Python JobSpec.addFile方法代码示例

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    def defineEvgen16Job(self, i):
        """Define an Evgen16 job based on predefined values and randomly generated names
        """

        job = JobSpec()
        job.computingSite = self.__site
        job.cloud = self.__cloud

        job.jobDefinitionID = int(time.time()) % 10000
        job.jobName = "%s_%d" % (uuid.uuid1(), i)
        job.AtlasRelease = 'Atlas-16.6.2'
        job.homepackage = 'AtlasProduction/16.6.2.1'
        job.transformation = 'Evgen_trf.py'
        job.destinationDBlock = self.__datasetName
        job.destinationSE = self.__destName
        job.currentPriority = 10000
        job.prodSourceLabel = 'test'
        job.cmtConfig = 'i686-slc5-gcc43-opt'

        #Output file
        fileO = FileSpec()
        fileO.lfn = "%s.evgen.pool.root" % job.jobName
        fileO.destinationDBlock = job.destinationDBlock
        fileO.destinationSE = job.destinationSE
        fileO.dataset = job.destinationDBlock
        fileO.destinationDBlockToken = 'ATLASDATADISK'
        fileO.type = 'output'
        job.addFile(fileO)

        #Log file
        fileL = FileSpec()
        fileL.lfn = "%s.job.log.tgz" % job.jobName
        fileL.destinationDBlock = job.destinationDBlock
        fileL.destinationSE = job.destinationSE
        fileL.dataset = job.destinationDBlock
        fileL.destinationDBlockToken = 'ATLASDATADISK'
        fileL.type = 'log'
        job.addFile(fileL)

        job.jobParameters = "2760 105048 19901 101 200 MC10.105048.PythiaB_ccmu3mu1X.py %s NONE NONE NONE MC10JobOpts-latest-test.tar.gz" % fileO.lfn
        return job

