def defineEvgen16Job(self, i):
"""Define an Evgen16 job based on predefined values and randomly generated names
"""
job = JobSpec()
job.computingSite = self.__site
job.cloud = self.__cloud
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (uuid.uuid1(), i)
job.AtlasRelease = 'Atlas-16.6.2'
job.homepackage = 'AtlasProduction/16.6.2.1'
job.transformation = 'Evgen_trf.py'
job.destinationDBlock = self.__datasetName
job.destinationSE = self.__destName
job.currentPriority = 10000
job.prodSourceLabel = 'test'
job.cmtConfig = 'i686-slc5-gcc43-opt'
#Output file
fileO = FileSpec()
fileO.lfn = "%s.evgen.pool.root" % job.jobName
fileO.destinationDBlock = job.destinationDBlock
fileO.destinationSE = job.destinationSE
fileO.dataset = job.destinationDBlock
fileO.destinationDBlockToken = 'ATLASDATADISK'
fileO.type = 'output'
job.addFile(fileO)
#Log file
fileL = FileSpec()
fileL.lfn = "%s.job.log.tgz" % job.jobName
fileL.destinationDBlock = job.destinationDBlock
fileL.destinationSE = job.destinationSE
fileL.dataset = job.destinationDBlock
fileL.destinationDBlockToken = 'ATLASDATADISK'
fileL.type = 'log'
job.addFile(fileL)
job.jobParameters = "2760 105048 19901 101 200 MC10.105048.PythiaB_ccmu3mu1X.py %s NONE NONE NONE MC10JobOpts-latest-test.tar.gz" % fileO.lfn
return job
def defineEvgen16Job(self, i):
"""Define an Evgen16 job based on predefined values and randomly generated names
"""
job = JobSpec()
job.computingSite = self.__site
job.cloud = self.__cloud
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (uuid.uuid1(), i)
job.AtlasRelease = 'Atlas-16.6.2'
job.homepackage = 'AtlasProduction/16.6.2.1'
job.transformation = 'Evgen_trf.py'
job.destinationDBlock = self.__datasetName
job.destinationSE = self.__destName
job.currentPriority = 10000
job.prodSourceLabel = 'test'
job.cmtConfig = 'i686-slc5-gcc43-opt'
#Output file
fileO = FileSpec()
fileO.lfn = "%s.evgen.pool.root" % job.jobName
fileO.destinationDBlock = job.destinationDBlock
fileO.destinationSE = job.destinationSE
fileO.dataset = job.destinationDBlock
fileO.destinationDBlockToken = 'ATLASDATADISK'
fileO.type = 'output'
job.addFile(fileO)
#Log file
fileL = FileSpec()
fileL.lfn = "%s.job.log.tgz" % job.jobName
fileL.destinationDBlock = job.destinationDBlock
fileL.destinationSE = job.destinationSE
fileL.dataset = job.destinationDBlock
fileL.destinationDBlockToken = 'ATLASDATADISK'
fileL.type = 'log'
job.addFile(fileL)
job.jobParameters = "2760 105048 19901 101 200 MC10.105048.PythiaB_ccmu3mu1X.py %s NONE NONE NONE MC10JobOpts-latest-test.tar.gz" % fileO.lfn
return job
def run(self):
for i in range(1):
prodDBlock = 'rome.004201.evgen.ZeeJimmy'
destinationDBlock = 'pandatest.000123.test.simul'
destinationSE = 'BNL_SE'
jobs = []
#for i in range(self.interval):
for i in range(2):
job = JobSpec()
job.jobDefinitionID=self.jobDefinitionID
job.AtlasRelease='Atlas-11.0.1'
job.prodDBlock=prodDBlock
job.destinationDBlock=destinationDBlock
job.destinationSE=destinationSE
job.currentPriority=i
lfnI = 'rome.004201.evgen.ZeeJimmy._00001.pool.root'
file = FileSpec()
file.lfn = lfnI
file.dataset = 'rome.004201.evgen.ZeeJimmy'
file.type = 'input'
file.prodDBlock = prodDBlock
file.dataset = prodDBlock
job.addFile(file)
lfnO ='%s.pool.root.1' % commands.getoutput('uuidgen')
file = FileSpec()
file.lfn = lfnO
file.type = 'output'
file.destinationDBlock = destinationDBlock
file.dataset = destinationDBlock
file.destinationSE = destinationSE
job.addFile(file)
job.homepackage='JobTransforms-11-00-01-01'
job.transformation='share/rome.g4sim.standard.trf'
job.jobParameters='%s %s 1 2 14268' % (lfnI,lfnO)
jobs.append(job)
self.taskbuffer.storeJobs(jobs,None)
time.sleep(self.interval)
def master_prepare(self,app,appmasterconfig):
# PandaTools
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaMCPandaRTHandler master_prepare called for %s', job.getFQID('.'))
usertag = configDQ2['usertag']
#usertag='user09'
nickname = getNickname(allowMissingNickname=True)
self.libDataset = '%s.%s.ganga.%s_%d.lib._%06d' % (usertag,nickname,commands.getoutput('hostname').split('.')[0],int(time.time()),job.id)
# self.userprefix='%s.%s.ganga' % (usertag,gridProxy.identity())
sources = 'sources.%s.tar.gz' % commands.getoutput('uuidgen 2> /dev/null')
self.library = '%s.lib.tgz' % self.libDataset
# check DBRelease
# if job.backend.dbRelease != '' and job.backend.dbRelease.find(':') == -1:
# raise ApplicationConfigurationError(None,"ERROR : invalid argument for backend.dbRelease. Must be 'DatasetName:FileName'")
# unpack library
logger.debug('Creating source tarball ...')
tmpdir = '/tmp/%s' % commands.getoutput('uuidgen 2> /dev/null')
os.mkdir(tmpdir)
inputbox=[]
if os.path.exists(app.transform_archive):
# must add a condition on size.
inputbox += [ File(app.transform_archive) ]
if app.evgen_job_option:
self.evgen_job_option=app.evgen_job_option
if os.path.exists(app.evgen_job_option):
# locally modified job option file to add to the input sand box
inputbox += [ File(app.evgen_job_option) ]
self.evgen_job_option=app.evgen_job_option.split("/")[-1]
# add input sandbox files
if (job.inputsandbox):
for file in job.inputsandbox:
inputbox += [ file ]
# add option files
for extFile in job.backend.extOutFile:
try:
shutil.copy(extFile,tmpdir)
except IOError:
os.makedirs(tmpdir)
shutil.copy(extFile,tmpdir)
# fill the archive
for opt_file in inputbox:
try:
shutil.copy(opt_file.name,tmpdir)
except IOError:
os.makedirs(tmpdir)
shutil.copy(opt_file.name,tmpdir)
# now tar it up again
inpw = job.getInputWorkspace()
rc, output = commands.getstatusoutput('tar czf %s -C %s .' % (inpw.getPath(sources),tmpdir))
if rc:
logger.error('Packing sources failed with status %d',rc)
logger.error(output)
raise ApplicationConfigurationError(None,'Packing sources failed.')
shutil.rmtree(tmpdir)
# upload sources
logger.debug('Uploading source tarball ...')
try:
cwd = os.getcwd()
os.chdir(inpw.getPath())
rc, output = Client.putFile(sources)
if output != 'True':
logger.error('Uploading sources %s failed. Status = %d', sources, rc)
logger.error(output)
raise ApplicationConfigurationError(None,'Uploading archive failed')
finally:
os.chdir(cwd)
# Use Panda's brokerage
## if job.inputdata and len(app.sites)>0:
## # update cloud, use inputdata's
## from dq2.info.TiersOfATLAS import whichCloud,ToACache
## inclouds=[]
## for site in app.sites:
## cloudSite=whichCloud(app.sites[0])
## if cloudSite not in inclouds:
## inclouds.append(cloudSite)
## # now converting inclouds content into proper brokering stuff.
## outclouds=[]
## for cloudSite in inclouds:
## for cloudID, eachCloud in ToACache.dbcloud.iteritems():
## if cloudSite==eachCloud:
## cloud=cloudID
## outclouds.append(cloud)
## break
## print outclouds
## # finally, matching with user's wishes
## if len(outclouds)>0:
## if not job.backend.requirements.cloud: # no user wish, update
## job.backend.requirements.cloud=outclouds[0]
## else:
## try:
## assert job.backend.requirements.cloud in outclouds
## except:
## raise ApplicationConfigurationError(None,'Input dataset not available in target cloud %s. Please try any of the following %s' % (job.backend.requirements.cloud, str(outclouds)))
from GangaPanda.Lib.Panda.Panda import runPandaBrokerage
runPandaBrokerage(job)
if job.backend.site == 'AUTO':
raise ApplicationConfigurationError(None,'site is still AUTO after brokerage!')
# output dataset preparation and registration
try:
outDsLocation = Client.PandaSites[job.backend.site]['ddm']
except:
raise ApplicationConfigurationError(None,"Could not extract output dataset location from job.backend.site value: %s. Aborting" % job.backend.site)
if not app.dryrun:
for outtype in app.outputpaths.keys():
dset=string.replace(app.outputpaths[outtype],"/",".")
dset=dset[1:]
# dataset registration must be done only once.
print "registering output dataset %s at %s" % (dset,outDsLocation)
try:
Client.addDataset(dset,False,location=outDsLocation)
dq2_set_dataset_lifetime(dset, location=outDsLocation)
except:
raise ApplicationConfigurationError(None,"Fail to create output dataset %s. Aborting" % dset)
# extend registration to build job lib dataset:
print "registering output dataset %s at %s" % (self.libDataset,outDsLocation)
try:
Client.addDataset(self.libDataset,False,location=outDsLocation)
dq2_set_dataset_lifetime(self.libDataset, outDsLocation)
except:
raise ApplicationConfigurationError(None,"Fail to create output dataset %s. Aborting" % self.libDataset)
###
cacheVer = "-AtlasProduction_" + str(app.prod_release)
logger.debug("master job submit?")
