| 1 |
|
import common |
| 2 |
+ |
from sets import Set |
| 3 |
|
from crab_exceptions import * |
| 4 |
|
from crab_util import * |
| 5 |
+ |
|
| 6 |
+ |
from WMCore.DataStructs.File import File |
| 7 |
+ |
from WMCore.DataStructs.Fileset import Fileset |
| 8 |
+ |
from WMCore.DataStructs.Run import Run |
| 9 |
+ |
from WMCore.DataStructs.Subscription import Subscription |
| 10 |
+ |
from WMCore.DataStructs.Workflow import Workflow |
| 11 |
+ |
from WMCore.JobSplitting.SplitterFactory import SplitterFactory |
| 12 |
|
from WMCore.SiteScreening.BlackWhiteListParser import SEBlackWhiteListParser |
| 13 |
|
|
| 14 |
|
class JobSplitter: |
| 17 |
|
self.args=args |
| 18 |
|
#self.maxEvents |
| 19 |
|
# init BlackWhiteListParser |
| 20 |
< |
seWhiteList = cfg_params.get('GRID.se_white_list',[]) |
| 20 |
> |
self.seWhiteList = cfg_params.get('GRID.se_white_list',[]) |
| 21 |
|
seBlackList = cfg_params.get('GRID.se_black_list',[]) |
| 22 |
< |
self.blackWhiteListParser = SEBlackWhiteListParser(seWhiteList, seBlackList, common.logger) |
| 22 |
> |
self.blackWhiteListParser = SEBlackWhiteListParser(self.seWhiteList, seBlackList, common.logger()) |
| 23 |
|
|
| 24 |
|
|
| 25 |
|
def checkUserSettings(self): |
| 51 |
|
self.selectTotalNumberEvents = 0 |
| 52 |
|
|
| 53 |
|
|
| 54 |
+ |
def ComputeSubBlockSites( self, blockSites ): |
| 55 |
+ |
""" |
| 56 |
+ |
""" |
| 57 |
+ |
sub_blockSites = {} |
| 58 |
+ |
for k,v in blockSites.iteritems(): |
| 59 |
+ |
sites=self.blackWhiteListParser.checkWhiteList(v) |
| 60 |
+ |
if sites : sub_blockSites[k]=v |
| 61 |
+ |
if len(sub_blockSites) < 1: |
| 62 |
+ |
msg = 'WARNING: the sites %s is not hosting any part of data.'%self.seWhiteList |
| 63 |
+ |
raise CrabException(msg) |
| 64 |
+ |
return sub_blockSites |
| 65 |
+ |
|
| 66 |
|
######################################################################## |
| 67 |
|
def jobSplittingByEvent( self ): |
| 68 |
|
""" |
| 77 |
|
self.list_of_args - File(s) job will run on (a list of lists) |
| 78 |
|
""" |
| 79 |
|
|
| 80 |
< |
jobDestination=[] |
| 80 |
> |
jobDestination=[] |
| 81 |
|
self.checkUserSettings() |
| 82 |
|
if ( (self.selectTotalNumberEvents + self.selectEventsPerJob + self.selectNumberOfJobs) != 2 ): |
| 83 |
|
msg = 'Must define exactly two of total_number_of_events, events_per_job, or number_of_jobs.' |
| 84 |
|
raise CrabException(msg) |
| 85 |
< |
|
| 86 |
< |
blockSites = self.args['blockSites'] |
| 85 |
> |
|
| 86 |
> |
blockSites = self.args['blockSites'] |
| 87 |
|
pubdata = self.args['pubdata'] |
| 88 |
|
filesbyblock=pubdata.getFiles() |
| 89 |
|
|
| 97 |
|
self.useParent = int(self.cfg_params.get('CMSSW.use_parent',0)) |
| 98 |
|
noBboundary = int(self.cfg_params.get('CMSSW.no_block_boundary',0)) |
| 99 |
|
|
| 100 |
+ |
if noBboundary == 1: |
| 101 |
+ |
if self.total_number_of_events== -1: |
| 102 |
+ |
msg = 'You are selecting no_block_boundary=1 which does not allow to set total_number_of_events=-1\n' |
| 103 |
+ |
