| 2 |
|
from crab_logger import Logger |
| 3 |
|
from crab_exceptions import * |
| 4 |
|
from crab_util import * |
| 5 |
< |
from BlackWhiteListParser import BlackWhiteListParser |
| 5 |
> |
from BlackWhiteListParser import SEBlackWhiteListParser |
| 6 |
|
import common |
| 7 |
|
import Scram |
| 8 |
|
from LFNBaseName import * |
| 14 |
|
JobType.__init__(self, 'CMSSW') |
| 15 |
|
common.logger.debug(3,'CMSSW::__init__') |
| 16 |
|
self.skip_blocks = skip_blocks |
| 17 |
< |
|
| 17 |
> |
|
| 18 |
|
self.argsList = [] |
| 19 |
|
|
| 20 |
|
self._params = {} |
| 21 |
|
self.cfg_params = cfg_params |
| 22 |
|
# init BlackWhiteListParser |
| 23 |
< |
self.blackWhiteListParser = BlackWhiteListParser(cfg_params) |
| 23 |
> |
self.blackWhiteListParser = SEBlackWhiteListParser(cfg_params) |
| 24 |
|
|
| 25 |
|
self.MaxTarBallSize = float(self.cfg_params.get('EDG.maxtarballsize',9.5)) |
| 26 |
|
|
| 60 |
|
if not cfg_params.has_key('CMSSW.datasetpath'): |
| 61 |
|
msg = "Error: datasetpath not defined " |
| 62 |
|
raise CrabException(msg) |
| 63 |
+ |
|
| 64 |
+ |
### Temporary: added to remove input file control in the case of PU |
| 65 |
+ |
self.dataset_pu = cfg_params.get('CMSSW.dataset_pu', None) |
| 66 |
+ |
|
| 67 |
|
tmp = cfg_params['CMSSW.datasetpath'] |
| 68 |
|
log.debug(6, "CMSSW::CMSSW(): datasetPath = "+tmp) |
| 69 |
|
if string.lower(tmp)=='none': |
| 99 |
|
self.output_file_sandbox.append(self.fjrFileName) |
| 100 |
|
|
| 101 |
|
# other output files to be returned via sandbox or copied to SE |
| 102 |
+ |
outfileflag = False |
| 103 |
|
self.output_file = [] |
| 104 |
|
tmp = cfg_params.get('CMSSW.output_file',None) |
| 105 |
|
if tmp : |
| 106 |
|
self.output_file = [x.strip() for x in tmp.split(',')] |
| 107 |
< |
else: |
| 108 |
< |
log.message("No output file defined: only stdout/err and the CRAB Framework Job Report will be available\n") |
| 109 |
< |
pass |
| 107 |
> |
outfileflag = True #output found |
| 108 |
> |
#else: |
| 109 |
> |
# log.message("No output file defined: only stdout/err and the CRAB Framework Job Report will be available\n") |
| 110 |
|
|
| 111 |
|
# script_exe file as additional file in inputSandbox |
| 112 |
|
self.scriptExe = cfg_params.get('USER.script_exe',None) |
| 120 |
|
msg ="Error. script_exe not defined" |
| 121 |
|
raise CrabException(msg) |
| 122 |
|
|
| 123 |
< |
# use parent files... |
| 123 |
> |
# use parent files... |
| 124 |
|
self.useParent = self.cfg_params.get('CMSSW.use_parent',False) |
| 125 |
|
|
| 126 |
|
## additional input files |
| 197 |
|
tmp.strip() |
| 198 |
|
self.incrementSeeds.append(tmp) |
| 199 |
|
|
| 200 |
< |
## Old method of dealing with seeds |
| 201 |
< |
## FUTURE: This is for old CMSSW and old CRAB. Can throw exceptions after a couple of CRAB releases and then |
| 197 |
< |
## remove |
| 198 |
< |
self.sourceSeed = cfg_params.get('CMSSW.pythia_seed',None) |
| 199 |
< |
if self.sourceSeed: |
| 200 |
< |
