending_slots'], 30)
return
def testInsertT0(self):
"""
_testInsertT0_
Test to see if we can insert the Tier-0 alone
with a single option
"""
self.createConfig()
resControlPath = os.path.join(WMCore.WMBase.getTestBase(),
"../../bin/wmagent-resource-control")
env = os.environ
env['PYTHONPATH'] = ":".join(sys.path)
cmdline = [resControlPath, "--add-T0"]
retval = subprocess.Popen(cmdline,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
env=env)
(_, _) = retval.communicate()
myResourceControl = ResourceControl()
result = myResourceControl.listThresholdsForSubmit()
self.assertTrue(len(result), 1)
self.assertTrue('CERN' in result)
for x in result:
self.assertEqual(len(result[x]['thresholds']), 10)
self.assertEqual(result[x]['total_pending_slots'], 500)
self.assertEqual(result[x]['total_running_slots'], -1)
for thresh in result[x]['thresholds']:
if thresh['task_type'] == 'Processing':
self.assertEqual(thresh['priority'], 1)
self.assertEqual(thresh['max_slots'], -1)
# Verify that sites with more than one SE were added correctly.
cernInfo = myResourceControl.listSiteInfo("CERN")
self.assertTrue(len(cernInfo["se_name"]) == 2)
return
class CursorLeakTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all of the
WMBS tables. Also add some dummy locations.
"""
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="se1.cern.ch")
locationAction.execute(siteName="se1.fnal.gov")
return
def tearDown(self):
"""
_tearDown_
Drop all the WMBS tables.
"""
self.testInit.clearDatabase()
def testCursor(self):
"""
_testCursor_
test the cursor closing is really affected
create 100 files with 5 parents and loop 100 times.
If the cursors are exhausted will crash.?
TODO: improve for more effective testing.
"""
raise nose.SkipTest
fileList = []
parentFile = None
for i in range(100):
testFile = File(lfn="/this/is/a/lfn%s" % i,
size=1024,
events=10,
checksums={"cksum": "1"})
testFile.addRun(Run(1, *[i]))
testFile.create()
for j in range(5):
parentFile = File(lfn="/this/is/a/lfnP%s" % j,
size=1024,
events=10,
checksums={"cksum": "1"})
parentFile.addRun(Run(1, *[j]))
parentFile.create()
testFile.addParent(parentFile['lfn'])
fileList.append(testFile)
for i in range(100):
for file in fileList:
file.loadData()
file.getAncestors(level=2)
file.getAncestors(level=2, type="lfn")
return
def testLotsOfAncestors(self):
"""
_testLotsOfAncestors_
Create a file with 15 parents with each parent having 100 parents to
verify that the query to return grandparents works correctly.
"""
raise nose.SkipTest
testFileA = File(lfn="/this/is/a/lfnA",
size=1024,
events=10,
checksums={"cksum": "1"},
locations="se1.fnal.gov")
testFileA.create()
for i in xrange(15):
testParent = File(lfn=makeUUID(),
size=1024,
events=10,
checksums={"cksum": "1"},
locations="se1.fnal.gov")
testParent.create()
testFileA.addParent(testParent["lfn"])
for i in xrange(100):
testGParent = File(lfn=makeUUID(),
size=1024,
events=10,
checksums={"cksum": "1"},
locations="se1.fnal.gov")
testGParent.create()
testParent.addParent(testGParent["lfn"])
assert len(testFileA.getAncestors(level=2, type="lfn")) == 1500, \
"ERROR: Incorrect grand parents returned"
return
class ExpressMergeTest(unittest.TestCase):
"""
_ExpressMergeTest_
Test for ExpressMerge job splitter
"""
def setUp(self):
"""
_setUp_
"""
import WMQuality.TestInit
WMQuality.TestInit.deleteDatabaseAfterEveryTest("I'm Serious")
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(
customModules=["WMComponent.DBS3Buffer", "T0.WMBS"])
self.splitterFactory = SplitterFactory(package="T0.JobSplitting")
myThread = threading.currentThread()
daoFactory = DAOFactory(package="T0.WMBS",
logger=logging,
dbinterface=myThread.dbi)
myThread.dbi.processData("""INSERT INTO wmbs_location
(id, site_name, state, state_time)
VALUES (1, 'SomeSite', 1, 1)
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_pnns
(id, pnn)
VALUES (2, 'SomePNN')
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_location_pnns
(location, pnn)
VALUES (1, 2)
""",
transaction=False)
insertRunDAO = daoFactory(classname="RunConfig.InsertRun")
insertRunDAO.execute(binds={
'RUN': 1,
'HLTKEY': "someHLTKey"
},
transaction=False)
insertLumiDAO = daoFactory(classname="RunConfig.InsertLumiSection")
for lumi in range(1, 5):
insertLumiDAO.execute(binds={
'RUN': 1,
'LUMI': lumi
},
transaction=False)
insertStreamDAO = daoFactory(classname="RunConfig.InsertStream")
insertStreamDAO.execute(binds={'STREAM': "Express"}, transaction=False)
insertStreamFilesetDAO = daoFactory(
classname="RunConfig.InsertStreamFileset")
insertStreamFilesetDAO.execute(1, "Express", "TestFileset1")
fileset1 = Fileset(name="TestFileset1")
self.fileset2 = Fileset(name="TestFileset2")
fileset1.load()
self.fileset2.create()
workflow1 = Workflow(spec="spec.xml",
owner="hufnagel",
name="TestWorkflow1",
task="Test")
workflow2 = Workflow(spec="spec.xml",
owner="hufnagel",
name="TestWorkflow2",
task="Test")
workflow1.create()
workflow2.create()
self.subscription1 = Subscription(fileset=fileset1,
workflow=workflow1,
split_algo="Express",
type="Express")
self.subscription2 = Subscription(fileset=self.fileset2,
workflow=workflow2,
split_algo="ExpressMerge",
type="ExpressMerge")
self.subscription1.create()
self.subscription2.create()
myThread.dbi.processData("""INSERT INTO wmbs_workflow_output
(WORKFLOW_ID, OUTPUT_IDENTIFIER, OUTPUT_FILESET)
VALUES (%d, 'SOMEOUTPUT', %d)
""" % (workflow1.id, self.fileset2.id),
transaction=False)
# keep for later
self.insertSplitLumisDAO = daoFactory(
classname="JobSplitting.InsertSplitLumis")
# default split parameters
self.splitArgs = {}
self.splitArgs['maxInputSize'] = 2 * 1024 * 1024 * 1024
self.splitArgs['maxInputFiles'] = 500,
self.splitArgs['maxLatency'] = 15 * 23
return
def tearDown(self):
"""
_tearDown_
"""
self.testInit.clearDatabase()
return
def deleteSplitLumis(self):
"""
_deleteSplitLumis_
"""
myThread = threading.currentThread()
myThread.dbi.processData("""DELETE FROM lumi_section_split_active
""",
transaction=False)
return
def test00(self):
"""
_test00_
Test that the job name prefix feature works
Test latency trigger (wait and 0)
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxLatency'] = 0
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertTrue(job['name'].startswith("ExpressMerge-"),
"ERROR: Job has wrong name")
return
def test01(self):
"""
_test01_
Test size and event triggers for single lumis (they are ignored)
Test latency trigger (timed out)
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
mySplitArgs['maxInputSize'] = 1
mySplitArgs['maxInputFiles'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
time.sleep(1)
mySplitArgs['maxLatency'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
return
def test02(self):
"""
_test02_
Test input files threshold on multi lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
mySplitArgs['maxInputFiles'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
time.sleep(1)
mySplitArgs['maxLatency'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
return
def test03(self):
"""
_test03_
Test input size threshold on multi lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
mySplitArgs['maxInputSize'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
time.sleep(1)
mySplitArgs['maxLatency'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
return
def test04(self):
"""
_test04_
Test multi lumis express merges
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
time.sleep(1)
mySplitArgs['maxLatency'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
return
def test05(self):
"""
_test05_
Test multi lumis express merges with holes
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 4]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
time.sleep(1)
mySplitArgs['maxLatency'] = 1
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
return
def test06(self):
"""
_test06_
Test active split lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1]:
for i in range(2):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave=False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription2)
self.insertSplitLumisDAO.execute(binds={
'SUB': self.subscription1['id'],
'LUMI': 1,
'NFILES': 5
})
mySplitArgs['maxLatency'] = 0
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
self.deleteSplitLumis()
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
return
class PhEDExInjectorPollerTest(unittest.TestCase):
"""
_PhEDExInjectorPollerTest_
Unit tests for the PhEDExInjector. Create some database inside DBSBuffer
and then have the PhEDExInjector upload the data to PhEDEx. Pull the data
back down and verify that everything is complete.
"""
def setUp(self):
"""
_setUp_
Install the DBSBuffer schema into the database and connect to PhEDEx.
"""
self.phedexURL = "https://cmsweb.cern.ch/phedex/datasvc/json/test"
self.dbsURL = "http://vocms09.cern.ch:8880/cms_dbs_int_local_yy_writer/servlet/DBSServlet"
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection(destroyAllDatabase=True)
self.testInit.setSchema(customModules=["WMComponent.DBS3Buffer"],
useDefault=False)
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMComponent.DBSBuffer.Database",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationAction = daofactory(classname="DBSBufferFiles.AddLocation")
locationAction.execute(siteName="srm-cms.cern.ch")
locationAction.execute(siteName="se.fnal.gov")
self.testFilesA = []
self.testFilesB = []
self.testDatasetA = "/%s/PromptReco-v1/RECO" % makeUUID()
self.testDatasetB = "/%s/CRUZET11-v1/RAW" % makeUUID()
self.phedex = PhEDEx({"endpoint": self.phedexURL}, "json")
return
def tearDown(self):
"""
_tearDown_
Delete the database.
"""
self.testInit.clearDatabase()
def stuffDatabase(self, spec="TestWorkload.pkl"):
"""
_stuffDatabase_
Fill the dbsbuffer with some files and blocks. We'll insert a total
of 5 files spanning two blocks. There will be a total of two datasets
inserted into the datbase.
We'll inject files with the location set as an SE name as well as a
PhEDEx node name as well.
"""
myThread = threading.currentThread()
buffer3Factory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
insertWorkflow = buffer3Factory(classname="InsertWorkflow")
insertWorkflow.execute("BogusRequest", "BogusTask", 0, 0, 0, 0)
checksums = {"adler32": "1234", "cksum": "5678"}
testFileA = DBSBufferFile(lfn=makeUUID(),
size=1024,
events=10,
checksums=checksums,
locations=set(["srm-cms.cern.ch"]))
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath(self.testDatasetA)
testFileA.addRun(Run(2, *[45]))
testFileA.create()
testFileB = DBSBufferFile(lfn=makeUUID(),
size=1024,
events=10,
checksums=checksums,
locations=set(["srm-cms.cern.ch"]))
testFileB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileB.setDatasetPath(self.testDatasetA)
testFileB.addRun(Run(2, *[45]))
testFileB.create()
testFileC = DBSBufferFile(lfn=makeUUID(),
size=1024,
events=10,
checksums=checksums,
locations=set(["srm-cms.cern.ch"]))
testFileC.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileC.setDatasetPath(self.testDatasetA)
testFileC.addRun(Run(2, *[45]))
testFileC.create()
self.testFilesA.append(testFileA)
self.testFilesA.append(testFileB)
self.testFilesA.append(testFileC)
testFileD = DBSBufferFile(lfn=makeUUID(),
size=1024,
events=10,
checksums=checksums,
locations=set(["srm-cms.cern.ch"]))
testFileD.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileD.setDatasetPath(self.testDatasetB)
testFileD.addRun(Run(2, *[45]))
testFileD.create()
testFileE = DBSBufferFile(lfn=makeUUID(),
size=1024,
events=10,
checksums=checksums,
locations=set(["srm-cms.cern.ch"]))
testFileE.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileE.setDatasetPath(self.testDatasetB)
testFileE.addRun(Run(2, *[45]))
testFileE.create()
self.testFilesB.append(testFileD)
self.testFilesB.append(testFileE)
uploadFactory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
datasetAction = uploadFactory(classname="NewDataset")
createAction = uploadFactory(classname="CreateBlocks")
datasetAction.execute(datasetPath=self.testDatasetA)
datasetAction.execute(datasetPath=self.testDatasetB)
self.blockAName = self.testDatasetA + "#" + makeUUID()
self.blockBName = self.testDatasetB + "#" + makeUUID()
newBlockA = DBSBlock(name=self.blockAName,
location="srm-cms.cern.ch",
das=None,
workflow=None)
newBlockA.setDataset(self.testDatasetA, 'data', 'VALID')
newBlockA.status = 'Closed'
newBlockB = DBSBlock(name=self.blockBName,
location="srm-cms.cern.ch",
das=None,
workflow=None)
newBlockB.setDataset(self.testDatasetB, 'data', 'VALID')
newBlockB.status = 'Closed'
createAction.execute(blocks=[newBlockA, newBlockB])
bufferFactory = DAOFactory(package="WMComponent.DBSBuffer.Database",
logger=myThread.logger,
dbinterface=myThread.dbi)
setBlock = bufferFactory(classname="DBSBufferFiles.SetBlock")
setBlock.execute(testFileA["lfn"], self.blockAName)
setBlock.execute(testFileB["lfn"], self.blockAName)
setBlock.execute(testFileC["lfn"], self.blockAName)
setBlock.execute(testFileD["lfn"], self.blockBName)
setBlock.execute(testFileE["lfn"], self.blockBName)
fileStatus = bufferFactory(classname="DBSBufferFiles.SetStatus")
fileStatus.execute(testFileA["lfn"], "LOCAL")
fileStatus.execute(testFileB["lfn"], "LOCAL")
fileStatus.execute(testFileC["lfn"], "LOCAL")
fileStatus.execute(testFileD["lfn"], "LOCAL")
fileStatus.execute(testFileE["lfn"], "LOCAL")
associateWorkflow = buffer3Factory(
classname="DBSBufferFiles.AssociateWorkflowToFile")
associateWorkflow.execute(testFileA["lfn"], "BogusRequest",
"BogusTask")
associateWorkflow.execute(testFileB["lfn"], "BogusRequest",
"BogusTask")
associateWorkflow.execute(testFileC["lfn"], "BogusRequest",
"BogusTask")
associateWorkflow.execute(testFileD["lfn"], "BogusRequest",
"BogusTask")
associateWorkflow.execute(testFileE["lfn"], "BogusRequest",
"BogusTask")
return
def createConfig(self):
"""
_createConfig_
Create a config for the PhEDExInjector with paths to the test DBS and
PhEDEx instances.
"""
config = self.testInit.getConfiguration()
config.component_("DBSInterface")
config.DBSInterface.globalDBSUrl = self.dbsURL
config.component_("PhEDExInjector")
config.PhEDExInjector.phedexurl = self.phedexURL
config.PhEDExInjector.subscribeMSS = True
config.PhEDExInjector.group = "Saturn"
config.PhEDExInjector.pollInterval = 30
config.PhEDExInjector.subscribeInterval = 60
return config
def retrieveReplicaInfoForBlock(self, blockName):
"""
_retrieveReplicaInfoForBlock_
Retrieve the replica information for a block. It takes several minutes
after a block is injected for the statistics to be calculated, so this
will block until that information is available.
"""
attempts = 0
while attempts < 15:
result = self.phedex.getReplicaInfoForFiles(block=blockName)
if "phedex" in result:
if "block" in result["phedex"]:
if len(result["phedex"]["block"]) != 0:
return result["phedex"]["block"][0]
attempts += 1
time.sleep(20)
logging.info("Could not retrieve replica info for block: %s" %
blockName)
return None
@attr("integration")
def testPoller(self):
"""
_testPoller_
Stuff the database and have the poller upload files to PhEDEx. Retrieve
replica information for the uploaded blocks and verify that all files
have been injected.
"""
return
self.stuffDatabase()
poller = PhEDExInjectorPoller(self.createConfig())
poller.setup(parameters=None)
poller.algorithm(parameters=None)
replicaInfo = self.retrieveReplicaInfoForBlock(self.blockAName)
goldenLFNs = []
for file in self.testFilesA:
goldenLFNs.append(file["lfn"])
for replicaFile in replicaInfo["file"]:
assert replicaFile["name"] in goldenLFNs, \
"Error: Extra file in replica block: %s" % replicaFile["name"]
goldenLFNs.remove(replicaFile["name"])
assert len(goldenLFNs) == 0, \
"Error: Files missing from PhEDEx replica: %s" % goldenLFNs
replicaInfo = self.retrieveReplicaInfoForBlock(self.blockBName)
goldenLFNs = []
for file in self.testFilesB:
goldenLFNs.append(file["lfn"])
for replicaFile in replicaInfo["file"]:
assert replicaFile["name"] in goldenLFNs, \
"Error: Extra file in replica block: %s" % replicaFile["name"]
goldenLFNs.remove(replicaFile["name"])
assert len(goldenLFNs) == 0, \
"Error: Files missing from PhEDEx replica: %s" % goldenLFNs
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMComponent.DBSUpload.Database",
logger=myThread.logger,
dbinterface=myThread.dbi)
setBlock = daofactory(classname="SetBlockStatus")
setBlock.execute(self.blockAName,
locations=None,
open_status="InGlobalDBS")
poller.algorithm(parameters=None)
replicaInfo = self.retrieveReplicaInfoForBlock(self.blockAName)
assert replicaInfo["is_open"] == "n", \
"Error: block should be closed."
replicaInfo = self.retrieveReplicaInfoForBlock(self.blockBName)
assert replicaInfo["is_open"] == "y", \
"Error: block should be open."
return
def test_CustodialSiteA(self):
"""
_CustodialSiteA_
Check the custodialSite stuff by DAO, since I don't have a cert
First make sure we properly handle having no custodialSite
"""
self.stuffDatabase()
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMComponent.PhEDExInjector.Database",
logger=myThread.logger,
dbinterface=myThread.dbi)
getUninjected = daofactory(classname="GetUninjectedFiles")
uninjectedFiles = getUninjected.execute()
self.assertEqual(uninjectedFiles.keys(), ['srm-cms.cern.ch'])
return
class RunJobTest(unittest.TestCase):
"""
_RunJobTest_
Test the RunJob object and accessors
"""
def setUp(self):
myThread = threading.currentThread()
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
#self.tearDown()
self.testInit.setSchema(customModules=[
"WMCore.WMBS", "WMCore.BossAir", "WMCore.ResourceControl",
"WMCore.Agent.Database"
],
useDefault=False)
self.daoFactory = DAOFactory(package="WMCore.BossAir",
logger=myThread.logger,
dbinterface=myThread.dbi)
resourceControl = ResourceControl()
resourceControl.insertSite(siteName='Xanadu',
pnn='se.Xanadu',
ceName='Xanadu',
plugin="TestPlugin")
resourceControl.insertThreshold(siteName = 'Xanadu', taskType = 'Processing', \
maxSlots = 10000, pendingSlots = 10000)
# Create user
wmbsFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
newuser = wmbsFactory(classname="Users.New")
newuser.execute(dn="mnorman",
group_name="phgroup",
role_name="cmsrole")
if PY3:
self.assertItemsEqual = self.assertCountEqual
def tearDown(self):
"""
Database deletion
"""
self.testInit.clearDatabase(modules=[
"WMCore.WMBS", "WMCore.BossAir", "WMCore.ResourceControl",
"WMCore.Agent.Database"
])
self.testInit.delWorkDir()
return
def createJobs(self, nJobs):
"""
Creates a series of jobGroups for submissions
"""
testWorkflow = Workflow(spec="dummy",
owner="mnorman",
name="dummy",
task="basicWorkload/Production")
testWorkflow.create()
# Create Fileset, Subscription, jobGroup
testFileset = Fileset(name="dummy")
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=testWorkflow,
type="Processing",
split_algo="FileBased")
testSubscription.create()
testJobGroup = JobGroup(subscription=testSubscription)
testJobGroup.create()
# Create jobs
for id in range(nJobs):
testJob = Job(name='Job_%i' % (id))
testJob['owner'] = "mnorman"
testJob['location'] = 'Xanadu'
testJob.create(testJobGroup)
testJobGroup.add(testJob)
testFileset.commit()
testJobGroup.commit()
return testJobGroup
def testA_BulkDAOs(self):
"""
_BulkDAOs_
Test the bulk DAO options, which is the only thing we should be using.
"""
myThread = threading.currentThread()
jobGroup = self.createJobs(nJobs=10)
runJobs = []
for job in jobGroup.jobs:
runJob = RunJob(jobid=job.exists())
runJob['status'] = 'New'
runJob['userdn'] = job['owner']
runJob['usergroup'] = 'phgroup'
runJob['userrole'] = 'cmsrole'
runJobs.append(runJob)
statusDAO = self.daoFactory(classname="NewState")
statusDAO.execute(states=['New', 'Gone', 'Dead'])
result = myThread.dbi.processData(
"SELECT name FROM bl_status")[0].fetchall()
self.assertItemsEqual(result, [('Dead', ), ('Gone', ), ('New', )])
newJobDAO = self.daoFactory(classname="NewJobs")
newJobDAO.execute(jobs=runJobs)
result = myThread.dbi.processData(
"SELECT wmbs_id FROM bl_runjob")[0].fetchall()
self.assertEqual(result, [(1, ), (2, ), (3, ), (4, ), (5, ), (6, ),
(7, ), (8, ), (9, ), (10, )])
loadJobsDAO = self.daoFactory(classname="LoadByStatus")
loadJobs = loadJobsDAO.execute(status="New")
self.assertEqual(len(loadJobs), 10)
idList = [x['id'] for x in loadJobs]
for job in loadJobs:
job['bulkid'] = 1001
updateDAO = self.daoFactory(classname="UpdateJobs")
updateDAO.execute(jobs=loadJobs)
loadJobs = loadJobsDAO.execute(status='New')
self.assertEqual(len(loadJobs), 10)
for job in loadJobs:
self.assertEqual(job['bulkid'], '1001')
loadWMBSDAO = self.daoFactory(classname="LoadByWMBSID")
for job in jobGroup.jobs:
jDict = loadWMBSDAO.execute(jobs=[job])
self.assertEqual(job['id'], jDict[0]['jobid'])
setStatusDAO = self.daoFactory(classname="SetStatus")
setStatusDAO.execute(jobs=idList, status='Dead')
result = loadJobsDAO.execute(status='Dead')
self.assertEqual(len(result), 10)
result = loadJobsDAO.execute(status='New')
self.assertEqual(len(result), 0)
runningJobDAO = self.daoFactory(classname="LoadRunning")
runningJobs = runningJobDAO.execute()
self.assertEqual(len(runningJobs), 10)
completeDAO = self.daoFactory(classname="CompleteJob")
completeDAO.execute(jobs=idList)
result = loadJobsDAO.execute(status='Dead', complete='0')
self.assertEqual(len(result), 10)
deleteDAO = self.daoFactory(classname="DeleteJobs")
deleteDAO.execute(jobs=idList)
result = loadJobsDAO.execute(status='Dead')
self.assertEqual(len(result), 0)
return
def testB_CheckWMBSBuild(self):
"""
_CheckWMBSBuild_
Trivial test that checks whether we can build
runJobs from WMBS jobs
"""
jobGroup = self.createJobs(nJobs=10)
for job in jobGroup.jobs:
rj = RunJob()
rj.buildFromJob(job=job)
self.assertEqual(job['id'], rj['jobid'])
self.assertEqual(job['retry_count'], rj['retry_count'])
job2 = rj.buildWMBSJob()
self.assertEqual(job['id'], job2['id'])
self.assertEqual(job['retry_count'], job2['retry_count'])
return
def testC_CheckWMBSBuildRoleAndGroup(self):
"""
_CheckWMBSBuild_
Trivial test that checks whether we can build
runJobs from WMBS jobs
"""
jobGroup = []
# Create jobs
for id in range(10):
testJob = Job(name='Job_%i' % (id))
testJob['owner'] = "mnorman"
testJob['usergroup'] = "mygroup_%i" % id
testJob['userrole'] = "myrole_%i" % id
testJob['location'] = 'Xanadu'
jobGroup.append(testJob)
for job in jobGroup:
rj = RunJob()
rj.buildFromJob(job=job)
self.assertEqual(job['usergroup'], rj['usergroup'])
self.assertEqual(job['userrole'], rj['userrole'])
job2 = rj.buildWMBSJob()
self.assertEqual(job['usergroup'], job2['usergroup'])
self.assertEqual(job['userrole'], job2['userrole'])
return
class ParentlessMergeBySizeTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Boiler plate DB setup.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMCore.WMBS"],
useDefault=False)
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
return
def tearDown(self):
"""
_tearDown_
Clear out WMBS.
