def submit_jobs(output_directory, tag, samples, events_total,
events_per_output, tarfile):
# Long loop to monitor jobs and resubmit failed jobs
for i in range(10000):
total_summary = {}
# Loop through samples
for sample in samples:
config_info = samples_config[sample]
executable = create_executable(tag, sample, config_info)
task = CondorTask(sample=DummySample(
N=int(float(events_total) / float(events_per_output)),
nevents=events_total,
dataset="/" + sample + "_NANO"),
tag=tag,
special_dir=output_directory,
events_per_output=events_per_output,
total_nevents=events_total,
split_within_files=True,
executable=executable,
open_dataset=False,
tarfile=tarfile,
condor_submit_params=CONDOR_SUBMIT_PARAMS)
task.process()
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/Hgg-MC-Production/").do()
time.sleep(300) # power nap
def submit(which):
total_summary = {}
extra_requirements = "true"
tag = "v0_noFilter"
pdname = "HVNoFilter"
events_per_point = int(1E4)
events_per_job = int(100)
cfgsDir = "psets_gensim_noFilter/"
modelsFile = cfgsDir + "/models.txt"
df = pd.read_csv(modelsFile)
for year in ["2016", "2017", "2018"]:
for iterr, row in df.iterrows():
# fmass = float(mass)
# mass = str(mass).replace(".","p")
epp = int(events_per_point)
reqname = "noFilter_m{}{}_ctau{}_xi_{}_{}_{}".format(
row.portal, row.mass, row.ctau, row.xi, tag, year)
njobs = epp // events_per_job
sample = DummySample(
dataset="/{}/params_{}_m_{}_ctau_{}mm_xi_{}_{}/LLPNTUPLE".
format(pdname, row.portal, row.mass, row.ctau * 10, row.xi,
year),
N=njobs,
nevents=epp)
task = CondorTask(
sample=sample,
output_name="output.root",
executable="executables/condor_executable_{}.sh".format(which),
tarfile="package_{}.tar.xz".format(year),
open_dataset=True,
files_per_output=1,
condor_submit_params={
"classads": [
["param_mass", row.mass],
["param_ctau", row.ctau],
["param_xi", str(row.xi).replace(".", "p")],
["param_portal", row.portal],
["param_year", year],
["param_nevents", events_per_job],
["metis_extraargs", ""],
["JobBatchName", reqname],
],
"requirements_line":
'Requirements = ((HAS_SINGULARITY=?=True) && (HAS_CVMFS_cms_cern_ch =?= true) && {extra_requirements})'
.format(extra_requirements=extra_requirements),
},
tag=tag,
recopy_inputs=True)
task.process()
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis_test/").do()
def submit():
requests = {
#'TTWJetsToLNuEWK_5f_NLO': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball_retired.tar.xz', # that's the SM point, but using the SMEFT model. No lepton filtering, so name is actually confusing
#'TTWJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWplusJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWplusJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWminusJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWminusJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWJetsToLNuEWK_5f_EFT_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_cpt8_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # one of the BSM points
#'TTWJetsToLNuEWK_5f_EFT_mix_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # EFT mix
'TTWJetsToLNuEWK_5f_EFT_cpq3_4_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks//TTWJetsToLNuEWK_5f_EFT_myNLO_cpq3_4_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # C_pq3 = 4
}
total_summary = {}
extra_requirements = "true"
tag = "v4"
events_per_point = 500000 # produced 500k events before
events_per_job = 2000 # up to 2000 works
#events_per_point = 50
#events_per_job = 10
njobs = int(events_per_point)//events_per_job
for reqname in requests:
gridpack = requests[reqname]
#reqname = "TTWJetsToLNuEWK_5f_EFT_myNLO"
#gridpack = '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz'
task = CondorTask(
sample = DummySample(dataset="/%s/RunIIAutumn18/NANO"%reqname,N=njobs,nevents=int(events_per_point)),
output_name = "nanoAOD.root",
executable = "executables/condor_executable_Autumn18.sh",
tarfile = "package.tar.gz",
#scram_arch = "slc7_amd64_gcc630",
open_dataset = False,
files_per_output = 1,
arguments = gridpack,
condor_submit_params = {
"sites":"T2_US_UCSD", #
"classads": [
["param_nevents",events_per_job],
["metis_extraargs",""],
["JobBatchName",reqname],
#["SingularityImage", "/cvmfs/singularity.opensciencegrid.org/cmssw/cms:rhel6-m202006"],
],
"requirements_line": 'Requirements = (HAS_SINGULARITY=?=True)' # && (HAS_CVMFS_cms_cern_ch =?= true) && {extra_requirements})'.format(extra_requirements=extra_requirements),
},
tag = tag,
min_completion_fraction = 0.90,
)
task.process()
total_summary[task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary, webdir="~/public_html/dump/tW_gen/").do()
def submit():
requests = {
'TTWJetsToLNuEWK_5f_EFT_myNLO': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # that's EFT
'TTWJetsToLNuEWK_5f_EFT_myNLO_cpt8': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_cpt8_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # that's EFT
'TTWJetsToLNuEWK_5f_NLO': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # that's the SM
}
total_summary = {}
extra_requirements = "true"
tag = "v1"
events_per_point = 500000
events_per_job = 10000
#events_per_point = 500
#events_per_job = 100
njobs = int(events_per_point)//events_per_job
for reqname in requests:
gridpack = requests[reqname]
#reqname = "TTWJetsToLNuEWK_5f_EFT_myNLO"
#gridpack = '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz'
task = CondorTask(
sample = DummySample(dataset="/%s/RunIIAutumn18/NANOGEN"%reqname,N=njobs,nevents=int(events_per_point)),
output_name = "output.root",
executable = "executables/condor_executable.sh",
tarfile = "package.tar.gz",
open_dataset = False,
files_per_output = 1,
arguments = gridpack,
condor_submit_params = {
"sites":"T2_US_UCSD", #
"classads": [
["param_nevents",events_per_job],
