cmsHarvester Namespace Reference

Classes
class	CMSHarvester
	CMSHarvester class. More...
class	CMSHarvesterHelpFormatter
	Helper class: CMSHarvesterHelpFormatter. More...
class	DBSXMLHandler
	Helper class: DBSXMLHandler. More...
class	Error
	Helper class: Error exception. More...
class	Usage
	Helper class: Usage exception. More...

Functions
def	build_dataset_ignore_list (self)
def	build_dataset_list (self, input_method, input_name)
def	build_dataset_use_list (self)
def	build_datasets_information (self)
def	build_runs_ignore_list (self)
def	build_runs_list (self, input_method, input_name)
def	build_runs_use_list (self)
def	check_cmssw (self)
def	check_dataset_list (self)
def	check_dbs (self)
def	check_globaltag (self, globaltag=None)
def	check_globaltag_contains_ref_hist_key (self, globaltag, connect_name)
def	check_globaltag_exists (self, globaltag, connect_name)
def	check_input_status (self)
def	check_ref_hist_mappings (self)
def	check_ref_hist_tag (self, tag_name)
def	create_and_check_castor_dir (self, castor_dir)
def	create_and_check_castor_dirs (self)
def	create_castor_path_name_common (self, dataset_name)
def	create_castor_path_name_special (self, dataset_name, run_number, castor_path_common)
def	create_config_file_name (self, dataset_name, run_number)
def	create_crab_config (self)
def	create_harvesting_config (self, dataset_name)
def	create_harvesting_config_file_name (self, dataset_name)
def	create_harvesting_output_file_name (self, dataset_name, run_number)
def	create_me_extraction_config (self, dataset_name)
def	create_me_summary_config_file_name (self, dataset_name)
def	create_me_summary_output_file_name (self, dataset_name)
def	create_multicrab_block_name (self, dataset_name, run_number, index)
def	create_multicrab_config (self)
def	create_output_file_name (self, dataset_name, run_number=None)
def	dbs_check_dataset_spread (self, dataset_name)
def	dbs_resolve_cmssw_version (self, dataset_name)
def	dbs_resolve_dataset_name (self, dataset_name)
def	dbs_resolve_datatype (self, dataset_name)
def	dbs_resolve_globaltag (self, dataset_name)
def	dbs_resolve_number_of_events (self, dataset_name, run_number=None)
def	dbs_resolve_runs (self, dataset_name)
def	escape_dataset_name (self, dataset_name)
def	load_ref_hist_mappings (self)
def	option_handler_caf_access (self, option, opt_str, value, parser)
def	option_handler_castor_dir (self, option, opt_str, value, parser)
def	option_handler_crab_submission (self, option, opt_str, value, parser)
def	option_handler_list_types (self, option, opt_str, value, parser)
def	option_handler_no_t1access (self, option, opt_str, value, parser)
def	option_handler_preferred_site (self, option, opt_str, value, parser)
def	option_handler_saveByLumiSection (self, option, opt_str, value, parser)
def	option_handler_sites (self, option, opt_str, value, parser)
def	parse_cmd_line_options (self)
def	pick_a_site (self, sites, cmssw_version)
def	process_dataset_ignore_list (self)
def	process_runs_use_and_ignore_lists (self)
def	ref_hist_mappings_needed (self, dataset_name=None)
def	run (self)
def	setup_dbs (self)
def	setup_harvesting_info (self)
def	show_exit_message (self)
def	singlify_datasets (self)
def	write_crab_config (self)
def	write_harvesting_config (self, dataset_name)
def	write_me_extraction_config (self, dataset_name)
def	write_multicrab_config (self)