"""
self.testInit.clearDatabase()
return
def stuffWMBS(self):
"""
_stuffWMBS_
Insert some dummy jobs, jobgroups, filesets, files and subscriptions
into WMBS to test job creation. Three completed job groups each
containing several files are injected. Another incomplete job group is
also injected. Also files are added to the "Mergeable" subscription as
well as to the output fileset for their jobgroups.
"""
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute(siteName="s1", seName="somese.cern.ch")
locationAction.execute(siteName="s1", seName="somese2.cern.ch")
changeStateDAO = self.daoFactory(classname="Jobs.ChangeState")
self.mergeFileset = Fileset(name="mergeFileset")
self.mergeFileset.create()
self.bogusFileset = Fileset(name="bogusFileset")
self.bogusFileset.create()
mergeWorkflow = Workflow(name="mergeWorkflow",
spec="bunk2",
owner="Steve",
task="Test")
mergeWorkflow.create()
markWorkflow = self.daoFactory(
classname="Workflow.MarkInjectedWorkflows")
markWorkflow.execute(names=[mergeWorkflow.name], injected=True)
self.mergeSubscription = Subscription(
fileset=self.mergeFileset,
workflow=mergeWorkflow,
split_algo="ParentlessMergeBySize")
self.mergeSubscription.create()
self.bogusSubscription = Subscription(
fileset=self.bogusFileset,
workflow=mergeWorkflow,
split_algo="ParentlessMergeBySize")
file1 = File(lfn="file1",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
file1.addRun(Run(1, *[45]))
file1.create()
file2 = File(lfn="file2",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
file2.addRun(Run(1, *[45]))
file2.create()
file3 = File(lfn="file3",
size=1024,
events=1024,
first_event=2048,
locations=set(["somese.cern.ch"]))
file3.addRun(Run(1, *[45]))
file3.create()
file4 = File(lfn="file4",
size=1024,
events=1024,
first_event=3072,
locations=set(["somese.cern.ch"]))
file4.addRun(Run(1, *[45]))
file4.create()
fileA = File(lfn="fileA",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
fileA.addRun(Run(1, *[46]))
fileA.create()
fileB = File(lfn="fileB",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
fileB.addRun(Run(1, *[46]))
fileB.create()
fileC = File(lfn="fileC",
size=1024,
events=1024,
first_event=2048,
locations=set(["somese.cern.ch"]))
fileC.addRun(Run(1, *[46]))
fileC.create()
fileI = File(lfn="fileI",
size=1024,
events=1024,
first_event=0,
locations=set(["somese.cern.ch"]))
fileI.addRun(Run(2, *[46]))
fileI.create()
fileII = File(lfn="fileII",
size=1024,
events=1024,
first_event=1024,
locations=set(["somese.cern.ch"]))
fileII.addRun(Run(2, *[46]))
fileII.create()
fileIII = File(lfn="fileIII",
size=1024,
events=102400,
first_event=2048,
locations=set(["somese.cern.ch"]))
fileIII.addRun(Run(2, *[46]))
fileIII.create()
fileIV = File(lfn="fileIV",
size=102400,
events=1024,
first_event=3072,
locations=set(["somese.cern.ch"]))
fileIV.addRun(Run(2, *[46]))
fileIV.create()
for file in [
file1, file2, file3, file4, fileA, fileB, fileC, fileI, fileII,
fileIII, fileIV
]:
self.mergeFileset.addFile(file)
self.bogusFileset.addFile(file)
self.mergeFileset.commit()
self.bogusFileset.commit()
return
def testMinMergeSize1(self):
"""
_testMinMergeSize1_
Set the minimum merge size to be 20,000 bytes which is more than the
sum of all file sizes in the WMBS instance. Verify that no merge jobs
will be produced.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=200000,
max_merge_size=2000000000,
max_merge_events=200000000)
assert len(result) == 0, \
"ERROR: No job groups should be returned."
return
def testMinMergeSize1(self):
"""
_testMinMergeSize1_
Set the minimum merge size to be 20,000 bytes which is more than the
sum of all file sizes in the WMBS instance. Verify that no merge jobs
will be produced.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=200000,
max_merge_size=2000000000,
max_merge_events=200000000)
assert len(result) == 0, \
"ERROR: No job groups should be returned."
return
def testMinMergeSize1a(self):
"""
_testMinMergeSize1a_
Set the minimum merge size to be 20,000 bytes which is more than the
sum of all file sizes in the WMBS instance and mark the fileset as
closed. Verify that one job containing all files is pushed out.
"""
self.stuffWMBS()
self.mergeFileset.markOpen(False)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=200000,
max_merge_size=2000000,
max_merge_events=2000000)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned: %s" % len(result)
assert len(result[0].jobs) == 1, \
"Error: One job should have been returned: %s" % len(result[0].jobs)
goldenFiles = [
"file1", "file2", "file3", "file4", "fileA", "fileB", "fileC",
"fileI", "fileII", "fileIII", "fileIV"
]
jobFiles = result[0].jobs[0].getFiles()
currentRun = 0
currentLumi = 0
currentEvent = 0
for file in jobFiles:
file.loadData()
assert file["lfn"] in goldenFiles, \
"Error: Unknown file: %s" % file["lfn"]
self.assertTrue(
file["locations"] == set(["somese.cern.ch",
"somese2.cern.ch"]),
"Error: File is missing a location.")
goldenFiles.remove(file["lfn"])
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
assert fileRun >= currentRun, \
"ERROR: Files not sorted by run."
if fileRun == currentRun:
assert fileLumi >= currentLumi, \
"ERROR: Files not ordered by lumi"
if fileLumi == currentLumi:
assert fileEvent >= currentEvent, \
"ERROR: Files not ordered by first event"
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
return
def testMaxMergeSize(self):
"""
_testMaxMergeSize_
Set the maximum merge size to be 100000 bytes. Verify that two merge
jobs are created, one for the one large file and another for the rest of
the files. Verify that each merge job contains the expected files and
that we merge across runs.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=1,
max_merge_size=100000,
max_merge_events=200000)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned: %s" % result
assert len(result[0].jobs) == 2, \
"ERROR: Two jobs should have been returned."
goldenFilesA = [
"file1", "file2", "file3", "file4", "fileA", "fileB", "fileC",
"fileI", "fileII", "fileIII"
]
goldenFilesB = ["fileIV"]
for job in result[0].jobs:
jobFiles = job.getFiles()
if jobFiles[0]["lfn"] in goldenFilesA:
goldenFiles = goldenFilesA
elif jobFiles[0]["lfn"] in goldenFilesB:
goldenFiles = goldenFilesB
currentRun = 0
currentLumi = 0
currentEvent = 0
for file in jobFiles:
assert file["lfn"] in goldenFiles, \
"Error: Unknown file in merge jobs."
assert file["locations"] == set(["somese.cern.ch"]), \
"Error: File is missing a location."
goldenFiles.remove(file["lfn"])
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
assert fileRun >= currentRun, \
"ERROR: Files not sorted by run."
if fileRun == currentRun:
assert fileLumi >= currentLumi, \
"ERROR: Files not ordered by lumi"
if fileLumi == currentLumi:
assert fileEvent >= currentEvent, \
"ERROR: Files not ordered by first event"
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
assert len(goldenFilesA) == 0 and len(goldenFilesB) == 0, \
"ERROR: Files missing from merge jobs."
return
def testMaxEvents(self):
"""
_testMaxEvents_
Verify the the max_merge_events parameter works and that we correctly
merge across runs.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=1,
max_merge_size=20000000,
max_merge_events=100000)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned: %s" % result
assert len(result[0].jobs) == 2, \
"ERROR: Two jobs should have been returned: %s" % len(result[0].jobs)
goldenFilesA = [
"file1", "file2", "file3", "file4", "fileA", "fileB", "fileC",
"fileI", "fileII", "fileIV"
]
goldenFilesB = ["fileIII"]
for job in result[0].jobs:
jobFiles = job.getFiles()
if jobFiles[0]["lfn"] in goldenFilesA:
goldenFiles = goldenFilesA
elif jobFiles[0]["lfn"] in goldenFilesB:
goldenFiles = goldenFilesB
currentRun = 0
currentLumi = 0
currentEvent = 0
for file in jobFiles:
assert file["lfn"] in goldenFiles, \
"Error: Unknown file in merge jobs."
assert file["locations"] == set(["somese.cern.ch"]), \
"Error: File is missing a location: %s" % file["locations"]
goldenFiles.remove(file["lfn"])
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
assert fileRun >= currentRun, \
"ERROR: Files not sorted by run: %s, %s" % (fileRun, currentRun)
if fileRun == currentRun:
assert fileLumi >= currentLumi, \
"ERROR: Files not ordered by lumi"
if fileLumi == currentLumi:
assert fileEvent >= currentEvent, \
"ERROR: Files not ordered by first event"
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
assert len(goldenFilesA) == 0 and len(goldenFilesB) == 0 and \
"ERROR: Files missing from merge jobs."
return
def testMinMergeSize1aNoRunMerge(self):
"""
_testMinMergeSize1aNoRunMerge_
Set the minimum merge size to be 20,000 bytes which is more than the
sum of all file sizes in the WMBS instance and mark the fileset as
closed. Verify that two jobs are pushed out and that we don't merge
accross run boundaries.
"""
self.stuffWMBS()
self.mergeFileset.markOpen(False)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=200000,
max_merge_size=2000000,
max_merge_events=2000000,
merge_across_runs=False)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned: %s" % len(result)
assert len(result[0].jobs) == 2, \
"Error: Two jobs should have been returned: %s" % len(result[0].jobs)
goldenFilesA = [
"file1", "file2", "file3", "file4", "fileA", "fileB", "fileC"
]
goldenFilesB = ["fileI", "fileII", "fileIII", "fileIV"]
goldenFilesA.sort()
goldenFilesB.sort()
for job in result[0].jobs:
currentRun = 0
currentLumi = 0
currentEvent = 0
jobLFNs = []
for file in job.getFiles():
file.loadData()
jobLFNs.append(file["lfn"])
self.assertTrue(
file["locations"] == set(
["somese.cern.ch", "somese2.cern.ch"]),
"Error: File is missing a location.")
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
assert fileRun >= currentRun, \
"ERROR: Files not sorted by run."
if fileRun == currentRun:
assert fileLumi >= currentLumi, \
"ERROR: Files not ordered by lumi"
if fileLumi == currentLumi:
assert fileEvent >= currentEvent, \
"ERROR: Files not ordered by first event"
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
jobLFNs.sort()
if jobLFNs == goldenFilesA:
goldenFilesA = []
else:
self.assertEqual(jobLFNs, goldenFilesB,
"Error: LFNs do not match.")
goldenFilesB = []
return
def testMaxMergeSizeNoRunMerge(self):
"""
_testMaxMergeSizeNoRunMerge_
Set the maximum merge size to be 100000 bytes. Verify that two merge
jobs are created, one for the one large file and another for the rest of
the files. Verify that each merge job contains the expected files and
that we don't merge across run boundaries.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=1,
max_merge_size=100000,
max_merge_events=200000,
merge_across_runs=False)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned: %s" % result
assert len(result[0].jobs) == 3, \
"ERROR: Three jobs should have been returned."
goldenFilesA = [
"file1", "file2", "file3", "file4", "fileA", "fileB", "fileC"
]
goldenFilesB = ["fileI", "fileII", "fileIII"]
goldenFilesC = ["fileIV"]
for job in result[0].jobs:
jobFiles = job.getFiles()
if jobFiles[0]["lfn"] in goldenFilesA:
goldenFiles = goldenFilesA
elif jobFiles[0]["lfn"] in goldenFilesB:
goldenFiles = goldenFilesB
else:
goldenFiles = goldenFilesC
currentRun = 0
currentLumi = 0
currentEvent = 0
for file in jobFiles:
self.assertTrue(file["lfn"] in goldenFiles,
"Error: Unknown file in merge jobs.")
self.assertTrue(file["locations"] == set(["somese.cern.ch"]),
"Error: File is missing a location.")
goldenFiles.remove(file["lfn"])
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
self.assertTrue(fileRun >= currentRun,
"ERROR: Files not sorted by run.")
if fileRun == currentRun:
self.assertTrue(fileLumi >= currentLumi,
"ERROR: Files not ordered by lumi")
if fileLumi == currentLumi:
self.assertTrue(fileEvent >= currentEvent,
"ERROR: Files not ordered by first event")
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
self.assertTrue(
len(goldenFilesA) == 0 and len(goldenFilesB) == 0,
"ERROR: Files missing from merge jobs.")
return
def testMaxEventsNoRunMerge(self):
"""
_testMaxEventsNoRunMerge_
Verify that the max events merge parameter works correctly and that we
don't merge accross run boundaries.
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=1,
max_merge_size=20000000,
max_merge_events=100000,
merge_across_runs=False)
self.assertTrue(
len(result) == 1,
"ERROR: More than one JobGroup returned: %s" % result)
self.assertTrue(
len(result[0].jobs) == 3,
"ERROR: Three jobs should have been returned: %s" %
len(result[0].jobs))
goldenFilesA = [
"file1",
"file2",
"file3",
"file4",
"fileA",
"fileB",
"fileC",
]
goldenFilesB = ["fileI", "fileII", "fileIV"]
goldenFilesC = ["fileIII"]
for job in result[0].jobs:
jobFiles = job.getFiles()
if jobFiles[0]["lfn"] in goldenFilesA:
goldenFiles = goldenFilesA
elif jobFiles[0]["lfn"] in goldenFilesB:
goldenFiles = goldenFilesB
else:
goldenFiles = goldenFilesC
currentRun = 0
currentLumi = 0
currentEvent = 0
for file in jobFiles:
self.assertTrue(file["lfn"] in goldenFiles,
"Error: Unknown file in merge jobs.")
self.assertTrue(
file["locations"] == set(["somese.cern.ch"]),
"Error: File is missing a location: %s" %
file["locations"])
goldenFiles.remove(file["lfn"])
fileRun = list(file["runs"])[0].run
fileLumi = min(list(file["runs"])[0])
fileEvent = file["first_event"]
if currentRun == 0:
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
continue
self.assertTrue(
fileRun >= currentRun,
"ERROR: Files not sorted by run: %s, %s" %
(fileRun, currentRun))
if fileRun == currentRun:
self.assertTrue(fileLumi >= currentLumi,
"ERROR: Files not ordered by lumi")
if fileLumi == currentLumi:
self.assertTrue(
fileEvent >= currentEvent,
"ERROR: Files not ordered by first event")
currentRun = fileRun
currentLumi = fileLumi
currentEvent = fileEvent
self.assertTrue(
len(goldenFilesA) == 0 and len(goldenFilesB) == 0
and len(goldenFilesC) == 0,
"ERROR: Files missing from merge jobs.")
return
def testLocationMerging(self):
"""
_testLocationMerging_
Verify that files residing on different SEs are not merged together in
the same job.
"""
self.stuffWMBS()
locationAction = self.daoFactory(classname="Locations.New")
locationAction.execute(siteName="s2", seName="somese2.cern.ch")
fileSite2 = File(lfn="fileSite2",
size=4098,
events=1024,
first_event=0,
locations=set(["somese2.cern.ch"]))
fileSite2.addRun(Run(1, *[46]))
fileSite2.create()
self.mergeFileset.addFile(fileSite2)
self.mergeFileset.commit()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=4097,
max_merge_size=99999999,
max_merge_events=999999999,
merge_across_runs=False)
assert len(result) == 1, \
"ERROR: More than one JobGroup returned."
assert len(result[0].jobs) == 3, \
"ERROR: Three jobs should have been returned."
for job in result[0].jobs:
firstInputFile = job.getFiles()[0]
baseLocation = list(firstInputFile["locations"])[0]
for inputFile in job.getFiles():
assert len(inputFile["locations"]) == 1, \
"Error: Wrong number of locations"
assert list(inputFile["locations"])[0] == baseLocation, \
"Error: Wrong location."
return
def testMaxWaitTime(self):
"""
_testMaxWaitTime_
Set the max wait times to be negative - this should force all files to merge
out immediately
Using the first setup as the first merge test which should normally produce
no jobGroups
"""
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=200000,
max_merge_size=2000000000,
max_merge_events=200000000,
max_wait_time=-10)
# Everything should be in one, small jobGroup
self.assertEqual(len(result), 1)
self.assertEqual(len(result[0].jobs), 1)
job = result[0].jobs[0]
# All files should be in one job
self.assertEqual(len(job.getFiles()), 11)
return
def testDifferentSubscritionIDs(self):
"""
_testDifferentSubscriptionIDs_
Make sure that the merge splitting still runs if the subscription ID
is not equal to the workflow ID.
