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 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
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])
print s
if s == 0:
for job in o:
if job == None:
continue
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.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])
print s
if s == 0:
for job in o:
if job == None:
continue
def main():
logger.info('Getting tasks with status send and running')
# tasks_list = Task.objects.all().filter(Q(status='send') | Q(status='running'))
tasks_list = Task.objects.all().filter(name='dvcs2016P09t2r13v1_mu+')
logger.info('Got list of %s tasks' % len(tasks_list))
for t in tasks_list:
logger.info('Getting jobs in status defined or failed for task %s' % t)
jobs_list_count = Job.objects.all().filter(task=t).count()
if jobs_list_count > 50:
jobs_list = Job.objects.all().filter(
task=t).order_by('id')[:max_send_amount]
else:
jobs_list = Job.objects.all().filter(
task=t).order_by('id')[:jobs_list_count]
logger.info('Got list of %s jobs' % len(jobs_list))
i = 0
for j in jobs_list:
if i >= max_send_amount:
break
logger.info('Going to send job %s of %s task' %
(j.file, j.task.name))
umark = commands.getoutput('uuidgen')
datasetName = 'panda.destDB.%s' % umark
destName = 'COMPASSPRODDISK' # PanDA will not try to move output data, data will be placed by pilot (based on schedconfig)
TMPRAWFILE = j.file[j.file.rfind('/') + 1:]
logger.info(TMPRAWFILE)
TMPMDSTFILE = 'mDST-%(runNumber)s-%(runChunk)s-%(prodSlt)s-%(phastVer)s.root' % {
'input_file': j.file,
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt,
'phastVer': j.task.phastver
}
logger.info(TMPMDSTFILE)
TMPHISTFILE = '%(runNumber)s-%(runChunk)s-%(prodSlt)s.root' % {
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt
}
logger.info(TMPHISTFILE)
TMPRICHFILE = 'gfile_%(runNumber)s-%(runChunk)s.gfile' % {
'runNumber': j.run_number,
'runChunk': j.chunk_number
}
logger.info(TMPRICHFILE)
EVTDUMPFILE = 'evtdump%(prodSlt)s-%(runChunk)s-%(runNumber)s.raw' % {
'prodSlt': j.task.prodslt,
'runNumber': j.run_number,
'runChunk': j.chunk_number
}
logger.info(EVTDUMPFILE)
STDOUTFILE = '%(prodNameOnly)s.%(runNumber)s-%(runChunk)s-%(prodSlt)s.stdout' % {
'prodNameOnly': j.task.soft,
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt
}
logger.info(STDOUTFILE)
STDERRFILE = '%(prodNameOnly)s.%(runNumber)s-%(runChunk)s-%(prodSlt)s.stderr' % {
'prodNameOnly': j.task.soft,
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt
}
logger.info(STDERRFILE)
try:
file_year = j.file.split('/')[5]
logger.info(file_year)
except:
logger.error('Error while splitting file to get year')
sys.exit(1)
ProdPathAndName = j.task.home + j.task.path + j.task.soft
job = JobSpec()
job.taskID = j.task.id
job.jobDefinitionID = 0
job.jobName = '%(prodName)s-%(fileYear)s--%(runNumber)s-%(runChunk)s-%(prodSlt)s-%(phastVer)s' % {
'prodName': j.task.soft,
'fileYear': file_year,
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt,
'phastVer': j.task.phastver
}
job.transformation = j.task.type # payload (can be URL as well)
job.destinationDBlock = datasetName
job.destinationSE = destName
job.currentPriority = 2000
job.prodSourceLabel = 'prod_test'
job.computingSite = site
job.attemptNr = j.attempt + 1
job.maxAttempt = j.task.max_attempts
if j.status == 'failed':
job.parentID = j.panda_id
head, tail = os.path.split(j.file)
# job.transferType = 'direct'
job.sourceSite = 'CERN_COMPASS_PROD'
# logs, and all files generated during execution will be placed in log (except output file)
#job.jobParameters='source /afs/cern.ch/project/eos/installation/compass/etc/setup.sh;export EOS_MGM_URL=root://eoscompass.cern.ch;export PATH=/afs/cern.ch/project/eos/installation/compass/bin:$PATH;ppwd=$(pwd);echo $ppwd;export TMPMDSTFILE=%(TMPMDSTFILE)s;export TMPHISTFILE=%(TMPHISTFILE)s;export TMPRICHFILE=%(TMPRICHFILE)s;coralpath=%(ProdPathAndName)s/coral;echo $coralpath;cd -P $coralpath;export coralpathsetup=$coralpath"/setup.sh";echo $coralpathsetup;source $coralpathsetup;cd $ppwd;$CORAL/../phast/coral/coral.exe %(ProdPathAndName)s/template.opt;xrdcp -np $ppwd/%(TMPMDSTFILE)s xroot://eoscompass.cern.ch//eos/compass/%(prodName)s/mDST/%(TMPMDSTFILE)s;xrdcp -np $ppwd/%(TMPHISTFILE)s xroot://eoscompass.cern.ch//eos/compass/%(prodName)s/histos/%(TMPHISTFILE)s;metadataxml=$(ls metadata-*);echo $metadataxml;cp $metadataxml $metadataxml.PAYLOAD;' % {'TMPMDSTFILE': TMPMDSTFILE, 'TMPHISTFILE': TMPHISTFILE, 'TMPRICHFILE': TMPRICHFILE, 'input_file': input_file, 'ProdPathAndName': ProdPathAndName, 'prodName': prodName}
