def convertToJobFileSpec(self,
datasetSpec,
setType=None,
useEventService=False):
jobFileSpec = JobFileSpec()
jobFileSpec.fileID = self.fileID
jobFileSpec.datasetID = datasetSpec.datasetID
jobFileSpec.jediTaskID = datasetSpec.jediTaskID
jobFileSpec.lfn = self.lfn
jobFileSpec.GUID = self.GUID
if setType is None:
jobFileSpec.type = self.type
else:
jobFileSpec.type = setType
jobFileSpec.scope = self.scope
jobFileSpec.fsize = self.fsize
jobFileSpec.checksum = self.checksum
jobFileSpec.attemptNr = self.attemptNr
# dataset attribute
if datasetSpec is not None:
# dataset
if datasetSpec.containerName not in [None, '']:
jobFileSpec.dataset = datasetSpec.containerName
else:
jobFileSpec.dataset = datasetSpec.datasetName
if self.type in datasetSpec.getInputTypes(
) or setType in datasetSpec.getInputTypes():
# prodDBlock
jobFileSpec.prodDBlock = datasetSpec.datasetName
# storage token
if datasetSpec.storageToken not in ['', None]:
jobFileSpec.dispatchDBlockToken = datasetSpec.storageToken
else:
# destinationDBlock
jobFileSpec.destinationDBlock = datasetSpec.datasetName
# storage token
if datasetSpec.storageToken not in ['', None]:
jobFileSpec.destinationDBlockToken = datasetSpec.storageToken.split(
'/')[0]
# destination
if datasetSpec.destination not in ['', None]:
jobFileSpec.destinationSE = datasetSpec.destination
# set prodDBlockToken for Event Service
if useEventService and datasetSpec.getObjectStore(
) is not None:
jobFileSpec.prodDBlockToken = 'objectstore^{0}'.format(
datasetSpec.getObjectStore())
# allow no output
if datasetSpec.isAllowedNoOutput():
jobFileSpec.allowNoOutput()
# return
return jobFileSpec
def convertToJobFileSpec(self,datasetSpec,setType=None,useEventService=False):
jobFileSpec = JobFileSpec()
jobFileSpec.fileID = self.fileID
jobFileSpec.datasetID = datasetSpec.datasetID
jobFileSpec.jediTaskID = datasetSpec.jediTaskID
jobFileSpec.lfn = self.lfn
jobFileSpec.GUID = self.GUID
if setType == None:
jobFileSpec.type = self.type
else:
jobFileSpec.type = setType
jobFileSpec.scope = self.scope
jobFileSpec.fsize = self.fsize
jobFileSpec.checksum = self.checksum
jobFileSpec.attemptNr = self.attemptNr
# dataset attribute
if datasetSpec != None:
# dataset
if not datasetSpec.containerName in [None,'']:
jobFileSpec.dataset = datasetSpec.containerName
else:
jobFileSpec.dataset = datasetSpec.datasetName
if self.type in datasetSpec.getInputTypes() or setType in datasetSpec.getInputTypes():
# prodDBlock
jobFileSpec.prodDBlock = datasetSpec.datasetName
# storage token
if not datasetSpec.storageToken in ['',None]:
jobFileSpec.dispatchDBlockToken = datasetSpec.storageToken
else:
# destinationDBlock
jobFileSpec.destinationDBlock = datasetSpec.datasetName
# storage token
if not datasetSpec.storageToken in ['',None]:
jobFileSpec.destinationDBlockToken = datasetSpec.storageToken.split('/')[0]
# destination
if not datasetSpec.destination in ['',None]:
jobFileSpec.destinationSE = datasetSpec.destination
# set prodDBlockToken for Event Service
if useEventService and datasetSpec.getObjectStore() != None:
jobFileSpec.prodDBlockToken = 'objectstore^{0}'.format(datasetSpec.getObjectStore())
# allow no output
if datasetSpec.isAllowedNoOutput():
jobFileSpec.allowNoOutput()
# return
return jobFileSpec
def prepare(self,app,appsubconfig,appmasterconfig,jobmasterconfig):
'''prepare the subjob specific configuration'''
from pandatools import Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaPandaRTHandler prepare called for %s', job.getFQID('.'))