self.outsite=job.backend.site
if app.se_name and app.se_name != "none" and not self.outsite:
self.outsite=app.se_name
# create build job
jspec = JobSpec()
jspec.jobDefinitionID = job.id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_rel
jspec.homepackage = 'AnalysisTransforms'+cacheVer#+nightVer
jspec.transformation = '%s/buildJob-00-00-03' % Client.baseURLSUB # common base to Athena and AthenaMC jobs: buildJob is a pilot job which takes care of all inputs for the real jobs (in prepare()
jspec.destinationDBlock = self.libDataset
jspec.destinationSE = job.backend.site
jspec.prodSourceLabel = 'panda'
jspec.assignedPriority = 2000
jspec.computingSite = job.backend.site
jspec.cloud = job.backend.requirements.cloud
# jspec.jobParameters = self.args not known yet
jspec.jobParameters = '-o %s' % (self.library)
if app.userarea:
print app.userarea
jspec.jobParameters += ' -i %s' % (os.path.basename(app.userarea))
else:
jspec.jobParameters += ' -i %s' % (sources)
jspec.cmtConfig = AthenaUtils.getCmtConfig(athenaVer=app.atlas_rel)
matchURL = re.search('(http.*://[^/]+)/',Client.baseURLSSL)
if matchURL:
jspec.jobParameters += ' --sourceURL %s' % matchURL.group(1)
fout = FileSpec()
fout.lfn = self.library
fout.type = 'output'
fout.dataset = self.libDataset
fout.destinationDBlock = self.libDataset
jspec.addFile(fout)
flog = FileSpec()
flog.lfn = '%s.log.tgz' % self.libDataset
flog.type = 'log'
flog.dataset = self.libDataset
flog.destinationDBlock = self.libDataset
jspec.addFile(flog)
#print "MASTER JOB DETAILS:",jspec.jobParameters
return jspec
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'BNL_ATLAS_2'
files = {
'EVNT.019128._00011.pool.root.1': None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), index)
job.AtlasRelease = 'Atlas-13.0.40'
job.homepackage = 'AtlasProduction/13.0.40.3'
job.transformation = 'csc_simul_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.computingSite = site
job.prodDBlock = 'valid1.005001.pythia_minbias.evgen.EVNT.e306_tid019128'
job.prodSourceLabel = 'test'
job.currentPriority = 10000
job.cloud = 'IT'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
fileI.lfn = lfn
import userinterface.Client as Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
if len(sys.argv)>2:
site = sys.argv[1]
cloud = sys.argv[2]
else:
site = "UTA_PAUL_TEST"
cloud = "US"
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_1" % commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-20.1.4'
job.homepackage = 'AtlasProduction/20.1.4.14'
#job.AtlasRelease = 'Atlas-20.20.8'
#job.homepackage = 'AtlasProduction/20.20.8.4'
job.transformation = 'Reco_tf.py'
job.destinationDBlock = 'panda.destDB.%s' % commands.getoutput('uuidgen')
job.destinationSE = 'AGLT2_TEST'
job.prodDBlock = 'user.mlassnig:user.mlassnig.pilot.test.single.hits'
job.currentPriority = 1000
#job.prodSourceLabel = 'ptest'
job.prodSourceLabel = 'user'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'x86_64-slc6-gcc48-opt'
job.specialHandling = 'ddm:rucio'
#job.transferType = 'direct'
def run(self):
try:
self.putLog('start %s' % self.evpFileName)
# lock evp file
self.evpFile = open(self.evpFileName)
try:
fcntl.flock(self.evpFile.fileno(),
fcntl.LOCK_EX | fcntl.LOCK_NB)
except:
# relase
self.putLog("cannot lock %s" % self.evpFileName)
self.evpFile.close()
return True
# options
runEvtList = []
eventPickDataType = ''
eventPickStreamName = ''
eventPickDS = []
eventPickAmiTag = ''
eventPickNumSites = 1
inputFileList = []
tagDsList = []
tagQuery = ''
tagStreamRef = ''
skipDaTRI = False
runEvtGuidMap = {}
# read evp file
for tmpLine in self.evpFile:
tmpMatch = re.search('^([^=]+)=(.+)$', tmpLine)
# check format
if tmpMatch == None:
continue
tmpItems = tmpMatch.groups()
if tmpItems[0] == 'runEvent':
# get run and event number
tmpRunEvt = tmpItems[1].split(',')
if len(tmpRunEvt) == 2:
runEvtList.append(tmpRunEvt)
elif tmpItems[0] == 'eventPickDataType':
# data type
eventPickDataType = tmpItems[1]
elif tmpItems[0] == 'eventPickStreamName':
# stream name
eventPickStreamName = tmpItems[1]
elif tmpItems[0] == 'eventPickDS':
# dataset pattern
eventPickDS = tmpItems[1].split(',')
elif tmpItems[0] == 'eventPickAmiTag':
# AMI tag
eventPickAmiTag = tmpItems[1]
elif tmpItems[0] == 'eventPickNumSites':
# the number of sites where datasets are distributed
try:
eventPickNumSites = int(tmpItems[1])
except:
pass
elif tmpItems[0] == 'userName':
# user name
self.userDN = tmpItems[1]
self.putLog("user=%s" % self.userDN)
elif tmpItems[0] == 'userTaskName':
# user task name
self.userTaskName = tmpItems[1]
elif tmpItems[0] == 'userDatasetName':
# user dataset name
self.userDatasetName = tmpItems[1]
elif tmpItems[0] == 'lockedBy':
# client name
self.lockedBy = tmpItems[1]
elif tmpItems[0] == 'creationTime':
# creation time
self.creationTime = tmpItems[1]
elif tmpItems[0] == 'params':
# parameters
self.params = tmpItems[1]
elif tmpItems[0] == 'inputFileList':
# input file list
inputFileList = tmpItems[1].split(',')
try:
inputFileList.remove('')
except:
pass
elif tmpItems[0] == 'tagDS':
# TAG dataset
tagDsList = tmpItems[1].split(',')
elif tmpItems[0] == 'tagQuery':
# query for TAG
tagQuery = tmpItems[1]
elif tmpItems[0] == 'tagStreamRef':
# StreamRef for TAG
tagStreamRef = tmpItems[1]
if not tagStreamRef.endswith('_ref'):
tagStreamRef += '_ref'
elif tmpItems[0] == 'runEvtGuidMap':
# GUIDs
try:
exec "runEvtGuidMap=" + tmpItems[1]
except:
pass
# extract task name
if self.userTaskName == '' and self.params != '':
try:
tmpMatch = re.search('--outDS(=| ) *([^ ]+)', self.params)
if tmpMatch != None:
self.userTaskName = tmpMatch.group(2)
if not self.userTaskName.endswith('/'):
self.userTaskName += '/'
except:
pass
# suppress DaTRI
if self.params != '':
if '--eventPickSkipDaTRI' in self.params:
skipDaTRI = True
# get compact user name
compactDN = self.taskBuffer.cleanUserID(self.userDN)
# get jediTaskID
self.jediTaskID = self.taskBuffer.getTaskIDwithTaskNameJEDI(
compactDN, self.userTaskName)
# convert
if tagDsList == [] or tagQuery == '':
# convert run/event list to dataset/file list
tmpRet, locationMap, allFiles = self.pd2p.convertEvtRunToDatasets(
runEvtList, eventPickDataType, eventPickStreamName,
eventPickDS, eventPickAmiTag, self.userDN, runEvtGuidMap)
if not tmpRet:
if 'isFatal' in locationMap and locationMap[
'isFatal'] == True:
self.ignoreError = False
self.endWithError(
'Failed to convert the run/event list to a dataset/file list'
)
return False
else:
# get parent dataset/files with TAG
tmpRet, locationMap, allFiles = self.pd2p.getTagParentInfoUsingTagQuery(
tagDsList, tagQuery, tagStreamRef)
if not tmpRet:
self.endWithError(
'Failed to get parent dataset/file list with TAG')
return False
# use only files in the list
if inputFileList != []:
tmpAllFiles = []
for tmpFile in allFiles:
if tmpFile['lfn'] in inputFileList:
tmpAllFiles.append(tmpFile)
allFiles = tmpAllFiles
# remove redundant CN from DN
tmpDN = self.userDN
tmpDN = re.sub('/CN=limited proxy', '', tmpDN)
tmpDN = re.sub('(/CN=proxy)+$', '', tmpDN)
# make dataset container
tmpRet = self.pd2p.registerDatasetContainerWithDatasets(
self.userDatasetName,
allFiles,
locationMap,
nSites=eventPickNumSites,
owner=tmpDN)
if not tmpRet:
self.endWithError('Failed to make a dataset container %s' %
self.userDatasetName)
return False
# skip DaTRI
if skipDaTRI:
# successfully terminated
self.putLog("skip DaTRI")
# update task
self.taskBuffer.updateTaskModTimeJEDI(self.jediTaskID)
else:
# get candidates
tmpRet, candidateMaps = self.pd2p.getCandidates(
self.userDatasetName, checkUsedFile=False, useHidden=True)
if not tmpRet:
self.endWithError(
'Failed to find candidate for destination')
return False
# collect all candidates
allCandidates = []
for tmpDS, tmpDsVal in candidateMaps.iteritems():
for tmpCloud, tmpCloudVal in tmpDsVal.iteritems():
for tmpSiteName in tmpCloudVal[0]:
if not tmpSiteName in allCandidates:
allCandidates.append(tmpSiteName)
if allCandidates == []:
self.endWithError('No candidate for destination')
return False
# get list of dataset (container) names
if eventPickNumSites > 1:
# decompose container to transfer datasets separately
tmpRet, tmpOut = self.pd2p.getListDatasetReplicasInContainer(
self.userDatasetName)
if not tmpRet:
self.endWithError('Failed to get the size of %s' %
self.userDatasetName)
return False
userDatasetNameList = tmpOut.keys()
else:
# transfer container at once
userDatasetNameList = [self.userDatasetName]
# loop over all datasets
sitesUsed = []
for tmpUserDatasetName in userDatasetNameList:
# get size of dataset container
tmpRet, totalInputSize = rucioAPI.getDatasetSize(
tmpUserDatasetName)
if not tmpRet:
self.endWithError('Failed to get the size of %s' %
tmpUserDatasetName)
return False
# run brokerage
tmpJob = JobSpec()
tmpJob.AtlasRelease = ''
self.putLog("run brokerage for %s" % tmpDS)
brokerage.broker.schedule([tmpJob],
self.taskBuffer,
self.siteMapper,
True,
allCandidates,
True,
datasetSize=totalInputSize)
if tmpJob.computingSite.startswith('ERROR'):
self.endWithError('brokerage failed with %s' %
tmpJob.computingSite)
return False
self.putLog("site -> %s" % tmpJob.computingSite)
# send transfer request
try:
tmpDN = rucioAPI.parse_dn(tmpDN)
tmpStatus, userInfo = rucioAPI.finger(tmpDN)
if not tmpStatus:
raise RuntimeError, 'user info not found for {0} with {1}'.format(
tmpDN, userInfo)
tmpDN = userInfo['nickname']
tmpDQ2ID = self.siteMapper.getSite(
tmpJob.computingSite).ddm
tmpMsg = "%s ds=%s site=%s id=%s" % (
'registerDatasetLocation for DaTRI ',
tmpUserDatasetName, tmpDQ2ID, tmpDN)
self.putLog(tmpMsg)
rucioAPI.registerDatasetLocation(
tmpDS, [tmpDQ2ID],
lifetime=14,
owner=tmpDN,
activity="User Subscriptions")
self.putLog('OK')
except:
errType, errValue = sys.exc_info()[:2]
tmpStr = 'Failed to send transfer request : %s %s' % (
errType, errValue)
tmpStr.strip()
tmpStr += traceback.format_exc()
self.endWithError(tmpStr)
return False
# list of sites already used
sitesUsed.append(tmpJob.computingSite)
self.putLog("used %s sites" % len(sitesUsed))
# set candidates
if len(sitesUsed) >= eventPickNumSites:
# reset candidates to limit the number of sites
allCandidates = sitesUsed
sitesUsed = []
else:
# remove site
allCandidates.remove(tmpJob.computingSite)
# send email notification for success
tmpMsg = 'A transfer request was successfully sent to Rucio.\n'
tmpMsg += 'Your task will get started once transfer is completed.'
self.sendEmail(True, tmpMsg)
try:
# unlock and delete evp file
fcntl.flock(self.evpFile.fileno(), fcntl.LOCK_UN)
self.evpFile.close()
os.remove(self.evpFileName)
except:
pass
# successfully terminated
self.putLog("end %s" % self.evpFileName)
return True
except:
errType, errValue = sys.exc_info()[:2]
self.endWithError('Got exception %s:%s %s' %
(errType, errValue, traceback.format_exc()))
return False
if len(sys.argv) > 1:
site = sys.argv[1]
else:
site = None
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'BNL_ATLAS_2'
jobList = []
for i in range(20):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-11.0.41'
#job.AtlasRelease = 'Atlas-11.0.3'
job.homepackage = 'AnalysisTransforms'
job.transformation = 'https://gridui01.usatlas.bnl.gov:24443/dav/test/runAthena'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 100
job.prodSourceLabel = 'user'
job.computingSite = site
#job.prodDBlock = "pandatest.b1599dfa-cd36-4fc5-92f6-495781a94c66"
job.prodDBlock = "pandatest.f228b051-077b-4f81-90bf-496340644379"
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
fileI.lfn = "lib.f228b051-077b-4f81-90bf-496340644379.tgz"
import time
import commands
import userinterface.Client as Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('/usr/bin/uuidgen')
job.AtlasRelease = 'Atlas-9.0.4'
job.prodDBlock = 'pandatest.000003.dd.input'
job.destinationDBlock = 'panda.destDB.%s' % commands.getoutput('/usr/bin/uuidgen')
job.destinationSE = 'BNL_SE'
ids = {'pandatest.000003.dd.input._00028.junk':'6c19e1fc-ee8c-4bae-bd4c-c9e5c73aca27',
'pandatest.000003.dd.input._00033.junk':'98f79ba1-1793-4253-aac7-bdf90a51d1ee',
'pandatest.000003.dd.input._00039.junk':'33660dd5-7cef-422a-a7fc-6c24cb10deb1'}
for lfn in ids.keys():
file = FileSpec()
file.lfn = lfn
file.GUID = ids[file.lfn]
file.dataset = 'pandatest.000003.dd.input'
file.type = 'input'
job.addFile(file)
s,o = Client.submitJobs([job])
print "---------------------"
print s
print o
print "---------------------"
s,o = Client.getJobStatus([4934, 4766, 4767, 4768, 4769])
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
files = {
inputFile:None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),index)
job.AtlasRelease = 'Atlas-15.3.1'
job.homepackage = 'AtlasProduction/15.3.1.5'
job.transformation = 'csc_atlasG4_trf.py'
job.destinationDBlock = datasetName
job.computingSite = site
job.prodDBlock = prodDBlock
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc4-gcc34-opt'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
files = {
inputFile:None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),index)
job.AtlasRelease = 'Atlas-17.0.5'
job.homepackage = 'AtlasProduction/17.0.5.6'
job.transformation = 'AtlasG4_trf.py'
job.destinationDBlock = datasetName
job.computingSite = site
job.prodDBlock = prodDBlock
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc5-gcc43-opt'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
destName = 'BNL_ATLAS_2'
files = {
'EVNT.023986._00001.pool.root.1': None,
#'EVNT.023989._00001.pool.root.1':None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), index)
job.AtlasRelease = 'Atlas-14.2.20'
job.homepackage = 'AtlasProduction/14.2.20.1'
job.transformation = 'csc_simul_reco_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.computingSite = site
job.prodDBlock = 'mc08.105031.Jimmy_jetsJ2.evgen.EVNT.e347_tid023986'
#job.prodDBlock = 'mc08.105034.Jimmy_jetsJ5.evgen.EVNT.e347_tid023989'
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
fileI = FileSpec()
fileI.dataset = job.prodDBlock
# instantiate JobSpecs
iJob = 0
jobList = []
for line in taskFile:
iJob += 1
job = JobSpec()
# job ID ###### FIXME
job.jobDefinitionID = int(time.time()) % 10000
# job name
job.jobName = "%s_%05d.job" % (taskName,iJob)
# AtlasRelease
if len(re.findall('\.',trfVer)) > 2:
match = re.search('^(\d+\.\d+\.\d+)',trfVer)
job.AtlasRelease = 'Atlas-%s' % match.group(1)
else:
job.AtlasRelease = 'Atlas-%s' % trfVer
# homepackage
vers = trfVer.split('.')
