'--max',
                    type=int,
                    help='Maximum number of events to process',
                    required=True)
parser.add_argument(
    '-c',
    '--check',
    type=str,
    help=
    'Check the given path and avoids to send the outputs already been processed',
    required=False,
    default='')

args = parser.parse_args()

config = Configuration()

config.sbatch_partition = 'cp3'
config.sbatch_qos = 'cp3'
#config.sbatch_workdir = '.'
config.sbatch_time = '0-03:00:00'
#config.sbatch_mem = '2048'
#config.sbatch_additionalOptions = []
config.inputSandboxContent = []  #['confs/*']
config.useJobArray = True
config.inputParamsNames = ['from', 'to', 'input', 'output']
config.inputParams = []

config.payload = """
{executable_path} --from ${{from}} --to ${{to}} --input ${{input}} --output ${{output}} 
"""
Exemplo n.º 2
0
def SlurmRunNano(path= None, outputDIR=None):
    config = Configuration()
    config.sbatch_partition = 'cp3'
    config.sbatch_qos = 'cp3'
    config.cmsswDir = os.path.dirname(os.path.abspath(__file__))
    config.sbatch_chdir = os.path.join(config.cmsswDir, outputDIR)
    config.sbatch_time = '0-06:00'
    sbatch_memPerCPU = '2000'
    #config.environmentType = 'cms'
    config.inputSandboxContent = ["gridpackTolheToNanoGen.sh"]
    config.stageoutFiles = ['*.root']
    config.stageoutDir = config.sbatch_chdir
    config.inputParamsNames = ["gridpack_path","NanoGEN"]
    config.inputParams = []
    
    for gridpack_path in glob.glob(os.path.join(os.path.dirname(os.path.abspath(__file__)), path, "*_tarball.tar.xz")):
        config.inputParams.append([gridpack_path, "%s%s"%(gridpack_path.split('/')[-1].split('_slc7')[0], ".root")])
        workDIR=os.path.dirname(os.path.abspath(__file__))
    config.payload = \
    """
            if [[ "${NanoGEN}" == *"200p00_125p00"* ]]; then
                eval suffix="lowmass_"
            else
                eval suffix=""
            fi

            if [[ "${NanoGEN}" == *"bbH"* ]]; then
                eval fragment="Hadronizer_TuneCP5_13TeV_aMCatNLO_2p_LHE_pythia8_cff.py"
            elif [[ "${NanoGEN}" == *"AToZHTo2L2B"* ]]; then
                eval fragment="Hadronizer_TuneCP5_13TeV_AToZHTo2L2B_${suffix}pythia8_PSweights_cff.py"
            else
                eval fragment="Hadronizer_TuneCP5_13TeV_HToZATo2L2B_${suffix}pythia8_PSweights_cff.py"
            fi
            pwd
            echo ${gridpack_path}
            echo ${NanoGEN}
            echo ${SLURM_ARRAY_JOB_ID}_${SLURM_ARRAY_TASK_ID}
            echo ${fragment}
            echo "****************************"
            cat ${workDIR}/python/${fragment}
            echo "****************************"
            bash gridpackTolheToNanoGen.sh ${fragment} ${NanoGEN} ${gridpack_path}
    """
    submitWorker = SubmitWorker(config, submit=True, yes=True, debug=True, quiet=True)
    submitWorker()
Exemplo n.º 3
0
def submit_on_slurm(name, args, debug=False):
    # Check arguments #
    GPU = args.find("--GPU") != -1
    output = args.find("--output") != -1

    config = Configuration()
    config.sbatch_partition = parameters.partition
    config.sbatch_qos = parameters.QOS
    config.sbatch_workdir = parameters.main_path
    config.sbatch_time = parameters.time
    #config.sbatch_mem = parameters.mem
    config.sbatch_additionalOptions = ['-n ' + str(parameters.tasks)]
    if GPU:
        config.sbatch_additionalOptions += ['--gres gpu:1', '--export=NONE']
    config.inputSandboxContent = []
    config.useJobArray = True
    config.inputParamsNames = []
    config.inputParams = []
    if output:
        config.inputParamsNames += ["--verbose"]
        config.inputParams += [[""]]
    if not output:
        config.inputParamsNames += ['scan', 'task']

    config.payload = """ """

    if GPU:
        config.payload += "export PYTHONPATH=/root6/lib:$PYTHONPATH\n"
        config.payload += "module load cp3\n"  # needed on gpu to load slurm_utils
        config.payload += "module load slurm/slurm_utils\n"
    config.payload += "python3 {script} "
    if not output:
        config.payload += "--scan ${{scan}} --task ${{task}} "
    config.payload += args

    timestamp = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
    out_dir = parameters.main_path

    slurm_config = copy.deepcopy(config)
    slurm_working_dir = os.path.join(out_dir, 'slurm', name + '_' + timestamp)

    slurm_config.batchScriptsDir = os.path.join(slurm_working_dir, 'scripts')
    slurm_config.inputSandboxDir = slurm_config.batchScriptsDir
    slurm_config.stageoutDir = os.path.join(slurm_working_dir, 'output')
    slurm_config.stageoutLogsDir = os.path.join(slurm_working_dir, 'logs')
    slurm_config.stageoutFiles = ["*.csv", "*.zip", "*.png"]

    slurm_config.payload = config.payload.format(
        script=os.path.join(out_dir, "ZAMachineLearning.py"))

    if not output:
        for f in glob.glob(
                os.path.join(parameters.main_path, 'split', name, '*.pkl')):
            task = os.path.basename(f)
            slurm_config.inputParams.append([name, task])

