def test_execute( self ):
job = Job()
job.setName( "helloWorld-test" )
job.setExecutable( find_all( "helloWorld.py", '.', 'Integration' )[0],
arguments = "This is an argument",
logFile = "aLogFileForTest.txt" ,
parameters=[('executable', 'string', '', "Executable Script"),
('arguments', 'string', '', 'Arguments for executable Script'),
( 'applicationLog', 'string', '', "Log file name" ),
( 'someCustomOne', 'string', '', "boh" )],
paramValues = [( 'someCustomOne', 'aCustomValue' )] )
job.setBannedSites( ['LCG.SiteA.com', 'DIRAC.SiteB.org'] )
job.setOwner( 'ownerName' )
job.setOwnerGroup( 'ownerGroup' )
job.setName( 'jobName' )
job.setJobGroup( 'jobGroup' )
job.setType( 'jobType' )
job.setDestination( 'DIRAC.someSite.ch' )
job.setCPUTime( 12345 )
job.setLogLevel( 'DEBUG' )
res = job.runLocal( self.d )
self.assertTrue( res['OK'] )
def test_execute(self):
job = Job()
job.setName("helloWorld-test")
job.setExecutable(find_all("helloWorld.py", '.', 'Integration')[0],
arguments="This is an argument",
logFile="aLogFileForTest.txt",
parameters=[('executable', 'string', '',
"Executable Script"),
('arguments', 'string', '',
'Arguments for executable Script'),
('applicationLog', 'string', '',
"Log file name"),
('someCustomOne', 'string', '', "boh")],
paramValues=[('someCustomOne', 'aCustomValue')])
job.setBannedSites(['LCG.SiteA.com', 'DIRAC.SiteB.org'])
job.setOwner('ownerName')
job.setOwnerGroup('ownerGroup')
job.setName('jobName')
job.setJobGroup('jobGroup')
job.setType('jobType')
job.setDestination('DIRAC.someSite.ch')
job.setCPUTime(12345)
job.setLogLevel('DEBUG')
res = job.runLocal(self.d)
self.assertTrue(res['OK'])
def createJob(self):
job = Job()
job.setName(self.__stepName)
job.setOutputSandbox(['*log'])
job.setExecutable('/usr/bin/wget',
arguments='"{0}/{1}"'.format(URL_ROOT,
self.__executable))
job.setExecutable('/bin/chmod',
arguments='+x "{0}"'.format(self.__executable))
arguments = '"{0}" "{1}" "{2}" "{3}" "{4}" "{5}" @{{JOB_ID}}'.format(
self.__softwareVersion, self.__application, self.__outputPath,
self.__outputPattern, self.__outputSE, self.__outputMode)
if self.__extraArgs:
arguments += ' ' + self.__extraArgs
job.setExecutable(self.__executable, arguments=arguments)
# failover for failed jobs
job.setExecutable('/bin/ls -l',
modulesList=['Script', 'FailoverRequest'])
if self.__inputData:
job.setInputData(self.__inputData)
if self.__site:
job.setDestination(self.__site)
if self.__bannedsite:
job.setBannedSites(self.__bannedsite)
job.setOutputSandbox(['app.out', 'app.err', 'Script3_CodeOutput.log'])
self.__job = job
def submit(name,
job_group,
task_id,
input_sandbox,
output_sandbox,
executable,
site=None,
banned_site=None,
sub_ids=[]):
dirac = Dirac()
submit_result = {'backend_job_ids': {}}
jobInfos = {}
for run in range(int((len(sub_ids) + 99) / 100)):
ids_this_run = [x for x in sub_ids[run * 100:(run + 1) * 100]]
job_names = ['%s.%s' % (name, sub_id) for sub_id in ids_this_run]
j = Job()
j.setName(name)
j.setExecutable(executable)
j.setParameterSequence('JobName', job_names, addToWorkflow=True)
j.setParameterSequence('arguments', ids_this_run, addToWorkflow=True)
if input_sandbox:
j.setInputSandbox(input_sandbox)
if output_sandbox:
j.setOutputSandbox(output_sandbox)
if job_group:
j.setJobGroup(job_group)
if site: # set destination to a certain site; list not allowed
j.setDestination(site)
if banned_site:
j.setBannedSites(banned_site)
result = dirac.submitJob(j)
if not result['OK']:
sys.stdout.write('DIRAC job submit error: %s\n' %
result['Message'])
sys.exit(1)
for sub_id, dirac_id in zip(ids_this_run, result['Value']):
submit_result['backend_job_ids'][sub_id] = dirac_id
jobInfos[dirac_id] = {'SubID': sub_id}