if int(vers[0]) <= 11:
job.homepackage = 'JobTransforms'
for ver in vers:
job.homepackage += "-%02d" % int(ver)
else:
job.homepackage = 'AtlasProduction/%s' % trfVer
# trf
job.transformation = trf
job.destinationDBlock = oDatasets[0]
# prod DBlock
job.prodDBlock = iDataset
import userinterface.Client as Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
if len(sys.argv) > 2:
site = sys.argv[1]
cloud = sys.argv[2]
else:
site = "UTA_PAUL_TEST"
cloud = "US"
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_1" % commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-20.1.4'
job.homepackage = 'AtlasProduction/20.1.4.14'
#job.AtlasRelease = 'Atlas-20.20.8'
#job.homepackage = 'AtlasProduction/20.20.8.4'
job.transformation = 'Reco_tf.py'
job.destinationDBlock = 'panda.destDB.%s' % commands.getoutput('uuidgen')
job.destinationSE = 'AGLT2_TEST'
job.prodDBlock = 'user.mlassnig:user.mlassnig.pilot.test.single.hits'
job.currentPriority = 1000
#job.prodSourceLabel = 'ptest'
job.prodSourceLabel = 'user'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'x86_64-slc6-gcc48-opt'
job.specialHandling = 'ddm:rucio'
#job.transferType = 'direct'
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'BNL_ATLAS_2'
files = {
'EVNT.019128._00011.pool.root.1':None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),index)
job.AtlasRelease = 'Atlas-13.0.40'
job.homepackage = 'AtlasProduction/13.0.40.3'
job.transformation = 'csc_simul_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.computingSite = site
job.prodDBlock = 'valid1.005001.pythia_minbias.evgen.EVNT.e306_tid019128'
job.prodSourceLabel = 'test'
job.currentPriority = 10000
job.cloud = 'IT'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
fileI.lfn = lfn
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
files = {
inputFile:None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),index)
job.AtlasRelease = 'Atlas-16.6.2'
job.homepackage = 'AtlasProduction/16.6.2.1'
job.transformation = 'AtlasG4_trf.py'
job.destinationDBlock = datasetName
job.computingSite = site
job.prodDBlock = prodDBlock
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc5-gcc43-opt'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
def master_prepare(self, app, appmasterconfig):
# PandaTools
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaMCPandaRTHandler master_prepare called for %s',
job.getFQID('.'))
usertag = configDQ2['usertag']
#usertag='user09'
nickname = getNickname(allowMissingNickname=True)
self.libDataset = '%s.%s.ganga.%s_%d.lib._%06d' % (
usertag, nickname, commands.getoutput('hostname').split('.')[0],
int(time.time()), job.id)
# self.userprefix='%s.%s.ganga' % (usertag,gridProxy.identity())
sources = 'sources.%s.tar.gz' % commands.getoutput(
'uuidgen 2> /dev/null')
self.library = '%s.lib.tgz' % self.libDataset
# check DBRelease
# if job.backend.dbRelease != '' and job.backend.dbRelease.find(':') == -1:
# raise ApplicationConfigurationError(None,"ERROR : invalid argument for backend.dbRelease. Must be 'DatasetName:FileName'")
# unpack library
logger.debug('Creating source tarball ...')
tmpdir = '/tmp/%s' % commands.getoutput('uuidgen 2> /dev/null')
os.mkdir(tmpdir)
inputbox = []
if os.path.exists(app.transform_archive):
# must add a condition on size.
inputbox += [File(app.transform_archive)]
if app.evgen_job_option:
self.evgen_job_option = app.evgen_job_option
if os.path.exists(app.evgen_job_option):
# locally modified job option file to add to the input sand box
inputbox += [File(app.evgen_job_option)]
self.evgen_job_option = app.evgen_job_option.split("/")[-1]
# add input sandbox files
if (job.inputsandbox):
for file in job.inputsandbox:
inputbox += [file]
# add option files
for extFile in job.backend.extOutFile:
try:
shutil.copy(extFile, tmpdir)
except IOError:
os.makedirs(tmpdir)
shutil.copy(extFile, tmpdir)
# fill the archive
for opt_file in inputbox:
try:
shutil.copy(opt_file.name, tmpdir)
except IOError:
os.makedirs(tmpdir)
shutil.copy(opt_file.name, tmpdir)
# now tar it up again
inpw = job.getInputWorkspace()
rc, output = commands.getstatusoutput('tar czf %s -C %s .' %
(inpw.getPath(sources), tmpdir))
if rc:
logger.error('Packing sources failed with status %d', rc)
logger.error(output)
raise ApplicationConfigurationError(None,
'Packing sources failed.')
shutil.rmtree(tmpdir)
# upload sources
logger.debug('Uploading source tarball ...')
try:
cwd = os.getcwd()
os.chdir(inpw.getPath())
rc, output = Client.putFile(sources)
if output != 'True':
logger.error('Uploading sources %s failed. Status = %d',
sources, rc)
logger.error(output)
raise ApplicationConfigurationError(
None, 'Uploading archive failed')
finally:
os.chdir(cwd)
# Use Panda's brokerage
## if job.inputdata and len(app.sites)>0:
## # update cloud, use inputdata's
## from dq2.info.TiersOfATLAS import whichCloud,ToACache
## inclouds=[]
## for site in app.sites:
## cloudSite=whichCloud(app.sites[0])
## if cloudSite not in inclouds:
## inclouds.append(cloudSite)
## # now converting inclouds content into proper brokering stuff.
## outclouds=[]
## for cloudSite in inclouds:
## for cloudID, eachCloud in ToACache.dbcloud.iteritems():
## if cloudSite==eachCloud:
## cloud=cloudID
## outclouds.append(cloud)
## break
## print outclouds
## # finally, matching with user's wishes
## if len(outclouds)>0:
## if not job.backend.requirements.cloud: # no user wish, update
## job.backend.requirements.cloud=outclouds[0]
## else:
## try:
## assert job.backend.requirements.cloud in outclouds
## except:
## raise ApplicationConfigurationError(None,'Input dataset not available in target cloud %s. Please try any of the following %s' % (job.backend.requirements.cloud, str(outclouds)))
from GangaPanda.Lib.Panda.Panda import runPandaBrokerage
runPandaBrokerage(job)
if job.backend.site == 'AUTO':
raise ApplicationConfigurationError(
None, 'site is still AUTO after brokerage!')
# output dataset preparation and registration
try:
outDsLocation = Client.PandaSites[job.backend.site]['ddm']
except:
raise ApplicationConfigurationError(
None,
"Could not extract output dataset location from job.backend.site value: %s. Aborting"
% job.backend.site)
if not app.dryrun:
for outtype in app.outputpaths.keys():
dset = string.replace(app.outputpaths[outtype], "/", ".")
dset = dset[1:]
# dataset registration must be done only once.
print "registering output dataset %s at %s" % (dset,
outDsLocation)
try:
Client.addDataset(dset, False, location=outDsLocation)
dq2_set_dataset_lifetime(dset, location=outDsLocation)
except:
raise ApplicationConfigurationError(
None,
"Fail to create output dataset %s. Aborting" % dset)
# extend registration to build job lib dataset:
print "registering output dataset %s at %s" % (self.libDataset,
outDsLocation)
try:
Client.addDataset(self.libDataset,
False,
location=outDsLocation)
dq2_set_dataset_lifetime(self.libDataset, outDsLocation)
except:
raise ApplicationConfigurationError(
None, "Fail to create output dataset %s. Aborting" %
self.libDataset)
###
cacheVer = "-AtlasProduction_" + str(app.prod_release)
logger.debug("master job submit?")