"""
myThread = threading.currentThread()
myThread.transaction.begin()
dummyWorkflow = Workflow(name="dummyWorkflow",
spec="bunk49",
owner="Steve",
task="Test2")
dummyWorkflow.create()
dummyFileset = Fileset(name="dummyFileset")
dummyFileset.create()
dummySubscription1 = Subscription(fileset=dummyFileset,
workflow=dummyWorkflow,
split_algo="ParentlessMergeBySize")
dummySubscription2 = Subscription(fileset=dummyFileset,
workflow=dummyWorkflow,
split_algo="ParentlessMergeBySize")
dummySubscription1.create()
dummySubscription2.create()
myThread.transaction.commit()
self.stuffWMBS()
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.mergeSubscription)
result = jobFactory(min_merge_size=4097,
max_merge_size=99999999,
max_merge_events=999999999,
merge_across_runs=False)
self.assertEqual(len(result), 1)
jobGroup = result[0]
self.assertEqual(len(jobGroup.jobs), 2)
return
class RepackMergeTest(unittest.TestCase):
"""
_RepackMergeTest_
Test for RepackMerge job splitter
"""
def setUp(self):
"""
_setUp_
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMComponent.DBS3Buffer", "T0.WMBS"])
self.splitterFactory = SplitterFactory(package = "T0.JobSplitting")
myThread = threading.currentThread()
daoFactory = DAOFactory(package = "T0.WMBS",
logger = logging,
dbinterface = myThread.dbi)
wmbsDaoFactory = DAOFactory(package = "WMCore.WMBS",
logger = logging,
dbinterface = myThread.dbi)
myThread.dbi.processData("""INSERT INTO wmbs_location
(id, site_name, state, state_time)
VALUES (1, 'SomeSite', 1, 1)
""", transaction = False)
myThread.dbi.processData("""INSERT INTO wmbs_pnns
(id, pnn)
VALUES (2, 'SomePNN')
""", transaction = False)
myThread.dbi.processData("""INSERT INTO wmbs_location_pnns
(location, pnn)
VALUES (1, 2)
""", transaction = False)
insertRunDAO = daoFactory(classname = "RunConfig.InsertRun")
insertRunDAO.execute(binds = { 'RUN' : 1,
'HLTKEY' : "someHLTKey" },
transaction = False)
insertLumiDAO = daoFactory(classname = "RunConfig.InsertLumiSection")
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 1 },
transaction = False)
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 2 },
transaction = False)
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 3 },
transaction = False)
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 4 },
transaction = False)
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 5 },
transaction = False)
insertStreamDAO = daoFactory(classname = "RunConfig.InsertStream")
insertStreamDAO.execute(binds = { 'STREAM' : "A" },
transaction = False)
insertCMSSVersionDAO = daoFactory(classname = "RunConfig.InsertCMSSWVersion")
insertCMSSVersionDAO.execute(binds = { 'VERSION' : "CMSSW_4_2_7" },
transaction = False)
insertStreamCMSSWVersionDAO = daoFactory(classname = "RunConfig.InsertStreamCMSSWVersion")
insertStreamCMSSWVersionDAO.execute(binds = { 'RUN' : 1,
'STREAM' : 'A',
'VERSION' : "CMSSW_4_2_7" },
transaction = False)
insertStreamerDAO = daoFactory(classname = "RunConfig.InsertStreamer")
insertStreamerDAO.execute(streamerPNN = "SomePNN",
binds = { 'RUN' : 1,
'P5_ID' : 1,
'LUMI' : 4,
'STREAM' : "A",
'LFN' : "/testLFN/A",
'FILESIZE' : 100,
'EVENTS' : 100,
'TIME' : int(time.time()) },
transaction = False)
insertStreamFilesetDAO = daoFactory(classname = "RunConfig.InsertStreamFileset")
insertStreamFilesetDAO.execute(1, "A", "TestFileset1")
self.fileset1 = Fileset(name = "TestFileset1")
self.fileset2 = Fileset(name = "TestFileset2")
self.fileset1.load()
self.fileset2.create()
workflow1 = Workflow(spec = "spec.xml", owner = "hufnagel", name = "TestWorkflow1", task="Test")
workflow2 = Workflow(spec = "spec.xml", owner = "hufnagel", name = "TestWorkflow2", task="Test")
workflow1.create()
workflow2.create()
self.subscription1 = Subscription(fileset = self.fileset1,
workflow = workflow1,
split_algo = "Repack",
type = "Repack")
self.subscription2 = Subscription(fileset = self.fileset2,
workflow = workflow2,
split_algo = "RepackMerge",
type = "RepackMerge")
self.subscription1.create()
self.subscription2.create()
myThread.dbi.processData("""INSERT INTO wmbs_workflow_output
(WORKFLOW_ID, OUTPUT_IDENTIFIER, OUTPUT_FILESET)
VALUES (%d, 'SOMEOUTPUT', %d)
""" % (workflow1.id, self.fileset2.id),
transaction = False)
# keep for later
self.insertSplitLumisDAO = daoFactory(classname = "JobSplitting.InsertSplitLumis")
self.insertClosedLumiDAO = daoFactory(classname = "RunLumiCloseout.InsertClosedLumi")
self.feedStreamersDAO = daoFactory(classname = "Tier0Feeder.FeedStreamers")
self.acquireFilesDAO = wmbsDaoFactory(classname = "Subscriptions.AcquireFiles")
self.completeFilesDAO = wmbsDaoFactory(classname = "Subscriptions.CompleteFiles")
self.currentTime = int(time.time())
# default split parameters
self.splitArgs = {}
self.splitArgs['minInputSize'] = 2.1 * 1024 * 1024 * 1024
self.splitArgs['maxInputSize'] = 4.0 * 1024 * 1024 * 1024
self.splitArgs['maxInputEvents'] = 100000000
self.splitArgs['maxInputFiles'] = 1000
self.splitArgs['maxEdmSize'] = 20 * 1024 * 1024 * 1024
self.splitArgs['maxOverSize'] = 10 * 1024 * 1024 * 1024
self.SplitArgs['maxLatency'] = 50000
return
def tearDown(self):
"""
_tearDown_
"""
self.testInit.clearDatabase()
return
def deleteSplitLumis(self):
"""
_deleteSplitLumis_
"""
myThread = threading.currentThread()
myThread.dbi.processData("""DELETE FROM lumi_section_split_active
""",
transaction = False)
return
def test00(self):
"""
_test00_
Test that the job name prefix feature works
Test max edm size threshold for single lumi
small lumi, followed by over-large lumi
expect 1 job for small lumi and 4 jobs for over-large
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2 * lumi):
newFile = File(makeUUID(), size = 1000 * lumi * lumi, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxEdmSize'] = 13000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 3,
"ERROR: JobFactory didn't create three jobs")
job = jobGroups[0].jobs[0]
self.assertTrue(job['name'].startswith("RepackMerge-"),
"ERROR: Job has wrong name")
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 3,
"ERROR: Job does not process 3 files")
job = jobGroups[0].jobs[2]
self.assertEqual(len(job.getFiles()), 1,
"ERROR: Job does not process 1 file")
return
def test01(self):
"""
_test01_
Test max size threshold for single lumi
small lumi, followed by large lumi
expect 1 job for small lumi and 1 job for large
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size = 1000 * lumi, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputSize'] = 3000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test02(self):
"""
_test02_
Test max event threshold for single lumi
small lumi, followed by large lumi
expect 1 job for small lumi and 1 job for large
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size = 1000, events = 100 * lumi)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputEvents'] = 300
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test03(self):
"""
_test03_
Test max input files threshold for single lumi
small lumi, followed by large lumi
expect 1 job for small lumi and 1 job for large
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(lumi * 2):
newFile = File(makeUUID(), size = 1000, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputFiles'] = 3
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
return
def test04(self):
"""
_test04_
Test max size threshold for multi lumi
3 same size lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 3]:
for i in range(2):
newFile = File(makeUUID(), size = 1000, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 3000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputSize'] = 5000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset2.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test05(self):
"""
_test05_
Test max event threshold for multi lumi
3 same size lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 3]:
for i in range(2):
newFile = File(makeUUID(), size = 1000, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 3000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputEvents'] = 500
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset2.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test06(self):
"""
_test06_
Test max input files threshold for multi lumi
3 same size lumis
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 3]:
for i in range(2):
newFile = File(makeUUID(), size = 1000, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 3000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputFiles'] = 5
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset2.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test07(self):
"""
_test07_
Test over merge
one small lumi, one large lumi (small below min size,
large below max size, but both together above max size)
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size = 1000 * lumi * lumi, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 3000
mySplitArgs['maxInputSize'] = 9000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
return
def test08(self):
"""
_test08_
Test under merge (over merge size threshold)
one small lumi, one large lumi (small below min size,
large below max size, but both together above max size)
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size = 1000 * lumi * lumi, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 3000
mySplitArgs['maxInputSize'] = 9000
mySplitArgs['maxOverSize'] = 9500
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
self.fileset2.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test09(self):
"""
_test09_
Test under merge (over merge event threshold)
one small lumi, one large lumi (small below min size,
large below max size, but both together above max size)
It was changed due to maxinputevents not being used anymore.
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2]:
for i in range(2):
newFile = File(makeUUID(), size = 1000 * lumi * lumi, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 1500
mySplitArgs['maxInputSize'] = 9000
mySplitArgs['maxOverSize'] = 9500
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
self.fileset2.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
return
def test10(self):
"""
_test10_
Test merging of multiple lumis with holes in the lumi sequence
Hole is due to no streamer files for the lumi
Multi lumi input
It only works with a single hole, as it creates a merged file even with it being of a smaller size than the mininputsize.
It was changed due to the maxinputevents not being used anymore
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 4]:
for i in range(2):
newFile = File(makeUUID(), size = 1000, events = 100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomePNN", immediateSave = False)
newFile.create()
self.fileset2.addFile(newFile)
self.fileset2.commit()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription2)
mySplitArgs['minInputSize'] = 100000
mySplitArgs['maxInputSize'] = 200000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
self.insertClosedLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 3,
'STREAM' : "A",
'FILECOUNT' : 0,
'INSERT_TIME' : self.currentTime,
'CLOSE_TIME' : self.currentTime },
transaction = False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
return
class HeartbeatTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all of the
Heartbeat tables. Also add some dummy locations.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging() # logLevel = logging.SQLDEBUG
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMCore.Agent.Database"],
useDefault=False)
self.heartbeat = HeartbeatAPI("testComponent")
def tearDown(self):
"""
_tearDown_
Drop all the Heartbeat tables.
"""
self.testInit.clearDatabase()
def testHeartbeat(self):
testComponent = HeartbeatAPI("testComponent")
testComponent.registerComponent()
self.assertEqual(testComponent.getHeartbeatInfo(), [])
testComponent.updateWorkerHeartbeat("testWorker")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 1)
self.assertEqual(result[0]['worker_name'], "testWorker")
time.sleep(1)
testComponent.updateWorkerHeartbeat("testWorker2")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 1)
self.assertEqual(result[0]['worker_name'], "testWorker2")
time.sleep(1)
testComponent.updateWorkerHeartbeat("testWorker")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 1)
self.assertEqual(result[0]['worker_name'], "testWorker")
testComponent = HeartbeatAPI("test2Component")
testComponent.registerComponent()
time.sleep(1)
testComponent.updateWorkerHeartbeat("test2Worker")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 2)
self.assertEqual(result[0]['worker_name'], "testWorker")
self.assertEqual(result[1]['worker_name'], "test2Worker")
time.sleep(1)
testComponent.updateWorkerHeartbeat("test2Worker2")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 2)
self.assertEqual(result[0]['worker_name'], "testWorker")
self.assertEqual(result[1]['worker_name'], "test2Worker2")
time.sleep(1)
testComponent.updateWorkerHeartbeat("test2Worker")
result = testComponent.getHeartbeatInfo()
self.assertEqual(len(result), 2)
self.assertEqual(result[0]['worker_name'], "testWorker")
self.assertEqual(result[1]['worker_name'], "test2Worker")
testComponent.updateWorkerError("test2Worker", "Error1")
result = testComponent.getHeartbeatInfo()
self.assertEqual(result[1]['error_message'], "Error1")
class ConditionTest(unittest.TestCase):
"""
_ExpressTest_
Test for Express job splitter
"""
def setUp(self):
"""
_setUp_
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["T0.WMBS"])
self.splitterFactory = SplitterFactory(package = "T0.JobSplitting")
myThread = threading.currentThread()
daoFactory = DAOFactory(package = "T0.WMBS",
logger = logging,
dbinterface = myThread.dbi)
wmbsDaoFactory = DAOFactory(package = "WMCore.WMBS",
logger = logging,
dbinterface = myThread.dbi)
myThread.dbi.processData("""INSERT INTO wmbs_location
(id, site_name, state)
VALUES (1, 'SomeSite', 1)
""", transaction = False)
myThread.dbi.processData("""INSERT INTO wmbs_location_senames
(location, se_name)
VALUES (1, 'SomeSE')
""", transaction = False)
insertRunDAO = daoFactory(classname = "RunConfig.InsertRun")
insertRunDAO.execute(binds = { 'RUN' : 1,
'TIME' : int(time.time()),
'HLTKEY' : "someHLTKey" },
transaction = False)
insertLumiDAO = daoFactory(classname = "RunConfig.InsertLumiSection")
insertLumiDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 1 },
transaction = False)
insertStreamDAO = daoFactory(classname = "RunConfig.InsertStream")
insertStreamDAO.execute(binds = { 'STREAM' : "Express" },
transaction = False)
insertStreamFilesetDAO = daoFactory(classname = "RunConfig.InsertStreamFileset")
insertStreamFilesetDAO.execute(1, "Express", "TestFileset1")
insertStreamerDAO = daoFactory(classname = "RunConfig.InsertStreamer")
insertStreamerDAO.execute(binds = { 'RUN' : 1,
'LUMI' : 1,
'STREAM' : "Express",
'TIME' : int(time.time()),
'LFN' : "/streamer",
'FILESIZE' : 0,
'EVENTS' : 0 },
transaction = False)
insertPromptCalibrationDAO = daoFactory(classname = "RunConfig.InsertPromptCalibration")
insertPromptCalibrationDAO.execute( { 'RUN' : 1,
'STREAM' : "Express" },
transaction = False)
self.fileset1 = Fileset(name = "TestFileset1")
self.fileset1.create()
workflow1 = Workflow(spec = "spec.xml", owner = "hufnagel", name = "TestWorkflow1", task="Test")
workflow1.create()
self.subscription1 = Subscription(fileset = self.fileset1,
workflow = workflow1,
split_algo = "Condition",
type = "Condition")
self.subscription1.create()
# set parentage chain and sqlite fileset
alcaRecoFile = File("/alcareco", size = 0, events = 0)
alcaRecoFile.addRun(Run(1, *[1]))
alcaRecoFile.setLocation("SomeSE", immediateSave = False)
alcaRecoFile.create()
alcaPromptFile = File("/alcaprompt", size = 0, events = 0)
alcaPromptFile.addRun(Run(1, *[1]))
alcaPromptFile.setLocation("SomeSE", immediateSave = False)
alcaPromptFile.create()
sqliteFile = File("/sqlite", size = 0, events = 0)
sqliteFile.create()
self.fileset1.addFile(sqliteFile)
self.fileset1.commit()
results = myThread.dbi.processData("""SELECT lfn FROM wmbs_file_details
""",
transaction = False)[0].fetchall()
setParentageDAO = wmbsDaoFactory(classname = "Files.SetParentage")
setParentageDAO.execute(binds = [ { 'parent' : "/streamer",
'child' : "/alcareco" },
{ 'parent' : "/alcareco",
'child' : "/alcaprompt" },
{ 'parent' : "/alcaprompt",
'child' : "/sqlite" } ],
transaction = False)
# default split parameters
self.splitArgs = {}
self.splitArgs['runNumber'] = 1
self.splitArgs['streamName'] = "Express"
return
def tearDown(self):
"""
_tearDown_
"""
self.testInit.clearDatabase()
return
def isPromptCalibFinished(self):
"""
_isPromptCalibFinished_
"""
myThread = threading.currentThread()
result = myThread.dbi.processData("""SELECT finished
FROM prompt_calib
""",
transaction = False)[0].fetchall()[0][0]
return result
def countPromptCalibFiles(self):
"""
_deleteSplitLumis_
"""
myThread = threading.currentThread()
result = myThread.dbi.processData("""SELECT COUNT(*)
FROM prompt_calib_file
""",
transaction = False)[0].fetchall()[0][0]
return result
def test00(self):
"""
_test00_
Make sure the job splitter behaves correctly.
Just make sure the job splitter does nothing
when the fileset is open and populates t0ast
data structures when it's closed. In the later
case all input files should be marked as
acquired without creating a job as well.
"""
mySplitArgs = self.splitArgs.copy()
jobFactory = self.splitterFactory(package = "WMCore.WMBS",
subscription = self.subscription1)
self.assertEqual(self.isPromptCalibFinished(), 0,
"ERROR: prompt_calib should not be finished")
self.assertEqual(self.countPromptCalibFiles(), 0,
"ERROR: there should be no prompt_calib_file")
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(self.isPromptCalibFinished(), 0,
"ERROR: prompt_calib should not be finished")
self.assertEqual(self.countPromptCalibFiles(), 1,
"ERROR: there should be one prompt_calib_file")
self.fileset1.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
self.assertEqual(self.isPromptCalibFinished(), 1,
"ERROR: prompt_calib should be finished")
self.assertEqual(self.countPromptCalibFiles(), 1,
"ERROR: there should be one prompt_calib_file")
return
class EventAwareLumiBasedTest(unittest.TestCase):
"""
_EventAwareLumiBasedTest_
Test event 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="T1_US_FNAL", pnn="T1_US_FNAL_Disk")
locationAction.execute(siteName="T2_CH_CERN", pnn="T2_CH_CERN")
self.testWorkflow = Workflow(spec="spec.xml",
owner="Steve",
name="wf001",
task="Test")
self.testWorkflow.create()
self.performanceParams = {
'timePerEvent': 12,
'memoryRequirement': 2300,
'sizePerEvent': 400
}
return
def tearDown(self):
"""
_tearDown_
Clear out WMBS.
"""
self.testInit.clearDatabase()
return
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, 'T1_US_FNAL_Disk')
newFile.create()
testFileset.addFile(newFile)
if twoSites:
for i in range(nFiles):
newFile = self.createFile('%s_%i_2' % (baseName, i),
nEventsPerFile, i, lumisPerFile,
'T2_CH_CERN')
newFile.create()
testFileset.addFile(newFile)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
return testSubscription
def createFile(self,
lfn,
events,
run,
lumis,
location,
lumiMultiplier=None):
"""
_createFile_
Create a file for testing
"""
if lumiMultiplier is None:
lumiMultiplier = run
newFile = File(lfn=lfn, size=1000, events=events)
lumiList = []
for lumi in range(lumis):
lumiList.append((lumiMultiplier * lumis) + lumi)
newFile.addRun(Run(run, *lumiList))
newFile.setLocation(location)
return newFile
def testA_FileSplitNoHardLimit(self):
"""
_testA_FileSplitNoHardLimit_
Simplest use case, there is only a self limit of events per job which
the algorithm must adapt to on a file by file basis. At most
one file per job so we don't have to pass information between files.
"""
splitter = SplitterFactory()
# Create 5 files with 7 lumi per file and 100 events per lumi on average.
testSubscription = self.createSubscription(nFiles=5,
lumisPerFile=7,
twoSites=False,
nEventsPerFile=700)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# First test, the optimal settings are 360 events per job
# As we have files with 100 events per lumi, this will configure the splitting to
# 3.6 lumis per job, which rounds to 3, the algorithm always approximates to the lower integer.
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
splitOnRun=True,
events_per_job=360,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 15, "There should be 15 jobs")
for idx, job in enumerate(jobs, start=1):
# Jobs may have 1 lumi or 2 check performance figures accordingly
self.assertEqual(job['estimatedMemoryUsage'], 2300)
if idx % 3 == 0:
self.assertEqual(job['estimatedDiskUsage'], 100 * 400)
self.assertEqual(job['estimatedJobTime'], 100 * 12)
else:
self.assertEqual(job['estimatedDiskUsage'], 3 * 100 * 400)
self.assertEqual(job['estimatedJobTime'], 3 * 100 * 12)
testSubscription = self.createSubscription(nFiles=5,
lumisPerFile=7,
twoSites=False,
nEventsPerFile=700)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# Now set the average to 200 events per job
# This results in the algorithm reducing the lumis per job to 2
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
splitOnRun=True,
events_per_job=200,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 20, "There should be 20 jobs")
for idx, job in enumerate(jobs, start=1):
# Jobs may have 1 lumi or 2 check performance figures accordingly
self.assertEqual(job['estimatedMemoryUsage'], 2300)
if idx % 4 == 0:
self.assertEqual(job['estimatedDiskUsage'], 100 * 400)
self.assertEqual(job['estimatedJobTime'], 100 * 12)
else:
self.assertEqual(job['estimatedDiskUsage'], 2 * 100 * 400)
self.assertEqual(job['estimatedJobTime'], 2 * 100 * 12)
# Check extremes, process a zero event files with lumis. It must be processed in one job
testSubscription = self.createSubscription(nFiles=5,
lumisPerFile=100,
twoSites=False,
nEventsPerFile=0)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
events_per_job=5000,
performance=self.performanceParams)
self.assertEqual(
len(jobGroups), 0,
"There are not enough events, so it should be 0 instead")
# we close this fileset to get it moving
fileset = testSubscription.getFileset()
fileset.markOpen(False)
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
events_per_job=5000,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 5, "There should be 5 jobs")
for job in jobs:
self.assertEqual(job['estimatedMemoryUsage'], 2300)
self.assertEqual(job['estimatedDiskUsage'], 0)
self.assertEqual(job['estimatedJobTime'], 0)
# Process files with 10k events per lumi, fallback to one lumi per job. We can't do better
testSubscription = self.createSubscription(nFiles=5,
lumisPerFile=5,
twoSites=False,
nEventsPerFile=50000)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
splitOnRun=True,
events_per_job=5000,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 25, "There should be 5 jobs")
for job in jobs:
self.assertEqual(job['estimatedMemoryUsage'], 2300)
self.assertEqual(job['estimatedDiskUsage'], 10000 * 400)
self.assertEqual(job['estimatedJobTime'], 10000 * 12)
def testB_NoFileSplitNoHardLimit(self):
"""
_testB_NoFileSplitNoHardLimit_
In this case we don't split on file boundaries, check different combination of files
make sure we make the most of the splitting, e.g. include many zero event files in
a single job.
"""
splitter = SplitterFactory()