# in case of a simple job get the dataset content, otherwise subjobs are filled by the splitter
if job.inputdata and not job._getRoot().subjobs:
if not job.inputdata.names:
contents = job.inputdata.get_contents(overlap=False, size=True)
for ds in contents.keys():
for f in contents[ds]:
job.inputdata.guids.append( f[0] )
job.inputdata.names.append( f[1][0] )
job.inputdata.sizes.append( f[1][1] )
job.inputdata.checksums.append( f[1][2] )
job.inputdata.scopes.append( f[1][3] )
site = job._getRoot().backend.site
job.backend.site = site
job.backend.actualCE = site
cloud = job._getRoot().backend.requirements.cloud
job.backend.requirements.cloud = cloud
# if no outputdata are given
if not job.outputdata:
job.outputdata = DQ2OutputDataset()
job.outputdata.datasetname = job._getRoot().outputdata.datasetname
#if not job.outputdata.datasetname:
else:
job.outputdata.datasetname = job._getRoot().outputdata.datasetname
if not job.outputdata.datasetname:
raise ApplicationConfigurationError('DQ2OutputDataset has no datasetname')
jspec = JobSpec()
jspec.jobDefinitionID = job._getRoot().id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.transformation = '%s/runGen-00-00-02' % Client.baseURLSUB
if job.inputdata:
jspec.prodDBlock = job.inputdata.dataset[0]
else:
jspec.prodDBlock = 'NULL'
jspec.destinationDBlock = job.outputdata.datasetname
if job.outputdata.location:
if not job._getRoot().subjobs or job.id == 0:
logger.warning('You have specified outputdata.location. Note that Panda may not support writing to a user-defined output location.')
jspec.destinationSE = job.outputdata.location
else:
jspec.destinationSE = site
jspec.prodSourceLabel = configPanda['prodSourceLabelRun']
jspec.processingType = configPanda['processingType']
jspec.assignedPriority = configPanda['assignedPriorityRun']
jspec.cloud = cloud
# memory
if job.backend.requirements.memory != -1:
jspec.minRamCount = job.backend.requirements.memory
# cputime
if job.backend.requirements.cputime != -1:
jspec.maxCpuCount = job.backend.requirements.cputime
jspec.computingSite = site
# library (source files)
if job.backend.libds:
flib = FileSpec()
flib.lfn = self.fileBO.lfn
flib.GUID = self.fileBO.GUID
flib.type = 'input'
flib.status = self.fileBO.status
flib.dataset = self.fileBO.destinationDBlock
flib.dispatchDBlock = self.fileBO.destinationDBlock
jspec.addFile(flib)
elif job.backend.bexec:
flib = FileSpec()
flib.lfn = self.library
flib.type = 'input'
flib.dataset = self.libDataset
flib.dispatchDBlock = self.libDataset
jspec.addFile(flib)
# input files FIXME: many more input types
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):
finp = FileSpec()
finp.lfn = lfn
finp.GUID = guid
finp.scope = scope
# finp.fsize =
# finp.md5sum =
finp.dataset = job.inputdata.dataset[0]
finp.prodDBlock = job.inputdata.dataset[0]
finp.dispatchDBlock = job.inputdata.dataset[0]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# output files
# outMap = {}
#FIXME: if options.outMeta != []:
self.rundirectory = "."
