from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from DIRAC.Core.Base.Script import parseCommandLine
parseCommandLine()
from DIRAC.tests.Utilities.utils import find_all
from DIRAC.Interfaces.API.Job import Job
# With a script that returns 0
scriptSHLocation = find_all("script-OK.sh", "..",
"/DIRAC/WorkloadManagementSystem/JobWrapper")[0]
j = Job()
j.setExecutable("sh %s" % scriptSHLocation)
jobXMLFile = "jobDescription-OK.xml"
with open(jobXMLFile, "w+") as fd:
fd.write(j._toXML())
# With a script that returns 0 - multiple steps
j = Job()
j.setExecutable("sh %s" % scriptSHLocation,
modulesList=["Script", "FailoverRequest"])
jobXMLFile = "jobDescription-OK-multiSteps.xml"
with open(jobXMLFile, "w+") as fd:
fd.write(j._toXML())
# With a script that returns 111
scriptSHLocation = find_all("script.sh", "..",
args = Script.getUnprocessedSwitches()
if len(args) != 1:
Script.showHelp()
directory = args[0]
UseFilter = None
for switch, switchValue in Script.getUnprocessedSwitches():
if switch == "UseFilter":
UseFilter = True if switchValue.lower() == "true" else False
# Let's first create the prodJobuction
prodJobType = 'Merge'
transName = 'testProduction_' + str(int(time.time()))
desc = 'just test'
prodJob = Job()
prodJob._addParameter(prodJob.workflow, 'PRODUCTION_ID', 'string', '00012345',
'ProductionID')
prodJob._addParameter(prodJob.workflow, 'JOB_ID', 'string', '00006789',
'ProductionJobID')
prodJob._addParameter(prodJob.workflow, 'eventType', 'string', 'TestEventType',
'Event Type of the prodJobuction')
prodJob._addParameter(prodJob.workflow, 'numberOfEvents', 'string', '-1',
'Number of events requested')
prodJob._addParameter(prodJob.workflow, 'ProcessingType', 'JDL', str('Test'),
'ProductionGroupOrType')
prodJob._addParameter(prodJob.workflow, 'Priority', 'JDL', str(9),
'UserPriority')
prodJob.setType(prodJobType)
prodJob.workflow.setName(transName)
prodJob.workflow.setDescrShort(desc)
def main():
"""
Launch job on the GRID
"""
# this thing pilots everything related to the GRID
dirac = Dirac()
if switches["output_type"] in "TRAINING":
print("Preparing submission for TRAINING data")
elif switches["output_type"] in "DL2":
print("Preparing submission for DL2 data")
else:
print("You have to choose either TRAINING or DL2 as output type!")
sys.exit()
# Read configuration file
cfg = load_config(switches["config_file"])
# Analysis
config_path = cfg["General"]["config_path"]
config_file = cfg["General"]["config_file"]
mode = cfg["General"]["mode"] # One mode naw
particle = cfg["General"]["particle"]
estimate_energy = cfg["General"]["estimate_energy"]
force_tailcut_for_extended_cleaning = cfg["General"][
"force_tailcut_for_extended_cleaning"]
# Take parameters from the analysis configuration file
ana_cfg = load_config(os.path.join(config_path, config_file))
config_name = ana_cfg["General"]["config_name"]
cam_id_list = ana_cfg["General"]["cam_id_list"]
# Regressor and classifier methods
regressor_method = ana_cfg["EnergyRegressor"]["method_name"]
classifier_method = ana_cfg["GammaHadronClassifier"]["method_name"]
# Someone might want to create DL2 without score or energy estimation
if regressor_method in ["None", "none", None]:
use_regressor = False
else:
use_regressor = True
if classifier_method in ["None", "none", None]:
use_classifier = False
else:
use_classifier = True
# GRID
outdir = os.path.join(cfg["GRID"]["outdir"], config_name)
n_file_per_job = cfg["GRID"]["n_file_per_job"]
n_jobs_max = cfg["GRID"]["n_jobs_max"]
model_dir = cfg["GRID"]["model_dir"]
training_dir_energy = cfg["GRID"]["training_dir_energy"]
training_dir_classification = cfg["GRID"]["training_dir_classification"]
dl2_dir = cfg["GRID"]["dl2_dir"]
home_grid = cfg["GRID"]["home_grid"]
user_name = cfg["GRID"]["user_name"]
banned_sites = cfg["GRID"]["banned_sites"]
# HACK
if force_tailcut_for_extended_cleaning is True:
print("Force tail cuts for extended cleaning!!!")