# Create 100 files with 7 lumi per file and 0 events per lumi on average.
testSubscription = self.createSubscription(nFiles=100,
lumisPerFile=7,
twoSites=False,
nEventsPerFile=0)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# First test, the optimal settings are 360 events per job
# As we have files with 0 events per lumi, this will configure the splitting to
# a single job containing all files
jobGroups = jobFactory(halt_job_on_file_boundaries=False,
splitOnRun=False,
events_per_job=360,
performance=self.performanceParams)
self.assertEqual(
len(jobGroups), 0,
"There aren't enough events, so it should have 0 job groups")
# we close this fileset to get it moving
fileset = testSubscription.getFileset()
fileset.markOpen(False)
jobGroups = jobFactory(halt_job_on_file_boundaries=False,
splitOnRun=False,
events_per_job=360,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 1, "There should be 1 job")
self.assertEqual(len(jobs[0]['input_files']), 100,
"All 100 files must be in the job")
self.assertEqual(jobs[0]['estimatedMemoryUsage'], 2300)
self.assertEqual(jobs[0]['estimatedDiskUsage'], 0)
self.assertEqual(jobs[0]['estimatedJobTime'], 0)
# Create 7 files, each one with different lumi/event distributions
testFileset = Fileset(name="FilesetA")
testFileset.create()
testFileA = self.createFile("/this/is/file1", 250, 0, 5, "T2_CH_CERN")
testFileB = self.createFile("/this/is/file2", 600, 1, 1, "T2_CH_CERN")
testFileC = self.createFile("/this/is/file3", 1200, 2, 2, "T2_CH_CERN")
testFileD = self.createFile("/this/is/file4", 100, 3, 1, "T2_CH_CERN")
testFileE = self.createFile("/this/is/file5", 30, 4, 1, "T2_CH_CERN")
testFileF = self.createFile("/this/is/file6", 10, 5, 1, "T2_CH_CERN")
testFileG = self.createFile("/this/is/file7", 151, 6, 3, "T2_CH_CERN")
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.addFile(testFileC)
testFileset.addFile(testFileD)
testFileset.addFile(testFileE)
testFileset.addFile(testFileF)
testFileset.addFile(testFileG)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# Optimal settings are: jobs with 150 events per job
# This means, the first file must be splitted in 3 lumis per job which would leave room
# for another lumi in the second job, but the second file has a lumi too big for that
# The 3rd job only contains the second file, the fourth and fifth job split the third file
jobGroups = jobFactory(halt_job_on_file_boundaries=False,
splitOnRun=False,
events_per_job=150,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 7, "7 jobs must be in the jobgroup")
self.assertEqual(jobs[0]["mask"].getRunAndLumis(), {0: [[0, 2]]},
"Wrong mask for the first job")
self.assertEqual(jobs[0]["estimatedJobTime"], 150 * 12)
self.assertEqual(jobs[0]["estimatedDiskUsage"], 150 * 400)
self.assertEqual(jobs[1]["mask"].getRunAndLumis(), {0: [[3, 4]]},
"Wrong mask for the second job")
self.assertEqual(jobs[1]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[1]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[2]["mask"].getRunAndLumis(), {1: [[1, 1]]},
"Wrong mask for the third job")
self.assertEqual(jobs[2]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[2]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[3]["mask"].getRunAndLumis(), {2: [[4, 4]]},
"Wrong mask for the fourth job")
self.assertEqual(jobs[3]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[3]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[4]["mask"].getRunAndLumis(), {2: [[5, 5]]},
"Wrong mask for the fifth job")
self.assertEqual(jobs[4]["estimatedJobTime"], 600 * 12)
self.assertEqual(jobs[4]["estimatedDiskUsage"], 600 * 400)
self.assertEqual(jobs[5]["mask"].getRunAndLumis(), {
3: [[3, 3]],
4: [[4, 4]],
5: [[5, 5]]
}, "Wrong mask for the sixth job")
self.assertEqual(jobs[5]["estimatedJobTime"], 140 * 12)
self.assertEqual(jobs[5]["estimatedDiskUsage"], 140 * 400)
self.assertEqual(jobs[6]["mask"].getRunAndLumis(), {6: [[18, 20]]},
"Wrong mask for the seventh job")
self.assertEqual(jobs[6]["estimatedJobTime"], 150 * 12)
self.assertEqual(jobs[6]["estimatedDiskUsage"], 150 * 400)
for job in jobs:
self.assertEqual(job["estimatedMemoryUsage"], 2300)
# Test interactions of this algorithm with splitOnRun = True
# Make 2 files, one with 3 runs and a second one with the last run of the first
fileA = File(lfn="/this/is/file1a", size=1000, events=2400)
lumiListA = []
lumiListB = []
lumiListC = []
for lumi in range(8):
lumiListA.append(1 + lumi)
lumiListB.append(1 + lumi)
lumiListC.append(1 + lumi)
fileA.addRun(Run(1, *lumiListA))
fileA.addRun(Run(2, *lumiListA))
fileA.addRun(Run(3, *lumiListA))
fileA.setLocation("T1_US_FNAL_Disk")
fileB = self.createFile('/this/is/file2a', 200, 3, 5,
"T1_US_FNAL_Disk")
testFileset = Fileset(name='FilesetB')
testFileset.create()
testFileset.addFile(fileA)
testFileset.addFile(fileB)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# The settings for this splitting are 700 events per job
jobGroups = jobFactory(splitOnRun=True,
halt_job_on_file_boundaries=False,
events_per_job=700,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 6, "Six jobs must be in the jobgroup")
self.assertEqual(jobs[0]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[0]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[1]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[1]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[2]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[2]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[3]["estimatedJobTime"], 100 * 12)
self.assertEqual(jobs[3]["estimatedDiskUsage"], 100 * 400)
self.assertEqual(jobs[4]["estimatedJobTime"], 700 * 12)
self.assertEqual(jobs[4]["estimatedDiskUsage"], 700 * 400)
self.assertEqual(jobs[5]["estimatedJobTime"], 300 * 12)
self.assertEqual(jobs[5]["estimatedDiskUsage"], 300 * 400)
def testC_HardLimitSplitting(self):
"""
_testC_HardLimitSplitting_
Test that we can specify a event limit, the
algorithm shall take single lumi files with more events than the limit
and mark them for failure
"""
splitter = SplitterFactory()
# Create 3 files, the one in the middle is a "bad" file
testFileset = Fileset(name="FilesetA")
testFileset.create()
testFileA = self.createFile("/this/is/file1", 1000, 0, 5,
"T1_US_FNAL_Disk")
testFileB = self.createFile("/this/is/file2", 1000, 1, 1,
"T1_US_FNAL_Disk")
testFileC = self.createFile("/this/is/file3", 1000, 2, 2,
"T1_US_FNAL_Disk")
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.addFile(testFileC)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# Settings are to split on job boundaries, to fail sing lumis with more than 800 events
# and to put 550 events per job
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
splitOnRun=True,
events_per_job=550,
max_events_per_lumi=800,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 6, "Six jobs must be in the jobgroup")
self.assertTrue(
jobs[3]['failedOnCreation'],
"The job processing the second file should me marked for failure")
self.assertEqual(
jobs[3]['failedReason'],
"File /this/is/file2 has too many events (1000) in 1 lumi(s)",
"The reason for the failure is not accurate")
def testD_HardLimitSplittingOnly(self):
"""
_testD_HardLimitSplittingOnly_
Checks that we can split a set of files where every file has a single
lumi too big to fit in a runnable job
"""
splitter = SplitterFactory()
# Create 3 single-big-lumi files
testFileset = Fileset(name="FilesetA")
testFileset.create()
testFileA = self.createFile("/this/is/file1", 1000, 0, 1,
"T1_US_FNAL_Disk")
testFileB = self.createFile("/this/is/file2", 1000, 1, 1,
"T1_US_FNAL_Disk")
testFileC = self.createFile("/this/is/file3", 1000, 2, 1,
"T1_US_FNAL_Disk")
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.addFile(testFileC)
testFileset.commit()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
# Settings are to split on job boundaries, to fail sing lumis with more than 800 events
# and to put 550 events per job
jobGroups = jobFactory(halt_job_on_file_boundaries=True,
splitOnRun=True,
events_per_job=550,
max_events_per_lumi=800,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1,
"There should be only one job group")
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 3, "Three jobs must be in the jobgroup")
for i in range(1, 4):
self.assertTrue(
jobs[i - 1]['failedOnCreation'],
"The job processing the second file should me marked for failure"
)
self.assertEqual(
jobs[i - 1]['failedReason'],
"File /this/is/file%d has too many events (1000) in 1 lumi(s)"
% i, "The reason for the failure is not accurate")
return
def test_NotEnoughEvents(self):
"""
_test_NotEnoughEvents_
Checks whether jobs are not created when there are not enough files (actually, events)
according to the events_per_job requested to the splitter algorithm
"""
splitter = SplitterFactory()
# Very small fileset (single file) without enough events
testSubscription = self.createSubscription(nFiles=1,
lumisPerFile=2,
nEventsPerFile=200)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(events_per_job=500,
performance=self.performanceParams,
splitOnRun=False)
self.assertEqual(len(jobGroups), 0)
# Still a small fileset (two files) without enough events
testSubscription = self.createSubscription(nFiles=2,
lumisPerFile=2,
nEventsPerFile=200)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(events_per_job=500,
performance=self.performanceParams,
splitOnRun=False)
self.assertEqual(len(jobGroups), 0)
# Finally an acceptable fileset size (three files) with enough events
testSubscription = self.createSubscription(nFiles=3,
lumisPerFile=2,
nEventsPerFile=200)
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(events_per_job=500,
performance=self.performanceParams,
splitOnRun=False)
self.assertEqual(len(jobGroups), 1)
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 2)
self.assertEqual(len(jobs[0]['input_files']), 3)
self.assertEqual(len(jobs[1]['input_files']), 1)
self.assertEqual(jobs[0]['mask'].getRunAndLumis(), {
0: [[0, 1]],
1: [[2, 3]],
2: [[4, 4]]
})
self.assertEqual(jobs[1]['mask'].getRunAndLumis(), {2: [[5, 5]]})
# Test fileset with a single run and splitOnRun=True
testFileset = Fileset(name="FilesetA")
testFileA = self.createFile("/this/is/file1",
200,
1,
2,
"T1_US_FNAL_Disk",
lumiMultiplier=0)
testFileB = self.createFile("/this/is/file2",
200,
1,
2,
"T1_US_FNAL_Disk",
lumiMultiplier=1)
testFileC = self.createFile("/this/is/file3",
200,
1,
2,
"T1_US_FNAL_Disk",
lumiMultiplier=2)
testFileset.addFile(testFileA)
testFileset.addFile(testFileB)
testFileset.addFile(testFileC)
testFileset.create()
testSubscription = Subscription(fileset=testFileset,
workflow=self.testWorkflow,
split_algo="EventAwareLumiBased",
type="Processing")
testSubscription.create()
jobFactory = splitter(package="WMCore.WMBS",
subscription=testSubscription)
jobGroups = jobFactory(events_per_job=500,
performance=self.performanceParams)
self.assertEqual(len(jobGroups), 1)
jobs = jobGroups[0].jobs
self.assertEqual(len(jobs), 2)
self.assertEqual(len(jobs[0]['input_files']), 3)
self.assertEqual(len(jobs[1]['input_files']), 1)
self.assertEqual(jobs[0]['mask'].getRunAndLumis(),
{1: [[0, 1], [2, 3], [4, 4]]})
self.assertEqual(jobs[1]['mask'].getRunAndLumis(), {1: [[5, 5]]})
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.assertEquals(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time=int(time.time()) * 2)
self.assertEquals(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time=int(time.time()) * 2)
self.assertEquals(jobGroups, [], "Should have returned a null set")
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time=int(time.time()) * 2)
self.assertEquals(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.assertEquals(len(jobGroups), 1)
self.assertEquals(len(jobGroups[0].jobs), 1)
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEquals(len(myfiles), 10)
self.multipleLumiSubscription.getFileset().markOpen(False)
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time=1)
self.assertEquals(len(jobGroups), 1)
self.assertEquals(len(jobGroups[0].jobs), 1)
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEquals(len(myfiles), 10)
#self.assertEquals(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.assertEquals(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.assertEquals(jobGroups, [], "Should have returned a null set")
self.multipleFileSubscription.acquireFiles(
self.multipleFileSubscription.availableFiles())
jobFactory = splitter(self.multipleFileSubscription)
jobGroups = jobFactory(trigger_time=1)
self.assertEquals(jobGroups, [], "Should have returned a null set")
self.multipleLumiSubscription.acquireFiles(
self.multipleLumiSubscription.availableFiles())
jobFactory = splitter(self.multipleLumiSubscription)
jobGroups = jobFactory(trigger_time=1)
self.assertEquals(jobGroups, [], "Should have returned a null set")
self.singleLumiSubscription.acquireFiles(
self.singleLumiSubscription.availableFiles())
jobFactory = splitter(self.singleLumiSubscription)
jobGroups = jobFactory(trigger_time=1)
self.assertEquals(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.assertEquals(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.assertEquals(len(jobGroups), 1, "Should have gotten one jobGroup")
self.assertEquals(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEquals(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.assertEquals(len(jobGroups), 1, "Should have gotten one jobGroup")
self.assertEquals(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEquals(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.assertEquals(len(jobGroups), 1, "Should have gotten one jobGroup")
self.assertEquals(len(jobGroups[0].jobs), 1, \
"JobFactory should have made one job")
myfiles = jobGroups[0].jobs[0].getFiles()
self.assertEquals(len(myfiles), 9, \
"JobFactory should have provides us with 9 files")
self.assertEquals(len(myfiles), 9)
class DBCoreTest(unittest.TestCase):
def setUp(self):
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMQuality.TestDB"],
useDefault=False)
return
def tearDown(self):
"""
Delete the databases
"""
self.testInit.clearDatabase()
return
def testBuildBinds(self):
"""
Test class for DBCore.buildbinds()
"""
#This class may become obselete soon. There is a TODO stuck onto DBCore.buildbinds()
#This just checks to see that the sequence properly packages the first value is set
#So that it sets the key seqname to the proper name in the files list
#Also right now tests that it sets the keys right in each dict, but this seems redundant
# -mnorman
seqname = 'file'
dictbinds = {'lumi': 123, 'test': 100}
files = ['testA', 'testB', 'testC']
myThread = threading.currentThread()
testInterface = myThread.dbi
binds = testInterface.buildbinds(files, seqname, dictbinds)
#This should return a dict for every value in files with additional elements
#from dictbinds. We then loop over every dict (which should be equal to the
#number of elements in files), and look to see that the filename matches and that
#At least one of the dictbinds keys matches to its proper element
for i in range(len(files)):
self.assertEqual(binds[i][seqname], files[i])
self.assertEqual(binds[i][dictbinds.keys()[0]],
dictbinds[dictbinds.keys()[0]])
return
def testProcessDataNoBinds(self):
"""
_testProcessDataNoBinds_
Verify that insert and select queries work when no binds are used.
"""
insertSQL = "INSERT INTO test_tablea VALUES (1, 2, 'three')"
selectSQL = "SELECT column1, column2, column3 from test_tablea"
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL)
resultSets = myThread.dbi.processData(selectSQL)
assert len(resultSets) == 1, \
"Error: Wrong number of ResultSets returned."
results = resultSets[0].fetchall()
assert len(results) == 1, \
"Error: Wrong number of rows returned."
assert len(results[0]) == 3, \
"Error: Wrong number of columns returned."
assert results[0][0] == 1, \
"Error: Column one is wrong."
assert results[0][1] == 2, \
"Error: Column two is wrong."
assert results[0][2] == "three", \
"Error: Column three is wrong."
return
def testProcessDataOneBind(self):
"""
_testProcessDataOneBind_
Verify that insert and select queries work with one set of bind variables.
"""
bindsA = {"one": 1, "two": 2, "three": "three"}
bindsB = {"one": 3, "two": 2, "three": "one"}
insertSQL = "INSERT INTO test_tablea VALUES (:one, :two, :three)"
selectSQL = \
"""SELECT column1, column2, column3 FROM test_tablea
WHERE column1 = :one AND column2 = :two AND column3 = :three"""
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL, binds=bindsA)
myThread.dbi.processData(insertSQL, binds=bindsB)
resultSets = myThread.dbi.processData(selectSQL, bindsA)
assert len(resultSets) == 1, \
"Error: Wrong number of ResultSets returned."
results = resultSets[0].fetchall()
assert len(results) == 1, \
"Error: Wrong number of rows returned."
assert len(results[0]) == 3, \
"Error: Wrong number of columns returned."
assert results[0][0] == 1, \
"Error: Column one is wrong."
assert results[0][1] == 2, \
"Error: Column two is wrong."
assert results[0][2] == "three", \
"Error: Column three is wrong."
resultSets = myThread.dbi.processData(selectSQL, bindsB)
assert len(resultSets) == 1, \
"Error: Wrong number of ResultSets returned."
results = resultSets[0].fetchall()
assert len(results) == 1, \
"Error: Wrong number of rows returned."
assert len(results[0]) == 3, \
"Error: Wrong number of columns returned."
assert results[0][0] == 3, \
"Error: Column one is wrong."
assert results[0][1] == 2, \
"Error: Column two is wrong."
assert results[0][2] == "one", \
"Error: Column three is wrong."
return
def testProcessDataSeveralBinds(self):
"""
_testProcessDataSeveralBinds_
Verify that insert and select queries work with several binds.
"""
bindsA = [{
"one": 1,
"two": 2,
"three": "three"
}, {
"one": 3,
"two": 2,
"three": "one"
}, {
"one": 4,
"two": 5,
"three": "six"
}, {
"one": 6,
"two": 5,
"three": "four"
}]
bindsB = [{
"one": 10,
"two": 11,
"three": "twelve"
}, {
"one": 12,
"two": 11,
"three": "ten"
}]
insertSQL = "INSERT INTO test_tablea VALUES (:one, :two, :three)"
selectSQL = \
"""SELECT column1, column2, column3 FROM test_tablea
WHERE column1 = :one AND column2 = :two AND column3 = :three"""
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL, binds=bindsA)
myThread.dbi.processData(insertSQL, binds=bindsB)
resultSets = myThread.dbi.processData(selectSQL, bindsA)
assert len(resultSets) == 1, \
"Error: Wrong number of ResultSets returned."
results = resultSets[0].fetchall()
assert len(results) == 4, \
"Error: Wrong number of rows returned."
for result in results:
assert len(result) == 3, \
"Error: Wrong number of columns returned."
for bind in bindsA:
if bind["one"] == result[0] and bind["two"] == result[1] and \
bind["three"] == result[2]:
bindsA.remove(bind)
break
assert len(bindsA) == 0, \
"Error: Missing rows from select."
resultSets = myThread.dbi.processData(selectSQL, bindsB)
assert len(resultSets) == 1, \
"Error: Wrong number of ResultSets returned."
results = resultSets[0].fetchall()
assert len(results) == 2, \
"Error: Wrong number of rows returned."
for result in results:
assert len(result) == 3, \
"Error: Wrong number of columns returned."
for bind in bindsB:
if bind["one"] == result[0] and bind["two"] == result[1] and \
bind["three"] == result[2]:
bindsB.remove(bind)
break
assert len(bindsB) == 0, \
"Error: Missing rows from select."
return
def testProcessDataHugeBinds(self):
"""
_testProcessDataHugeBinds_
Verify that select and insert queries work with several thousand binds.
"""
bindsA = []
bindsB = []
for i in range(3001):
bindsA.append({"one": i, "two": i * 2, "three": str(i * 3)})
for i in range(1501):
bindsB.append({"one": (i + 1) * 2, "two": i, "three": str(i * 5)})
insertSQL = "INSERT INTO test_tablea VALUES (:one, :two, :three)"
selectSQL = \
"""SELECT column1, column2, column3 FROM test_tablea
WHERE column1 = :one AND column2 = :two AND column3 = :three"""
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL, binds=bindsA)
myThread.dbi.processData(insertSQL, binds=bindsB)
resultSets = myThread.dbi.processData(selectSQL, bindsA)
results = []
for resultSet in resultSets:
results.extend(resultSet.fetchall())
assert len(results) == 3001, \
"Error: Wrong number of rows returned: %d" % len(results)
for result in results:
assert len(result) == 3, \
"Error: Wrong number of columns returned."
for bind in bindsA:
if bind["one"] == result[0] and bind["two"] == result[1] and \
bind["three"] == result[2]:
bindsA.remove(bind)
break
assert len(bindsA) == 0, \
"Error: Missing rows from select."
resultSets = myThread.dbi.processData(selectSQL, bindsB)
results = []
for resultSet in resultSets:
results.extend(resultSet.fetchall())
assert len(results) == 1501, \
"Error: Wrong number of rows returned."
for result in results:
assert len(result) == 3, \
"Error: Wrong number of columns returned."
for bind in bindsB:
if bind["one"] == result[0] and bind["two"] == result[1] and \
bind["three"] == result[2]:
bindsB.remove(bind)
break
assert len(bindsB) == 0, \
"Error: Missing rows from select."
return
def testInsertHugeNumber(self):
"""
_testInsertHugeNumber_
Verify that we can insert and select huge numbers.
"""
insertSQL = "INSERT INTO test_bigcol VALUES(:val1)"
selectSQL = "SELECT * FROM test_bigcol"
bindsA = {"val1": 2012211901}
bindsB = {"val1": 20122119010}
myThread = threading.currentThread()
myThread.dbi.processData(insertSQL, binds=bindsA)
myThread.dbi.processData(insertSQL, binds=bindsB)
resultSets = myThread.dbi.processData(selectSQL)
results = []
for resultSet in resultSets:
for row in resultSet.fetchall():
results.append(row[0])
assert len(results) == 2, \
"Error: Wrong number of results."
assert bindsA["val1"] in results, \
"Error: Value one is missing."
assert bindsB["val1"] in results, \
"Error: Value one is missing."
return
class EventBasedTest(unittest.TestCase):
"""
_EventBasedTest_
Test event 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", 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="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="T2_CH_CERN")
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleFileFileset.addFile(newFile)
self.singleFileFileset.commit()
self.multipleFileRunset = Fileset(name="TestFileset3")
self.multipleFileRunset.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations="T2_CH_CERN")
newFile.addRun(Run(i / 3, *[45]))
newFile.create()
self.multipleFileRunset.addFile(newFile)
self.multipleFileRunset.commit()
self.singleRunFileset = Fileset(name="TestFileset4")
self.singleRunFileset.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations="T2_CH_CERN")
newFile.addRun(Run(1, *[45]))
newFile.create()
self.singleRunFileset.addFile(newFile)
self.singleRunFileset.commit()
self.singleRunMultipleLumi = Fileset(name="TestFileset5")
self.singleRunMultipleLumi.create()
for i in range(10):
newFile = File(makeUUID(),
size=1000,
events=100,
locations="T2_CH_CERN")
newFile.addRun(Run(1, *[45 + i]))
newFile.create()
self.singleRunMultipleLumi.addFile(newFile)
self.singleRunMultipleLumi.commit()
testWorkflow = Workflow(spec="spec.xml",
owner="mnorman",
name="wf001",
task="Test")
testWorkflow.create()
self.multipleFileSubscription = Subscription(
fileset=self.multipleFileFileset,
workflow=testWorkflow,
split_algo="RunBased",
type="Processing")
self.singleFileSubscription = Subscription(
fileset=self.singleFileFileset,
workflow=testWorkflow,
split_algo="RunBased",
type="Processing")
self.multipleRunSubscription = Subscription(
fileset=self.multipleFileRunset,
workflow=testWorkflow,
split_algo="RunBased",
type="Processing")
self.singleRunSubscription = Subscription(
fileset=self.singleRunFileset,
workflow=testWorkflow,
split_algo="RunBased",
type="Processing")
self.singleRunMultipleLumiSubscription = Subscription(
fileset=self.singleRunMultipleLumi,
workflow=testWorkflow,
split_algo="RunBased",
type="Processing")
self.multipleFileSubscription.create()
self.singleFileSubscription.create()
self.multipleRunSubscription.create()
self.singleRunSubscription.create()
self.singleRunMultipleLumiSubscription.create()
return
def tearDown(self):
"""
_tearDown_
Tear down WMBS architechture.
"""
self.testInit.clearDatabase()
return
def testExactRuns(self):
"""
_testExactRuns_
Test run based job splitting when the number of events per job is
exactly the same as the number of events in the input file.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleFileSubscription)
jobGroups = jobFactory(files_per_job=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."
return
def testMoreRuns(self):
"""
_testMoreEvents_
Test run based job splitting when the number of runs per job is
greater than the number of runs in the input file.
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleFileSubscription)
jobGroups = jobFactory(files_per_job=2)
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."
return
def testMultipleRuns(self):
"""
_testMultipleRuns_
Test run based job splitting when the number of runs is
equal to the number in each input file, with multiple files
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleFileSubscription)
jobGroups = jobFactory(files_per_job=1)
assert len(jobGroups) == 10, \
"ERROR: JobFactory didn't return one JobGroup per run."
assert len(jobGroups[0].jobs) == 1, \
"ERROR: JobFactory didn't put each run in a file."
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 1)
return
def testMultipleRunsCombine(self):
"""
_testMultipleRunsCombine_
Test run based job splitting when the number of jobs is
less then the number of files, with multiple files
"""
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.multipleRunSubscription)
jobGroups = jobFactory(files_per_job=2)
assert len(jobGroups) == 4, \
"ERROR: JobFactory didn't return one JobGroup per run."
assert len(jobGroups[1].jobs) == 2, \
"ERROR: JobFactory didn't put only one job in the first job"
#Last one in the queue should have one job, previous two (three files per run)
self.assertEqual(len(jobGroups[1].jobs.pop().getFiles(type="lfn")), 1)
self.assertEqual(len(jobGroups[1].jobs.pop().getFiles(type="lfn")), 2)
return
def testSingleRunsCombineUneven(self):
"""
_testSingleRunsCombineUneven_
Test run based job splitting when the number of jobs is
less then and indivisible by the number of files, with multiple files.
"""
#This should return two jobs, one with 8 and one with 2 files
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleRunSubscription)
jobGroups = jobFactory(files_per_job=8)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 8)
return
def testPersistSingleRunsCombineUneven(self):
"""
_testPerisistSingleRunsCombineUneven_
Test run based job splitting when the number of jobs is
less then and indivisible by the number of files, with multiple files.
"""
#This should return two jobs, one with 8 and one with 2 files
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS",
subscription=self.singleRunSubscription)
jobGroups = jobFactory(files_per_job=8)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 8)
return
def testSingleRunsMultipleLumiCombineUneven(self):
"""
_testSingleRunsMultipeLumiCombineUneven_
Test run based job splitting when the number of jobs is
less then and indivisible by the number of files, with multiple files.
"""
#This should return two jobs, one with 8 and one with 2 files
splitter = SplitterFactory()
jobFactory = splitter(
package="WMCore.WMBS",
subscription=self.singleRunMultipleLumiSubscription)
jobGroups = jobFactory(files_per_job=8)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 2)
self.assertEqual(len(jobGroups[0].jobs.pop().getFiles(type="lfn")), 8)
return
class StoreResultsTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Initialize the database.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMCore.WMBS"],
useDefault=False)
self.testDir = self.testInit.generateWorkDir()
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package="WMCore.WMBS",
logger=myThread.logger,
dbinterface=myThread.dbi)
self.listTasksByWorkflow = self.daoFactory(
classname="Workflow.LoadFromName")
self.listFilesets = self.daoFactory(classname="Fileset.List")
self.listSubsMapping = self.daoFactory(
classname="Subscriptions.ListSubsAndFilesetsFromWorkflow")
return
def tearDown(self):
"""
_tearDown_
Clear out the database.
