def validateOptions(self):
SubCommand.validateOptions(self)
#check the format of jobids
self.jobids = ''
if getattr(self.options, 'jobids', None):
self.jobids = validateJobids(self.options.jobids)
def __call__(self): # pylint: disable=arguments-differ
if self.options.short:
taskname = self.cachedinfo['RequestName']
inputlist = {'subresource': 'search', 'workflow': taskname}
server = self.crabserver
webdir = getProxiedWebDir(crabserver=self.crabserver, task=taskname, logFunction=self.logger.debug)
dictresult, status, reason = server.get(api='task', data=inputlist)
if not webdir:
webdir = dictresult['result'][0]
self.logger.info('Server result: %s' % webdir)
if status != 200:
msg = "Problem retrieving information from the server:\ninput:%s\noutput:%s\nreason:%s" % (str(inputlist), str(dictresult), str(reason))
raise RESTCommunicationException(msg)
splitting = getColumn(dictresult, 'tm_split_algo')
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids, splitting != 'Automatic')
self.setDestination()
self.logger.info("Setting the destination to %s " % self.dest)
failed, success = self.retrieveShortLogs(webdir, self.proxyfilename)
if failed:
msg = "%sError%s: Failed to retrieve the following files: %s" % (colors.RED, colors.NORMAL, failed)
self.logger.info(msg)
else:
self.logger.info("%sSuccess%s: All files successfully retrieved." % (colors.GREEN, colors.NORMAL))
returndict = {'success': success, 'failed': failed}
else:
# Different from the old getlog code: set 'logs2' as subresource so that 'getcommand' uses the new logic.
returndict = getcommand.__call__(self, subresource='logs2')
if ('success' in returndict and not returndict['success']) or \
('failed' in returndict and returndict['failed']):
msg = "You can use the --short option to retrieve a short version of the log files from the Grid scheduler."
self.logger.info(msg)
return returndict
def validateOptions(self):
SubCommand.validateOptions(self)
#check the format of jobids
self.jobids = ''
if getattr(self.options, 'jobids', None):
self.jobids = validateJobids(self.options.jobids)
def validateOptions(self):
"""
Check if the sitelist parameter is a comma separater list of cms sitenames,
and put the strings to be passed to the server to self
"""
SubCommand.validateOptions(self)
## Check the format of the jobids option.
if getattr(self.options, 'jobids'):
jobidstuple = validateJobids(self.options.jobids)
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
#Checking if the sites provided by the user are valid cmsnames. Doing this because with only the
#server error handling we get:
# Server answered with: Invalid input parameter
# Reason is: Incorrect 'siteblacklist' parameter
#which is not really user friendly.
#Moreover, I prefer to be independent from Lexicon. I'll the regex here.
sn_re = "^T[1-3]_[A-Z]{2}(_[A-Za-z0-9]+)+$" #sn_re => SiteName_RegularExpression
sn_rec = re.compile(sn_re) #sn_rec => SiteName_RegularExpressionCompiled
for sitelist in ['sitewhitelist', 'siteblacklist']:
if getattr(self.options, sitelist) is not None:
for i, site_name in enumerate(getattr(self.options, sitelist).split(',')):
if not sn_rec.match(site_name):
msg = "The site name %s does not look like a valid CMS site name" % (site_name)
msg += " (it is not matching the regular expression %s)." % (sn_re)
raise ConfigurationException(msg)
setattr(self, sitelist, getattr(self.options, sitelist).split(','))