self.outsite = job.backend.site
if app.se_name and app.se_name != "none" and not self.outsite:
self.outsite = app.se_name
# create build job
jspec = JobSpec()
jspec.jobDefinitionID = job.id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_rel
jspec.homepackage = 'AnalysisTransforms' + cacheVer #+nightVer
jspec.transformation = '%s/buildJob-00-00-03' % Client.baseURLSUB # common base to Athena and AthenaMC jobs: buildJob is a pilot job which takes care of all inputs for the real jobs (in prepare()
jspec.destinationDBlock = self.libDataset
jspec.destinationSE = job.backend.site
jspec.prodSourceLabel = 'panda'
jspec.assignedPriority = 2000
jspec.computingSite = job.backend.site
jspec.cloud = job.backend.requirements.cloud
# jspec.jobParameters = self.args not known yet
jspec.jobParameters = '-o %s' % (self.library)
if app.userarea:
print app.userarea
jspec.jobParameters += ' -i %s' % (os.path.basename(app.userarea))
else:
jspec.jobParameters += ' -i %s' % (sources)
jspec.cmtConfig = AthenaUtils.getCmtConfig(athenaVer=app.atlas_rel)
matchURL = re.search('(http.*://[^/]+)/', Client.baseURLSSL)
if matchURL:
jspec.jobParameters += ' --sourceURL %s' % matchURL.group(1)
fout = FileSpec()
fout.lfn = self.library
fout.type = 'output'
fout.dataset = self.libDataset
fout.destinationDBlock = self.libDataset
jspec.addFile(fout)
flog = FileSpec()
flog.lfn = '%s.log.tgz' % self.libDataset
flog.type = 'log'
flog.dataset = self.libDataset
flog.destinationDBlock = self.libDataset
jspec.addFile(flog)
#print "MASTER JOB DETAILS:",jspec.jobParameters
return jspec
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
site = sys.argv[1]
cloud = sys.argv[2]
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = None
jobList = []
for i in range(1):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), i)
job.AtlasRelease = 'Atlas-17.0.5'
job.homepackage = 'AtlasProduction/17.0.5.6'
job.transformation = 'Evgen_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 10000
job.prodSourceLabel = 'test'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'i686-slc5-gcc43-opt'
file = FileSpec()
file.lfn = "%s.evgen.pool.root" % job.jobName
file.destinationDBlock = job.destinationDBlock
file.destinationSE = job.destinationSE
file.dataset = job.destinationDBlock
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
files = {
inputFile: None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), index)
job.AtlasRelease = 'Atlas-15.3.1'
job.homepackage = 'AtlasProduction/15.3.1.5'
job.transformation = 'csc_atlasG4_trf.py'
job.destinationDBlock = datasetName
job.computingSite = site
job.prodDBlock = prodDBlock
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc4-gcc34-opt'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
def run(self):
try:
# get job
tmpJobs = self.taskBuffer.getFullJobStatus([self.rPandaID])
if tmpJobs == [] or tmpJobs[0] == None:
_logger.debug("cannot find job for PandaID=%s" % self.rPandaID)
return
self.job = tmpJobs[0]
_logger.debug("%s start %s:%s:%s" % (self.token,self.job.jobDefinitionID,self.job.prodUserName,self.job.computingSite))
# using output container
if not self.job.destinationDBlock.endswith('/'):
_logger.debug("%s ouput dataset container is required" % self.token)
_logger.debug("%s end" % self.token)
return
# FIXEME : dont' touch group jobs for now
if self.job.destinationDBlock.startswith('group') and (not self.userRequest):
_logger.debug("%s skip group jobs" % self.token)
_logger.debug("%s end" % self.token)
return
# check processingType
typesForRebro = ['pathena','prun','ganga','ganga-rbtest']
if not self.job.processingType in typesForRebro:
_logger.debug("%s skip processingType=%s not in %s" % \
(self.token,self.job.processingType,str(typesForRebro)))
_logger.debug("%s end" % self.token)
return
# check jobsetID
if self.job.jobsetID in [0,'NULL',None]:
_logger.debug("%s jobsetID is undefined" % self.token)
_logger.debug("%s end" % self.token)
return
# check metadata
if self.job.metadata in [None,'NULL']:
_logger.debug("%s metadata is unavailable" % self.token)
_logger.debug("%s end" % self.token)
return
# check --disableRebrokerage
match = re.search("--disableRebrokerage",self.job.metadata)
if match != None and (not self.simulation) and (not self.forceOpt) \
and (not self.userRequest):
_logger.debug("%s diabled rebrokerage" % self.token)
_logger.debug("%s end" % self.token)
return
# check --site
match = re.search("--site",self.job.metadata)
if match != None and (not self.simulation) and (not self.forceOpt) \
and (not self.userRequest):
_logger.debug("%s --site is used" % self.token)
_logger.debug("%s end" % self.token)
return
# check --libDS
match = re.search("--libDS",self.job.metadata)
if match != None:
_logger.debug("%s --libDS is used" % self.token)
_logger.debug("%s end" % self.token)
return
# check --workingGroup since it is site-specific
match = re.search("--workingGroup",self.job.metadata)
if match != None:
_logger.debug("%s workingGroup is specified" % self.token)
_logger.debug("%s end" % self.token)
return
# avoid too many rebrokerage
if not self.checkRev():
_logger.debug("%s avoid too many rebrokerage" % self.token)
_logger.debug("%s end" % self.token)
return
# check if multiple JobIDs use the same libDS
if self.bPandaID != None and self.buildStatus not in ['finished','failed']:
if self.minPandaIDlibDS == None or self.maxPandaIDlibDS == None:
_logger.debug("%s max/min PandaIDs are unavailable for the libDS" % self.token)
_logger.debug("%s end" % self.token)
return
tmpPandaIDsForLibDS = self.taskBuffer.getFullJobStatus([self.minPandaIDlibDS,self.maxPandaIDlibDS])
if len(tmpPandaIDsForLibDS) != 2 or tmpPandaIDsForLibDS[0] == None or tmpPandaIDsForLibDS[1] == None:
_logger.debug("%s failed to get max/min PandaIDs for the libDS" % self.token)
_logger.debug("%s end" % self.token)
return
# check
if tmpPandaIDsForLibDS[0].jobDefinitionID != tmpPandaIDsForLibDS[1].jobDefinitionID:
_logger.debug("%s multiple JobIDs use the libDS %s:%s %s:%s" % (self.token,tmpPandaIDsForLibDS[0].jobDefinitionID,
self.minPandaIDlibDS,tmpPandaIDsForLibDS[1].jobDefinitionID,
self.maxPandaIDlibDS))
_logger.debug("%s end" % self.token)
return
# check excludedSite
if self.excludedSite == None:
self.excludedSite = []
match = re.search("--excludedSite( +|=)\s*(\'|\")*([^ \"\';$]+)",self.job.metadata)
if match != None:
self.excludedSite = match.group(3).split(',')
# remove empty
try:
self.excludedSite.remove('')
except:
pass
_logger.debug("%s excludedSite=%s" % (self.token,str(self.excludedSite)))
# check cloud
if self.cloud == None:
match = re.search("--cloud( +|=)\s*(\'|\")*([^ \"\';$]+)",self.job.metadata)
if match != None:
self.cloud = match.group(3)
_logger.debug("%s cloud=%s" % (self.token,self.cloud))
# get inDS/LFNs
status,tmpMapInDS,maxFileSize = self.taskBuffer.getInDatasetsForReBrokerage(self.jobID,self.userName)
if not status:
# failed
_logger.error("%s failed to get inDS/LFN from DB" % self.token)
return
status,inputDS = self.getListDatasetsUsedByJob(tmpMapInDS)
if not status:
# failed
_logger.error("%s failed" % self.token)
return
# get relicas
replicaMap = {}
unknownSites = {}
for tmpDS in inputDS:
if tmpDS.endswith('/'):
# container
status,tmpRepMaps = self.getListDatasetReplicasInContainer(tmpDS)
else:
# normal dataset
status,tmpRepMap = self.getListDatasetReplicas(tmpDS)
tmpRepMaps = {tmpDS:tmpRepMap}
if not status:
# failed
_logger.debug("%s failed" % self.token)
return
# make map per site
for tmpDS,tmpRepMap in tmpRepMaps.iteritems():
for tmpSite,tmpStat in tmpRepMap.iteritems():
# ignore special sites
if tmpSite in ['CERN-PROD_TZERO','CERN-PROD_DAQ','CERN-PROD_TMPDISK']:
continue
# ignore tape sites
if tmpSite.endswith('TAPE'):
continue
# keep sites with unknown replica info
if tmpStat[-1]['found'] == None:
if not unknownSites.has_key(tmpDS):
unknownSites[tmpDS] = []
unknownSites[tmpDS].append(tmpSite)
# ignore ToBeDeleted
if tmpStat[-1]['archived'] in ['ToBeDeleted',]:
continue
# change EOS
if tmpSite.startswith('CERN-PROD_EOS'):
tmpSite = 'CERN-PROD_EOS'
# change EOS TMP
if tmpSite.startswith('CERN-PROD_TMP'):
tmpSite = 'CERN-PROD_TMP'
# change DISK to SCRATCHDISK
tmpSite = re.sub('_[^_-]+DISK$','',tmpSite)
# change PERF-XYZ to SCRATCHDISK
tmpSite = re.sub('_PERF-[^_-]+$','',tmpSite)
# change PHYS-XYZ to SCRATCHDISK
tmpSite = re.sub('_PHYS-[^_-]+$','',tmpSite)
# patch for BNLPANDA
if tmpSite in ['BNLPANDA']:
tmpSite = 'BNL-OSG2'
# add to map
if not replicaMap.has_key(tmpSite):
replicaMap[tmpSite] = {}
replicaMap[tmpSite][tmpDS] = tmpStat[-1]
_logger.debug("%s replica map -> %s" % (self.token,str(replicaMap)))
# refresh replica info in needed
self.refreshReplicaInfo(unknownSites)
# instantiate SiteMapper
siteMapper = SiteMapper(self.taskBuffer)
# get original DDM
origSiteDDM = self.getAggName(siteMapper.getSite(self.job.computingSite).ddm)
# check all datasets
maxDQ2Sites = []
if inputDS != []:
# loop over all sites
for tmpSite,tmpDsVal in replicaMap.iteritems():
# loop over all datasets
appendFlag = True
for tmpOrigDS in inputDS:
# check completeness
if tmpDsVal.has_key(tmpOrigDS) and tmpDsVal[tmpOrigDS]['found'] != None and \
tmpDsVal[tmpOrigDS]['total'] == tmpDsVal[tmpOrigDS]['found']:
pass
else:
appendFlag = False
# append
if appendFlag:
if not tmpSite in maxDQ2Sites:
maxDQ2Sites.append(tmpSite)
_logger.debug("%s candidate DQ2s -> %s" % (self.token,str(maxDQ2Sites)))
if inputDS != [] and maxDQ2Sites == []:
_logger.debug("%s no DQ2 candidate" % self.token)
else:
maxPandaSites = []
# original maxinputsize
origMaxInputSize = siteMapper.getSite(self.job.computingSite).maxinputsize
# look for Panda siteIDs
for tmpSiteID,tmpSiteSpec in siteMapper.siteSpecList.iteritems():
# use ANALY_ only
if not tmpSiteID.startswith('ANALY_'):
continue
# remove test and local
if re.search('_test',tmpSiteID,re.I) != None:
continue
if re.search('_local',tmpSiteID,re.I) != None:
continue
# avoid same site
if self.avoidSameSite and self.getAggName(tmpSiteSpec.ddm) == origSiteDDM:
continue
# check DQ2 ID
if self.cloud in [None,tmpSiteSpec.cloud] \
and (self.getAggName(tmpSiteSpec.ddm) in maxDQ2Sites or inputDS == []):
# excluded sites
excludedFlag = False
for tmpExcSite in self.excludedSite:
if re.search(tmpExcSite,tmpSiteID) != None:
excludedFlag = True
break
if excludedFlag:
_logger.debug("%s skip %s since excluded" % (self.token,tmpSiteID))
continue
# use online only
if tmpSiteSpec.status != 'online':
_logger.debug("%s skip %s status=%s" % (self.token,tmpSiteID,tmpSiteSpec.status))
continue
# check maxinputsize
if (maxFileSize == None and origMaxInputSize > siteMapper.getSite(tmpSiteID).maxinputsize) or \
maxFileSize > siteMapper.getSite(tmpSiteID).maxinputsize:
_logger.debug("%s skip %s due to maxinputsize" % (self.token,tmpSiteID))
continue
# append
if not tmpSiteID in maxPandaSites:
maxPandaSites.append(tmpSiteID)
# choose at most 20 sites randomly to avoid too many lookup
random.shuffle(maxPandaSites)
maxPandaSites = maxPandaSites[:20]
_logger.debug("%s candidate PandaSites -> %s" % (self.token,str(maxPandaSites)))
# no Panda siteIDs
if maxPandaSites == []:
_logger.debug("%s no Panda site candidate" % self.token)
else:
# set AtlasRelease and cmtConfig to dummy job
tmpJobForBrokerage = JobSpec()
if self.job.AtlasRelease in ['NULL',None]:
tmpJobForBrokerage.AtlasRelease = ''
else:
tmpJobForBrokerage.AtlasRelease = self.job.AtlasRelease
# use nightlies
matchNight = re.search('^AnalysisTransforms-.*_(rel_\d+)$',self.job.homepackage)
if matchNight != None:
tmpJobForBrokerage.AtlasRelease += ':%s' % matchNight.group(1)
# use cache
else:
matchCache = re.search('^AnalysisTransforms-([^/]+)',self.job.homepackage)
