def __get_job_options(self, runNumbers):
if not self.engine().submit_hook():
logging.warning(
"A job with the name {j} has already been submitted.".format(
j=self.engine().job_name()))
return
CreateDirectory(self.engine().config_dir(), True)
for r in runNumbers:
jobFolder = os.path.join(self.__joboptions_dir,
"{ddd}xxx".format(ddd=str(r)[:3]))
if not os.path.isdir(jobFolder):
logging.warning(
"Job option folder {f} for DSID {r} does not exist. Skipping {r}..."
.format(f=jobFolder, r=r))
continue
dir_to_copy = os.path.join(jobFolder, str(r))
if len(dir_to_copy) == 0:
continue
shutil.copytree(dir_to_copy,
os.path.join(self.engine().config_dir(), str(r)))
# assemble the config file for the job option
seeds = []
while len(seeds) < self.__nJobs:
s = random.uniform(100000, 500000)
if s not in seeds:
seeds += [s]
jo = [os.path.join(self.engine().config_dir(), str(r))][0]
out_dir = os.path.join(self.evgen_dir(), str(r))
WriteList(
(ReadListFromFile(self.seed_file()) if os.path.exists(
self.seed_file()) else []) + ["%d" % (i) for i in seeds],
self.seed_file(),
)
WriteList(
(ReadListFromFile(self.run_file())
if os.path.exists(self.run_file()) else []) +
["%d" % (r) for i in range(self.__nJobs)],
self.run_file(),
)
WriteList(
(ReadListFromFile(self.job_file())
if os.path.exists(self.job_file()) else []) +
[jo for i in range(self.__nJobs)],
self.job_file(),
)
WriteList(
(ReadListFromFile(self.out_file())
if os.path.exists(self.out_file()) else []) +
[out_dir for i in range(self.__nJobs)],
self.out_file(),
)
# submit the job array
self.__n_scheduled += self.__nJobs
logging.info("INFO <__get_job_options> Found %s" % (jo))
def submit_job(self):
if self.__submitted: return False
job_array = WriteList(
self.merge_lists(),
"%s/%s.txt" % (self.engine().config_dir(), id_generator(31)))
final_merge_name = WriteList(
self.temporary_files(),
"%s/%s.txt" % (self.engine().config_dir(), id_generator(30)))
if not self.engine().submit_array(
script="ClusterSubmission/Merge.sh",
sub_job=self.job_name(),
mem=self.engine().merge_mem(),
env_vars=[
("JobConfigList", job_array),
("OutFileList", final_merge_name),
("ALRB_rootVersion", ROOTVERSION),
],
hold_jobs=self.hold_jobs(),
run_time=self.engine().merge_time(),
array_size=len(self.merge_lists())):
return False
self.__submitted = True
if not self.child(): return True
if not self.child().submit_job(): return False
return self.engine().submit_clean_job(
hold_jobs=[self.engine().subjob_name(self.child().job_name())],
to_clean=self.temporary_files(),
sub_job=self.job_name())
def __extract_seeds(self, run):
try:
EVNT_DIR = [
os.path.join(self.evgen_dir(), R)
for R in os.listdir(self.evgen_dir()) if R.startswith(str(run))
][0]
except:
return
logging.info(
"<__extract_seeds> Searching {evntdir} for EVNT files not already processed in derivation format {d}."
.format(evntdir=EVNT_DIR, d=self.__derivation))
DERIVATION_DIR = os.path.join(self.aod_dir(),
EVNT_DIR[EVNT_DIR.rfind("/") + 1:])
CreateDirectory(DERIVATION_DIR, False)
Evnt_Seeds = [
int(E[E.find("EVNT") + 5:E.find(".pool")])
for E in os.listdir(EVNT_DIR) if E.endswith(".root")
]
DAOD_Seeds = [
int(A.split(".")[-2]) for A in os.listdir(DERIVATION_DIR)
if A.find(self.__derivation) != -1 and A.endswith(".root")
]
Non_ProcSeeds = [seed for seed in Evnt_Seeds if seed not in DAOD_Seeds]
if len(Non_ProcSeeds) == 0:
return
logging.info("Extracted seeds for run {r}:".format(r=run))
logging.info(" +-=- {s}".format(
s=", ".join([str(seed) for seed in Non_ProcSeeds])))
WriteList(
(ReadListFromFile(self.seed_file())
if os.path.exists(self.seed_file()) else []) +
[str(seed) for seed in Non_ProcSeeds],
self.seed_file(),
)
WriteList(
(ReadListFromFile(self.run_file())
if os.path.exists(self.run_file()) else []) +
[str(run) for seed in Non_ProcSeeds],
self.run_file(),
)
WriteList(
(ReadListFromFile(self.in_file()) if os.path.exists(self.in_file())
else []) + [EVNT_DIR for seed in Non_ProcSeeds],
self.in_file(),
)
self.__n_scheduled += len(Non_ProcSeeds)
def main():
CheckRucioSetup()
CheckRemainingProxyTime()
""""""
RunOptions = getArgumentParser().parse_args()
all_files = []
if RunOptions.single_out_file and len(RunOptions.out_file_name) == 0:
logging.error("Please provide a file name if you run with --single-out_file")
exit(1)