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    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]
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
 def doBrokerage(self,inputList,vo,prodSourceLabel,workQueue):
     # list with a lock
     inputListWorld = ListWithLock([])
     # variables for submission
     maxBunchTask = 100
     # make logger
     tmpLog = MsgWrapper(logger)
     tmpLog.debug('start doBrokerage')
     # return for failure
     retFatal    = self.SC_FATAL
     retTmpError = self.SC_FAILED
     tmpLog.debug('vo={0} label={1} queue={2} nTasks={3}'.format(vo,prodSourceLabel,
                                                                 workQueue.queue_name,
                                                                 len(inputList)))
     # loop over all tasks
     allRwMap    = {}
     prioMap     = {}
     tt2Map      = {}
     expRWs      = {}
     jobSpecList = []
     for tmpJediTaskID,tmpInputList in inputList:
         for taskSpec,cloudName,inputChunk in tmpInputList:
             # collect tasks for WORLD
             if taskSpec.useWorldCloud():
                 inputListWorld.append((taskSpec,inputChunk))
                 continue
             # make JobSpec to be submitted for TaskAssigner
             jobSpec = JobSpec()
             jobSpec.taskID     = taskSpec.jediTaskID
             jobSpec.jediTaskID = taskSpec.jediTaskID
             # set managed to trigger TA
             jobSpec.prodSourceLabel  = 'managed'
             jobSpec.processingType   = taskSpec.processingType
             jobSpec.workingGroup     = taskSpec.workingGroup
             jobSpec.metadata         = taskSpec.processingType
             jobSpec.assignedPriority = taskSpec.taskPriority
             jobSpec.currentPriority  = taskSpec.currentPriority
             jobSpec.maxDiskCount     = (taskSpec.getOutDiskSize() + taskSpec.getWorkDiskSize()) / 1024 / 1024
             if taskSpec.useWorldCloud():
                 # use destinationSE to trigger task brokerage in WORLD cloud
                 jobSpec.destinationSE = taskSpec.cloud
             prodDBlock = None
             setProdDBlock = False
             for datasetSpec in inputChunk.getDatasets():
                 prodDBlock = datasetSpec.datasetName
                 if datasetSpec.isMaster():
                     jobSpec.prodDBlock = datasetSpec.datasetName
                     setProdDBlock = True
                 for fileSpec in datasetSpec.Files:
                     tmpInFileSpec = fileSpec.convertToJobFileSpec(datasetSpec)
                     jobSpec.addFile(tmpInFileSpec)
             # use secondary dataset name as prodDBlock
             if setProdDBlock == False and prodDBlock != None:
                 jobSpec.prodDBlock = prodDBlock
             # append
             jobSpecList.append(jobSpec)
             prioMap[jobSpec.taskID] = jobSpec.currentPriority
             tt2Map[jobSpec.taskID]  = jobSpec.processingType
             # get RW for a priority
             if not allRwMap.has_key(jobSpec.currentPriority):
                 tmpRW = self.taskBufferIF.calculateRWwithPrio_JEDI(vo,prodSourceLabel,workQueue,
                                                                    jobSpec.currentPriority) 
                 if tmpRW == None:
                     tmpLog.error('failed to calculate RW with prio={0}'.format(jobSpec.currentPriority))
                     return retTmpError
                 allRwMap[jobSpec.currentPriority] = tmpRW
             # get expected RW
             expRW = self.taskBufferIF.calculateTaskRW_JEDI(jobSpec.jediTaskID)
             if expRW == None:
                 tmpLog.error('failed to calculate RW for jediTaskID={0}'.format(jobSpec.jediTaskID))
                 return retTmpError
             expRWs[jobSpec.taskID] = expRW
     # for old clouds
     if jobSpecList != []:
         # get fullRWs
         fullRWs = self.taskBufferIF.calculateRWwithPrio_JEDI(vo,prodSourceLabel,None,None)
         if fullRWs == None:
             tmpLog.error('failed to calculate full RW')
             return retTmpError
         # set metadata
         for jobSpec in jobSpecList:
             rwValues = allRwMap[jobSpec.currentPriority]
             jobSpec.metadata = "%s;%s;%s;%s;%s;%s" % (jobSpec.metadata,
                                                       str(rwValues),str(expRWs),
                                                       str(prioMap),str(fullRWs),
                                                       str(tt2Map))
         tmpLog.debug('run task assigner for {0} tasks'.format(len(jobSpecList)))
         nBunchTask = 0
         while nBunchTask < len(jobSpecList):
             # get a bunch
             jobsBunch = jobSpecList[nBunchTask:nBunchTask+maxBunchTask]
             strIDs = 'jediTaskID='
             for tmpJobSpec in jobsBunch:
                 strIDs += '{0},'.format(tmpJobSpec.taskID)
             strIDs = strIDs[:-1]
             tmpLog.debug(strIDs)
             # increment index
             nBunchTask += maxBunchTask
             # run task brokerge
             stS,outSs = PandaClient.runTaskAssignment(jobsBunch)
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    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)
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]

#.........这里部分代码省略.........
##                     if cloudSite==eachCloud:
##                         cloud=cloudID
##                         outclouds.append(cloud)
##                         break
                    
##             print outclouds
##             # finally, matching with user's wishes
##             if len(outclouds)>0:
##                 if not job.backend.requirements.cloud: # no user wish, update
##                     job.backend.requirements.cloud=outclouds[0]
##                 else:
##                     try:
##                         assert job.backend.requirements.cloud in outclouds
##                     except:
##                         raise ApplicationConfigurationError(None,'Input dataset not available in target cloud %s. Please try any of the following %s' % (job.backend.requirements.cloud, str(outclouds)))
                                                            
        from GangaPanda.Lib.Panda.Panda import runPandaBrokerage
        
        runPandaBrokerage(job)
        
        if job.backend.site == 'AUTO':
            raise ApplicationConfigurationError(None,'site is still AUTO after brokerage!')

