def testXSetBlock(self):
"""
_testSetBlock_
Verify that the [Set|Get]Block DAOs work correctly.
"""
myThread = threading.currentThread()
uploadFactory = DAOFactory(package = "WMComponent.DBSUpload.Database",
logger = myThread.logger,
dbinterface = myThread.dbi)
createAction = uploadFactory(classname = "SetBlockStatus")
createAction.execute(block = "someblockname", locations = ["se1.cern.ch"])
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("/Cosmics/CRUZET09-PromptReco-v1/RECO")
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 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 getFiles(self,
name,
tier='RECO',
nFiles=12,
site="malpaquet",
nLumis=1):
"""
Create some quick dummy test files
"""
files = []
for f in range(nFiles):
testFile = DBSBufferFile(
lfn='/data/store/random/random/RANDOM/test/0/%s-%s-%i.root' %
(name, site, f),
size=1024,
events=20,
checksums={'cksum': 1})
testFile.setAlgorithm(appName=name,
appVer="CMSSW_3_1_1",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/%s/%s/%s" % (name, name, tier))
lumis = []
for i in range(nLumis):
lumis.append((f * 100000) + i)
testFile.addRun(Run(1, *lumis))
testFile.setAcquisitionEra(name.split('-')[0])
testFile.setProcessingVer("0")
testFile.setGlobalTag("Weird")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
count = 0
for f in files:
count += 1
testFileChild = DBSBufferFile(
lfn=
'/data/store/random/random/RANDOM/test/0/%s-%s-%i-child.root' %
(name, site, count),
size=1024,
events=10,
checksums={'cksum': 1})
testFileChild.setAlgorithm(appName=name,
appVer="CMSSW_3_1_1",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChild.setDatasetPath("/%s/%s_2/RECO" % (name, name))
testFileChild.addRun(Run(1, *[45]))
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([f['lfn']])
return files
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 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 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 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()
# There are no accessors for these things because load is never called
action = self.daoFactory2(classname="LoadInfoFromDAS")
das = action.execute(ids=[1])[0]
self.assertEqual(das['Parent'], 'Parent')
self.assertEqual(das['GlobalTag'], 'GlobalTag')
self.assertEqual(das['ValidStatus'], 'VALID')
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 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 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 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 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()
checksums = {"adler32": "adler32", "cksum": "cksum", "md5": "md5"}
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 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 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 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 testXSetBlock(self):
"""
_testSetBlock_
Verify that the [Set|Get]Block DAOs work correctly.
"""
myThread = threading.currentThread()
uploadFactory = DAOFactory(package = "WMComponent.DBSUpload.Database",
logger = myThread.logger,
dbinterface = myThread.dbi)
createAction = uploadFactory(classname = "SetBlockStatus")
createAction.execute(block = "someblockname", locations = ["se1.cern.ch"])
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("/Cosmics/CRUZET09-PromptReco-v1/RECO")
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 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 testSetLocationByLFN(self):
"""
_testSetLocationByLFN_
"""
myThread = threading.currentThread()
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", 'sename': 'se1.cern.ch'})
testFileB = DBSBufferFile(id = testFileA["id"])
testFileB.load()
self.assertEqual(testFileB['locations'], set(['se1.cern.ch']))
return
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()
# There are no accessors for these things because load is never called
action = self.daoFactory2(classname = "LoadInfoFromDAS")
das = action.execute(ids = [1])[0]
self.assertEqual(das['Parent'], 'Parent')
