def trigger_stage_out(self, jobspec):
"""Trigger the stage-out procedure for the job.
Create a dummy output file to force harvester to wait until aCT
job is done
:param jobspec: job specifications
:type jobspec: JobSpec
:return: A tuple of return code (True: success, False: fatal failure, None: temporary failure)
and error dialog
:rtype: (bool, string)
"""
fileSpec = FileSpec()
fileSpec.PandaID = jobspec.PandaID
fileSpec.taskID = jobspec.taskID
fileSpec.lfn = 'dummy.{0}'.format(jobspec.PandaID)
fileSpec.scope = 'dummy'
fileSpec.fileType = 'output'
jobspec.add_in_file(fileSpec)
return True, ''
def run(self):
while True:
mainLog = self.make_logger(_logger, 'id={0}'.format(self.get_pid()), method_name='run')
mainLog.debug('getting number of jobs to be fetched')
# get number of jobs to be fetched
nJobsPerQueue = self.dbProxy.get_num_jobs_to_fetch(harvester_config.jobfetcher.nQueues,
harvester_config.jobfetcher.lookupTime)
mainLog.debug('got {0} queues'.format(len(nJobsPerQueue)))
# loop over all queues
for queueName, nJobs in iteritems(nJobsPerQueue):
# check queue
if not self.queueConfigMapper.has_queue(queueName):
continue
tmpLog = self.make_logger(_logger, 'queueName={0}'.format(queueName),
method_name='run')
# get queue
queueConfig = self.queueConfigMapper.get_queue(queueName)
# upper limit
if nJobs > harvester_config.jobfetcher.maxJobs:
nJobs = harvester_config.jobfetcher.maxJobs
# get jobs
tmpLog.debug('getting {0} jobs'.format(nJobs))
sw = core_utils.get_stopwatch()
siteName = queueConfig.siteName
jobs, errStr = self.communicator.get_jobs(siteName, self.nodeName,
queueConfig.get_source_label(),
self.nodeName, nJobs,
queueConfig.getJobCriteria)
tmpLog.info('got {0} jobs with {1} {2}'.format(len(jobs), errStr, sw.get_elapsed_time()))
# convert to JobSpec
if len(jobs) > 0:
jobSpecs = []
fileStatMap = dict()
sw_startconvert = core_utils.get_stopwatch()
for job in jobs:
timeNow = datetime.datetime.utcnow()
jobSpec = JobSpec()
jobSpec.convert_job_json(job)
jobSpec.computingSite = queueName
jobSpec.status = 'starting'
jobSpec.subStatus = 'fetched'
jobSpec.creationTime = timeNow
jobSpec.stateChangeTime = timeNow
jobSpec.configID = queueConfig.configID
jobSpec.set_one_attribute('schedulerID',
'harvester-{0}'.format(harvester_config.master.harvester_id))
if queueConfig.zipPerMB is not None and jobSpec.zipPerMB is None:
jobSpec.zipPerMB = queueConfig.zipPerMB
for tmpLFN, fileAttrs in iteritems(jobSpec.get_input_file_attributes()):
# check file status
if tmpLFN not in fileStatMap:
fileStatMap[tmpLFN] = self.dbProxy.get_file_status(tmpLFN, 'input',
queueConfig.ddmEndpointIn,
'starting')
# make file spec
fileSpec = FileSpec()
fileSpec.PandaID = jobSpec.PandaID
fileSpec.taskID = jobSpec.taskID
fileSpec.lfn = tmpLFN
fileSpec.endpoint = queueConfig.ddmEndpointIn
fileSpec.scope = fileAttrs['scope']
# set preparing to skip stage-in if the file is (being) taken care of by another job
if 'ready' in fileStatMap[tmpLFN] or 'preparing' in fileStatMap[tmpLFN] \
or 'to_prepare' in fileStatMap[tmpLFN]:
fileSpec.status = 'preparing'
else:
fileSpec.status = 'to_prepare'
if fileSpec.status not in fileStatMap[tmpLFN]:
fileStatMap[tmpLFN][fileSpec.status] = 0
fileStatMap[tmpLFN][fileSpec.status] += 1