print "pythia_seed is a deprecated parameter. Use preserve_seeds or increment_seeds in the future.\n","Added to increment_seeds." |
| 201 |
< |
self.incrementSeeds.append('sourceSeed') |
| 202 |
< |
self.incrementSeeds.append('theSource') |
| 203 |
< |
|
| 200 |
> |
## FUTURE: Can remove in CRAB 2.4.0 |
| 201 |
> |
self.sourceSeed = cfg_params.get('CMSSW.pythia_seed',None) |
| 202 |
|
self.sourceSeedVtx = cfg_params.get('CMSSW.vtx_seed',None) |
| 203 |
< |
if self.sourceSeedVtx: |
| 206 |
< |
print "vtx_seed is a deprecated parameter. Use preserve_seeds or increment_seeds in the future.\n","Added to increment_seeds." |
| 207 |
< |
self.incrementSeeds.append('VtxSmeared') |
| 208 |
< |
|
| 209 |
< |
self.sourceSeedG4 = cfg_params.get('CMSSW.g4_seed',None) |
| 210 |
< |
if self.sourceSeedG4: |
| 211 |
< |
print "g4_seed is a deprecated parameter. Use preserve_seeds or increment_seeds in the future.\n","Added to increment_seeds." |
| 212 |
< |
self.incrementSeeds.append('g4SimHits') |
| 213 |
< |
|
| 203 |
> |
self.sourceSeedG4 = cfg_params.get('CMSSW.g4_seed',None) |
| 204 |
|
self.sourceSeedMix = cfg_params.get('CMSSW.mix_seed',None) |
| 205 |
< |
if self.sourceSeedMix: |
| 206 |
< |
print "mix_seed is a deprecated parameter. Use preserve_seeds or increment_seeds in the future.\n","Added to increment_seeds." |
| 207 |
< |
self.incrementSeeds.append('mix') |
| 205 |
> |
if self.sourceSeed or self.sourceSeedVtx or self.sourceSeedG4 or self.sourceSeedMix: |
| 206 |
> |
msg = 'pythia_seed, vtx_seed, g4_seed, and mix_seed are no longer valid settings. You must use increment_seeds or preserve_seeds' |
| 207 |
> |
raise CrabException(msg) |
| 208 |
|
|
| 209 |
|
self.firstRun = cfg_params.get('CMSSW.first_run',None) |
| 210 |
|
|
| 221 |
– |
|
| 211 |
|
# Copy/return |
| 212 |
|
self.copy_data = int(cfg_params.get('USER.copy_data',0)) |
| 213 |
|
self.return_data = int(cfg_params.get('USER.return_data',0)) |
| 247 |
|
## If present, add TFileService to output files |
| 248 |
|
if not int(cfg_params.get('CMSSW.skip_TFileService_output',0)): |
| 249 |
|
tfsOutput = PsetEdit.getTFileService() |
| 250 |
< |
if tfsOutput: |
| 250 |
> |
if tfsOutput: |
| 251 |
|
if tfsOutput in self.output_file: |
| 252 |
|
common.logger.debug(5,"Output from TFileService "+tfsOutput+" already in output files") |
| 253 |
|
else: |
| 254 |
+ |
outfileflag = True #output found |
| 255 |
|
self.output_file.append(tfsOutput) |
| 256 |
|
common.logger.message("Adding "+tfsOutput+" to output files (from TFileService)") |
| 257 |
+ |
pass |
| 258 |
+ |
pass |
| 259 |
+ |
## If present and requested, add PoolOutputModule to output files |
| 260 |
+ |
if int(cfg_params.get('CMSSW.get_edm_output',0)): |
| 261 |
+ |
edmOutput = PsetEdit.getPoolOutputModule() |
| 262 |
+ |
if edmOutput: |
| 263 |
+ |
if edmOutput in self.output_file: |
| 264 |
+ |
common.logger.debug(5,"Output from PoolOutputModule "+edmOutput+" already in output files") |