self.assertEqual(das['GlobalTag'], 'GlobalTag')
self.assertEqual(das['ValidStatus'], 'VALID')
def getFiles(self, name, tier, nFiles=12, site="malpaquet"):
"""
Create some quick dummy test files
"""
files = []
for f in range(0, nFiles):
testFile = DBSBufferFile(lfn="%s-%s-%i" % (name, site, f), size=1024, events=20, checksums={"cksum": 1})
testFile.setAlgorithm(
appName=name, appVer="CMSSW_3_1_1", appFam="RECO", psetHash="GIBBERISH", configContent=self.configURL
)
testFile.setDatasetPath("/%s/%s/%s" % (name, name, tier))
testFile.addRun(Run(1, *[f]))
testFile.setGlobalTag("aGlobalTag")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
testFileChild = DBSBufferFile(lfn="%s-%s-child" % (name, site), size=1024, events=10, checksums={"cksum": 1})
testFileChild.setAlgorithm(
appName=name, appVer="CMSSW_3_1_1", appFam="RECO", psetHash="GIBBERISH", configContent=self.configURL
)
testFileChild.setDatasetPath("/%s/%s_2/RECO" % (name, name))
testFileChild.addRun(Run(1, *[45]))
testFileChild.setGlobalTag("aGlobalTag")
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([x["lfn"] for x in files])
return files
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()
checksums = {"adler32": "adler32", "cksum": "cksum", "md5": "md5"}
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 createTestFiles(self):
"""
_createTestFiles_
Create some dbsbuffer test files with different statuses
:return:
"""
phedexStatus = self.dbsbufferFactory(classname="DBSBufferFiles.SetPhEDExStatus")
for i in range(0, 4):
lfn = "/path/to/some/lfn" + str(i)
# Two files should be InDBS, two files should be NOTUPLOADED
if i in [0,2]:
status = 'InDBS'
else:
status = 'NOTUPLOADED'
testDBSFile = DBSBufferFile(lfn=lfn, size=600000, events=60000, status=status, workflowId=1)
testDBSFile.setAlgorithm(appName="cmsRun", appVer="UNKNOWN",
appFam="RECO", psetHash="SOMEHASH" + str(i),
configContent="SOMECONTENT")
testDBSFile.setDatasetPath("/path/to/some/dataset")
testDBSFile.create()
# Create all four combinations of status(InDBS,NOTUPLOADED) and in_phedex(0,1)
if i in [0,1]:
phedexStatus.execute(lfn, 1)
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 prepareDBSFiles(self):
"""
_prepareDBSFiles_
Retrieve the information from the JSON input data
and create DBSFile objects that can be registered in the
database.
"""
timestamp = time.strftime('%m%d%y_%H%M%S')
for fileEntry in self.inputData:
# Get all the info out of a standard named dataset
datasetInfo = str(fileEntry["dataset"])
tokens = datasetInfo.split('/')
primDs = tokens[1]
procDs = tokens[2]
dataTier = tokens[3]
procDsTokens = procDs.split('-')
acqEra = procDsTokens[0]
procVer = procDsTokens[-1][1:]
ckSumInfo = fileEntry["checksums"]
for entry in ckSumInfo:
ckSumInfo[entry] = str(ckSumInfo[entry])
# Build the basic dbsBuffer file
dbsFile = DBSBufferFile(lfn = str(fileEntry["lfn"]),
size = int(fileEntry.get("size", 0)),
events = int(fileEntry.get("events", 0)),
checksums = ckSumInfo,
status = "NOTUPLOADED")
dbsFile.setAlgorithm(appName = "cmsRun",
appVer = str(fileEntry.get("cmssw", "LEGACY")),
appFam = "Legacy",
psetHash = "GIBBERISH",
configContent = "None;;None;;None")
dbsFile.setDatasetPath("/%s/%s/%s" % (primDs,
procDs,
dataTier))
dbsFile.setValidStatus(validStatus = "PRODUCTION")
dbsFile.setProcessingVer(ver = procVer)
dbsFile.setAcquisitionEra(era = acqEra)
dbsFile.setGlobalTag(globalTag = str(fileEntry.get('globalTag', "LEGACY")))
# Build a representative task name
dbsFile['task'] = '/LegacyInsertionTask_%s/Insertion' % timestamp
# Get the runs and lumis
runsAndLumis = fileEntry.get("runsAndLumis", {})
for run in runsAndLumis:
newRun = Run(runNumber = int(run))
newRun.extend([int(x) for x in runsAndLumis[run]])
dbsFile.addRun(newRun)
# Complete the file information with the location and queue it
dbsFile.setLocation(se = str(fileEntry["location"]), immediateSave = False)
self.dbsFilesToCreate.append(dbsFile)
self.inputData = None
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 getFiles(self,
name,
tier,
nFiles=12,
site="malpaquet",
workflowName=None,