# Do we have one dataset, or a file with a list of them?
if os.path.exists(RunOptions.dataset):
with open(RunOptions.dataset) as dsfile:
for line in dsfile:
# Ignore comment lines and empty lines
if line.startswith('#'): continue
realline = line.strip()
if realline.find("_tid") > -1: realline = realline[0:realline.find("_tid")]
if not realline: continue # Ignore whitespace
if not RunOptions.single_out_file:
createFileList(realline, RunOptions)
else:
all_files += GetDataSetFiles(realline, RunOptions.RSE, RunOptions.protocols)
else:
createFileList(RunOptions.dataset, RunOptions)
if len(all_files) > 0:
WriteList(all_files, options.out_file)
def downloadDataSets(InputDatasets, Destination, RSE="", use_singularity=False):
### Apparently rucio does no longer work in combination with AthAnalysis. So let's
### execute it from a singulartity container
Cmds = []
image_to_choose = setupBatchSubmitArgParser().get_default("SingularityImage")
home_dir = setupBatchSubmitArgParser().get_default("BaseFolder") + "/TMP/.singularity/"
CreateDirectory(Destination, False)
if use_singularity:
CreateDirectory(home_dir, False)
to_clean = []
for DS in InputDatasets:
if not use_singularity:
Cmds += ["rucio download %s --ndownloader 32 %s --dir %s" % (DS, "" if len(RSE) == 0 else "--rse %s" % (RSE), Destination)]
else:
singularity_dir = home_dir + "/" + id_generator(21)
to_clean += [singularity_dir]
singularity_script = WriteList([
"#!/bin/bash",
"export ATLAS_LOCAL_ROOT_BASE=/cvmfs/atlas.cern.ch/repo/ATLASLocalRootBase",
"source ${ATLAS_LOCAL_ROOT_BASE}/user/atlasLocalSetup.sh",
"lsetup rucio",
"echo 'rucio download %s --ndownloader 32 %s --dir %s'" % (DS, "" if len(RSE) == 0 else "--rse %s" % (RSE), Destination),
"rucio download %s --ndownloader 32 %s --dir %s" % (DS, "" if len(RSE) == 0 else "--rse %s" % (RSE), Destination),
], "%s/to_exec.sh" % (singularity_dir))
os.system("chmod 0777 " + singularity_script)
Cmds += [
"singularity exec --cleanenv -H %s:/alrb -B %s:/srv %s/%s %s" %
(singularity_dir, Destination, SINGULARITY_DIR, image_to_choose, singularity_script)
]
ExecuteCommands(ListOfCmds=Cmds, MaxCurrent=8)
for c in to_clean:
os.system("rm -rf %s" % (c))
def write_ship_file(self, env_vars):
ship_file_name = WriteList(["#!/bin/bash"] + [
"export %s='%s'" % (var, val)
for var, val in env_vars + self.common_env_vars()
], "%s/%s.sh" % (self.config_dir(), id_generator(74)))
os.system("chmod 0700 %s" % (ship_file_name))
return ship_file_name
def submit_move_job(self,
hold_jobs=[],
to_move=[],
destination="",
source_dir="",
sub_job=""):
move_cfg = ""
if len(to_move) > 0:
move_cfg = "%s/Move_%s.txt" % (self.config_dir(), id_generator(35))
WriteList(to_move, move_cfg)
elif len(source_dir) > 0:
move_cfg = source_dir
else:
logging.error("<submit_move_job> Nothing to move")
return False
if len(destination) == 0:
logging.error(
"<submit_move_job> No destination where to move provided")
return False
return self.submit_job(script="ClusterSubmission/Move.sh",
mem=100,
env_vars=[
("DestinationDir", destination),
("FromDir", move_cfg),
],
hold_jobs=hold_jobs,
sub_job="Move%s%s" %
("" if len(sub_job) == 0 else "-", sub_job),
run_time="01:00:00")
def updateBlackList(black_list):
current_black = getPRWblackList()
if IsListIn(black_list, current_black): return
current_black = ClearFromDuplicates(current_black + black_list)
current_dir = os.getcwd()
FileName = os.path.realpath(
ResolvePath("XAMPPbase/BlackListedPRWdatasets.txt"))
Pkg_Dir = os.path.realpath(ResolvePath("XAMPPbase"))
###############################################################################
# Find out the current branch to propagage only #
# the updated List to the main repository. Other changes regarding #
# side developments of the package should not be propagated yet #
###############################################################################
upstream = setupGITupstream()
current_branch = getBranch()
os.chdir(Pkg_Dir)
new_branch = "PRW_%s_%s" % (time.strftime("%Y%m%d"), USERNAME)
if current_branch:
os.system(
"git commit -am \"Commit changes of all files in order to push the 'BlackListedPRWdatasets.txt'\""
)
print "INFO: Create new branch %s to update the BlackListedPRWdatasets " % (
new_branch)
os.system("git checkout -b %s %s/master" % (new_branch, upstream))
WriteList(sorted(current_black), FileName)
os.system("git add BlackListedPRWdatasets.txt")
os.system(