msg +='\tYou shoud get the number of event from DBS web interface and use it for your configuration.' |
| 104 |
+ |
raise CrabException(msg) |
| 105 |
+ |
if len(self.seWhiteList) == 0 or len(self.seWhiteList.split(',')) != 1: |
| 106 |
+ |
msg = 'You are selecting no_block_boundary=1 which requires to choose one and only one site.\n' |
| 107 |
+ |
msg += "\tPlease set se_white_list with the site's storage element name." |
| 108 |
+ |
raise CrabException(msg) |
| 109 |
+ |
blockSites = self.ComputeSubBlockSites(blockSites) |
| 110 |
+ |
|
| 111 |
|
# ---- Handle the possible job splitting configurations ---- # |
| 112 |
|
if (self.selectTotalNumberEvents): |
| 113 |
|
totalEventsRequested = self.total_number_of_events |
| 157 |
|
parString = "" |
| 158 |
|
pString = "" |
| 159 |
|
filesEventCount = 0 |
| 160 |
+ |
msg='' |
| 161 |
|
|
| 162 |
|
# ---- Iterate over the blocks in the dataset until ---- # |
| 163 |
|
# ---- we've met the requested total # of events ---- # |
| 229 |
|
list_of_lists.append([fullString,str(-1),str(jobSkipEventCount)]) |
| 230 |
|
msg += "Job %s can run over %s events (last file in block).\n"%(str(jobCount+1), str(filesEventCount - jobSkipEventCount)) |
| 231 |
|
jobDestination.append(blockSites[block]) |
| 232 |
< |
msg += "Job %s Destination: %s\n"%(str(jobCount+1),str(jobDestination[jobCount])) |
| 232 |
> |
msg += "Job %s Destination: %s\n"%(str(jobCount+1),str(SE2CMS(jobDestination[jobCount]))) |
| 233 |
|
# fill jobs of block dictionary |
| 234 |
|
jobsOfBlock[block].append(jobCount+1) |
| 235 |
|
# reset counter |
| 258 |
|
list_of_lists.append([fullString,str(eventsPerJobRequested),str(jobSkipEventCount)]) |
| 259 |
|
msg += "Job %s can run over %s events.\n"%(str(jobCount+1),str(eventsPerJobRequested)) |
| 260 |
|
jobDestination.append(blockSites[block]) |
| 261 |
< |
msg+= "Job %s Destination: %s\n"%(str(jobCount+1),str(jobDestination[jobCount])) |
| 261 |
> |
msg+= "Job %s Destination: %s\n"%(str(jobCount+1),str(SE2CMS(jobDestination[jobCount]))) |
| 262 |
|
jobsOfBlock[block].append(jobCount+1) |
| 263 |
|
# reset counter |
| 264 |
|
jobCount = jobCount + 1 |
| 283 |
|
list_of_lists.append([fullString,str(eventsPerJobRequested),str(jobSkipEventCount)]) |
| 284 |
|
msg += "Job %s can run over %s events.\n"%(str(jobCount+1),str(eventsPerJobRequested)) |
| 285 |
|
jobDestination.append(blockSites[block]) |
| 286 |
< |
msg+= "Job %s Destination: %s\n"%(str(jobCount+1),str(jobDestination[jobCount])) |
| 286 |
> |
msg+= "Job %s Destination: %s\n"%(str(jobCount+1),str(SE2CMS(jobDestination[jobCount]))) |
| 287 |
|
jobsOfBlock[block].append(jobCount+1) |
| 288 |
|
# increase counter |
| 289 |
|
jobCount = jobCount + 1 |
| 297 |
|
pString_tmp='' |
| 298 |
|
if self.useParent==1: |
| 299 |
|
for f in parent : pString_tmp += f + ',' |
| 300 |
< |
pString = pString_tmp |
| 300 |
> |
pString = pString_tmp |
| 301 |
|
parString = file + ',' |
| 302 |
|
pass # END if |
| 303 |
|
pass # END while (iterate over files in the block) |
| 307 |
|
if (eventsRemaining > 0 and jobCount < totalNumberOfJobs ): |