taskPath=None,
noChild=False):
"""
Create some quick dummy test files
"""
if workflowName is not None and taskPath is not None:
workflowId = self.injectWorkflow(workflowName=workflowName,
taskPath=taskPath)
else:
workflowId = self.injectWorkflow()
files = []
for f in range(0, nFiles):
testFile = DBSBufferFile(lfn='%s-%s-%i' % (name, site, f),
size=1024,
events=20,
checksums={'cksum': 1},
workflowId=workflowId)
testFile.setAlgorithm(appName=name,
appVer="CMSSW_3_1_1",
appFam="RECO",
psetHash="GIBBERISH",
configContent=self.configURL)
testFile.setDatasetPath("/%s/%s/%s" % (name, name, tier))
testFile.addRun(Run(1, *[f]))
testFile.setGlobalTag("aGlobalTag")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
if not noChild:
testFileChild = DBSBufferFile(lfn='%s-%s-child' % (name, site),
size=1024,
events=10,
checksums={'cksum': 1},
workflowId=workflowId)
testFileChild.setAlgorithm(appName=name,
appVer="CMSSW_3_1_1",
appFam="RECO",
psetHash="GIBBERISH",
configContent=self.configURL)
testFileChild.setDatasetPath("/%s/%s_2/RECO" % (name, name))
testFileChild.addRun(Run(1, *[45]))
testFileChild.setGlobalTag("aGlobalTag")
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([x['lfn'] for x in files])
return files
def createFilesWithChildren(self, moreParentFiles, acqEra):
"""
_createFilesWithChildren_
Create several parentless files and then create child files.
"""
parentFiles = []
childFiles = []
baseLFN = "/store/data/%s/Cosmics/RAW/v1/000/143/316/" % (acqEra)
for i in range(10):
testFile = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root", size = 1024,
events = 20, checksums = {"cksum": 1})
testFile.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RAW", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-v1/RAW" % (acqEra))
lumis = []
for j in range(10):
lumis.append((i * 10) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
parentFiles.append(testFile)
baseLFN = "/store/data/%s/Cosmics/RECO/v1/000/143/316/" % (acqEra)
for i in range(5):
testFile = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root", size = 1024,
events = 20, checksums = {"cksum": 1})
testFile.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-v1/RECO" % (acqEra))
lumis = []
for j in range(20):
lumis.append((i * 20) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
testFile.addParents([parentFiles[i * 2]["lfn"],
parentFiles[i * 2 + 1]["lfn"]])
testFile.addParents([moreParentFiles[i * 2]["lfn"],
moreParentFiles[i * 2 + 1]["lfn"]])
childFiles.append(testFile)
return (parentFiles, childFiles)
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 getFiles(self, name, tier, nFiles=12, site="malpaquet", nLumis=1):
"""
_getFiles_
Create some dummy test files.
"""
files = []
(acqEra, procVer) = name.split("-")
baseLFN = "/store/data/%s/Cosmics/RECO/%s/000/143/316/" % (acqEra,
procVer)
for f in range(nFiles):
testFile = DBSBufferFile(lfn=baseLFN + makeUUID() + ".root",
size=1024,
events=20,
checksums={"cksum": 1})
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_3_1_1",
appFam="RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-%s/RECO" % (acqEra, procVer))
lumis = []
for i in range(nLumis):
lumis.append((f * 1000000) + i)
testFile.addRun(Run(1, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("0")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
baseLFN = "/store/data/%s/Cosmics/RAW-RECO/%s/000/143/316/" % (acqEra,
procVer)
testFileChild = DBSBufferFile(lfn=baseLFN + makeUUID() + ".root",
size=1024,
events=10,
checksums={'cksum': 1})
testFileChild.setAlgorithm(appName="cmsRun",
appVer="CMSSW_3_1_1",
appFam="RAW-RECO",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFileChild.setDatasetPath("/Cosmics/%s-%s/RAW-RECO" %
(acqEra, procVer))
testFileChild.addRun(Run(1, *[45]))
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([x['lfn'] for x in files])
return files
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 addFileToDBS(self, jobReportFile, task, errorDataset=False):
"""
_addFileToDBS_
Add a file that was output from a job to the DBS buffer.