["metis_extraargs",""],
["JobBatchName",reqname],
],
"requirements_line": 'Requirements = ((HAS_SINGULARITY=?=True) && (HAS_CVMFS_cms_cern_ch =?= true) && {extra_requirements})'.format(extra_requirements=extra_requirements),
},
tag = tag,
)
task.process()
total_summary[task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary, webdir="~/public_html/dump/tW_gen/").do()
def test1():
dataset_names = [
# "/TT_TuneCUETP8M2T4_13TeV-powheg-pythia8/RunIISummer17MiniAOD-92X_upgrade2017_realistic_v10_ext1-v1/MINIAODSIM",
"/Dummy_test_StopBabyMaker_v25/CMS4",
]
basedir = "/hadoop/cms/store/user/{0}/metis_test/example/".format(
os.getenv("USER"))
for i in range(10000):
total_summary = {}
for dsname in dataset_names:
try:
task = CondorTask(
sample=DirectorySample(
location=basedir,
dataset=dsname,
),
files_per_output=1,
output_name="merged_ntuple.root",
tag="v25",
scram_arch="slc6_amd64_gcc700",
executable="condor_executable.sh",
arguments="testarg1",
tarfile="input.tar.gz",
condor_submit_params={"sites": "UAF,T2_US_UCSD,UCSB"},
no_load_from_backup=
True, # for the purpose of the example, don't use a backup
)
task.process()
except:
traceback_string = traceback.format_exc()
print("Runtime error:\n{0}".format(traceback_string))
# send_email(subject="metis error", body=traceback_string)
total_summary[dsname] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis/").do()
time.sleep(5)
def submit(which):
total_summary = {}
tag = "ZMuMuSkim_v0"
task = CMSSWTask(
sample = DBSSample(dataset="/SingleMuon/Run2017F-ZMu-17Nov2017-v1/RAW-RECO"),
events_per_output = 10000,
pset = "/home/users/mcitron/CMSSW_9_4_17/src/cms_lpc_llp/llp_ntupler/python/displacedJetMuon_ntupler_Data_2017_RAWRECO.py",
cmssw_version = "CMSSW_9_4_17",
scram_arch = "slc6_amd64_gcc700",
tag = tag,
# dont_check_tree = True,
executable = "/home/users/mcitron/exeForMetis/condor_genprod_exe.sh",
condor_submit_params = {
"container": "/cvmfs/singularity.opensciencegrid.org/cmssw/cms:rhel6-m202006",
},
recopy_inputs=True
)
task.process()
total_summary[task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary, webdir="~/public_html/dump/metis_test/").do()
merge_task.reset_io_mapping()
merge_task.update_mapping()
merge_task.process()
# save some information for the dashboard
total_summary[maker_task.get_sample().get_datasetname(
)] = maker_task.get_task_summary()
total_summary[merge_task.get_sample().get_datasetname(
)] = merge_task.get_task_summary()
# Aggregate whether all tasks are complete
all_tasks_complete = all_tasks_complete and maker_task.complete(
) and merge_task.complete()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
os.system("msummary -s name -r | tee summary.txt")
os.system("chmod -R 755 {}".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
def submit_metis(job_tag, samples_map, sample_list=[], arguments_map="", exec_script="metis.sh", tar_files=[], hadoop_dirname="testjobs", files_per_output=1, globber="*.root", sites="T2_US_UCSD"):
import time
import json
import metis
from time import sleep
from metis.Sample import DirectorySample
from metis.CondorTask import CondorTask
from metis.StatsParser import StatsParser
import os
import glob
import subprocess
# file/dir paths
main_dir = os.getcwd()
metis_path = os.path.dirname(os.path.dirname(metis.__file__))
tar_path = os.path.join(metis_path, "package.tar")
tar_gz_path = tar_path + ".gz"
metis_dashboard_path = os.path.join(metis_path, "dashboard")
exec_path = os.path.join(main_dir, exec_script)
hadoop_path = "metis/{}/{}".format(hadoop_dirname, job_tag) # The output goes to /hadoop/cms/store/user/$USER/"hadoop_path"
# Create tarball
os.chdir(main_dir)
print os.getcwd()
print "tar -chzf {} {}".format(tar_gz_path, " ".join(tar_files))
os.system("tar -chzf {} {}".format(tar_gz_path, " ".join(tar_files)))
# Change directory to metis
os.chdir(metis_path)
total_summary = {}
# if no sample_list is provided then we form it via the keys of the samples_map
if len(sample_list) == 0:
for key in samples_map:
sample_list.append(key)
samples_to_run = []
for key in sample_list:
samples_to_run.append(
DirectorySample(
dataset=key,
location=samples_map[key],
globber=globber,
)
)
files_per_output_config_list = []
if isinstance(files_per_output, dict):
for key in sample_list:
files_per_output_config_list.append(files_per_output[key])
else:
for key in sample_list:
files_per_output_config_list.append(files_per_output)
# Loop over datasets to submit
while True:
all_tasks_complete = True
#for sample in sorted(samples):
for index, sample in enumerate(samples_to_run):
# define the task
maker_task = CondorTask(
sample = sample,
tag = job_tag,
arguments = arguments_map[sample.get_datasetname()] if arguments_map else "",
executable = exec_path,
tarfile = tar_gz_path,
special_dir = hadoop_path,
output_name = "output.root",
files_per_output = files_per_output_config_list[index],
condor_submit_params = {"sites" : sites},
open_dataset = False,
flush = True,
#no_load_from_backup = True,
)
# process the job (either submits, checks for resubmit, or finishes etc.)
maker_task.process()
# save some information for the dashboard
total_summary["["+job_tag+"] "+maker_task.get_sample().get_datasetname()] = maker_task.get_task_summary()
# Aggregate whether all tasks are complete
all_tasks_complete = all_tasks_complete and maker_task.complete()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
os.system("msummary -r -p {} | tee summary.txt".format(job_tag))
os.system("chmod -R 755 {}".format(metis_dashboard_path))
os.system("chmod 644 {}/images/*".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
for i in range(0,60):
sleep(1) # could use a backward counter to be preeety :)
except KeyboardInterrupt:
raw_input("Press Enter to force update, or Ctrl-C to quit.")