| 308 |
|
common.logger.info("Could not run on all requested events because some blocks not hosted at allowed sites.") |
| 309 |
|
common.logger.info(str(jobCount)+" job(s) can run on "+str(totalEventCount)+" events.\n") |
| 310 |
< |
|
| 310 |
> |
|
| 311 |
|
# skip check on block with no sites DD |
| 312 |
|
if noBboundary == 0 : self.checkBlockNoSite(blocks,jobsOfBlock,blockSites) |
| 313 |
|
|
| 323 |
|
|
| 324 |
|
# keep trace of block with no sites to print a warning at the end |
| 325 |
|
|
| 326 |
< |
def checkBlockNoSite(self,blocks,jobsOfBlock,blockSites): |
| 326 |
> |
def checkBlockNoSite(self,blocks,jobsOfBlock,blockSites): |
| 327 |
|
# screen output |
| 328 |
|
screenOutput = "List of jobs and available destination sites:\n\n" |
| 329 |
|
noSiteBlock = [] |
| 335 |
|
if block in jobsOfBlock.keys() : |
| 336 |
|
blockCounter += 1 |
| 337 |
|
allBlock.append( blockCounter ) |
| 338 |
+ |
sites=self.blackWhiteListParser.checkWhiteList(self.blackWhiteListParser.checkBlackList(blockSites[block],[block]),[block]) |
| 339 |
|
screenOutput += "Block %5i: jobs %20s: sites: %s\n" % (blockCounter,spanRanges(jobsOfBlock[block]), |
| 340 |
< |
','.join(self.blackWhiteListParser.checkWhiteList(self.blackWhiteListParser.checkBlackList(blockSites[block],[block]),[block]))) |
| 341 |
< |
if len(self.blackWhiteListParser.checkWhiteList(self.blackWhiteListParser.checkBlackList(blockSites[block],[block]),[block])) == 0: |
| 340 |
> |
', '.join(SE2CMS(sites))) |
| 341 |
> |
if len(sites) == 0: |
| 342 |
|
noSiteBlock.append( spanRanges(jobsOfBlock[block]) ) |
| 343 |
|
bloskNoSite.append( blockCounter ) |
| 344 |
|
|
| 350 |
|
virgola = "," |
| 351 |
|
for block in bloskNoSite: |
| 352 |
|
msg += ' ' + str(block) + virgola |
| 353 |
< |
msg += '\n Related jobs:\n ' |
| 353 |
> |
msg += '\n\t\tRelated jobs:\n ' |
| 354 |
|
virgola = "" |
| 355 |
|
if len(noSiteBlock) > 1: |
| 356 |
|
virgola = "," |
| 357 |
|
for range_jobs in noSiteBlock: |
| 358 |
|
msg += str(range_jobs) + virgola |
| 359 |
< |
msg += '\n will not be submitted and this block of data can not be analyzed!\n' |
| 359 |
> |
msg += '\n\t\twill not be submitted and this block of data can not be analyzed!\n' |
| 360 |
|
if self.cfg_params.has_key('GRID.se_white_list'): |
| 361 |
< |
msg += 'WARNING: SE White List: '+self.cfg_params['GRID.se_white_list']+'\n' |
| 362 |
< |
msg += '(Hint: By whitelisting you force the job to run at this particular site(s).\n' |
| 363 |
< |
msg += 'Please check if the dataset is available at this site!)\n' |
| 361 |
> |
msg += '\tWARNING: SE White List: '+self.cfg_params['GRID.se_white_list']+'\n' |
| 362 |
> |
msg += '\t(Hint: By whitelisting you force the job to run at this particular site(s).\n' |
| 363 |
> |
msg += '\tPlease check if the dataset is available at this site!)' |
| 364 |
|
if self.cfg_params.has_key('GRID.ce_white_list'): |
| 365 |
< |
msg += 'WARNING: CE White List: '+self.cfg_params['GRID.ce_white_list']+'\n' |
| 366 |
< |