# log files
flog = FileSpec()
flog.lfn = '%s._$PANDAID.log.tgz' % job.outputdata.datasetname
flog.type = 'log'
flog.dataset = job.outputdata.datasetname
flog.destinationDBlock = job.outputdata.datasetname
flog.destinationSE = job.backend.site
jspec.addFile(flog)
# job parameters
param = ''
# source URL
matchURL = re.search("(http.*://[^/]+)/",Client.baseURLCSRVSSL)
srcURL = ""
if matchURL != None:
srcURL = matchURL.group(1)
param += " --sourceURL %s " % srcURL
param += '-r "%s" ' % self.rundirectory
exe_name = job.application.exe
if job.backend.bexec == '':
if hasattr(job.application.exe, "name"):
exe_name = os.path.basename(job.application.exe.name)
# set jobO parameter
if job.application.args:
param += ' -j "" -p "%s %s" '%(exe_name,urllib.quote(" ".join(job.application.args)))
else:
param += ' -j "" -p "%s" '%exe_name
if self.inputsandbox:
param += ' -a %s '%self.inputsandbox
else:
param += '-l %s ' % self.library
param += '-j "" -p "%s %s" ' % ( exe_name,urllib.quote(" ".join(job.application.args)))
if job.inputdata:
param += '-i "%s" ' % job.inputdata.names
# fill outfiles
outfiles = {}
for f in self.extOutFile:
tarnum = 1
if f.find('*') != -1:
# archive *
outfiles[f] = "outputbox%i.%s.%s.tar.gz" % (tarnum, job.getFQID('.'), time.strftime("%Y%m%d%H%M%S") )
tarnum += 1
else:
outfiles[f] = "%s.%s.%s" %(f, job.getFQID('.'), time.strftime("%Y%m%d%H%M%S"))
fout = FileSpec()
fout.lfn = outfiles[f]
fout.type = 'output'
fout.dataset = job.outputdata.datasetname
fout.destinationDBlock = job.outputdata.datasetname
fout.destinationSE = job.backend.site
jspec.addFile(fout)
param += '-o "%s" ' % (outfiles) # must be double quotes, because python prints strings in 'single quotes'
for file in jspec.Files:
if file.type in [ 'output', 'log'] and configPanda['chirpconfig']:
file.dispatchDBlockToken = configPanda['chirpconfig']
logger.debug('chirp file %s',file)
jspec.jobParameters = param
return jspec
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 run(self, data):
datasetName = 'panda:panda.destDB.%s' % commands.getoutput('uuidgen')
destName = 'ANALY_RRC-KI-HPC'
site = 'ANALY_RRC-KI-HPC'
scope = config['DEFAULT_SCOPE']
distributive = data['distributive']
release = data['release']
parameters = data['parameters']
input_type = data['input_type']
input_params = data['input_params']
input_files = data['input_files']
output_type = data['output_type']
output_params = data['output_params']
output_files = data['output_files']
jobid = data['jobid']
_logger.debug('Jobid: ' + str(jobid))
job = JobSpec()
job.jobDefinitionID = int(time.time()) % 10000
job.jobName = commands.getoutput('uuidgen')
job.transformation = config['DEFAULT_TRF']
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 1000
job.prodSourceLabel = 'user'
job.computingSite = site
job.cloud = 'RU'
job.prodDBlock = "%s:%s.%s" % (scope, scope, job.jobName)
job.jobParameters = '%s %s "%s"' % (release, distributive, parameters)
params = {}
_logger.debug('MoveData')
ec = 0
ec, uploaded_input_files = movedata(
params=params,
fileList=input_files,
fromType=input_type,
fromParams=input_params,
toType='hpc',
toParams={'dest': '/' + re.sub(':', '/', job.prodDBlock)})
if ec != 0:
_logger.error('Move data error: ' + ec[1])
return
for file in uploaded_input_files:
fileIT = FileSpec()
fileIT.lfn = file
fileIT.dataset = job.prodDBlock
fileIT.prodDBlock = job.prodDBlock
fileIT.type = 'input'
fileIT.scope = scope
fileIT.status = 'ready'
fileIT.GUID = commands.getoutput('uuidgen')
job.addFile(fileIT)
for file in output_files:
fileOT = FileSpec()
fileOT.lfn = file
fileOT.destinationDBlock = job.prodDBlock
fileOT.destinationSE = job.destinationSE
fileOT.dataset = job.prodDBlock
fileOT.type = 'output'
fileOT.scope = scope
fileOT.GUID = commands.getoutput('uuidgen')
job.addFile(fileOT)
fileOL = FileSpec()
fileOL.lfn = "%s.log.tgz" % job.jobName
fileOL.destinationDBlock = job.destinationDBlock
fileOL.destinationSE = job.destinationSE
fileOL.dataset = job.destinationDBlock
fileOL.type = 'log'
fileOL.scope = 'panda'
job.addFile(fileOL)
self.jobList.append(job)
#submitJob
o = self.submitJobs(self.jobList)
x = o[0]
#update PandaID
conn = MySQLdb.connect(
host=self.dbhost,
db=self.dbname,
# port=self.dbport, connect_timeout=self.dbtimeout,
user=self.dbuser,
passwd=self.dbpasswd)
cur = conn.cursor()
try:
varDict = {}
PandaID = int(x[0])
varDict['id'] = jobid
varDict['pandaId'] = PandaID
sql = "UPDATE %s SET %s.pandaId=%s WHERE %s.id=%s" % (
self.table_jobs, self.table_jobs, varDict['pandaId'],
self.table_jobs, varDict['id'])
cur.execute(sql, varDict)
except:
_logger.error('SENDJOB: Incorrect server response')
try:
conn.commit()
return True
except:
_logger.error("commit error")
return False
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
'''prepare the subjob specific configuration'''
from pandatools import Client
from taskbuffer.JobSpec import JobSpec
from taskbuffer.FileSpec import FileSpec
job = app._getParent()
logger.debug('AthenaPandaRTHandler prepare called for %s',
job.getFQID('.'))