fileSpec.fileType = 'input'
jobSpec.add_in_file(fileSpec)
jobSpec.trigger_propagation()
jobSpecs.append(jobSpec)
# insert to DB
tmpLog.debug("Converting of {0} jobs {1}".format(len(jobs),sw_startconvert.get_elapsed_time()))
sw_insertdb =core_utils.get_stopwatch()
self.dbProxy.insert_jobs(jobSpecs)
tmpLog.debug('Insert of {0} jobs {1}'.format(len(jobSpecs), sw_insertdb.get_elapsed_time()))
mainLog.debug('done')
# check if being terminated
if self.terminated(harvester_config.jobfetcher.sleepTime):
mainLog.debug('terminated')
return
queueConfig = queueConfigMapper.get_queue(queueName)
print queueConfig.stager
print queueConfig.stager['Globus_srcPath']
print queueConfig.stager['srcEndpoint']
print queueConfig.stager['Globus_dstPath']
print queueConfig.stager['dstEndpoint']
print queueConfig.stager['zipDir']
print "Initial queueConfig.stager = ", queueConfig.stager
queueConfig.stager['module'] = 'pandaharvester.harvesterstager.go_stager'
queueConfig.stager['name'] = 'GlobusStager'
print "Modified queueConfig.stager = ", queueConfig.stager
scope = 'panda'
fileSpec = FileSpec()
fileSpec.fileType = 'es_output'
fileSpec.lfn = 'panda.sgotest.' + uuid.uuid4().hex + '.gz'
fileSpec.fileAttributes = {}
assFileSpec = FileSpec()
assFileSpec.lfn = 'panda.sgotest.' + uuid.uuid4().hex
assFileSpec.fileType = 'es_output'
assFileSpec.fsize = random.randint(10, 100)
# create source file
hash = hashlib.md5()
hash.update('%s:%s' % (scope, fileSpec.lfn))
hash_hex = hash.hexdigest()
correctedscope = "/".join(scope.split('.'))
assFileSpec.path = "{endPoint}/{scope}/{hash1}/{hash2}/{lfn}".format(
endPoint=queueConfig.stager['Globus_srcPath'],
scope=correctedscope,
file_prefix = 'panda.sgotest.'
def exit_func():
for f in os.listdir('.'):
if f.startswith(file_prefix):
os.remove(f)
atexit.register(exit_func)
queueName = sys.argv[1]
queueConfigMapper = QueueConfigMapper()
queueConfig = queueConfigMapper.get_queue(queueName)
fileSpec = FileSpec()
fileSpec.fileType = 'output'
fileSpec.lfn = file_prefix + uuid.uuid4().hex + '.gz'
fileSpec.fileAttributes = {'guid': str(uuid.uuid4())}
fileSpec.checksum = '0d439274'
assFileSpec = FileSpec()
assFileSpec.lfn = file_prefix + uuid.uuid4().hex
assFileSpec.fileType = 'es_output'
assFileSpec.fsize = random.randint(10, 100)
assFileSpec.path = os.getcwd() + '/' + assFileSpec.lfn
oFile = open(assFileSpec.lfn, 'w')
oFile.write(''.join(random.choice(string.ascii_uppercase + string.digits) for _ in range(assFileSpec.fsize)))
oFile.close()
fileSpec.add_associated_file(assFileSpec)
jobSpec = JobSpec()
jobSpec.jobParams = {'outFiles': fileSpec.lfn + ',log',
jobSpec.jobParams = {
'scopeLog': 'panda',
'logFile': 'log',
}
jobSpec.computingSite = queueName
jobSpec.PandaID = job_id
jobSpec.modificationTime = datetime.datetime.now()
realDataset = 'panda.sgotest.' + uuid.uuid4().hex
ddmEndPointOut = 'BNL-OSG2_DATADISK'
outFiles_scope_str = ''
outFiles_str = ''
realDatasets_str = ''
ddmEndPointOut_str = ''
# create up 5 files for output
for index in range(random.randint(1, 5)):
fileSpec = FileSpec()
assFileSpec = FileSpec()
fileSpec.fileType = 'es_output'
assFileSpec.lfn = 'panda.sgotest.' + uuid.uuid4().hex
fileSpec.lfn = assFileSpec.lfn + '.gz'
fileSpec.scope = 'panda'