msg += '(Hint: By whitelisting you force the job to run at this particular site(s).\n' |
| 367 |
< |
msg += 'Please check if the dataset is available at this site!)\n' |
| 365 |
> |
msg += '\tWARNING: CE White List: '+self.cfg_params['GRID.ce_white_list']+'\n' |
| 366 |
> |
msg += '\t(Hint: By whitelisting you force the job to run at this particular site(s).\n' |
| 367 |
> |
msg += '\tPlease check if the dataset is available at this site!)\n' |
| 368 |
|
|
| 369 |
|
common.logger.info(msg) |
| 370 |
|
|
| 371 |
|
if bloskNoSite == allBlock: |
| 372 |
< |
raise CrabException('No jobs created') |
| 372 |
> |
raise CrabException('No jobs created') |
| 373 |
|
|
| 374 |
|
return |
| 375 |
|
|
| 376 |
|
|
| 377 |
|
######################################################################## |
| 378 |
< |
def jobSplittingByRun(self): |
| 378 |
> |
def jobSplittingByRun(self): |
| 379 |
|
""" |
| 380 |
|
""" |
| 348 |
– |
from sets import Set |
| 349 |
– |
from WMCore.JobSplitting.RunBased import RunBased |
| 350 |
– |
from WMCore.DataStructs.Workflow import Workflow |
| 351 |
– |
from WMCore.DataStructs.File import File |
| 352 |
– |
from WMCore.DataStructs.Fileset import Fileset |
| 353 |
– |
from WMCore.DataStructs.Subscription import Subscription |
| 354 |
– |
from WMCore.JobSplitting.SplitterFactory import SplitterFactory |
| 355 |
– |
from WMCore.DataStructs.Run import Run |
| 381 |
|
|
| 382 |
|
self.checkUserSettings() |
| 383 |
< |
blockSites = self.args['blockSites'] |
| 383 |
> |
blockSites = self.args['blockSites'] |
| 384 |
|
pubdata = self.args['pubdata'] |
| 385 |
|
|
| 386 |
|
if self.selectNumberOfJobs == 0 : |
| 387 |
|
self.theNumberOfJobs = 9999999 |
| 388 |
|
blocks = {} |
| 389 |
< |
runList = [] |
| 389 |
> |
runList = [] |
| 390 |
|
thefiles = Fileset(name='FilesToSplit') |
| 391 |
|
fileList = pubdata.getListFiles() |
| 392 |
|
for f in fileList: |
| 393 |
|
block = f['Block']['Name'] |
| 394 |
< |
try: |
| 394 |
> |
try: |
| 395 |
|
f['Block']['StorageElementList'].extend(blockSites[block]) |
| 396 |
|
except: |
| 397 |
|
continue |
| 399 |
|
[ wmbsFile['locations'].add(x) for x in blockSites[block] ] |
| 400 |
|
wmbsFile['block'] = block |
| 401 |
|
runNum = f['RunsList'][0]['RunNumber'] |
| 402 |
< |
runList.append(runNum) |
| 402 |
> |
runList.append(runNum) |
| 403 |
|
myRun = Run(runNumber=runNum) |
| 404 |
|
wmbsFile.addRun( myRun ) |
| 405 |
|
thefiles.addFile( |
| 406 |
|
wmbsFile |
| 407 |
|
) |
| 408 |
< |
|
| 408 |
> |
|
| 409 |
|
work = Workflow() |
| 410 |
|
subs = Subscription( |
| 411 |
|
fileset = thefiles, |
| 414 |
|
type = "Processing") |
| 415 |
|
splitter = SplitterFactory() |
| 416 |
|
jobfactory = splitter(subs) |
| 417 |
< |
|
| 418 |
< |
#loop over all runs |
| 417 |
> |
|
| 418 |
> |
#loop over all runs |
| 419 |
|
set = Set(runList) |
| 420 |
|
list_of_lists = [] |
| 421 |
|
jobDestination = [] |
| 423 |
|
for jobGroup in jobfactory(): |
| 424 |
|
if count < self.theNumberOfJobs: |
| 425 |
|
res = self.getJobInfo(jobGroup) |
| 426 |
< |
parString = '' |
| 426 |