if matchCache != None:
tmpJobForBrokerage.AtlasRelease = matchCache.group(1).replace('_','-')
if not self.job.cmtConfig in ['NULL',None]:
tmpJobForBrokerage.cmtConfig = self.job.cmtConfig
# memory size
if not self.job.minRamCount in ['NULL',None,0]:
tmpJobForBrokerage.minRamCount = self.job.minRamCount
# CPU count
if not self.job.maxCpuCount in ['NULL',None,0]:
tmpJobForBrokerage.maxCpuCount = self.job.maxCpuCount
# run brokerage
brokerage.broker.schedule([tmpJobForBrokerage],self.taskBuffer,siteMapper,forAnalysis=True,
setScanSiteList=maxPandaSites,trustIS=True,reportLog=True)
newSiteID = tmpJobForBrokerage.computingSite
self.brokerageInfo += tmpJobForBrokerage.brokerageErrorDiag
_logger.debug("%s runBrokerage - > %s" % (self.token,newSiteID))
# unknown site
if not siteMapper.checkSite(newSiteID):
_logger.error("%s unknown site" % self.token)
_logger.debug("%s failed" % self.token)
return
# get new site spec
newSiteSpec = siteMapper.getSite(newSiteID)
# avoid repetition
if self.getAggName(newSiteSpec.ddm) == origSiteDDM:
_logger.debug("%s assigned to the same site %s " % (self.token,newSiteID))
_logger.debug("%s end" % self.token)
return
# simulation mode
if self.simulation:
_logger.debug("%s end simulation" % self.token)
return
# prepare jobs
status = self.prepareJob(newSiteID,newSiteSpec)
if status:
# run SetUpper
statusSetUp = self.runSetUpper()
if not statusSetUp:
_logger.debug("%s runSetUpper failed" % self.token)
else:
_logger.debug("%s successfully assigned to %s" % (self.token,newSiteID))
_logger.debug("%s end" % self.token)
except:
errType,errValue,errTraceBack = sys.exc_info()
_logger.error("%s run() : %s %s" % (self.token,errType,errValue))
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
site = sys.argv[1]
cloud = sys.argv[2]
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = None
jobList = []
for i in range(1):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), i)
job.AtlasRelease = 'Atlas-15.6.10'
job.homepackage = 'AtlasProduction/15.6.10.1'
job.transformation = 'Evgen_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 10000
job.prodSourceLabel = 'test'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'i686-slc5-gcc43-opt'
file = FileSpec()
file.lfn = "%s.evgen.pool.root" % job.jobName
file.destinationDBlock = job.destinationDBlock
file.destinationSE = job.destinationSE
file.dataset = job.destinationDBlock
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = None
files = {
'daq.ATLAS.0092045.physics.RPCwBeam.LB0016.SFO-2._0009.data': None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), index)
job.AtlasRelease = 'Atlas-14.4.0'
job.homepackage = 'AtlasTier0/14.4.0.2'
job.transformation = 'Reco_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.computingSite = site
job.prodDBlock = 'data08_cos.00092045.physics_RPCwBeam.daq.RAW.o4_T1224560091'
job.prodSourceLabel = 'test'
job.processingType = 'reprocessing'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc4-gcc34-opt'
origParams = """inputBSFile=daq.ATLAS.0092045.physics.RPCwBeam.LB0016.SFO-2._0009.data maxEvents=5 skipEvents=0 autoConfiguration=FieldAndGeo preInclude=RecExCommission/RecExCommission.py,RecExCommission/MinimalCommissioningSetup.py,RecJobTransforms/UseOracle.py preExec="jetFlags.Enabled.set_Value_and_Lock(False)" DBRelease=DBRelease-6.2.1.5.tar.gz conditionsTag=COMCOND-ES1C-000-00 RunNumber=92045 beamType=cosmics AMITag=r595 projectName=data08_cos trigStream=physics_RPCwBeam outputTypes=DPDCOMM outputESDFile=ESD.029868._01110.pool.root outputTAGComm=TAG_COMM.029868._01110.pool.root outputAODFile=AOD.029868._01110.pool.root outputMergedDQMonitorFile=DQM_MERGED.029868._01110.root DPD_PIXELCOMM=DPD_PIXELCOMM.029868._01110.pool.root DPD_SCTCOMM=DPD_SCTCOMM.029868._01110.pool.root DPD_IDCOMM=DPD_IDCOMM.029868._01110.pool.root DPD_IDPROJCOMM=DPD_IDPROJCOMM.029868._01110.pool.root DPD_CALOCOMM=DPD_CALOCOMM.029868._01110.pool.root DPD_TILECOMM=DPD_TILECOMM.029868._01110.pool.root DPD_EMCLUSTCOMM=DPD_EMCLUSTCOMM.029868._01110.pool.root DPD_EGAMMACOMM=DPD_EGAMMACOMM.029868._01110.pool.root DPD_RPCCOMM=DPD_RPCCOMM.029868._01110.pool.root DPD_TGCCOMM=DPD_TGCCOMM.029868._01110.pool.root --ignoreunknown"""
def run(self):
try:
self.putLog('start %s' % self.evpFileName)
# lock evp file
self.evpFile = open(self.evpFileName)
try:
fcntl.flock(self.evpFile.fileno(),fcntl.LOCK_EX|fcntl.LOCK_NB)
except:
# relase
self.putLog("cannot lock %s" % self.evpFileName)
self.evpFile.close()
return True
# options
runEvtList = []
eventPickDataType = ''
eventPickStreamName = ''
eventPickDS = []
eventPickAmiTag = ''
eventPickNumSites = 1
inputFileList = []
tagDsList = []
tagQuery = ''
tagStreamRef = ''
skipDaTRI = False
runEvtGuidMap = {}
ei_api = ''
# read evp file
for tmpLine in self.evpFile:
tmpMatch = re.search('^([^=]+)=(.+)$',tmpLine)
# check format
if tmpMatch == None:
continue
tmpItems = tmpMatch.groups()
if tmpItems[0] == 'runEvent':
# get run and event number
tmpRunEvt = tmpItems[1].split(',')
if len(tmpRunEvt) == 2:
runEvtList.append(tmpRunEvt)
elif tmpItems[0] == 'eventPickDataType':
# data type
eventPickDataType = tmpItems[1]
elif tmpItems[0] == 'eventPickStreamName':
# stream name
eventPickStreamName = tmpItems[1]
elif tmpItems[0] == 'eventPickDS':
# dataset pattern
eventPickDS = tmpItems[1].split(',')
elif tmpItems[0] == 'eventPickAmiTag':
# AMI tag
eventPickAmiTag = tmpItems[1]
elif tmpItems[0] == 'eventPickNumSites':
# the number of sites where datasets are distributed
try:
eventPickNumSites = int(tmpItems[1])
except:
pass
elif tmpItems[0] == 'userName':
# user name
self.userDN = tmpItems[1]
self.putLog("user=%s" % self.userDN)
elif tmpItems[0] == 'userTaskName':
# user task name
self.userTaskName = tmpItems[1]
elif tmpItems[0] == 'userDatasetName':
# user dataset name
self.userDatasetName = tmpItems[1]
elif tmpItems[0] == 'lockedBy':
# client name
self.lockedBy = tmpItems[1]
elif tmpItems[0] == 'creationTime':
# creation time
self.creationTime = tmpItems[1]
elif tmpItems[0] == 'params':
# parameters
self.params = tmpItems[1]
elif tmpItems[0] == 'ei_api':
# ei api parameter for MC
ei_api = tmpItems[1]
elif tmpItems[0] == 'inputFileList':
# input file list
inputFileList = tmpItems[1].split(',')
try:
inputFileList.remove('')
except:
pass
elif tmpItems[0] == 'tagDS':
# TAG dataset
tagDsList = tmpItems[1].split(',')
elif tmpItems[0] == 'tagQuery':
# query for TAG
tagQuery = tmpItems[1]
elif tmpItems[0] == 'tagStreamRef':
# StreamRef for TAG
tagStreamRef = tmpItems[1]
if not tagStreamRef.endswith('_ref'):
tagStreamRef += '_ref'
elif tmpItems[0] == 'runEvtGuidMap':
# GUIDs
try:
exec "runEvtGuidMap="+tmpItems[1]
except:
pass
# extract task name
if self.userTaskName == '' and self.params != '':
try:
tmpMatch = re.search('--outDS(=| ) *([^ ]+)',self.params)
if tmpMatch != None:
self.userTaskName = tmpMatch.group(2)
if not self.userTaskName.endswith('/'):
self.userTaskName += '/'
except:
pass
# suppress DaTRI
if self.params != '':
if '--eventPickSkipDaTRI' in self.params:
skipDaTRI = True
# get compact user name
compactDN = self.taskBuffer.cleanUserID(self.userDN)
# get jediTaskID
self.jediTaskID = self.taskBuffer.getTaskIDwithTaskNameJEDI(compactDN,self.userTaskName)
# convert
if tagDsList == [] or tagQuery == '':
# convert run/event list to dataset/file list
tmpRet,locationMap,allFiles = self.pd2p.convertEvtRunToDatasets(runEvtList,
eventPickDataType,
eventPickStreamName,
eventPickDS,
eventPickAmiTag,
self.userDN,
runEvtGuidMap,
ei_api
)
if not tmpRet:
if 'isFatal' in locationMap and locationMap['isFatal'] == True:
self.ignoreError = False
self.endWithError('Failed to convert the run/event list to a dataset/file list')
return False
else:
# get parent dataset/files with TAG
tmpRet,locationMap,allFiles = self.pd2p.getTagParentInfoUsingTagQuery(tagDsList,tagQuery,tagStreamRef)
if not tmpRet:
self.endWithError('Failed to get parent dataset/file list with TAG')
return False
# use only files in the list
if inputFileList != []:
tmpAllFiles = []
for tmpFile in allFiles:
if tmpFile['lfn'] in inputFileList:
tmpAllFiles.append(tmpFile)
allFiles = tmpAllFiles
# remove redundant CN from DN
tmpDN = self.userDN
tmpDN = re.sub('/CN=limited proxy','',tmpDN)
tmpDN = re.sub('(/CN=proxy)+$','',tmpDN)
# make dataset container
tmpRet = self.pd2p.registerDatasetContainerWithDatasets(self.userDatasetName,allFiles,
locationMap,
nSites=eventPickNumSites,
owner=tmpDN)
if not tmpRet:
self.endWithError('Failed to make a dataset container %s' % self.userDatasetName)
return False
# skip DaTRI
if skipDaTRI:
# successfully terminated
self.putLog("skip DaTRI")
# update task
self.taskBuffer.updateTaskModTimeJEDI(self.jediTaskID)
else:
# get candidates
tmpRet,candidateMaps = self.pd2p.getCandidates(self.userDatasetName,checkUsedFile=False,
useHidden=True)
if not tmpRet:
self.endWithError('Failed to find candidate for destination')
return False
# collect all candidates
allCandidates = []
for tmpDS,tmpDsVal in candidateMaps.iteritems():
for tmpCloud,tmpCloudVal in tmpDsVal.iteritems():
for tmpSiteName in tmpCloudVal[0]:
if not tmpSiteName in allCandidates:
allCandidates.append(tmpSiteName)
if allCandidates == []:
self.endWithError('No candidate for destination')
return False
# get list of dataset (container) names
if eventPickNumSites > 1:
# decompose container to transfer datasets separately
tmpRet,tmpOut = self.pd2p.getListDatasetReplicasInContainer(self.userDatasetName)
if not tmpRet:
self.endWithError('Failed to get the size of %s' % self.userDatasetName)
return False
userDatasetNameList = tmpOut.keys()
else:
# transfer container at once
userDatasetNameList = [self.userDatasetName]
# loop over all datasets
sitesUsed = []
for tmpUserDatasetName in userDatasetNameList:
# get size of dataset container
tmpRet,totalInputSize = rucioAPI.getDatasetSize(tmpUserDatasetName)
if not tmpRet:
self.endWithError('Failed to get the size of %s' % tmpUserDatasetName)
return False
# run brokerage
tmpJob = JobSpec()
tmpJob.AtlasRelease = ''
self.putLog("run brokerage for %s" % tmpDS)
brokerage.broker.schedule([tmpJob],self.taskBuffer,self.siteMapper,True,allCandidates,
True,datasetSize=totalInputSize)
if tmpJob.computingSite.startswith('ERROR'):
self.endWithError('brokerage failed with %s' % tmpJob.computingSite)
return False
self.putLog("site -> %s" % tmpJob.computingSite)
# send transfer request
try:
tmpDN = rucioAPI.parse_dn(tmpDN)
tmpStatus,userInfo = rucioAPI.finger(tmpDN)
if not tmpStatus:
raise RuntimeError,'user info not found for {0} with {1}'.format(tmpDN,userInfo)
tmpDN = userInfo['nickname']
tmpDQ2ID = self.siteMapper.getSite(tmpJob.computingSite).ddm_input
tmpMsg = "%s ds=%s site=%s id=%s" % ('registerDatasetLocation for DaTRI ',
tmpUserDatasetName,
tmpDQ2ID,
tmpDN)
self.putLog(tmpMsg)
rucioAPI.registerDatasetLocation(tmpDS,[tmpDQ2ID],lifetime=14,owner=tmpDN,
activity="User Subscriptions")
self.putLog('OK')
except:
errType,errValue = sys.exc_info()[:2]
tmpStr = 'Failed to send transfer request : %s %s' % (errType,errValue)
tmpStr.strip()
tmpStr += traceback.format_exc()
self.endWithError(tmpStr)
return False
# list of sites already used
sitesUsed.append(tmpJob.computingSite)
self.putLog("used %s sites" % len(sitesUsed))
# set candidates
if len(sitesUsed) >= eventPickNumSites:
# reset candidates to limit the number of sites
allCandidates = sitesUsed
sitesUsed = []
else:
# remove site
allCandidates.remove(tmpJob.computingSite)
# send email notification for success
tmpMsg = 'A transfer request was successfully sent to Rucio.\n'
tmpMsg += 'Your task will get started once transfer is completed.'