"git commit BlackListedPRWdatasets.txt -m \"Updated the list of black prw files\""
)
os.system("git push %s %s" % (upstream, new_branch))
if current_branch: os.system("git checkout %s" % (current_branch))
os.chdir(current_dir)
def insertPRWUser(user):
Users = getUsersSubmittedPRW()
if user in Users: return
Users += [user]
current_dir = os.getcwd()
FileName = os.path.realpath(
ResolvePath("XAMPPbase/UsersWhoSubmittedPRW.txt"))
Pkg_Dir = os.path.realpath(ResolvePath("XAMPPbase"))
###############################################################################
# Find out the current branch to propagage only #
# the updated List to the main repository. Other changes regarding #
# side developments of the package should not be propagated yet #
###############################################################################
upstream = setupGITupstream()
current_branch = getBranch()
os.chdir(Pkg_Dir)
new_branch = "PRW_" + user.replace(".", "_")
if current_branch:
os.system(
"git commit -am \"Commit changes of all files in order to push the 'UsersWhoSubmittedPRW.txt'\""
)
print "INFO: Create new branch %s to update the UsersWhoSubmittedPRW " % (
new_branch)
os.system("git checkout -b %s %s/master" % (new_branch, upstream))
print "INFO: %s submitted to the grid prw_config jobs. Add him to the common list such that others can download his files" % (
user)
WriteList(sorted(Users), FileName)
os.system("git add UsersWhoSubmittedPRW.txt")
os.system(
"git commit UsersWhoSubmittedPRW.txt -m \"Added %s to the list of users who submitted a prw config creation job\""
% (user))
os.system("git push %s %s" % (upstream, new_branch))
if current_branch: os.system("git checkout %s" % (current_branch))
os.chdir(current_dir)
def run(self):
CreateDirectory(self.download_dir(), False)
CreateDirectory(self.final_directory(), False)
#self.__datasets = sorted(self.__datasets, key=lambda x: GetPRW_datasetID(x))
DownloadList = [
'rucio download --ndownloader 5 --dir %s %s' %
(self.download_dir(), ds) for ds in self.__datasets
if ds not in self.__already_on_disk
]
ExecuteCommands(ListOfCmds=DownloadList, MaxCurrent=16)
self.clearFromDuplicates(self.download_dir())
Files = []
for dir in os.listdir(self.download_dir()):
dir_path = "%s/%s" % (self.download_dir(), dir)
if not os.path.isdir(dir_path): continue
if not self.hasDataset(dir): continue
Files += [
"%s/%s" % (dir_path, F) for F in os.listdir(dir_path)
if IsROOTFile(F)
]
WriteList(sorted(Files), "%s/temp_in.txt" % (self.download_dir()))
# only 1 entry in the MCPileupReweighting tree per Channel/RunNumber combination is actually needed
# thus, remove all others but one in order to significantly reduce the files size!
# This is done by the SlimPRWFile macro in XAMPPbase/utils/
MergeCmd = "SlimPRWFile --InList %s/temp_in.txt --outFile %s" % (
self.download_dir(), self.final_file())
print MergeCmd
os.system(MergeCmd)
print "INFO: Clean up the temporary file"
os.system("rm %s/temp_in.txt " % (self.download_dir()))
self.standaloneCheck()
print "INFO: Done"
def write_status_log(file_location, grid_jobs):
if len(grid_jobs) > 0:
max_task_letters = max([len(str(job.jediTaskID)) for job in grid_jobs])
max_status_letters = max(
[len(str(job.taskStatus)) for job in grid_jobs])
max_job_name_letters = max([len(job.jobName) for job in grid_jobs])
max_progress_letters = max(
[len(make_progess(job)) for job in grid_jobs])
log_file = []
for job in sorted(grid_jobs,
key=lambda x: get_progress(x)[0],
reverse=True):
proc_frac, done_jobs, all_jobs = get_progress(job)
log_file += [
"https://bigpanda.cern.ch/task/%s/ %s %s %s %s %s %s %s(%.2f%%) %s"
%
(job.jediTaskID,
FillWhiteSpaces(max_task_letters - len(str(job.jediTaskID))),
job.jobName,
FillWhiteSpaces(max_job_name_letters - len(job.jobName)),
job.taskStatus,
FillWhiteSpaces(max_status_letters - len(job.taskStatus)),
make_progess(job),
FillWhiteSpaces(max_progress_letters -
len(make_progess(job))), proc_frac, job.inDS)
]
WriteList(log_file, file_location)
def submit_copy_job(
self,
hold_jobs=[],
to_copy=[], ### Give particular files to copy
destination="",
source_dir="", ### Optional
sub_job=""):
copy_cfg = ""
if len(to_copy) > 0:
copy_cfg = "%s/Copy_%s.txt" % (self.config_dir(), id_generator(35))
WriteList(to_copy, copy_cfg)
elif len(source_dir) > 0:
copy_cfg = source_dir
else:
print "<submit_copy_job> Nothing to copy"
return False
if len(destination) == 0:
print "<Submit_copy_job> Where to copy everything?"