> |
parString = '' |
| 427 |
|
for file in res['lfns']: |
| 428 |
|
parString += file + ',' |
| 429 |
|
fullString = parString[:-1] |
| 430 |
< |
list_of_lists.append([fullString,str(-1),str(0)]) |
| 430 |
> |
list_of_lists.append([fullString,str(-1),str(0)]) |
| 431 |
|
#need to check single file location |
| 432 |
< |
jobDestination.append(res['locations']) |
| 432 |
> |
jobDestination.append(res['locations']) |
| 433 |
|
count +=1 |
| 434 |
|
# prepare dict output |
| 435 |
|
dictOut = {} |
| 442 |
|
|
| 443 |
|
def getJobInfo( self,jobGroup ): |
| 444 |
|
res = {} |
| 445 |
< |
lfns = [] |
| 446 |
< |
locations = [] |
| 445 |
> |
lfns = [] |
| 446 |
> |
locations = [] |
| 447 |
|
tmp_check=0 |
| 448 |
|
for job in jobGroup.jobs: |
| 449 |
|
for file in job.getFiles(): |
| 450 |
< |
lfns.append(file['lfn']) |
| 450 |
> |
lfns.append(file['lfn']) |
| 451 |
|
for loc in file['locations']: |
| 452 |
|
if tmp_check < 1 : |
| 453 |
|
locations.append(loc) |
| 454 |
< |
tmp_check = tmp_check + 1 |
| 455 |
< |
### qui va messo il check per la locations |
| 456 |
< |
res['lfns'] = lfns |
| 457 |
< |
res['locations'] = locations |
| 458 |
< |
return res |
| 459 |
< |
|
| 454 |
> |
tmp_check = tmp_check + 1 |
| 455 |
> |
### qui va messo il check per la locations |
| 456 |
> |
res['lfns'] = lfns |
| 457 |
> |
res['locations'] = locations |
| 458 |
> |
return res |
| 459 |
> |
|
| 460 |
|
######################################################################## |
| 461 |
< |
def jobSplittingNoInput(self): |
| 461 |
> |
def prepareSplittingNoInput(self): |
| 462 |
|
""" |
| 438 |
– |
Perform job splitting based on number of event per job |
| 463 |
|
""" |
| 440 |
– |
common.logger.debug('Splitting per events') |
| 441 |
– |
self.checkUserSettings() |
| 442 |
– |
jobDestination=[] |
| 443 |
– |
if ( (self.selectTotalNumberEvents + self.selectEventsPerJob + self.selectNumberOfJobs) != 2 ): |
| 444 |
– |
msg = 'Must define exactly two of total_number_of_events, events_per_job, or number_of_jobs.' |
| 445 |
– |
raise CrabException(msg) |
| 446 |
– |
|
| 447 |
– |
managedGenerators =self.args['managedGenerators'] |
| 448 |
– |
generator = self.args['generator'] |
| 449 |
– |
firstRun = self.cfg_params.get('CMSSW.first_run',None) |
| 450 |
– |
|
| 464 |
|
if (self.selectEventsPerJob): |
| 465 |
|
common.logger.info('Required '+str(self.eventsPerJob)+' events per job ') |
| 466 |
|
if (self.selectNumberOfJobs): |
| 483 |
|
self.total_number_of_jobs = self.theNumberOfJobs |
| 484 |
|
self.eventsPerJob = int(self.total_number_of_events/self.total_number_of_jobs) |
| 485 |
|
|
| 486 |
+ |
|
| 487 |
+ |
def jobSplittingNoInput(self): |
| 488 |
+ |
""" |
| 489 |
+ |
Perform job splitting based on number of event per job |
| 490 |
+ |
""" |
| 491 |
+ |
common.logger.debug('Splitting per events') |
| 492 |
+ |
self.checkUserSettings() |
| 493 |
+ |
jobDestination=[] |
| 494 |
+ |
if ( (self.selectTotalNumberEvents + self.selectEventsPerJob + self.selectNumberOfJobs) != 2 ): |
| 495 |
+ |