Variables
	__author__
	__version__
	all_file_names
	all_sites_found
	CAF More...
	caf_access
	castor_base_dir
	castor_path_checks_cache
	self.logger.debug("Path is now `%s'" % \ path) More...
	castor_path_common
	DEBUG DEBUG DEBUG This is probably only useful to make sure we don't muck things up, right? Figure out across how many sites this sample has been spread. More...
	castor_paths
	cmd
	cmd_line_opts
	cmssw_version
	complete_sites
	site_names_ref = set(files_info[run_number].values()[0][1]) for site_names_tmp in files_info[run_number].values()[1:]: if set(site_names_tmp[1]) != site_names_ref: mirrored = False break More...
	config_contents
	In case this file is the second step (the real harvesting step) of the two-step harvesting we have to tell it to use our local files. More...
	config_file_name
	Only add the alarming piece to the file name if this is a spread-out dataset. More...
	crab_submission
	dataset_names_after_checks
	dataset_names_after_checks_tmp
	datasets_information
	datasets_to_ignore
	datasets_to_use
	dbs_api
	empty_runs
	exit_code
	file_name
	files_at_site
	files_info
	files_without_sites
	globaltag
	harvesting_info
	harvesting_mode
	harvesting_type
	Jsonfilename
	Jsonlumi
	CRAB More...
	mirrored
	msg
	class Handler(xml.sax.handler.ContentHandler): def startElement(self, name, attrs): if name == "result": site_name = str(attrs["STORAGEELEMENT_SENAME"]) TODO TODO TODO Ugly hack to get around cases like this: $ dbs search –query="find dataset, site, file.count where dataset=/RelValQCD_Pt_3000_3500/CMSSW_3_3_0_pre1-STARTUP31X_V4-v1/GEN-SIM-RECO" Using DBS instance at: http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet Processing ... More...
	nevents
	non_t1access
	nr_max_sites
	num_events_catalog
	num_events_dataset
	num_sites
	if self.datasets_information[dataset_name]["num_events"][run_number] != 0: pdb.set_trace() DEBUG DEBUG DEBUG end More...
	option_parser
	output
	path
	else: Piece not in the list, fine. More...
	permissions
	permissions_new
	permissions_target
	preferred_site
	ref_hist_mappings_file_name
	run_number
	runs_to_ignore
	runs_to_use
	saveByLumiSection
	site_names
	sites_with_complete_copies
	skip_this_path_piece
	self.logger.debug("Checking CASTOR path piece `%s'" % \ piece) More...
	status
	tmp
	TODO TODO TODO Need to think about where this should go, but somewhere we have to move over the fact that we want to process all runs for each dataset that we're considering. More...
	traceback_string
	twiki_url

Function Documentation

build_dataset_ignore_list()

def cmsHarvester.build_dataset_ignore_list

(

self

)

Build a list of datasets to ignore.

NOTE: We should always have a list of datasets to process, but
it may be that we don't have a list of datasets to ignore.

Definition at line 3443 of file cmsHarvester.py.

References info().

    def build_dataset_ignore_list(self):
        """Build a list of datasets to ignore.

        NOTE: We should always have a list of datasets to process, but
        it may be that we don't have a list of datasets to ignore.

        """

        self.logger.info("Building list of datasets to ignore...")

        input_method = self.input_method["datasets"]["ignore"]
        input_name = self.input_name["datasets"]["ignore"]
        dataset_names = self.build_dataset_list(input_method,
                                                input_name)
        self.datasets_to_ignore = dict(list(zip(dataset_names,
                                           [None] * len(dataset_names))))

        self.logger.info("  found %d dataset(s) to ignore:" % \
                         len(dataset_names))
        for dataset in dataset_names:
            self.logger.info("  `%s'" % dataset)

        # End of build_dataset_ignore_list.

build_dataset_list()

def cmsHarvester.build_dataset_list	(		self,
			input_method,
			input_name
	)

Build a list of all datasets to be processed.

Definition at line 3357 of file cmsHarvester.py.

References dbs_resolve_dataset_name(), info(), and print().

    def build_dataset_list(self, input_method, input_name):
        """Build a list of all datasets to be processed.