        # output dataset preparation and registration
        try:
            outDsLocation = Client.PandaSites[job.backend.site]['ddm']
        except:
            raise ApplicationConfigurationError(None,"Could not extract output dataset location from job.backend.site value: %s. Aborting" % job.backend.site)
        if not app.dryrun:
            for outtype in app.outputpaths.keys():
                dset=string.replace(app.outputpaths[outtype],"/",".")
                dset=dset[1:]
                # dataset registration must be done only once.
                print "registering output dataset %s at %s" % (dset,outDsLocation)
                try:
                    Client.addDataset(dset,False,location=outDsLocation)
                    dq2_set_dataset_lifetime(dset, location=outDsLocation)
                except:
                    raise ApplicationConfigurationError(None,"Fail to create output dataset %s. Aborting" % dset)
            # extend registration to build job lib dataset:
            print "registering output dataset %s at %s" % (self.libDataset,outDsLocation)

            try:
                Client.addDataset(self.libDataset,False,location=outDsLocation)
                dq2_set_dataset_lifetime(self.libDataset, outDsLocation)
            except:
                raise ApplicationConfigurationError(None,"Fail to create output dataset %s. Aborting" % self.libDataset)


        ###
        cacheVer = "-AtlasProduction_" + str(app.prod_release)
            
        logger.debug("master job submit?")
        self.outsite=job.backend.site
        if app.se_name and app.se_name != "none" and not self.outsite:
            self.outsite=app.se_name

       
        #       create build job
        jspec = JobSpec()
        jspec.jobDefinitionID   = job.id
        jspec.jobName           = commands.getoutput('uuidgen 2> /dev/null')
        jspec.AtlasRelease      = 'Atlas-%s' % app.atlas_rel
        jspec.homepackage       = 'AnalysisTransforms'+cacheVer#+nightVer
        jspec.transformation    = '%s/buildJob-00-00-03' % Client.baseURLSUB # common base to Athena and AthenaMC jobs: buildJob is a pilot job which takes care of all inputs for the real jobs (in prepare()
        jspec.destinationDBlock = self.libDataset
        jspec.destinationSE     = job.backend.site
        jspec.prodSourceLabel   = 'panda'
        jspec.assignedPriority  = 2000
        jspec.computingSite     = job.backend.site
        jspec.cloud             = job.backend.requirements.cloud
#        jspec.jobParameters     = self.args not known yet
        jspec.jobParameters     = '-o %s' % (self.library)
        if app.userarea:
            print app.userarea
            jspec.jobParameters     += ' -i %s' % (os.path.basename(app.userarea))
        else:
            jspec.jobParameters     += ' -i %s' % (sources)
        jspec.cmtConfig         = AthenaUtils.getCmtConfig(athenaVer=app.atlas_rel)
        
        matchURL = re.search('(http.*://[^/]+)/',Client.baseURLSSL)
        if matchURL:
            jspec.jobParameters += ' --sourceURL %s' % matchURL.group(1)

        fout = FileSpec()
        fout.lfn  = self.library
        fout.type = 'output'
        fout.dataset = self.libDataset
        fout.destinationDBlock = self.libDataset
        jspec.addFile(fout)

        flog = FileSpec()
        flog.lfn = '%s.log.tgz' % self.libDataset
        flog.type = 'log'
        flog.dataset = self.libDataset
        flog.destinationDBlock = self.libDataset
        jspec.addFile(flog)
        #print "MASTER JOB DETAILS:",jspec.jobParameters