msg = 'Must define exactly two of total_number_of_events, events_per_job, or number_of_jobs.' |
| 496 |
+ |
raise CrabException(msg) |
| 497 |
+ |
|
| 498 |
+ |
managedGenerators =self.args['managedGenerators'] |
| 499 |
+ |
generator = self.args['generator'] |
| 500 |
+ |
firstRun = self.cfg_params.get('CMSSW.first_run',None) |
| 501 |
+ |
|
| 502 |
+ |
self.prepareSplittingNoInput() |
| 503 |
+ |
|
| 504 |
|
common.logger.debug('N jobs '+str(self.total_number_of_jobs)) |
| 505 |
|
|
| 506 |
|
# is there any remainder? |
| 522 |
|
## pythia first run |
| 523 |
|
args.append(str(firstRun)+str(i)) |
| 524 |
|
if (generator in managedGenerators): |
| 525 |
< |
if (generator == 'comphep' and i == 0): |
| 525 |
> |
args.append(generator) |
| 526 |
> |
if (generator == 'comphep' and i == 0): |
| 527 |
|
# COMPHEP is brain-dead and wants event #'s like 1,100,200,300 |
| 528 |
|
args.append('1') |
| 529 |
< |
else: |
| 529 |
> |
else: |
| 530 |
|
args.append(str(i*self.eventsPerJob)) |
| 531 |
|
args.append(str(self.eventsPerJob)) |
| 532 |
|
self.list_of_args.append(args) |
| 536 |
|
dictOut['params'] = ['MaxEvents'] |
| 537 |
|
if (firstRun): |
| 538 |
|
dictOut['params'] = ['FirstRun','MaxEvents'] |
| 539 |
< |
if ( generator in managedGenerators ) : dictOut['params'] = ['FirstRun', 'FirstEvent', 'MaxEvents'] |
| 540 |
< |
else: |
| 541 |
< |
if (generator in managedGenerators) : dictOut['params'] = ['FirstEvent', 'MaxEvents'] |
| 539 |
> |
if ( generator in managedGenerators ) : |
| 540 |
> |
dictOut['params'] = ['FirstRun', 'Generator', 'FirstEvent', 'MaxEvents'] |
| 541 |
> |
else: |
| 542 |
> |
if (generator in managedGenerators) : |
| 543 |
> |
dictOut['params'] = ['Generator', 'FirstEvent', 'MaxEvents'] |
| 544 |
|
dictOut['args'] = self.list_of_args |
| 545 |
|
dictOut['jobDestination'] = jobDestination |
| 546 |
|
dictOut['njobs']=self.total_number_of_jobs |
| 560 |
|
common.logger.debug('Splitting per job') |
| 561 |
|
common.logger.info('Required '+str(self.theNumberOfJobs)+' jobs in total ') |
| 562 |
|
|
| 563 |
< |
self.total_number_of_jobs = self.theNumberOfJobs |
| 563 |
> |
# self.total_number_of_jobs = self.theNumberOfJobs |
| 564 |
> |
|
| 565 |
> |
self.prepareSplittingNoInput() |
| 566 |
|
|
| 567 |
|
common.logger.debug('N jobs '+str(self.total_number_of_jobs)) |
| 568 |
|
|
| 569 |
|
common.logger.info(str(self.total_number_of_jobs)+' jobs can be created') |
| 570 |
|
|
| 571 |
|
# argument is seed number.$i |
| 572 |
< |
#self.list_of_args = [] |
| 572 |
> |
self.list_of_args = [] |
| 573 |
|
for i in range(self.total_number_of_jobs): |
| 574 |
+ |
args=[] |
| 575 |
|
jobDestination.append([""]) |
| 576 |
< |
# self.list_of_args.append([str(i)]) |
| 576 |
> |
if self.eventsPerJob != 0 : |
| 577 |
> |
args.append(str(self.eventsPerJob)) |
| 578 |
> |
self.list_of_args.append(args) |
| 579 |
|
|
| 580 |
|
# prepare dict output |
| 581 |
|
dictOut = {} |
| 582 |
< |
dictOut['args'] = [] # self.list_of_args |
| 582 |
> |
dictOut['params'] = ['MaxEvents'] |