"""
self.testInit.clearDatabase()
self.testInit.delWorkDir()
return
def testStoreResults(self):
"""
_testStoreResults_
Create a StoreResults workflow and verify it installs into WMBS
correctly.
"""
arguments = StoreResultsWorkloadFactory.getTestArguments()
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": "USER"}
for goldenOutputMod, tier in goldenOutputMods.items():
fset = goldenOutputMod + tier
mergedOutput = testWorkflow.outputMap[fset][0][
"merged_output_fileset"]
unmergedOutput = testWorkflow.outputMap[fset][0]["output_fileset"]
mergedOutput.loadData()
unmergedOutput.loadData()
self.assertEqual(
mergedOutput.name,
"/TestWorkload/StoreResults/merged-%s" % fset,
"Error: Merged output fileset is wrong: %s" %
mergedOutput.name)
self.assertEqual(
unmergedOutput.name,
"/TestWorkload/StoreResults/merged-%s" % fset,
"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 testFilesets(self):
"""
Test workflow tasks, filesets and subscriptions creation
"""
# expected tasks, filesets, subscriptions, etc
expOutTasks = ['/TestWorkload/StoreResults']
expWfTasks = [
'/TestWorkload/StoreResults',
'/TestWorkload/StoreResults/StoreResultsLogCollect'
]
expFsets = [
'TestWorkload-StoreResults-/MinimumBias/ComissioningHI-v1/RAW',
'/TestWorkload/StoreResults/merged-MergedUSER',
'/TestWorkload/StoreResults/merged-logArchive'
]
subMaps = [
(2, '/TestWorkload/StoreResults/merged-logArchive',
'/TestWorkload/StoreResults/StoreResultsLogCollect',
'MinFileBased', 'LogCollect'),
(1, 'TestWorkload-StoreResults-/MinimumBias/ComissioningHI-v1/RAW',
'/TestWorkload/StoreResults', 'ParentlessMergeBySize', 'Merge')
]
testArguments = StoreResultsWorkloadFactory.getTestArguments()
factory = StoreResultsWorkloadFactory()
testWorkload = factory.factoryWorkloadConstruction(
"TestWorkload", testArguments)
testWMBSHelper = WMBSHelper(testWorkload,
"StoreResults",
blockName=testArguments['InputDataset'],
cachepath=self.testInit.testDir)
testWMBSHelper.createTopLevelFileset()
testWMBSHelper._createSubscriptionsInWMBS(
testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)
self.assertItemsEqual(testWorkload.listOutputProducingTasks(),
expOutTasks)
workflows = self.listTasksByWorkflow.execute(workflow="TestWorkload")
self.assertItemsEqual([item['task'] for item in workflows], expWfTasks)
# returns a tuple of id, name, open and last_update
filesets = self.listFilesets.execute()
self.assertItemsEqual([item[1] for item in filesets], expFsets)
subscriptions = self.listSubsMapping.execute(workflow="TestWorkload",
returnTuple=True)
self.assertItemsEqual(subscriptions, subMaps)
class ThreadPoolTest(unittest.TestCase):
"""
_ThreadPool_t_
Unit tests for threadpool
"""
_nrOfThreads = 10
_nrOfPools = 5
def setUp(self):
"make a logger instance and create tables"
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema()
def tearDown(self):
"""
Deletion of database
"""
# FIXME: this might not work if your not using socket.
self.testInit.clearDatabase()
def testA(self):
"""
__testSubscribe__
Test subscription of a component.
"""
raise nose.SkipTest
myThread = threading.currentThread()
# create a 'fake' component that contains a arg dictionary.
component = Dummy()
# load default parameters.
config = self.testInit.getConfiguration()
# normally assigned by the harness of the test component.
config.Agent.componentName = "TestComponent"
component.config = config
threadPools = []
for i in xrange(0, ThreadPoolTest._nrOfPools):
threadPool = ThreadPool("WMCore.ThreadPool.ThreadSlave", \
component, 'MyPool_'+str(i), ThreadPoolTest._nrOfThreads)
threadPools.append(threadPool)
# this is how you would use the threadpool. The threadpool retrieves
# events/payloads from the message service. If a thread is available
# it is dispatched, otherwise it is stored in the trheadpool.
# make the number of tasks bigger than number of threads to tesT
# the persistent queue.
for i in xrange(0, ThreadPoolTest._nrOfThreads * 10):
event = 'eventNr_' + str(i)
payload = 'payloadNr_' + str(i)
# normally you would have different events per threadpool and
# even different objects per pool. the payload part will be
# pickled into the database enabling flexibility in passing
# information.
for j in xrange(0, ThreadPoolTest._nrOfPools):
threadPools[j].enqueue(event, \
{'event' : event, 'payload' : payload})
# this commit you want to be in the agent harness, so the message is
# actual removed from the msgService. we can do this as the threadpool
# acts as a dispatcher and is a shortlived action: dispatch to thread
# or queue and tell agent harness it is finished.
finished = False
timeout = 60 # secs
currenttime = 0
while not finished:
print('waiting for threads to finishs. Work left:')
for j in xrange(0, ThreadPoolTest._nrOfPools):
print('pool_' + str(j) + ':' + str(threadPools[j].callQueue))
time.sleep(1)
finished = True
currenttime += 1
if (timeout == currenttime):
raise RuntimeError
for j in xrange(0, ThreadPoolTest._nrOfPools):
if (len(threadPools[j].resultsQueue) <
ThreadPoolTest._nrOfThreads * 10):
finished = False
break
# check if the tables are really empty and all messages
# have been processed.
for j in xrange(0, ThreadPoolTest._nrOfPools):
assert len(threadPools[j].resultsQueue) == \
ThreadPoolTest._nrOfThreads*10
myThread.transaction.begin()
for j in xrange(0, ThreadPoolTest._nrOfPools):
self.assertEqual(threadPools[j].countMessages(), 0)
myThread.transaction.commit()
class WMAgentTest(unittest.TestCase):
"""
_WMAgentTest_
Global unittest for all WMAgent components
"""
# This is an integration test
__integration__ = "Any old bollocks"
sites = ['T2_US_Florida', 'T2_US_UCSD', 'T2_TW_Taiwan', 'T1_CH_CERN']
components = ['JobCreator', 'JobSubmitter', 'JobTracker',
'JobAccountant', 'JobArchiver', 'TaskArchiver',
'RetryManager', 'ErrorHandler']
def setUp(self):
"""
_setUp_
Set up vital components
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMCore.WMBS",'WMCore.MsgService',
'WMCore.ResourceControl', 'WMCore.ThreadPool',
'WMCore.Agent.Database'],
useDefault = False)
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package = "WMCore.WMBS",
logger = myThread.logger,
dbinterface = myThread.dbi)
locationAction = self.daoFactory(classname = "Locations.New")
pendingSlots = self.daoFactory(classname = "Locations.SetPendingSlots")
for site in self.sites:
locationAction.execute(siteName = site, seName = 'se.%s' % (site), ceName = site)
pendingSlots.execute(siteName = site, pendingSlots = 1000)
#Create sites in resourceControl
resourceControl = ResourceControl()
for site in self.sites:
resourceControl.insertSite(siteName = site, seName = 'se.%s' % (site), ceName = site)
resourceControl.insertThreshold(siteName = site, taskType = 'Processing', \
maxSlots = 10000, pendingSlots = 10000)
self.testDir = self.testInit.generateWorkDir()
# Set heartbeat
for component in self.components:
heartbeatAPI = HeartbeatAPI(component)
heartbeatAPI.registerComponent()
return
def tearDown(self):
"""
_tearDown_
Tear down everything and go home.
"""
self.testInit.clearDatabase()
self.testInit.delWorkDir()
return
def createTestWorkload(self, workloadName = 'Test', emulator = True):
"""
_createTestWorkload_
Creates a test workload for us to run on, hold the basic necessities.
"""
workload = testWorkload("TestWorkload")
rereco = workload.getTask("ReReco")
taskMaker = TaskMaker(workload, os.path.join(self.testDir, 'workloadTest'))
taskMaker.skipSubscription = True
taskMaker.processWorkload()
workload.save(workloadName)
return workload
def getConfig(self):
"""
_getConfig_
This is the global test configuration object
"""
config = Configuration()
config.component_("Agent")
config.Agent.WMSpecDirectory = self.testDir
config.Agent.agentName = 'testAgent'
config.Agent.componentName = 'test'
# First the general stuff
config.section_("General")
config.General.workDir = os.getenv("TESTDIR", self.testDir)
# Now the CoreDatabase information
# This should be the dialect, dburl, etc
config.section_("CoreDatabase")
config.CoreDatabase.connectUrl = os.getenv("DATABASE")
config.CoreDatabase.socket = os.getenv("DBSOCK")
# JobCreator
config.component_("JobCreator")
config.JobCreator.namespace = 'WMComponent.JobCreator.JobCreator'
config.JobCreator.logLevel = 'DEBUG'
config.JobCreator.maxThreads = 1
config.JobCreator.UpdateFromResourceControl = True
config.JobCreator.pollInterval = 10
config.JobCreator.jobCacheDir = self.testDir
config.JobCreator.defaultJobType = 'processing' #Type of jobs that we run, used for resource control
config.JobCreator.workerThreads = 2
config.JobCreator.componentDir = os.path.join(os.getcwd(), 'Components')
# JobSubmitter
config.component_("JobSubmitter")
config.JobSubmitter.namespace = 'WMComponent.JobSubmitter.JobSubmitter'
config.JobSubmitter.logLevel = 'INFO'
config.JobSubmitter.maxThreads = 1
config.JobSubmitter.pollInterval = 10
config.JobSubmitter.pluginName = 'CondorGlobusPlugin'
config.JobSubmitter.pluginDir = 'JobSubmitter.Plugins'
config.JobSubmitter.submitDir = os.path.join(self.testDir, 'submit')
config.JobSubmitter.submitNode = os.getenv("HOSTNAME", 'badtest.fnal.gov')
config.JobSubmitter.submitScript = os.path.join(getWMBASE(),
'test/python/WMComponent_t/JobSubmitter_t',
'submit.sh')
config.JobSubmitter.componentDir = os.path.join(os.getcwd(), 'Components')
config.JobSubmitter.workerThreads = 2
config.JobSubmitter.jobsPerWorker = 200
# JobTracker
config.component_("JobTracker")
config.JobTracker.logLevel = 'DEBUG'
config.JobTracker.pollInterval = 10
config.JobTracker.trackerName = 'CondorTracker'
config.JobTracker.pluginDir = 'WMComponent.JobTracker.Plugins'
config.JobTracker.componentDir = os.path.join(os.getcwd(), 'Components')
config.JobTracker.runTimeLimit = 7776000 #Jobs expire after 90 days
config.JobTracker.idleTimeLimit = 7776000
config.JobTracker.heldTimeLimit = 7776000
config.JobTracker.unknTimeLimit = 7776000
# JobAccountant
config.component_("JobAccountant")
config.JobAccountant.pollInterval = 60
config.JobAccountant.componentDir = os.path.join(os.getcwd(), 'Components')
config.JobAccountant.logLevel = 'INFO'
# JobArchiver
config.component_("JobArchiver")
config.JobArchiver.pollInterval = 60
config.JobArchiver.logLevel = 'INFO'
config.JobArchiver.logDir = os.path.join(self.testDir, 'logs')
config.JobArchiver.componentDir = os.path.join(os.getcwd(), 'Components')
config.JobArchiver.numberOfJobsToCluster = 1000
# Task Archiver
config.component_("TaskArchiver")
config.TaskArchiver.componentDir = self.testInit.generateWorkDir()
config.TaskArchiver.WorkQueueParams = {}
config.TaskArchiver.pollInterval = 60
config.TaskArchiver.logLevel = 'INFO'
config.TaskArchiver.timeOut = 0
# JobStateMachine
config.component_('JobStateMachine')
config.JobStateMachine.couchurl = os.getenv('COUCHURL',
'mnorman:[email protected]:5984')
config.JobStateMachine.couchDBName = "mnorman_test"
# Needed, because this is a test
os.makedirs(config.JobSubmitter.submitDir)
return config
def createFileCollection(self, name, nSubs, nFiles, workflowURL = 'test', site = None):
"""
_createFileCollection_
Create a collection of files for splitting into jobs
"""
myThread = threading.currentThread()
testWorkflow = Workflow(spec = workflowURL, owner = "mnorman",
name = name, task="/TestWorkload/ReReco")
testWorkflow.create()
for sub in range(nSubs):
nameStr = '%s-%i' % (name, sub)
testFileset = Fileset(name = nameStr)
testFileset.create()
for f in range(nFiles):
# pick a random site
if not site:
tmpSite = 'se.%s' % (random.choice(self.sites))
else:
tmpSite = 'se.%s' % (site)
testFile = File(lfn = "/lfn/%s/%i" % (nameStr, f), size = 1024, events = 10)
testFile.setLocation(tmpSite)
testFile.create()
testFileset.addFile(testFile)
testFileset.commit()
testFileset.markOpen(isOpen = 0)
testSubscription = Subscription(fileset = testFileset,
workflow = testWorkflow,
type = "Processing",
split_algo = "FileBased")
testSubscription.create()
return
def createReports(self, jobs, retryCount = 0):
"""
_createReports_
Create some dummy job reports for each job
"""
report = Report()
report.addStep('testStep', 0)
for job in jobs:
#reportPath = os.path.join(job['cache_dir'], 'Report.%i.pkl' % (retryCount))
reportPath = job['fwjr_path']
if os.path.exists(reportPath):
os.remove(reportPath)
report.save(reportPath)
return
def testA_StraightThrough(self):
"""
_StraightThrough_
Just run everything straight through without any variations
"""
# Do pre-submit job check
nRunning = getCondorRunningJobs()
self.assertEqual(nRunning, 0, "User currently has %i running jobs. Test will not continue" % (nRunning))
myThread = threading.currentThread()
workload = self.createTestWorkload()
config = self.getConfig()
name = 'WMAgent_Test1'
site = self.sites[0]
nSubs = 5
nFiles = 10
workloadPath = os.path.join(self.testDir, 'workloadTest',
'TestWorkload', 'WMSandbox',
'WMWorkload.pkl')
# Create a collection of files
self.createFileCollection(name = name, nSubs = nSubs,
nFiles = nFiles,
workflowURL = workloadPath,
site = site)
############################################################
# Test the JobCreator
config.Agent.componentName = 'JobCreator'
testJobCreator = JobCreatorPoller(config = config)
testJobCreator.algorithm()
time.sleep(5)
# Did all jobs get created?
getJobsAction = self.daoFactory(classname = "Jobs.GetAllJobs")
result = getJobsAction.execute(state = 'Created', jobType = "Processing")
self.assertEqual(len(result), nSubs*nFiles)
# Count database objects
result = myThread.dbi.processData('SELECT * FROM wmbs_sub_files_acquired')[0].fetchall()
self.assertEqual(len(result), nSubs * nFiles)
# Find the test directory
testDirectory = os.path.join(self.testDir, 'TestWorkload', 'ReReco')
self.assertTrue('JobCollection_1_0' in os.listdir(testDirectory))
self.assertTrue(len(os.listdir(testDirectory)) <= 20)
groupDirectory = os.path.join(testDirectory, 'JobCollection_1_0')
# First job should be in here
self.assertTrue('job_1' in os.listdir(groupDirectory))
jobFile = os.path.join(groupDirectory, 'job_1', 'job.pkl')
self.assertTrue(os.path.isfile(jobFile))
f = open(jobFile, 'r')
job = cPickle.load(f)
f.close()
self.assertEqual(job['workflow'], name)
self.assertEqual(len(job['input_files']), 1)
self.assertEqual(os.path.basename(job['sandbox']), 'TestWorkload-Sandbox.tar.bz2')
###############################################################
# Now test the JobSubmitter
config.Agent.componentName = 'JobSubmitter'
testJobSubmitter = JobSubmitterPoller(config = config)
testJobSubmitter.algorithm()
# Check that jobs are in the right state
result = getJobsAction.execute(state = 'Created', jobType = "Processing")
self.assertEqual(len(result), 0)
result = getJobsAction.execute(state = 'Executing', jobType = "Processing")
self.assertEqual(len(result), nSubs * nFiles)
# Check assigned locations
getLocationAction = self.daoFactory(classname = "Jobs.GetLocation")
for id in result:
loc = getLocationAction.execute(jobid = id)
self.assertEqual(loc, [[site]])
# Check to make sure we have running jobs
nRunning = getCondorRunningJobs()
self.assertEqual(nRunning, nFiles * nSubs)
#################################################################
# Now the JobTracker
config.Agent.componentName = 'JobTracker'
testJobTracker = JobTrackerPoller(config = config)
testJobTracker.setup()
testJobTracker.algorithm()
# Running the algo without removing the jobs should do nothing
result = getJobsAction.execute(state = 'Executing', jobType = "Processing")
self.assertEqual(len(result), nSubs * nFiles)
condorRM()
time.sleep(1)
# All jobs gone?
nRunning = getCondorRunningJobs()
self.assertEqual(nRunning, 0)
testJobTracker.algorithm()
time.sleep(5)
# Running the algo without removing the jobs should do nothing
result = getJobsAction.execute(state = 'Executing', jobType = "Processing")
self.assertEqual(len(result), 0)
result = getJobsAction.execute(state = 'Complete', jobType = "Processing")
self.assertEqual(len(result), nSubs * nFiles)
#################################################################
# Now the JobAccountant
# First you need to load all jobs
self.getFWJRAction = self.daoFactory(classname = "Jobs.GetFWJRByState")
completeJobs = self.getFWJRAction.execute(state = "complete")
# Create reports for all jobs
self.createReports(jobs = completeJobs, retryCount = 0)
config.Agent.componentName = 'JobAccountant'
testJobAccountant = JobAccountantPoller(config = config)
testJobAccountant.setup()
# It should do something with the jobs
testJobAccountant.algorithm()
# All the jobs should be done now
result = getJobsAction.execute(state = 'Complete', jobType = "Processing")
self.assertEqual(len(result), 0)
result = getJobsAction.execute(state = 'Success', jobType = "Processing")
self.assertEqual(len(result), nSubs * nFiles)
#######################################################################
# Now the JobArchiver
config.Agent.componentName = 'JobArchiver'
testJobArchiver = JobArchiverPoller(config = config)
testJobArchiver.algorithm()
# All the jobs should be cleaned up
result = getJobsAction.execute(state = 'Success', jobType = "Processing")
self.assertEqual(len(result), 0)
result = getJobsAction.execute(state = 'Cleanout', jobType = "Processing")
self.assertEqual(len(result), nSubs * nFiles)
logDir = os.path.join(self.testDir, 'logs')
for job in completeJobs:
self.assertFalse(os.path.exists(job['fwjr_path']))
jobFolder = 'JobCluster_%i' \
% (int(job['id']/config.JobArchiver.numberOfJobsToCluster))
jobPath = os.path.join(logDir, jobFolder, 'Job_%i.tar' %(job['id']))
self.assertTrue(os.path.isfile(jobPath))
self.assertTrue(os.path.getsize(jobPath) > 0)
###########################################################################
# Now the TaskAchiver
config.Agent.componentName = 'TaskArchiver'
testTaskArchiver = TaskArchiverPoller(config = config)
testTaskArchiver.algorithm()
result = getJobsAction.execute(state = 'Cleanout', jobType = "Processing")
self.assertEqual(len(result), 0)
for jdict in completeJobs:
job = Job(id = jdict['id'])
self.assertFalse(job.exists())
if os.path.isdir('testDir'):
shutil.rmtree('testDir')
shutil.copytree('%s' %self.testDir, os.path.join(os.getcwd(), 'testDir'))
return
class FeederManagerTest(unittest.TestCase):
"""
TestCase for TestFeederManager module
"""
_maxMessage = 10
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all needed
tables.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.generateWorkDir()
self.testInit.setSchema(customModules = \
['WMCore.Agent.Database',
'WMComponent.FeederManager.Database',
'WMCore.ThreadPool',
'WMCore.WMBS'],
useDefault = False)
return
def tearDown(self):
"""
_tearDown_
Database deletion
"""
self.testInit.clearDatabase()
return
def getConfig(self):
"""
_createConfig_
Create a config for the JobAccountant. This config needs to include
information for connecting to the database as the component will create
it's own database connections. These parameters are still pulled from
the environment.
"""
config = self.testInit.getConfiguration()
self.testInit.generateWorkDir(config)
config.component_("FeederManager")
config.FeederManager.logLevel = "INFO"
config.FeederManager.componentName = "FeederManager"
config.FeederManager.componentDir = \
os.path.join(os.getenv("TESTDIR"), "FeederManager")
config.FeederManager.addDatasetWatchHandler = \
'WMComponent.FeederManager.Handler.DefaultAddDatasetWatch'
# The maximum number of threads to process each message type
config.FeederManager.maxThreads = 10
# The poll interval at which to look for new fileset/feeder association
config.FeederManager.pollInterval = 60
return config
def testA(self):
"""
_testA_
Handle AddDatasetWatch events
"""
myThread = threading.currentThread()
config = self.getConfig()
testFeederManager = FeederManager(config)
testFeederManager.prepareToStart()
for i in xrange(0, FeederManagerTest._maxMessage):
for j in xrange(0, 3):
feederManagerdict = {'payload':{'FeederType':'NO Feeder',
'dataset' : 'NO DATASET', 'FileType' : 'NO FILE TYPE',
'StartRun' : 'NO START RUN' }}
testFeederManager.handleMessage( type = 'AddDatasetWatch',
payload = feederManagerdict )
time.sleep(30)
myThread.workerThreadManager.terminateWorkers()
while threading.activeCount() > 1:
print('Currently: '+str(threading.activeCount())+\
' Threads. Wait until all our threads have finished')
time.sleep(1)
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)
self.nSites = 2
locationAction = daofactory(classname="Locations.New")
for site in range(self.nSites):
locationAction.execute(siteName="site%i" % site,
pnn="T2_CH_CERN_%i" % site)
return
def tearDown(self):
"""
_tearDown_
Clear out WMBS.
"""
self.testInit.clearDatabase(modules=["WMCore.WMBS"])
return
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(["T2_CH_CERN_%i" % 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 testA_ExactFiles(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.
"""
nFiles = 5
sub = self.createTestSubscription(nFiles=nFiles)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=sub)
jobGroups = jobFactory(files_per_job=nFiles)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 1)
self.assertEqual(len(jobGroups[0].jobs[0]['input_files']), nFiles)
return
def testB_LessFilesOpen(self):
"""
_LessFilesOpen_
Test with less files then required.