        return jspec

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    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]
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
 def doBrokerage(self, inputList, vo, prodSourceLabel, workQueue):
     # variables for submission
     maxBunchTask = 100
     # make logger
     tmpLog = MsgWrapper(logger)
     tmpLog.debug("start doBrokerage")
     # return for failure
     retFatal = self.SC_FATAL
     retTmpError = self.SC_FAILED
     tmpLog.debug("vo={0} label={1} queue={2}".format(vo, prodSourceLabel, workQueue.queue_name))
     # loop over all tasks
     allRwMap = {}
     prioMap = {}
     tt2Map = {}
     expRWs = {}
     jobSpecList = []
     for tmpJediTaskID, tmpInputList in inputList:
         for taskSpec, cloudName, inputChunk in tmpInputList:
             # make JobSpec to be submitted for TaskAssigner
             jobSpec = JobSpec()
             jobSpec.taskID = taskSpec.jediTaskID
             jobSpec.jediTaskID = taskSpec.jediTaskID
             # set managed to trigger TA
             jobSpec.prodSourceLabel = "managed"
             jobSpec.processingType = taskSpec.processingType
             jobSpec.workingGroup = taskSpec.workingGroup
             jobSpec.metadata = taskSpec.processingType
             jobSpec.assignedPriority = taskSpec.taskPriority
             jobSpec.currentPriority = taskSpec.currentPriority
             jobSpec.maxDiskCount = (taskSpec.getOutDiskSize() + taskSpec.getWorkDiskSize()) / 1024 / 1024
             if taskSpec.useWorldCloud():
                 # use destinationSE to trigger task brokerage in WORLD cloud
                 jobSpec.destinationSE = taskSpec.cloud
             prodDBlock = None
             setProdDBlock = False
             for datasetSpec in inputChunk.getDatasets():
                 prodDBlock = datasetSpec.datasetName
                 if datasetSpec.isMaster():
                     jobSpec.prodDBlock = datasetSpec.datasetName
                     setProdDBlock = True
                 for fileSpec in datasetSpec.Files:
                     tmpInFileSpec = fileSpec.convertToJobFileSpec(datasetSpec)
                     jobSpec.addFile(tmpInFileSpec)
             # use secondary dataset name as prodDBlock
             if setProdDBlock == False and prodDBlock != None:
                 jobSpec.prodDBlock = prodDBlock
             # append
             jobSpecList.append(jobSpec)
             prioMap[jobSpec.taskID] = jobSpec.currentPriority
             tt2Map[jobSpec.taskID] = jobSpec.processingType
             # get RW for a priority
             if not allRwMap.has_key(jobSpec.currentPriority):
                 tmpRW = self.taskBufferIF.calculateRWwithPrio_JEDI(
                     vo, prodSourceLabel, workQueue, jobSpec.currentPriority
                 )
                 if tmpRW == None:
                     tmpLog.error("failed to calculate RW with prio={0}".format(jobSpec.currentPriority))
                     return retTmpError
                 allRwMap[jobSpec.currentPriority] = tmpRW
             # get expected RW
             expRW = self.taskBufferIF.calculateTaskRW_JEDI(jobSpec.jediTaskID)
             if expRW == None:
                 tmpLog.error("failed to calculate RW for jediTaskID={0}".format(jobSpec.jediTaskID))
                 return retTmpError
             expRWs[jobSpec.taskID] = expRW
     # get fullRWs
     fullRWs = self.taskBufferIF.calculateRWwithPrio_JEDI(vo, prodSourceLabel, None, None)
     if fullRWs == None:
         tmpLog.error("failed to calculate full RW")
         return retTmpError
     # set metadata
     for jobSpec in jobSpecList:
         rwValues = allRwMap[jobSpec.currentPriority]
         jobSpec.metadata = "%s;%s;%s;%s;%s;%s" % (
             jobSpec.metadata,
             str(rwValues),
             str(expRWs),
             str(prioMap),
             str(fullRWs),
             str(tt2Map),
         )
     tmpLog.debug("run task assigner for {0} tasks".format(len(jobSpecList)))
     nBunchTask = 0
     while nBunchTask < len(jobSpecList):
         # get a bunch
         jobsBunch = jobSpecList[nBunchTask : nBunchTask + maxBunchTask]
         strIDs = "jediTaskID="
         for tmpJobSpec in jobsBunch:
             strIDs += "{0},".format(tmpJobSpec.taskID)
         strIDs = strIDs[:-1]
         tmpLog.debug(strIDs)
         # increment index
         nBunchTask += maxBunchTask
         # run task brokerge
         stS, outSs = PandaClient.runTaskAssignment(jobsBunch)
         tmpLog.debug("{0}:{1}".format(stS, str(outSs)))
     # return
     tmpLog.debug("done")
     return self.SC_SUCCEEDED