## Sanity checks for task sizes. Limits are purposely fairly generous to provide
## some level of future-proofing. The server may restrict further.
if self.options.maxjobruntime is not None:
if self.options.maxjobruntime < 60 or self.options.maxjobruntime > 336*60:
msg = "The requested maximum job runtime (%d minutes) must be between 60 and 20160 minutes." % (self.options.maxjobruntime)
raise ConfigurationException(msg)
self.maxjobruntime = str(self.options.maxjobruntime)
if self.options.maxmemory is not None:
if self.options.maxmemory < 30 or self.options.maxmemory > 1024*30:
msg = "The requested per-job memory (%d MB) must be between 30 and 30720 MB." % (self.options.maxmemory)
raise ConfigurationException(msg)
self.maxmemory = str(self.options.maxmemory)
if self.options.numcores is not None:
if self.options.numcores < 1 or self.options.numcores > 128:
msg = "The requested number of cores (%d) must be between 1 and 128." % (self.options.numcores)
raise ConfigurationException(msg)
self.numcores = str(self.options.numcores)
if self.options.priority is not None:
self.priority = str(self.options.priority)
def __call__(self):
if self.options.short:
#Check if splitting is automatic
try:
splitting=self.cachedinfo['OriginalConfig'].Data.splitting
except AttributeError: #Default setting is 'Automatic'
splitting='Automatic'
except KeyError: #crab remade task does not have 'OriginalConfig' key, need to fetch from DB
splitting='Unknown'
taskname = self.cachedinfo['RequestName']
inputlist = {'subresource': 'webdir', 'workflow': taskname}
serverFactory = CRABClient.Emulator.getEmulator('rest')
server = serverFactory(self.serverurl, self.proxyfilename, self.proxyfilename, version=__version__)
uri = self.getUrl(self.instance, resource = 'task')
webdir=None
if splitting!='Unknown':
webdir = getProxiedWebDir(taskname, self.serverurl, uri, self.proxyfilename, self.logger.debug)
if not webdir:
dictresult, status, reason = server.get(uri, data = inputlist)
if status != 200:
msg = "Problem retrieving information from the server:\ninput:%s\noutput:%s\nreason:%s" % (str(inputlist), str(dictresult), str(reason))
raise RESTCommunicationException(msg)
if splitting=='Unknown':
splitting=getColumn(dictresult,'tm_split_algo')
webdir = dictresult['result'][0]
self.logger.info('Server result: %s' % webdir)
self.setDestination()
self.logger.info("Setting the destination to %s " % self.dest)
#check the format of jobids
self.options.jobids = validateJobids(self.options.jobids,splitting!='Automatic')
failed, success = self.retrieveShortLogs(webdir, self.proxyfilename)
if failed:
msg = "%sError%s: Failed to retrieve the following files: %s" % (colors.RED,colors.NORMAL,failed)
self.logger.info(msg)
else:
self.logger.info("%sSuccess%s: All files successfully retrieved." % (colors.GREEN,colors.NORMAL))
returndict = {'success': success, 'failed': failed}
else:
# Different from the old getlog code: set 'logs2' as subresource so that 'getcommand' uses the new logic.
returndict = getcommand.__call__(self, subresource = 'logs2')
if ('success' in returndict and not returndict['success']) or \
('failed' in returndict and returndict['failed']):
msg = "You can use the --short option to retrieve a short version of the log files from the Grid scheduler."
self.logger.info(msg)
return returndict
def validateOptions(self):
#Figuring out the destination directory
SubCommand.validateOptions(self)
self.dest = None
if self.options.outputpath is not None:
if re.match("^[a-z]+://", self.options.outputpath):
self.dest = self.options.outputpath
elif not os.path.isabs( self.options.outputpath ):
self.dest = os.path.abspath( self.options.outputpath )
else:
self.dest = self.options.outputpath
#convert all to -1
if getattr(self.options, 'quantity', None) == 'all':
self.options.quantity = -1
#check the format of jobids
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids)
def validateOptions(self):
#Figuring out the destination directory
SubCommand.validateOptions(self)
self.dest = None
if self.options.outputpath is not None:
if re.match("^[a-z]+://", self.options.outputpath):
self.dest = self.options.outputpath
elif not os.path.isabs(self.options.outputpath):
self.dest = os.path.abspath(self.options.outputpath)
else:
self.dest = self.options.outputpath
#convert all to -1
if getattr(self.options, 'quantity', None) == 'all':
self.options.quantity = -1
#check the format of jobids
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids)
self.dump = self.options.dump
def validateOptions(self):
SubCommand.validateOptions(self)
if self.options.sort is not None:
sortOpts = [
"state", "site", "runtime", "memory", "cpu", "retries",
"waste", "exitcode"
]
if self.options.sort not in sortOpts:
msg = "%sError%s:" % (colors.RED, colors.NORMAL)
msg += " Only the following values are accepted for --sort option: %s" % (
sortOpts)
raise ConfigurationException(msg)
if self.options.jobids:
jobidstuple = validateJobids(self.options.jobids)
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
if self.options.jobids and not (self.options.long
or self.options.sort):
raise ConfigurationException(
"Parameter --jobids can only be used in combination "
"with --long or --sort options.")
