def insertFilesForDBSBuffer(
self,
files,
procDataset,
algos,
jobType="NotMerge",
insertDetectorData=False,
maxFiles=100,
maxSize=99999999,
timeOut=None,
fileCommitLength=5,
):
"""
_insertFiles_
list of files inserted in DBS
"""
# TODO: Whats the purpose of insertDetectorData
if len(files) < 1:
return
affectedBlocks = []
insertFiles = []
addedRuns = []
seName = None
# Get the algos in insertable form
# logging.error("About to input algos")
# logging.error(algos)
ialgos = [DBSWriterObjects.createAlgorithmForInsert(dict(algo)) for algo in algos]
# print ialgos
for outFile in files:
# //
# // Convert each file into a DBS File object
# //
lumiList = []
# Somehing similar should be the real deal when multiple runs/lumi could be returned from wmbs file
for runlumiinfo in outFile.getRuns():
lrun = long(runlumiinfo.run)
run = DbsRun(
RunNumber=lrun,
NumberOfEvents=0,
NumberOfLumiSections=0,
TotalLuminosity=0,
StoreNumber=0,
StartOfRun=0,
EndOfRun=0,
)
# Only added if not added by another file in this loop, why waste a call to DBS
if lrun not in addedRuns:
self.dbs.insertRun(run)
addedRuns.append(lrun) # save it so we do not try to add it again to DBS
logging.debug("run %s added to DBS " % str(lrun))
for alsn in runlumiinfo:
lumi = DbsLumiSection(
LumiSectionNumber=long(alsn),
StartEventNumber=0,
EndEventNumber=0,
LumiStartTime=0,
LumiEndTime=0,
RunNumber=lrun,
)
lumiList.append(lumi)
logging.debug("lumi list created for the file")
dbsfile = DbsFile(
# Checksum = str(outFile['cksum']),
NumberOfEvents=outFile["events"],
LogicalFileName=outFile["lfn"],
FileSize=int(outFile["size"]),
Status="VALID",
ValidationStatus="VALID",
FileType="EDM",
Dataset=procDataset,
TierList=DBSWriterObjects.makeTierList(procDataset["Path"].split("/")[3]),
AlgoList=ialgos,
LumiList=lumiList,
ParentList=outFile.getParentLFNs(),
# BranchHash = outFile['BranchHash'],
)
# Set checksums by hand
# dbsfile['Checksum'] = 0 #Set a default?
for entry in outFile["checksums"].keys():
# This should be a dictionary with a cktype key and cksum value
if entry.lower() == "cksum":
dbsfile["Checksum"] = str(outFile["checksums"][entry])
elif entry.lower() == "adler32":
dbsfile["Adler32"] = str(outFile["checksums"][entry])
elif entry.lower() == "md5":
dbsfile["Md5"] = str(outFile["checksums"][entry])
# This check comes from ProdAgent, not sure if its required
if len(outFile["locations"]) > 0:
seName = list(outFile["locations"])[0]
logging.debug("SEname associated to file is: %s" % seName)
else:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No SEname associated to file"
# print "FAKING seName for now"
# seName="cmssrm.fnal.gov"
raise DBSWriterError(msg)
insertFiles.append(dbsfile)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
# // fileblocks
sumSize = 0
sumFiles = 0
tmpFiles = []
blockList = []
# First, get the block. See if the block already exists
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(self.dbs, procDataset, seName)
fileBlock["files"] = []
# if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
except DbsException, ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset["Path"]
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
def insertFiles(self, fwkJobRep, insertDetectorData=False):
"""
_insertFiles_
Process the files in the FwkJobReport instance and insert
them into the associated datasets
A list of affected fileblock names is returned both for merged
and unmerged fileblocks. Only merged blocks will have to be managed.
#for merged file
#blocks to facilitate management of those blocks.
#This list is not populated for processing jobs since we dont really
#care about the processing job blocks.