# in case of a simple job get the dataset content, otherwise subjobs are filled by the splitter
if job.inputdata and not job._getRoot().subjobs:
if not job.inputdata.names:
contents = job.inputdata.get_contents(overlap=False, size=True)
for ds in contents.keys():
for f in contents[ds]:
job.inputdata.guids.append(f[0])
job.inputdata.names.append(f[1][0])
job.inputdata.sizes.append(f[1][1])
job.inputdata.checksums.append(f[1][2])
job.inputdata.scopes.append(f[1][3])
site = job._getRoot().backend.site
job.backend.site = site
job.backend.actualCE = site
cloud = job._getRoot().backend.requirements.cloud
job.backend.requirements.cloud = cloud
# if no outputdata are given
if not job.outputdata:
job.outputdata = DQ2OutputDataset()
job.outputdata.datasetname = job._getRoot().outputdata.datasetname
#if not job.outputdata.datasetname:
else:
job.outputdata.datasetname = job._getRoot().outputdata.datasetname
if not job.outputdata.datasetname:
raise ApplicationConfigurationError(
None, 'DQ2OutputDataset has no datasetname')
jspec = JobSpec()
jspec.jobDefinitionID = job._getRoot().id
jspec.jobName = commands.getoutput('uuidgen 2> /dev/null')
jspec.transformation = '%s/runGen-00-00-02' % Client.baseURLSUB
if job.inputdata:
jspec.prodDBlock = job.inputdata.dataset[0]
else:
jspec.prodDBlock = 'NULL'
jspec.destinationDBlock = job.outputdata.datasetname
if job.outputdata.location:
if not job._getRoot().subjobs or job.id == 0:
logger.warning(
'You have specified outputdata.location. Note that Panda may not support writing to a user-defined output location.'
)
jspec.destinationSE = job.outputdata.location
else:
jspec.destinationSE = site
jspec.prodSourceLabel = configPanda['prodSourceLabelRun']
jspec.processingType = configPanda['processingType']
jspec.assignedPriority = configPanda['assignedPriorityRun']
jspec.cloud = cloud
# memory
if job.backend.requirements.memory != -1:
jspec.minRamCount = job.backend.requirements.memory
# cputime
if job.backend.requirements.cputime != -1:
jspec.maxCpuCount = job.backend.requirements.cputime
jspec.computingSite = site
# library (source files)
if job.backend.libds:
flib = FileSpec()
flib.lfn = self.fileBO.lfn
flib.GUID = self.fileBO.GUID
flib.type = 'input'
flib.status = self.fileBO.status
flib.dataset = self.fileBO.destinationDBlock
flib.dispatchDBlock = self.fileBO.destinationDBlock
jspec.addFile(flib)
elif job.backend.bexec:
flib = FileSpec()
flib.lfn = self.library
flib.type = 'input'
flib.dataset = self.libDataset
flib.dispatchDBlock = self.libDataset
jspec.addFile(flib)
# input files FIXME: many more input types
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):
finp = FileSpec()
finp.lfn = lfn
finp.GUID = guid
finp.scope = scope
# finp.fsize =
# finp.md5sum =
finp.dataset = job.inputdata.dataset[0]
finp.prodDBlock = job.inputdata.dataset[0]
finp.dispatchDBlock = job.inputdata.dataset[0]
finp.type = 'input'
finp.status = 'ready'
jspec.addFile(finp)
# output files
# outMap = {}
#FIXME: if options.outMeta != []:
self.rundirectory = "."