        """

        dataset_names = []

        # It may be, but only for the list of datasets to ignore, that
        # the input method and name are None because nothing was
        # specified. In that case just an empty list is returned.
        if input_method is None:
            pass
        elif input_method == "dataset":
            # Input comes from a dataset name directly on the command
            # line. But, this can also contain wildcards so we need
            # DBS to translate it conclusively into a list of explicit
            # dataset names.
            self.logger.info("Asking DBS for dataset names")
            dataset_names = self.dbs_resolve_dataset_name(input_name)
        elif input_method == "datasetfile":
            # In this case a file containing a list of dataset names
            # is specified. Still, each line may contain wildcards so
            # this step also needs help from DBS.
            # NOTE: Lines starting with a `#' are ignored.
            self.logger.info("Reading input from list file `%s'" % \
                             input_name)
            try:
                listfile = open("/afs/cern.ch/cms/CAF/CMSCOMM/COMM_DQM/harvesting/bin/%s" %input_name, "r")                print("open listfile")
                for dataset in listfile:
                    # Skip empty lines.
                    dataset_stripped = dataset.strip()
                    if len(dataset_stripped) < 1:
                        continue
                    # Skip lines starting with a `#'.
                    if dataset_stripped[0] != "#":
                        dataset_names.extend(self. \
                                             dbs_resolve_dataset_name(dataset_stripped))
                listfile.close()
            except IOError:
                msg = "ERROR: Could not open input list file `%s'" % \
                      input_name
                self.logger.fatal(msg)
                raise Error(msg)
        else:
            # DEBUG DEBUG DEBUG
            # We should never get here.
            assert False, "Unknown input method `%s'" % input_method
            # DEBUG DEBUG DEBUG end

        # Remove duplicates from the dataset list.
        # NOTE: There should not be any duplicates in any list coming
        # from DBS, but maybe the user provided a list file with less
        # care.
        # Store for later use.
        dataset_names = sorted(set(dataset_names))


        # End of build_dataset_list.
        return dataset_names

build_dataset_use_list()

def cmsHarvester.build_dataset_use_list

(

self

)

Build a list of datasets to process.

Definition at line 3420 of file cmsHarvester.py.

References info(), and reco.zip().

    def build_dataset_use_list(self):
        """Build a list of datasets to process.

        """

        self.logger.info("Building list of datasets to consider...")

        input_method = self.input_method["datasets"]["use"]
        input_name = self.input_name["datasets"]["use"]
        dataset_names = self.build_dataset_list(input_method,
                                                input_name)
        self.datasets_to_use = dict(list(zip(dataset_names,
                                        [None] * len(dataset_names))))

        self.logger.info("  found %d dataset(s) to process:" % \
                         len(dataset_names))
        for dataset in dataset_names:
            self.logger.info("  `%s'" % dataset)

        # End of build_dataset_use_list.

build_datasets_information()

def cmsHarvester.build_datasets_information

(

self

)

Obtain all information on the datasets that we need to run.

Use DBS to figure out all required information on our
datasets, like the run numbers and the GlobalTag. All
information is stored in the datasets_information member
variable.

Definition at line 5285 of file cmsHarvester.py.

    def build_datasets_information(self):
        """Obtain all information on the datasets that we need to run.

        Use DBS to figure out all required information on our
        datasets, like the run numbers and the GlobalTag. All
        information is stored in the datasets_information member
        variable.

        """

        # Get a list of runs in the dataset.
        # NOTE: The harvesting has to be done run-by-run, so we
        # split up datasets based on the run numbers. Strictly
        # speaking this is not (yet?) necessary for Monte Carlo
        # since all those samples use run number 1. Still, this
        # general approach should work for all samples.