return False
return self.submit_job(script="ClusterSubmission/Copy.sh",
mem=100,
env_vars=[
("DestinationDir", destination),
("FromDir", copy_cfg),
],
hold_jobs=hold_jobs,
sub_job="Copy%s%s" %
("" if len(sub_job) == 0 else "-", sub_job),
run_time="01:00:00")
def submit_clean_job(self, hold_jobs=[], to_clean=[], sub_job=""):
clean_cfg = "%s/Clean_%s.txt" % (self.config_dir(), id_generator(35))
WriteList(to_clean, clean_cfg)
return self.submit_job(script="ClusterSubmission/Clean.sh",
mem=100,
env_vars=[("ToClean", clean_cfg)],
hold_jobs=hold_jobs,
sub_job="Clean%s%s" %
("" if len(sub_job) == 0 else "-", sub_job),
run_time="01:00:00")
def __assemble_merge_list(self, files_to_merge):
copied_in = [x for x in files_to_merge]
if self.__shuffle_files: shuffle(copied_in)
merge_lists = []
merge_in = []
for i, fi in enumerate(copied_in):
if i > 0 and i % self.__files_per_job == 0:
merge_name = "%s/%s.txt" % (self.engine().config_dir(),
id_generator(85))
WriteList(merge_in, merge_name)
merge_lists += [merge_name]
merge_in = []
merge_in += [fi]
### Pack the last remenants into a last merge job
if len(merge_in) > 0:
merge_name = "%s/%s.txt" % (self.engine().config_dir(),
id_generator(85))
WriteList(merge_in, merge_name)
merge_lists += [merge_name]
return merge_lists
def createFileList(dsname, options):
logging.info('Creating file list for ' + dsname)
DS = GetDataSetFiles(dsname, options.RSE, options.protocols)
if len(DS) == 0:
logging.error("No datasets found")
return
if dsname.find(":") > -1:
dsname = dsname[dsname.find(":") + 1:len(dsname)]
CreateDirectory(options.OutDir, False)
filelistname = options.OutDir + "/" + dsname.rstrip('/') + ".txt"
if os.path.exists(filelistname) == True:
logging.info("Remove the old FileList")
os.system("rm " + filelistname)
WriteList(DS, filelistname)
def createFileList(dsname, options):
prettyPrint('Creating file list for', dsname)
DS = GetDataSetFiles(dsname, options.RSE, options.protocols)
if len(DS) == 0:
print "No datasets found"
return
if dsname.find(":") > -1:
dsname = dsname[dsname.find(":") + 1:len(dsname)]
OutDir = options.OutDir
if os.path.exists(OutDir) == False:
print "mkdir -p " + OutDir
os.system("mkdir -p " + OutDir)
filelistname = OutDir + "/" + dsname.rstrip('/') + ".txt"
if os.path.exists(filelistname) == True:
print "Remove the old FileList"
os.system("rm " + filelistname)
WriteList(DS, filelistname)
def pack_environment(self, env_vars, script):
exec_script = self.link_to_copy_area(script)
if not exec_script: return False
ship_file = self.write_ship_file(env_vars)
if self.run_singularity():
ship_file = self.write_ship_file([
("CONTAINER_SCRIPT", exec_script),
("CONTAINER_IMAGE", self.singularity_container()),
("CONTAINER_SHIPING_FILE", ship_file),
])
exec_script = self.link_to_copy_area(
ResolvePath("ClusterSubmission/Singularity.sh"))
env_script = WriteList([
"#!/bin/bash",
"source %s" % (ship_file),
"source %s" % (exec_script)
], "%s/EnvScript_%s.sh" % (self.config_dir(), id_generator(50)))
os.system("chmod 0700 %s" % (env_script))
return env_script
def submit_job(self):
WriteList(
[
D.replace(self.evgen_dir(), self.aod_dir())
for D in ReadListFromFile(self.in_file())
],
self.out_file(),
)
extra_args = ""
if len(self.__preExec) > 0:
extra_args += ' --preExec "%s" ' % (self.__preExec)
if len(self.__preInclude) > 0:
extra_args += ' --preInclude "%s" ' % (self.__preInclude)
if len(self.__postExec) > 0:
extra_args += ' --postExec "%s" ' % (self.__postExec)
if len(self.__postInclude) > 0:
extra_args += ' --postInclude "%s" ' % (self.__postInclude)
if not self.engine().submit_array(
sub_job=self.__derivation,
script="SubmitMC/batch_derivation.sh",
mem=self.__mem,
env_vars=[
("SeedFile", self.seed_file()),
("RunFile", self.run_file()),
("InFile", self.in_file()),
("OutFile", self.out_file()),
("DERIVATION_DIR", self.aod_dir()),
