Python MonteCarlo.MonteCarloWorkloadFactory類代碼示例

    def testMCWithPileup(self):
        """
        _testMCWithPileup_

        Create a Monte Carlo workflow and verify that it is injected correctly
        into WMBS and invoke its detailed test.
        The input configuration includes pileup input files.
        """
        defaultArguments = MonteCarloWorkloadFactory.getTestArguments()
        defaultArguments["CouchURL"] = os.environ["COUCHURL"]
        defaultArguments["CouchDBName"] = TEST_DB_NAME
        defaultArguments["ConfigCacheID"] = self.injectMonteCarloConfig()

        # Add pileup inputs
        defaultArguments["MCPileup"] = COSMICS_PU
        defaultArguments["DataPileup"] = DATA_PU
        defaultArguments["DeterministicPileup"] = True

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)

        testWMBSHelper = WMBSHelper(testWorkload, "Production", "SomeBlock", cachepath=self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        self._commonMonteCarloTest()
        productionTask = testWorkload.getTaskByPath('/TestWorkload/Production')
        cmsRunStep = productionTask.getStep("cmsRun1").getTypeHelper()
        pileupData = cmsRunStep.getPileup()
        self.assertEqual(pileupData.data.dataset, [DATA_PU])
        self.assertEqual(pileupData.mc.dataset, [COSMICS_PU])

        splitting = productionTask.jobSplittingParameters()
        self.assertTrue(splitting["deterministicPileup"])
        return

    def testMCWithLHE(self):
        """
        _testMCWithLHE_

        Create a MonteCarlo workflow with a variation on the type of work
        done, this refers to the previous LHEStepZero where the input
        can be .lhe files and there is more than one lumi per job.
        """
        defaultArguments = MonteCarloWorkloadFactory.getTestArguments()
        defaultArguments["CouchURL"] = os.environ["COUCHURL"]
        defaultArguments["CouchDBName"] = TEST_DB_NAME
        defaultArguments["ConfigCacheID"] = self.injectMonteCarloConfig()
        defaultArguments["LheInputFiles"] = "True"
        defaultArguments["EventsPerJob"] = 200
        defaultArguments["EventsPerLumi"] = 50

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)

        testWMBSHelper = WMBSHelper(testWorkload, "Production", "SomeBlock", cachepath=self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        self._commonMonteCarloTest()

        productionTask = testWorkload.getTaskByPath('/TestWorkload/Production')
        splitting = productionTask.jobSplittingParameters()
        self.assertEqual(splitting["events_per_job"], 200)
        self.assertEqual(splitting["events_per_lumi"], 50)
        self.assertEqual(splitting["lheInputFiles"], True)
        self.assertFalse(splitting["deterministicPileup"])

        return

    def testPileupFetcherOnMC(self):
        pileupMcArgs = MonteCarloWorkloadFactory.getTestArguments()
        pileupMcArgs["PileupConfig"] = {"cosmics": ["/Mu/PenguinsPenguinsEverywhere-SingleMu-HorriblyJaundicedYellowEyedPenginsSearchingForCarrots-v31/RECO"],
                                        "minbias": ["/Mu/PenguinsPenguinsEverywhere-SingleMu-HorriblyJaundicedYellowEyedPenginsSearchingForCarrots-v31/RECO"]}
        pileupMcArgs["CouchURL"] = os.environ["COUCHURL"]
        pileupMcArgs["CouchDBName"] = "pileupfetcher_t"
        pileupMcArgs["ConfigCacheID"] = self.injectGenerationConfig()

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", pileupMcArgs)