"""
insertLists = {}
orderedHashes = []
affectedBlocks = set()
if len(fwkJobRep.files) <= 0:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No files found in FrameWorkJobReport for:\n"
msg += "==> JobSpecId: %s" % fwkJobRep.jobSpecId
msg += " Workflow: %s" % fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
for outFile in fwkJobRep.sortFiles():
# //
# // Convert each file into a DBS File object
#//
seName = None
if "SEName" in outFile:
if outFile['SEName']:
seName = outFile['SEName']
logging.debug("SEname associated to file is: %s" % seName)
## remove the fallback to site se-name if no SE is associated to File
## because it's likely that there is some stage out problem if there
## is no SEName associated to the file.
# if not seName:
# if fwkJobRep.siteDetails.has_key("se-name"):
# seName = fwkJobRep.siteDetails['se-name']
# seName = str(seName)
# logging.debug("site SEname: %s"%seName)
if not seName:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No SEname associated to files in FrameWorkJobReport for "
# msg += "No SEname found in FrameWorkJobReport for "
msg += "==> JobSpecId: %s" % fwkJobRep.jobSpecId
msg += " Workflow: %s" % fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
try:
if (insertDetectorData):
dbsFiles = DBSWriterObjects.createDBSFiles(
outFile, fwkJobRep.jobType, self.dbs)
else:
dbsFiles = DBSWriterObjects.createDBSFiles(
outFile, fwkJobRep.jobType)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles:\n"
msg += "Error creating DbsFile instances for file:\n"
msg += "%s\n" % outFile['LFN']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if len(dbsFiles) <= 0:
msg = "No DbsFile instances created. Not enough info in the FrameWorkJobReport for"
msg += "==> JobSpecId: %s" % fwkJobRep.jobSpecId
msg += " Workflow: %s" % fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
for f in dbsFiles:
datasetName = makeDBSDSName(f)
hashName = "%s-%s" % (seName, datasetName)
if hashName not in insertLists:
insertLists[hashName] = _InsertFileList(
seName, datasetName)
insertLists[hashName].append(f)
if not orderedHashes.count(hashName):
orderedHashes.append(hashName)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
#// fileblocks
for hash in orderedHashes:
fileList = insertLists[hash]
procDataset = fileList[0]['Dataset']
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs, procDataset, fileList.seName)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset
msg += "In Storage Element:\n %s\n" % fileList.seName
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if fwkJobRep.jobType == "Merge":
# //
# // Merge files
#//
for mergedFile in fileList:
mergedFile['Block'] = fileBlock
affectedBlocks.add(fileBlock['Name'])
msg = "calling: self.dbs.insertMergedFile(%s, %s)" % (str(
mergedFile['ParentList']), str(mergedFile))
logging.debug(msg)
try:
self.dbs.insertMergedFile(mergedFile['ParentList'],
mergedFile)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert merged file:\n"
msg += " %s\n" % mergedFile['LogicalFileName']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
logging.debug(
"Inserted merged file: %s to FileBlock: %s" %
(mergedFile['LogicalFileName'], fileBlock['Name']))
else:
# //
# // Processing files
#//
affectedBlocks.add(fileBlock['Name'])
msg = "calling: self.dbs.insertFiles(%s, %s, %s)" % (
str(procDataset), str(list(fileList)), str(fileBlock))
logging.debug(msg)
try:
self.dbs.insertFiles(procDataset, list(fileList),
fileBlock)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % ([x['LogicalFileName']
for x in fileList], )
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
logging.debug("Inserted files: %s to FileBlock: %s" %
(([x['LogicalFileName']
for x in fileList]), fileBlock['Name']))
return list(affectedBlocks)
# First see if the block is full
if self.manageFileBlock(
fileBlock=fileBlock,
maxFiles=maxFiles,
maxSize=maxSize,
timeOut=timeOut,
algos=ialgos,
filesToCommit=filesToCommit,
procDataset=procDataset,
):
fileBlock["OpenForWriting"] = 0