("DerivationRelease", self.__derivRelease),
("DerivationCache", self.__derivCache),
("ReductionConf", self.__derivation),
("ExtraArgs", extra_args),
],
hold_jobs=self.hold_jobs(),
run_time=self.__run_time,
array_size=self.n_scheduled(),
):
return False
return True
def submit_build_job(self):
if self.__submitted_build:
print "ERROR <submit_build_job>: Build job is already submitted"
return False
if not self.submit_hook(): return False
if self.send_build_job() and not self.submit_job(
script="ClusterSubmission/Build.sh",
sub_job="Build",
mem=self.__buildMem,
env_vars=[("CleanOut", self.out_dir()),
("CleanLogs", self.log_dir()),
("CleanTmp", self.tmp_dir()),
("nCoresToUse", self.__buildCores),
("COPYAREA", self.build_dir())],
run_time=self.__buildTime,
hold_jobs=self.__holdBuild):
return False
elif not self.send_build_job():
if not CreateDirectory(self.log_dir(),
False) or not CreateDirectory(
self.out_dir(), False):
return False
Dummy_Job = "%s/%s.sh" % (self.config_dir(), id_generator(35))
WriteList([
"#!/bin/bash",
"echo \"I'm a dummy build job. Will wait 60 seconds until everything is scheduled\"",
"sleep 120"
], Dummy_Job)
if not self.submit_job(script=Dummy_Job,
sub_job="Build",
mem=100,
env_vars=[],
run_time="00:05:00",
hold_jobs=self.__holdBuild):
return False
self.__submitted_build = True
self.lock_area()
return True
def finish(self):
if len(self.__job_dependency_dict) == 0:
logging.debug("Nothing has been scheduled")
return False
dag_content = []
for job in self.__job_dependency_dict:
dag_content += job.get_job_config_str()
dag_content += ["\n\n\n"]
for job in self.__job_dependency_dict:
dag_content += job.get_dependency_str()
dag_dir = self.log_dir() + "/DAG/"
dag_location = WriteList(dag_content,
"%s/%s.dag" % (dag_dir, self.job_name()))
os.chdir(dag_dir)
cmd = "condor_submit_dag -verbose -maxidle %d %s %s.dag" % (
self.max_running_per_array(),
("-append '+MyProject = \"%s\"'" % self.accountinggroup()
if self.accountinggroup() else ""), self.job_name())
return not os.system(cmd)
def submit_build_job(self):
if self.check_submitted_build():
logging.warning(
"<submit_build_job>: Build job is already submitted")
return True
if not self.submit_hook(): return False
### Few cluster engines go crazy if the log files of the own jobs are deleted
### Make sure that the build job deletes the log dir before submission
if not CreateDirectory(self.log_dir(), True): return False
if self.send_build_job() and not self.submit_job(
script="ClusterSubmission/Build.sh",
sub_job="Build",
mem=self.get_build_mem(),
n_cores=self.get_build_cores(),
env_vars=[("CleanOut", self.out_dir()),
("CleanTmp", self.tmp_dir()),
("nCoresToUse", self.get_build_cores()),
("COPYAREA", self.build_dir())],
run_time=self.get_build_time(),
hold_jobs=self.get_build_hold_jobs()):
return False
elif not self.send_build_job():
if not CreateDirectory(self.out_dir(), False): return False
Dummy_Job = WriteList([
"#!/bin/bash",
"echo \"I'm a dummy build job. Will wait 15 seconds until everything is scheduled\"",
"sleep 15"
], "%s/%s.sh" % (self.config_dir(), id_generator(35)))
if not self.submit_job(script=Dummy_Job,
sub_job="Build",
mem=100,
run_time="00:05:00",
hold_jobs=self.__holdBuild):
return False
self.__submitted_build = True
self.lock_area()
return True
def download_ci_files(options):
### Retrieve first the EOS token
getEOS_token(options)
### Check first whether the CI dir actually exits
smp_dir = "%s/datasamples/" % (options.ciDir)
if not os.path.isdir(smp_dir):
print "ERROR: The path to look up for the data samples %s does not exists. Where is my data" % (
smp_dir)
exit(1)
### Create first the directory to store the temporary files in there
### Clean the old remants
CreateDirectory(options.TEMPdir, True)
downloaded_smp = []
for smp in os.listdir(smp_dir):
smp_name = smp[:smp.rfind(".")]