If the fileset is open, this should produce no jobs.
"""
sub = self.createTestSubscription(nFiles=5)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=sub)
jobGroups = jobFactory(files_per_job=10)
self.assertEqual(len(jobGroups), 0)
return
def testC_LessFilesClosed(self):
"""
_LessFilesClosed_
Test with less files then required.
If the fileset is closed, this should produce one job.
"""
sub = self.createTestSubscription(nFiles=5, closeFileset=True)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=sub)
jobGroups = jobFactory(files_per_job=10)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 1)
self.assertEqual(len(jobGroups[0].jobs[0]['input_files']), 5)
return
def testD_MoreFilesOpen(self):
"""
_MoreFilesOpen_
If you pass it more files then files_per_job, it should produce
jobs until it hits the limit, then stop.
"""
sub = self.createTestSubscription(nFiles=10)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=sub)
jobGroups = jobFactory(files_per_job=3)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 3)
for job in jobGroups[0].jobs:
self.assertEqual(len(job['input_files']), 3)
return
def testE_MoreFilesClosed(self):
"""
_MoreFilesClosed_
If you pass it more files then files_per_job, it should produce
jobs enough to hold all the files in the fileset if the
fileset is closed
"""
sub = self.createTestSubscription(nFiles=10, closeFileset=True)
splitter = SplitterFactory()
jobFactory = splitter(package="WMCore.WMBS", subscription=sub)
jobGroups = jobFactory(files_per_job=3)
self.assertEqual(len(jobGroups), 1)
self.assertEqual(len(jobGroups[0].jobs), 4)
for job in jobGroups[0].jobs[:3]:
self.assertEqual(len(job['input_files']), 3)
self.assertEqual(len(jobGroups[0].jobs[3]['input_files']), 1)
return
class DBSUploadTest(unittest.TestCase):
"""
TestCase for DBSUpload module
"""
def setUp(self):
"""
_setUp_
setUp function for unittest
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMComponent.DBS3Buffer"],
useDefault = False)
self.testDir = self.testInit.generateWorkDir(deleteOnDestruction = False)
self.configFile = EmulatorSetup.setupWMAgentConfig()
myThread = threading.currentThread()
self.bufferFactory = DAOFactory(package = "WMComponent.DBS3Buffer",
logger = myThread.logger,
dbinterface = myThread.dbi)
locationAction = self.bufferFactory(classname = "DBSBufferFiles.AddLocation")
locationAction.execute(siteName = "se1.cern.ch")
locationAction.execute(siteName = "se1.fnal.gov")
locationAction.execute(siteName = "malpaquet")
self.dbsUrl = "https://*****:*****@attr("integration")
def testBasicUpload(self):
"""
_testBasicUpload_
Verify that we can successfully upload to DBS3. Also verify that the
uploader correctly handles files parentage when uploading.
"""
self.dbsApi = DbsApi(url = self.dbsUrl)
config = self.getConfig()
dbsUploader = DBSUploadPoller(config = config)
# First test verifies that uploader will poll and then not do anything
# as the database is empty.
dbsUploader.algorithm()
acqEra = "Summer%s" % (int(time.time()))
parentFiles = self.createParentFiles(acqEra)
# The algorithm needs to be run twice. On the first iteration it will
# create all the blocks and upload one. On the second iteration it will
# timeout and upload the second block.
dbsUploader.algorithm()
time.sleep(5)
dbsUploader.algorithm()
time.sleep(5)
# Verify the files made it into DBS3.
self.verifyData(parentFiles[0]["datasetPath"], parentFiles)
# Inject some more parent files and some child files into DBSBuffer.
# Run the uploader twice, only the parent files should be added to DBS3.
(moreParentFiles, childFiles) = \
self.createFilesWithChildren(parentFiles, acqEra)
dbsUploader.algorithm()
time.sleep(5)
dbsUploader.algorithm()
time.sleep(5)
self.verifyData(parentFiles[0]["datasetPath"],
parentFiles + moreParentFiles)
# Run the uploader another two times to upload the child files. Verify
# that the child files were uploaded.
dbsUploader.algorithm()
time.sleep(5)
dbsUploader.algorithm()
time.sleep(5)
self.verifyData(childFiles[0]["datasetPath"], childFiles)
return
@attr("integration")
def testDualUpload(self):
"""
_testDualUpload_
Verify that the dual upload mode works correctly.
"""
self.dbsApi = DbsApi(url = self.dbsUrl)
config = self.getConfig()
dbsUploader = DBSUploadPoller(config = config)
dbsUtil = DBSBufferUtil()
# First test verifies that uploader will poll and then not do anything
# as the database is empty.
dbsUploader.algorithm()
acqEra = "Summer%s" % (int(time.time()))
parentFiles = self.createParentFiles(acqEra)
(moreParentFiles, childFiles) = \
self.createFilesWithChildren(parentFiles, acqEra)
allFiles = parentFiles + moreParentFiles
allBlocks = []
for i in range(4):
DBSBufferDataset(parentFiles[0]["datasetPath"]).create()
blockName = parentFiles[0]["datasetPath"] + "#" + makeUUID()
dbsBlock = DBSBufferBlock(blockName,
location = "malpaquet",
datasetpath = None)
dbsBlock.status = "Open"
dbsBlock.setDataset(parentFiles[0]["datasetPath"], 'data', 'VALID')
dbsUtil.createBlocks([dbsBlock])
for file in allFiles[i * 5 : (i * 5) + 5]:
dbsBlock.addFile(file, 'data', 'VALID')
dbsUtil.setBlockFiles({"block": blockName, "filelfn": file["lfn"]})
if i < 2:
dbsBlock.status = "InDBS"
dbsUtil.updateBlocks([dbsBlock])
dbsUtil.updateFileStatus([dbsBlock], "InDBS")
allBlocks.append(dbsBlock)
DBSBufferDataset(childFiles[0]["datasetPath"]).create()
blockName = childFiles[0]["datasetPath"] + "#" + makeUUID()
dbsBlock = DBSBufferBlock(blockName,
location = "malpaquet",
datasetpath = None)
dbsBlock.status = "InDBS"
dbsBlock.setDataset(childFiles[0]["datasetPath"], 'data', 'VALID')
dbsUtil.createBlocks([dbsBlock])
for file in childFiles:
dbsBlock.addFile(file, 'data', 'VALID')
dbsUtil.setBlockFiles({"block": blockName, "filelfn": file["lfn"]})
dbsUtil.updateFileStatus([dbsBlock], "InDBS")
dbsUploader.algorithm()
time.sleep(5)
dbsUploader.algorithm()
time.sleep(5)
self.verifyData(parentFiles[0]["datasetPath"], parentFiles)
# Change the status of the rest of the parent blocks so we can upload
# them and the children.
for dbsBlock in allBlocks:
dbsBlock.status = "InDBS"
dbsUtil.updateBlocks([dbsBlock])
dbsUploader.algorithm()
time.sleep(5)
self.verifyData(parentFiles[0]["datasetPath"], parentFiles + moreParentFiles)
# Run the uploader one more time to upload the children.
dbsUploader.algorithm()
time.sleep(5)
self.verifyData(childFiles[0]["datasetPath"], childFiles)
return
def testCloseSettingsPerWorkflow(self):
"""
_testCloseSettingsPerWorkflow_
Test the block closing mechanics in the DBS3 uploader,
this uses a fake dbs api to avoid reliance on external services.
"""
# Signal trapExit that we are a friend
os.environ["DONT_TRAP_EXIT"] = "True"
try:
# Monkey patch the imports of DbsApi
from WMComponent.DBS3Buffer import DBSUploadPoller as MockDBSUploadPoller
MockDBSUploadPoller.DbsApi = MockDbsApi
# Set the poller and the dbsUtil for verification
myThread = threading.currentThread()
(_, dbsFilePath) = mkstemp(dir = self.testDir)
self.dbsUrl = dbsFilePath
config = self.getConfig()
dbsUploader = MockDBSUploadPoller.DBSUploadPoller(config = config)
dbsUtil = DBSBufferUtil()
# First test is event based limits and timeout with no new files.
# Set the files and workflow
acqEra = "TropicalSeason%s" % (int(time.time()))
workflowName = 'TestWorkload%s' % (int(time.time()))
taskPath = '/%s/TestProcessing' % workflowName
self.injectWorkflow(workflowName, taskPath,
MaxWaitTime = 2, MaxFiles = 100,
MaxEvents = 150)
self.createParentFiles(acqEra, nFiles = 20,
workflowName = workflowName,
taskPath = taskPath)
# The algorithm needs to be run twice. On the first iteration it will
# create all the blocks and upload one with less than 150 events.
# On the second iteration the second block is uploaded.
dbsUploader.algorithm()
dbsUploader.checkBlocks()
openBlocks = dbsUtil.findOpenBlocks()
self.assertEqual(len(openBlocks), 1)
globalFiles = myThread.dbi.processData("SELECT id FROM dbsbuffer_file WHERE status = 'InDBS'")[0].fetchall()
notUploadedFiles = myThread.dbi.processData("SELECT id FROM dbsbuffer_file WHERE status = 'NOTUPLOADED'")[0].fetchall()
self.assertEqual(len(globalFiles), 14)
self.assertEqual(len(notUploadedFiles), 6)
# Check the fake DBS for data
fakeDBS = open(self.dbsUrl, 'r')
fakeDBSInfo = json.load(fakeDBS)
fakeDBS.close()
self.assertEqual(len(fakeDBSInfo), 2)
for block in fakeDBSInfo:
self.assertTrue('block_events' not in block['block'])
self.assertEqual(block['block']['file_count'], 7)
self.assertEqual(block['block']['open_for_writing'], 0)
self.assertTrue('close_settings' not in block)
time.sleep(3)
dbsUploader.algorithm()
dbsUploader.checkBlocks()
openBlocks = dbsUtil.findOpenBlocks()
self.assertEqual(len(openBlocks), 0)
fakeDBS = open(self.dbsUrl, 'r')
fakeDBSInfo = json.load(fakeDBS)
fakeDBS.close()
self.assertEqual(len(fakeDBSInfo), 3)
for block in fakeDBSInfo:
if block['block']['file_count'] != 6:
self.assertEqual(block['block']['file_count'], 7)
self.assertTrue('block_events' not in block['block'])
self.assertEqual(block['block']['open_for_writing'], 0)
self.assertTrue('close_settings' not in block)
# Now check the limit by size and timeout with new files
acqEra = "TropicalSeason%s" % (int(time.time()))
workflowName = 'TestWorkload%s' % (int(time.time()))
taskPath = '/%s/TestProcessing' % workflowName
self.injectWorkflow(workflowName, taskPath,
MaxWaitTime = 2, MaxFiles = 5,
MaxEvents = 200000000)
self.createParentFiles(acqEra, nFiles = 16,
workflowName = workflowName,
taskPath = taskPath)
dbsUploader.algorithm()
dbsUploader.checkBlocks()
openBlocks = dbsUtil.findOpenBlocks()
self.assertEqual(len(openBlocks), 1)
fakeDBS = open(self.dbsUrl, 'r')
fakeDBSInfo = json.load(fakeDBS)
fakeDBS.close()
self.assertEqual(len(fakeDBSInfo), 6)
for block in fakeDBSInfo:
if acqEra in block['block']['block_name']:
self.assertEqual(block['block']['file_count'], 5)
self.assertTrue('block_events' not in block['block'])
self.assertTrue('close_settings' not in block)
self.assertEqual(block['block']['open_for_writing'], 0)
# Put more files, they will go into the same block and then it will be closed
# after timeout
time.sleep(3)
self.createParentFiles(acqEra, nFiles = 3,
workflowName = workflowName,
taskPath = taskPath)
dbsUploader.algorithm()
dbsUploader.checkBlocks()
openBlocks = dbsUtil.findOpenBlocks()
self.assertEqual(len(openBlocks), 0)
fakeDBS = open(self.dbsUrl, 'r')
fakeDBSInfo = json.load(fakeDBS)
fakeDBS.close()
self.assertEqual(len(fakeDBSInfo), 7)
for block in fakeDBSInfo:
if acqEra in block['block']['block_name']:
if block['block']['file_count'] < 5:
self.assertEqual(block['block']['file_count'], 4)
else:
self.assertEqual(block['block']['file_count'], 5)
self.assertTrue('block_events' not in block['block'])
self.assertEqual(block['block']['open_for_writing'], 0)
self.assertTrue('close_settings' not in block)
# Finally test size limits
acqEra = "TropicalSeason%s" % (int(time.time()))
workflowName = 'TestWorkload%s' % (int(time.time()))
taskPath = '/%s/TestProcessing' % workflowName
self.injectWorkflow(workflowName, taskPath,
MaxWaitTime = 1, MaxFiles = 500,
MaxEvents = 200000000, MaxSize = 2048)
self.createParentFiles(acqEra, nFiles = 7,
workflowName = workflowName,
taskPath = taskPath)
dbsUploader.algorithm()
dbsUploader.checkBlocks()
time.sleep(2)
dbsUploader.algorithm()
dbsUploader.checkBlocks()
self.assertEqual(len(openBlocks), 0)
fakeDBS = open(self.dbsUrl, 'r')
fakeDBSInfo = json.load(fakeDBS)
fakeDBS.close()
self.assertEqual(len(fakeDBSInfo), 11)
for block in fakeDBSInfo:
if acqEra in block['block']['block_name']:
if block['block']['file_count'] != 1:
self.assertEqual(block['block']['block_size'], 2048)
self.assertEqual(block['block']['file_count'], 2)
self.assertTrue('block_events' not in block['block'])
self.assertEqual(block['block']['open_for_writing'], 0)
self.assertTrue('close_settings' not in block)
except:
self.fail("We failed at some point in the test")
finally:
# We don't trust anyone else with _exit
del os.environ["DONT_TRAP_EXIT"]
return
class RucioInjectorPollerTest(EmulatedUnitTestCase):
"""
Tests for the RucioInjectorPoller component
"""
def setUp(self):
"""
Install the DBSBuffer schema into the database and connect to Rucio.
"""
super(RucioInjectorPollerTest, self).setUp()
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection(destroyAllDatabase=True)
self.testInit.setSchema(customModules=["WMComponent.DBS3Buffer"],
useDefault=False)
myThread = threading.currentThread()
daofactory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationAction = daofactory(classname="DBSBufferFiles.AddLocation")
self.locations = ["T2_CH_CERN", "T1_US_FNAL_Disk"]
for rse in self.locations:
locationAction.execute(siteName=rse)
self.testFilesA = []
self.testFilesB = []
self.testDatasetA = "/SampleA/PromptReco-v1/RECO"
self.testDatasetB = "/SampleB/CRUZET11-v1/RAW"
if PY3:
self.assertItemsEqual = self.assertCountEqual
return
def tearDown(self):
"""
Delete the database.
"""
self.testInit.clearDatabase()
def createConfig(self):
"""
Create a basic configuration for the component
"""
config = self.testInit.getConfiguration()
config.component_("RucioInjector")
config.RucioInjector.pollInterval = 300
config.RucioInjector.pollIntervalRules = 43200
config.RucioInjector.cacheExpiration = 2 * 24 * 60 * 60 # two days
config.RucioInjector.createBlockRules = True
config.RucioInjector.RSEPostfix = False # enable it to append _Test to the RSE names
config.RucioInjector.metaDIDProject = "Production"
config.RucioInjector.containerDiskRuleParams = {"weight": "ddm_quota", "copies": 2, "grouping": "DATASET"}
config.RucioInjector.containerDiskRuleRSEExpr = "(tier=2|tier=1)&cms_type=real&rse_type=DISK"
config.RucioInjector.rucioAccount = "wma_test"
config.RucioInjector.rucioUrl = "http://cmsrucio-int.cern.ch"
config.RucioInjector.rucioAuthUrl = "https://cmsrucio-auth-int.cern.ch"
return config
def stuffDatabase(self):
"""
Fill the dbsbuffer tables with some files and blocks. We'll insert a total
of 5 files spanning two blocks. There will be a total of two datasets
inserted into the database.
We'll inject files with the location set as an SE name as well as a
PhEDEx node name as well.
"""
myThread = threading.currentThread()
# Create the DAOs factory and the relevant instances
buffer3Factory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
setBlock = buffer3Factory(classname="DBSBufferFiles.SetBlock")
fileStatus = buffer3Factory(classname="DBSBufferFiles.SetStatus")
associateWorkflow = buffer3Factory(classname="DBSBufferFiles.AssociateWorkflowToFile")
insertWorkflow = buffer3Factory(classname="InsertWorkflow")
datasetAction = buffer3Factory(classname="NewDataset")
createAction = buffer3Factory(classname="CreateBlocks")
# Create workflow in the database
insertWorkflow.execute("BogusRequest", "BogusTask", 0, 0, 0, 0)
# First file on first block
checksums = {"adler32": "1234", "cksum": "5678"}
testFileA = DBSBufferFile(lfn=makeUUID(), size=1024, events=10,
checksums=checksums,
locations=set(["T2_CH_CERN"]))
testFileA.setAlgorithm(appName="cmsRun", appVer="CMSSW_2_1_8",
appFam="RECO", psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath(self.testDatasetA)
testFileA.addRun(Run(2, *[45]))
testFileA.create()
# Second file on first block
testFileB = DBSBufferFile(lfn=makeUUID(), size=1024, events=10,
checksums=checksums,
locations=set(["T2_CH_CERN"]))
testFileB.setAlgorithm(appName="cmsRun", appVer="CMSSW_2_1_8",
appFam="RECO", psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileB.setDatasetPath(self.testDatasetA)
testFileB.addRun(Run(2, *[45]))
testFileB.create()
# Third file on first block
testFileC = DBSBufferFile(lfn=makeUUID(), size=1024, events=10,
checksums=checksums,
locations=set(["T2_CH_CERN"]))
testFileC.setAlgorithm(appName="cmsRun", appVer="CMSSW_2_1_8",
appFam="RECO", psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileC.setDatasetPath(self.testDatasetA)
testFileC.addRun(Run(2, *[45]))
testFileC.create()
self.testFilesA.append(testFileA)
self.testFilesA.append(testFileB)
self.testFilesA.append(testFileC)
# First file on second block
testFileD = DBSBufferFile(lfn=makeUUID(), size=1024, events=10,
checksums=checksums,
locations=set(["T1_US_FNAL_Disk"]))
testFileD.setAlgorithm(appName="cmsRun", appVer="CMSSW_2_1_8",
appFam="RECO", psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileD.setDatasetPath(self.testDatasetB)
testFileD.addRun(Run(2, *[45]))
testFileD.create()
# Second file on second block
testFileE = DBSBufferFile(lfn=makeUUID(), size=1024, events=10,
checksums=checksums,
locations=set(["T1_US_FNAL_Disk"]))
testFileE.setAlgorithm(appName="cmsRun", appVer="CMSSW_2_1_8",
appFam="RECO", psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileE.setDatasetPath(self.testDatasetB)
testFileE.addRun(Run(2, *[45]))
testFileE.create()
self.testFilesB.append(testFileD)
self.testFilesB.append(testFileE)
# insert datasets in the dbsbuffer table
datasetAction.execute(datasetPath=self.testDatasetA)
datasetAction.execute(datasetPath=self.testDatasetB)
self.blockAName = self.testDatasetA + "#" + makeUUID()
self.blockBName = self.testDatasetB + "#" + makeUUID()
# create and insert blocks into dbsbuffer table
newBlockA = DBSBufferBlock(name=self.blockAName,
location="T2_CH_CERN",
datasetpath=None)
newBlockA.setDataset(self.testDatasetA, 'data', 'VALID')
newBlockA.status = 'Closed'
newBlockB = DBSBufferBlock(name=self.blockBName,
location="T1_US_FNAL_Disk",
datasetpath=None)
newBlockB.setDataset(self.testDatasetB, 'data', 'VALID')
newBlockB.status = 'Closed'
createAction.execute(blocks=[newBlockA, newBlockB])
# associate files to their correspondent block id
setBlock.execute(testFileA["lfn"], self.blockAName)
setBlock.execute(testFileB["lfn"], self.blockAName)
setBlock.execute(testFileC["lfn"], self.blockAName)
setBlock.execute(testFileD["lfn"], self.blockBName)
setBlock.execute(testFileE["lfn"], self.blockBName)
# set file status to LOCAL
fileStatus.execute(testFileA["lfn"], "LOCAL")
fileStatus.execute(testFileB["lfn"], "LOCAL")
fileStatus.execute(testFileC["lfn"], "LOCAL")
fileStatus.execute(testFileD["lfn"], "LOCAL")
fileStatus.execute(testFileE["lfn"], "LOCAL")
# associate files to a given workflow
associateWorkflow.execute(testFileA["lfn"], "BogusRequest", "BogusTask")
associateWorkflow.execute(testFileB["lfn"], "BogusRequest", "BogusTask")
associateWorkflow.execute(testFileC["lfn"], "BogusRequest", "BogusTask")
associateWorkflow.execute(testFileD["lfn"], "BogusRequest", "BogusTask")
associateWorkflow.execute(testFileE["lfn"], "BogusRequest", "BogusTask")
return
def testBadConfig(self):
"""
Test wrong component configuration
"""
config = self.createConfig()
config.RucioInjector.metaDIDProject = "Very invalid project name"
with self.assertRaises(RucioInjectorException):
RucioInjectorPoller(config)
def testActivityMap(self):
"""
Initialize a RucioInjectorPoller object and test `_activityMap` method
"""
poller = RucioInjectorPoller(self.createConfig())
# test production agent and non-Tape endpoint
activity = poller._activityMap("T1_US_FNAL_Disk")
self.assertEquals(activity, "Production Output")
activity = poller._activityMap("T1_US_FNAL_Test")
self.assertEquals(activity, "Production Output")
# test production agent and Tape endpoint (which is forbidden at the moment)
with self.assertRaises(WMRucioException):
poller._activityMap("T1_US_FNAL_Tape")
# now pretend it to be a T0 agent/component
poller.isT0agent = True
# test T0 agent and non-Tape endpoint
activity = poller._activityMap("T1_US_FNAL_Disk")
self.assertEquals(activity, "T0 Export")
activity = poller._activityMap("T1_US_FNAL_Test")
self.assertEquals(activity, "T0 Export")
# test T0 agent and Tape endpoint
activity = poller._activityMap("T1_US_FNAL_Tape")
self.assertEquals(activity, "T0 Tape")
def testLoadingFiles(self):
"""
Initialize a RucioInjectorPoller object and load uninjected files
"""
self.stuffDatabase()
poller = RucioInjectorPoller(self.createConfig())
poller.setup(parameters=None)
uninjectedFiles = poller.getUninjected.execute()
self.assertItemsEqual(list(uninjectedFiles), self.locations)
self.assertEquals(list(uninjectedFiles["T2_CH_CERN"]), [self.testDatasetA])
self.assertEquals(list(uninjectedFiles["T1_US_FNAL_Disk"]), [self.testDatasetB])
class ProcessPoolTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection(destroyAllDatabase = True)
self.testInit.setSchema(customModules = ["WMCore.Agent.Database"],
useDefault = False)
return
def tearDown(self):
"""
_tearDown_
"""
self.testInit.clearDatabase()
return
def testA_ProcessPool(self):
"""
_testProcessPool_
"""
raise nose.SkipTest
config = self.testInit.getConfiguration()
config.Agent.useHeartbeat = False
self.testInit.generateWorkDir(config)
processPool = ProcessPool("ProcessPool_t.ProcessPoolTestWorker",
totalSlaves = 1,
componentDir = config.General.workDir,
config = config,
namespace = "WMCore_t")
processPool.enqueue(["One", "Two", "Three"])
result = processPool.dequeue(3)
self.assertEqual(len(result), 3, "Error: Expected three items back.")
self.assertTrue( "One" in result)
self.assertTrue( "Two" in result)
self.assertTrue( "Three" in result)
return
def testB_ProcessPoolStress(self):
"""
_testProcessPoolStress_
"""
raise nose.SkipTest
config = self.testInit.getConfiguration()
config.Agent.useHeartbeat = False
self.testInit.generateWorkDir(config)
processPool = ProcessPool("ProcessPool_t.ProcessPoolTestWorker",
totalSlaves = 1,
componentDir = config.General.workDir,
namespace = "WMCore_t",
config = config)
result = None
input = None
for i in range(1000):
input = []
while i > 0:
input.append("COMMAND%s" % i)
i -= 1
processPool.enqueue(input)
result = processPool.dequeue(len(input))
self.assertEqual(len(result), len(input),
"Error: Wrong number of results returned.")
for k in result:
self.assertTrue(k in input)
return
def testC_MultiPool(self):
"""
_testMultiPool_
Run a test with multiple workers
"""
raise nose.SkipTest
config = self.testInit.getConfiguration()
config.Agent.useHeartbeat = False
self.testInit.generateWorkDir(config)
processPool = ProcessPool("ProcessPool_t.ProcessPoolTestWorker",
totalSlaves = 3,
componentDir = config.General.workDir,
namespace = "WMCore_t",
config = config)
for i in range(100):
input = []
while i > 0:
input.append("COMMAND%s" % i)
i -= 1
processPool.enqueue(input)
result = processPool.dequeue(len(input))
self.assertEqual(len(result), len(input),
"Error: Wrong number of results returned.")
class DBSBufferFileTest(unittest.TestCase):
def setUp(self):
"""
_setUp_
Setup the database and logging connection. Try to create all of the
DBSBuffer tables. Also add some dummy locations.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["WMComponent.DBS3Buffer"],
useDefault=False)
myThread = threading.currentThread()
self.daoFactory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
locationAction = self.daoFactory(
classname="DBSBufferFiles.AddLocation")
locationAction.execute(siteName="se1.cern.ch")
locationAction.execute(siteName="se1.fnal.gov")
def tearDown(self):
"""
_tearDown_
Drop all the DBSBuffer tables.