"""
datasetInfo = jobReportFile["dataset"]
dbsFile = DBSBufferFile(lfn=jobReportFile["lfn"],
size=jobReportFile["size"],
events=jobReportFile["events"],
checksums=jobReportFile["checksums"],
status="NOTUPLOADED")
dbsFile.setAlgorithm(appName=datasetInfo["applicationName"],
appVer=datasetInfo["applicationVersion"],
appFam=jobReportFile["module_label"],
psetHash="GIBBERISH",
configContent=jobReportFile.get('configURL'))
if errorDataset:
dbsFile.setDatasetPath(
"/%s/%s/%s" %
(datasetInfo["primaryDataset"] + "-Error",
datasetInfo["processedDataset"], datasetInfo["dataTier"]))
else:
dbsFile.setDatasetPath(
"/%s/%s/%s" %
(datasetInfo["primaryDataset"],
datasetInfo["processedDataset"], datasetInfo["dataTier"]))
dbsFile.setValidStatus(
validStatus=jobReportFile.get("validStatus", None))
dbsFile.setProcessingVer(ver=jobReportFile.get('processingVer', None))
dbsFile.setAcquisitionEra(
era=jobReportFile.get('acquisitionEra', None))
dbsFile.setGlobalTag(globalTag=jobReportFile.get('globalTag', None))
#TODO need to find where to get the prep id
dbsFile.setPrepID(prep_id=jobReportFile.get('prep_id', None))
dbsFile['task'] = task
for run in jobReportFile["runs"]:
newRun = Run(runNumber=run.run)
newRun.extend(run.lumis)
dbsFile.addRun(newRun)
dbsFile.setLocation(pnn=list(jobReportFile["locations"])[0],
immediateSave=False)
self.dbsFilesToCreate.append(dbsFile)
return
def injectFilesFromDBS(inputFileset, datasetPath):
"""
_injectFilesFromDBS_
"""
print("injecting files from %s into %s, please wait..." %
(datasetPath, inputFileset.name))
args = {}
args["url"] = "https://cmsweb.cern.ch/dbs/prod/global/DBSReader"
args["mode"] = "GET"
dbsApi = DbsApi(args)
dbsResults = dbsApi.listFileArray(
path=datasetPath, retriveList=["retrive_lumi", "retrive_run"])
# Limiter on number of files
dbsResults = dbsResults[0:20]
print(" found %d files, inserting into wmbs..." % (len(dbsResults)))
for dbsResult in dbsResults:
myFile = File(lfn=dbsResult["LogicalFileName"],
size=dbsResult["FileSize"],
events=dbsResult["NumberOfEvents"],
checksums={"cksum": dbsResult["Checksum"]},
locations="cmssrm.fnal.gov",
merged=True)
myRun = Run(runNumber=dbsResult["LumiList"][0]["RunNumber"])
for lumi in dbsResult["LumiList"]:
myRun.appendLumi(lumi["LumiSectionNumber"])
myFile.addRun(myRun)
myFile.create()
inputFileset.addFile(myFile)
dbsFile = DBSBufferFile(lfn=dbsResult["LogicalFileName"],
size=dbsResult["FileSize"],
events=dbsResult["NumberOfEvents"],
checksums={"cksum": dbsResult["Checksum"]},
locations="cmssrm.fnal.gov",
status="NOTUPLOADED")
dbsFile.setDatasetPath(datasetPath)
dbsFile.setAlgorithm(appName="cmsRun",
appVer="Unknown",
appFam="Unknown",
psetHash="Unknown",
configContent="Unknown")
dbsFile.create()
inputFileset.commit()
inputFileset.markOpen(False)
return
def createParentFiles(self,
acqEra,
nFiles=10,
workflowName='TestWorkload',
taskPath='/TestWorkload/DataTest'):
"""
_createParentFiles_
Create several parentless files in DBSBuffer. This simulates raw files
in the T0.