# Prepare command to launch script
source_ctapipe = "source /cvmfs/cta.in2p3.fr/software/conda/dev/setupConda.sh"
source_ctapipe += " && conda activate ctapipe_v0.11.0"
if switches["output_type"] in "TRAINING":
execute = "data_training.py"
script_args = [
"--config_file={}".format(config_file),
"--estimate_energy={}".format(str(estimate_energy)),
"--regressor_config={}.yaml".format(regressor_method),
"--regressor_dir=./",
"--outfile {outfile}",
"--indir ./ --infile_list={infile_name}",
"--max_events={}".format(switches["max_events"]),
"--{mode}",
"--cam_ids {}".format(cam_id_list),
]
output_filename_template = "TRAINING"
elif switches["output_type"] in "DL2":
execute = "write_dl2.py"
script_args = [
"--config_file={}".format(config_file),
"--regressor_config={}.yaml".format(regressor_method),
"--regressor_dir=./",
"--classifier_config={}.yaml".format(classifier_method),
"--classifier_dir=./",
"--outfile {outfile}",
"--indir ./ --infile_list={infile_name}",
"--max_events={}".format(switches["max_events"]),
"--{mode}",
"--force_tailcut_for_extended_cleaning={}".format(
force_tailcut_for_extended_cleaning),
"--cam_ids {}".format(cam_id_list),
]
output_filename_template = "DL2"
# Make the script save also the full calibrated images if required
if switches["save_images"] is True:
script_args.append("--save_images")
# Make the script print debug information if required
if switches["debug_script"] is True:
script_args.append("--debug")
cmd = [source_ctapipe, "&&", "./" + execute]
cmd += script_args
pilot_args_write = " ".join(cmd)
# For table merging if multiple runs
pilot_args_merge = " ".join([
source_ctapipe,
"&&",
"./merge_tables.py",
"--template_file_name",
"{in_name}",
"--outfile",
"{out_name}",
])
prod3b_filelist = dict()
if estimate_energy is False and switches["output_type"] in "TRAINING":
prod3b_filelist["gamma"] = cfg["EnergyRegressor"]["gamma_list"]
elif estimate_energy is True and switches["output_type"] in "TRAINING":
prod3b_filelist["gamma"] = cfg["GammaHadronClassifier"]["gamma_list"]
prod3b_filelist["proton"] = cfg["GammaHadronClassifier"]["proton_list"]
elif switches["output_type"] in "DL2":
prod3b_filelist["gamma"] = cfg["Performance"]["gamma_list"]
prod3b_filelist["proton"] = cfg["Performance"]["proton_list"]
prod3b_filelist["electron"] = cfg["Performance"]["electron_list"]
# from IPython import embed
# embed()
# Split list of files according to stoprage elements
with open(prod3b_filelist[particle]) as f:
filelist = f.readlines()
filelist = ["{}".format(_.replace("\n", "")) for _ in filelist]
res = dirac.splitInputData(filelist, n_file_per_job)
list_run_to_loop_on = res["Value"]