        # Now loop over all datasets in the list and process them.
        # NOTE: This processing has been split into several loops
        # to be easier to follow, sacrificing a bit of efficiency.
        self.datasets_information = {}
        self.logger.info("Collecting information for all datasets to process")
        dataset_names = sorted(self.datasets_to_use.keys())
        for dataset_name in dataset_names:

            # Tell the user which dataset: nice with many datasets.
            sep_line = "-" * 30
            self.logger.info(sep_line)
            self.logger.info("  `%s'" % dataset_name)
            self.logger.info(sep_line)

            runs = self.dbs_resolve_runs(dataset_name)
            self.logger.info("    found %d run(s)" % len(runs))
            if len(runs) > 0:
                self.logger.debug("      run number(s): %s" % \
                                  ", ".join([str(i) for i in runs]))
            else:
                # DEBUG DEBUG DEBUG
                # This should never happen after the DBS checks.
                self.logger.warning("  --> skipping dataset "
                                    "without any runs")
                assert False, "Panic: found a dataset without runs " \
                       "after DBS checks!"
                # DEBUG DEBUG DEBUG end

            cmssw_version = self.dbs_resolve_cmssw_version(dataset_name)
            self.logger.info("    found CMSSW version `%s'" % cmssw_version)

            # Figure out if this is data or MC.
            datatype = self.dbs_resolve_datatype(dataset_name)
            self.logger.info("    sample is data or MC? --> %s" % \
                             datatype)

            

            # Try and figure out the GlobalTag to be used.
            if self.globaltag is None:
                globaltag = self.dbs_resolve_globaltag(dataset_name)
            else:
                globaltag = self.globaltag

            self.logger.info("    found GlobalTag `%s'" % globaltag)

            # DEBUG DEBUG DEBUG
            if globaltag == "":
                # Actually we should not even reach this point, after
                # our dataset sanity checks.
                assert datatype == "data", \
                       "ERROR Empty GlobalTag for MC dataset!!!"
            # DEBUG DEBUG DEBUG end

            

            # DEBUG DEBUG DEBUG
            #tmp = self.dbs_check_dataset_spread_old(dataset_name)
            # DEBUG DEBUG DEBUG end
            sites_catalog = self.dbs_check_dataset_spread(dataset_name)

            # Extract the total event counts.
            num_events = {}
            for run_number in sites_catalog.keys():
                num_events[run_number] = sites_catalog \
                                         [run_number]["all_sites"]
                del sites_catalog[run_number]["all_sites"]

            # Extract the information about whether or not datasets
            # are mirrored.
            mirror_catalog = {}
            for run_number in sites_catalog.keys():
                mirror_catalog[run_number] = sites_catalog \
                                             [run_number]["mirrored"]
                del sites_catalog[run_number]["mirrored"]

            # BUG BUG BUG
            # I think I could now get rid of that and just fill the
            # "sites" entry with the `inverse' of this
            # num_events_catalog(?).
            #num_sites = self.dbs_resolve_dataset_number_of_sites(dataset_name)
            #sites_catalog = self.dbs_check_dataset_spread(dataset_name)
            #sites_catalog = dict(zip(num_events_catalog.keys(),
            #                         [[j for i in num_events_catalog.values() for j in i.keys()]]))
            # BUG BUG BUG end

build_runs_ignore_list()

def cmsHarvester.build_runs_ignore_list

(

self

)

Build a list of runs to ignore.

NOTE: We should always have a list of runs to process, but
it may be that we don't have a list of runs to ignore.

Definition at line 3541 of file cmsHarvester.py.

References info().

    def build_runs_ignore_list(self):
        """Build a list of runs to ignore.

        NOTE: We should always have a list of runs to process, but
        it may be that we don't have a list of runs to ignore.

        """

        self.logger.info("Building list of runs to ignore...")

        input_method = self.input_method["runs"]["ignore"]
        input_name = self.input_name["runs"]["ignore"]
        runs = self.build_runs_list(input_method, input_name)
        self.runs_to_ignore = dict(list(zip(runs, [None] * len(runs))))

        self.logger.info("  found %d run(s) to ignore:" % \
                         len(runs))
        if len(runs) > 0:
            self.logger.info("  %s" % ", ".join([str(i) for i in runs]))

        # End of build_runs_ignore_list().

build_runs_list()

def cmsHarvester.build_runs_list	(		self,
			input_method,
			input_name
	)

Definition at line 3469 of file cmsHarvester.py.