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
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)
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    job.homepackage       = 'AnalysisTransforms'
    job.transformation    = 'https://gridui01.usatlas.bnl.gov:24443/dav/test/runAthena'
    job.destinationDBlock = datasetName
    job.destinationSE     = destName
    job.currentPriority   = 100
    job.prodSourceLabel   = 'user'
    job.computingSite     = site
    #job.prodDBlock        = "pandatest.b1599dfa-cd36-4fc5-92f6-495781a94c66"
    job.prodDBlock        = "pandatest.f228b051-077b-4f81-90bf-496340644379"
    
    fileI = FileSpec()
    fileI.dataset    = job.prodDBlock
    fileI.prodDBlock = job.prodDBlock
    fileI.lfn = "lib.f228b051-077b-4f81-90bf-496340644379.tgz"
    fileI.type = 'input'
    job.addFile(fileI)
    
    fileOL = FileSpec()
    fileOL.lfn = "%s.job.log.tgz" % commands.getoutput('uuidgen') 
    fileOL.destinationDBlock = job.destinationDBlock
    fileOL.destinationSE     = job.destinationSE
    fileOL.dataset           = job.destinationDBlock
    fileOL.type = 'log'
    job.addFile(fileOL)

    fileOZ = FileSpec()
    fileOZ.lfn = "%s.pool.root" % commands.getoutput('uuidgen') 
    fileOZ.destinationDBlock = job.destinationDBlock
    fileOZ.destinationSE     = job.destinationSE
    fileOZ.dataset           = job.destinationDBlock
    fileOZ.type = 'output'

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
pool = DBProxyPool('adbpro.usatlas.bnl.gov',passwd,2)

proxy = pool.getProxy()

import sys
import commands

job1 = JobSpec()
job1.PandaID='NULL'
job1.jobStatus='unknown'
job1.computingSite="aaa"
f11 = FileSpec()
f11.lfn = 'in1.pool.root'
f11.type = 'input'
job1.addFile(f11)
f12 = FileSpec()
f12.lfn = 'out1.pool.root'
f12.type = 'output'
job1.addFile(f12)

job2 = JobSpec()
job2.PandaID='NULL'
job2.jobStatus='unknown'
job2.computingSite="bbb"
f21 = FileSpec()
f21.lfn = 'in2.pool.root'
f21.type = 'input'
job2.addFile(f21)
f22 = FileSpec()
f22.lfn = 'out2.pool.root'

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]

#.........这里部分代码省略.........

        pandajob = JobSpec()
        ## always setting a job definition ID
        pandajob.jobDefinitionID = jobdef if jobdef else -1
        ## always setting a job set ID
        pandajob.jobsetID = jobset if jobset else -1
        pandajob.jobName = jobname
        pandajob.prodUserID = task['tm_user_dn']
        pandajob.destinationDBlock = outdataset
        pandajob.prodDBlock = task['tm_input_dataset']
        pandajob.prodSourceLabel = 'user'
        pandajob.computingSite = site
        pandajob.cloud = getSite(pandajob.computingSite)
        pandajob.destinationSE = 'local'
        pandajob.transformation = task['tm_transformation']
        ## need to initialize this
        pandajob.metadata = ''

        def outFileSpec(of=None, log=False):
            """Local routine to create an FileSpec for the an job output/log file