        # Since this is test of the fetcher - The loading from WMBS isn't
        # really necessary because the fetching happens before the workflow
        # is inserted into WMBS: feed the workload instance directly into fetcher:
        fetcher = PileupFetcher()
        creator = SandboxCreator()
        pathBase = "%s/%s" % (self.testDir, testWorkload.name())
        for topLevelTask in testWorkload.taskIterator():
            for taskNode in topLevelTask.nodeIterator():
                # this is how the call to PileupFetcher is happening
                # from the SandboxCreator test
                task = WMTask.WMTaskHelper(taskNode)
                taskPath = "%s/WMSandbox/%s" % (pathBase, task.name())
                fetcher.setWorkingDirectory(taskPath)
                # create Sandbox for the fetcher ...
                creator._makePathonPackage(taskPath)
                fetcher(task)
                self._queryPileUpConfigFile(pileupMcArgs, task, taskPath)

    def testPileupFetcherOnMC(self):
        pileupMcArgs = MonteCarloWorkloadFactory.getTestArguments()
        pileupMcArgs["MCPileup"] = "/Cosmics/ComissioningHI-PromptReco-v1/RECO"
        pileupMcArgs["DataPileup"] = "/HighPileUp/Run2011A-v1/RAW"
        pileupMcArgs["CouchURL"] = os.environ["COUCHURL"]
        pileupMcArgs["CouchDBName"] = "pileupfetcher_t"
        pileupMcArgs["ConfigCacheID"] = self.injectGenerationConfig()

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", pileupMcArgs)

        # now that the workload was created and args validated, we can add this PileupConfig
        pileupMcArgs["PileupConfig"] = parsePileupConfig(pileupMcArgs["MCPileup"], pileupMcArgs["DataPileup"])

        # Since this is test of the fetcher - The loading from WMBS isn't
        # really necessary because the fetching happens before the workflow
        # is inserted into WMBS: feed the workload instance directly into fetcher:
        fetcher = PileupFetcher()
        creator = SandboxCreator()
        pathBase = "%s/%s" % (self.testDir, testWorkload.name())
        for topLevelTask in testWorkload.taskIterator():
            for taskNode in topLevelTask.nodeIterator():
                # this is how the call to PileupFetcher is happening
                # from the SandboxCreator test
                task = WMTask.WMTaskHelper(taskNode)
                taskPath = "%s/WMSandbox/%s" % (pathBase, task.name())
                fetcher.setWorkingDirectory(taskPath)
                # create Sandbox for the fetcher ...
                creator._makePathonPackage(taskPath)
                fetcher(task)
                self._queryPileUpConfigFile(pileupMcArgs, task, taskPath)

    def validateSchema(self, schema):
        """
        _validateSchema_

        Check for required fields, and some skim facts
        """
        MonteCarloWorkloadFactory.validateSchema(self, schema)
        #Now let's validate new stuff
        self.validateEventsPerLumi(schema)

    def __call__(self, workloadName, arguments):
        """
        Create a workload instance for a LHE Step0 request
        Just pass it down to MonteCarlo
        """
        #Override splitting arguments
        # Splitting arguments
        timePerEvent     = int(arguments.get('TimePerEvent', 60))
        filterEfficiency = float(arguments.get('FilterEfficiency', 1.0))
        totalTime        = int(arguments.get('TotalTime', 9 * 3600))
        self.totalEvents = int(int(arguments['RequestNumEvents']) / filterEfficiency)
        if arguments.get("LheInputFiles", False) == True \
             or arguments.get("LheInputFiles", False) == "True":
            self.lheInputFiles = True

        # These are mostly place holders because the job splitting algo and
        # parameters will be updated after the workflow has been created.
        eventsPerJob = int(totalTime/timePerEvent/filterEfficiency)
        self.prodJobSplitAlgo  = arguments.setdefault("ProdJobSplitAlgo", 'EventBased')
        self.prodJobSplitArgs  = arguments.setdefault("ProdJobSplitArgs",
                                               {"events_per_job": eventsPerJob,
                                                "events_per_lumi": arguments['EventsPerLumi']})
        self.prodJobSplitArgs.setdefault("lheInputFiles", self.lheInputFiles)
        mcWorkload = MonteCarloWorkloadFactory.__call__(self, workloadName, arguments)
        mcWorkload.setBlockCloseSettings(mcWorkload.getBlockCloseMaxWaitTime(), 5,
                                         250000000, mcWorkload.getBlockCloseMaxSize())

        return mcWorkload

    def testMemCoresSettings(self):
        """
        _testMemCoresSettings_

        Make sure the multicore and memory setings are properly propagated to
        all tasks and steps.
        """
        defaultArguments = MonteCarloWorkloadFactory.getTestArguments()
        defaultArguments["CouchURL"] = os.environ["COUCHURL"]
        defaultArguments["CouchDBName"] = TEST_DB_NAME
        defaultArguments["ConfigCacheID"] = self.injectMonteCarloConfig()