# log files
flog = FileSpec()
flog.lfn = '%s._$PANDAID.log.tgz' % job.outputdata.datasetname
flog.type = 'log'
flog.dataset = job.outputdata.datasetname
flog.destinationDBlock = job.outputdata.datasetname
flog.destinationSE = job.backend.site
jspec.addFile(flog)
# job parameters
param = ''
# source URL
matchURL = re.search("(http.*://[^/]+)/", Client.baseURLCSRVSSL)
srcURL = ""
if matchURL != None:
srcURL = matchURL.group(1)
param += " --sourceURL %s " % srcURL
param += '-r "%s" ' % self.rundirectory
exe_name = job.application.exe
if job.backend.bexec == '':
if hasattr(job.application.exe, "name"):
exe_name = os.path.basename(job.application.exe.name)
# set jobO parameter
if job.application.args:
param += ' -j "" -p "%s %s" ' % (
exe_name, urllib.quote(" ".join(job.application.args)))
else:
param += ' -j "" -p "%s" ' % exe_name
if self.inputsandbox:
param += ' -a %s ' % self.inputsandbox
else:
param += '-l %s ' % self.library
param += '-j "" -p "%s %s" ' % (
exe_name, urllib.quote(" ".join(job.application.args)))
if job.inputdata:
param += '-i "%s" ' % job.inputdata.names
# fill outfiles
outfiles = {}
for f in self.extOutFile:
tarnum = 1
if f.find('*') != -1:
# archive *
outfiles[f] = "outputbox%i.%s.%s.tar.gz" % (
tarnum, job.getFQID('.'), time.strftime("%Y%m%d%H%M%S"))
tarnum += 1
else:
outfiles[f] = "%s.%s.%s" % (f, job.getFQID('.'),
time.strftime("%Y%m%d%H%M%S"))
fout = FileSpec()
fout.lfn = outfiles[f]
fout.type = 'output'
fout.dataset = job.outputdata.datasetname
fout.destinationDBlock = job.outputdata.datasetname
fout.destinationSE = job.backend.site
jspec.addFile(fout)
param += '-o "%s" ' % (
outfiles
) # must be double quotes, because python prints strings in 'single quotes'
for file in jspec.Files:
if file.type in ['output', 'log'] and configPanda['chirpconfig']:
file.dispatchDBlockToken = configPanda['chirpconfig']
logger.debug('chirp file %s', file)
jspec.jobParameters = param
return jspec
def send_job(jobid, siteid):
_logger.debug('Jobid: ' + str(jobid))
site = sites_.get(siteid)
job = jobs_.get(int(jobid))
cont = job.container
files_catalog = cont.files
fscope = getScope(job.owner.username)
datasetName = '{}:{}'.format(fscope, cont.guid)
distributive = job.distr.name
release = job.distr.release
# Prepare runScript
parameters = job.distr.command
parameters = parameters.replace("$COMMAND$", job.params)
parameters = parameters.replace("$USERNAME$", job.owner.username)
parameters = parameters.replace("$WORKINGGROUP$", job.owner.working_group)
# Prepare metadata
metadata = dict(user=job.owner.username)
# Prepare PanDA Object
pandajob = JobSpec()
pandajob.jobDefinitionID = int(time.time()) % 10000
pandajob.jobName = cont.guid
pandajob.transformation = client_config.DEFAULT_TRF
pandajob.destinationDBlock = datasetName
pandajob.destinationSE = site.se
pandajob.currentPriority = 1000
pandajob.prodSourceLabel = 'user'
pandajob.computingSite = site.ce
pandajob.cloud = 'RU'
pandajob.VO = 'atlas'
pandajob.prodDBlock = "%s:%s" % (fscope, pandajob.jobName)
pandajob.coreCount = job.corecount
pandajob.metadata = json.dumps(metadata)
#pandajob.workingGroup = job.owner.working_group
if site.encode_commands:
# It requires script wrapper on cluster side
pandajob.jobParameters = '%s %s %s "%s"' % (cont.guid, release,
distributive, parameters)
else:
pandajob.jobParameters = parameters
has_input = False
for fcc in files_catalog:
if fcc.type == 'input':
f = fcc.file
guid = f.guid
fileIT = FileSpec()
fileIT.lfn = f.lfn
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = guid
pandajob.addFile(fileIT)
has_input = True
if fcc.type == 'output':
f = fcc.file
fileOT = FileSpec()
fileOT.lfn = f.lfn
fileOT.destinationDBlock = pandajob.prodDBlock
fileOT.destinationSE = pandajob.destinationSE
fileOT.dataset = pandajob.prodDBlock
fileOT.type = 'output'
fileOT.scope = fscope
fileOT.GUID = f.guid
pandajob.addFile(fileOT)
# Save replica meta
fc.new_replica(f, site)
if not has_input:
# Add fake input
fileIT = FileSpec()
fileIT.lfn = "fake.input"
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = "fake.guid"
pandajob.addFile(fileIT)
# Prepare lof file
fileOL = FileSpec()
fileOL.lfn = "%s.log.tgz" % pandajob.jobName
fileOL.destinationDBlock = pandajob.destinationDBlock
fileOL.destinationSE = pandajob.destinationSE
fileOL.dataset = '{}:logs'.format(fscope)
fileOL.type = 'log'
fileOL.scope = 'panda'
pandajob.addFile(fileOL)
# Save log meta
log = File()
log.scope = fscope
log.lfn = fileOL.lfn
log.guid = getGUID(log.scope, log.lfn)
log.type = 'log'
log.status = 'defined'
files_.save(log)
# Save replica meta
fc.new_replica(log, site)
# Register file in container
fc.reg_file_in_cont(log, cont, 'log')
# Submit job
o = submitJobs([pandajob])
x = o[0]
try:
#update PandaID
PandaID = int(x[0])
job.pandaid = PandaID
job.ce = site.ce
except:
job.status = 'submit_error'
jobs_.save(job)
return 0
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
job.prodSourceLabel = 'user'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'x86_64-slc6-gcc48-opt'
job.specialHandling = 'ddm:rucio'
#job.transferType = 'direct'
ifile = 'HITS.06828093._000096.pool.root.1'
fileI = FileSpec()
fileI.GUID = 'AC5B3759-B606-BA42-8681-4BD86455AE02'
fileI.checksum = 'ad:5d000974'
fileI.dataset = 'user.mlassnig:user.mlassnig.pilot.test.single.hits'
fileI.fsize = 94834717
fileI.lfn = ifile
fileI.prodDBlock = job.prodDBlock
fileI.scope = 'mc15_13TeV'
fileI.type = 'input'
job.addFile(fileI)
ofile = 'RDO_%s.root' % commands.getoutput('uuidgen')
fileO = FileSpec()
fileO.dataset = job.destinationDBlock
fileO.destinationDBlock = job.destinationDBlock
fileO.destinationSE = job.destinationSE
fileO.lfn = ofile
fileO.type = 'output'
job.addFile(fileO)
fileOL = FileSpec()
fileOL.lfn = "%s.job.log.tgz" % job.jobName
fileOL.destinationDBlock = job.destinationDBlock
def send_job(jobid, siteid):
_logger.debug('Jobid: ' + str(jobid))
site = sites_.get(siteid)
job = jobs_.get(int(jobid))
cont = job.container
files_catalog = cont.files
fscope = getScope(job.owner.username)
datasetName = '{}:{}'.format(fscope, cont.guid)
distributive = job.distr.name
release = job.distr.release
# Prepare runScript
parameters = job.distr.command
parameters = parameters.replace("$COMMAND$", job.params)
parameters = parameters.replace("$USERNAME$", job.owner.username)
parameters = parameters.replace("$WORKINGGROUP$", job.owner.working_group)
# Prepare metadata
metadata = dict(user=job.owner.username)