def validateOptions(self):
#Figuring out the destination directory
SubCommand.validateOptions(self)
self.dest = None
if self.options.outputpath is not None:
if re.match("^[a-z]+://", self.options.outputpath):
self.dest = self.options.outputpath
elif not os.path.isabs(self.options.outputpath):
self.dest = os.path.abspath(self.options.outputpath)
else:
self.dest = self.options.outputpath
#convert all to -1
if getattr(self.options, 'quantity', None) == 'all':
self.options.quantity = -1
#check the format of jobids
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids)
if hasattr(self.options, 'command') and self.options.command != None:
AvailableCommands = ['LCG', 'GFAL']
self.command = self.options.command.upper()
if self.command not in AvailableCommands:
msg = "You specified to use %s command and it is not allowed. Available commands are: %s " % (
self.command, str(AvailableCommands))
ex = ConfigurationException(msg)
raise ex
else:
self.command = None
if hasattr(self.options, 'checksum'):
if re.match('^yes$|^no$', self.options.checksum):
self.checksum = 'ADLER32' if self.options.checksum == 'yes' else None
else:
msg = "You specified to use %s checksum. Only lowercase yes/no is accepted to turn ADLER32 checksum" % self.options.checksum
ex = ConfigurationException(msg)
raise ex
def validateOptions(self):
#Figuring out the destination directory
SubCommand.validateOptions(self)
self.dest = None
if self.options.outputpath is not None:
if re.match("^[a-z]+://", self.options.outputpath):
self.dest = self.options.outputpath
elif not os.path.isabs( self.options.outputpath ):
self.dest = os.path.abspath( self.options.outputpath )
else:
self.dest = self.options.outputpath
#convert all to -1
if getattr(self.options, 'quantity', None) == 'all':
self.options.quantity = -1
#check the format of jobids
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids)
if hasattr(self.options, 'command') and self.options.command != None:
AvailableCommands = ['LCG', 'GFAL']
self.command = self.options.command.upper()
if self.command not in AvailableCommands:
msg = "You specified to use %s command and it is not allowed. Available commands are: %s " % (self.command, str(AvailableCommands))
ex = ConfigurationException(msg)
raise ex
else:
self.command = None
if hasattr(self.options, 'checksum'):
if re.match('^yes$|^no$', self.options.checksum):
self.checksum = 'ADLER32' if self.options.checksum == 'yes' else None
else:
msg = "You specified to use %s checksum. Only lowercase yes/no is accepted to turn ADLER32 checksum" % self.options.checksum
ex = ConfigurationException(msg)
raise ex
def validateOptions(self):
"""
Check if the sitelist parameter is a comma separater list of cms sitenames,
and put the strings to be passed to the server to self
"""
SubCommand.validateOptions(self)
if self.options.publication:
if self.options.sitewhitelist is not None or self.options.siteblacklist is not None or \
self.options.maxjobruntime is not None or self.options.maxmemory is not None or \
self.options.numcores is not None or self.options.priority is not None:
msg = "The options --sitewhitelist, --siteblacklist,"
msg += " --maxjobruntime, --maxmemory, --numcores and --priority"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit (failed) publications,"
msg += " in which case all of the first options make no sense."
raise ConfigurationException(msg)
if self.options.jobids:
msg = "The option --jobids"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit (failed) publications,"
msg += " which does not allow yet filtering on job ids (ALL failed publications will be resubmitted)."
raise ConfigurationException(msg)
if self.options.force:
msg = "The option --force"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit failed publications."
msg += " Publications in a status other than 'failed' can not be resubmitted."