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)

        # test default values
        taskObj = testWorkload.getTask('Production')
        for step in ('cmsRun1', 'stageOut1', 'logArch1'):
            stepHelper = taskObj.getStepHelper(step)
            self.assertEqual(stepHelper.getNumberOfCores(), 1)
            self.assertEqual(stepHelper.getNumberOfStreams(), 0)
        # then test Memory requirements
        perfParams = taskObj.jobSplittingParameters()['performance']
        self.assertEqual(perfParams['memoryRequirement'], 2300.0)

        # now test case where args are provided
        defaultArguments["Multicore"] = 6
        defaultArguments["Memory"] = 4600.0
        defaultArguments["EventStreams"] = 3
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)
        taskObj = testWorkload.getTask('Production')
        for step in ('cmsRun1', 'stageOut1', 'logArch1'):
            stepHelper = taskObj.getStepHelper(step)
            if step == 'cmsRun1':
                self.assertEqual(stepHelper.getNumberOfCores(), defaultArguments["Multicore"])
                self.assertEqual(stepHelper.getNumberOfStreams(), defaultArguments["EventStreams"])
            else:
                self.assertEqual(stepHelper.getNumberOfCores(), 1)
                self.assertEqual(stepHelper.getNumberOfStreams(), 0)
        # then test Memory requirements
        perfParams = taskObj.jobSplittingParameters()['performance']
        self.assertEqual(perfParams['memoryRequirement'], defaultArguments["Memory"])

        return

    def testMonteCarloExtension(self):
        """
        _testMonteCarloExtension_

        Create a Monte Carlo workflow and verify that it is injected correctly
        into WMBS and invoke its detailed test. This uses a non-zero first
        lumi. Check that the splitting arguments are correctly
        set for the lfn counter.
        """
        defaultArguments = MonteCarloWorkloadFactory.getTestArguments()
        defaultArguments["CouchURL"] = os.environ["COUCHURL"]
        defaultArguments["CouchDBName"] = TEST_DB_NAME
        defaultArguments["ConfigCacheID"] = self.injectMonteCarloConfig()
        defaultArguments["FirstLumi"] = 10001
        defaultArguments["EventsPerJob"] = 100
        defaultArguments["FirstEvent"] = 10001
        # defaultArguments["FirstEvent"] = 10001

        initial_lfn_counter = (
            100
        )  # EventsPerJob == EventsPerLumi, then the number of previous jobs is equal to the number of the initial lumi

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)

        testWMBSHelper = WMBSHelper(testWorkload, "Production", "SomeBlock", cachepath=self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        self._commonMonteCarloTest()

        productionTask = testWorkload.getTaskByPath("/TestWorkload/Production")
        productionSplitting = productionTask.jobSplittingParameters()
        self.assertTrue("initial_lfn_counter" in productionSplitting, "No initial lfn counter was stored")
        self.assertEqual(productionSplitting["initial_lfn_counter"], initial_lfn_counter, "Wrong initial LFN counter")

        for outputMod in ["OutputA", "OutputB"]:
            mergeTask = testWorkload.getTaskByPath("/TestWorkload/Production/ProductionMerge%s" % outputMod)
            mergeSplitting = mergeTask.jobSplittingParameters()
            self.assertTrue("initial_lfn_counter" in mergeSplitting, "No initial lfn counter was stored")
            self.assertEqual(mergeSplitting["initial_lfn_counter"], initial_lfn_counter, "Wrong initial LFN counter")