job.jobParameters = 'export EOS_MGM_URL=root://eoscompass.cern.ch;ppwd=$(pwd);export COMPASS_SW_PREFIX=/eos/experiment/compass/;export COMPASS_SW_PATH=%(prodPath)s;export COMPASS_PROD_NAME=%(prodName)s;export TMPRAWFILE=%(TMPRAWFILE)s;export TMPMDSTFILE=%(TMPMDSTFILE)s;export TMPHISTFILE=%(TMPHISTFILE)s;export TMPRICHFILE=%(TMPRICHFILE)s;export prodSlt=%(prodSlt)s;export EVTDUMPFILE=%(EVTDUMPFILE)s;xrdcp -N -f root://castorpublic.cern.ch/%(input_file)s\?svcClass=compasscdr .;coralpath=%(ProdPathAndName)s/coral;cd -P $coralpath;export coralpathsetup=$coralpath"/setup.sh";source $coralpathsetup;cd $ppwd;$CORAL/../phast/coral/coral.exe %(ProdPathAndName)s/%(template)s;if [ ! -s testevtdump.raw ]; then echo "PanDA message: the file is empty">testevtdump.raw; fi;cp payload_stderr.txt payload_stderr.out;cp payload_stdout.txt payload_stdout.out;rm %(tail)s' % {
'TMPRAWFILE': TMPRAWFILE,
'TMPMDSTFILE': TMPMDSTFILE,
'TMPHISTFILE': TMPHISTFILE,
'TMPRICHFILE': TMPRICHFILE,
'input_file': j.file,
'ProdPathAndName': ProdPathAndName,
'prodPath': j.task.path,
'prodName': j.task.soft,
'template': j.task.template,
'tail': tail,
'prodSlt': j.task.prodslt,
'EVTDUMPFILE': EVTDUMPFILE,
'STDOUTFILE': STDOUTFILE,
'STDERRFILE': STDERRFILE
}
fileIRaw = FileSpec()
fileIRaw.lfn = "%s" % (j.file)
fileIRaw.GUID = '5874a461-61d3-4543-8f34-6fd7a4624e78'
fileIRaw.fsize = 1073753368
fileIRaw.checksum = '671608be'
fileIRaw.destinationDBlock = job.destinationDBlock
fileIRaw.destinationSE = job.destinationSE
fileIRaw.dataset = job.destinationDBlock
fileIRaw.type = 'input'
job.addFile(fileIRaw)
fileOstdout = FileSpec()
fileOstdout.lfn = "payload_stdout.txt"
fileOstdout.destinationDBlock = job.destinationDBlock
fileOstdout.destinationSE = job.destinationSE
fileOstdout.dataset = job.destinationDBlock
fileOstdout.type = 'output'
job.addFile(fileOstdout)
fileOstderr = FileSpec()
fileOstderr.lfn = "payload_stderr.txt"
fileOstderr.destinationDBlock = job.destinationDBlock
fileOstderr.destinationSE = job.destinationSE
fileOstderr.dataset = job.destinationDBlock
fileOstderr.type = 'output'
job.addFile(fileOstderr)
fileOLog = FileSpec()
fileOLog.lfn = "%(prodName)s-%(runNumber)s-%(runChunk)s-%(prodSlt)s-%(phastVer)s.job.log.tgz" % {
'prodName': j.task.soft,
'runNumber': j.run_number,
'runChunk': j.chunk_number,
'prodSlt': j.task.prodslt,
'phastVer': j.task.phastver
}
fileOLog.destinationDBlock = job.destinationDBlock
fileOLog.destinationSE = job.destinationSE
fileOLog.dataset = job.destinationDBlock
fileOLog.type = 'log'
job.addFile(fileOLog)
fileOmDST = FileSpec()
fileOmDST.lfn = "%s" % (TMPMDSTFILE)
fileOmDST.destinationDBlock = job.destinationDBlock
fileOmDST.destinationSE = job.destinationSE
fileOmDST.dataset = job.destinationDBlock
fileOmDST.type = 'output'
job.addFile(fileOmDST)
fileOTrafdic = FileSpec()
fileOTrafdic.lfn = "%s" % (TMPHISTFILE)
fileOTrafdic.destinationDBlock = job.destinationDBlock
fileOTrafdic.destinationSE = job.destinationSE
fileOTrafdic.dataset = job.destinationDBlock
fileOTrafdic.type = 'output'
job.addFile(fileOTrafdic)
fileOtestevtdump = FileSpec()
fileOtestevtdump.lfn = "testevtdump.raw"
fileOtestevtdump.destinationDBlock = job.destinationDBlock
fileOtestevtdump.destinationSE = job.destinationSE
fileOtestevtdump.dataset = job.destinationDBlock
fileOtestevtdump.type = 'output'
job.addFile(fileOtestevtdump)
s, o = Client.submitJobs([job], srvID=aSrvID)
logger.info(s)
logger.info(o)
# for x in o:
# logger.info("PandaID=%s" % x[0])
# today = datetime.datetime.today()
#
# if x[0] != 0 and x[0] != 'NULL':
# j_update = Job.objects.get(id=j.id)
# j_update.panda_id = x[0]
# j_update.status = 'sent'
# j_update.attempt = j_update.attempt + 1
# j_update.date_updated = today
#
# try:
# j_update.save()
# logger.info('Job %s with PandaID %s updated at %s' % (j.id, x[0], today))
# except IntegrityError as e:
# logger.exception('Unique together catched, was not saved')
# except DatabaseError as e:
# logger.exception('Something went wrong while saving: %s' % e.message)
# else:
# logger.info('Job %s was not added to PanDA' % j.id)
i += 1
logger.info('done')
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'
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
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.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'
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
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