outFiles_scope_str += 'panda,'
outFiles_str += fileSpec.lfn + ','
realDatasets_str += realDataset + ","
ddmEndPointOut_str += ddmEndPointOut + ","
assFileSpec.fileType = 'es_output'
assFileSpec.fsize = random.randint(10, 100)
# create source file
hash = hashlib.md5()
hash.update('%s:%s' % (scope, fileSpec.lfn))
hash_hex = hash.hexdigest()
def run(self):
while True:
mainLog = self.make_logger(_logger,
'id={0}'.format(self.get_pid()),
method_name='run')
mainLog.debug('getting number of jobs to be fetched')
# get number of jobs to be fetched
nJobsPerQueue = self.dbProxy.get_num_jobs_to_fetch(
harvester_config.jobfetcher.nQueues,
harvester_config.jobfetcher.lookupTime)
mainLog.debug('got {0} queues'.format(len(nJobsPerQueue)))
# loop over all queues
for queueName, nJobs in iteritems(nJobsPerQueue):
# check queue
if not self.queueConfigMapper.has_queue(queueName):
continue
tmpLog = self.make_logger(_logger,
'queueName={0}'.format(queueName),
method_name='run')
# get queue
queueConfig = self.queueConfigMapper.get_queue(queueName)
# upper limit
if nJobs > harvester_config.jobfetcher.maxJobs:
nJobs = harvester_config.jobfetcher.maxJobs
# get jobs
default_prodSourceLabel = queueConfig.get_source_label()
pdpm = getattr(queueConfig,
'prodSourceLabelRandomWeightsPermille', {})
choice_list = core_utils.make_choice_list(
pdpm=pdpm, default=default_prodSourceLabel)
prodSourceLabel = random.choice(choice_list)
tmpLog.debug('getting {0} jobs for prodSourceLabel {1}'.format(
nJobs, prodSourceLabel))
sw = core_utils.get_stopwatch()
siteName = queueConfig.siteName
jobs, errStr = self.communicator.get_jobs(
siteName, self.nodeName, prodSourceLabel, self.nodeName,
nJobs, queueConfig.getJobCriteria)
tmpLog.info('got {0} jobs with {1} {2}'.format(
len(jobs), errStr, sw.get_elapsed_time()))
# convert to JobSpec
if len(jobs) > 0:
# get extractor plugin
if hasattr(queueConfig, 'extractor'):
extractorCore = self.pluginFactory.get_plugin(
queueConfig.extractor)
else:
extractorCore = None
jobSpecs = []
fileStatMap = dict()
sw_startconvert = core_utils.get_stopwatch()
for job in jobs:
timeNow = datetime.datetime.utcnow()
jobSpec = JobSpec()
jobSpec.convert_job_json(job)
jobSpec.computingSite = queueName
jobSpec.status = 'starting'
jobSpec.subStatus = 'fetched'
jobSpec.creationTime = timeNow
jobSpec.stateChangeTime = timeNow
jobSpec.configID = queueConfig.configID
jobSpec.set_one_attribute(
'schedulerID', 'harvester-{0}'.format(
harvester_config.master.harvester_id))
if queueConfig.zipPerMB is not None and jobSpec.zipPerMB is None:
jobSpec.zipPerMB = queueConfig.zipPerMB
fileGroupDictList = [
jobSpec.get_input_file_attributes()
]
if extractorCore is not None:
fileGroupDictList.append(
extractorCore.get_aux_inputs(jobSpec))
for fileGroupDict in fileGroupDictList:
for tmpLFN, fileAttrs in iteritems(fileGroupDict):
# check file status
if tmpLFN not in fileStatMap:
fileStatMap[
tmpLFN] = self.dbProxy.get_file_status(
tmpLFN, 'input',
queueConfig.ddmEndpointIn,
'starting')
# make file spec
fileSpec = FileSpec()
fileSpec.PandaID = jobSpec.PandaID
fileSpec.taskID = jobSpec.taskID
fileSpec.lfn = tmpLFN
fileSpec.endpoint = queueConfig.ddmEndpointIn
fileSpec.scope = fileAttrs['scope']