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
# Then we need a new block
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(self.dbs, procDataset, seName)
fileBlock["files"] = []
except DbsException, ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset["Path"]
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
fileBlock["files"].append(file["LogicalFileName"])
filesToCommit.append(file)
if len(filesToCommit) >= fileCommitLength:
# Only commit the files if there are more of them then the maximum length
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
def insertFilesForDBSBuffer(self,
files,
procDataset,
algos,
jobType="NotMerge",
insertDetectorData=False,
maxFiles=100,
maxSize=99999999,
timeOut=None,
fileCommitLength=5):
"""
_insertFiles_
list of files inserted in DBS
"""
#TODO: Whats the purpose of insertDetectorData
if len(files) < 1:
return
affectedBlocks = []
insertFiles = []
addedRuns = []
seName = None
#Get the algos in insertable form
# logging.error("About to input algos")
# logging.error(algos)
ialgos = [
DBSWriterObjects.createAlgorithmForInsert(dict(algo))
for algo in algos
]
#print ialgos
for outFile in files:
# //
# // Convert each file into a DBS File object
#//
lumiList = []
#Somehing similar should be the real deal when multiple runs/lumi could be returned from wmbs file
for runlumiinfo in outFile.getRuns():
lrun = long(runlumiinfo.run)
run = DbsRun(
RunNumber=lrun,
NumberOfEvents=0,
NumberOfLumiSections=0,
TotalLuminosity=0,
StoreNumber=0,
StartOfRun=0,
EndOfRun=0,
)
#Only added if not added by another file in this loop, why waste a call to DBS
if lrun not in addedRuns:
self.dbs.insertRun(run)
addedRuns.append(
lrun) #save it so we do not try to add it again to DBS
logging.debug("run %s added to DBS " % str(lrun))
for alsn in runlumiinfo:
lumi = DbsLumiSection(
LumiSectionNumber=long(alsn),
StartEventNumber=0,
EndEventNumber=0,
LumiStartTime=0,
LumiEndTime=0,
RunNumber=lrun,
)
lumiList.append(lumi)
logging.debug("lumi list created for the file")
dbsfile = DbsFile(
#Checksum = str(outFile['cksum']),
NumberOfEvents=outFile['events'],
LogicalFileName=outFile['lfn'],
FileSize=int(outFile['size']),
Status="VALID",
ValidationStatus='VALID',
FileType='EDM',
Dataset=procDataset,
TierList=DBSWriterObjects.makeTierList(
procDataset['Path'].split('/')[3]),
AlgoList=ialgos,
LumiList=lumiList,
ParentList=outFile.getParentLFNs(),
#BranchHash = outFile['BranchHash'],
)
#Set checksums by hand
#dbsfile['Checksum'] = 0 #Set a default?
for entry in outFile['checksums'].keys():
#This should be a dictionary with a cktype key and cksum value
if entry.lower() == 'cksum':
dbsfile['Checksum'] = str(outFile['checksums'][entry])
elif entry.lower() == 'adler32':
dbsfile['Adler32'] = str(outFile['checksums'][entry])
elif entry.lower() == 'md5':
dbsfile['Md5'] = str(outFile['checksums'][entry])
#This check comes from ProdAgent, not sure if its required
if len(outFile["locations"]) > 0:
seName = list(outFile["locations"])[0]
logging.debug("SEname associated to file is: %s" % seName)
else:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No SEname associated to file"
#print "FAKING seName for now"
#seName="cmssrm.fnal.gov"
raise DBSWriterError(msg)
insertFiles.append(dbsfile)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
#// fileblocks
sumSize = 0
sumFiles = 0
tmpFiles = []
blockList = []
#First, get the block. See if the block already exists
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs, procDataset, seName)
fileBlock['files'] = []
#if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
except DbsException as ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
filesToCommit = []
for file in insertFiles:
# First see if the block is full
if self.manageFileBlock(fileBlock=fileBlock,
maxFiles=maxFiles,
maxSize=maxSize,
timeOut=timeOut,
algos=ialgos,
filesToCommit=filesToCommit,
procDataset=procDataset):
fileBlock['OpenForWriting'] = 0
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