raise ConfigurationException(msg)
## The --jobids option indicates which jobs have to be resubmitted. If it is not
## given, then all jobs in the task that are not running or successfully
## completed are resubmitted. If the user provides a list of job ids, then also
## successfully completed jobs can be resubmitted.
## Check the format of the jobids option.
if self.options.jobids:
jobidstuple = validateJobids(self.options.jobids)
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
## The --force option should not be accepted unless combined with a user-given
## list of job ids via --jobids.
if self.options.force and not self.jobids:
msg = "Option --force can only be used in combination with option --jobids."
raise ConfigurationException(msg)
## Covention used for the job parameters that the user can set when doing job
## resubmission (i.e. siteblacklist, sitewhitelist, maxjobruntime, maxmemory,
## numcores and priority):
## - If the user doesn't set a parameter we don't pass it to the server and the
## the server copies the original value the parameter had at task submission.
## It copies it from the Task DB. Therefore we need to keep these parameters
## in separate columns of the Task DB containing their original values.
## - For the site black- and whitelists, if the user passes an empty string,
## e.g. --siteblacklist='', we pass to the server siteblacklist=empty and the
## server interprets this as and empty list ([]). If the user passes a given
## list of sites, this new list overwrittes the original one.
## - The values of the parameters are used only for the resubmitted jobs (for
## their first resubmission and all next automatic resubmissions).
#Checking if the sites provided by the user are valid cmsnames. Doing this because with only the
#server error handling we get:
# Server answered with: Invalid input parameter
# Reason is: Incorrect 'siteblacklist' parameter
#which is not really user friendly.
#Moreover, I prefer to be independent from Lexicon. I'll the regex here.
sn_re = "^T[1-3]_[A-Z]{2}(_[A-Za-z0-9]+)+$" #sn_re => SiteName_RegularExpression
sn_rec = re.compile(sn_re) #sn_rec => SiteName_RegularExpressionCompiled
for sitelist in ['sitewhitelist', 'siteblacklist']:
if getattr(self.options, sitelist) is not None:
if getattr(self.options, sitelist) != "":
for site_name in getattr(self.options, sitelist).split(','):
if '*' not in site_name and not sn_rec.match(site_name):
msg = "The site name '%s' does not look like a valid CMS site name" % (site_name)
msg += " (it is not matching the regular expression '%s')." % (sn_re)
raise ConfigurationException(msg)
setattr(self, sitelist, getattr(self.options, sitelist).split(','))
else:
setattr(self, sitelist, [])
## Sanity checks for task sizes. Limits are purposely fairly generous to provide
## some level of future-proofing. The server may restrict further.
if self.options.maxjobruntime is not None:
if self.options.maxjobruntime < 60 or self.options.maxjobruntime > 336*60:
msg = "The requested maximum job runtime (%d minutes) must be between 60 and 20160 minutes." % (self.options.maxjobruntime)
raise ConfigurationException(msg)
if self.options.maxmemory is not None:
if self.options.maxmemory < 30 or self.options.maxmemory > 1024*30:
msg = "The requested per-job memory (%d MB) must be between 30 and 30720 MB." % (self.options.maxmemory)
raise ConfigurationException(msg)
if self.options.numcores is not None:
if self.options.numcores < 1 or self.options.numcores > 128:
msg = "The requested number of cores (%d) must be between 1 and 128." % (self.options.numcores)
raise ConfigurationException(msg)
if self.options.priority is not None:
if self.options.priority < 1:
msg = "The requested priority (%d) must be greater than 0." % (self.options.priority)
raise ConfigurationException(msg)
def __call__(self, **argv): # pylint: disable=arguments-differ
## Retrieve the transferLogs parameter from the task database.
taskdbparam, configparam = '', ''
if argv.get('subresource') in ['logs', 'logs2']:
taskdbparam = 'tm_save_logs'
configparam = "General.transferLogs"
elif argv.get('subresource') in ['data', 'data2']:
taskdbparam = 'tm_transfer_outputs'
configparam = "General.transferOutputs"
transferFlag = 'unknown'
inputlist = {'subresource': 'search', 'workflow': self.cachedinfo['RequestName']}
serverFactory = CRABClient.Emulator.getEmulator('rest')
server = serverFactory(self.serverurl, self.proxyfilename, self.proxyfilename, version=__version__)
uri = getUrl(self.instance, resource = 'task')
dictresult, status, _ = server.get(uri, data = inputlist)
self.logger.debug('Server result: %s' % dictresult)
splitting = None
if status == 200:
if 'desc' in dictresult and 'columns' in dictresult['desc']:
position = dictresult['desc']['columns'].index(taskdbparam)
transferFlag = dictresult['result'][position] #= 'T' or 'F'
position = dictresult['desc']['columns'].index('tm_split_algo')
splitting = dictresult['result'][position]
else:
self.logger.debug("Unable to locate %s in server result." % (taskdbparam))