"""
workflowId = self.injectWorkflow(workflowName=workflowName,
taskPath=taskPath)
parentlessFiles = []
baseLFN = "/store/data/%s/Cosmics/RAW/v1/000/143/316/" % (acqEra)
for i in range(nFiles):
testFile = DBSBufferFile(lfn=baseLFN + makeUUID() + ".root",
size=1024,
events=20,
checksums={"cksum": 1},
workflowId=workflowId)
testFile.setAlgorithm(appName="cmsRun",
appVer="CMSSW_3_1_1",
appFam="RAW",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-v1/RAW" % (acqEra))
testFile['block_close_max_wait_time'] = 1000000
testFile['block_close_max_events'] = 1000000
testFile['block_close_max_size'] = 1000000
testFile['block_close_max_files'] = 1000000
lumis = []
for j in range(10):
lumis.append((i * 10) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
parentlessFiles.append(testFile)
return parentlessFiles
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 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 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 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 addFileToDBS(self, jobReportFile, task, errorDataset = False):
"""
_addFileToDBS_
Add a file that was output from a job to the DBS buffer.
"""
datasetInfo = jobReportFile["dataset"]
dbsFile = DBSBufferFile(lfn = jobReportFile["lfn"],
size = jobReportFile["size"],
events = jobReportFile["events"],
checksums = jobReportFile["checksums"],
status = "NOTUPLOADED")
dbsFile.setAlgorithm(appName = datasetInfo["applicationName"],
appVer = datasetInfo["applicationVersion"],
appFam = jobReportFile["module_label"],
psetHash = "GIBBERISH",
configContent = jobReportFile.get('configURL'))
if errorDataset:
dbsFile.setDatasetPath("/%s/%s/%s" % (datasetInfo["primaryDataset"] + "-Error",
datasetInfo["processedDataset"],
datasetInfo["dataTier"]))
else:
dbsFile.setDatasetPath("/%s/%s/%s" % (datasetInfo["primaryDataset"],
datasetInfo["processedDataset"],
datasetInfo["dataTier"]))
dbsFile.setValidStatus(validStatus = jobReportFile.get("validStatus", None))
dbsFile.setProcessingVer(ver = jobReportFile.get('processingVer', None))
dbsFile.setAcquisitionEra(era = jobReportFile.get('acquisitionEra', None))
dbsFile.setGlobalTag(globalTag = jobReportFile.get('globalTag', None))
#TODO need to find where to get the prep id
dbsFile.setPrepID(prep_id = jobReportFile.get('prep_id', None))
dbsFile['task'] = task
for run in jobReportFile["runs"]:
newRun = Run(runNumber = run.run)
newRun.extend(run.lumis)
dbsFile.addRun(newRun)
dbsFile.setLocation(pnn = list(jobReportFile["locations"])[0], immediateSave = False)
self.dbsFilesToCreate.append(dbsFile)
return
def getFiles(self, name, tier, nFiles = 12, site = "malpaquet", workflowName = None, taskPath = None,
noChild = False):
"""
Create some quick dummy test files
"""
if workflowName is not None and taskPath is not None:
workflowId = self.injectWorkflow(workflowName = workflowName,
taskPath = taskPath)
else:
workflowId = self.injectWorkflow()
files = []
for f in range(0, nFiles):
testFile = DBSBufferFile(lfn = '%s-%s-%i' % (name, site, f), size = 1024,
events = 20, checksums = {'cksum': 1}, workflowId = workflowId)
testFile.setAlgorithm(appName = name, appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = self.configURL)
testFile.setDatasetPath("/%s/%s/%s" % (name, name, tier))
testFile.addRun(Run( 1, *[f]))
testFile.setGlobalTag("aGlobalTag")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
if not noChild:
testFileChild = DBSBufferFile(lfn = '%s-%s-child' %(name, site), size = 1024,
events = 10, checksums = {'cksum': 1},
workflowId = workflowId)
testFileChild.setAlgorithm(appName = name, appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = self.configURL)