# set preparing to skip stage-in if the file is (being) taken care of by another job
if 'ready' in fileStatMap[tmpLFN] or 'preparing' in fileStatMap[tmpLFN] \
or 'to_prepare' in fileStatMap[tmpLFN]:
fileSpec.status = 'preparing'
else:
fileSpec.status = 'to_prepare'
if fileSpec.status not in fileStatMap[tmpLFN]:
fileStatMap[tmpLFN][fileSpec.status] = 0
fileStatMap[tmpLFN][fileSpec.status] += 1
if 'INTERNAL_FileType' in fileAttrs:
fileSpec.fileType = fileAttrs[
'INTERNAL_FileType']
jobSpec.auxInput = JobSpec.AUX_hasAuxInput
else:
fileSpec.fileType = 'input'
if 'INTERNAL_URL' in fileAttrs:
fileSpec.url = fileAttrs['INTERNAL_URL']
jobSpec.add_in_file(fileSpec)
jobSpec.trigger_propagation()
jobSpecs.append(jobSpec)
# insert to DB
tmpLog.debug("Converting of {0} jobs {1}".format(
len(jobs), sw_startconvert.get_elapsed_time()))
sw_insertdb = core_utils.get_stopwatch()
self.dbProxy.insert_jobs(jobSpecs)
tmpLog.debug('Insert of {0} jobs {1}'.format(
len(jobSpecs), sw_insertdb.get_elapsed_time()))
mainLog.debug('done')
# check if being terminated
if self.terminated(harvester_config.jobfetcher.sleepTime):
mainLog.debug('terminated')
return
jobSpec.modificationTime = datetime.datetime.now()
realDataset = 'panda.sgotest.' + uuid.uuid4().hex
ddmEndPointIn = 'BNL-OSG2_DATADISK'
inFiles_scope_str = ''
inFiles_str = ''
realDatasets_str = ''
realDatasetsIn_str = ''
ddmEndPointIn_str = ''
GUID_str = ''
fsize_str = ''
checksum_str = ''
scope_in_str = ''
# create up 5 files for input
for index in range(random.randint(1, 5)):
fileSpec = FileSpec()
assFileSpec = FileSpec()
fileSpec.fileType = 'input'
assFileSpec.lfn = 'panda.sgotest.' + uuid.uuid4().hex
fileSpec.lfn = assFileSpec.lfn
fileSpec.scope = 'panda'
inFiles_scope_str += 'panda,'
inFiles_str += fileSpec.lfn + ','
realDatasets_str += realDataset + ","
realDatasetsIn_str += realDataset + ","
ddmEndPointIn_str += ddmEndPointIn + ","
# some dummy inputs
GUID_str += 'd82e8e5e301b77489fd4da04bcdd6565,'
fsize_str += '3084569129,'
checksum_str += 'ad:9f60d29f,'
scope_in_str += 'panda,'
file_prefix = 'panda.sgotest.'
def exit_func():
for f in os.listdir('.'):
if f.startswith(file_prefix):
os.remove(f)
atexit.register(exit_func)
queueName = sys.argv[1]
queueConfigMapper = QueueConfigMapper()
queueConfig = queueConfigMapper.get_queue(queueName)
fileSpec = FileSpec()
fileSpec.fileType = 'output'
fileSpec.lfn = file_prefix + uuid.uuid4().hex + '.gz'
fileSpec.fileAttributes = {'guid': str(uuid.uuid4())}
fileSpec.checksum = '0d439274'
assFileSpec = FileSpec()
assFileSpec.lfn = file_prefix + uuid.uuid4().hex
assFileSpec.fileType = 'es_output'
assFileSpec.fsize = random.randint(10, 100)
assFileSpec.path = os.getcwd() + '/' + assFileSpec.lfn
oFile = open(assFileSpec.lfn, 'w')
oFile.write(''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(assFileSpec.fsize)))
oFile.close()
fileSpec.add_associated_file(assFileSpec)
def run(self):
while True:
mainLog = self.make_logger(_logger,
'id={0}'.format(self.ident),
method_name='run')
mainLog.debug('getting number of jobs to be fetched')
# get number of jobs to be fetched
nJobsPerQueue = self.dbProxy.get_num_jobs_to_fetch(
harvester_config.jobfetcher.nQueues,
harvester_config.jobfetcher.lookupTime)
mainLog.debug('got {0} queues'.format(len(nJobsPerQueue)))