self.sendEmail(True,tmpMsg)
try:
# unlock and delete evp file
fcntl.flock(self.evpFile.fileno(),fcntl.LOCK_UN)
self.evpFile.close()
os.remove(self.evpFileName)
except:
pass
# successfully terminated
self.putLog("end %s" % self.evpFileName)
return True
except:
errType,errValue = sys.exc_info()[:2]
self.endWithError('Got exception %s:%s %s' % (errType,errValue,traceback.format_exc()))
return False
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = None
files = {
'daq.ATLAS.0092045.physics.RPCwBeam.LB0016.SFO-2._0009.data':None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),index)
job.AtlasRelease = 'Atlas-14.4.0'
job.homepackage = 'AtlasTier0/14.4.0.2'
job.transformation = 'Reco_trf.py'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.computingSite = site
job.prodDBlock = 'data08_cos.00092045.physics_RPCwBeam.daq.RAW.o4_T1224560091'
job.prodSourceLabel = 'test'
job.processingType = 'reprocessing'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc4-gcc34-opt'
origParams = """inputBSFile=daq.ATLAS.0092045.physics.RPCwBeam.LB0016.SFO-2._0009.data maxEvents=5 skipEvents=0 autoConfiguration=FieldAndGeo preInclude=RecExCommission/RecExCommission.py,RecExCommission/MinimalCommissioningSetup.py,RecJobTransforms/UseOracle.py preExec="jetFlags.Enabled.set_Value_and_Lock(False)" DBRelease=DBRelease-6.2.1.5.tar.gz conditionsTag=COMCOND-ES1C-000-00 RunNumber=92045 beamType=cosmics AMITag=r595 projectName=data08_cos trigStream=physics_RPCwBeam outputTypes=DPDCOMM outputESDFile=ESD.029868._01110.pool.root outputTAGComm=TAG_COMM.029868._01110.pool.root outputAODFile=AOD.029868._01110.pool.root outputMergedDQMonitorFile=DQM_MERGED.029868._01110.root DPD_PIXELCOMM=DPD_PIXELCOMM.029868._01110.pool.root DPD_SCTCOMM=DPD_SCTCOMM.029868._01110.pool.root DPD_IDCOMM=DPD_IDCOMM.029868._01110.pool.root DPD_IDPROJCOMM=DPD_IDPROJCOMM.029868._01110.pool.root DPD_CALOCOMM=DPD_CALOCOMM.029868._01110.pool.root DPD_TILECOMM=DPD_TILECOMM.029868._01110.pool.root DPD_EMCLUSTCOMM=DPD_EMCLUSTCOMM.029868._01110.pool.root DPD_EGAMMACOMM=DPD_EGAMMACOMM.029868._01110.pool.root DPD_RPCCOMM=DPD_RPCCOMM.029868._01110.pool.root DPD_TGCCOMM=DPD_TGCCOMM.029868._01110.pool.root --ignoreunknown"""
int(taskID))
# instantiate JobSpecs
iJob = 0
jobList = []
for line in taskFile:
iJob += 1
job = JobSpec()
# job ID ###### FIXME
job.jobDefinitionID = int(time.time()) % 10000
# job name
job.jobName = "%s_%05d.job" % (taskName, iJob)
# AtlasRelease
if len(re.findall('\.', trfVer)) > 2:
match = re.search('^(\d+\.\d+\.\d+)', trfVer)
job.AtlasRelease = 'Atlas-%s' % match.group(1)
else:
job.AtlasRelease = 'Atlas-%s' % trfVer
# homepackage
vers = trfVer.split('.')
if int(vers[0]) <= 11:
job.homepackage = 'JobTransforms'
for ver in vers:
job.homepackage += "-%02d" % int(ver)
else:
job.homepackage = 'AtlasProduction/%s' % trfVer
# trf
job.transformation = trf
job.destinationDBlock = oDatasets[0]
# prod DBlock
job.prodDBlock = iDataset
import time
import commands
import userinterface.Client as Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('/usr/bin/uuidgen')
job.AtlasRelease = 'Atlas-9.0.4'
job.prodDBlock = 'pandatest.000003.dd.input'
job.destinationDBlock = 'panda.destDB.%s' % commands.getoutput(
'/usr/bin/uuidgen')
job.destinationSE = 'BNL_SE'
ids = {
'pandatest.000003.dd.input._00028.junk':
'6c19e1fc-ee8c-4bae-bd4c-c9e5c73aca27',
'pandatest.000003.dd.input._00033.junk':
'98f79ba1-1793-4253-aac7-bdf90a51d1ee',
'pandatest.000003.dd.input._00039.junk':
'33660dd5-7cef-422a-a7fc-6c24cb10deb1'
}
for lfn in ids.keys():
file = FileSpec()
file.lfn = lfn
file.GUID = ids[file.lfn]
file.dataset = 'pandatest.000003.dd.input'
file.type = 'input'
job.addFile(file)
if len(sys.argv)>1:
site = sys.argv[1]
else:
site = None
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'BNL_ATLAS_2'
jobList = []
for i in range(20):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-11.0.41'
#job.AtlasRelease = 'Atlas-11.0.3'
job.homepackage = 'AnalysisTransforms'
job.transformation = 'https://gridui01.usatlas.bnl.gov:24443/dav/test/runAthena'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 100
job.prodSourceLabel = 'user'
job.computingSite = site
#job.prodDBlock = "pandatest.b1599dfa-cd36-4fc5-92f6-495781a94c66"
job.prodDBlock = "pandatest.f228b051-077b-4f81-90bf-496340644379"
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
fileI.lfn = "lib.f228b051-077b-4f81-90bf-496340644379.tgz"
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
files = {
inputFile: None,
}
jobList = []
index = 0
for lfn in files.keys():
index += 1
job = JobSpec()
job.jobDefinitionID = (time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'), index)
job.AtlasRelease = 'Atlas-16.6.2'
job.homepackage = 'AtlasProduction/16.6.2.1'
job.transformation = 'AtlasG4_trf.py'
job.destinationDBlock = datasetName
job.computingSite = site
job.prodDBlock = prodDBlock
job.prodSourceLabel = 'test'
job.processingType = 'test'
job.currentPriority = 10000
job.cloud = cloud
job.cmtConfig = 'i686-slc5-gcc43-opt'
fileI = FileSpec()
fileI.dataset = job.prodDBlock
fileI.prodDBlock = job.prodDBlock
from taskbuffer.FileSpec import FileSpec
if len(sys.argv) > 1:
site = sys.argv[1]
else:
site = None
jobList = []
for i in range(2):
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'ANALY_BNL_ATLAS_1'
job = JobSpec()
job.jobDefinitionID = 1
job.jobName = commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-12.0.2'
job.homepackage = 'AnalysisTransforms'
job.transformation = 'https://gridui01.usatlas.bnl.gov:24443/dav/test/runAthena2'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 3000
job.prodSourceLabel = 'user'
job.computingSite = site
job.prodDBlock = 'testIdeal_06.005001.pythia_minbias.recon.AOD.v12000103'
fileOL = FileSpec()
fileOL.lfn = "%s.job.log.tgz" % commands.getoutput('uuidgen')
fileOL.destinationDBlock = job.destinationDBlock
fileOL.destinationSE = job.destinationSE
fileOL.dataset = job.destinationDBlock
fileOL.type = 'log'
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""Prepare the specific aspec of each subjob.
Returns: subjobconfig list of objects understood by backends."""
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
from GangaAtlas.Lib.ATLASDataset.DQ2Dataset import dq2_set_dataset_lifetime
from GangaPanda.Lib.Panda.Panda import refreshPandaSpecs
# make sure we have the correct siteType
refreshPandaSpecs()
job = app._getParent()
masterjob = job._getRoot()
logger.debug('ProdTransPandaRTHandler prepare called for %s',
job.getFQID('.'))
job.backend.actualCE = job.backend.site
job.backend.requirements.cloud = Client.PandaSites[
job.backend.site]['cloud']
# check that the site is in a submit-able status
if not job.splitter or job.splitter._name != 'DQ2JobSplitter':
allowed_sites = job.backend.list_ddm_sites()
try:
outDsLocation = Client.PandaSites[job.backend.site]['ddm']
tmpDsExist = False
if (configPanda['processingType'].startswith('gangarobot') or
configPanda['processingType'].startswith('hammercloud')):
#if Client.getDatasets(job.outputdata.datasetname):
if getDatasets(job.outputdata.datasetname):
tmpDsExist = True
logger.info('Re-using output dataset %s' %
job.outputdata.datasetname)
if not configPanda[
'specialHandling'] == 'ddm:rucio' and not configPanda[
'processingType'].startswith(
'gangarobot'
) and not configPanda['processingType'].startswith(
'hammercloud') and not configPanda[
'processingType'].startswith('rucio_test'):
Client.addDataset(job.outputdata.datasetname,
False,
location=outDsLocation,
allowProdDisk=True,
dsExist=tmpDsExist)
logger.info('Output dataset %s registered at %s' %
(job.outputdata.datasetname, outDsLocation))
dq2_set_dataset_lifetime(job.outputdata.datasetname, outDsLocation)
except exceptions.SystemExit:
raise BackendError(
'Panda', 'Exception in adding dataset %s: %s %s' %
(job.outputdata.datasetname, sys.exc_info()[0],
sys.exc_info()[1]))
# JobSpec.
jspec = JobSpec()
jspec.currentPriority = app.priority
jspec.jobDefinitionID = masterjob.id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.coreCount = app.core_count
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_release
jspec.homepackage = app.home_package
jspec.transformation = app.transformation
jspec.destinationDBlock = job.outputdata.datasetname
if job.outputdata.location:
jspec.destinationSE = job.outputdata.location
else:
jspec.destinationSE = job.backend.site
if job.inputdata:
jspec.prodDBlock = job.inputdata.dataset[0]
else:
jspec.prodDBlock = 'NULL'
if app.prod_source_label:
jspec.prodSourceLabel = app.prod_source_label
else:
jspec.prodSourceLabel = configPanda['prodSourceLabelRun']
jspec.processingType = configPanda['processingType']
jspec.specialHandling = configPanda['specialHandling']
jspec.computingSite = job.backend.site
jspec.cloud = job.backend.requirements.cloud
jspec.cmtConfig = app.atlas_cmtconfig
if app.dbrelease == 'LATEST':
try:
latest_dbrelease = getLatestDBReleaseCaching()
except:
from pandatools import Client
latest_dbrelease = Client.getLatestDBRelease()
m = re.search('(.*):DBRelease-(.*)\.tar\.gz', latest_dbrelease)
if m:
self.dbrelease_dataset = m.group(1)
self.dbrelease = m.group(2)
else:
raise ApplicationConfigurationError(
None,
"Error retrieving LATEST DBRelease. Try setting application.dbrelease manually."