print "Force updating..."
os.chdir(main_dir)
output_is_tree=True,
tag=tag,
condor_submit_params={"sites": "T2_US_UCSD,UAF"},
cmssw_version="CMSSW_10_2_9",
scram_arch="slc6_amd64_gcc700",
min_completion_fraction=1.00,
)
maker_tasks.append(maker_task)
if not args.dryRun:
for i in range(100):
total_summary = {}
#for maker_task, merge_task in zip(maker_tasks,merge_tasks):
for maker_task in maker_tasks:
maker_task.process()
frac = maker_task.complete(return_fraction=True)
total_summary[maker_task.get_sample().get_datasetname(
)] = maker_task.get_task_summary()
print(frac)
# parse the total summary and write out the dashboard
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis_tW_scattering/").do()
# 15 min power nap
time.sleep(15. * 60)
def submit():
requests = {
#'TTWJetsToLNuEWK_5f_NLO': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball_retired.tar.xz', # that's the SM point, but using the SMEFT model. No lepton filtering, so name is actually confusing
#'TTWJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWplusJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWplusJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWminusJetsToLNuEWK_5f_NLO_v2': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWminusJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # that's the actual SM
#'TTWJetsToLNuEWK_5f_EFT_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_cpt8_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # one of the BSM points
#'TTWJetsToLNuEWK_5f_EFT_mix_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc6_amd64_gcc630_CMSSW_9_3_16_tarball.tar.xz', # EFT mix
#'TTWJetsToLNuEWK_5f_EFT_cpq3_4_myNLO_full': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks//TTWJetsToLNuEWK_5f_EFT_myNLO_cpq3_4_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz', # C_pq3 = 4
#'TTWJetsToLNuEWK_5f_NLO': '/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_NLO_slc7_amd64_gcc730_CMSSW_9_3_16_tarball.tar.xz',
'TTWJetsToLNuEWK_5f_SMEFTatNLO_weight':
'/hadoop/cms/store/user/dspitzba/tW_scattering/gridpacks/TTWJetsToLNuEWK_5f_EFT_myNLO_slc7_amd64_gcc700_CMSSW_10_6_19_tarball.tar.xz',
}
total_summary = {}
extra_requirements = "true"
# v6+ is UL
tag = "UL18_v7"
#events_per_point = 250000
#events_per_job = 250
#events_per_point = 2000000
events_per_point = 4000000
events_per_job = 2000
njobs = int(events_per_point) // events_per_job
for reqname in requests:
gridpack = requests[reqname]
task = CondorTask(
sample=DummySample(dataset="/%s/RunIIAutumn18/NANO" % reqname,
N=njobs,
nevents=int(events_per_point)),
output_name="nanoAOD.root",
executable="executables/condor_executable_UL18.sh",
#executable = "executables/condor_executable_UL17.sh",
#executable = "executables/condor_executable_UL16_postVFP.sh",
tarfile="package.tar.gz",
additional_input_files=[gridpack],
open_dataset=False,
files_per_output=1,
arguments=gridpack.split('/')[-1],
condor_submit_params={
"sites":
"T2_US_UCSD", #
#"memory": 1950,
#"cpus": 1,
"memory":
15600,
"cpus":
8,
"classads": [
["param_nevents", events_per_job],
["metis_extraargs", ""],
["JobBatchName", reqname],
["IS_CLOUD_JOB", "yes"],
],
"requirements_line":
'Requirements = (HAS_SINGULARITY=?=True)'
},
tag=tag,
min_completion_fraction=0.90,
)
task.process()
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/tW_gen/").do()
tag=tag_mc,
extra_args="--year {}".format(year)))
# # test 3/4 body
# locations = []
# locations += glob.glob("/hadoop/cms/store/user/namin/testscoutingmc/BToPhi_params_year*_m*_ctau*mm_RAWSIM_v*/")
# for location in locations:
# tag_mc = location.rsplit("_",1)[-1].replace("/","")
# taskname = location.rstrip("/").rsplit("/")[-1]
# dataset = "/{}/{}/BABY".format(
# taskname.split("_",1)[0],
# taskname.split("_",1)[1].split("_RAWSIM")[0],
# )
# year = 2018
# if "year2017" in dataset:
# year = 2017
# infos.append(dict(location=location, dataset=dataset, isdata=False, tag=tag_mc, extra_args="--year {}".format(year)))
tasks = get_tasks(infos)
for _ in range(500):
total_summary = {}
for task in tasks:
task.process()
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/scouting/").do()
# time.sleep(30*60)
time.sleep(4 * 60 * 60)
recopy_inputs=False,
)
merge_task = LocalMergeTask(
input_filenames=task.get_outputs(),
output_filename="{}/{}.root".format(merged_dir, shortname),
ignore_bad=skip_tail,
)
nsamples += 1
if not task.complete():
task.additional_input_files = [
"/home/users/namin/2018/fourtop/all/FTAnalysis/babymaking/batch/condor_chirp"
]
task.process()
else:
if not merge_task.complete():
merge_task.process()
else:
ndone += 1
total_summary[dsname.split("|")[0]] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/FTmetis/").do(
custom_event_rate_parser=get_event_rate, )
if ndone == nsamples:
print "All done!"