               :arg str of: output file base name
               :return: FileSpec object for the output file."""
            outfile = FileSpec()
            if log:
                outfile.lfn = "job.log_%d_%s.tgz" % (jobid, lfnhanger)
                outfile.type = 'log'
            else:
                outfile.lfn = '%s_%d_%s%s' %(os.path.splitext(of)[0], jobid, lfnhanger, os.path.splitext(of)[1])
                outfile.type = 'output'
            outfile.destinationDBlock = pandajob.destinationDBlock
            outfile.destinationSE = task['tm_asyncdest']
            outfile.dataset = pandajob.destinationDBlock
            return outfile

        alloutfiles = []
        outjobpar = {}
        outfilestring = ''
        for outputfile in task['tm_outfiles']:
            outfilestring += '%s,' % outputfile
            filespec = outFileSpec(outputfile)
            alloutfiles.append(filespec)
            #pandajob.addFile(filespec)
            outjobpar[outputfile] = filespec.lfn
        for outputfile in task['tm_tfile_outfiles']:
            outfilestring += '%s,' % outputfile
            filespec = outFileSpec(outputfile)
            alloutfiles.append(filespec)
            #pandajob.addFile(filespec)
            outjobpar[outputfile] = filespec.lfn
        for outputfile in task['tm_edm_outfiles']:
            outfilestring += '%s,' % outputfile
            filespec = outFileSpec(outputfile)
            alloutfiles.append(filespec)
            #pandajob.addFile(filespec)
            outjobpar[outputfile] = filespec.lfn
        outfilestring = outfilestring[:-1]

        infiles = []
        for inputfile in job['input_files']:
            infiles.append( inputfile['lfn'] )

        pandajob.jobParameters = '-a %s ' % task['tm_user_sandbox']
        pandajob.jobParameters += '--sourceURL %s ' % task['tm_cache_url']
        pandajob.jobParameters += '--jobNumber=%s ' % jobid
        pandajob.jobParameters += '--cmsswVersion=%s ' % task['tm_job_sw']
        pandajob.jobParameters += '--scramArch=%s ' % task['tm_job_arch']
        pandajob.jobParameters += '--inputFile=\'%s\' ' % json.dumps(infiles)

        self.jobParametersSetting(pandajob, job, self.jobtypeMapper[task['tm_job_type']])

        pandajob.jobParameters += '-o "%s" ' % str(outjobpar)
        pandajob.jobParameters += '--dbs_url=%s ' % task['tm_dbs_url']
        pandajob.jobParameters += '--publish_dbs_url=%s ' % task['tm_publish_dbs_url']
        pandajob.jobParameters += '--publishFiles=%s ' % ('True' if task['tm_publication'] == 'T' else 'False')
        pandajob.jobParameters += '--saveLogs=%s ' % ('True' if task['tm_save_logs'] == 'T' else 'False')
        pandajob.jobParameters += '--availableSites=\'%s\' ' %json.dumps(allsites)

        pandajob.jobParameters += '--group=%s ' % (task['tm_user_group'] if task['tm_user_group'] else '')
        pandajob.jobParameters += '--role=%s ' % (task['tm_user_role'] if task['tm_user_role'] else '')

        self.logger.info(type(task['tm_user_infiles']))
        self.logger.info(task['tm_user_infiles'])

        if task['tm_user_infiles']:
            addinfilestring = ''
            for addinfile in task['tm_user_infiles']:
                addinfilestring += '%s,' % addinfile
            pandajob.jobParameters += '--userFiles=%s ' % ( addinfilestring[:-1] )

        pandajob.jobName = '%s' % task['tm_taskname'] #Needed by ASO and Dashboard

        if 'panda_oldjobid' in job and job['panda_oldjobid']:
            pandajob.parentID = job['panda_oldjobid']

        pandajob.addFile(outFileSpec(log=True))
        for filetoadd in alloutfiles:
            pandajob.addFile(filetoadd)