)
else:
self.dbrelease_dataset = app.dbrelease_dataset
self.dbrelease = app.dbrelease
jspec.jobParameters = app.job_parameters
if self.dbrelease:
if self.dbrelease == 'current':
jspec.jobParameters += ' --DBRelease=current'
else:
if jspec.transformation.endswith(
"_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --DBRelease=DBRelease-%s.tar.gz' % (
self.dbrelease, )
else:
jspec.jobParameters += ' DBRelease=DBRelease-%s.tar.gz' % (
self.dbrelease, )
dbspec = FileSpec()
dbspec.lfn = 'DBRelease-%s.tar.gz' % self.dbrelease
dbspec.dataset = self.dbrelease_dataset
dbspec.prodDBlock = jspec.prodDBlock
dbspec.type = 'input'
jspec.addFile(dbspec)
if job.inputdata:
m = re.search('(.*)\.(.*)\.(.*)\.(.*)\.(.*)\.(.*)',
job.inputdata.dataset[0])
if not m:
logger.error("Error retrieving run number from dataset name")
#raise ApplicationConfigurationError(None, "Error retrieving run number from dataset name")
runnumber = 105200
else:
runnumber = int(m.group(2))
if jspec.transformation.endswith(
"_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --runNumber %d' % runnumber
else:
jspec.jobParameters += ' RunNumber=%d' % runnumber
# Output files.
randomized_lfns = []
ilfn = 0
for lfn, lfntype in zip(app.output_files, app.output_type):
ofspec = FileSpec()
if app.randomize_lfns:
randomized_lfn = lfn + (
'.%s.%d.%s' %
(job.backend.site, int(time.time()),
commands.getoutput('uuidgen 2> /dev/null')[:4]))
else:
randomized_lfn = lfn
ofspec.lfn = randomized_lfn
randomized_lfns.append(randomized_lfn)
ofspec.destinationDBlock = jspec.destinationDBlock
ofspec.destinationSE = jspec.destinationSE
ofspec.dataset = jspec.destinationDBlock
ofspec.type = 'output'
jspec.addFile(ofspec)
if jspec.transformation.endswith(
"_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --output%sFile %s' % (
lfntype, randomized_lfns[ilfn])
else:
jspec.jobParameters += ' output%sFile=%s' % (
lfntype, randomized_lfns[ilfn])
ilfn = ilfn + 1
# Input files.
if job.inputdata:
for guid, lfn, size, checksum, scope in zip(
job.inputdata.guids, job.inputdata.names,
job.inputdata.sizes, job.inputdata.checksums,
job.inputdata.scopes):
ifspec = FileSpec()
ifspec.lfn = lfn
ifspec.GUID = guid
ifspec.fsize = size
ifspec.md5sum = checksum
ifspec.scope = scope
ifspec.dataset = jspec.prodDBlock
ifspec.prodDBlock = jspec.prodDBlock
ifspec.type = 'input'
jspec.addFile(ifspec)
if app.input_type:
itype = app.input_type
else:
itype = m.group(5)
if jspec.transformation.endswith(
"_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --input%sFile %s' % (itype, ','.join(
job.inputdata.names))
else:
jspec.jobParameters += ' input%sFile=%s' % (itype, ','.join(
job.inputdata.names))
# Log files.
lfspec = FileSpec()
lfspec.lfn = '%s.job.log.tgz' % jspec.jobName
lfspec.destinationDBlock = jspec.destinationDBlock
lfspec.destinationSE = jspec.destinationSE
lfspec.dataset = jspec.destinationDBlock
lfspec.type = 'log'
jspec.addFile(lfspec)
return jspec
def prepare(self,app,appconfig,appmasterconfig,jobmasterconfig):
'''prepare the subjob specific configuration'''
# PandaTools
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaMCPandaRTHandler prepare called for %s', job.getFQID('.'))
try:
assert self.outsite
except:
logger.error("outsite not set. Aborting")
raise Exception()
job.backend.site = self.outsite
job.backend.actualCE = self.outsite
cloud = job._getRoot().backend.requirements.cloud
job.backend.requirements.cloud = cloud
# now just filling the job from AthenaMC data
jspec = JobSpec()
jspec.jobDefinitionID = job._getRoot().id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_rel
if app.transform_archive:
jspec.homepackage = 'AnalysisTransforms'+app.transform_archive
elif app.prod_release:
jspec.homepackage = 'AnalysisTransforms-AtlasProduction_'+str(app.prod_release)
jspec.transformation = '%s/runAthena-00-00-11' % Client.baseURLSUB
#---->???? prodDBlock and destinationDBlock when facing several input / output datasets?
jspec.prodDBlock = 'NULL'
if job.inputdata and len(app.inputfiles)>0 and app.inputfiles[0] in app.dsetmap:
jspec.prodDBlock = app.dsetmap[app.inputfiles[0]]
# How to specify jspec.destinationDBlock when more than one type of output is available? Panda prod jobs seem to specify only the last output dataset
outdset=""
for type in ["EVNT","RDO","HITS","AOD","ESD","NTUP"]:
if type in app.outputpaths.keys():
outdset=string.replace(app.outputpaths[type],"/",".")
outdset=outdset[1:-1]
break
if not outdset:
try:
assert len(app.outputpaths.keys())>0
except:
logger.error("app.outputpaths is empty: check your output datasets")
raise
type=app.outputpaths.keys()[0]
outdset=string.replace(app.outputpaths[type],"/",".")
outdset=outdset[1:-1]
jspec.destinationDBlock = outdset
jspec.destinationSE = self.outsite
jspec.prodSourceLabel = 'user'
jspec.assignedPriority = 1000
jspec.cloud = cloud
# memory
if job.backend.requirements.memory != -1:
jspec.minRamCount = job.backend.requirements.memory
jspec.computingSite = self.outsite
jspec.cmtConfig = AthenaUtils.getCmtConfig(athenaVer=app.atlas_rel)
# library (source files)
flib = FileSpec()
flib.lfn = self.library
# flib.GUID =
flib.type = 'input'
# flib.status =
flib.dataset = self.libDataset
flib.dispatchDBlock = self.libDataset
jspec.addFile(flib)
# input files FIXME: many more input types
for lfn in app.inputfiles:
useguid=app.turls[lfn].replace("guid:","")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# add dbfiles if any:
for lfn in app.dbfiles:
useguid=app.dbturls[lfn].replace("guid:","")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# then minbias files
for lfn in app.mbfiles:
useguid=app.minbias_turls[lfn].replace("guid:","")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# then cavern files
for lfn in app.cavernfiles:
useguid=app.cavern_turls[lfn].replace("guid:","")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# output files( this includes the logfiles)
# Output files
jidtag=""
job = app._getParent() # Returns job or subjob object
if job._getRoot().subjobs:
jidtag = job._getRoot().id
else:
jidtag = "%d" % job.id
outfiles=app.subjobsOutfiles[job.id]
pandaOutfiles={}
for type in outfiles.keys():
pandaOutfiles[type]=outfiles[type]+"."+str(jidtag)
if type=="LOG":
pandaOutfiles[type]+=".tgz"
#print pandaOutfiles
for outtype in pandaOutfiles.keys():
fout = FileSpec()
dset=string.replace(app.outputpaths[outtype],"/",".")
dset=dset[1:-1]
fout.dataset=dset
fout.lfn=pandaOutfiles[outtype]
fout.type = 'output'
# fout.destinationDBlock = jspec.destinationDBlock
fout.destinationDBlock = fout.dataset
fout.destinationSE = jspec.destinationSE
if outtype=='LOG':
fout.type='log'
fout.destinationDBlock = fout.dataset
fout.destinationSE = job.backend.site
jspec.addFile(fout)
# job parameters
param = '-l %s ' % self.library # user tarball.
# use corruption checker
if job.backend.requirements.corCheck:
param += '--corCheck '
# disable to skip missing files
if job.backend.requirements.notSkipMissing:
param += '--notSkipMissing '
# transform parameters
# need to update arglist with final output file name...
newArgs=[]
if app.mode == "evgen":
app.args[3]=app.args[3]+" -t "
if app.verbosity:
app.args[3]=app.args[3]+" -l %s " % app.verbosity
for arg in app.args[3:]:
for type in outfiles.keys():
if arg.find(outfiles[type])>-1:
arg=arg.replace(outfiles[type],pandaOutfiles[type])
newArgs.append(arg)
arglist=string.join(newArgs," ")
# print "Arglist:",arglist
param += ' -r ./ '
param += ' -j "%s"' % urllib.quote(arglist)
allinfiles=app.inputfiles+app.dbfiles
# Input files.
param += ' -i "%s" ' % allinfiles
if len(app.mbfiles)>0:
param+= ' -m "%s" ' % app.mbfiles
if len(app.cavernfiles)>0:
param+= ' -n "%s" ' % app.cavernfiles
# param += '-m "[]" ' #%minList FIXME
# param += '-n "[]" ' #%cavList FIXME
del pandaOutfiles["LOG"] # logfiles do not appear in IROOT block, and this one is not needed anymore...
param += ' -o "{\'IROOT\':%s }"' % str(pandaOutfiles.items())
# source URL
matchURL = re.search("(http.*://[^/]+)/",Client.baseURLSSL)
if matchURL != None:
param += " --sourceURL %s " % matchURL.group(1)
param += " --trf"
jspec.jobParameters = param
jspec.metadata="--trf \"%s\"" % arglist
#print "SUBJOB DETAILS:",jspec.values()
if app.dryrun:
print "job.application.dryrun activated, printing out job parameters"
print jspec.values()
return
return jspec
if len(sys.argv) > 1:
site = sys.argv[1]
else:
site = None
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'BNL_SE'
jobListE = []
lfnListE = []
for i in range(2):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('uuidgen')
job.AtlasRelease = 'Atlas-11.0.3'
job.homepackage = 'JobTransforms-11-00-03-03'
job.transformation = 'share/csc.evgen.trf'
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 1000
job.prodSourceLabel = 'test'
job.computingSite = site
file = FileSpec()
file.lfn = "%s.evgen.pool.root" % commands.getoutput('uuidgen')
lfnListE.append(file.lfn)
file.lfn += ('.%d' % (i + 1))
file.destinationDBlock = job.destinationDBlock
file.destinationSE = job.destinationSE
file.dataset = job.destinationDBlock
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""Prepare the specific aspec of each subjob.
Returns: subjobconfig list of objects understood by backends."""
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
from GangaAtlas.Lib.ATLASDataset.DQ2Dataset import dq2_set_dataset_lifetime
from GangaPanda.Lib.Panda.Panda import refreshPandaSpecs
# make sure we have the correct siteType
refreshPandaSpecs()
job = app._getParent()
masterjob = job._getRoot()
logger.debug('ProdTransPandaRTHandler prepare called for %s',
job.getFQID('.'))
job.backend.actualCE = job.backend.site
job.backend.requirements.cloud = Client.PandaSites[job.backend.site]['cloud']
# check that the site is in a submit-able status
if not job.splitter or job.splitter._name != 'DQ2JobSplitter':
allowed_sites = job.backend.list_ddm_sites()
try:
outDsLocation = Client.PandaSites[job.backend.site]['ddm']
tmpDsExist = False
if (configPanda['processingType'].startswith('gangarobot') or configPanda['processingType'].startswith('hammercloud')):
#if Client.getDatasets(job.outputdata.datasetname):
if getDatasets(job.outputdata.datasetname):
tmpDsExist = True
logger.info('Re-using output dataset %s'%job.outputdata.datasetname)
if not configPanda['specialHandling']=='ddm:rucio' and not configPanda['processingType'].startswith('gangarobot') and not configPanda['processingType'].startswith('hammercloud') and not configPanda['processingType'].startswith('rucio_test'):
Client.addDataset(job.outputdata.datasetname,False,location=outDsLocation,allowProdDisk=True,dsExist=tmpDsExist)
logger.info('Output dataset %s registered at %s'%(job.outputdata.datasetname,outDsLocation))
dq2_set_dataset_lifetime(job.outputdata.datasetname, outDsLocation)
except exceptions.SystemExit:
raise BackendError('Panda','Exception in adding dataset %s: %s %s'%(job.outputdata.datasetname,sys.exc_info()[0],sys.exc_info()[1]))
# JobSpec.
jspec = JobSpec()
jspec.currentPriority = app.priority
jspec.jobDefinitionID = masterjob.id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.coreCount = app.core_count
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_release
jspec.homepackage = app.home_package
jspec.transformation = app.transformation
jspec.destinationDBlock = job.outputdata.datasetname
if job.outputdata.location:
jspec.destinationSE = job.outputdata.location
else:
jspec.destinationSE = job.backend.site
if job.inputdata:
jspec.prodDBlock = job.inputdata.dataset[0]
else:
jspec.prodDBlock = 'NULL'
if app.prod_source_label:
jspec.prodSourceLabel = app.prod_source_label
else:
jspec.prodSourceLabel = configPanda['prodSourceLabelRun']
jspec.processingType = configPanda['processingType']
jspec.specialHandling = configPanda['specialHandling']
jspec.computingSite = job.backend.site
jspec.cloud = job.backend.requirements.cloud
jspec.cmtConfig = app.atlas_cmtconfig
if app.dbrelease == 'LATEST':
try:
latest_dbrelease = getLatestDBReleaseCaching()
except:
from pandatools import Client
latest_dbrelease = Client.getLatestDBRelease()
m = re.search('(.*):DBRelease-(.*)\.tar\.gz', latest_dbrelease)
if m:
self.dbrelease_dataset = m.group(1)
self.dbrelease = m.group(2)
else:
raise ApplicationConfigurationError(None, "Error retrieving LATEST DBRelease. Try setting application.dbrelease manually.")