# define a template name for the file that's going to be written out.
# the placeholder braces are going to get set during the file-loop
output_filename = output_filename_template
output_path = outdir
if estimate_energy is False and switches["output_type"] in "TRAINING":
output_path += "/{}/".format(training_dir_energy)
step = "energy"
if estimate_energy is True and switches["output_type"] in "TRAINING":
output_path += "/{}/".format(training_dir_classification)
step = "classification"
if switches["output_type"] in "DL2":
if force_tailcut_for_extended_cleaning is False:
output_path += "/{}/".format(dl2_dir)
else:
output_path += "/{}_force_tc_extended_cleaning/".format(dl2_dir)
step = ""
output_filename += "_{}.h5"
# sets all the local files that are going to be uploaded with the job
# plus the pickled classifier
# if file name starts with `LFN:`, it will be copied from the GRID
input_sandbox = [
# Utility to assign one job to one command...
os.path.expandvars("$GRID/pilot.sh"),
os.path.expandvars("$PROTOPIPE/protopipe/"),
os.path.expandvars("$GRID/merge_tables.py"),
# python wrapper for the mr_filter wavelet cleaning
# os.path.expandvars("$PYWI/pywi/"),
# os.path.expandvars("$PYWICTA/pywicta/"),
# script that is being run
os.path.expandvars("$PROTOPIPE/protopipe/scripts/" + execute),
# Configuration file
os.path.expandvars(os.path.join(config_path, config_file)),
]
models_to_upload = []
configs_to_upload = []
if estimate_energy is True and switches["output_type"] in "TRAINING":
config_path_template = "LFN:" + os.path.join(home_grid, outdir,
model_dir, "{}.yaml")
config_to_upload = config_path_template.format(regressor_method)
model_path_template = "LFN:" + os.path.join(
home_grid, outdir, model_dir, "regressor_{}_{}.pkl.gz")
for cam_id in cam_id_list:
model_to_upload = model_path_template.format(
cam_id, regressor_method) # TBC
print("The following model(s) will be uploaded to the GRID:")
print(model_to_upload)
models_to_upload.append(model_to_upload)
print(
"The following configs(s) for such models will be uploaded to the GRID:"
)
print(config_to_upload)
configs_to_upload.append(config_to_upload)
# input_sandbox.append(model_to_upload)
elif estimate_energy is False and switches["output_type"] in "TRAINING":
pass
else: # Charge also classifer for DL2
model_type_list = ["regressor", "classifier"]
model_method_list = [regressor_method, classifier_method]
config_path_template = "LFN:" + os.path.join(home_grid, outdir,
model_dir, "{}.yaml")
model_path_template = "LFN:" + os.path.join(
home_grid, outdir, model_dir, "{}_{}_{}.pkl.gz")
if force_tailcut_for_extended_cleaning is True:
force_mode = mode.replace("wave", "tail")
print("################")
print(force_mode)
else:
force_mode = mode
for idx, model_type in enumerate(model_type_list):
print(
"The following configuration file will be uploaded to the GRID:"
)
config_to_upload = config_path_template.format(
model_method_list[idx])
print(config_to_upload)
configs_to_upload.append(config_to_upload) # upload only 1 copy
print(
"The following model(s) related to such configuration file will be uploaded to the GRID:"
)
for cam_id in cam_id_list:
if model_type in "regressor" and use_regressor is False:
print("Do not upload regressor model on GRID!!!")
continue
if model_type in "classifier" and use_classifier is False:
print("Do not upload classifier model on GRID!!!")
continue
model_to_upload = model_path_template.format(
model_type_list[idx], cam_id, model_method_list[idx])
print(model_to_upload)
models_to_upload.append(model_to_upload)
# input_sandbox.append(model_to_upload)
# summary before submitting
print("\nDEBUG> running as:")
print(pilot_args_write)
print("\nDEBUG> with input_sandbox:")
print(input_sandbox)
print("\nDEBUG> with output file:")
print(output_filename.format("{job_name}"))
print("\nDEBUG> Particles:")
print(particle)
print("\nDEBUG> Energy estimation:")
print(estimate_energy)
# ######## ## ## ## ## ## ## #### ## ## ######
# ## ## ## ## ### ## ### ## ## ### ## ## ##
# ## ## ## ## #### ## #### ## ## #### ## ##
# ######## ## ## ## ## ## ## ## ## ## ## ## ## ## ####
# ## ## ## ## ## #### ## #### ## ## #### ## ##
# ## ## ## ## ## ### ## ### ## ## ### ## ##
# ## ## ####### ## ## ## ## #### ## ## ######
# list of files on the GRID SE space
# not submitting jobs where we already have the output
batcmd = "dirac-dms-user-lfns --BaseDir {}".format(
os.path.join(home_grid, output_path))
result = subprocess.check_output(batcmd, shell=True)
try:
grid_filelist = open(result.split()[-1]).read()
except IOError:
raise IOError("ERROR> cannot read GRID filelist...")