        return pandajob

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
    job.transformation    = 'csc_simul_trf.py'
    job.destinationDBlock = datasetName
    job.cloud = cloud
    job.computingSite     = site
#    job.prodDBlock        = 'mc12.005001.pythia_minbias.evgen.EVNT.v12000701_tid012303'
    job.prodDBlock        = files[lfn]
    job.prodSourceLabel   = 'test'
#    job.prodSourceLabel   = 'cloudtest'
    job.currentPriority   = 1001

    fileI = FileSpec()
    fileI.dataset    = job.prodDBlock
    fileI.prodDBlock = job.prodDBlock
    fileI.lfn = lfn
    fileI.type = 'input'
    job.addFile(fileI)

    fileD = FileSpec()
    fileD.dataset    = 'ddo.000001.Atlas.Ideal.DBRelease.v030101'
    fileD.prodDBlock = 'ddo.000001.Atlas.Ideal.DBRelease.v030101'
    fileD.lfn = 'DBRelease-3.1.1.tar.gz'
    fileD.type = 'input'
    job.addFile(fileD)


    fileOE = FileSpec()
    fileOE.lfn = "%s.HITS.pool.root" % job.jobName
    fileOE.destinationDBlock = job.destinationDBlock
    fileOE.destinationSE     = job.destinationSE
    fileOE.dataset           = job.destinationDBlock
    fileOE.type = 'output'

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
            newName = '%s.%s%s' % (job.jobName,prefix,suffix)
            outMap[arg] = (var,newName)
        # DBRelease
        elif arg == 'DBRelease':
            dbrMap = (arg,var)
        # input
        elif arg.startswith('input') and arg.endswith('File'):
            inputMap = (arg,var)


    fileI = FileSpec()
    fileI.dataset    = job.prodDBlock
    fileI.prodDBlock = job.prodDBlock
    fileI.lfn = lfn
    fileI.type = 'input'
    job.addFile(fileI)

    fileD = FileSpec()
    fileD.dataset    = 'ddo.000001.Atlas.Ideal.DBRelease.v06020105'
    fileD.prodDBlock = fileD.dataset
    fileD.lfn = 'DBRelease-6.2.1.5.tar.gz'
    fileD.type = 'input'
    job.addFile(fileD)

    newParams = origParams
    newParams = newParams.replace(dbrMap[0]+'='+dbrMap[1],dbrMap[0]+'='+fileD.lfn)
    newParams = newParams.replace(inputMap[0]+'='+inputMap[1],inputMap[0]+'='+fileI.lfn)    

    for arg,vars in outMap.iteritems():
        fileO = FileSpec()
        fileO.lfn = vars[1]

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]