        return

def getMCArgs():
    mcArgs = MonteCarloWorkloadFactory.getTestArguments()
    mcArgs.update({
                   "CouchURL": None,
                   "CouchDBName": None,
                   "ConfigCacheDoc" : None
                   })
    mcArgs.pop('ConfigCacheDoc')

    return mcArgs

    def testMonteCarlo(self):
        """
        _testMonteCarlo_

        Create a Monte Carlo workflow and verify that it is injected correctly
        into WMBS and invoke its detailed test.
        """
        defaultArguments = MonteCarloWorkloadFactory.getTestArguments()
        defaultArguments["CouchURL"] = os.environ["COUCHURL"]
        defaultArguments["ConfigCacheID"] = self.injectMonteCarloConfig()

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", defaultArguments)

        testWMBSHelper = WMBSHelper(testWorkload, "Production", "SomeBlock", cachepath = self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        self._commonMonteCarloTest()

        return

    def __call__(self, workloadName, arguments):
        """
        Create a workload instance for a LHE Step0 request
        Just pass it down to MonteCarlo
        """
        #Override splitting arguments
        # Splitting arguments
        timePerEvent     = int(arguments.get('TimePerEvent', 60))
        filterEfficiency = float(arguments.get('FilterEfficiency', 1.0))
        totalTime        = int(arguments.get('TotalTime', 9 * 3600))
        self.totalEvents = int(int(arguments['RequestNumEvents']) / filterEfficiency)

        # These are mostly place holders because the job splitting algo and
        # parameters will be updated after the workflow has been created.
        eventsPerJob = int(totalTime/timePerEvent/filterEfficiency)
        self.prodJobSplitAlgo  = arguments.setdefault("ProdJobSplitAlgo", 'EventBased')
        self.prodJobSplitArgs  = arguments.setdefault("ProdJobSplitArgs",
                                               {"events_per_job": eventsPerJob,
                                                "events_per_lumi": arguments['EventsPerLumi']})
        return MonteCarloWorkloadFactory.__call__(self, workloadName, arguments)

    def testFilesets(self):
        """
        Test workflow tasks, filesets and subscriptions creation
        """
        # expected tasks, filesets, subscriptions, etc
        expOutTasks = ['/TestWorkload/Production',
                       '/TestWorkload/Production/ProductionMergeOutputB',
                       '/TestWorkload/Production/ProductionMergeOutputA']
        expWfTasks = ['/TestWorkload/Production',
                      '/TestWorkload/Production/LogCollect',
                      '/TestWorkload/Production/ProductionCleanupUnmergedOutputA',
                      '/TestWorkload/Production/ProductionCleanupUnmergedOutputB',
                      '/TestWorkload/Production/ProductionMergeOutputA',
                      '/TestWorkload/Production/ProductionMergeOutputA/ProductionOutputAMergeLogCollect',
                      '/TestWorkload/Production/ProductionMergeOutputB',
                      '/TestWorkload/Production/ProductionMergeOutputB/ProductionOutputBMergeLogCollect']
        expFsets = ['FILESET_DEFINED_DURING_RUNTIME',
                    '/TestWorkload/Production/unmerged-OutputBUSER',
                    '/TestWorkload/Production/ProductionMergeOutputA/merged-logArchive',
                    '/TestWorkload/Production/ProductionMergeOutputA/merged-MergedRECO',
                    '/TestWorkload/Production/ProductionMergeOutputB/merged-logArchive',
                    '/TestWorkload/Production/ProductionMergeOutputB/merged-MergedUSER',
                    '/TestWorkload/Production/unmerged-logArchive',
                    '/TestWorkload/Production/unmerged-OutputARECO']
        subMaps = ['FILESET_DEFINED_DURING_RUNTIME',
                   (6,
                    '/TestWorkload/Production/ProductionMergeOutputA/merged-logArchive',
                    '/TestWorkload/Production/ProductionMergeOutputA/ProductionOutputAMergeLogCollect',
                    'MinFileBased',
                    'LogCollect'),
                   (3,
                    '/TestWorkload/Production/ProductionMergeOutputB/merged-logArchive',
                    '/TestWorkload/Production/ProductionMergeOutputB/ProductionOutputBMergeLogCollect',
                    'MinFileBased',
                    'LogCollect'),
                   (8,
                    '/TestWorkload/Production/unmerged-logArchive',
                    '/TestWorkload/Production/LogCollect',
                    'MinFileBased',
                    'LogCollect'),
                   (7,
                    '/TestWorkload/Production/unmerged-OutputARECO',
                    '/TestWorkload/Production/ProductionCleanupUnmergedOutputA',
                    'SiblingProcessingBased',
                    'Cleanup'),
                   (5,
                    '/TestWorkload/Production/unmerged-OutputARECO',
                    '/TestWorkload/Production/ProductionMergeOutputA',
                    'ParentlessMergeBySize',
                    'Merge'),
                   (4,
                    '/TestWorkload/Production/unmerged-OutputBUSER',
                    '/TestWorkload/Production/ProductionCleanupUnmergedOutputB',
                    'SiblingProcessingBased',
                    'Cleanup'),
                   (2,
                    '/TestWorkload/Production/unmerged-OutputBUSER',
                    '/TestWorkload/Production/ProductionMergeOutputB',
                    'ParentlessMergeBySize',
                    'Merge')]