#Register on Task-manager Webapp of IHEPDIRAC
task = RPCClient('WorkloadManagement/TaskManager')
taskInfo = {'TaskName': name, 'JobGroup': job_group, 'JSUB-ID': task_id}
task_result = task.createTask(name, taskInfo, jobInfos)
task_web_id = task_result['Value']
submit_result['backend_task_id'] = task_web_id
return submit_result
def test_execute(self):
job = Job()
job._siteSet = {"DIRAC.someSite.ch"}
job.setName("helloWorld-test")
job.setExecutable(
self.helloWorld,
arguments="This is an argument",
logFile="aLogFileForTest.txt",
parameters=[
("executable", "string", "", "Executable Script"),
("arguments", "string", "", "Arguments for executable Script"),
("applicationLog", "string", "", "Log file name"),
("someCustomOne", "string", "", "boh"),
],
paramValues=[("someCustomOne", "aCustomValue")],
)
job.setBannedSites(["LCG.SiteA.com", "DIRAC.SiteB.org"])
job.setOwner("ownerName")
job.setOwnerGroup("ownerGroup")
job.setName("jobName")
job.setJobGroup("jobGroup")
job.setType("jobType")
job.setDestination("DIRAC.someSite.ch")
job.setCPUTime(12345)
job.setLogLevel("DEBUG")
try:
# This is the standard location in Jenkins
job.setInputSandbox(
find_all("pilot.cfg",
os.environ["WORKSPACE"] + "/PilotInstallDIR")[0])
except (IndexError, KeyError):
job.setInputSandbox(find_all("pilot.cfg", rootPath)[0])
job.setConfigArgs("pilot.cfg")
res = job.runLocal(self.d)
self.assertTrue(res["OK"])
def test_execute(self):
job = Job()
job._siteSet = {'DIRAC.someSite.ch'}
job.setName("helloWorld-test")
job.setExecutable(self.helloWorld,
arguments="This is an argument",
logFile="aLogFileForTest.txt",
parameters=[('executable', 'string', '',
"Executable Script"),
('arguments', 'string', '',
'Arguments for executable Script'),
('applicationLog', 'string', '',
"Log file name"),
('someCustomOne', 'string', '', "boh")],
paramValues=[('someCustomOne', 'aCustomValue')])
job.setBannedSites(['LCG.SiteA.com', 'DIRAC.SiteB.org'])
job.setOwner('ownerName')
job.setOwnerGroup('ownerGroup')
job.setName('jobName')
job.setJobGroup('jobGroup')
job.setType('jobType')
job.setDestination('DIRAC.someSite.ch')
job.setCPUTime(12345)
job.setLogLevel('DEBUG')
try:
# This is the standard location in Jenkins
job.setInputSandbox(
find_all('pilot.cfg',
os.environ['WORKSPACE'] + '/PilotInstallDIR')[0])
except (IndexError, KeyError):
job.setInputSandbox(find_all('pilot.cfg', rootPath)[0])
job.setConfigArgs('pilot.cfg')
res = job.runLocal(self.d)
self.assertTrue(res['OK'])
gLogger.error("No executable defined.")
dexit(1)
j.setName("MC job")
if not opts.name is None:
j.setName(opts.name)
j.setInputSandbox(input_sandbox_files) # all input files in the sandbox
j.setOutputSandbox(output_sandbox_files)
j.setCPUTime(opts.cpu)
if not opts.site is None:
j.setDestination(opts.site.split(","))#can also be a list
if not opts.bannedSites is None:
j.setBannedSites(opts.bannedSites.split(","))
if not opts.release is None:
tag = opts.release
cl = SoftwareTagClient()
result = cl.getSitesForTag(tag,'Valid') # keyword doesn't work there.
if not result['OK']:
gLogger.error("*ERROR* Could not get sites for Tag %s"%tag,result['Message'])
dexit(1)
sites = result[ 'Value' ]
j.setDestination(sites)
if not opts.stagein is None:
input_stage_files = []
# we do add. input staging
files = opts.stagein.split(",")
startsensor = idx * 4
numsensors = 4
if idx == 47:
numsensors = 1
args = visit + ' ' + insidename + ' ' + str(startsensor) + ' ' + str(numsensors) + ' ' + str(idx)
outputname = 'fits_' + visit + '_' + str(idx) + '.tar'
j.setCPUTime(1209600)
j.setExecutable('runimsim2.1.sh', arguments=args)
j.stderr="std.err"
j.stdout="std.out"
#!!! May need the 2.1i directory here depending on visit number !!!