# get jobs from today and yesterday...
days = []
for i in range(2): # how many days do you want to look back?
days.append(
(datetime.date.today() - datetime.timedelta(days=i)).isoformat())
# get list of run_tokens that are currently running / waiting
running_ids = set()
running_names = []
for status in ["Waiting", "Running", "Checking"]:
for day in days:
try:
[
running_ids.add(id) for id in dirac.selectJobs(
status=status, date=day, owner=user_name)["Value"]
]
except KeyError:
pass
n_jobs = len(running_ids)
if n_jobs > 0:
print("Scanning {} running/waiting jobs... please wait...".format(
n_jobs))
for i, id in enumerate(running_ids):
if ((100 * i) / n_jobs) % 5 == 0:
print("\r{} %".format(((20 * i) / n_jobs) * 5)),
jobname = dirac.getJobAttributes(id)["Value"]["JobName"]
running_names.append(jobname)
else:
print("\n... done")
for bunch in list_run_to_loop_on:
# for bunch in bunches_of_run:
# from IPython import embed
# embed()
# this selects the `runxxx` part of the first and last file in the run
# list and joins them with a dash so that we get a nice identifier in
# the output file name.
# if there is only one file in the list, use only that one
# run_token = re.split('_', bunch[+0])[3] # JLK JLK
run_token = re.split("_", bunch[0])[3]
if len(bunch) > 1:
run_token = "-".join([run_token, re.split("_", bunch[-1])[3]])
print("-" * 50)
print("-" * 50)
# setting output name
output_filenames = dict()
if switches["output_type"] in "DL2":
job_name = "protopipe_{}_{}_{}_{}_{}".format(
config_name, switches["output_type"], particle, run_token,
mode)
output_filenames[mode] = output_filename.format("_".join(
[particle, mode, run_token]))
else:
job_name = "protopipe_{}_{}_{}_{}_{}_{}".format(
config_name, switches["output_type"], step, particle,
run_token, mode)
output_filenames[mode] = output_filename.format("_".join(
[step, particle, mode, run_token]))
# if job already running / waiting, skip
if job_name in running_names:
print("\n WARNING> {} still running\n".format(job_name))
continue
print("Output file name: {}".format(output_filenames[mode]))
# if file already in GRID storage, skip
# (you cannot overwrite it there, delete it and resubmit)
# (assumes tail and wave will always be written out together)
already_exist = False
file_on_grid = os.path.join(output_path, output_filenames[mode])
print("DEBUG> check for existing file on GRID...")
if file_on_grid in grid_filelist:
print("\n WARNING> {} already on GRID SE\n".format(job_name))
continue
if n_jobs_max == 0:
print("WARNING> maximum number of jobs to submit reached")
print("WARNING> breaking loop now")
break
else:
n_jobs_max -= 1
j = Job()
# runtime in seconds times 8 (CPU normalisation factor)
j.setCPUTime(6 * 3600 * 8)
j.setName(job_name)
j.setInputSandbox(input_sandbox)
if banned_sites:
j.setBannedSites(banned_sites)
# Add simtel files as input data
j.setInputData(bunch)
for run_file in bunch:
file_token = re.split("_", run_file)[3]