# Then we need a new block
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs, procDataset, seName)
fileBlock['files'] = []
except DbsException as ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
fileBlock['files'].append(file['LogicalFileName'])
filesToCommit.append(file)
if len(filesToCommit) >= fileCommitLength:
# Only commit the files if there are more of them then the maximum length
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
logging.debug("Inserted files: %s to FileBlock: %s" \
% ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % (
[x['LogicalFileName'] for x in insertFiles], )
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if len(filesToCommit) > 0:
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
logging.debug("Inserted files: %s to FileBlock: %s" \
% ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % ([x['LogicalFileName']
for x in insertFiles], )
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
## Do bulk inserts now for DBS
#filesToCommit = []
#count = 0
#count2 = 0
#for file in insertFiles:
# count += 1
# #Try and close the box
# logging.error("Should have a file")
# logging.error(len(filesToCommit))
# count2 += len(filesToCommit)
# if self.manageFileBlock(fileBlock = fileBlock, maxFiles = maxFiles,
# maxSize = maxSize, timeOut = timeOut, algos = ialgos,
# filesToCommit = filesToCommit, procDataset = procDataset):
# fileBlock['OpenForWriting'] = '0'
# if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
#
#
#
# # Then we need a new block
# try:
# fileBlock = DBSWriterObjects.getDBSFileBlock(
# self.dbs,
# procDataset,
# seName)
# fileBlock['files'] = []
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
# msg += "Cannot retrieve FileBlock for dataset:\n"
# msg += " %s\n" % procDataset['Path']
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
# #At this point, we should commit the block as is
# fileBlock['files'].append(file['LogicalFileName'])
# if jobType == "MergeSpecial":
# for file in fileList:
# file['Block'] = fileBlock
# msg="calling: self.dbs.insertMergedFile(%s, %s)" % (str(file['ParentList']),str(file))
# logging.debug(msg)
# try:
# #
# #
# # NOTE To Anzar From Anzar (File cloning as in DBS API can be done here and then I can use Bulk insert on Merged files as well)
# self.dbs.insertMergedFile(file['ParentList'],
# file)
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert merged file:\n"
# msg += " %s\n" % file['LogicalFileName']
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
# logging.debug("Inserted merged file: %s to FileBlock: %s"%(file['LogicalFileName'],fileBlock['Name']))
# else:
# filesToCommit.append(file)
# if len(filesToCommit) >= fileCommitLength:
# # Only commit the files if there are more of them then the maximum length
# try:
# logging.error("About to commit %i files" %(len(filesToCommit)))
# count2 += len(filesToCommit)
# self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
# filesToCommit = []
# logging.debug("Inserted files: %s to FileBlock: %s" \
# % ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert processed files:\n"
# msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
#
#
#
#
## If we still have files to commit, commit them
#logging.error("Got to the end of the loop")
#logging.error(len(filesToCommit))
#logging.error(count2)
#if len(filesToCommit) > 0:
# try:
# logging.error("About to insert some files")
# self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
# filesToCommit = []
# logging.debug("Inserted files: %s to FileBlock: %s" \
# % ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert processed files:\n"
# msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
return list(affectedBlocks)
def insertFilesForDBSBuffer(self,
files,
procDataset,
algos,
jobType="NotMerge",
insertDetectorData=False,
maxFiles=100,
maxSize=99999999,
timeOut=None,
fileCommitLength=5):
"""
_insertFiles_
list of files inserted in DBS
"""
#TODO: Whats the purpose of insertDetectorData
if len(files) < 1:
return
affectedBlocks = []
insertFiles = []