sys.exit()
time.sleep(600 if i < 10 else 1.5 * 60 * 60)
cmssw_version = cmsswver,
tarfile = tarfile,
special_dir = "run2_moriond17_cms4/ProjectMetis",
# min_completion_fraction = 0.94,
publish_to_dis = True,
sparms = sparms,
)
tasks.append(task)
return tasks
if __name__ == "__main__":
for i in range(10000):
total_summary = {}
total_counts = {}
tasks = []
tasks.extend(get_tasks())
for task in tasks:
dsname = task.get_sample().get_datasetname()
try:
if not task.complete():
task.process()
except:
traceback_string = traceback.format_exc()
print "Runtime error:\n{0}".format(traceback_string)
send_email(subject="metis error", body=traceback_string)
total_summary[dsname] = task.get_task_summary()
StatsParser(data=total_summary, webdir="~/public_html/dump/metis/", make_plots=False).do()
time.sleep(1.*3600)
tag=tag,
# condor_submit_params = {"sites": "T2_US_UCSD", "use_xrootd":True, "classads": [ ["metis_extraargs", "fetch_nano"] ]},
condor_submit_params={
"sites": "T2_US_UCSD",
"use_xrootd": True
},
cmssw_version="CMSSW_10_2_13",
scram_arch="slc7_amd64_gcc700",
input_executable=
"condor_executable_metis.sh", # your condor executable here
tarfile=tarfile, # your tarfile with assorted goodies here
special_dir="VBSHWWNanoSkim/{}".format(
tag
), # output files into /hadoop/cms/store/<user>/<special_dir>
)
# Straightforward logic
if not task.complete():
task.process()
# Set task summary
task_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
# Parse the summary and make a summary.txt that will be used to pretty status of the jobs
os.system("rm -f web_summary.json")
os.system("rm -f summary.json")
webdir = "~/public_html/VBSHWWNanoSkimmerDashboard"
StatsParser(data=task_summary, webdir=webdir).do()
os.system("chmod -R 755 {}".format(webdir))
os.system("msummary -r -i {}/web_summary.json".format(webdir))
while True:
allcomplete = True
for ds, fpo, args in dsdefs[:]:
sample = DBSSample(dataset=ds)
task = CondorTask(sample=sample,
open_dataset=False,
files_per_output=fpo,
output_name="test_skim.root"
if DOSKIM else "myMicroAODOutputFile.root",
tag=job_tag,
executable=exec_path,
tarfile=tar_path,
condor_submit_params={"sites": "T2_US_UCSD"},
special_dir=hadoop_path,
arguments=args.replace(" ", "|"))
task.process()
allcomplete = allcomplete and task.complete()
# save some information for the dashboard
total_summary[ds] = task.get_task_summary()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis_microaod_80x/").do()
os.system("chmod -R 755 ~/public_html/dump/metis_microaod_80x")
if allcomplete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
print "Sleeping 1800 seconds ..."
time.sleep(1800)
tag=job_tag,
cmssw_version="CMSSW_10_2_22", # doesn't do anything
arguments=job_args,
executable=exec_path,
tarfile=tar_path,
condor_submit_params={
"sites":
"T2_US_UCSD",
"container":
"/cvmfs/singularity.opensciencegrid.org/cmssw/cms:rhel7-m20201010"
})
attrs = vars(task)
# check attrs
print(', '.join("%s: %s" % item for item in attrs.items()))
task.process()
total_summary = {}
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
# Web dashboard
StatsParser(data=total_summary, webdir="~/public_html/v4/").do()
os.system("chmod -R 755 ~/public_html/v4/")
print("Sleeping 10seconds ...")
time.sleep(10)
# Straightforward logic
if not task.complete():
task.process()
else:
if not merge_task.complete():
merge_task.process()
# Aggregate whether all tasks are complete
all_tasks_complete = all_tasks_complete and task.complete()
# Set task summary
task_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
# Parse the summary and make a summary.txt that will be used to pretty status of the jobs
StatsParser(data=task_summary,
webdir="~/public_html/VVVNanoLooperDashboard").do()
os.system("chmod -R 755 ~/public_html/VVVNanoLooperDashboard")
os.system("msummary -r | tee summary.txt")
# If all done exit the loop
if all_tasks_complete:
print("")
print("All job finished")
print("")
break
# Neat trick to not exit the script for force updating
print(
'Press Ctrl-C to force update, otherwise will sleep for 600 seconds'
)
try:
def submit(dinfos, version_tag, dotestrun=False, files_per_output_func=UNITY):
# Loop
while True:
tasks = [] # List to hold all the Metis Tasks
total_summary = {} # Summary for task status report to access via http
# Boolean to aggregate all the task compelete status
all_tasks_complete = True
# Loop over all the campaigns
for campaign in dinfos:
create_tar_ball(dinfos[campaign]["year"], dinfos[campaign]["baby_type"], version_tag) # Create tar ball
# Get all the tasks for this campaign
this_set_of_tasks = get_tasks(
samples_dictionary = dinfos[campaign]["samples"],
year = dinfos[campaign]["year"],
baby_type = dinfos[campaign]["baby_type"],
baby_version_tag = version_tag,
dotestrun = dotestrun,
files_per_output_func = files_per_output_func,
)
# Need separate dictionaries for each tag in order to let StatsParsers uniquely handle same dataset name jobs.
if len(this_set_of_tasks) > 0:
print this_set_of_tasks[0].tag
total_summary[this_set_of_tasks[0].tag] = {}
# Add to the master list of tasks
tasks += this_set_of_tasks
# Loop over all the tasks
for task in tasks:
# Process the task (submit or resubmit)
task.process()
# save some information for the dashboard
total_summary[task.tag][task.get_sample().get_datasetname()] = task.get_task_summary()
# Aggregate task complete booleans
all_tasks_complete = all_tasks_complete and task.complete()
# parse the total summary and write out the dashboard
#metis_dashboard_path = os.path.join(os.path.dirname(os.path.dirname(metis.__file__)), "dashboard")
metis_dashboard_path = "~/public_html/dump"
# Go through each tag and do StatsParser
for tag in total_summary:
StatsParser(data=total_summary[tag], webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
#os.system("msummary -r -p MAKER | tee summary.txt")
os.system("msummary -r | tee summary.txt")
os.system("chmod -R 755 {}".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "All job finished"
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
for i in range(0,300):
sleep(1) # could use a backward counter to be preeety :)
except KeyboardInterrupt:
raw_input("Press Enter to force update, or Ctrl-C to quit.")