# wait for a random number of seconds (up to five minutes) before
# starting to add a bit more entropy in the starting times of the
# dirac queries.
# if too many jobs try in parallel to access the SEs,
# the interface crashes
# #sleep = random.randint(0, 20) # seconds
# #j.setExecutable('sleep', str(sleep))
# JLK: Try to stop doing that
# consecutively downloads the data files, processes them,
# deletes the input
# and goes on to the next input file;
# afterwards, the output files are merged
# j.setExecutable('dirac-dms-get-file', "LFN:" + run_file)
# source the miniconda ctapipe environment and
# run the python script with all its arguments
if switches["output_type"] in "DL2":
output_filename_temp = output_filename.format("_".join(
[particle, mode, file_token]))
if switches["output_type"] in "TRAINING":
output_filename_temp = output_filename.format("_".join(
[step, particle, mode, file_token]))
j.setExecutable(
"./pilot.sh",
pilot_args_write.format(
outfile=output_filename_temp,
infile_name=os.path.basename(run_file),
mode=mode,
),
)
# remove the current file to clear space
j.setExecutable("rm", os.path.basename(run_file))
# simple `ls` for good measure
j.setExecutable("ls", "-lh")
# if there is more than one file per job, merge the output tables
if len(bunch) > 1:
names = []
names.append(("*_{}_".format(particle), output_filenames[mode]))
for in_name, out_name in names:
print("in_name: {}, out_name: {}".format(in_name, out_name))
j.setExecutable(
"./pilot.sh",
pilot_args_merge.format(in_name=in_name,
out_name=out_name),
)
print("DEBUG> args append: {}".format(
pilot_args_merge.format(in_name=in_name,
out_name=out_name)))
bunch.extend(models_to_upload)
bunch.extend(configs_to_upload)
j.setInputData(bunch)
print("Input data set to job = {}".format(bunch))
outputs = []
outputs.append(output_filenames[mode])
print("Output file path: {}{}".format(output_path,
output_filenames[mode]))
j.setOutputData(outputs, outputSE=None, outputPath=output_path)
# check if we should somehow stop doing what we are doing
if switches["dry"] is True:
print("\nThis is a DRY RUN! -- NO job has been submitted!")
print("Name of the job: {}".format(job_name))
print("Name of the output file: {}".format(outputs))
print("Output path from GRID home: {}".format(output_path))
break
# this sends the job to the GRID and uploads all the
# files into the input sandbox in the process
print("\nSUBMITTING job with the following INPUT SANDBOX:")
print(input_sandbox)
print("Submission RESULT: {}\n".format(dirac.submitJob(j)["Value"]))
# break if this is only a test submission
if switches["test"] is True:
print("This is a TEST RUN! -- Only ONE job will be submitted!")
print("Name of the job: {}".format(job_name))
print("Name of the output file: {}".format(outputs))
print("Output path from GRID home: {}".format(output_path))
break
# since there are two nested loops, need to break again
if switches["test"] is True:
break
try:
os.remove("datapipe.tar.gz")
os.remove("modules.tar.gz")
except:
pass
# Upload analysis configuration file for provenance
SE_LIST = ['CC-IN2P3-USER', 'DESY-ZN-USER', 'CNAF-USER', 'CEA-USER']
analysis_config_local = os.path.join(config_path, config_file)
# the configuration file is uploaded to the data directory because
# the training samples (as well as their cleaning settings) are independent
analysis_config_dirac = os.path.join(home_grid, output_path, config_file)
print("Uploading {} to {}...".format(analysis_config_local,
analysis_config_dirac))
if switches["dry"] is False:
# Upload this file to all Dirac Storage Elements in SE_LIST
for se in SE_LIST:
# the uploaded config file overwrites any old copy
ana_cfg_upload_cmd = "dirac-dms-add-file -f {} {} {}".format(
analysis_config_dirac, analysis_config_local, se)
ana_cfg_upload_result = subprocess.check_output(ana_cfg_upload_cmd,
shell=True)
print(ana_cfg_upload_result)
else:
print("This is a DRY RUN! -- analysis.yaml has NOT been uploaded.")