## If transferFlag = False, there is nothing to retrieve.
if transferFlag == 'F':
msg = "No files to retrieve. Files not transferred to storage since task configuration parameter %s is False." % (configparam)
self.logger.info(msg)
return {'success': {}, 'failed': {}}
## Retrieve tm_edm_outfiles, tm_tfile_outfiles and tm_outfiles from the task database and check if they are empty.
if argv.get('subresource') in ['data', 'data2'] and status == 200:
if 'desc' in dictresult and 'columns' in dictresult['desc']:
position = dictresult['desc']['columns'].index('tm_edm_outfiles')
tm_edm_outfiles = dictresult['result'][position]
position = dictresult['desc']['columns'].index('tm_tfile_outfiles')
tm_tfile_outfiles = dictresult['result'][position]
position = dictresult['desc']['columns'].index('tm_outfiles')
tm_outfiles = dictresult['result'][position]
if tm_edm_outfiles == '[]' and tm_tfile_outfiles == '[]' and tm_outfiles == '[]':
msg = "%sWarning%s:" % (colors.RED, colors.NORMAL)
msg += " There are no output files to retrieve, because CRAB could not detect any in the CMSSW configuration"
msg += " nor was any explicitly specified in the CRAB configuration."
self.logger.warning(msg)
#check the format of jobids
if getattr(self.options, 'jobids', None):
self.options.jobids = validateJobids(self.options.jobids, splitting != 'Automatic')
self.processAndStoreJobIds()
#Retrieving output files location from the server
self.logger.debug('Retrieving locations for task %s' % self.cachedinfo['RequestName'])
inputlist = [('workflow', self.cachedinfo['RequestName'])]
inputlist.extend(list(argv.iteritems()))
if getattr(self.options, 'quantity', None):
self.logger.debug('Retrieving %s file locations' % self.options.quantity)
inputlist.append(('limit', self.options.quantity))
else:
self.logger.debug('Retrieving all file locations')
inputlist.append(('limit', -1))
if getattr(self.options, 'jobids', None):
self.logger.debug('Retrieving jobs %s' % self.options.jobids)
inputlist.extend(self.options.jobids)
serverFactory = CRABClient.Emulator.getEmulator('rest')
server = serverFactory(self.serverurl, self.proxyfilename, self.proxyfilename, version=__version__)
dictresult, status, reason = server.get(self.uri, data = urllib.urlencode(inputlist))
self.logger.debug('Server result: %s' % dictresult)
if status != 200:
msg = "Problem retrieving information from the server:\ninput:%s\noutput:%s\nreason:%s" % (str(inputlist), str(dictresult), str(reason))
raise RESTCommunicationException(msg)
totalfiles = len(dictresult['result'])
fileInfoList = dictresult['result']
self.insertXrootPfns(fileInfoList)
if len(fileInfoList) > 0:
if self.options.dump or self.options.xroot:
self.logger.debug("Getting url info")
else:
self.setDestination()
self.logger.info("Setting the destination to %s " % self.dest)
if self.options.xroot:
self.logger.debug("XRootD urls are requested")
xrootlfn = ["root://cms-xrd-global.cern.ch/%s" % link['lfn'] for link in fileInfoList]
self.logger.info("\n".join(xrootlfn))
returndict = {'xrootd': xrootlfn}
elif self.options.dump:
jobid_pfn_lfn_list = sorted(map(lambda x: (x['jobid'], x['pfn'], x['lfn']), fileInfoList)) # pylint: disable=deprecated-lambda
lastjobid = -1
filecounter = 1
msg = ""
for jobid, pfn, lfn in jobid_pfn_lfn_list:
if jobid != lastjobid:
msg += "%s=== Files from job %s:" % ('\n' if lastjobid != -1 else '', jobid)
lastjobid = jobid
filecounter = 1
msg += "\n%d) PFN: %s" % (filecounter, pfn)
msg += "\n%s LFN: %s" % (' '*(len(str(filecounter))), lfn)
filecounter += 1
self.logger.info(msg)