testFileChild.setDatasetPath("/%s/%s_2/RECO" %(name, name))
testFileChild.addRun(Run( 1, *[45]))
testFileChild.setGlobalTag("aGlobalTag")
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([x['lfn'] for x in files])
return files
def getFiles(self, name, tier = 'RECO', nFiles = 12, site = "malpaquet", nLumis = 1):
"""
Create some quick dummy test files
"""
files = []
for f in range(nFiles):
testFile = DBSBufferFile(lfn = '/data/store/random/random/RANDOM/test/0/%s-%s-%i.root' % (name, site, f), size = 1024,
events = 20, checksums = {'cksum': 1})
testFile.setAlgorithm(appName = name, appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/%s/%s/%s" % (name, name, tier))
lumis = []
for i in range(nLumis):
lumis.append((f * 100000) + i)
testFile.addRun(Run( 1, *lumis))
testFile.setAcquisitionEra(name.split('-')[0])
testFile.setProcessingVer("0")
testFile.setGlobalTag("Weird")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
count = 0
for f in files:
count += 1
testFileChild = DBSBufferFile(lfn = '/data/store/random/random/RANDOM/test/0/%s-%s-%i-child.root' %(name, site, count), size = 1024,
events = 10, checksums = {'cksum': 1})
testFileChild.setAlgorithm(appName = name, appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFileChild.setDatasetPath("/%s/%s_2/RECO" %(name, name))
testFileChild.addRun(Run( 1, *[45]))
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([f['lfn']])
return files
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 _addToDBSBuffer(self, dbsFile, checksums, locations):
"""
This step is just for increase the performance for
Accountant doesn't neccessary to check the parentage
"""
dbsBuffer = DBSBufferFile(lfn=dbsFile["LogicalFileName"],
size=dbsFile["FileSize"],
events=dbsFile["NumberOfEvents"],
checksums=checksums,
locations=locations,
status="GLOBAL",
inPhedex=1)
dbsBuffer.setDatasetPath('bogus')
dbsBuffer.setAlgorithm(appName="cmsRun", appVer="Unknown",
appFam="Unknown", psetHash="Unknown",
configContent="Unknown")
self.dbsFilesToCreate.add(dbsBuffer)
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 type(testFileB["id"]) == int or type(testFileB["id"]) == long, \
"ERROR: File id is not an integer type."
assert type(testFileB["size"]) == int or type(testFileB["size"]) == long, \
"ERROR: File size is not an integer type."
assert type(testFileB["events"]) == int or type(testFileB["events"]) == long, \
"ERROR: File events is not an integer type."
assert type(testFileC["id"]) == int or type(testFileC["id"]) == long, \
"ERROR: File id is not an integer type."
assert type(testFileC["size"]) == int or type(testFileC["size"]) == long, \
"ERROR: File size is not an integer type."
assert type(testFileC["events"]) == int or type(testFileC["events"]) == long, \
"ERROR: File events is not an integer type."
testFileA.delete()
return
def _addToDBSBuffer(self, dbsFile, checksums, locations):
"""
This step is just for increase the performance for
Accountant doesn't neccessary to check the parentage
"""
dbsBuffer = DBSBufferFile(lfn=dbsFile["LogicalFileName"],
size=dbsFile["FileSize"],
events=dbsFile["NumberOfEvents"],
checksums=checksums,
locations=locations,
status="GLOBAL",
inPhedex=1)
dbsBuffer.setDatasetPath('bogus')
dbsBuffer.setAlgorithm(appName="cmsRun", appVer="Unknown",
appFam="Unknown", psetHash="Unknown",
configContent="Unknown")
self.dbsFilesToCreate.add(dbsBuffer)
return
def createParentFiles(self, acqEra, nFiles=10, workflowName="TestWorkload", taskPath="/TestWorkload/DataTest"):
"""
_createParentFiles_
Create several parentless files in DBSBuffer. This simulates raw files
in the T0.