| 265 |
+ |
else: |
| 266 |
+ |
self.output_file.append(edmOutput) |
| 267 |
+ |
common.logger.message("Adding "+edmOutput+" to output files (from PoolOutputModule)") |
| 268 |
+ |
pass |
| 269 |
+ |
pass |
| 270 |
|
except CrabException: |
| 271 |
|
msg='Error while manipulating ParameterSet: exiting...' |
| 272 |
|
raise CrabException(msg) |
| 273 |
< |
## Prepare inputSandbox TarBall (only the first time) |
| 273 |
> |
## Prepare inputSandbox TarBall (only the first time) |
| 274 |
|
self.tgzNameWithPath = self.getTarBall(self.executable) |
| 275 |
|
|
| 276 |
|
def DataDiscoveryAndLocation(self, cfg_params): |
| 478 |
|
if self.useParent: |
| 479 |
|
fullParentString = pString[:-2] |
| 480 |
|
list_of_lists.append([fullString,fullParentString,str(eventsPerJobRequested),str(jobSkipEventCount)]) |
| 481 |
< |
else: |
| 481 |
> |
else: |
| 482 |
|
list_of_lists.append([fullString,str(eventsPerJobRequested),str(jobSkipEventCount)]) |
| 483 |
|
common.logger.debug(3,"Job "+str(jobCount+1)+" can run over "+str(eventsPerJobRequested)+" events.") |
| 484 |
|
self.jobDestination.append(blockSites[block]) |
| 794 |
|
common.logger.debug(5,"Files added to "+self.tgzNameWithPath+" : "+str(tar.getnames())) |
| 795 |
|
|
| 796 |
|
tar.close() |
| 797 |
+ |
except IOError: |
| 798 |
+ |
raise CrabException('Could not create tar-ball '+self.tgzNameWithPath) |
| 799 |
|
except tarfile.TarError: |
| 800 |
|
raise CrabException('Could not create tar-ball '+self.tgzNameWithPath) |
| 801 |
|
|
| 876 |
|
txt += 'DatasetPath='+self.datasetPath+'\n' |
| 877 |
|
|
| 878 |
|
datasetpath_split = self.datasetPath.split("/") |
| 879 |
< |
|
| 879 |
> |
### FEDE FOR NEW LFN ### |
| 880 |
> |
self.primaryDataset = datasetpath_split[1] |
| 881 |
> |
######################## |
| 882 |
|
txt += 'PrimaryDataset='+datasetpath_split[1]+'\n' |
| 883 |
|
txt += 'DataTier='+datasetpath_split[2]+'\n' |
| 884 |
|
txt += 'ApplicationFamily=cmsRun\n' |
| 885 |
|
|
| 886 |
|
else: |
| 887 |
|
txt += 'DatasetPath=MCDataTier\n' |
| 888 |
+ |
### FEDE FOR NEW LFN ### |
| 889 |
+ |
self.primaryDataset = 'null' |
| 890 |
+ |
######################## |
| 891 |
|
txt += 'PrimaryDataset=null\n' |
| 892 |
|
txt += 'DataTier=null\n' |
| 893 |
|
txt += 'ApplicationFamily=MCDataTier\n' |
| 897 |
|
txt += 'cp $RUNTIME_AREA/'+pset+' .\n' |
| 898 |
|
if (self.datasetPath): # standard job |
| 899 |
|
txt += 'InputFiles=${args[1]}; export InputFiles\n' |
| 900 |
< |
if (self.useParent): |
| 900 |
> |
if (self.useParent): |
| 901 |
|
txt += 'ParentFiles=${args[2]}; export ParentFiles\n' |
| 902 |
|
txt += 'MaxEvents=${args[3]}; export MaxEvents\n' |
| 903 |
|
txt += 'SkipEvents=${args[4]}; export SkipEvents\n' |
| 928 |
|
txt += 'cat ' + psetName + '\n' |
| 929 |
|
txt += 'echo "****** end ' + psetName + ' ********"\n' |
| 930 |
|
txt += '\n' |
| 931 |
< |
txt += 'PSETHASH=`edmConfigHash < ' + psetName + '` \n' |
| 931 |
> |