| 583 |
> |
dictOut['args'] = self.list_of_args |
| 584 |
|
dictOut['jobDestination'] = jobDestination |
| 585 |
|
dictOut['njobs']=self.total_number_of_jobs |
| 586 |
|
return dictOut |
| 547 |
– |
|
| 587 |
|
|
| 588 |
< |
def jobSplittingByLumi(self): |
| 588 |
> |
|
| 589 |
> |
def jobSplittingByLumi(self): |
| 590 |
|
""" |
| 591 |
+ |
Split task into jobs by Lumi section paying attention to which |
| 592 |
+ |
lumis should be run (according to the analysis dataset). |
| 593 |
+ |
This uses WMBS job splitting which does not split files over jobs |
| 594 |
+ |
so the job will have AT LEAST as many lumis as requested, perhaps |
| 595 |
+ |
more |
| 596 |
|
""" |
| 597 |
< |
return |
| 597 |
> |
|
| 598 |
> |
common.logger.debug('Splitting by Lumi') |
| 599 |
> |
self.checkUserSettings() # FIXME need one for lumis |
| 600 |
> |
|
| 601 |
> |
blockSites = self.args['blockSites'] |
| 602 |
> |
pubdata = self.args['pubdata'] |
| 603 |
> |
|
| 604 |
> |
if self.selectNumberOfJobs == 0 : |
| 605 |
> |
self.theNumberOfJobs = 9999999 |
| 606 |
> |
lumisPerFile = pubdata.getLumis() |
| 607 |
> |
|
| 608 |
> |
# Make the list of WMBS files for job splitter |
| 609 |
> |
fileList = pubdata.getListFiles() |
| 610 |
> |
thefiles = Fileset(name='FilesToSplit') |
| 611 |
> |
for jobFile in fileList: |
| 612 |
> |
block = jobFile['Block']['Name'] |
| 613 |
> |
try: |
| 614 |
> |
jobFile['Block']['StorageElementList'].extend(blockSites[block]) |
| 615 |
> |
except: |
| 616 |
> |
continue |
| 617 |
> |
wmbsFile = File(jobFile['LogicalFileName']) |
| 618 |
> |
[ wmbsFile['locations'].add(x) for x in blockSites[block] ] |
| 619 |
> |
wmbsFile['block'] = block |
| 620 |
> |
for lumi in lumisPerFile[jobFile['LogicalFileName']]: |
| 621 |
> |
wmbsFile.addRun(Run(lumi[0], lumi[1])) |
| 622 |
> |
thefiles.addFile(wmbsFile) |
| 623 |
> |
|
| 624 |
> |
work = Workflow() |
| 625 |
> |
subs = Subscription(fileset = thefiles, workflow = work, |
| 626 |
> |
split_algo = 'LumiBased', type = "Processing") |
| 627 |
> |
splitter = SplitterFactory() |
| 628 |
> |
jobFactory = splitter(subs) |
| 629 |
> |
|
| 630 |
> |
list_of_lists = [] |
| 631 |
> |
jobDestination = [] |
| 632 |
> |
jobCount = 0 |
| 633 |
> |
self.theNumberOfJobs = 20 #FIXME |
| 634 |
> |
for jobGroup in jobFactory(lumis_per_job = 50): #FIXME |
| 635 |
> |
for job in jobGroup.jobs: |
| 636 |
> |
if jobCount < self.theNumberOfJobs: |
| 637 |
> |
lumis = [] |
| 638 |
> |
lfns = [] |
| 639 |
> |
locations = [] |
| 640 |
> |
firstFile = True |
| 641 |
> |
# Collect information from all the files |
| 642 |
> |
for jobFile in job.getFiles(): |
| 643 |
> |
if firstFile: # Get locations from first file in the job |
| 644 |
> |
for loc in jobFile['locations']: |
| 645 |
> |
locations.append(loc) |
| 646 |
> |
firstFile = False |
| 647 |
> |
# Accumulate Lumis from all files |
| 648 |
> |
for lumiList in jobFile['runs']: |
| 649 |
> |
theRun = lumiList.run |
| 650 |
> |
for theLumi in list(lumiList): |
| 651 |
> |
lumis.append( (theRun, theLumi) ) |