addedRuns = []
seName = None
#Get the algos in insertable form
# logging.error("About to input algos")
# logging.error(algos)
ialgos = [
DBSWriterObjects.createAlgorithmForInsert(dict(algo))
for algo in algos
]
#print ialgos
for outFile in files:
# //
# // Convert each file into a DBS File object
#//
lumiList = []
#Somehing similar should be the real deal when multiple runs/lumi could be returned from wmbs file
for runlumiinfo in outFile.getRuns():
lrun = long(runlumiinfo.run)
run = DbsRun(
RunNumber=lrun,
NumberOfEvents=0,
NumberOfLumiSections=0,
TotalLuminosity=0,
StoreNumber=0,
StartOfRun=0,
EndOfRun=0,
)
#Only added if not added by another file in this loop, why waste a call to DBS
if lrun not in addedRuns:
self.dbs.insertRun(run)
addedRuns.append(
lrun) #save it so we do not try to add it again to DBS
logging.debug("run %s added to DBS " % str(lrun))
for alsn in runlumiinfo:
lumi = DbsLumiSection(
LumiSectionNumber=long(alsn),
StartEventNumber=0,
EndEventNumber=0,
LumiStartTime=0,
LumiEndTime=0,
RunNumber=lrun,
)
lumiList.append(lumi)
logging.debug("lumi list created for the file")
dbsfile = DbsFile(
#Checksum = str(outFile['cksum']),
NumberOfEvents=outFile['events'],
LogicalFileName=outFile['lfn'],
FileSize=int(outFile['size']),
Status="VALID",
ValidationStatus='VALID',
FileType='EDM',
Dataset=procDataset,
TierList=DBSWriterObjects.makeTierList(
procDataset['Path'].split('/')[3]),
AlgoList=ialgos,
LumiList=lumiList,
ParentList=outFile.getParentLFNs(),
#BranchHash = outFile['BranchHash'],
)
#Set checksums by hand
#dbsfile['Checksum'] = 0 #Set a default?
for entry in outFile['checksums'].keys():
#This should be a dictionary with a cktype key and cksum value
if entry.lower() == 'cksum':
dbsfile['Checksum'] = str(outFile['checksums'][entry])
elif entry.lower() == 'adler32':
dbsfile['Adler32'] = str(outFile['checksums'][entry])
elif entry.lower() == 'md5':
dbsfile['Md5'] = str(outFile['checksums'][entry])
#This check comes from ProdAgent, not sure if its required
if len(outFile["locations"]) > 0:
seName = list(outFile["locations"])[0]
logging.debug("SEname associated to file is: %s" % seName)
else:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No SEname associated to file"
#print "FAKING seName for now"
#seName="cmssrm.fnal.gov"
raise DBSWriterError(msg)
insertFiles.append(dbsfile)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
#// fileblocks
sumSize = 0
sumFiles = 0
tmpFiles = []
blockList = []
#First, get the block. See if the block already exists
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs, procDataset, seName)
fileBlock['files'] = []
#if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
except DbsException, ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
filesToCommit = []
for file in insertFiles:
# First see if the block is full
if self.manageFileBlock(fileBlock=fileBlock,
maxFiles=maxFiles,
maxSize=maxSize,
timeOut=timeOut,
algos=ialgos,
filesToCommit=filesToCommit,
procDataset=procDataset):
fileBlock['OpenForWriting'] = 0
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
# Then we need a new block
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs, procDataset, seName)
fileBlock['files'] = []
except DbsException, ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
fileBlock['files'].append(file['LogicalFileName'])
filesToCommit.append(file)
if len(filesToCommit) >= fileCommitLength:
# Only commit the files if there are more of them then the maximum length
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
def insertFiles(self, fwkJobRep, insertDetectorData = False):
"""
_insertFiles_
Process the files in the FwkJobReport instance and insert
them into the associated datasets
A list of affected fileblock names is returned both for merged
and unmerged fileblocks. Only merged blocks will have to be managed.
#for merged file
#blocks to facilitate management of those blocks.
#This list is not populated for processing jobs since we dont really
#care about the processing job blocks.