References info(), and createfilelist.int.

    def build_runs_list(self, input_method, input_name):

        runs = []

        # A list of runs (either to use or to ignore) is not
        # required. This protects against `empty cases.'
        if input_method is None:
            pass
        elif input_method == "runs":
            # A list of runs was specified directly from the command
            # line.
            self.logger.info("Reading list of runs from the " \
                             "command line")
            runs.extend([int(i.strip()) \
                         for i in input_name.split(",") \
                         if len(i.strip()) > 0])
        elif input_method == "runslistfile":
            # We were passed a file containing a list of runs.
            self.logger.info("Reading list of runs from file `%s'" % \
                             input_name)
            try:
                listfile = open(input_name, "r")                for run in listfile:
                    # Skip empty lines.
                    run_stripped = run.strip()
                    if len(run_stripped) < 1:
                        continue
                    # Skip lines starting with a `#'.
                    if run_stripped[0] != "#":
                        runs.append(int(run_stripped))
                listfile.close()
            except IOError:
                msg = "ERROR: Could not open input list file `%s'" % \
                      input_name
                self.logger.fatal(msg)
                raise Error(msg)

        else:
            # DEBUG DEBUG DEBUG
            # We should never get here.
            assert False, "Unknown input method `%s'" % input_method
            # DEBUG DEBUG DEBUG end

        # Remove duplicates, sort and done.
        runs = list(set(runs))

        # End of build_runs_list().
        return runs

build_runs_use_list()

def cmsHarvester.build_runs_use_list

(

self

)

Build a list of runs to process.

Definition at line 3520 of file cmsHarvester.py.

References info().

    def build_runs_use_list(self):
        """Build a list of runs to process.

        """

        self.logger.info("Building list of runs to consider...")

        input_method = self.input_method["runs"]["use"]
        input_name = self.input_name["runs"]["use"]
        runs = self.build_runs_list(input_method, input_name)
        self.runs_to_use = dict(list(zip(runs, [None] * len(runs))))

        self.logger.info("  found %d run(s) to process:" % \
                         len(runs))
        if len(runs) > 0:
            self.logger.info("  %s" % ", ".join([str(i) for i in runs]))

        # End of build_runs_list().

check_cmssw()

def cmsHarvester.check_cmssw

(

self

)

Check if CMSSW is setup.

Definition at line 2333 of file cmsHarvester.py.

    def check_cmssw(self):
        """Check if CMSSW is setup.

        """

        # Try to access the CMSSW_VERSION environment variable. If
        # it's something useful we consider CMSSW to be set up
        # properly. Otherwise we raise an error.
        cmssw_version = os.getenv("CMSSW_VERSION")
        if cmssw_version is None:
            self.logger.fatal("It seems CMSSW is not setup...")
            self.logger.fatal("($CMSSW_VERSION is empty)")
            raise Error("ERROR: CMSSW needs to be setup first!")

        self.cmssw_version = cmssw_version
        self.logger.info("Found CMSSW version %s properly set up" % \
                          self.cmssw_version)

        # End of check_cmsssw.
        return True

check_dataset_list()

def cmsHarvester.check_dataset_list

(

self

)

Check list of dataset names for impossible ones.

Two kinds of checks are done:
- Checks for things that do not make sense. These lead to
  errors and skipped datasets.
- Sanity checks. For these warnings are issued but the user is
  considered to be the authoritative expert.

Checks performed:
- The CMSSW version encoded in the dataset name should match
  self.cmssw_version. This is critical.
- There should be some events in the dataset/run. This is
  critical in the sense that CRAB refuses to create jobs for
  zero events. And yes, this does happen in practice. E.g. the
  reprocessed CRAFT08 datasets contain runs with zero events.
- A cursory check is performed to see if the harvesting type
  makes sense for the data type. This should prevent the user
  from inadvertently running RelVal for data.
- It is not possible to run single-step harvesting jobs on
  samples that are not fully contained at a single site.
- Each dataset/run has to be available at at least one site.

Definition at line 3794 of file cmsHarvester.py.