#.........这里部分代码省略.........
                     if self.pandaDDM and self.job.prodSourceLabel=='managed':
                         # instantiate SiteMapper
                         if self.siteMapper == None:
                             self.siteMapper = SiteMapper(self.taskBuffer)
                         # get file list for PandaDDM
                         retList = self.taskBuffer.queryFilesWithMap({'destinationDBlock':destinationDBlock})
                         lfnsStr = ''
                         guidStr = ''
                         for tmpFile in retList:
                             if tmpFile.type in ['log','output']:
                                 lfnsStr += '%s,' % tmpFile.lfn
                                 guidStr += '%s,' % tmpFile.GUID
                         if lfnsStr != '':
                             guidStr = guidStr[:-1]
                             lfnsStr = lfnsStr[:-1]
                             # create a DDM job
                             ddmjob = JobSpec()
                             ddmjob.jobDefinitionID   = int(time.time()) % 10000
                             ddmjob.jobName           = "%s" % commands.getoutput('uuidgen')
                             ddmjob.transformation    = 'http://pandaserver.cern.ch:25080/trf/mover/run_dq2_cr'
                             ddmjob.destinationDBlock = 'testpanda.%s' % ddmjob.jobName
                             ddmjob.computingSite     = "BNL_ATLAS_DDM"
                             ddmjob.destinationSE     = ddmjob.computingSite
                             ddmjob.currentPriority   = 200000
                             ddmjob.prodSourceLabel   = 'ddm'
                             ddmjob.transferType      = 'sub'
                             # append log file
                             fileOL = FileSpec()
                             fileOL.lfn = "%s.job.log.tgz" % ddmjob.jobName
                             fileOL.destinationDBlock = ddmjob.destinationDBlock
                             fileOL.destinationSE     = ddmjob.destinationSE
                             fileOL.dataset           = ddmjob.destinationDBlock
                             fileOL.type = 'log'
                             ddmjob.addFile(fileOL)
                             # make arguments
                             dstDQ2ID = 'BNLPANDA'
                             srcDQ2ID = self.siteMapper.getSite(self.job.computingSite).ddm
                             callBackURL = 'https://%s:%s/server/panda/datasetCompleted?vuid=%s&site=%s' % \
                                           (panda_config.pserverhost,panda_config.pserverport,
                                            dataset.vuid,dstDQ2ID)
                             _logger.debug(callBackURL)
                             # set src/dest
                             ddmjob.sourceSite      = srcDQ2ID
                             ddmjob.destinationSite = dstDQ2ID
                             # if src==dst, send callback without ddm job
                             if dstDQ2ID == srcDQ2ID:
                                 comout = commands.getoutput('curl -k %s' % callBackURL)
                                 _logger.debug(comout)
                             else:
                                 # run dq2_cr
                                 callBackURL = urllib.quote(callBackURL)
                                 # get destination dir
                                 destDir = brokerage.broker_util._getDefaultStorage(self.siteMapper.getSite(self.job.computingSite).dq2url)
                                 argStr = "-s %s -r %s --guids %s --lfns %s --callBack %s -d %s/%s %s" % \
                                          (srcDQ2ID,dstDQ2ID,guidStr,lfnsStr,callBackURL,destDir,
                                           destinationDBlock,destinationDBlock)
                                 # set job parameters
                                 ddmjob.jobParameters = argStr
                                 _logger.debug('%s pdq2_cr %s' % (self.pandaID,ddmjob.jobParameters))
                                 ddmJobs.append(ddmjob)
                     # start Activator
                     if re.search('_sub\d+$',dataset.name) == None:
                         if self.job.prodSourceLabel=='panda' and self.job.processingType in ['merge','unmerge']:
                             # don't trigger Activator for merge jobs
                             pass
                         else:

# 需要导入模块: from taskbuffer.JobSpec import JobSpec [as 别名]
# 或者: from taskbuffer.JobSpec.JobSpec import addFile [as 别名]
job.computingSite     = site

job.prodDBlock        = 'valid1.005200.T1_McAtNlo_Jimmy.simul.HITS.e322_s429_tid022081'
    
job.prodSourceLabel   = 'test'    
job.currentPriority   = 10000
job.cloud             = 'US'

for lfn in ['HITS.022081._00001.pool.root','HITS.022081._00002.pool.root']:
    fileI = FileSpec()
    fileI.dataset    = job.prodDBlock
    fileI.prodDBlock = job.prodDBlock
    fileI.lfn = lfn
    fileI.type = 'input'
    job.addFile(fileI)

fileD1 = FileSpec()
fileD1.dataset    = 'ddo.000001.Atlas.Ideal.DBRelease.v050001'
fileD1.prodDBlock = fileD1.dataset
fileD1.lfn = 'DBRelease-5.0.1.tar.gz'
fileD1.type = 'input'
job.addFile(fileD1)

fileD2 = FileSpec()
fileD2.dataset    = 'ddo.000001.Atlas.Ideal.DBRelease.v050101'
fileD2.prodDBlock = fileD2.dataset
fileD2.lfn = 'DBRelease-5.1.1.tar.gz'
fileD2.type = 'input'
job.addFile(fileD2)