print "Force updating..."
condor_submit_params={"sites": "T2_US_UCSD"} if
(args.use_wget) else {
"sites":
"T2_US_UCSD,T2_US_CALTECH,T2_US_MIT,T2_US_WISCONSIN,T2_US_Nebraska,T2_US_Vanderbilt,T2_US_Florida,T3_US_UCR,T3_US_OSG,T3_US_Baylor,T3_US_Colorado,T3_US_Rice,T3_US_PuertoRico,T3_US_Cornell,T3_US_FIT,T3_US_FIU,T3_US_OSU,T3_US_Rutgers,T3_US_TAMU,T3_US_TAMU,T3_US_TTU,T3_US_UCD,T3_US_UMD,T3_US_UMiss"
},
special_dir=hadoop_path,
arguments=conds_dict[year])
task.process()
if not task.complete():
allcomplete = False
total_summary[dataset] = task.get_task_summary()
if first_run:
print "Sleeping 30s to space out jobs"
os.system("sleep 30s")
StatsParser(data=total_summary,
summary_fname="summary_%s.json" % job_tag,
webdir="~/public_html/dump/ttH_BabyMaker_v2_%s/" %
job_tag).do()
os.system("chmod -R 755 ~/public_html/dump/ttH_BabyMaker_v2_%s" % job_tag)
if allcomplete:
print("")
print("Job={} finished".format(job_tag))
print("")
break
first_run = False
print("Sleeping 1000 seconds ...")
time.sleep(1000)
signal_strings = [
"ttHJetToGG", "ttHToGG", "THQ", "THW", "VBF", "GluGluHToGG", "VHToGG",
"FCNC", "ttHiggs"
]
def main():
if data_year == "2016":
samples_to_run = dataset.hww_samples_to_run_2016
samples_short_name = dataset.samples_short_name_2016
dslocs = dataset.dslocscms4_2016_allpf
# file/dir paths
main_dir = os.path.dirname(os.path.abspath(__file__))
metis_path = os.path.dirname(os.path.dirname(metis.__file__))
tar_path = os.path.join(metis_path, "package.tar")
tar_gz_path = tar_path + ".gz"
metis_dashboard_path = os.path.join(metis_path, "dashboard")
exec_path = os.path.join(main_dir, "metis.sh")
merge_exec_path = os.path.join(main_dir, "merge.sh")
hadoop_path = "metis/baby/{}".format(
job_tag
) # The output goes to /hadoop/cms/store/user/$USER/"hadoop_path"
if job_tag.find("HWW") != -1:
args = "0" # HWWBaby
# Create tarball
os.chdir(main_dir)
os.system(
"tar -chzf {} localsetup.sh processBaby *.so *.pcm rooutil/lib*.so coreutil/data coreutil/lib*.so *.txt btagsf MVAinput jetCorrections leptonSFs puWeight2016.root pileup_jul21_nominalUpDown.root ../CORE/Tools/ mergeHadoopFiles.C rooutil/hadd.py fastjet/fastjet-install/lib"
.format(tar_gz_path))
# Change directory to metis
os.chdir(metis_path)
total_summary = {}
# Loop over datasets to submit
while True:
all_tasks_complete = True
for sample in samples_to_run:
loc = dslocs[sample]
# define the task
maker_sample_name = "/MAKER_" + sample[1:]
maker_task = CondorTask(
sample=DirectorySample(dataset=maker_sample_name,
location=loc),
tag=job_tag,
arguments=args,
executable=exec_path,
tarfile=tar_gz_path,
special_dir=hadoop_path,
output_name="output.root",
files_per_output=1,
condor_submit_params={"sites": "T2_US_UCSD"},
open_dataset=False,
flush=True,
#min_completion_fraction = 0.5,
#no_load_from_backup = True,
)
# process the job (either submits, checks for resubmit, or finishes etc.)
maker_task.process()
# save some information for the dashboard
total_summary[maker_task.get_sample().get_datasetname(
)] = maker_task.get_task_summary()
# define the task
merge_sample_name = "/MERGE_" + sample[1:]
merge_task = CondorTask(
sample=DirectorySample(dataset=merge_sample_name,
location=maker_task.get_outputdir()),
tag=job_tag,
executable=merge_exec_path,
tarfile=tar_gz_path,
files_per_output=100000,
output_dir=maker_task.get_outputdir() + "/merged",
output_name=samples_short_name[sample] + ".root",
condor_submit_params={"sites": "T2_US_UCSD"},
open_dataset=False,
flush=True,
output_is_tree=True,
cmssw_version="CMSSW_9_2_0",
scram_arch="slc6_amd64_gcc530",
#no_load_from_backup = True,
)
if maker_task.complete():
# process the job (either submits, checks for resubmit, or finishes etc.)
merge_task.process()
# save some information for the dashboard
total_summary[merge_task.get_sample().get_datasetname(
)] = merge_task.get_task_summary()
# Aggregate whether all tasks are complete
all_tasks_complete = all_tasks_complete and maker_task.complete()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
os.system("msummary -r | tee summary.txt")
os.system("chmod -R 755 {}".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
for i in range(0, 300):
sleep(1) # could use a backward counter to be preeety :)
except KeyboardInterrupt:
raw_input("Press Enter to force update, or Ctrl-C to quit.")