print("\nall done -- exiting now")
exit()
print("Usage %s <scriptName> <jobName> <nbJobs>" % sys.argv[0])
sys.exit(1)
scriptName = sys.argv[1]
jobName = sys.argv[2]
nbJobs = int(sys.argv[3])
if not os.path.exists(jobName):
os.makedirs(jobName)
os.makedirs("%s/Done" % jobName)
os.makedirs("%s/Failed" % jobName)
else:
print("Folder %s exists" % jobName)
sys.exit(1)
f = open("%s/jobIdList.txt" % jobName, 'w')
for i in xrange(nbJobs):
j = Job()
j.setCPUTime(10000)
j.setExecutable(scriptName)
j.addToOutputSandbox.append('myLog.txt')
j.addToOutputSandbox.append('clock.txt')
j.addToOutputSandbox.append('time.txt')
dirac = Dirac()
jobID = dirac.submitJob(j)
realId = jobID.get('JobID')
f.write("%s\n" % realId)
f.close()
from DIRAC.tests.Utilities.utils import find_all
from DIRAC.Interfaces.API.Job import Job
from DIRAC.Interfaces.API.Dirac import Dirac
#from tests.Workflow.Integration.Test_UserJobs import createJob
gLogger.setLevel('DEBUG')
cwd = os.path.realpath('.')
dirac = Dirac()
# Simple Hello Word job to DIRAC.Jenkins.ch
gLogger.info("\n Submitting hello world job targeting DIRAC.Jenkins.ch")
helloJ = Job()
helloJ.setName("helloWorld-TEST-TO-Jenkins")
helloJ.setInputSandbox([find_all('exe-script.py', '..', '/DIRAC/tests/Workflow/')[0]])
helloJ.setExecutable("exe-script.py", "", "helloWorld.log")
helloJ.setCPUTime(17800)
helloJ.setDestination('DIRAC.Jenkins.ch')
result = dirac.submitJob(helloJ)
gLogger.info("Hello world job: ", result)
if not result['OK']:
gLogger.error("Problem submitting job", result['Message'])
exit(1)
# Simple Hello Word job to DIRAC.Jenkins.ch, that needs to be matched by a MP WN
gLogger.info("\n Submitting hello world job targeting DIRAC.Jenkins.ch and a MP WN")
helloJMP = Job()
helloJMP.setName("helloWorld-TEST-TO-Jenkins-MP")
def submitTS():
########################################
# Modify here with your dirac username
owner = 'user02'
########################################
########################################
# Job description
########################################
job = Job()
job.setName('mandelbrot raw')
job.setOutputSandbox(['*log'])
job.workflow.addParameter(
Parameter("JOB_ID", "000000", "string", "", "", True, False,
"Initialize JOB_ID"))
## define the job workflow in 3 steps
# job step1: setup software
job.setExecutable('git clone https://github.com/bregeon/mandel4ts.git')
# job step2: run mandelbrot application
job.setExecutable('./mandel4ts/mandelbrot.py',
arguments="-P 0.0005 -M 1000 -L @{JOB_ID} -N 200")
outputPath = os.path.join('/vo.france-grilles.fr/user', owner[0], owner,
'mandelbrot/images/raw')
outputPattern = 'data_*txt'
outputSE = 'DIRAC-USER'
outputMetadata = json.dumps({
"application": "mandelbrot",
"image_format": "ascii",
"image_width": 7680,
"image_height": 200,
"owner": owner
})
# job step3: upload data and set metadata
job.setExecutable('./mandel4ts/dirac-add-files.py',
arguments="%s '%s' %s '%s'" %
(outputPath, outputPattern, outputSE, outputMetadata))
########################################
# Transformation definition
########################################
t = Transformation()
t.setTransformationName(owner + '_step1')
t.setType("MCSimulation")
t.setDescription("Mandelbrot images production")
t.setLongDescription("Mandelbrot images production")
# set the job workflow to the transformation
t.setBody(job.workflow.toXML())
########################################
# Transformation submission
########################################
res = t.addTransformation()
if not res['OK']:
print(res['Message'])
DIRAC.exit(-1)
t.setStatus("Active")
t.setAgentType("Automatic")
return res
"../tools/ND280GRID.py",
"../tools/ND280Job.py",
"../tools/ND280Software.py",
"../tools/pexpect.py",
"../tools/StorageElement.py"
]
# these files are created by the job
outputSandbox =\
[
stdout,
stderr,
logfile
]
# the executible here '' is later set, so don't confuse users later on
diracJob = Job('', stdout, stderr)
# give a descriptive name
diracJob.setName('ND280Custom')