        testArguments = MonteCarloWorkloadFactory.getTestArguments()
        testArguments["CouchURL"] = os.environ["COUCHURL"]
        testArguments["CouchDBName"] = TEST_DB_NAME
        testArguments["ConfigCacheID"] = self.injectMonteCarloConfig()

        factory = MonteCarloWorkloadFactory()
        testWorkload = factory.factoryWorkloadConstruction("TestWorkload", testArguments)

        myMask = Mask(FirstRun=1, FirstLumi=1, FirstEvent=1, LastRun=1, LastLumi=10, LastEvent=1000)
        testWMBSHelper = WMBSHelper(testWorkload, "Production", mask=myMask,
                                    cachepath=self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        self.assertItemsEqual(testWorkload.listOutputProducingTasks(), expOutTasks)

        workflows = self.listTasksByWorkflow.execute(workflow="TestWorkload")
        self.assertItemsEqual([item['task'] for item in workflows], expWfTasks)

        # same function as in WMBSHelper, otherwise we cannot know which fileset name is
        maskString = ",".join(["%s=%s" % (x, myMask[x]) for x in sorted(myMask)])
        topFilesetName = 'TestWorkload-Production-%s' % md5(maskString).hexdigest()
        expFsets[0] = topFilesetName
        # returns a tuple of id, name, open and last_update
        filesets = self.listFilesets.execute()
        self.assertItemsEqual([item[1] for item in filesets], expFsets)

        subMaps[0] = (1, topFilesetName, '/TestWorkload/Production', 'EventBased', 'Production')
        subscriptions = self.listSubsMapping.execute(workflow="TestWorkload", returnTuple=True)
        self.assertItemsEqual(subscriptions, subMaps)

        ### create another top level subscription
        myMask = Mask(FirstRun=1, FirstLumi=11, FirstEvent=1001, LastRun=1, LastLumi=20, LastEvent=2000)
        testWMBSHelper = WMBSHelper(testWorkload, "Production", mask=myMask,
                                    cachepath=self.testInit.testDir)
        testWMBSHelper.createTopLevelFileset()
        testWMBSHelper._createSubscriptionsInWMBS(testWMBSHelper.topLevelTask, testWMBSHelper.topLevelFileset)

        workflows = self.listTasksByWorkflow.execute(workflow="TestWorkload")
#.........這裏部分代碼省略.........

 def __init__(self):
     MonteCarloWorkloadFactory.__init__(self)

 def __init__(self):
     MonteCarloWorkloadFactory.__init__(self)
     self.lheInputFiles = False

 def __call__(self, workflowName, args):
     workload = MonteCarloWorkloadFactory.__call__(self, workflowName, args)
     #delattr(workload.taskIterator().next().steps().data.application.configuration,
     #        'configCacheUrl')
     return workload