    # Submit job!

    logging.info("Submitting job...")
    if not debug:
        submitWorker = SubmitWorker(slurm_config,
                                    submit=True,
                                    yes=True,
                                    debug=False,
                                    quiet=False)
        submitWorker()
        logging.info("Done")
    else:
        logging.debug(slurm_config.payload)
        logging.debug(slurm_config.inputParamsNames)
        logging.debug(slurm_config.inputParams)
        logging.info('... don\'t worry, jobs not sent')
Exemplo n.º 4
0
def submit_on_slurm(name,args,debug=False):
    # Check arguments #
    GPU = args.find("--GPU") != -1
    output = args.find("--output") != -1

    config = Configuration()
    config.sbatch_partition = parameters.partition
    config.sbatch_qos = parameters.QOS
    config.sbatch_chdir = parameters.main_path
    config.sbatch_time = parameters.time
    config.sbatch_additionalOptions = [parameters.additional_options]
    config.sbatch_memPerCPU = parameters.mem
    if parameters.partition == 'cp3-gpu':
        config.sbatch_additionalOptions += ['--export=NONE']
        #if parameters.cpus > 1:
        #    config.sbatch_additionalOptions += ["--cpus-per-gpu={}".format(parameters.cpus)]
        #config.sbatch_additionalOptions += ['--mem-per-gpu={}'.format(parameters.mem)]
    elif parameters.partition == 'gpu':
        config.sbatch_additionalOptions += ['--gres=gpu:TeslaV100:{}'.format(parameters.gpus),'--export=NONE']
        #config.sbatch_additionalOptions += ['--mem-per-gpu={}'.format(parameters.mem)]
        #if parameters.cpus > 1:
        #    config.sbatch_additionalOptions += ["--cpus-per-gpu={}".format(parameters.cpus)]
        #if parameters.cpus > 1:
        #    config.sbatch_additionalOptions += ["--cpus-per-gpu={}".format(parameters.cpus)]
        #    config.sbatch_additionalOptions += ["--cpus-per-task={}".format(parameters.cpus)]
            #config.sbatch_additionalOptions += ["-c {}".format(parameters.cpus)]
    else:
        if parameters.tasks > 1:
            config.sbatch_additionalOptions += ["-n={}".format(parameters.tasks)]
        if parameters.cpus > 1:
            config.sbatch_additionalOptions += ["--cpus-per-task={}".format(parameters.cpus)]
        
    config.inputSandboxContent = []
    config.useJobArray = True
    config.inputParamsNames = []
    config.inputParams = []
    if output:
        config.inputParamsNames += ["--verbose"]
        config.inputParams += [[""]]
    if not output:
        config.inputParamsNames += ['scan','task']
        if parameters.crossvalidation and parameters.split_per_model:
            config.inputParamsNames += ['modelId']

    config.payload = ""

    if parameters.partition == 'cp3-gpu':
        config.payload += "export PYTHONPATH=/python3/lib/python3.6/site-packages/:$PYTHONPATH\n" # GPU tf
        config.payload += "export PYTHONPATH=/root6/lib:$PYTHONPATH\n" # ROOT
        config.payload += "module load cp3\n" # needed on gpu to load slurm_utils
        config.payload += "module load python/python36_sl7_gcc73\n" 
        config.payload += "module load slurm/slurm_utils\n"
    if parameters.partition == 'gpu':
        config.payload += "module load releases/2019b_test \n"
        config.payload += "module load cp3\n" # needed on gpu to load slurm_utils
        config.payload += "module load root/6.12.04-sl7_gcc73 \n"
        config.payload += "module load root_numpy \n"
        config.payload += "module load TensorFlow \n"
        config.payload += "module load slurm/slurm_utils\n"
        
    config.payload += "python3 {script} "
    if not output:
        config.payload += "--scan ${{scan}} --task ${{task}} "
    if parameters.crossvalidation and parameters.split_per_model:
        config.payload += "--modelId ${{modelId}}"
    config.payload += args

    timestamp = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
    out_dir = parameters.main_path

    slurm_config = copy.deepcopy(config)
    slurm_working_dir = os.path.join(out_dir,'slurm',name+'_'+timestamp)