print "Force updating..."
def runall(special_dir, tag, total_nevents, events_per_output, config):
for _ in range(25):
proc_tag = "v1"
#special_dir = "workflowtest/ProjectMetis"
#special_dir = "miniaod_runII/JHUSample_ttH"
cmssw_v_gensim = config["cmssw_v_gensim"]
pset_gensim = config["pset_gensim"]
scram_arch_gensim = config["scram_arch_gensim"]
cmssw_v_aodsim = config["cmssw_v_aodsim"]
pset_aodsim = config["pset_aodsim"]
scram_arch_aodsim = config["scram_arch_aodsim"]
pset_aodsim2 = config["pset_aodsim2"]
cmssw_v_aodsim2 = cmssw_v_aodsim #config["pset_aodsim2"]
scram_arch_aodsim2 = scram_arch_aodsim #config["scram_arch_aodsim2"]
cmssw_v_miniaodsim = config["cmssw_v_miniaodsim"]
pset_miniaodsim = config["pset_miniaodsim"]
scram_arch_miniaodsim = config["scram_arch_miniaodsim"]
step1 = CMSSWTask(
# Change dataset to something more meaningful (but keep STEP1, as we use this
# for string replacement later); keep N=1
sample=DummySample(N=1, dataset="/" + tag + "_STEP1"),
# A unique identifier
tag=proc_tag,
special_dir=special_dir,
# Probably want to beef up the below two numbers to control splitting,
# but note that step2 is the bottleneck, so don't put too many events
# in one output file here
events_per_output=events_per_output,
total_nevents=total_nevents,
#events_per_output = 50,
#total_nevents = 1000,
# We have one input dummy file, so this must be True
split_within_files=True,
pset="psets/" + pset_gensim,
cmssw_version=cmssw_v_gensim,
scram_arch=scram_arch_gensim,
)
step2 = CMSSWTask(
sample=DirectorySample(
location=step1.get_outputdir(),
dataset=step1.get_sample().get_datasetname().replace(
"STEP1", "STEP2"),
),
tag=proc_tag,
special_dir=special_dir,
open_dataset=True,
files_per_output=1,
pset="psets/" + pset_aodsim,
cmssw_version=cmssw_v_aodsim,
scram_arch=scram_arch_aodsim,
)
step3 = CMSSWTask(
sample=DirectorySample(
location=step2.get_outputdir(),
dataset=step2.get_sample().get_datasetname().replace(
"STEP2", "STEP3"),
),
tag=proc_tag,
special_dir=special_dir,
open_dataset=True,
files_per_output=1,
pset="psets/" + pset_aodsim2,
cmssw_version=cmssw_v_aodsim2,
scram_arch=scram_arch_aodsim2,
)
step4 = CMSSWTask(
sample=DirectorySample(
location=step3.get_outputdir(),
dataset=step3.get_sample().get_datasetname().replace(
"STEP3", "STEP4"),
),
tag=proc_tag,
special_dir=special_dir,
open_dataset=True,
files_per_output=1,
output_name="step4.root",
pset="psets/" + pset_miniaodsim,
cmssw_version=cmssw_v_miniaodsim,
scram_arch=scram_arch_miniaodsim,
# condor_submit_params = {"sites":"UAF,UCSD"},
)
'''
step5 = CMSSWTask(
sample = DirectorySample(
location = step4.get_outputdir(),
dataset = step4.get_sample().get_datasetname().replace("STEP4","STEP5"),
),
tag = proc_tag,
special_dir = special_dir,
open_dataset = True,
files_per_output = 1,
pset = "psets/TOP-RunIIFall17NanoAODv7-00001_1_cfg.py",
# The below two lines should match output file names in the pset
output_name = "step5.root",
#other_outputs = ["step3_inMINIAODSIM.root","step3_inDQM.root"],
cmssw_version = "CMSSW_10_2_22",
scram_arch = "slc6_amd64_gcc700",
# condor_submit_params = {"sites":"UAF,UCSD"},
)
'''
#for _ in range(25):
total_summary = {}
for task in [step1, step2, step3, step4]:
task.process()
summary = task.get_task_summary()
total_summary[task.get_sample().get_datasetname()] = summary
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis/").do()
time.sleep(600)
files_per_output=6,
output_name="merged.root",
tag=job_tag,
#cmssw_version = "CMSSW_9_2_1", # doesn't do anything
arguments=args,
executable=exec_path,
tarfile=tar_path,
special_dir=hadoop_path,
#condor_submit_params = {"universe" : "local"}
condor_submit_params={"sites": "UAF"})
task.process()
# save some information for the dashboard
total_summary["WWW_v0_1_%s_%s" %
(baby_version, job_tag)] = task.get_task_summary()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary,
webdir="~/public_html/tasutil/Metis_WWW/").do()
if task.complete():
print ""
print "Job=%s finished" % job_tag
print ""
break
time.sleep(2 * 60)
#eof
total_summary = {}
fracs = []
for task in tasks:
#for maker_task in maker_tasks:
task.process()
frac = task.complete(return_fraction=True)
total_summary[
task.get_sample().get_datasetname()] = task.get_task_summary()
fracs.append(frac)
# parse the total summary and write out the dashboard
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis_%s/" % model).do()
print "%s/%s jobs are completed." % (sum(fracs), len(fracs))
if sum(fracs) / len(fracs) == 1:
print "Done."