print "INFO: Download the files from sample %s" % (smp_name)
download_to = "%s/%s" % (options.TEMPdir, smp_name)
CreateDirectory(download_to, False)
### Download the files first
for file_to_load in ReadListFromFile("%s/%s" % (smp_dir, smp)):
destination_file = "%s/%s" % (
download_to, file_to_load[file_to_load.rfind("/") + 1:])
CopyCmd = "xrdcp %s/%s %s" % (options.EOSpath, file_to_load,
destination_file)
if os.path.exists(destination_file):
print "INFO: Omit do download %s" % (file_to_load)
elif os.system(CopyCmd) != 0:
print "ERROR: Failed to download %s" % (file_to_load)
exit(1)
### Write the file list for the analysis
file_list = "%s/FileList_%s.txt" % (options.TEMPdir, smp_name)
WriteList([
"%s/%s" % (download_to, f[f.rfind("/") + 1:])
for f in ReadListFromFile("%s/%s" % (smp_dir, smp))
], file_list)
downloaded_smp += [smp_name]
return downloaded_smp
def write_broken_log(options, broken_jobs):
if len(broken_jobs) > 0:
logging.info("Found %d unhealthy jobs will prompt them below" %
(len(broken_jobs)))
max_task_letters = max(
[len(str(job.jediTaskID)) for job in broken_jobs])
max_status_letters = max(
[len(str(job.taskStatus)) for job in broken_jobs])
max_job_name_letters = max([len(job.jobName) for job in broken_jobs])
log_file = []
for job in sorted(broken_jobs, key=lambda x: x.jobName):
log_file += [
"https://bigpanda.cern.ch/task/%s/ %s %s %s %s %s %s" %
(job.jediTaskID,
FillWhiteSpaces(max_task_letters - len(str(job.jediTaskID))),
job.taskStatus,
FillWhiteSpaces(max_status_letters - len(job.taskStatus)),
job.jobName,
FillWhiteSpaces(max_job_name_letters - len(job.jobName)),
job.inDS)
]
logging.info(log_file[-1])
WriteList(log_file, options.broken_log_file)
chan_in_old = readPRWchannels("%s/%s" % (RunOptions.oldPRWDir, new))
chan_in_new = readPRWchannels("%s/%s" % (RunOptions.newPRWDir, new))
messages = []
AnythingNew = False
### Compare the prw channels of both files
for c in chan_in_new:
if not c in chan_in_old:
messages += ["INFO: Channel %d (%s) has been added through the last iteration to %s" % (c, MyxSecDB.name(c), new)]
AnythingNew = True
else:
chan_in_old.remove(c)
### The old file somehow contains additional channels. We need to double check
if len(chan_in_old) > 0:
messages += [
"WARNING: The following channels were merged into the old file %s/%s but are no longer present in %s/%s" %
(RunOptions.oldPRWDir, new, RunOptions.newPRWDir, new)
]
for c in chan_in_old:
messages += [" -=-=- %d (%s)" % (c, MyxSecDB.name(c))]
if not AnythingNew:
messages += ["INFO: Nothing new has been added to %s/%s w.r.t %s/%s" % (RunOptions.oldPRWDir, new, RunOptions.newPRWDir, new)]
WriteList(messages, "PRWcheck_%s.log" % (new[:new.rfind(".")]))
if len(chan_in_old) > 0 and AnythingNew and RunOptions.uniteFiles:
Cmd = "SlimPRWFile --inFile %s/%s --inFile %s/%s --outFile %s/%s --InIsSlimmed" % (
RunOptions.newPRWDir, new, RunOptions.oldPRWDir, new, RunOptions.newPRWDir, new)
os.system(Cmd)
def _write_submission_file(self,
sub_job,
exec_script,
env_vars=[],
mem=1,
run_time='00:00:01',
nproc=1,
arraylength=-1):
self.set_cluster_control_module(
"ClusterSubmission/ClusterControlHTCONDOR.sh")
if not exec_script:
logging.error("<_write_submission_file> No exec_script was given!")
return False
if mem < 0:
logging.error(
"<_write_submission_file> No memory requirement for the job was specified."