j.setInputSandbox(["runimsim2.1.sh","run_imsim_nersc.py","LFN:/lsst/user/j/james.perry/instcats/2.1.1i/" + instcatname])
j.setOutputSandbox(["std.out","std.err"])
j.setTag(["4Processors"])
j.setOutputData([visit + "/" + outputname], outputPath="", outputSE=["UKI-NORTHGRID-LANCS-HEP-disk"])
j.setPlatform("AnyPlatform")
j.setBannedSites(["VAC.UKI-NORTHGRID-MAN-HEP.uk", "LCG.IN2P3-CC.fr"])
#print("Would submit job for sensors", sensorstring)
jobID = dirac.submitJob(j)
print("Submitted job as ID " + str(jobID))
print "Status is:", dirac.status(jobID['JobID'])
joblistfile.write(str(jobID['JobID']) + '\n')
joblistfile.close()
def dirLUT(args=None):
from DIRAC.Interfaces.API.Dirac import Dirac
from DIRAC.Interfaces.API.Job import Job
user_script = './dirLUT.sh'
if (len(args) != 3):
Script.showHelp()
direction = args[0]
zenith = args[1]
diffuse = args[2]
if diffuse == "0":
diffName = "point"
elif diffuse == "1":
diffName = "diff"
else:
print "Invalid extension definition!"
Script.showHelp()
return 1
if zenith == "40":
zenName = "40deg"
elif zenith == "20":
zenName = "20deg"
else:
print "Invalid zenith definition!"
Script.showHelp()
return 1
if direction == "N":
directionName = "north"
# deg = "180"
elif direction == "S":
directionName = "south"
# deg = "0"
else:
print 'Wrong direction. It can only be "N" or "S".'
Script.showHelp()
return 1
listname = './training/gamma_trainLUT_%s_%s_%s.lfns' % (zenName, diffName,
direction)
with open(listname) as f:
totalEntries = sum(1 for _ in f)
# Number of files used per job
runN = 20
runMin = 0
runMax = totalEntries / runN
for i in range(runMin, runMax):
jobName = "%s_%s_%s_%s_%s" % (user_script, direction, zenName,
diffName, i)
jobOut = "%s_%s_%s%s.out" % (user_script, directionName, diffName, i)
script_args = "%s %s %s %s %s" % (direction, zenName, diffName, i,
runN)
j = Job()
j.setInputSandbox([
user_script, listname, "setupPackageMARS.sh", "CheckFileZombie.C"
])
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'
])
Script.gLogger.info(j._toJDL())
print "Submitting job %s" % (script_args)
Dirac().submit(j)
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()
""" simple hello world job
"""
from DIRAC.Interfaces.API.Job import Job
from DIRAC.Interfaces.API.Dirac import Dirac
from DIRAC.DataManagementSystem.Utilities.DMSHelpers import DMSHelpers
j = Job()
j.setName("helloWorld-test")
j.setExecutable("exe-script.py", "", "Executable.log")
# <-- user settings
j.setCPUTime(172800)
tier1s = DMSHelpers().getTiers(tier=(0, 1))
j.setBannedSites(tier1s)
# user settings -->
# print j.workflow
# submit the job to dirac
result = Dirac().submitJob(j)
print(result)
if NJobs == 0:
print("maximum number of jobs to submit reached")
print("breaking loop now")
break
else:
NJobs -= 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)
# mr_filter loses its executable property by uploading it to the GRID SE; reset
j.setExecutable('chmod', '+x mr_filter')
j.setExecutable('ls -lah')
for run_file in run_filelist:
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 querries.
# if too many jobs try in parallel to access the SEs, the interface crashes
sleep = random.randint(0, 5 * 60)
j.setExecutable('sleep', str(sleep))
if NJobs == 0:
print("maximum number of jobs to submit reached")
print("breaking loop now")
break
else:
NJobs -= 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)
# j.setDestination( 'LCG.IN2P3-CC.fr' )
# mr_filter loses its executable property by uploading it to the GRID SE; reset
j.setExecutable('chmod', '+x mr_filter')
j.setExecutable('ls -lah')
for run_file in run_filelist:
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 querries.