break
# 60 min power nap
time.sleep(60. * 60)
for signal in signals:
#signal['spin'] = signal['name'].split('_')[0]
print signal['name']
try:
res = readResFile(signal['workspace'])
files = [f.name for f in sample.get_files()]
print(len(files))
task = CondorTask(sample=sample,
open_dataset=False,
flush=True,
files_per_output=info["fpo"],
output_name="merged_ntuple.root",
tag=job_tag,
cmssw_version=cmssw_ver,
executable=exec_path,
tarfile=tar_path,
condor_submit_params={"sites": "T2_US_UCSD"},
special_dir=hadoop_path,
arguments=info["meta_conditions"])
task.process()
if not task.complete():
allcomplete = False
# save some information for the dashboard
total_summary[dataset] = task.get_task_summary()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary,
webdir="~/public_html/dump/ttH_BabyMaker/").do()
os.system("chmod -R 755 ~/public_html/dump/ttH_BabyMaker")
if allcomplete:
print("")
print("Job={} finished".format(job_tag))
print("")
break
print("Sleeping 300 seconds ...")
time.sleep(300)
def main():
main_dir = os.path.dirname(os.path.abspath(__file__))
metis_path = os.path.dirname(os.path.dirname(metis.__file__))
exec_path = main_dir + "/metis.sh"
hadoop_path = "metis/"
metis_dashboard_path = os.path.join(metis_path, "dashboard")
job_tag = ""
total_summary = {}
os.chdir(metis_path)
while True:
masspoints = [125]
tasks = []
for mass in masspoints:
miniaod = CondorTask(
## Dummy sample as no input is needed in generating the events
#sample = DummySample(
# N=3000,
# dataset="/VHToNonbb_M125_13TeV_amcatnloFXFX_madspin_pythia8/PRIVATE-RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6_ext1-v1/MINIAODSIM"
# ),
# Dummy sample as no input is needed in generating the events
sample = DummySample(
N=3500,
dataset="/WWW_4F_TuneCP5_13TeV-amcatnlo-pythia8/PRIVATE-RunIIFall17MiniAODv2-PU2017_12Apr2018_94X_mc2017_realistic_v14-v1/MINIAODSIM"
),
tag = job_tag,
executable = exec_path,
special_dir = hadoop_path + "/private_miniaod",
output_name = "output.root",
files_per_output = 1,
condor_submit_params = {"sites" : "T2_US_UCSD"},
open_dataset = True,
flush = True,
#no_load_from_backup = True,
)
tasks.extend([miniaod])
all_tasks_complete = True
for task in tasks:
task.process()
summary = task.get_task_summary()
total_summary[task.get_sample().get_datasetname()] = summary
all_tasks_complete = all_tasks_complete and task.complete()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
os.system("msummary -v0 -p ext2 | tee summary.txt")
os.system("chmod -R 755 {}".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
for i in range(0,300):
sleep(1) # could use a backward counter to be preeety :)
except KeyboardInterrupt:
raw_input("Press Enter to force update, or Ctrl-C to quit.")
print "Force updating..."
# tasks.extend(moriondremc_isotracks.get_tasks())
# tasks.extend(data2018_102x.get_tasks()) # NOTE open dataset because WTF DAS can add files after the datset is marked as valid? NOTE remember to flush eventually
for task in tasks:
dsname = task.get_sample().get_datasetname()
try:
if not task.complete():
# condor chirp for expert monitoring
# task.additional_input_files = ["/home/users/namin/2017/ProjectMetis/metis/executables/condor_chirp"]
task.process()
except:
traceback_string = traceback.format_exc()
print "Runtime error:\n{0}".format(traceback_string)
send_email(subject="metis error", body=traceback_string)
total_summary[dsname] = task.get_task_summary()
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis/",
make_plots=False).do(show_progress_bar=True)
interruptible_sleep(3 * 3600,
reload_modules=[
data2016_94x_v2,
data2017_94x_v2,
mc2017_94x_v2,
mc2016_94x_v3,
data2018_102x,
mc2018_102x_v1,
prompt10x_data,
moriondremc,
moriondremc_isotracks,
moriondremc_private,
])
f.name for f in sample.get_files() if f.name not in corrupt_files
]
sample.set_files(files)
task = CondorTask(
sample=sample,
open_dataset=False,
flush=True,
files_per_output=1,
output_name="merged_ntuple.root",
tag=job_tag,
cmssw_version="CMSSW_9_2_1", # doesn't do anything
executable=exec_path,
tarfile=tar_path,
condor_submit_params={"sites": "T2_US_UCSD"},
special_dir=hadoop_path)
task.process()
if not task.complete():
allcomplete = False
# save some information for the dashboard
total_summary[ds] = task.get_task_summary()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir="~/public_html/dump/deepiso/").do()
os.system("chmod -R 755 ~/public_html/dump/deepiso")
if allcomplete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
print "Sleeping 300 seconds ..."
