Python FileSpec.destinationDBlock方法代码示例

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

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
 def convertToJobFileSpec(self,datasetSpec,setType=None,useEventService=False):
     jobFileSpec = JobFileSpec()
     jobFileSpec.fileID     = self.fileID
     jobFileSpec.datasetID  = datasetSpec.datasetID
     jobFileSpec.jediTaskID = datasetSpec.jediTaskID
     jobFileSpec.lfn        = self.lfn
     jobFileSpec.GUID       = self.GUID
     if setType == None:
         jobFileSpec.type   = self.type
     else:
         jobFileSpec.type   = setType
     jobFileSpec.scope      = self.scope
     jobFileSpec.fsize      = self.fsize
     jobFileSpec.checksum   = self.checksum
     jobFileSpec.attemptNr  = self.attemptNr
     # dataset attribute
     if datasetSpec != None:
         # dataset
         if not datasetSpec.containerName in [None,'']:
             jobFileSpec.dataset = datasetSpec.containerName
         else:
             jobFileSpec.dataset = datasetSpec.datasetName
         if self.type in datasetSpec.getInputTypes() or setType in datasetSpec.getInputTypes():
             # prodDBlock
             jobFileSpec.prodDBlock = datasetSpec.datasetName
             # storage token    
             if not datasetSpec.storageToken in ['',None]:
                 jobFileSpec.dispatchDBlockToken = datasetSpec.storageToken 
         else:
             # destinationDBlock
             jobFileSpec.destinationDBlock = datasetSpec.datasetName
             # storage token    
             if not datasetSpec.storageToken in ['',None]:
                 jobFileSpec.destinationDBlockToken = datasetSpec.storageToken.split('/')[0] 
             # destination
             if not datasetSpec.destination in ['',None]:
                 jobFileSpec.destinationSE = datasetSpec.destination
             # set prodDBlockToken for Event Service
             if useEventService and datasetSpec.getObjectStore() != None:
                 jobFileSpec.prodDBlockToken = 'objectstore^{0}'.format(datasetSpec.getObjectStore())
             # allow no output
             if datasetSpec.isAllowedNoOutput():
                 jobFileSpec.allowNoOutput()
     # return
     return jobFileSpec

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
    job.homepackage       = 'AtlasProduction/12.0.7.1'

    job.transformation    = 'csc_evgen_trf.py'
    job.destinationDBlock = datasetName
#    job.destinationSE     = destName
#    job.cloud             = 'CA'
    job.cloud             = cloud
    job.taskID = taskid
    job.currentPriority   = 1000
    job.prodSourceLabel   = 'test'
#    job.prodSourceLabel   = 'cloudtest'
    job.computingSite     = site
    
    file = FileSpec()
    file.lfn = "%s.evgen.pool.root" % job.jobName
    file.destinationDBlock = job.destinationDBlock
    file.destinationSE     = job.destinationSE
    file.dataset           = job.destinationDBlock
    file.type = 'output'
    job.addFile(file)
    
    fileOL = FileSpec()
    fileOL.lfn = "%s.job.log.tgz" % job.jobName
    fileOL.destinationDBlock = job.destinationDBlock
    fileOL.destinationSE     = job.destinationSE
    fileOL.dataset           = job.destinationDBlock
    fileOL.type = 'log'
    job.addFile(fileOL)
    
    job.jobParameters="8072 0 5000 1 DC3.008072.JimmyPhotonJet1.py %s NONE NONE NONE" % file.lfn
    jobList.append(job)

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

#.........这里部分代码省略.........
        elif job.splitter._name not in ['DQ2JobSplitter', 'ArgSplitter', 'ArgSplitterTask']:
            raise ApplicationConfigurationError('Panda splitter must be DQ2JobSplitter or ArgSplitter')
        
        if job.backend.site == 'AUTO':
            raise ApplicationConfigurationError('site is still AUTO after brokerage!')

#       output dataset
        if job.outputdata:
            if job.outputdata._name != 'DQ2OutputDataset':
                raise ApplicationConfigurationError('Panda backend supports only DQ2OutputDataset')
        else:
            logger.info('Adding missing DQ2OutputDataset')
            job.outputdata = DQ2OutputDataset()

        job.outputdata.datasetname,outlfn = dq2outputdatasetname(job.outputdata.datasetname, job.id, job.outputdata.isGroupDS, job.outputdata.groupname)

        self.outDsLocation = Client.PandaSites[job.backend.site]['ddm']

        try:
            Client.addDataset(job.outputdata.datasetname,False,location=self.outDsLocation)
            logger.info('Output dataset %s registered at %s'%(job.outputdata.datasetname,self.outDsLocation))
            dq2_set_dataset_lifetime(job.outputdata.datasetname, location=self.outDsLocation)
        except exceptions.SystemExit:
            raise BackendError('Panda','Exception in Client.addDataset %s: %s %s'%(job.outputdata.datasetname,sys.exc_info()[0],sys.exc_info()[1]))