)
return False
job_name = self.subjob_name(sub_job)
if len([
x for x in self.__job_dependency_dict
if x.getJobName() == job_name
]):
logging.error(
"The job %s has already been defined. Please ensure unique job names"
% (job_name))
return False
log_string = "%s/%s%s" % (
self.log_dir(), sub_job if len(sub_job) else job_name,
"_$(CONDOR_TASK_ID)" if arraylength > 0 else "")
exec_script = self.pack_environment(env_vars, exec_script)
submision_content = []
submision_content += [
"universe = vanilla",
"executable = %s" % (exec_script),
"output = %s.out" % (log_string),
"error = %s.err" % (log_string),
"log = %s.log" % (log_string),
#"transfer_executable = True",
"notification = Error",
"notify_user = %s" % (self.mail_user()),
"request_memory = %d" % (mem),
"on_exit_remove = (ExitBySignal == False) && (ExitCode == 0)",
"request_cpus = %d" % (nproc),
#### Extra attributes
"+MaxRuntime = %d" %
(TimeToSeconds(run_time)), ### CERN cloud
"+RequestRuntime = %d" %
(TimeToSeconds(run_time)), ### DESY cloud
"+MyProject = %s" %
(self.accountinggroup()) if self.accountinggroup() else "",
]
if arraylength > 0:
submision_content += [
"environment = CONDOR_TASK_ID=$(CONDOR_TASK_ID)"
]
submision_content += [
"queue",
]
self.__job_dependency_dict += [
HTCondorJob(job_name=job_name,
submission_file=WriteList(
submision_content, "%s/%s.sub" %
(self.config_dir(), id_generator(25))),
arraylength=arraylength,
engine=self,
abbreviation=self.__assign_abb_letter())
]
self.__submitted_jobs += 1 if arraylength <= 1 else arraylength
return True
def lock_area(self):
WriteList(
["###Hook file to prevent double submission of the same job"],
"%s/.job.lck" % (self.tmp_dir()))
while AOD.rfind("_r") != AOD.find("_r"):
AOD = AOD[:AOD.rfind("_r")]
### Remote the double e -tag
while AOD.rfind(".e") < AOD.rfind("_e"):
uscore_pos = AOD.find("_", AOD.rfind(".e"))
AOD = AOD[:uscore_pos] + AOD[AOD.find("_", uscore_pos + 1):]
return AOD
if __name__ == '__main__':
OutDir = os.getcwd()
parser = argparse.ArgumentParser(
description='This script converts DAOD filelists to AOD filelists which then can be used for creating pileup reweighting files.',
prog='CreateAODFromDAODList',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--datasets', '-d', '-D', help='DAOD filelist to be converted into AOD', required=True)
parser.add_argument('--outFile', help="pipe the output into a script into a file", default='')
RunOptions = parser.parse_args()
logging.info('The following DAODs are converted into ADOs:\n')
DAODsToConvert = [convertToAOD(daod) for daod in ReadListFromFile(RunOptions.datasets)]
logging.info('\nThe ADOs are:\n')
for daod in DAODsToConvert:
logging.info(" --- %s" % (daod))
if len(RunOptions.outFile) > 0: WriteList(DAODsToConvert, RunOptions.outFile)
def main():
"""List datasets located at a RSE location."""
CheckRucioSetup()
CheckRemainingProxyTime()
RunOptions = getArgumentParser().parse_args()
Today = time.strftime("%Y-%m-%d")
Patterns = RunOptions.pattern
OutDir = RunOptions.OutDir
RSE = RunOptions.RSE
if ',' in RSE:
RSE = RSE.split(
','
)[0] # in case people have more than one RSE in their environment variable for grid submits
Prefix = ''
if RunOptions.MyRequests:
Prefix = 'MyRequestTo_'
DS = ListUserRequests(RSE, RunOptions.rucio)
else:
DS = ListDisk(RSE)
### MetaFile = open("Content_%s.txt"%(RSE), 'w')
### for DataSet, Size in ListDiskWithSize(RSE):
### Owner, ID = GetDataSetInfo(DataSet,RSE)
### line = "%s | %s | %s | %.2f GB"%(ID, Owner,DataSet, Size)
### MetaFile.write("%s\n"%(line))
### print line
### MetaFile.close()
### exit(0)
if len(DS) == 0:
logging.warning("Disk is empty.")
exit(0)
CreateDirectory(OutDir, False)
###########
# Define the file list name
###########
FileList = "%s%s_%s" % (Prefix, RSE, Today)
if len(Patterns) > 0: FileList += "_%s" % ('_'.join(Patterns))
if len(RunOptions.exclude) > 0:
FileList += "_exl_%s" % ('_'.join(RunOptions.exclude))
FileList += '.txt'
Write = []
for d in sorted(DS):
allPatternsFound = True
for Pattern in Patterns:
if not Pattern in d:
allPatternsFound = False
break
for Pattern in RunOptions.exclude:
if Pattern in d:
allPatternsFound = False
break
if allPatternsFound:
IsInWrite = False
if d.split(".")[-1].isdigit(): d = d[:d.rfind(".")]