returndict = {'pfn': [pfn for _, pfn, _ in jobid_pfn_lfn_list], 'lfn': [lfn for _, _, lfn in jobid_pfn_lfn_list]}
else:
self.logger.info("Retrieving %s files" % (totalfiles))
arglist = ['--destination', self.dest, '--input', fileInfoList, '--dir', self.options.projdir, \
'--proxy', self.proxyfilename, '--parallel', self.options.nparallel, '--wait', self.options.waittime, \
'--checksum', self.checksum, '--command', self.command]
copyoutput = remote_copy(self.logger, arglist)
successdict, faileddict = copyoutput()
#need to use deepcopy because successdict and faileddict are dict that is under the a manage dict, accessed multithreadly
returndict = {'success': copy.deepcopy(successdict) , 'failed': copy.deepcopy(faileddict)}
if totalfiles == 0:
self.logger.info("No files to retrieve.")
returndict = {'success': {} , 'failed': {}}
if transferFlag == 'unknown':
if ('success' in returndict and not returndict['success']) and \
('failed' in returndict and not returndict['failed']):
msg = "This is normal behavior if %s = False in the task configuration." % (configparam)
self.logger.info(msg)
return returndict
def validateOptions(self):
"""
Check if the sitelist parameter is a comma separater list of cms sitenames,
and put the strings to be passed to the server to self
"""
SubCommand.validateOptions(self)
serverFactory = CRABClient.Emulator.getEmulator('rest')
self.server = serverFactory(self.serverurl,
self.proxyfilename,
self.proxyfilename,
version=__version__)
uri = getUrl(self.instance, resource='task')
crabDBInfo, _, _ = self.server.get(uri,
data={
'subresource':
'search',
'workflow':
self.cachedinfo['RequestName']
})
self.splitting = getColumn(crabDBInfo, 'tm_split_algo')
if self.options.publication:
if self.options.sitewhitelist is not None or self.options.siteblacklist is not None or \
self.options.maxjobruntime is not None or self.options.maxmemory is not None or \
self.options.numcores is not None or self.options.priority is not None:
msg = "The options --sitewhitelist, --siteblacklist,"
msg += " --maxjobruntime, --maxmemory, --numcores and --priority"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit (failed) publications,"
msg += " in which case all of the first options make no sense."
raise ConfigurationException(msg)
if self.options.jobids:
msg = "The option --jobids"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit (failed) publications,"
msg += " which does not allow yet filtering on job ids (ALL failed publications will be resubmitted)."
raise ConfigurationException(msg)
if self.options.force:
msg = "The option --force"
msg += " can not be specified together with the option --publication."
msg += " The last option is to only resubmit failed publications."
msg += " Publications in a status other than 'failed' can not be resubmitted."
raise ConfigurationException(msg)
## The --jobids option indicates which jobs have to be resubmitted. If it is not
## given, then all jobs in the task that are not running or successfully
## completed are resubmitted. If the user provides a list of job ids, then also
## successfully completed jobs can be resubmitted.
## Check the format of the jobids option.
if self.options.jobids:
jobidstuple = validateJobids(self.options.jobids,
self.splitting != 'Automatic')
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
## The --force option should not be accepted unless combined with a user-given
## list of job ids via --jobids.
if self.options.force and not self.jobids:
msg = "Option --force can only be used in combination with option --jobids."
