def finalizeTask(self, update=False, debug=False):
outlfn = self.crabConfig.Data.outLFNDirBase.split("/store/user/")[1]
if outlfn.endswith("/"):
outlfn = outlfn[:-1]
crab = CrabController()
# Check files for each job
dCacheFiles = gridFunctions.getdcachelist(outlfn)
# ~ success , failed = crab.getlog( sample )
if debug:
cmd = "crab log %s" % crab._prepareFoldername(self.name)
else:
# CRAB likes to be verbos, even if you tell it to be quiet...
cmd = "crab --quiet log %s" % crab._prepareFoldername(self.name)
p = subprocess.Popen(
cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True
) # ,shell=True,universal_newlines=True)
(out, err) = p.communicate()
# ~ print out
crabFolder = crab._prepareFoldername(self.name)
# ~ print crabFolder
logArchs = glob.glob("%s/%s/results/*.log.tar.gz" % (self.crabConfig.General.workArea, crabFolder))
finalFiles = []
totalEvents = 0
# ~ sys.exit()
for logArchName in logArchs:
JobNumber = logArchName.split("/")[-1].split("_")[1].split(".")[0]
# ~ print logArchName
log = self.readLogArch(logArchName)
# check if file on dCache
dfile = []
for layer in dCacheFiles:
dfile += [s for s in layer if "%s_%s.pxlio" % (self.name, JobNumber) in s]
if len(dfile) > 0 and log["readEvents"] > 0:
finalFiles.append({"path": dfile[0], "nevents": log["readEvents"]})
totalEvents += log["readEvents"]
self.finalFiles = finalFiles
self.totalEvents = totalEvents
if self.isData:
self.addData2db(True)
else:
self.addMC2db(True)
with open("finalSample", "a") as outfile:
outfile.write("%s:%s\n" % (self.name, self.crabConfig.Data.inputDataset))
self.state = "FINAL"
def finalizeTask(self , update = False, debug= False ):
outlfn = self.crabConfig.Data.outLFNDirBase.split('/store/user/')[1]
if outlfn.endswith("/"): outlfn =outlfn[:-1]
crab = CrabController()
# Check files for each job
dCacheFiles = gridFunctions.getdcachelist( outlfn )
#~ success , failed = crab.getlog( sample )
if debug:
cmd = 'crab log %s' % crab._prepareFoldername( self.name )
else:
# CRAB likes to be verbos, even if you tell it to be quiet...
cmd = 'crab --quiet log %s' % crab._prepareFoldername( self.name )
p = subprocess.Popen(cmd,stdout=subprocess.PIPE,stderr=subprocess.PIPE, shell=True)#,shell=True,universal_newlines=True)
(out,err) = p.communicate()
#~ print out
crabFolder = crab._prepareFoldername( self.name )
#~ print crabFolder
logArchs = glob.glob("%s/%s/results/*.log.tar.gz" % (self.crabConfig.General.workArea,crabFolder))
finalFiles = []
totalEvents = 0
#~ sys.exit()
for logArchName in logArchs:
JobNumber = logArchName.split("/")[-1].split("_")[1].split(".")[0]
#~ print logArchName
log = self.readLogArch( logArchName )
# check if file on dCache
dfile = []
for layer in dCacheFiles:
dfile += [s for s in layer if "%s_%s.pxlio" %( self.name, JobNumber ) in s ]
if len(dfile) > 0 and log['readEvents'] > 0 :
finalFiles.append( {'path':dfile[0], 'nevents':log['readEvents']} )
totalEvents += log['readEvents']
self.finalFiles = finalFiles
self.totalEvents = totalEvents
if self.isData:
self.addData2db( True )
else:
self.addMC2db( True )
with open('finalSample','a') as outfile:
outfile.write("%s:%s\n" % ( self.name, self.crabConfig.Data.inputDataset))
self.state = "FINAL"
def getFilesfromFile(cfgFile):
sampleList={}
file = open(cfgFile,'r')
user,tag,sample,config,mainFolder=["","","","",""]
known_folders=dict()
if not os.path.exists("data"):
os.mkdir("data")
#if os.path.exists("data/fileList.pkl"):
#known_folders=readDcachePickle("data/fileList.pkl")
for line in file:
if line[0]=="#" or len(line.split())==0:
continue
if "tag=" in line:
tag=line.split("=")[1].strip()
continue
if "user="******"=")[1].strip()
continue
if "config=" in line:
config=line.split("=")[1].strip()
continue
if "mainFolder=" in line:
mainFolder=line.split("=")[1].strip()
continue
sample=line.strip()
log.debug( " ".join([user,tag,sample,config]))
if mainFolder=="":
mainFolder="MUSiC"
folder="/%s/%s/%s/%s" % (user,mainFolder,tag,sample)
sampleFolder=folder.replace("/","")
sampleFolder+=".pkl"
if os.path.exists("data/"+sampleFolder):
known_folders=readDcachePickle("data/"+sampleFolder)
file_lists_1=known_folders[folder]
file_lists=[]
for j in file_lists_1:
#file_lists.append([ i.replace("dcap://grid-dcap.physik.rwth-aachen.de/pnfs","/pnfs") for i in j])
file_lists.append([ i for i in j])
else:
#time.sleep(4)
file_lists=[]
if "Data" in sample:
file_lists_1 = getdcachelist( folder, sample,mem_limit = 1000000000 )
else:
file_lists_1 = getdcachelist( folder, sample,mem_limit = 4000000000 )
for j in file_lists_1:
#file_lists.append([ i.replace("dcap://grid-dcap.physik.rwth-aachen.de/pnfs","/pnfs") for i in j])
file_lists.append([ i for i in j])
outfile = open( "data/"+sampleFolder, 'a+b' )
cPickle.dump( {folder:file_lists}, outfile, -1 )
outfile.close()
if len(file_lists)>0:
file_lists=filter(lambda x: "failed" not in x, file_lists)
sampleList.update({sample:[file_lists,config]})
else:
print sample
#raise IOError( 'No sample in List for folder '+sample )
return sampleList