"""
self.testInit.clearDatabase()
def testCreateDeleteExists(self):
"""
_testCreateDeleteExists_
Test the create(), delete() and exists() methods of the file class
by creating and deleting a file. The exists() method will be
called before and after creation and after deletion.
"""
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
assert testFile.exists() == False, \
"ERROR: File exists before it was created"
testFile.addRun(Run(1, *[45]))
testFile.create()
assert testFile.exists() > 0, \
"ERROR: File does not exist after it was created"
testFile.delete()
assert testFile.exists() == False, \
"ERROR: File exists after it has been deleted"
return
def testCreateTransaction(self):
"""
_testCreateTransaction_
Begin a transaction and then create a file in the database. Afterwards,
rollback the transaction. Use the File class's exists() method to
to verify that the file doesn't exist before it was created, exists
after it was created and doesn't exist after the transaction was rolled
back.
"""
myThread = threading.currentThread()
myThread.transaction.begin()
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
assert testFile.exists() == False, \
"ERROR: File exists before it was created"
testFile.addRun(Run(1, *[45]))
testFile.create()
assert testFile.exists() > 0, \
"ERROR: File does not exist after it was created"
myThread.transaction.rollback()
assert testFile.exists() == False, \
"ERROR: File exists after transaction was rolled back."
return
def testDeleteTransaction(self):
"""
_testDeleteTransaction_
Create a file and commit it to the database. Start a new transaction
and delete the file. Rollback the transaction after the file has been
deleted. Use the file class's exists() method to verify that the file
does not exist after it has been deleted but does exist after the
transaction is rolled back.
"""
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
assert testFile.exists() == False, \
"ERROR: File exists before it was created"
testFile.addRun(Run(1, *[45]))
testFile.create()
assert testFile.exists() > 0, \
"ERROR: File does not exist after it was created"
myThread = threading.currentThread()
myThread.transaction.begin()
testFile.delete()
assert testFile.exists() == False, \
"ERROR: File exists after it has been deleted"
myThread.transaction.rollback()
assert testFile.exists() > 0, \
"ERROR: File does not exist after transaction was rolled back."
return
def testGetParentLFNs(self):
"""
_testGetParentLFNs_
Create three files and set them to be parents of a fourth file. Check
to make sure that getParentLFNs() on the child file returns the correct
LFNs.
"""
testFileParentA = DBSBufferFile(lfn="/this/is/a/parent/lfnA",
size=1024,
events=20)
testFileParentA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentA.addRun(Run(1, *[45]))
testFileParentB = DBSBufferFile(lfn="/this/is/a/parent/lfnB",
size=1024,
events=20)
testFileParentB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentB.addRun(Run(1, *[45]))
testFileParentC = DBSBufferFile(lfn="/this/is/a/parent/lfnC",
size=1024,
events=20)
testFileParentC.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentC.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentC.addRun(Run(1, *[45]))
testFileParentA.create()
testFileParentB.create()
testFileParentC.create()
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.addRun(Run(1, *[45]))
testFile.create()
testFile.addParents([
testFileParentA["lfn"], testFileParentB["lfn"],
testFileParentC["lfn"]
])
parentLFNs = testFile.getParentLFNs()
assert len(parentLFNs) == 3, \
"ERROR: Child does not have the right amount of parents"
goldenLFNs = [
"/this/is/a/parent/lfnA", "/this/is/a/parent/lfnB",
"/this/is/a/parent/lfnC"
]
for parentLFN in parentLFNs:
assert parentLFN in goldenLFNs, \
"ERROR: Unknown parent lfn"
goldenLFNs.remove(parentLFN)
testFile.delete()
testFileParentA.delete()
testFileParentB.delete()
testFileParentC.delete()
return
def testLoad(self):
"""
_testLoad_
Test the loading of file meta data using the ID of a file and the
LFN of a file.
"""
checksums = {"adler32": "adler32", "cksum": "cksum", "md5": "md5"}
testFileA = DBSBufferFile(lfn="/this/is/a/lfn",
size=1024,
events=10,
checksums=checksums)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.create()
testFileB = DBSBufferFile(lfn=testFileA["lfn"])
testFileB.load()
testFileC = DBSBufferFile(id=testFileA["id"])
testFileC.load()
assert testFileA == testFileB, \
"ERROR: File load by LFN didn't work"
assert testFileA == testFileC, \
"ERROR: File load by ID didn't work"
assert isinstance(testFileB["id"], int) or isinstance(testFileB["id"], long), \
"ERROR: File id is not an integer type."
assert isinstance(testFileB["size"], int) or isinstance(testFileB["size"], long), \
"ERROR: File size is not an integer type."
assert isinstance(testFileB["events"], int) or isinstance(testFileB["events"], long), \
"ERROR: File events is not an integer type."
assert isinstance(testFileC["id"], int) or isinstance(testFileC["id"], long), \
"ERROR: File id is not an integer type."
assert isinstance(testFileC["size"], int) or isinstance(testFileC["size"], long), \
"ERROR: File size is not an integer type."
assert isinstance(testFileC["events"], int) or isinstance(testFileC["events"], long), \
"ERROR: File events is not an integer type."
testFileA.delete()
return
def testFilesize(self):
"""
_testFilesize_
Test storing and loading the file information from dbsbuffer_file.
Make sure filesize can be bigger than 32 bits
"""
checksums = {"adler32": "adler32", "cksum": "cksum"}
testFileA = DBSBufferFile(lfn="/this/is/a/lfn",
size=3221225472,
events=1500000,
checksums=checksums)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_7_6_0",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.create()
testFileB = DBSBufferFile(lfn=testFileA["lfn"])
testFileB.load()
self.assertEqual(testFileB["size"], 3221225472,
"Error: the filesize should be 3GB")
self.assertEqual(testFileB["events"], 1500000,
"Error: the number of events should be 1.5M")
def testAddChild(self):
"""
_testAddChild_
Add a child to some parent files and make sure that all the parentage
information is loaded/stored correctly from the database.
"""
testFileParentA = DBSBufferFile(lfn="/this/is/a/parent/lfnA",
size=1024,
events=20)
testFileParentA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentA.addRun(Run(1, *[45]))
testFileParentB = DBSBufferFile(lfn="/this/is/a/parent/lfnB",
size=1024,
events=20)
testFileParentB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentB.addRun(Run(1, *[45]))
testFileParentA.create()
testFileParentB.create()
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
testFileParentA.addChildren("/this/is/a/lfn")
testFileParentB.addChildren("/this/is/a/lfn")
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load(parentage=1)
goldenFiles = [testFileParentA, testFileParentB]
for parentFile in testFileB["parents"]:
assert parentFile in goldenFiles, \
"ERROR: Unknown parent file"
goldenFiles.remove(parentFile)
assert len(goldenFiles) == 0, \
"ERROR: Some parents are missing"
return
def testAddChildTransaction(self):
"""
_testAddChildTransaction_
Add a child to some parent files and make sure that all the parentage
information is loaded/stored correctly from the database. Rollback the
addition of one of the childs and then verify that it does in fact only
have one parent.
"""
testFileParentA = DBSBufferFile(lfn="/this/is/a/parent/lfnA",
size=1024,
events=20)
testFileParentA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentA.addRun(Run(1, *[45]))
testFileParentB = DBSBufferFile(lfn="/this/is/a/parent/lfnB",
size=1024,
events=20)
testFileParentB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileParentB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileParentB.addRun(Run(1, *[45]))
testFileParentA.create()
testFileParentB.create()
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
testFileParentA.addChildren("/this/is/a/lfn")
myThread = threading.currentThread()
myThread.transaction.begin()
testFileParentB.addChildren("/this/is/a/lfn")
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load(parentage=1)
goldenFiles = [testFileParentA, testFileParentB]
for parentFile in testFileB["parents"]:
assert parentFile in goldenFiles, \
"ERROR: Unknown parent file"
goldenFiles.remove(parentFile)
assert len(goldenFiles) == 0, \
"ERROR: Some parents are missing"
myThread.transaction.rollback()
testFileB.load(parentage=1)
goldenFiles = [testFileParentA]
for parentFile in testFileB["parents"]:
assert parentFile in goldenFiles, \
"ERROR: Unknown parent file"
goldenFiles.remove(parentFile)
assert len(goldenFiles) == 0, \
"ERROR: Some parents are missing"
return
def testSetLocation(self):
"""
_testSetLocation_
Create a file and add a couple locations. Load the file from the
database to make sure that the locations were set correctly.
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
testFileA.setLocation(["se1.fnal.gov", "se1.cern.ch"])
testFileA.setLocation(["bunkse1.fnal.gov", "bunkse1.cern.ch"],
immediateSave=False)
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load()
goldenLocations = ["se1.fnal.gov", "se1.cern.ch"]
for location in testFileB["locations"]:
assert location in goldenLocations, \
"ERROR: Unknown file location"
goldenLocations.remove(location)
assert len(goldenLocations) == 0, \
"ERROR: Some locations are missing"
return
def testSetLocationTransaction(self):
"""
_testSetLocationTransaction_
Create a file at specific locations and commit everything to the
database. Reload the file from the database and verify that the
locations are correct. Rollback the database transaction and once
again reload the file. Verify that the original locations are back.
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
testFileA.setLocation(["se1.fnal.gov"])
myThread = threading.currentThread()
myThread.transaction.begin()
testFileA.setLocation(["se1.cern.ch"])
testFileA.setLocation(["bunkse1.fnal.gov", "bunkse1.cern.ch"],
immediateSave=False)
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load()
goldenLocations = ["se1.fnal.gov", "se1.cern.ch"]
for location in testFileB["locations"]:
assert location in goldenLocations, \
"ERROR: Unknown file location"
goldenLocations.remove(location)
assert len(goldenLocations) == 0, \
"ERROR: Some locations are missing"
myThread.transaction.rollback()
testFileB.load()
goldenLocations = ["se1.fnal.gov"]
for location in testFileB["locations"]:
assert location in goldenLocations, \
"ERROR: Unknown file location"
goldenLocations.remove(location)
assert len(goldenLocations) == 0, \
"ERROR: Some locations are missing"
return
def testLocationsConstructor(self):
"""
_testLocationsConstructor_
Test to make sure that locations passed into the File() constructor
are loaded from and save to the database correctly. Also test to make
sure that the class behaves well when the location is passed in as a
single string instead of a set.
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn",
size=1024,
events=10,
locations=set(["se1.fnal.gov"]))
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
testFileB = DBSBufferFile(lfn="/this/is/a/lfn2",
size=1024,
events=10,
locations="se1.fnal.gov")
testFileB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileB.addRun(Run(1, *[45]))
testFileB.create()
testFileC = DBSBufferFile(id=testFileA["id"])
testFileC.load()
goldenLocations = ["se1.fnal.gov"]
for location in testFileC["locations"]:
assert location in goldenLocations, \
"ERROR: Unknown file location"
goldenLocations.remove(location)
assert len(goldenLocations) == 0, \
"ERROR: Some locations are missing"
testFileC = DBSBufferFile(id=testFileB["id"])
testFileC.load()
goldenLocations = ["se1.fnal.gov"]
for location in testFileC["locations"]:
assert location in goldenLocations, \
"ERROR: Unknown file location"
goldenLocations.remove(location)
assert len(goldenLocations) == 0, \
"ERROR: Some locations are missing"
return
def testAddRunSet(self):
"""
_testAddRunSet_
Test the ability to add run and lumi information to a file.
"""
testFile = DBSBufferFile(lfn="/this/is/a/lfn",
size=1024,
events=10,
locations="se1.fnal.gov")
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.create()
runSet = set()
runSet.add(Run(1, *[45]))
runSet.add(Run(2, *[67, 68]))
testFile.addRunSet(runSet)
assert (runSet - testFile["runs"]) == set(), \
"Error: addRunSet is not updating set correctly"
def testSetBlock(self):
"""
_testSetBlock_
Verify that the [Set|Get]Block DAOs work correctly.
"""
myThread = threading.currentThread()
dataset = "/Cosmics/CRUZET09-PromptReco-v1/RECO"
uploadFactory = DAOFactory(package="WMComponent.DBS3Buffer",
logger=myThread.logger,
dbinterface=myThread.dbi)
datasetAction = uploadFactory(classname="NewDataset")
createAction = uploadFactory(classname="CreateBlocks")
datasetAction.execute(datasetPath=dataset)
newBlock = DBSBufferBlock(name="someblockname",
location="se1.cern.ch",
datasetpath=None)
newBlock.setDataset(dataset, 'data', 'VALID')
createAction.execute(blocks=[newBlock])
setBlockAction = self.daoFactory(classname="DBSBufferFiles.SetBlock")
getBlockAction = self.daoFactory(classname="DBSBufferFiles.GetBlock")
testFile = DBSBufferFile(lfn="/this/is/a/lfn",
size=1024,
events=10,
locations="se1.fnal.gov")
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath(dataset)
testFile.create()
setBlockAction.execute(lfn=testFile["lfn"], blockName="someblockname")
blockName = getBlockAction.execute(lfn=testFile["lfn"])
assert blockName[0][0] == "someblockname", \
"Error: Incorrect block returned: %s" % blockName[0][0]
return
def testCountFilesDAO(self):
"""
_testCountFilesDAO_
Verify that the CountFiles DAO object functions correctly.
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfnA",
size=1024,
events=10,
locations="se1.fnal.gov")
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.create()
testFileB = DBSBufferFile(lfn="/this/is/a/lfnB",
size=1024,
events=10,
locations="se1.fnal.gov")
testFileB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileB.create()
testFileC = DBSBufferFile(lfn="/this/is/a/lfnC",
size=1024,
events=10,
locations="se1.fnal.gov")
testFileC.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileC.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileC.create()
countAction = self.daoFactory(classname="CountFiles")
assert countAction.execute() == 3, \
"Error: Wrong number of files counted in DBS Buffer."
return
def testAddParents(self):
"""
_testAddParents_
Verify that the addParents() method works correctly even if the parents
do not already exist in the database.
"""
myThread = threading.currentThread()
testFile = DBSBufferFile(lfn="/this/is/a/lfnA",
size=1024,
events=10,
locations="se1.fnal.gov")
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.create()
testParent = DBSBufferFile(lfn="/this/is/a/lfnB",
size=1024,
events=10,
locations="se1.fnal.gov")
testParent.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testParent.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RAW")
testParent.create()
goldenLFNs = ["lfn1", "lfn2", "lfn3", "/this/is/a/lfnB"]
testFile.addParents(goldenLFNs)
verifyFile = DBSBufferFile(id=testFile["id"])
verifyFile.load(parentage=1)
parentLFNs = verifyFile.getParentLFNs()
for parentLFN in parentLFNs:
self.assertTrue(parentLFN in goldenLFNs,
"Error: unknown lfn %s" % parentLFN)
goldenLFNs.remove(parentLFN)
self.assertEqual(len(goldenLFNs), 0, "Error: missing LFNs...")
# Check that the bogus dataset is listed as inDBS
sqlCommand = """SELECT in_dbs FROM dbsbuffer_algo_dataset_assoc das
INNER JOIN dbsbuffer_dataset ds ON das.dataset_id = ds.id
WHERE ds.path = 'bogus'"""
status = myThread.dbi.processData(sqlCommand)[0].fetchall()[0][0]
self.assertEqual(status, 1)
# Now make sure the dummy files are listed as being in DBS
sqlCommand = """SELECT status FROM dbsbuffer_file df
INNER JOIN dbsbuffer_algo_dataset_assoc das ON das.id = df.dataset_algo
INNER JOIN dbsbuffer_dataset ds ON das.dataset_id = ds.id
WHERE ds.path = '/bogus/dataset/path' """
status = myThread.dbi.processData(sqlCommand)[0].fetchall()
for entry in status:
self.assertEqual(entry, ('AlreadyInDBS', ))
return
def testGetChildrenDAO(self):
"""
_testGetChildrenDAO_
Verify that the GetChildren DAO correctly returns the LFNs of a file's
children.
"""
testFileChildA = DBSBufferFile(lfn="/this/is/a/child/lfnA",
size=1024,
events=20)
testFileChildA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildB = DBSBufferFile(lfn="/this/is/a/child/lfnB",
size=1024,
events=20)
testFileChildB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildC = DBSBufferFile(lfn="/this/is/a/child/lfnC",
size=1024,
events=20)
testFileChildC.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildC.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildA.create()
testFileChildB.create()
testFileChildC.create()
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.create()
testFileChildA.addParents([testFile["lfn"]])
testFileChildB.addParents([testFile["lfn"]])
testFileChildC.addParents([testFile["lfn"]])
getChildrenAction = self.daoFactory(
classname="DBSBufferFiles.GetChildren")
childLFNs = getChildrenAction.execute(testFile["lfn"])
assert len(childLFNs) == 3, \
"ERROR: Parent does not have the right amount of children."
goldenLFNs = [
"/this/is/a/child/lfnA", "/this/is/a/child/lfnB",
"/this/is/a/child/lfnC"
]
for childLFN in childLFNs:
assert childLFN in goldenLFNs, \
"ERROR: Unknown child lfn"
goldenLFNs.remove(childLFN)
return
def testGetParentStatusDAO(self):
"""
_testGetParentStatusDAO_
Verify that the GetParentStatus DAO correctly returns the status of a
file's children.
"""
testFileChild = DBSBufferFile(lfn="/this/is/a/child/lfnA",
size=1024,
events=20)
testFileChild.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChild.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChild.create()
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.create()
testFileChild.addParents([testFile["lfn"]])
getStatusAction = self.daoFactory(
classname="DBSBufferFiles.GetParentStatus")
parentStatus = getStatusAction.execute(testFileChild["lfn"])
assert len(parentStatus) == 1, \
"ERROR: Wrong number of statuses returned."
assert parentStatus[0] == "NOTUPLOADED", \
"ERROR: Wrong status returned."
return
def testSetLocationByLFN(self):
"""
_testSetLocationByLFN_
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.addRun(Run(1, *[45]))
testFileA.create()
setLocationAction = self.daoFactory(
classname="DBSBufferFiles.SetLocationByLFN")
setLocationAction.execute(binds={
'lfn': "/this/is/a/lfn",
'pnn': 'se1.cern.ch'
})
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load()
self.assertEqual(testFileB['locations'], set(['se1.cern.ch']))
return
def testAddCKSumByLFN(self):
"""
_testAddCKSumByLFN_
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.create()
setCksumAction = self.daoFactory(
classname="DBSBufferFiles.AddChecksumByLFN")
binds = [{
'lfn': "/this/is/a/lfn",
'cktype': 'adler32',
'cksum': 201
}, {
'lfn': "/this/is/a/lfn",
'cktype': 'cksum',
'cksum': 101
}]
setCksumAction.execute(bulkList=binds)
testFileB = DBSBufferFile(id=testFileA["id"])
testFileB.load()
self.assertEqual(testFileB['checksums'], {
'adler32': '201',
'cksum': '101'
})
return
def testBulkLoad(self):
"""
_testBulkLoad_
Can we load in bulk?