time.sleep(300)
def main():
# file/dir paths
main_dir = os.path.dirname(os.path.abspath(__file__))
metis_path = os.path.dirname(os.path.dirname(metis.__file__))
tar_path = os.path.join(metis_path, "package.tar")
tar_gz_path = tar_path + ".gz"
metis_dashboard_path = os.path.join(metis_path, "dashboard")
exec_path = os.path.join(main_dir, "metis.sh")
hadoop_path = "metis/baby/{}".format(
job_tag
) # The output goes to /hadoop/cms/store/user/$USER/"hadoop_path"
# Create tarball
os.chdir(main_dir)
os.system("tar -chzf {} a.out setup.sh *.xml".format(tar_gz_path))
# Change directory to metis
os.chdir(metis_path)
total_summary = {}
samples_to_run = [
DirectorySample(
dataset=
"/GluGluHToWWToLNuQQ_M125_13TeV_powheg_JHUGenV628_pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber="GluGluHToWWToLNuQQ_M125_13TeV_powheg_JHUGenV628*.root",
),
DirectorySample(
dataset=
"/VBFHToWWToLNuQQ_M125_13TeV_powheg_JHUGenV628_pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber="VBFHToWWToLNuQQ_M125_13TeV_powheg_JHUGenV628*.root",
),
DirectorySample(
dataset=
"/TT_TuneCUETP8M2T4_13TeV-powheg-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber="TT_TuneCUETP8M2T4_13TeV-powheg*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber="WJetsToLNu_TuneCUETP8M1_13TeV-madgraphMLM*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_HT-400To600_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6_ext1-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber=
"WJetsToLNu_HT-400To600_TuneCUETP8M1_13TeV-madgraphMLM_ext1*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_HT-600To800_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6_ext1-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber=
"WJetsToLNu_HT-600To800_TuneCUETP8M1_13TeV-madgraphMLM_ext1*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_HT-800To1200_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6_ext1-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber=
"WJetsToLNu_HT-800To1200_TuneCUETP8M1_13TeV-madgraphMLM_ext1*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_HT-1200To2500_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber=
"WJetsToLNu_HT-1200To2500_TuneCUETP8M1_13TeV-madgraphMLM*.root",
),
DirectorySample(
dataset=
"/WJetsToLNu_HT-2500ToInf_TuneCUETP8M1_13TeV-madgraphMLM-pythia8/RunIISummer16MiniAODv2-PUMoriond17_80X_mcRun2_asymptotic_2016_TrancheIV_v6_ext1-v1/MINIAODSIM",
location=
"/hadoop/cms/store/user/phchang/metis/baby/HWW2016_v1.15.1/merged4",
globber=
"WJetsToLNu_HT-2500ToInf_TuneCUETP8M1_13TeV-madgraphMLM_ext1*.root",
),
]
# Loop over datasets to submit
while True:
all_tasks_complete = True
#for sample in sorted(samples):
for sample in samples_to_run:
# define the task
maker_task = CondorTask(
sample=sample,
tag=job_tag,
arguments="",
executable=exec_path,
tarfile=tar_gz_path,
special_dir=hadoop_path,
output_name="output.root",
files_per_output=1,
condor_submit_params={"sites": "T2_US_UCSD"},
open_dataset=False,
flush=True,
#no_load_from_backup = True,
)
# process the job (either submits, checks for resubmit, or finishes etc.)
maker_task.process()
# save some information for the dashboard
total_summary[maker_task.get_sample().get_datasetname(
)] = maker_task.get_task_summary()
# Aggregate whether all tasks are complete
all_tasks_complete = all_tasks_complete and maker_task.complete()
# parse the total summary and write out the dashboard
StatsParser(data=total_summary, webdir=metis_dashboard_path).do()
# Print msummary table so I don't have to load up website
os.system("msummary -r | tee summary.txt")
os.system("chmod -R 755 {}".format(metis_dashboard_path))
# If all done exit the loop
if all_tasks_complete:
print ""
print "Job={} finished".format(job_tag)
print ""
break
# Neat trick to not exit the script for force updating
print 'Press Ctrl-C to force update, otherwise will sleep for 300 seconds'
try:
for i in range(0, 300):
sleep(1) # could use a backward counter to be preeety :)
except KeyboardInterrupt:
raw_input("Press Enter to force update, or Ctrl-C to quit.")
print "Force updating..."
calc_neg_events=False,
)
jra_flatpu = CMSSWTask(
sample=DirectorySample(
location=indir_flatpu,
dataset=
"/QCD_Pt-15to7000_TuneCP5_Flat_13TeV_pythia8/privateMC_fastsim_FlatPU_94x/JRA"
),
open_dataset=True,
files_per_output=10,
pset="../run_JRA_cfg_update.py",
cmssw_version="CMSSW_9_4_12",
scram_arch="slc6_amd64_gcc630",
tag=tag,
tarfile="lib_JRA.tar.xz",
dont_check_tree=True,
calc_neg_events=False,
)
tasks = [jra_nopu, jra_flatpu]
for task in tasks:
task.process()
summary = task.get_task_summary()
total_summary[task.get_sample().get_datasetname()] = summary
StatsParser(data=total_summary,
webdir="~/public_html/dump/metis_jra/").do()
time.sleep(30 * 60)
flush = True,
files_per_output = fpo(sample),
output_name = "merged_ntuple.root",
tag = job_tag,
cmssw_version = cmssw_ver,
executable = "condor_exe_%s.sh" % job_tag if (args.use_xrdcp or args.use_gridftp or args.use_wget) else "condor_exe.sh",
tarfile = "empty" if (args.use_xrdcp or args.use_gridftp or args.use_wget) else tar_path,
condor_submit_params = {"sites" : "T2_US_UCSD"} if (args.use_wget) else {"sites" : "T2_US_UCSD,T2_US_CALTECH,T2_US_MIT,T2_US_WISCONSIN,T2_US_Nebraska,T2_US_Purdue,T2_US_Vanderbilt,T2_US_Florida"},
special_dir = hadoop_path,
arguments = conds_dict[year]
)
task.process()
if not task.complete():
allcomplete = False
total_summary[dataset] = task.get_task_summary()
if first_run:
print "Sleeping 30s to space out jobs"
os.system("sleep 30s")
StatsParser(data=total_summary, webdir="~/public_html/dump/ttH_BabyMaker_v2_%s/" % job_tag).do()
os.system("chmod -R 755 ~/public_html/dump/ttH_BabyMaker_v2_%s" % job_tag)
if allcomplete:
print("")
print("Job={} finished".format(job_tag))
print("")
break
first_run = False
print("Sleeping 1000 seconds ...")
time.sleep(1000)
signal_strings = ["ttHJetToGG", "ttHToGG", "THQ", "THW", "VBF", "GluGluHToGG", "VHToGG", "FCNC", "ttHiggs"]