"""
insertLists = {}
orderedHashes = []
affectedBlocks = set()
if len(fwkJobRep.files)<=0:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No files found in FrameWorkJobReport for:\n"
msg += "==> JobSpecId: %s"%fwkJobRep.jobSpecId
msg += " Workflow: %s"%fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
for outFile in fwkJobRep.sortFiles():
# //
# // Convert each file into a DBS File object
#//
pnn = None
if "PNN" in outFile:
if outFile['PNN'] :
pnn = outFile['PNN']
logging.debug("PNN associated to file is: %s"%pnn)
## remove the fallback to site se-name if no SE is associated to File
## because it's likely that there is some stage out problem if there
## is no SEName associated to the file.
# if not seName:
# if fwkJobRep.siteDetails.has_key("se-name"):
# seName = fwkJobRep.siteDetails['se-name']
# seName = str(seName)
# logging.debug("site SEname: %s"%seName)
if not pnn:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No PNN associated to files in FrameWorkJobReport for "
# msg += "No SEname found in FrameWorkJobReport for "
msg += "==> JobSpecId: %s"%fwkJobRep.jobSpecId
msg += " Workflow: %s"%fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
try:
if ( insertDetectorData ):
dbsFiles = DBSWriterObjects.createDBSFiles(outFile,
fwkJobRep.jobType,
self.dbs)
else:
dbsFiles = DBSWriterObjects.createDBSFiles(outFile,
fwkJobRep.jobType)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles:\n"
msg += "Error creating DbsFile instances for file:\n"
msg += "%s\n" % outFile['LFN']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if len(dbsFiles)<=0:
msg="No DbsFile instances created. Not enough info in the FrameWorkJobReport for"
msg += "==> JobSpecId: %s"%fwkJobRep.jobSpecId
msg += " Workflow: %s"%fwkJobRep.workflowSpecId
raise DBSWriterError(msg)
for f in dbsFiles:
datasetName = makeDBSDSName(f)
hashName = "%s-%s" % (pnn, datasetName)
if hashName not in insertLists:
insertLists[hashName] = _InsertFileList(pnn,
datasetName)
insertLists[hashName].append(f)
if not orderedHashes.count(hashName):
orderedHashes.append(hashName)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
#// fileblocks
for hash in orderedHashes:
fileList = insertLists[hash]
procDataset = fileList[0]['Dataset']
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs,
procDataset,
fileList.pnn)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset
# msg += "In Storage Element:\n %s\n" % fileList.seName
msg += "In PNN:\n %s\n" % fileList.pnn
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if fwkJobRep.jobType == "Merge":
# //
# // Merge files
#//
for mergedFile in fileList:
mergedFile['Block'] = fileBlock
affectedBlocks.add(fileBlock['Name'])
msg="calling: self.dbs.insertMergedFile(%s, %s)" % (str(mergedFile['ParentList']),str(mergedFile))
logging.debug(msg)
try:
self.dbs.insertMergedFile(mergedFile['ParentList'],
mergedFile)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert merged file:\n"
msg += " %s\n" % mergedFile['LogicalFileName']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
logging.debug("Inserted merged file: %s to FileBlock: %s"%(mergedFile['LogicalFileName'],fileBlock['Name']))
else:
# //
# // Processing files
#//
affectedBlocks.add(fileBlock['Name'])
msg="calling: self.dbs.insertFiles(%s, %s, %s)" % (str(procDataset),str(list(fileList)),str(fileBlock))
logging.debug(msg)
try:
self.dbs.insertFiles(procDataset, list(fileList),
fileBlock)
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % (
[ x['LogicalFileName'] for x in fileList ],
)
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
logging.debug("Inserted files: %s to FileBlock: %s"%( ([ x['LogicalFileName'] for x in fileList ]),fileBlock['Name']))
return list(affectedBlocks)
def insertFilesForDBSBuffer(self, files, procDataset, algos,
jobType = "NotMerge", insertDetectorData = False,