        # handle the libds
        if job.backend.libds:
            self.libDataset = job.backend.libds
            self.fileBO = getLibFileSpecFromLibDS(self.libDataset)
            self.library = self.fileBO.lfn
        elif job.backend.bexec:
            self.libDataset = job.outputdata.datasetname+'.lib'
            self.library = '%s.tgz' % self.libDataset
            try:
                Client.addDataset(self.libDataset,False,location=self.outDsLocation)
                dq2_set_dataset_lifetime(self.libDataset, location=self.outDsLocation)
                logger.info('Lib dataset %s registered at %s'%(self.libDataset,self.outDsLocation))
            except exceptions.SystemExit:
                raise BackendError('Panda','Exception in Client.addDataset %s: %s %s'%(self.libDataset,sys.exc_info()[0],sys.exc_info()[1]))

        # collect extOutFiles
        self.extOutFile = []
        for tmpName in job.outputdata.outputdata:
            if tmpName != '':
                self.extOutFile.append(tmpName)

        for tmpName in job.outputsandbox:
            if tmpName != '':
                self.extOutFile.append(tmpName)

        for tmpName in job.backend.extOutFile:
            if tmpName != '':
                self.extOutFile.append(tmpName)

        # create build job
        if job.backend.bexec != '':
            jspec = JobSpec()
            jspec.jobDefinitionID   = job.id
            jspec.jobName           = commands.getoutput('uuidgen 2> /dev/null')
            jspec.transformation    = '%s/buildGen-00-00-01' % Client.baseURLSUB
            if Client.isDQ2free(job.backend.site):
                jspec.destinationDBlock = '%s/%s' % (job.outputdata.datasetname,self.libDataset)
                jspec.destinationSE     = 'local'
            else:
                jspec.destinationDBlock = self.libDataset
                jspec.destinationSE     = job.backend.site
            jspec.prodSourceLabel   = configPanda['prodSourceLabelBuild']
            jspec.processingType    = configPanda['processingType']
            jspec.assignedPriority  = configPanda['assignedPriorityBuild']
            jspec.computingSite     = job.backend.site
            jspec.cloud             = job.backend.requirements.cloud
            jspec.jobParameters     = '-o %s' % (self.library)
            if self.inputsandbox:
                jspec.jobParameters     += ' -i %s' % (self.inputsandbox)
            else:
                raise ApplicationConfigurationError('Executable on Panda with build job defined, but inputsandbox is emtpy !')
            matchURL = re.search('(http.*://[^/]+)/',Client.baseURLCSRVSSL)
            if matchURL:
                jspec.jobParameters += ' --sourceURL %s ' % matchURL.group(1)
            if job.backend.bexec != '':
                jspec.jobParameters += ' --bexec "%s" ' % urllib.quote(job.backend.bexec)
                jspec.jobParameters += ' -r %s ' % '.'
                

            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)
            return jspec
        else:
            return None

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [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.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [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.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [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.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [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.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
    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'
    job.addFile(fileOE)

    fileOA = FileSpec()
    fileOA.lfn = "%s.RDO.pool.root" %  job.jobName
    fileOA.destinationDBlock = job.destinationDBlock
    fileOA.destinationSE     = job.destinationSE
    fileOA.dataset           = job.destinationDBlock
    fileOA.type = 'output'
    job.addFile(fileOA)