"""
workflowId = self.injectWorkflow(workflowName=workflowName, taskPath=taskPath)
parentlessFiles = []
baseLFN = "/store/data/%s/Cosmics/RAW/v1/000/143/316/" % (acqEra)
for i in range(nFiles):
testFile = DBSBufferFile(
lfn=baseLFN + makeUUID() + ".root", size=1024, events=20, checksums={"cksum": 1}, workflowId=workflowId
)
testFile.setAlgorithm(
appName="cmsRun",
appVer="CMSSW_3_1_1",
appFam="RAW",
psetHash="GIBBERISH",
configContent="MOREGIBBERISH",
)
testFile.setDatasetPath("/Cosmics/%s-v1/RAW" % (acqEra))
testFile["block_close_max_wait_time"] = 1000000
testFile["block_close_max_events"] = 1000000
testFile["block_close_max_size"] = 1000000
testFile["block_close_max_files"] = 1000000
lumis = []
for j in range(10):
lumis.append((i * 10) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
parentlessFiles.append(testFile)
return parentlessFiles
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
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 getFiles(self, name, tier, nFiles = 12, site = "malpaquet", nLumis = 1):
"""
_getFiles_
Create some dummy test files.
"""
files = []
(acqEra, procVer) = name.split("-")
baseLFN = "/store/data/%s/Cosmics/RECO/%s/000/143/316/" % (acqEra, procVer)
for f in range(nFiles):
testFile = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root", size = 1024,
events = 20, checksums = {"cksum": 1})
testFile.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-%s/RECO" % (acqEra, procVer))
lumis = []
for i in range(nLumis):
lumis.append((f * 1000000) + i)
testFile.addRun(Run( 1, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("0")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation(site)
files.append(testFile)
baseLFN = "/store/data/%s/Cosmics/RAW-RECO/%s/000/143/316/" % (acqEra, procVer)
testFileChild = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root",
size = 1024, events = 10, checksums = {'cksum': 1})
testFileChild.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RAW-RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFileChild.setDatasetPath("/Cosmics/%s-%s/RAW-RECO" %(acqEra, procVer))
testFileChild.addRun(Run( 1, *[45]))
testFileChild.create()
testFileChild.setLocation(site)
testFileChild.addParents([x['lfn'] for x in files])
return files
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
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 injectFilesFromDBS(inputFileset, datasetPath):
"""
_injectFilesFromDBS_
"""
print "injecting files from %s into %s, please wait..." % (datasetPath, inputFileset.name)
args={}
args["url"] = "https://cmsweb.cern.ch/dbs/prod/global/DBSReader"
args["version"] = "DBS_2_0_9"
args["mode"] = "GET"
dbsApi = DbsApi(args)
dbsResults = dbsApi.listFileArray(path = datasetPath, retriveList = ["retrive_lumi", "retrive_run"])
dbsResults = dbsResults[0:10]
print " found %d files, inserting into wmbs..." % (len(dbsResults))
for dbsResult in dbsResults:
myFile = File(lfn = dbsResult["LogicalFileName"], size = dbsResult["FileSize"],
events = dbsResult["NumberOfEvents"], checksums = {"cksum": dbsResult["Checksum"]},
locations = "cmssrm.fnal.gov", merged = True)
myRun = Run(runNumber = dbsResult["LumiList"][0]["RunNumber"])
for lumi in dbsResult["LumiList"]:
myRun.lumis.append(lumi["LumiSectionNumber"])
myFile.addRun(myRun)
myFile.create()
inputFileset.addFile(myFile)
dbsFile = DBSBufferFile(lfn = dbsResult["LogicalFileName"], size = dbsResult["FileSize"],