# loop over all queues
for queueName, nJobs in iteritems(nJobsPerQueue):
# check queue
if not self.queueConfigMapper.has_queue(queueName):
continue
tmpLog = self.make_logger(_logger,
'queueName={0}'.format(queueName),
method_name='run')
# get queue
queueConfig = self.queueConfigMapper.get_queue(queueName)
# upper limit
if nJobs > harvester_config.jobfetcher.maxJobs:
nJobs = harvester_config.jobfetcher.maxJobs
# get jobs
tmpLog.debug('getting {0} jobs'.format(nJobs))
siteName = queueConfig.siteName
jobs, errStr = self.communicator.get_jobs(
siteName, self.nodeName, queueConfig.get_source_label(),
self.nodeName, nJobs, queueConfig.getJobCriteria)
tmpLog.info('got {0} jobs with {1}'.format(len(jobs), errStr))
# convert to JobSpec
if len(jobs) > 0:
jobSpecs = []
fileStatMap = dict()
for job in jobs:
timeNow = datetime.datetime.utcnow()
jobSpec = JobSpec()
jobSpec.convert_job_json(job)
jobSpec.computingSite = queueName
jobSpec.status = 'starting'
jobSpec.subStatus = 'fetched'
jobSpec.creationTime = timeNow
jobSpec.stateChangeTime = timeNow
jobSpec.configID = queueConfig.configID
jobSpec.set_one_attribute(
'schedulerID', 'harvester-{0}'.format(
harvester_config.master.harvester_id))
if queueConfig.zipPerMB is not None and jobSpec.zipPerMB is None:
jobSpec.zipPerMB = queueConfig.zipPerMB
for tmpLFN, fileAttrs in iteritems(
jobSpec.get_input_file_attributes()):
# check file status
if tmpLFN not in fileStatMap:
fileStatMap[
tmpLFN] = self.dbProxy.get_file_status(
tmpLFN, 'input',
queueConfig.ddmEndpointIn, 'starting')
# make file spec
fileSpec = FileSpec()
fileSpec.PandaID = jobSpec.PandaID
fileSpec.taskID = jobSpec.taskID
fileSpec.lfn = tmpLFN
fileSpec.endpoint = queueConfig.ddmEndpointIn
fileSpec.scope = fileAttrs['scope']
# set preparing to skip stage-in if the file is (being) taken care of by another job
if 'ready' in fileStatMap[tmpLFN] or 'preparing' in fileStatMap[tmpLFN] \
or 'to_prepare' in fileStatMap[tmpLFN]:
fileSpec.status = 'preparing'
else:
fileSpec.status = 'to_prepare'
if fileSpec.status not in fileStatMap[tmpLFN]:
fileStatMap[tmpLFN][fileSpec.status] = 0
fileStatMap[tmpLFN][fileSpec.status] += 1
fileSpec.fileType = 'input'
jobSpec.add_in_file(fileSpec)
jobSpec.trigger_propagation()
jobSpecs.append(jobSpec)
# insert to DB
self.dbProxy.insert_jobs(jobSpecs)
mainLog.debug('done')
# check if being terminated
if self.terminated(harvester_config.jobfetcher.sleepTime):
mainLog.debug('terminated')
return
import sys
import uuid
from pandaharvester.harvestercore.job_spec import JobSpec
from pandaharvester.harvestercore.file_spec import FileSpec
from pandaharvester.harvestercore.event_spec import EventSpec
from pandaharvester.harvestercore.communicator_pool import CommunicatorPool
rID = sys.argv[1]
taskid = rID.split('-')[0]
pandaid = long(rID.split('-')[1])
job = JobSpec()
job.PandaID = pandaid
event = EventSpec()
file = FileSpec()
file.status = 'finished'
file.objstoreID = 9575
file.pathConvention = 1000
file.lfn = str(uuid.uuid4().hex) + '.zip'
file.fsize = 555
file.chksum = '0d2a9dc9'
event.eventRangeID = rID
event.eventStatus = 'finished'
job.zipEventMap = {1: {'events':[event],
'zip':file}}
a = CommunicatorPool()
a.update_jobs([job])