# Prepare PanDA Object
pandajob = JobSpec()
pandajob.jobDefinitionID = int(time.time()) % 10000
pandajob.jobName = cont.guid
pandajob.transformation = client_config.DEFAULT_TRF
pandajob.destinationDBlock = datasetName
pandajob.destinationSE = site.se
pandajob.currentPriority = 1000
pandajob.prodSourceLabel = 'user'
pandajob.computingSite = site.ce
pandajob.cloud = 'RU'
pandajob.VO = 'atlas'
pandajob.prodDBlock = "%s:%s" % (fscope, pandajob.jobName)
pandajob.coreCount = job.corecount
pandajob.metadata = json.dumps(metadata)
#pandajob.workingGroup = job.owner.working_group
if site.encode_commands:
# It requires script wrapper on cluster side
pandajob.jobParameters = '%s %s %s "%s"' % (cont.guid, release, distributive, parameters)
else:
pandajob.jobParameters = parameters
has_input = False
for fcc in files_catalog:
if fcc.type == 'input':
f = fcc.file
guid = f.guid
fileIT = FileSpec()
fileIT.lfn = f.lfn
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = guid
pandajob.addFile(fileIT)
has_input = True
if fcc.type == 'output':
f = fcc.file
fileOT = FileSpec()
fileOT.lfn = f.lfn
fileOT.destinationDBlock = pandajob.prodDBlock
fileOT.destinationSE = pandajob.destinationSE
fileOT.dataset = pandajob.prodDBlock
fileOT.type = 'output'
fileOT.scope = fscope
fileOT.GUID = f.guid
pandajob.addFile(fileOT)
# Save replica meta
fc.new_replica(f, site)
if not has_input:
# Add fake input
fileIT = FileSpec()
fileIT.lfn = "fake.input"
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = "fake.guid"
pandajob.addFile(fileIT)
# Prepare lof file
fileOL = FileSpec()
fileOL.lfn = "%s.log.tgz" % pandajob.jobName
fileOL.destinationDBlock = pandajob.destinationDBlock
fileOL.destinationSE = pandajob.destinationSE
fileOL.dataset = '{}:logs'.format(fscope)
fileOL.type = 'log'
fileOL.scope = 'panda'
pandajob.addFile(fileOL)
# Save log meta
log = File()
log.scope = fscope
log.lfn = fileOL.lfn
log.guid = getGUID(log.scope, log.lfn)
log.type = 'log'
log.status = 'defined'
files_.save(log)
# Save replica meta
fc.new_replica(log, site)
# Register file in container
fc.reg_file_in_cont(log, cont, 'log')
# Submit job
o = submitJobs([pandajob])
x = o[0]
try:
#update PandaID
PandaID = int(x[0])
job.pandaid = PandaID
job.ce = site.ce
except:
job.status = 'submit_error'
jobs_.save(job)
return 0
job.prodSourceLabel = 'user'
job.computingSite = site
job.cloud = cloud
job.cmtConfig = 'x86_64-slc6-gcc48-opt'
job.specialHandling = 'ddm:rucio'
#job.transferType = 'direct'
ifile = 'HITS.06828093._000096.pool.root.1'
fileI = FileSpec()
fileI.GUID = 'AC5B3759-B606-BA42-8681-4BD86455AE02'
fileI.checksum = 'ad:5d000974'
fileI.dataset = 'user.mlassnig:user.mlassnig.pilot.test.single.hits'
fileI.fsize = 94834717
fileI.lfn = ifile
fileI.prodDBlock = job.prodDBlock
fileI.scope = 'mc15_13TeV'
fileI.type = 'input'
job.addFile(fileI)
ofile = 'RDO_%s.root' % commands.getoutput('uuidgen')
fileO = FileSpec()
fileO.dataset = job.destinationDBlock
fileO.destinationDBlock = job.destinationDBlock
fileO.destinationSE = job.destinationSE
fileO.lfn = ofile
fileO.type = 'output'
job.addFile(fileO)
fileOL = FileSpec()
fileOL.lfn = "%s.job.log.tgz" % job.jobName
fileOL.destinationDBlock = job.destinationDBlock