events = dbsResult["NumberOfEvents"], checksums = {"cksum": dbsResult["Checksum"]},
locations = "cmssrm.fnal.gov", status = "LOCAL")
dbsFile.setDatasetPath(datasetPath)
dbsFile.setAlgorithm(appName = "cmsRun", appVer = "Unknown", appFam = "Unknown",
psetHash = "Unknown", configContent = "Unknown")
dbsFile.create()
inputFileset.commit()
inputFileset.markOpen(False)
return
def createTestFiles(self):
"""
_createTestFiles_
Create some dbsbuffer test files with different statuses
:return:
"""
phedexStatus = self.dbsbufferFactory(
classname="DBSBufferFiles.SetPhEDExStatus")
for i in range(0, 4):
lfn = "/path/to/some/lfn" + str(i)
# Two files should be InDBS, two files should be NOTUPLOADED
if i in [0, 2]:
status = 'InDBS'
else:
status = 'NOTUPLOADED'
testDBSFile = DBSBufferFile(lfn=lfn,
size=600000,
events=60000,
status=status,
workflowId=1)
testDBSFile.setAlgorithm(appName="cmsRun",
appVer="UNKNOWN",
appFam="RECO",
psetHash="SOMEHASH" + str(i),
configContent="SOMECONTENT")
testDBSFile.setDatasetPath("/path/to/some/dataset")
testDBSFile.create()
# Create all four combinations of status(InDBS,NOTUPLOADED) and in_phedex(0,1)
if i in [0, 1]:
phedexStatus.execute(lfn, 1)
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 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 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 createFilesWithChildren(self, moreParentFiles, acqEra):
"""
_createFilesWithChildren_
Create several parentless files and then create child files.
"""
parentFiles = []
childFiles = []
baseLFN = "/store/data/%s/Cosmics/RAW/v1/000/143/316/" % (acqEra)
for i in range(10):
testFile = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root", size = 1024,
events = 20, checksums = {"cksum": 1})
testFile.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RAW", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-v1/RAW" % (acqEra))
testFile['block_close_max_wait_time'] = 1000000
testFile['block_close_max_events'] = 1000000
testFile['block_close_max_size'] = 1000000
testFile['block_close_max_files'] = 1000000
lumis = []
for j in range(10):
lumis.append((i * 10) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
parentFiles.append(testFile)
baseLFN = "/store/data/%s/Cosmics/RECO/v1/000/143/316/" % (acqEra)
for i in range(5):
testFile = DBSBufferFile(lfn = baseLFN + makeUUID() + ".root", size = 1024,
events = 20, checksums = {"cksum": 1})
testFile.setAlgorithm(appName = "cmsRun", appVer = "CMSSW_3_1_1",
appFam = "RECO", psetHash = "GIBBERISH",
configContent = "MOREGIBBERISH")
testFile.setDatasetPath("/Cosmics/%s-v1/RECO" % (acqEra))
testFile['block_close_max_wait_time'] = 1000000
testFile['block_close_max_events'] = 1000000
testFile['block_close_max_size'] = 1000000
testFile['block_close_max_files'] = 1000000
lumis = []
for j in range(20):
lumis.append((i * 20) + j)
testFile.addRun(Run(143316, *lumis))
testFile.setAcquisitionEra(acqEra)
testFile.setProcessingVer("1")
testFile.setGlobalTag("START54::All")
testFile.create()
testFile.setLocation("malpaquet")
testFile.addParents([parentFiles[i * 2]["lfn"],
parentFiles[i * 2 + 1]["lfn"]])
testFile.addParents([moreParentFiles[i * 2]["lfn"],
moreParentFiles[i * 2 + 1]["lfn"]])
childFiles.append(testFile)
return (parentFiles, childFiles)
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 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 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