maxFiles = 100, maxSize = 99999999, timeOut = None,
fileCommitLength = 5):
"""
_insertFiles_
list of files inserted in DBS
"""
#TODO: Whats the purpose of insertDetectorData
if len(files) < 1:
return
affectedBlocks = []
insertFiles = []
addedRuns=[]
pnn = None
#Get the algos in insertable form
# logging.error("About to input algos")
# logging.error(algos)
ialgos = [DBSWriterObjects.createAlgorithmForInsert(dict(algo)) for algo in algos ]
#print ialgos
for outFile in files:
# //
# // Convert each file into a DBS File object
#//
lumiList = []
#Somehing similar should be the real deal when multiple runs/lumi could be returned from wmbs file
for runlumiinfo in outFile.getRuns():
lrun=long(runlumiinfo.run)
run = DbsRun(
RunNumber = lrun,
NumberOfEvents = 0,
NumberOfLumiSections = 0,
TotalLuminosity = 0,
StoreNumber = 0,
StartOfRun = 0,
EndOfRun = 0,
)
#Only added if not added by another file in this loop, why waste a call to DBS
if lrun not in addedRuns:
self.dbs.insertRun(run)
addedRuns.append(lrun) #save it so we do not try to add it again to DBS
logging.debug("run %s added to DBS " % str(lrun))
for alsn in runlumiinfo:
lumi = DbsLumiSection(
LumiSectionNumber = long(alsn),
StartEventNumber = 0,
EndEventNumber = 0,
LumiStartTime = 0,
LumiEndTime = 0,
RunNumber = lrun,
)
lumiList.append(lumi)
logging.debug("lumi list created for the file")
dbsfile = DbsFile(
#Checksum = str(outFile['cksum']),
NumberOfEvents = outFile['events'],
LogicalFileName = outFile['lfn'],
FileSize = int(outFile['size']),
Status = "VALID",
ValidationStatus = 'VALID',
FileType = 'EDM',
Dataset = procDataset,
TierList = DBSWriterObjects.makeTierList(procDataset['Path'].split('/')[3]),
AlgoList = ialgos,
LumiList = lumiList,
ParentList = outFile.getParentLFNs(),
#BranchHash = outFile['BranchHash'],
)
#Set checksums by hand
#dbsfile['Checksum'] = 0 #Set a default?
for entry in outFile['checksums'].keys():
#This should be a dictionary with a cktype key and cksum value
if entry.lower() == 'cksum':
dbsfile['Checksum'] = str(outFile['checksums'][entry])
elif entry.lower() == 'adler32':
dbsfile['Adler32'] = str(outFile['checksums'][entry])
elif entry.lower() == 'md5':
dbsfile['Md5'] = str(outFile['checksums'][entry])
#This check comes from ProdAgent, not sure if its required
if len(outFile["locations"]) > 0:
pnn = list(outFile["locations"])[0]
logging.debug("PNN associated to file is: %s"%pnn)
else:
msg = "Error in DBSWriter.insertFiles\n"
msg += "No PNN associated to file"
#print "FAKING seName for now"
#seName="cmssrm.fnal.gov"
raise DBSWriterError(msg)
insertFiles.append(dbsfile)
# //Processing Jobs:
# // Insert the lists of sorted files into the appropriate
#// fileblocks
sumSize = 0
sumFiles = 0
tmpFiles = []
blockList = []
#First, get the block. See if the block already exists
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs,
procDataset,
pnn)
fileBlock['files'] = []
#if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
except DbsException as ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
filesToCommit = []
for file in insertFiles:
# First see if the block is full
if self.manageFileBlock(fileBlock = fileBlock, maxFiles = maxFiles,
maxSize = maxSize, timeOut = timeOut, algos = ialgos,
filesToCommit = filesToCommit, procDataset = procDataset):
fileBlock['OpenForWriting'] = 0
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
# Then we need a new block
try:
fileBlock = DBSWriterObjects.getDBSFileBlock(
self.dbs,
procDataset,
pnn)
fileBlock['files'] = []
except DbsException as ex:
msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
msg += "Cannot retrieve FileBlock for dataset:\n"
msg += " %s\n" % procDataset['Path']
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
fileBlock['files'].append(file['LogicalFileName'])