if d.find("_tid") != -1: d = d[0:d.rfind("_tid")]
if len([w for w in Write if w.find(d) != -1]) > 0: continue
logging.info("Write dataset %s" % (d))
Write.append(d)
if len(Write) == 0:
logging.error("No datasets containing given pattern(s) found!")
exit(0)
WriteList(Write, "%s/%s" % (OutDir, FileList))
logging.info("Datasets written to file %s/%s" % (OutDir, FileList))
OutDir = os.getcwd()
parser = argparse.ArgumentParser(
description=
'This script converts DAOD filelists to AOD filelists which then can be used for creating pileup reweighting files.',
prog='CreateAODFromDAODList',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--datasets',
'-d',
'-D',
help='DAOD filelist to be converted into AOD',
required=True)
parser.add_argument('--outFile',
help="pipe the output into a script into a file",
default='')
RunOptions = parser.parse_args()
print 'The following DAODs are converted into ADOs:\n'
DAODsToConvert = [
convertToAOD(daod) for daod in ReadListFromFile(RunOptions.datasets)
]
print '\nThe ADOs are:\n'
for daod in DAODsToConvert:
print " --- %s" % (daod)
if len(RunOptions.outFile) > 0:
WriteList(DAODsToConvert, RunOptions.outFile)
def __prepare_input(self, in_ds=""):
print "INFO <_prepare_input>: Assemble configuration for %s" % (in_ds)
### Name to be piped to the job
out_name = in_ds[in_ds.rfind("/") + 1:in_ds.rfind(".")] if IsTextFile(
in_ds) or IsROOTFile(in_ds) else in_ds
split_dir = "%s/Datasets/%s" % (self.split_cfg_dir(), out_name)
root_files = []
### Now we need to find the corresponding ROOT files
### 1) The dataset is a root file itself
if IsROOTFile(in_ds):
root_files += [in_ds]
### 2) The given dataset is a .txt file
elif IsTextFile(in_ds):
### Find the root files from there
root_files = self.__extract_root_files(in_ds)
if len(root_files) == 0: return False
### 3) The given dataset is a directory
elif os.path.isdir(in_ds):
if in_ds.endswith("/"):
in_ds = in_ds[:in_ds.rfind("/")]
out_name = in_ds[in_ds.rfind("/") + 1:]
split_dir = "%s/Directory/%s" % (self.split_cfg_dir(), out_name)
root_files = [
"%s/%s" % (in_ds, F) for F in os.listdir(in_ds)
if IsROOTFile(F)
]
### 4) It's a logical dataset stored on d-cache
else:
root_files = self.__find_on_dcache(in_ds)
if len(root_files) == 0:
print "ERROR: Could not associate anything to %s" % (in_ds)
return False
if len(out_name) == 0:
print "ERROR: How should the output be called %s" % (in_ds)
return False
### Assemble the splitting of the jobs
main_list = "%s/AllROOTFiles.main" % (split_dir)
files_in_main = ReadListFromFile(main_list) if os.path.exists(
main_list) else []
### The list is unkown or the content of ROOT files has changed
### Redo the splitting again ;-)
if len(files_in_main) != len(root_files) or not IsListIn(
files_in_main, root_files):
print "INFO: Assemble new split for %s" % (in_ds)
CreateDirectory(split_dir, True)
WriteList(root_files, main_list)
os.system("CreateBatchJobSplit -I %s -O %s -EpJ %i" %
(main_list, split_dir, self.__events_per_job))
### Each of the lists contains the ROOT files to process per each sub job
split_lists = [
"%s/%s" % (split_dir, F) for F in os.listdir(split_dir)
if IsTextFile(F)
]
n_jobs = len(split_lists)
subjob_outs = [
"%s/%s_%d.root" % (self.engine().tmp_dir(), out_name, d)
for d in range(n_jobs)
]
assembled_in = [] if not os.path.exists(
self.job_input()) else ReadListFromFile(self.job_input())
assembled_out = [] if not os.path.exists(
self.job_out_names()) else ReadListFromFile(self.job_out_names())
start_reg = len(assembled_in)
### Write what we've
WriteList(assembled_in + split_lists, self.job_input())
WriteList(assembled_out + subjob_outs, self.job_out_names())
#### Submit the merge jobs
self.__merge_interfaces += [
self.engine().create_merge_interface(
out_name=out_name,
files_to_merge=subjob_outs,
hold_jobs=[(self.engine().job_name(),
[start_reg + i + 1 for i in range(n_jobs)])],
files_per_job=self.__files_per_merge_itr,
final_split=self.__final_split)
]
self.__nsheduled += n_jobs
return True