# set the program/executable, arguments, logFile, ...
diracJob.setExecutable(exe, arguments=args, logFile=logfile)
# set the job length
diracJob.setCPUTime(3600)
diracJob.setExecutionEnv(environmentDict)
diracJob.setInputSandbox(inputSandbox)
diracJob.setOutputSandbox(outputSandbox)
print 'job being submitted...'
def Flux(args=None):
from DIRAC.Interfaces.API.Dirac import Dirac
from DIRAC.Interfaces.API.Job import Job
import time
import os.path
user_script = './flux.sh'
modmacro = './CTAflux_speeed.C'
site = "PARANAL"
if (len(args) != 5):
Script.showHelp()
typeofdata = "test"
particle = args[0]
direction = args[1]
MOD = args[2]
exten = args[3]
zenName = args[4]
# List of files over which flux should be run
LFN_file = "./stereofiles/lfn_%s_%s_%s_%s.lfns" % (particle, exten,
zenName, direction)
fileLength = sum(1 for line in open(LFN_file))
f = open(LFN_file, 'r')
if particle == "proton":
filesPerJob = 10
else:
filesPerJob = 20
fileList = []
text_file_name = "lfnStereoFiles_%s_%s_%s_%s.txt" % (particle, exten,
typeofdata, direction)
text_file = open(text_file_name, "w")
# File containing the id number of files already produced. The relaunch of these jobs will be skipped
done_file_name = "./stereofiles/done/done_%s_%s_%s_%s.lfns" % (
particle, exten, zenName, direction)
if os.path.exists(done_file_name):
done_content = [
int(line.strip()) for line in open(done_file_name, 'r')
]
else:
done_content = []
loop = 0
iJob = 0
for line in f:
loop = loop + 1
infileLFN = line.strip()
fileList.append(infileLFN)
text_file.write("%s\n" % infileLFN)
remain = loop % filesPerJob
if remain == 0 or loop == fileLength:
iJob = iJob + 1
# Skipping of already finished jobs
if iJob in done_content:
text_file.close()
fileList = []
text_file = open(text_file_name, "w")
continue
else:
j = Job()
text_file.close()
j.setInputSandbox([
user_script, "setupPackageMARS.sh", "CheckFileZombie.C",
text_file_name, modmacro
])
jobName = "%s_%s_%s_%s_%s_%s_%s" % (user_script, site,
particle, direction, iJob,
exten, zenName)
jobOut = "%s_%s_%s_%s_%s.out" % (user_script, site, particle,
direction, iJob)
script_args = "%s %s %s %s %s %s %s" % (
particle, site, iJob, direction, MOD, exten, zenName)
j.setExecutable(user_script, script_args)
j.setOutputSandbox([jobOut, "applicationLog.txt"])
j.setName(jobName)
j.setBannedSites([
'LCG.MSFG.fr', 'LCG.M3PEC.fr', 'LCG.OBSPM.fr',
'LCG.UNI-DORTMUND.de', 'LCG.UNIV-LILLE.fr',
'LCG.Prague.cz', 'LCG.GRIF.fr'
])
Script.gLogger.info(j._toJDL())
print "Submitting job %s %s %s %s %s %s" % (
user_script, zenName, particle, direction, site, iJob)
time.sleep(3)
Dirac().submit(j)
fileList = []
text_file = open(text_file_name, "w")
def setUp(self):
self.job = Job()
pass