filesToCommit.append(file)
if len(filesToCommit) >= fileCommitLength:
# Only commit the files if there are more of them then the maximum length
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
logging.debug("Inserted files: %s to FileBlock: %s" \
% ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if len(filesToCommit) > 0:
try:
self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
filesToCommit = []
logging.debug("Inserted files: %s to FileBlock: %s" \
% ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
except DbsException as ex:
msg = "Error in DBSWriter.insertFiles\n"
msg += "Cannot insert processed files:\n"
msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
msg += "%s\n" % formatEx(ex)
raise DBSWriterError(msg)
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
## Do bulk inserts now for DBS
#filesToCommit = []
#count = 0
#count2 = 0
#for file in insertFiles:
# count += 1
# #Try and close the box
# logging.error("Should have a file")
# logging.error(len(filesToCommit))
# count2 += len(filesToCommit)
# if self.manageFileBlock(fileBlock = fileBlock, maxFiles = maxFiles,
# maxSize = maxSize, timeOut = timeOut, algos = ialgos,
# filesToCommit = filesToCommit, procDataset = procDataset):
# fileBlock['OpenForWriting'] = '0'
# if not fileBlock in affectedBlocks:
# affectedBlocks.append(fileBlock)
#
#
#
# # Then we need a new block
# try:
# fileBlock = DBSWriterObjects.getDBSFileBlock(
# self.dbs,
# procDataset,
# seName)
# fileBlock['files'] = []
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFilesForDBSBuffer\n"
# msg += "Cannot retrieve FileBlock for dataset:\n"
# msg += " %s\n" % procDataset['Path']
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
# #At this point, we should commit the block as is
# fileBlock['files'].append(file['LogicalFileName'])
# if jobType == "MergeSpecial":
# for file in fileList:
# file['Block'] = fileBlock
# msg="calling: self.dbs.insertMergedFile(%s, %s)" % (str(file['ParentList']),str(file))
# logging.debug(msg)
# try:
# #
# #
# # NOTE To Anzar From Anzar (File cloning as in DBS API can be done here and then I can use Bulk insert on Merged files as well)
# self.dbs.insertMergedFile(file['ParentList'],
# file)
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert merged file:\n"
# msg += " %s\n" % file['LogicalFileName']
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
# logging.debug("Inserted merged file: %s to FileBlock: %s"%(file['LogicalFileName'],fileBlock['Name']))
# else:
# filesToCommit.append(file)
# if len(filesToCommit) >= fileCommitLength:
# # Only commit the files if there are more of them then the maximum length
# try:
# logging.error("About to commit %i files" %(len(filesToCommit)))
# count2 += len(filesToCommit)
# self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
# filesToCommit = []
# logging.debug("Inserted files: %s to FileBlock: %s" \
# % ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert processed files:\n"
# msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
#
#
#
#
## If we still have files to commit, commit them
#logging.error("Got to the end of the loop")
#logging.error(len(filesToCommit))
#logging.error(count2)
#if len(filesToCommit) > 0:
# try:
# logging.error("About to insert some files")
# self.dbs.insertFiles(procDataset, filesToCommit, fileBlock)
# filesToCommit = []
# logging.debug("Inserted files: %s to FileBlock: %s" \
# % ( ([ x['LogicalFileName'] for x in insertFiles ]),fileBlock['Name']))
#
# except DbsException, ex:
# msg = "Error in DBSWriter.insertFiles\n"
# msg += "Cannot insert processed files:\n"
# msg += " %s\n" % ([ x['LogicalFileName'] for x in insertFiles ],)
# msg += "%s\n" % formatEx(ex)
# raise DBSWriterError(msg)
if not fileBlock in affectedBlocks:
affectedBlocks.append(fileBlock)
return list(affectedBlocks)