    slurm_config.batchScriptsDir = os.path.join(slurm_working_dir, 'scripts')
    slurm_config.inputSandboxDir = slurm_config.batchScriptsDir
    slurm_config.stageoutDir = os.path.join(slurm_working_dir, 'output')
    slurm_config.stageoutLogsDir = os.path.join(slurm_working_dir, 'logs')
    slurm_config.stageoutFiles = ["*.csv","*.zip","*.png"]

    slurm_config.payload = config.payload.format(script=os.path.join(out_dir,"HHMachineLearning.py"))

    if not output:
        for f in glob.glob(os.path.join(parameters.main_path,'split',name,'*.pkl')):
            task = os.path.basename(f)
            if parameters.crossvalidation and parameters.split_per_model:
                for N in range(parameters.N_models):
                    slurm_config.inputParams.append([name,task,N])
            else:
                slurm_config.inputParams.append([name,task])

    # Submit job!

    logging.info("Submitting job...")
    if not debug:
        submitWorker = SubmitWorker(slurm_config, submit=True, yes=True, debug=False, quiet=False)
        submitWorker()
        logging.info("Done")
    else:
        logging.info("Number of jobs : %d"%len(slurm_config.inputParams))
        logging.info(slurm_config.payload)
        logging.info(slurm_config.inputParamsNames)
        for inputParam in slurm_config.inputParams:
            logging.info(inputParam)
        logging.info('... don\'t worry, jobs not sent')
Exemplo n.º 5
0
#######USER DEFINED : Number of jobs is the length of the config.json file.
import json
nnnJobs = 0
with open("config.json", "r") as f:
    jjj = json.loads(f.read())
    nnnJobs = len(jjj)

from CP3SlurmUtils.Configuration import Configuration

config = Configuration()

#--------------------------------------------------------------------------------
# 1. SLURM sbatch command options
#--------------------------------------------------------------------------------

config.sbatch_partition = 'cp3'
config.sbatch_qos = 'cp3'
config.sbatch_workdir = './slurm/'
config.sbatch_time = '0-02:30:00'
config.sbatch_mem = '1024'

config.sbatch_output = ''
config.sbatch_error = ''

# Example: config.sbatch_additionalOptions = ['--mail-type=END', '--mail-user=<your-email-address>']
config.sbatch_additionalOptions = []

#--------------------------------------------------------------------------------
# 2. User batch script parameters that are same for all jobs
#--------------------------------------------------------------------------------
Exemplo n.º 6
0
def ClusterParameterEstimator_4SLURM(yml=None,
                                     outputdir=None,
                                     task=None,
                                     isTest=False):
    config = Configuration()
    config.sbatch_partition = 'cp3'
    config.sbatch_qos = 'cp3'
    config.cmsswDir = os.path.dirname(os.path.abspath(__file__))
    config.sbatch_chdir = os.path.join(config.cmsswDir, outputdir)
    config.sbatch_time = '0-02:00'
    config.sbatch_memPerCPU = '2000'
    config.batchScriptsFilename = "slurmSubmission.sh"
    #config.environmentType = 'cms'
    config.inputSandboxContent = [
        "skimProducer.py" if task == "skim" else
        ("SiStripHitResol.py" if task == "hitresolution" else
         ("CPEstimator.py"))
    ]
    config.stageoutFiles = ['*.root']
    config.stageoutDir = config.sbatch_chdir
    config.inputParamsNames = ["inputFiles", "outputFile", "task", "sample"]

    analysisCfgs = os.path.join(config.cmsswDir, yml)
    config.inputParams = getTasks(task=task,
                                  analysisCfgs=analysisCfgs,
                                  cmsswDir=config.cmsswDir,
                                  stageoutDir=config.stageoutDir,
                                  isTest=isTest)
    shutil.copyfile(analysisCfgs, config.stageoutDir + "/analysis.yml")
    config.payload = \
    """
    echo ${SLURM_ARRAY_JOB_ID}_${SLURM_ARRAY_TASK_ID}
    if [[ "$task" == *"skim"* ]]; then
        cmsRun skimProducer.py inputFiles=${inputFiles} outputFile=${outputFile}
    elif [[ "$task" == "hitresolution" ]]; then
        cmsRun SiStripHitResol.py inputFiles=${inputFiles} outputFile=${outputFile}
    else
        cmsRun CPEstimator.py inputFiles=${inputFiles} outputFile=${outputFile}
    fi
    """

    submitWorker = SubmitWorker(config,
                                submit=True,
                                yes=True,
                                debug=True,
                                quiet=True)
    submitWorker()
    logger.warning(
        'Work still in progress for better workflow ...\n'
        'To hadd files and produce plots. Please run as follow when the jobs finish running\n'
        'python postprocessing.py --workdir {}\n'
        'squeue -u user_name  : allows you to check your submitted jobs status\n'
        .format(outputdir))