# if too many jobs try in parallel to access the SEs, the interface crashes
sleep = random.randint(0, 5 * 60)
j.setExecutable('sleep', str(sleep))
""" simple hello world job """ from DIRAC.Interfaces.API.Job import Job from DIRAC.Interfaces.API.Dirac import Dirac from DIRAC.DataManagementSystem.Utilities.DMSHelpers import DMSHelpers j = Job() j.setName( "helloWorld-test" ) j.setExecutable( "exe-script.py", "", "Executable.log" ) # <-- user settings j.setCPUTime( 172800 ) tier1s = DMSHelpers().getTiers( tier = ( 0, 1 ) ) j.setBannedSites( tier1s ) # user settings --> # print j.workflow # submit the job to dirac result = Dirac().submitJob(j) print result
def Stereo(args=None):
from DIRAC.Interfaces.API.Dirac import Dirac
from DIRAC.Interfaces.API.Job import Job
user_script = './stereo.sh'
macro = './CTAstereo.C'
if (len(args) != 5):
Script.showHelp()
particle = args[0]
typeofdata = args[1]
direction = args[2]
zenith = args[3]
diffuse = args[4]
if typeofdata == 'train':
# The master layout with all the telescopes
candidates = './Prod3_3HB9All_Candidates.txt'
elif typeofdata == 'test':
# Different layouts
candidates = './Prod3_3HB9_Candidates_Full.txt'
else:
print "Invalid type of data definition!"
Script.showHelp()
return 1
if diffuse == "0":
diffName = "point"
elif diffuse == "1":
diffName = "diff"
else:
print "Invalid extension definition!"
Script.showHelp()
return 1
if zenith == "40":
zenName = "40deg"
elif zenith == "20":
zenName = "20deg"
else:
print "Invalid zenith definition!"
Script.showHelp()
return 1
if direction == "N":
directionName = "north"
# deg = "180"
elif direction == "S":
directionName = "south"
# deg = "0"
else:
print 'Wrong direction. It can only be "N" or "S".'
Script.showHelp()
return 1
filesPerJob = 5
site = "PARANAL"
listname = './training/gamma_trainLUT_%s_%s_%s.lfns' % (zenName, diffName, direction)
loop = 0
iJob = 0
# totalEntries /= (2*filesPerJob)
# print totalEntries
f = open(listname, 'r')
totalEntries = sum(1 for _ in f)
f = open(listname, 'r')
fileList = []
text_file_name = "lfnFiles_%s_%s_%s_%s.txt" % (particle, direction, zenName, diffuse)
text_file = open(text_file_name, "w")
for line in f:
loop = loop+1
infileLFN = line.strip()
# filein = os.path.basename(infileLFN)
fileList.append(infileLFN)
text_file.write("%s\n" % infileLFN)
remain = loop % filesPerJob
if iJob == 10:
break
if loop == totalEntries:
remain = 0
if remain == 0:
iJob = iJob+1
j = Job()
text_file.close()
j.setInputSandbox([user_script, "setupPackageMARS.sh", text_file_name, candidates, macro])
jobName = "%s %s %s %s %s %s %s %s" % (user_script, site, particle, typeofdata, directionName, zenName, diffName, iJob)
jobOut = "%s_%s_%s_%s_%s.out" % (user_script, site, typeofdata, directionName, iJob)
script_args = "%s %s %s %s %s %s %s" % (particle, typeofdata, direction, zenName, diffName, site, iJob)
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.GRIF.fr', 'ARC.SE-SNIC-T2.se'])
Script.gLogger.info(j._toJDL())
print "Submitting job %s" % (jobName)
Dirac().submit(j)
fileList = []
text_file = open(text_file_name, "w")
# dirac job created by ganga
from DIRAC.Interfaces.API.Job import Job
from DIRAC.Interfaces.API.Dirac import Dirac
j = Job()
dirac = Dirac()
# default commands added by ganga
j.setName("helloWorld-test")
j.setInputSandbox( ['/afs/cern.ch/user/f/fstagni/userJobs/_inputHello.tar.bz2', '/afs/cern.ch/user/f/fstagni/userJobs/hello-script.py'] )
j.setExecutable("exe-script.py","","Ganga_Executable.log")
# <-- user settings
j.setCPUTime(172800)
j.setBannedSites(['LCG.CERN.ch', 'LCG.CNAF.it', 'LCG.GRIDKA.de',
'LCG.IN2P3.fr', 'LCG.NIKHEF.nl', 'LCG.PIC.es', 'LCG.RAL.uk',
'LCG.SARA.nl'])
# user settings -->
#print j.workflow
# submit the job to dirac
result = dirac.submit(j)
print result
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")