raise ConfigurationException(msg)
## Covention used for the job parameters that the user can set when doing job
## resubmission (i.e. siteblacklist, sitewhitelist, maxjobruntime, maxmemory,
## numcores and priority):
## - If the user doesn't set a parameter we don't pass it to the server and the
## the server copies the original value the parameter had at task submission.
## It copies it from the Task DB. Therefore we need to keep these parameters
## in separate columns of the Task DB containing their original values.
## - For the site black- and whitelists, if the user passes an empty string,
## e.g. --siteblacklist='', we pass to the server siteblacklist=empty and the
## server interprets this as and empty list ([]). If the user passes a given
## list of sites, this new list overwrittes the original one.
## - The values of the parameters are used only for the resubmitted jobs (for
## their first resubmission and all next automatic resubmissions).
#Checking if the sites provided by the user are valid cmsnames. Doing this because with only the
#server error handling we get:
# Server answered with: Invalid input parameter
# Reason is: Incorrect 'siteblacklist' parameter
#which is not really user friendly.
#Moreover, I prefer to be independent from Lexicon. I'll the regex here.
sn_re = "^T[1-3]_[A-Z]{2}(_[A-Za-z0-9]+)+$" #sn_re => SiteName_RegularExpression
sn_rec = re.compile(
sn_re) #sn_rec => SiteName_RegularExpressionCompiled
for sitelist in ['sitewhitelist', 'siteblacklist']:
if getattr(self.options, sitelist) is not None:
if getattr(self.options, sitelist) != "":
for site_name in getattr(self.options,
sitelist).split(','):
if '*' not in site_name and not sn_rec.match(
site_name):
msg = "The site name '%s' does not look like a valid CMS site name" % (
site_name)
msg += " (it is not matching the regular expression '%s')." % (
sn_re)
raise ConfigurationException(msg)
setattr(self, sitelist,
getattr(self.options, sitelist).split(','))
else:
setattr(self, sitelist, [])
## Sanity checks for task sizes. Limits are purposely fairly generous to provide
## some level of future-proofing. The server may restrict further.
if self.options.maxjobruntime is not None:
if self.options.maxjobruntime < 60 or self.options.maxjobruntime > 336 * 60:
msg = "The requested maximum job runtime (%d minutes) must be between 60 and 20160 minutes." % (
self.options.maxjobruntime)
raise ConfigurationException(msg)
if self.options.maxmemory is not None:
if self.options.maxmemory < 30 or self.options.maxmemory > 1024 * 30:
msg = "The requested per-job memory (%d MB) must be between 30 and 30720 MB." % (
self.options.maxmemory)
raise ConfigurationException(msg)
if self.options.numcores is not None:
if self.options.numcores < 1 or self.options.numcores > 128:
msg = "The requested number of cores (%d) must be between 1 and 128." % (
self.options.numcores)
raise ConfigurationException(msg)
if self.options.priority is not None:
if self.options.priority < 1:
msg = "The requested priority (%d) must be greater than 0." % (
self.options.priority)
raise ConfigurationException(msg)
def processJobIds(self, jobList):
"""
If this is a publication resubmission, return None since jobIds are not taken
into account for publication resubmissions.
If the user provides a list of jobIds to be resubmitted, validate it and
return the same list in case of success.
If no jobIds are provided, create a list of jobs that need resubmitting and
return it.
"""
if self.options.publication:
return None
automatic = any('-' in n for _, n in jobList)
def consider(jobId):
if automatic and (jobId.startswith('0-') or '-' not in jobId):
return False
return True
# Build a dictionary from the jobList
jobStatusDict = {}
for jobStatus, jobId in jobList:
if consider(jobId):
jobStatusDict[jobId] = jobStatus
failedJobStatus = 'failed'
finishedJobStatus = 'finished'
possibleToResubmitJobIds = []
for jobStatus, jobId in jobList:
if ((self.options.force and jobStatus == finishedJobStatus)
or jobStatus == failedJobStatus) and consider(jobId):
possibleToResubmitJobIds.append(jobId)
allowedJobStates = [failedJobStatus]
if self.jobids:
# Automatic splitting does not work with lists... probe- and
# tail-job ids have a '-' in them, so re-split the joblist.
if any(('-' in jobId for _, jobId in jobList)):
jobidstuple = validateJobids(self.options.jobids, False)
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
msg = "Requesting resubmission of jobs %s in task %s" % (
self.jobids, self.cachedinfo['RequestName'])
self.logger.debug(msg)
if self.options.force:
allowedJobStates += [finishedJobStatus]
# Go through the jobids and check if it's possible to resubmit them
for jobId in self.jobids:
if (jobId not in jobStatusDict) or (jobStatusDict[jobId]
not in allowedJobStates):
possibleAndWantedJobIds = list(
set(possibleToResubmitJobIds) & set(self.jobids))
notPossibleAndWantedJobIds = list(
set(self.jobids) - set(possibleAndWantedJobIds))
msg = "Not possible to resubmit the following jobs:\n%s\n" % notPossibleAndWantedJobIds
msg += "Only jobs in status %s can be resubmitted. " % failedJobStatus
msg += "Jobs in status %s can also be resubmitted, " % finishedJobStatus
msg += "but only if the jobid is specified and the force option is set."
raise ConfigurationException(msg)
return self.jobids
else:
msg = "Requesting resubmission of failed jobs in task %s" % (
self.cachedinfo['RequestName'])
self.logger.debug(msg)
if not possibleToResubmitJobIds:
msg = "Found no jobs to resubmit. Only jobs in status %s can be resubmitted. " % failedJobStatus
msg += "Jobs in status %s can also be resubmitted, but only if the jobids " % finishedJobStatus
msg += "are specified and the force option is set."
raise ConfigurationException(msg)
return possibleToResubmitJobIds
def processJobIds(self, jobList):
"""
If this is a publication resubmission, return None since jobIds are not taken
into account for publication resubmissions.
If the user provides a list of jobIds to be resubmitted, validate it and
return the same list in case of success.
If no jobIds are provided, create a list of jobs that need resubmitting and
return it.
"""
if self.options.publication:
return None
automatic = any('-' in n for _, n in jobList)
def consider(jobId):
if automatic and (jobId.startswith('0-') or '-' not in jobId):
return False
return True
# Build a dictionary from the jobList
jobStatusDict = {}
for jobStatus, jobId in jobList:
if consider(jobId):
jobStatusDict[jobId] = jobStatus
failedJobStatus = 'failed'
finishedJobStatus = 'finished'
possibleToResubmitJobIds = []
for jobStatus, jobId in jobList:
if ((self.options.force and jobStatus == finishedJobStatus) or jobStatus == failedJobStatus) and consider(jobId):
possibleToResubmitJobIds.append(jobId)
allowedJobStates = [failedJobStatus]
if self.jobids:
# Automatic splitting does not work with lists... probe- and
# tail-job ids have a '-' in them, so re-split the joblist.
if any(('-' in jobId for _, jobId in jobList)):
jobidstuple = validateJobids(self.options.jobids, False)
self.jobids = [str(jobid) for (_, jobid) in jobidstuple]
msg = "Requesting resubmission of jobs %s in task %s" % (self.jobids, self.cachedinfo['RequestName'])
self.logger.debug(msg)
if self.options.force:
allowedJobStates += [finishedJobStatus]
# Go through the jobids and check if it's possible to resubmit them
for jobId in self.jobids:
if (jobId not in jobStatusDict) or (jobStatusDict[jobId] not in allowedJobStates):
possibleAndWantedJobIds = list(set(possibleToResubmitJobIds) & set(self.jobids))
notPossibleAndWantedJobIds = list(set(self.jobids) - set(possibleAndWantedJobIds))
msg = "Not possible to resubmit the following jobs:\n%s\n" % notPossibleAndWantedJobIds
msg += "Only jobs in status %s can be resubmitted. " % failedJobStatus
msg += "Jobs in status %s can also be resubmitted, " % finishedJobStatus
msg += "but only if the jobid is specified and the force option is set."
raise ConfigurationException(msg)
return self.jobids
else:
msg = "Requesting resubmission of failed jobs in task %s" % (self.cachedinfo['RequestName'])
self.logger.debug(msg)
if not possibleToResubmitJobIds:
msg = "Found no jobs to resubmit. Only jobs in status %s can be resubmitted. " % failedJobStatus
msg += "Jobs in status %s can also be resubmitted, but only if the jobids " % finishedJobStatus
msg += "are specified and the force option is set."
raise ConfigurationException(msg)
return possibleToResubmitJobIds