else:
self.dbrelease_dataset = app.dbrelease_dataset
self.dbrelease = app.dbrelease
jspec.jobParameters = app.job_parameters
if self.dbrelease:
if self.dbrelease == 'current':
jspec.jobParameters += ' --DBRelease=current'
else:
if jspec.transformation.endswith("_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --DBRelease=DBRelease-%s.tar.gz' % (self.dbrelease,)
else:
jspec.jobParameters += ' DBRelease=DBRelease-%s.tar.gz' % (self.dbrelease,)
dbspec = FileSpec()
dbspec.lfn = 'DBRelease-%s.tar.gz' % self.dbrelease
dbspec.dataset = self.dbrelease_dataset
dbspec.prodDBlock = jspec.prodDBlock
dbspec.type = 'input'
jspec.addFile(dbspec)
if job.inputdata:
m = re.search('(.*)\.(.*)\.(.*)\.(.*)\.(.*)\.(.*)',
job.inputdata.dataset[0])
if not m:
logger.error("Error retrieving run number from dataset name")
#raise ApplicationConfigurationError(None, "Error retrieving run number from dataset name")
runnumber = 105200
else:
runnumber = int(m.group(2))
if jspec.transformation.endswith("_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --runNumber %d' % runnumber
else:
jspec.jobParameters += ' RunNumber=%d' % runnumber
# Output files.
randomized_lfns = []
ilfn = 0
for lfn, lfntype in zip(app.output_files,app.output_type):
ofspec = FileSpec()
if app.randomize_lfns:
randomized_lfn = lfn + ('.%s.%d.%s' % (job.backend.site, int(time.time()), commands.getoutput('uuidgen 2> /dev/null')[:4] ) )
else:
randomized_lfn = lfn
ofspec.lfn = randomized_lfn
randomized_lfns.append(randomized_lfn)
ofspec.destinationDBlock = jspec.destinationDBlock
ofspec.destinationSE = jspec.destinationSE
ofspec.dataset = jspec.destinationDBlock
ofspec.type = 'output'
jspec.addFile(ofspec)
if jspec.transformation.endswith("_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --output%sFile %s' % (lfntype, randomized_lfns[ilfn])
else:
jspec.jobParameters += ' output%sFile=%s' % (lfntype, randomized_lfns[ilfn])
ilfn=ilfn+1
# Input files.
if job.inputdata:
for guid, lfn, size, checksum, scope in zip(job.inputdata.guids, job.inputdata.names, job.inputdata.sizes, job.inputdata.checksums, job.inputdata.scopes):
ifspec = FileSpec()
ifspec.lfn = lfn
ifspec.GUID = guid
ifspec.fsize = size
ifspec.md5sum = checksum
ifspec.scope = scope
ifspec.dataset = jspec.prodDBlock
ifspec.prodDBlock = jspec.prodDBlock
ifspec.type = 'input'
jspec.addFile(ifspec)
if app.input_type:
itype = app.input_type
else:
itype = m.group(5)
if jspec.transformation.endswith("_tf.py") or jspec.transformation.endswith("_tf"):
jspec.jobParameters += ' --input%sFile %s' % (itype, ','.join(job.inputdata.names))
else:
jspec.jobParameters += ' input%sFile=%s' % (itype, ','.join(job.inputdata.names))
# Log files.
lfspec = FileSpec()
lfspec.lfn = '%s.job.log.tgz' % jspec.jobName
lfspec.destinationDBlock = jspec.destinationDBlock
lfspec.destinationSE = jspec.destinationSE
lfspec.dataset = jspec.destinationDBlock
lfspec.type = 'log'
jspec.addFile(lfspec)
return jspec
if len(sys.argv)>1:
site = sys.argv[1]
else:
site = None
datasetName = 'panda.destDB.%s' % commands.getoutput('uuidgen')
#destName = 'BNL_SE'
jobList = []
for i in [999905,999906,999907]:
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),i)
job.AtlasRelease = 'Atlas-14.1.0'
job.homepackage = 'AtlasProduction/12.0.6.2'
job.transformation = 'csc_evgen_trf.py'
job.destinationDBlock = datasetName
#job.destinationSE = destName
job.currentPriority = 1000
job.prodSourceLabel = 'managed'
#job.prodSourceLabel = 'test'
#job.computingSite = site
job.cmtConfig = 'i686-slc4-gcc34-opt'
job.metadata = 'evgen;%s;%s;%s' % (str({'FR': 46, 'NL': 45, 'NDGF': 300, 'CERN': 19, 'TW': 44110, 'CA': 2922, 'DE': 9903, 'IT': 1168, 'US': 6226, 'UK': 1026, 'ES': 26619}),str({999907:100,999906:200,999905:300}),str({999905:100,999906:910,999907:500}))
#job.metadata = 'evgen;%s' % str({'FR': 46, 'NL': 45, 'NDGF': 300, 'CERN': 19, 'TW': 44110, 'CA': 2922, 'DE': 9903, 'IT': 1168, 'US': 6226, 'UK': 1026, 'ES': 26619})
#job.cloud = "UK"
job.taskID = i
else:
site = None
cloud = None
datasetName = 'panda.destDB.%s_tid999991' % commands.getoutput('uuidgen')
taskid = 999989
jobList = []
for i in range(1):
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = "%s_%d" % (commands.getoutput('uuidgen'),i)
# job.AtlasRelease = 'Atlas-12.0.6'
# job.homepackage = 'AtlasProduction/12.0.6.5'
job.AtlasRelease = 'Atlas-12.0.7'
job.homepackage = 'AtlasProduction/12.0.7.1'
job.transformation = 'csc_evgen_trf.py'
job.destinationDBlock = datasetName
# job.destinationSE = destName
# job.cloud = 'CA'
job.cloud = cloud
job.taskID = taskid
job.currentPriority = 1000
job.prodSourceLabel = 'test'
# job.prodSourceLabel = 'cloudtest'
job.computingSite = site
file = FileSpec()
file.lfn = "%s.evgen.pool.root" % job.jobName
def prepare(self, app, appconfig, appmasterconfig, jobmasterconfig):
'''prepare the subjob specific configuration'''
# PandaTools
from pandatools import Client
from pandatools import AthenaUtils
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaMCPandaRTHandler prepare called for %s',
job.getFQID('.'))
try:
assert self.outsite
except:
logger.error("outsite not set. Aborting")
raise Exception()
job.backend.site = self.outsite
job.backend.actualCE = self.outsite
cloud = job._getRoot().backend.requirements.cloud
job.backend.requirements.cloud = cloud
# now just filling the job from AthenaMC data
jspec = JobSpec()
jspec.jobDefinitionID = job._getRoot().id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.AtlasRelease = 'Atlas-%s' % app.atlas_rel
if app.transform_archive:
jspec.homepackage = 'AnalysisTransforms' + app.transform_archive
elif app.prod_release:
jspec.homepackage = 'AnalysisTransforms-AtlasProduction_' + str(
app.prod_release)
jspec.transformation = '%s/runAthena-00-00-11' % Client.baseURLSUB
#---->???? prodDBlock and destinationDBlock when facing several input / output datasets?
jspec.prodDBlock = 'NULL'
if job.inputdata and len(
app.inputfiles) > 0 and app.inputfiles[0] in app.dsetmap:
jspec.prodDBlock = app.dsetmap[app.inputfiles[0]]
# How to specify jspec.destinationDBlock when more than one type of output is available? Panda prod jobs seem to specify only the last output dataset
outdset = ""
for type in ["EVNT", "RDO", "HITS", "AOD", "ESD", "NTUP"]:
if type in app.outputpaths.keys():
outdset = string.replace(app.outputpaths[type], "/", ".")
outdset = outdset[1:-1]
break
if not outdset:
try:
assert len(app.outputpaths.keys()) > 0
except:
logger.error(
"app.outputpaths is empty: check your output datasets")
raise
type = app.outputpaths.keys()[0]
outdset = string.replace(app.outputpaths[type], "/", ".")
outdset = outdset[1:-1]
jspec.destinationDBlock = outdset
jspec.destinationSE = self.outsite
jspec.prodSourceLabel = 'user'
jspec.assignedPriority = 1000
jspec.cloud = cloud
# memory
if job.backend.requirements.memory != -1:
jspec.minRamCount = job.backend.requirements.memory
jspec.computingSite = self.outsite
jspec.cmtConfig = AthenaUtils.getCmtConfig(athenaVer=app.atlas_rel)
# library (source files)
flib = FileSpec()
flib.lfn = self.library
# flib.GUID =
flib.type = 'input'
# flib.status =
flib.dataset = self.libDataset
flib.dispatchDBlock = self.libDataset
jspec.addFile(flib)
# input files FIXME: many more input types
for lfn in app.inputfiles:
useguid = app.turls[lfn].replace("guid:", "")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# add dbfiles if any:
for lfn in app.dbfiles:
useguid = app.dbturls[lfn].replace("guid:", "")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# then minbias files
for lfn in app.mbfiles:
useguid = app.minbias_turls[lfn].replace("guid:", "")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# then cavern files
for lfn in app.cavernfiles:
useguid = app.cavern_turls[lfn].replace("guid:", "")
finp = FileSpec()
finp.lfn = lfn
finp.GUID = useguid
finp.dataset = app.dsetmap[lfn]
finp.prodDBlock = app.dsetmap[lfn]
finp.prodDBlockToken = 'local'
finp.dispatchDBlock = app.dsetmap[lfn]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# output files( this includes the logfiles)
# Output files
jidtag = ""
job = app._getParent() # Returns job or subjob object
if job._getRoot().subjobs:
jidtag = job._getRoot().id
else:
jidtag = "%d" % job.id
outfiles = app.subjobsOutfiles[job.id]
pandaOutfiles = {}
for type in outfiles.keys():
pandaOutfiles[type] = outfiles[type] + "." + str(jidtag)
if type == "LOG":
pandaOutfiles[type] += ".tgz"
#print pandaOutfiles
for outtype in pandaOutfiles.keys():
fout = FileSpec()
dset = string.replace(app.outputpaths[outtype], "/", ".")
dset = dset[1:-1]
fout.dataset = dset
fout.lfn = pandaOutfiles[outtype]
fout.type = 'output'
# fout.destinationDBlock = jspec.destinationDBlock
fout.destinationDBlock = fout.dataset
fout.destinationSE = jspec.destinationSE
if outtype == 'LOG':
fout.type = 'log'
fout.destinationDBlock = fout.dataset
fout.destinationSE = job.backend.site
jspec.addFile(fout)
# job parameters
param = '-l %s ' % self.library # user tarball.
# use corruption checker
if job.backend.requirements.corCheck:
param += '--corCheck '
# disable to skip missing files
if job.backend.requirements.notSkipMissing:
param += '--notSkipMissing '
# transform parameters
# need to update arglist with final output file name...
newArgs = []
if app.mode == "evgen":
app.args[3] = app.args[3] + " -t "
if app.verbosity:
app.args[3] = app.args[3] + " -l %s " % app.verbosity
for arg in app.args[3:]:
for type in outfiles.keys():
if arg.find(outfiles[type]) > -1:
arg = arg.replace(outfiles[type], pandaOutfiles[type])
newArgs.append(arg)
arglist = string.join(newArgs, " ")
# print "Arglist:",arglist
param += ' -r ./ '
param += ' -j "%s"' % urllib.quote(arglist)
allinfiles = app.inputfiles + app.dbfiles
# Input files.
param += ' -i "%s" ' % allinfiles
if len(app.mbfiles) > 0:
param += ' -m "%s" ' % app.mbfiles
if len(app.cavernfiles) > 0:
param += ' -n "%s" ' % app.cavernfiles
# param += '-m "[]" ' #%minList FIXME
# param += '-n "[]" ' #%cavList FIXME
del pandaOutfiles[
"LOG"] # logfiles do not appear in IROOT block, and this one is not needed anymore...
param += ' -o "{\'IROOT\':%s }"' % str(pandaOutfiles.items())
# source URL
matchURL = re.search("(http.*://[^/]+)/", Client.baseURLSSL)
if matchURL != None:
param += " --sourceURL %s " % matchURL.group(1)
param += " --trf"
jspec.jobParameters = param
jspec.metadata = "--trf \"%s\"" % arglist
#print "SUBJOB DETAILS:",jspec.values()
if app.dryrun:
print "job.application.dryrun activated, printing out job parameters"
print jspec.values()
return
return jspec