if (self.CMSSW_major >= 2 and self.CMSSW_minor >= 1) or (self.CMSSW_major >= 3): |
| 932 |
> |
txt += 'PSETHASH=`edmConfigHash ' + psetName + '` \n' |
| 933 |
> |
else: |
| 934 |
> |
txt += 'PSETHASH=`edmConfigHash < ' + psetName + '` \n' |
| 935 |
|
txt += 'echo "PSETHASH = $PSETHASH" \n' |
| 936 |
|
txt += '\n' |
| 937 |
|
return txt |
| 1059 |
|
txt = '\n#Written by cms_cmssw::wsRenameOutput\n' |
| 1060 |
|
txt += 'echo ">>> current directory (SOFTWARE_DIR): $SOFTWARE_DIR" \n' |
| 1061 |
|
txt += 'echo ">>> current directory content:"\n' |
| 1062 |
< |
if self.debug_wrapper: |
| 1062 |
> |
if self.debug_wrapper: |
| 1063 |
|
txt += 'ls -Al\n' |
| 1064 |
|
txt += '\n' |
| 1065 |
|
|
| 1091 |
|
txt += '\n' |
| 1092 |
|
txt += 'echo ">>> current directory (SOFTWARE_DIR): $SOFTWARE_DIR" \n' |
| 1093 |
|
txt += 'echo ">>> current directory content:"\n' |
| 1094 |
< |
if self.debug_wrapper: |
| 1094 |
> |
if self.debug_wrapper: |
| 1095 |
|
txt += 'ls -Al\n' |
| 1096 |
|
txt += '\n' |
| 1097 |
|
txt += 'cd $RUNTIME_AREA\n' |
| 1113 |
|
'", other.GlueHostApplicationSoftwareRunTimeEnvironment)' |
| 1114 |
|
|
| 1115 |
|
req = req + ' && (other.GlueHostNetworkAdapterOutboundIP)' |
| 1116 |
< |
if common.scheduler.name() == "glitecoll": |
| 1116 |
> |
if ( common.scheduler.name() == "glitecoll" ) or ( common.scheduler.name() == "glite"): |
| 1117 |
|
req += ' && other.GlueCEStateStatus == "Production" ' |
| 1118 |
|
|
| 1119 |
|
return req |
| 1192 |
|
publish_data = int(self.cfg_params.get('USER.publish_data',0)) |
| 1193 |
|
if (publish_data == 1): |
| 1194 |
|
processedDataset = self.cfg_params['USER.publish_data_name'] |
| 1195 |
< |
LFNBaseName = LFNBase(processedDataset) |
| 1195 |
> |
if (self.primaryDataset == 'null'): |
| 1196 |
> |
self.primaryDataset = processedDataset |
| 1197 |
> |
if (common.scheduler.name().upper() == "CAF" or common.scheduler.name().upper() == "LSF"): |
| 1198 |
> |
### FEDE FOR NEW LFN ### |
| 1199 |
> |
LFNBaseName = LFNBase(self.primaryDataset, processedDataset, LocalUser=True) |
| 1200 |
> |
self.user = getUserName(LocalUser=True) |
| 1201 |
> |
######################## |
| 1202 |
> |
else : |
| 1203 |
> |
### FEDE FOR NEW LFN ### |
| 1204 |
> |
LFNBaseName = LFNBase(self.primaryDataset, processedDataset) |
| 1205 |
> |
self.user = getUserName() |
| 1206 |
> |
######################## |
| 1207 |
|
|
| 1208 |
|
txt += 'if [ $copy_exit_status -eq 0 ]; then\n' |
| 1209 |
< |
txt += ' FOR_LFN=%s_${PSETHASH}/\n'%(LFNBaseName) |
| 1209 |
> |
### FEDE FOR NEW LFN ### |
| 1210 |
> |
#txt += ' FOR_LFN=%s_${PSETHASH}/\n'%(LFNBaseName) |
| 1211 |
> |
txt += ' FOR_LFN=%s/${PSETHASH}/\n'%(LFNBaseName) |
| 1212 |
> |
######################## |
| 1213 |
|
txt += 'else\n' |
| 1214 |
|
txt += ' FOR_LFN=/copy_problems/ \n' |
| 1215 |
|
txt += ' SE=""\n' |
| 1217 |
|
txt += 'fi\n' |
| 1218 |
|
|
| 1219 |
|
txt += 'echo ">>> Modify Job Report:" \n' |
| 1220 |
< |