    fileOL = FileSpec()
    fileOL.lfn = "%s.job.log.tgz" % job.jobName

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
 def run(self):
     try:
         _logger.debug('%s Start %s' % (self.pandaID,self.job.jobStatus))
         flagComplete    = True
         ddmJobs         = []
         topUserDsList   = []
         usingMerger     = False        
         disableNotifier = False
         firstIndvDS     = True
         finalStatusDS   = []
         for destinationDBlock in self.destinationDBlocks:
             dsList = []
             _logger.debug('%s start %s' % (self.pandaID,destinationDBlock))
             # ignore tid datasets
             if re.search('_tid[\d_]+$',destinationDBlock):
                 _logger.debug('%s skip %s' % (self.pandaID,destinationDBlock))                
                 continue
             # ignore HC datasets
             if re.search('^hc_test\.',destinationDBlock) != None or re.search('^user\.gangarbt\.',destinationDBlock) != None:
                 if re.search('_sub\d+$',destinationDBlock) == None and re.search('\.lib$',destinationDBlock) == None:
                     _logger.debug('%s skip HC %s' % (self.pandaID,destinationDBlock))                
                     continue
             # query dataset
             if self.datasetMap.has_key(destinationDBlock):
                 dataset = self.datasetMap[destinationDBlock]
             else:
                 dataset = self.taskBuffer.queryDatasetWithMap({'name':destinationDBlock})
             if dataset == None:
                 _logger.error('%s Not found : %s' % (self.pandaID,destinationDBlock))
                 flagComplete = False
                 continue
             # skip tobedeleted/tobeclosed 
             if dataset.status in ['cleanup','tobeclosed','completed']:
                 _logger.debug('%s skip %s due to %s' % (self.pandaID,destinationDBlock,dataset.status))
                 continue
             dsList.append(dataset)
             # sort
             dsList.sort()
             # count number of completed files
             notFinish = self.taskBuffer.countFilesWithMap({'destinationDBlock':destinationDBlock,
                                                            'status':'unknown'})
             if notFinish < 0:
                 _logger.error('%s Invalid DB return : %s' % (self.pandaID,notFinish))
                 flagComplete = False                
                 continue
             # check if completed
             _logger.debug('%s notFinish:%s' % (self.pandaID,notFinish))
             if self.job.destinationSE == 'local' and self.job.prodSourceLabel in ['user','panda']:
                 # close non-DQ2 destinationDBlock immediately
                 finalStatus = 'closed'
             elif self.job.lockedby == 'jedi' and self.isTopLevelDS(destinationDBlock):
                 # set it closed in order not to trigger DDM cleanup. It will be closed by JEDI
                 finalStatus = 'closed'
             elif self.job.prodSourceLabel in ['user'] and "--mergeOutput" in self.job.jobParameters \
                      and self.job.processingType != 'usermerge':
                 # merge output files
                 if firstIndvDS:
                     # set 'tobemerged' to only the first dataset to avoid triggering many Mergers for --individualOutDS
                     finalStatus = 'tobemerged'
                     firstIndvDS = False
                 else:
                     finalStatus = 'tobeclosed'
                 # set merging to top dataset
                 usingMerger = True
                 # disable Notifier
                 disableNotifier = True
             elif self.job.produceUnMerge():
                 finalStatus = 'doing'
             else:
                 # set status to 'tobeclosed' to trigger DQ2 closing
                 finalStatus = 'tobeclosed'
             if notFinish==0: 
                 _logger.debug('%s set %s to dataset : %s' % (self.pandaID,finalStatus,destinationDBlock))
                 # set status
                 dataset.status = finalStatus
                 # update dataset in DB
                 retT = self.taskBuffer.updateDatasets(dsList,withLock=True,withCriteria="status<>:crStatus AND status<>:lockStatus ",
                                                       criteriaMap={':crStatus':finalStatus,':lockStatus':'locked'})
                 if len(retT) > 0 and retT[0]==1:
                     finalStatusDS += dsList
                     # close user datasets
                     if self.job.prodSourceLabel in ['user'] and self.job.destinationDBlock.endswith('/') \
                            and (dataset.name.startswith('user') or dataset.name.startswith('group')):
                         # get top-level user dataset 
                         topUserDsName = re.sub('_sub\d+$','',dataset.name)
                         # update if it is the first attempt
                         if topUserDsName != dataset.name and not topUserDsName in topUserDsList and self.job.lockedby != 'jedi':
                             topUserDs = self.taskBuffer.queryDatasetWithMap({'name':topUserDsName})
                             if topUserDs != None:
                                 # check status
                                 if topUserDs.status in ['completed','cleanup','tobeclosed',
                                                         'tobemerged','merging']:
                                     _logger.debug('%s skip %s due to status=%s' % (self.pandaID,topUserDsName,topUserDs.status))
                                 else:
                                     # set status
                                     if self.job.processingType.startswith('gangarobot') or \
                                            self.job.processingType.startswith('hammercloud'):
                                         # not trigger freezing for HC datasets so that files can be appended
                                         topUserDs.status = 'completed'
                                     elif not usingMerger:
#.........这里部分代码省略.........

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
            file.dataset    = iDataset
            file.prodDBlock = iDataset
            job.addFile(file)
            break
    # output files
    for oDataset in oDatasets:
        # remove _tidXXX
        pat = re.sub('_tid\d+$','',oDataset)
        # search
        m = re.search('('+pat+'\S+)',line)
        if m != None:
            file = FileSpec()
            file.lfn  = m.group(1)
            file.type = 'output'
            file.dataset           = oDataset
            file.destinationDBlock = oDataset
            job.addFile(file)
    # log
    file = FileSpec()
    file.lfn  = "%s._%05d.log.tgz" % (lDataset,iJob)
    file.type = 'log'
    file.dataset           = lDataset
    file.destinationDBlock = lDataset
    job.addFile(file)

    # job par
    job.jobParameters = line[:-1]

    """
    print job.values()
    for file in job.Files:

# 需要导入模块: from taskbuffer.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [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.FileSpec import FileSpec [as 别名]
# 或者: from taskbuffer.FileSpec.FileSpec import destinationDBlock [as 别名]
    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)


    fileO1 = FileSpec()
    fileO1.lfn = "%s.ESD.pool.root" % job.jobName
    fileO1.destinationDBlock = job.destinationDBlock
    fileO1.destinationSE     = job.destinationSE
    fileO1.dataset           = job.destinationDBlock
    fileO1.type = 'output'
    job.addFile(fileO1)

    fileO2 = FileSpec()
    fileO2.lfn = "%s.ESDF.pool.root" %  job.jobName
    fileO2.destinationDBlock = job.destinationDBlock
    fileO2.destinationSE     = job.destinationSE
    fileO2.dataset           = job.destinationDBlock
    fileO2.type = 'output'
#    job.addFile(fileO2)

    fileO3 = FileSpec()
    fileO3.lfn = "%s.NTUP.pool.root" %  job.jobName