| 652 |
> |
|
| 653 |
> |
lfns.append(jobFile['lfn']) |
| 654 |
> |
fileString = ','.join(lfns) |
| 655 |
> |
lumiString = compressLumiString(lumis) |
| 656 |
> |
common.logger.debug('Job %s will run on %s files and %s lumis ' |
| 657 |
> |
% (jobCount, len(lfns), len(lumis) )) |
| 658 |
> |
list_of_lists.append([fileString, str(-1), str(0), lumiString]) |
| 659 |
> |
|
| 660 |
> |
jobDestination.append(locations) |
| 661 |
> |
jobCount += 1 |
| 662 |
> |
# Prepare dict output matching back to non-WMBS job creation |
| 663 |
> |
dictOut = {} |
| 664 |
> |
dictOut['params'] = ['InputFiles', 'MaxEvents', 'SkipEvents', 'Lumis'] |
| 665 |
> |
dictOut['args'] = list_of_lists |
| 666 |
> |
dictOut['jobDestination'] = jobDestination |
| 667 |
> |
dictOut['njobs'] = jobCount |
| 668 |
> |
|
| 669 |
> |
return dictOut |
| 670 |
> |
|
| 671 |
> |
|
| 672 |
|
def Algos(self): |
| 673 |
|
""" |
| 674 |
|
Define key splittingType matrix |
| 675 |
|
""" |
| 676 |
< |
SplitAlogs = { |
| 677 |
< |
'EventBased' : self.jobSplittingByEvent, |
| 676 |
> |
SplitAlogs = { |
| 677 |
> |
'EventBased' : self.jobSplittingByEvent, |
| 678 |
|
'RunBased' : self.jobSplittingByRun, |
| 679 |
< |
'LumiBased' : self.jobSplittingByLumi, |
| 680 |
< |
'NoInput' : self.jobSplittingNoInput, |
| 679 |
> |
'LumiBased' : self.jobSplittingByLumi, |
| 680 |
> |
'NoInput' : self.jobSplittingNoInput, |
| 681 |
|
'ForScript' : self.jobSplittingForScript |
| 682 |
< |
} |
| 682 |
> |
} |
| 683 |
|
return SplitAlogs |
| 684 |
|
|
| 685 |
+ |
|
| 686 |
+ |
|
| 687 |
+ |
def compressLumiString(lumis): |
| 688 |
+ |
""" |
| 689 |
+ |
Turn a list of 2-tuples of run/lumi numbers into a list of the format |
| 690 |
+ |
R1:L1,R2:L2-R3:L3 which is acceptable to CMSSW LumiBlockRange variable |
| 691 |
+ |
""" |
| 692 |
+ |
|
| 693 |
+ |
lumis.sort() |
| 694 |
+ |
parts = [] |
| 695 |
+ |
startRange = None |
| 696 |
+ |
endRange = None |
| 697 |
+ |
|
| 698 |
+ |
for lumiBlock in lumis: |
| 699 |
+ |
if not startRange: # This is the first one |
| 700 |
+ |
startRange = lumiBlock |
| 701 |
+ |
endRange = lumiBlock |
| 702 |
+ |
elif lumiBlock == endRange: # Same Lumi (different files?) |
| 703 |
+ |
pass |
| 704 |
+ |
elif lumiBlock[0] == endRange[0] and lumiBlock[1] == endRange[1] + 1: # This is a continuation |
| 705 |
+ |
endRange = lumiBlock |
| 706 |
+ |
else: # This is the start of a new range |
| 707 |
+ |
part = ':'.join(map(str, startRange)) |
| 708 |
+ |
if startRange != endRange: |
| 709 |
+ |
part += '-' + ':'.join(map(str, endRange)) |
| 710 |
+ |
parts.append(part) |
| 711 |
+ |
startRange = lumiBlock |
| 712 |
+ |
endRange = lumiBlock |
| 713 |
+ |
|
| 714 |
+ |
# Put out what's left |
| 715 |
+ |
if startRange: |
| 716 |
+ |
part = ':'.join(map(str, startRange)) |
| 717 |
+ |
if startRange != endRange: |
| 718 |
+ |
part += '-' + ':'.join(map(str, endRange)) |
| 719 |
+ |
parts.append(part) |
| 720 |
+ |
|
| 721 |
+ |
output = ','.join(parts) |
| 722 |
+ |
return output |
| 723 |
+ |
|
| 724 |
+ |
|