txt += 'chmod a+x $SOFTWARE_DIR/ProdCommon/ProdCommon/FwkJobRep/ModifyJobReport.py\n' |
| 1220 |
> |
txt += 'chmod a+x $RUNTIME_AREA/ProdCommon/FwkJobRep/ModifyJobReport.py\n' |
| 1221 |
|
txt += 'ProcessedDataset='+processedDataset+'\n' |
| 1222 |
|
txt += 'echo "ProcessedDataset = $ProcessedDataset"\n' |
| 1223 |
|
txt += 'echo "SE = $SE"\n' |
| 1224 |
|
txt += 'echo "SE_PATH = $SE_PATH"\n' |
| 1225 |
|
txt += 'echo "FOR_LFN = $FOR_LFN" \n' |
| 1226 |
|
txt += 'echo "CMSSW_VERSION = $CMSSW_VERSION"\n\n' |
| 1227 |
< |
txt += 'echo "$SOFTWARE_DIR/ProdCommon/ProdCommon/FwkJobRep/ModifyJobReport.py $RUNTIME_AREA/crab_fjr_$NJob.xml $NJob $FOR_LFN $PrimaryDataset $DataTier $ProcessedDataset $ApplicationFamily $executable $CMSSW_VERSION $PSETHASH $SE $SE_PATH"\n' |
| 1228 |
< |
txt += '$SOFTWARE_DIR/ProdCommon/ProdCommon/FwkJobRep/ModifyJobReport.py $RUNTIME_AREA/crab_fjr_$NJob.xml $NJob $FOR_LFN $PrimaryDataset $DataTier $ProcessedDataset $ApplicationFamily $executable $CMSSW_VERSION $PSETHASH $SE $SE_PATH\n' |
| 1227 |
> |
### FEDE FOR NEW LFN ### |
| 1228 |
> |
txt += 'echo "$RUNTIME_AREA/ProdCommon/FwkJobRep/ModifyJobReport.py $RUNTIME_AREA/crab_fjr_$NJob.xml $NJob $FOR_LFN $PrimaryDataset $DataTier ' + self.user + '-$ProcessedDataset-$PSETHASH $ApplicationFamily $executable $CMSSW_VERSION $PSETHASH $SE $SE_PATH"\n' |
| 1229 |
> |
txt += '$RUNTIME_AREA/ProdCommon/FwkJobRep/ModifyJobReport.py $RUNTIME_AREA/crab_fjr_$NJob.xml $NJob $FOR_LFN $PrimaryDataset $DataTier ' + self.user + '-$ProcessedDataset-$PSETHASH $ApplicationFamily $executable $CMSSW_VERSION $PSETHASH $SE $SE_PATH\n' |
| 1230 |
> |
######################## |
| 1231 |
|
txt += 'modifyReport_result=$?\n' |
| 1232 |
|
txt += 'if [ $modifyReport_result -ne 0 ]; then\n' |
| 1233 |
|
txt += ' modifyReport_result=70500\n' |
| 1253 |
|
txt += ' executable_exit_status=`python $RUNTIME_AREA/parseCrabFjr.py --input $RUNTIME_AREA/crab_fjr_$NJob.xml --exitcode`\n' |
| 1254 |
|
txt += ' if [ $executable_exit_status -eq 50115 ];then\n' |
| 1255 |
|
txt += ' echo ">>> crab_fjr.xml contents: "\n' |
| 1256 |
< |
txt += ' cat $RUNTIME_AREA/crab_fjr_NJob.xml\n' |
| 1256 |
> |
txt += ' cat $RUNTIME_AREA/crab_fjr_$NJob.xml\n' |
| 1257 |
|
txt += ' echo "Wrong FrameworkJobReport --> does not contain useful info. ExitStatus: $executable_exit_status"\n' |
| 1258 |
|
txt += ' elif [ $executable_exit_status -eq -999 ];then\n' |
| 1259 |
|
txt += ' echo "ExitStatus from FrameworkJobReport not available. not available. Using exit code of executable from command line."\n' |
| 1265 |
|
txt += ' fi\n' |
| 1266 |
|
#### Patch to check input data reading for CMSSW16x Hopefully we-ll remove it asap |
| 1267 |
|
|
| 1268 |
< |
if self.datasetPath: |
| 1268 |
> |
if (self.datasetPath and not (self.dataset_pu or self.useParent) : |
| 1269 |
|
# VERIFY PROCESSED DATA |
| 1270 |
|
txt += ' if [ $executable_exit_status -eq 0 ];then\n' |
| 1271 |
|
txt += ' echo ">>> Verify list of processed files:"\n' |