"""
addToBuffer = DBSBufferUtil()
testFileChildA = DBSBufferFile(lfn="/this/is/a/child/lfnA",
size=1024,
events=20)
testFileChildA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildB = DBSBufferFile(lfn="/this/is/a/child/lfnB",
size=1024,
events=20)
testFileChildB.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildB.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildC = DBSBufferFile(lfn="/this/is/a/child/lfnC",
size=1024,
events=20)
testFileChildC.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChildC.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileChildA.create()
testFileChildB.create()
testFileChildC.create()
testFileChildA.setLocation(["se1.fnal.gov", "se1.cern.ch"])
testFileChildB.setLocation(["se1.fnal.gov", "se1.cern.ch"])
testFileChildC.setLocation(["se1.fnal.gov", "se1.cern.ch"])
runSet = set()
runSet.add(Run(1, *[45]))
runSet.add(Run(2, *[67, 68]))
testFileChildA.addRunSet(runSet)
testFileChildB.addRunSet(runSet)
testFileChildC.addRunSet(runSet)
testFileChildA.save()
testFileChildB.save()
testFileChildC.save()
setCksumAction = self.daoFactory(
classname="DBSBufferFiles.AddChecksumByLFN")
binds = [{
'lfn': "/this/is/a/child/lfnA",
'cktype': 'adler32',
'cksum': 201
}, {
'lfn': "/this/is/a/child/lfnA",
'cktype': 'cksum',
'cksum': 101
}, {
'lfn': "/this/is/a/child/lfnB",
'cktype': 'adler32',
'cksum': 201
}, {
'lfn': "/this/is/a/child/lfnB",
'cktype': 'cksum',
'cksum': 101
}, {
'lfn': "/this/is/a/child/lfnC",
'cktype': 'adler32',
'cksum': 201
}, {
'lfn': "/this/is/a/child/lfnC",
'cktype': 'cksum',
'cksum': 101
}]
setCksumAction.execute(bulkList=binds)
testFile = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFile.create()
testFileChildA.addParents([testFile["lfn"]])
testFileChildB.addParents([testFile["lfn"]])
testFileChildC.addParents([testFile["lfn"]])
binds = [{
'id': testFileChildA.exists()
}, {
'id': testFileChildB.exists()
}, {
'id': testFileChildC.exists()
}]
listOfFiles = addToBuffer.loadDBSBufferFilesBulk(fileObjs=binds)
# print listOfFiles
compareList = [
'locations', 'psetHash', 'configContent', 'appName', 'appVer',
'appFam', 'events', 'datasetPath', 'runs'
]
for f in listOfFiles:
self.assertTrue(
f['lfn'] in [
"/this/is/a/child/lfnA", "/this/is/a/child/lfnB",
"/this/is/a/child/lfnC"
], "Unknown file in loaded results")
self.assertEqual(f['checksums'], {
'adler32': '201',
'cksum': '101'
})
for parent in f['parents']:
self.assertEqual(parent['lfn'], testFile['lfn'])
for key in compareList:
self.assertEqual(f[key], testFileChildA[key])
def testProperties(self):
"""
_testProperties_
Test added tags that use DBSBuffer to transfer from workload to DBS
"""
testFileA = DBSBufferFile(lfn="/this/is/a/lfn", size=1024, events=10)
testFileA.setAlgorithm(appName="cmsRun",
appVer="CMSSW_2_1_8",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileA.setDatasetPath("/Cosmics/CRUZET09-PromptReco-v1/RECO")
testFileA.setValidStatus(validStatus="VALID")
testFileA.setProcessingVer(ver="ProcVer")
testFileA.setAcquisitionEra(era="AcqEra")
testFileA.setGlobalTag(globalTag="GlobalTag")
testFileA.setDatasetParent(datasetParent="Parent")
testFileA.create()
return
class RepackTest(unittest.TestCase):
"""
_RepackTest_
Test for Repack job splitter
"""
def setUp(self):
"""
_setUp_
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules=["T0.WMBS"])
self.splitterFactory = SplitterFactory(package="T0.JobSplitting")
myThread = threading.currentThread()
daoFactory = DAOFactory(package="T0.WMBS",
logger=logging,
dbinterface=myThread.dbi)
myThread.dbi.processData("""INSERT INTO wmbs_location
(id, site_name, state)
VALUES (1, 'SomeSite', 1)
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_location_senames
(location, se_name)
VALUES (1, 'SomeSE')
""",
transaction=False)
myThread.dbi.processData("""INSERT INTO wmbs_location_senames
(location, se_name)
VALUES (1, 'SomeSE2')
""",
transaction=False)
insertRunDAO = daoFactory(classname="RunConfig.InsertRun")
insertRunDAO.execute(binds={
'RUN': 1,
'TIME': int(time.time()),
'HLTKEY': "someHLTKey"
},
transaction=False)
insertLumiDAO = daoFactory(classname="RunConfig.InsertLumiSection")
for lumi in [1, 2, 3, 4]:
insertLumiDAO.execute(binds={
'RUN': 1,
'LUMI': lumi
},
transaction=False)
insertStreamDAO = daoFactory(classname="RunConfig.InsertStream")
insertStreamDAO.execute(binds={'STREAM': "A"}, transaction=False)
insertStreamFilesetDAO = daoFactory(
classname="RunConfig.InsertStreamFileset")
insertStreamFilesetDAO.execute(1, "A", "TestFileset1")
self.fileset1 = Fileset(name="TestFileset1")
self.fileset1.load()
workflow1 = Workflow(spec="spec.xml",
owner="hufnagel",
name="TestWorkflow1",
task="Test")
workflow1.create()
self.subscription1 = Subscription(fileset=self.fileset1,
workflow=workflow1,
split_algo="Repack",
type="Repack")
self.subscription1.create()
# keep for later
self.insertClosedLumiDAO = daoFactory(
classname="RunLumiCloseout.InsertClosedLumi")
self.currentTime = int(time.time())
# default split parameters
self.splitArgs = {}
self.splitArgs['maxSizeSingleLumi'] = 20 * 1024 * 1024 * 1024
self.splitArgs['maxSizeMultiLumi'] = 10 * 1024 * 1024 * 1024
self.splitArgs['maxInputEvents'] = 500000
self.splitArgs['maxInputFiles'] = 1000
return
def tearDown(self):
"""
_tearDown_
"""
self.testInit.clearDatabase()
return
def getNumActiveSplitLumis(self):
"""
_getNumActiveSplitLumis_
helper function that counts the number of active split lumis
"""
myThread = threading.currentThread()
results = myThread.dbi.processData("""SELECT COUNT(*)
FROM lumi_section_split_active
""",
transaction=False)[0].fetchall()
return results[0][0]
def test00(self):
"""
_test00_
Test that the job name prefix feature works
Test multi lumi size threshold
Multi lumi input
"""
mySplitArgs = self.splitArgs.copy()
for lumi in [1, 2, 3, 4]:
filecount = 2
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
mySplitArgs['maxSizeMultiLumi'] = self.splitArgs['maxSizeMultiLumi']
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxSizeMultiLumi'] = 5000
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertTrue(job['name'].startswith("Repack-"),
"ERROR: Job has wrong name")
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset1.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertTrue(job['name'].startswith("Repack-"),
"ERROR: Job has wrong name")
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.assertEqual(self.getNumActiveSplitLumis(), 0,
"ERROR: Split lumis were created")
return
def test01(self):
"""
_test01_
Test multi lumi event threshold
Multi lumi input
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1, 2, 3, 4]:
filecount = 2
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputEvents'] = 500
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset1.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.assertEqual(self.getNumActiveSplitLumis(), 0,
"ERROR: Split lumis were created")
return
def test02(self):
"""
_test02_
Test single lumi size threshold
Single lumi input
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1]:
filecount = 8
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxSizeSingleLumi'] = 6500
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 6,
"ERROR: Job does not process 6 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
self.assertEqual(self.getNumActiveSplitLumis(), 1,
"ERROR: Split lumis were not created")
return
def test03(self):
"""
_test03_
Test single lumi event threshold
Single lumi input
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1]:
filecount = 8
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputEvents'] = 650
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 2,
"ERROR: JobFactory didn't create two jobs")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 6,
"ERROR: Job does not process 6 files")
job = jobGroups[0].jobs[1]
self.assertEqual(len(job.getFiles()), 2,
"ERROR: Job does not process 2 files")
self.assertEqual(self.getNumActiveSplitLumis(), 1,
"ERROR: Split lumis were not created")
return
def test04(self):
"""
_test04_
Test streamer count threshold (only multi lumi)
Multi lumi input
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1, 2, 3, 4]:
filecount = 2
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
mySplitArgs['maxInputFiles'] = 5
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.fileset1.markOpen(False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create a single job")
job = jobGroups[0].jobs[0]
self.assertEqual(len(job.getFiles()), 4,
"ERROR: Job does not process 4 files")
self.assertEqual(self.getNumActiveSplitLumis(), 0,
"ERROR: Split lumis were created")
return
def test05(self):
"""
_test05_
Test repacking of multiple lumis with holes in the lumi sequence
Multi lumi input
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1, 2, 4]:
filecount = 2
for i in range(filecount):
newFile = File(makeUUID(), size=1000, events=100)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
mySplitArgs['maxInputFiles'] = 5
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 0,
"ERROR: JobFactory should have returned no JobGroup")
self.insertClosedLumiDAO.execute(binds={
'RUN': 1,
'LUMI': 3,
'STREAM': "A",
'FILECOUNT': 0,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
},
transaction=False)
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 1,
"ERROR: JobFactory didn't create one job")
self.assertEqual(len(jobGroups[0].jobs[0].getFiles()), 4,
"ERROR: first job does not process 4 files")
return
def test06(self):
"""
_test06_
Test repacking of 3 lumis
2 small lumis (single job), followed by a big one (multiple jobs)
files for lumi 1 and 2 are below multi-lumi thresholds
files for lumi 3 are above single-lumi threshold
"""
mySplitArgs = self.splitArgs.copy()
insertClosedLumiBinds = []
for lumi in [1, 2, 3]:
filecount = 2
for i in range(filecount):
if lumi == 3:
nevents = 500
else:
nevents = 100
newFile = File(makeUUID(), size=1000, events=nevents)
newFile.addRun(Run(1, *[lumi]))
newFile.setLocation("SomeSE", immediateSave=False)
newFile.create()
self.fileset1.addFile(newFile)
insertClosedLumiBinds.append({
'RUN': 1,
'LUMI': lumi,
'STREAM': "A",
'FILECOUNT': filecount,
'INSERT_TIME': self.currentTime,
'CLOSE_TIME': self.currentTime
})
self.fileset1.commit()
jobFactory = self.splitterFactory(package="WMCore.WMBS",
subscription=self.subscription1)
self.insertClosedLumiDAO.execute(binds=insertClosedLumiBinds,
transaction=False)
mySplitArgs['maxInputEvents'] = 900
jobGroups = jobFactory(**mySplitArgs)
self.assertEqual(len(jobGroups), 1,
"ERROR: JobFactory didn't return one JobGroup")
self.assertEqual(len(jobGroups[0].jobs), 3,
"ERROR: JobFactory didn't create three jobs")
self.assertEqual(len(jobGroups[0].jobs[0].getFiles()), 4,
"ERROR: first job does not process 4 files")
self.assertEqual(len(jobGroups[0].jobs[1].getFiles()), 1,
"ERROR: second job does not process 1 file")
self.assertEqual(len(jobGroups[0].jobs[2].getFiles()), 1,
"ERROR: third job does not process 1 file")
return
class ReportIntegrationTest(unittest.TestCase):
"""
_ReportIntegrationTest_
"""
def setUp(self):
"""
_setUp_
Setup the database and WMBS for the test.
"""
self.testInit = TestInit(__file__)
self.testInit.setLogging()
self.testInit.setDatabaseConnection()
self.testInit.setSchema(customModules = ["WMComponent.DBS3Buffer",
"WMCore.WMBS"],
useDefault = False)
myThread = threading.currentThread()
self.daofactory = DAOFactory(package = "WMCore.WMBS",
logger = myThread.logger,
dbinterface = myThread.dbi)
self.dbsfactory = DAOFactory(package = "WMComponent.DBS3Buffer",
logger = myThread.logger,
dbinterface = myThread.dbi)
locationAction = self.daofactory(classname = "Locations.New")
locationAction.execute(siteName = "site1", pnn = "T1_US_FNAL_Disk")
inputFile = File(lfn = "/path/to/some/lfn", size = 10, events = 10,
locations = "T1_US_FNAL_Disk")
inputFile.create()
inputFileset = Fileset(name = "InputFileset")
inputFileset.create()
inputFileset.addFile(inputFile)
inputFileset.commit()
unmergedFileset = Fileset(name = "UnmergedFileset")
unmergedFileset.create()
mergedFileset = Fileset(name = "MergedFileset")
mergedFileset.create()
procWorkflow = Workflow(spec = "wf001.xml", owner = "Steve",
name = "TestWF", task = "/TestWF/None")
procWorkflow.create()
procWorkflow.addOutput("outputRECORECO", unmergedFileset)
mergeWorkflow = Workflow(spec = "wf002.xml", owner = "Steve",
name = "MergeWF", task = "/MergeWF/None")
mergeWorkflow.create()
mergeWorkflow.addOutput("Merged", mergedFileset)
insertWorkflow = self.dbsfactory(classname = "InsertWorkflow")
insertWorkflow.execute("TestWF", "/TestWF/None", 0, 0, 0, 0)
insertWorkflow.execute("MergeWF", "/MergeWF/None", 0, 0, 0, 0)
self.procSubscription = Subscription(fileset = inputFileset,
workflow = procWorkflow,
split_algo = "FileBased",
type = "Processing")
self.procSubscription.create()
self.procSubscription.acquireFiles()
self.mergeSubscription = Subscription(fileset = unmergedFileset,
workflow = mergeWorkflow,
split_algo = "WMBSMergeBySize",
type = "Merge")
self.mergeSubscription.create()
self.procJobGroup = JobGroup(subscription = self.procSubscription)
self.procJobGroup.create()
self.mergeJobGroup = JobGroup(subscription = self.mergeSubscription)
self.mergeJobGroup.create()
self.testJob = Job(name = "testJob", files = [inputFile])
self.testJob.create(group = self.procJobGroup)
self.testJob["state"] = "complete"
myThread = threading.currentThread()
self.daofactory = DAOFactory(package = "WMCore.WMBS",
logger = myThread.logger,
dbinterface = myThread.dbi)
self.stateChangeAction = self.daofactory(classname = "Jobs.ChangeState")
self.setFWJRAction = self.daofactory(classname = "Jobs.SetFWJRPath")
self.getJobTypeAction = self.daofactory(classname = "Jobs.GetType")
locationAction = self.daofactory(classname = "Locations.New")
locationAction.execute(siteName = "cmssrm.fnal.gov")
self.stateChangeAction.execute(jobs = [self.testJob])
self.tempDir = tempfile.mkdtemp()
return
def tearDown(self):
"""
_tearDown_
Clear out the database and the pickled report file.
"""
self.testInit.clearDatabase()
try:
os.remove(os.path.join(self.tempDir, "ProcReport.pkl"))
os.remove(os.path.join(self.tempDir, "MergeReport.pkl"))
except Exception as ex:
pass
try:
os.rmdir(self.tempDir)
except Exception as ex:
pass
return
def createConfig(self, workerThreads):
"""
_createConfig_
Create a config for the JobAccountant with the given number of worker
threads. This config needs to include information for connecting to the
database as the component will create it's own database connections.
These parameters are still pulled from the environment.
"""
config = self.testInit.getConfiguration()
self.testInit.generateWorkDir(config)
config.section_("JobStateMachine")
config.JobStateMachine.couchurl = os.getenv("COUCHURL")
config.JobStateMachine.couchDBName = "report_integration_t"
config.JobStateMachine.jobSummaryDBName = "report_integration_wmagent_summary_t"
config.component_("JobAccountant")
config.JobAccountant.pollInterval = 60
config.JobAccountant.workerThreads = workerThreads
config.JobAccountant.componentDir = os.getcwd()
config.JobAccountant.logLevel = 'SQLDEBUG'
config.component_("TaskArchiver")
config.TaskArchiver.localWMStatsURL = "%s/%s" % (config.JobStateMachine.couchurl, config.JobStateMachine.jobSummaryDBName)
return config
def verifyJobSuccess(self, jobID):
"""
_verifyJobSuccess_
Verify that the metadata for a successful job is correct. This will
check the outcome, retry count and state.
"""
testJob = Job(id = jobID)
testJob.load()
assert testJob["state"] == "success", \
"Error: test job in wrong state: %s" % testJob["state"]
assert testJob["retry_count"] == 0, \
"Error: test job has wrong retry count: %s" % testJob["retry_count"]
assert testJob["outcome"] == "success", \
"Error: test job has wrong outcome: %s" % testJob["outcome"]
return
def verifyFileMetaData(self, jobID, fwkJobReportFiles):
"""
_verifyFileMetaData_
Verify that all the files that were output by a job made it into WMBS
correctly. Compare the contents of WMBS to the files in the frameworks
job report.
Note that fwkJobReportFiles is a list of DataStructs File objects.
"""
testJob = Job(id = jobID)
testJob.loadData()
inputLFNs = []
for inputFile in testJob["input_files"]:
inputLFNs.append(inputFile["lfn"])
for fwkJobReportFile in fwkJobReportFiles:
outputFile = File(lfn = fwkJobReportFile["lfn"])
outputFile.loadData(parentage = 1)
assert outputFile["events"] == int(fwkJobReportFile["events"]), \
"Error: Output file has wrong events: %s, %s" % \
(outputFile["events"], fwkJobReportFile["events"])
assert outputFile["size"] == int(fwkJobReportFile["size"]), \
"Error: Output file has wrong size: %s, %s" % \
(outputFile["size"], fwkJobReportFile["size"])
for ckType in fwkJobReportFile["checksums"]:
assert ckType in outputFile["checksums"], \
"Error: Output file is missing checksums: %s" % ckType
assert outputFile["checksums"][ckType] == fwkJobReportFile["checksums"][ckType], \
"Error: Checksums don't match."
assert len(fwkJobReportFile["checksums"]) == \
len(outputFile["checksums"]), \
"Error: Wrong number of checksums."
jobType = self.getJobTypeAction.execute(jobID = jobID)
if jobType == "Merge":
assert str(outputFile["merged"]) == "True", \
"Error: Merge jobs should output merged files."
else:
assert outputFile["merged"] == fwkJobReportFile["merged"], \
"Error: Output file merged output is wrong: %s, %s" % \
(outputFile["merged"], fwkJobReportFile["merged"])
assert len(outputFile["locations"]) == 1, \
"Error: outputfile should have one location: %s" % outputFile["locations"]
assert list(outputFile["locations"])[0] == list(fwkJobReportFile["locations"])[0], \
"Error: wrong location for file."
assert len(outputFile["parents"]) == len(inputLFNs), \
"Error: Output file has wrong number of parents."
for outputParent in outputFile["parents"]:
assert outputParent["lfn"] in inputLFNs, \
"Error: Unknown parent file: %s" % outputParent["lfn"]
fwjrRuns = {}
for run in fwkJobReportFile["runs"]:
fwjrRuns[run.run] = run.lumis
for run in outputFile["runs"]:
assert run.run in fwjrRuns, \
"Error: Extra run in output: %s" % run.run
for lumi in run:
assert lumi in fwjrRuns[run.run], \
"Error: Extra lumi: %s" % lumi
fwjrRuns[run.run].remove(lumi)
if len(fwjrRuns[run.run]) == 0:
del fwjrRuns[run.run]
assert len(fwjrRuns) == 0, \
"Error: Missing runs, lumis: %s" % fwjrRuns
testJobGroup = JobGroup(id = testJob["jobgroup"])
testJobGroup.loadData()
jobGroupFileset = testJobGroup.output
jobGroupFileset.loadData()
assert outputFile["id"] in jobGroupFileset.getFiles(type = "id"), \
"Error: output file not in jobgroup fileset."
if testJob["mask"]["FirstEvent"] == None:
assert outputFile["first_event"] == 0, \
"Error: first event not set correctly: 0, %s" % \
outputFile["first_event"]
else:
assert testJob["mask"]["FirstEvent"] == outputFile["first_event"], \
"Error: last event not set correctly: %s, %s" % \
(testJob["mask"]["FirstEvent"], outputFile["first_event"])
return
def testReportHandling(self):
"""
_testReportHandling_
Verify that we're able to parse a CMSSW report, convert it to a Report()
style report, pickle it and then have the accountant process it.
"""
self.procPath = os.path.join(WMCore.WMBase.getTestBase(),
"WMCore_t/FwkJobReport_t/CMSSWProcessingReport.xml")
myReport = Report("cmsRun1")
myReport.parse(self.procPath)
# Fake some metadata that should be added by the stageout scripts.
for fileRef in myReport.getAllFileRefsFromStep("cmsRun1"):
fileRef.size = 1024
fileRef.location = "cmssrm.fnal.gov"
fwjrPath = os.path.join(self.tempDir, "ProcReport.pkl")
cmsRunStep = myReport.retrieveStep("cmsRun1")
cmsRunStep.status = 0
myReport.setTaskName('/TestWF/None')
myReport.persist(fwjrPath)
self.setFWJRAction.execute(jobID = self.testJob["id"], fwjrPath = fwjrPath)
pFile = DBSBufferFile(lfn = "/path/to/some/lfn", size = 600000, events = 60000)
pFile.setAlgorithm(appName = "cmsRun", appVer = "UNKNOWN",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
pFile.setDatasetPath("/bogus/dataset/path")
#pFile.addRun(Run(1, *[45]))
pFile.create()
config = self.createConfig(workerThreads = 1)
accountant = JobAccountantPoller(config)
accountant.setup()
accountant.algorithm()
self.verifyJobSuccess(self.testJob["id"])
self.verifyFileMetaData(self.testJob["id"], myReport.getAllFilesFromStep("cmsRun1"))
inputFile = File(lfn = "/store/backfill/2/unmerged/WMAgentCommissioining10/MinimumBias/RECO/rereco_GR09_R_34X_V5_All_v1/0000/outputRECORECO.root")
inputFile.load()
self.testMergeJob = Job(name = "testMergeJob", files = [inputFile])
self.testMergeJob.create(group = self.mergeJobGroup)
self.testMergeJob["state"] = "complete"
self.stateChangeAction.execute(jobs = [self.testMergeJob])
self.mergePath = os.path.join(WMCore.WMBase.getTestBase(),
"WMCore_t/FwkJobReport_t/CMSSWMergeReport.xml")
myReport = Report("mergeReco")
myReport.parse(self.mergePath)
# Fake some metadata that should be added by the stageout scripts.
for fileRef in myReport.getAllFileRefsFromStep("mergeReco"):
fileRef.size = 1024
fileRef.location = "cmssrm.fnal.gov"
fileRef.dataset = {"applicationName": "cmsRun", "applicationVersion": "CMSSW_3_4_2_patch1",
"primaryDataset": "MinimumBias", "processedDataset": "Rereco-v1",
"dataTier": "RECO"}
fwjrPath = os.path.join(self.tempDir, "MergeReport.pkl")
myReport.setTaskName('/MergeWF/None')
cmsRunStep = myReport.retrieveStep("mergeReco")
cmsRunStep.status = 0
myReport.persist(fwjrPath)
self.setFWJRAction.execute(jobID = self.testMergeJob["id"], fwjrPath = fwjrPath)
accountant.algorithm()
self.verifyJobSuccess(self.testMergeJob["id"])
self.verifyFileMetaData(self.testMergeJob["id"], myReport.getAllFilesFromStep("mergeReco"))
return