References info(), relativeConstraints.keys, and MessageLogger_cfi.warning.

    def check_dataset_list(self):
        """Check list of dataset names for impossible ones.

        Two kinds of checks are done:
        - Checks for things that do not make sense. These lead to
          errors and skipped datasets.
        - Sanity checks. For these warnings are issued but the user is
          considered to be the authoritative expert.

        Checks performed:
        - The CMSSW version encoded in the dataset name should match
          self.cmssw_version. This is critical.
        - There should be some events in the dataset/run. This is
          critical in the sense that CRAB refuses to create jobs for
          zero events. And yes, this does happen in practice. E.g. the
          reprocessed CRAFT08 datasets contain runs with zero events.
        - A cursory check is performed to see if the harvesting type
          makes sense for the data type. This should prevent the user
          from inadvertently running RelVal for data.
        - It is not possible to run single-step harvesting jobs on
          samples that are not fully contained at a single site.
        - Each dataset/run has to be available at at least one site.

        """

        self.logger.info("Performing sanity checks on dataset list...")

        dataset_names_after_checks = copy.deepcopy(self.datasets_to_use)

        for dataset_name in self.datasets_to_use.keys():

            # Check CMSSW version.
            version_from_dataset = self.datasets_information[dataset_name] \
                                   ["cmssw_version"]
            if version_from_dataset != self.cmssw_version:
                msg = "  CMSSW version mismatch for dataset `%s' " \
                      "(%s vs. %s)" % \
                      (dataset_name,
                       self.cmssw_version, version_from_dataset)
                if self.force_running:
                    # Expert mode: just warn, then continue.
                    self.logger.warning("%s " \
                                        "--> `force mode' active: " \
                                        "run anyway" % msg)
                else:
                    del dataset_names_after_checks[dataset_name]
                    self.logger.warning("%s " \
                                        "--> skipping" % msg)
                    continue

            

            # Check that the harvesting type makes sense for the
            # sample. E.g. normally one would not run the DQMOffline
            # harvesting on Monte Carlo.
            # TODO TODO TODO
            # This should be further refined.
            suspicious = False
            datatype = self.datasets_information[dataset_name]["datatype"]
            if datatype == "data":
                # Normally only DQM harvesting is run on data.
                if self.harvesting_type != "DQMOffline":
                    suspicious = True
            elif datatype == "mc":
                if self.harvesting_type == "DQMOffline":
                    suspicious = True
            else:
                # Doh!
                assert False, "ERROR Impossible data type `%s' " \
                       "for dataset `%s'" % \
                       (datatype, dataset_name)
            if suspicious:
                msg = "  Normally one does not run `%s' harvesting " \
                      "on %s samples, are you sure?" % \
                      (self.harvesting_type, datatype)
                if self.force_running:
                    self.logger.warning("%s " \
                                        "--> `force mode' active: " \
                                        "run anyway" % msg)
                else:
                    del dataset_names_after_checks[dataset_name]
                    self.logger.warning("%s " \
                                        "--> skipping" % msg)
                    continue

            # TODO TODO TODO end

            

            # BUG BUG BUG
            # For the moment, due to a problem with DBS, I cannot
            # figure out the GlobalTag for data by myself. (For MC
            # it's no problem.) This means that unless a GlobalTag was
            # specified from the command line, we will have to skip
            # any data datasets.

            if datatype == "data":
                if self.globaltag is None:
                    msg = "For data datasets (like `%s') " \
                          "we need a GlobalTag" % \
                          dataset_name
                    del dataset_names_after_checks[dataset_name]
                    self.logger.warning("%s " \
                                        "--> skipping" % msg)
                    continue

            # BUG BUG BUG end

            

            # Check if the GlobalTag exists and (if we're using
            # reference histograms) if it's ready to be used with
            # reference histograms.
            globaltag = self.datasets_information[dataset_name]["globaltag"]
            if not globaltag in self.globaltag_check_cache:
                if self.check_globaltag(globaltag):
                    self.globaltag_check_cache.append(globaltag)
                else:
                    msg = "Something is wrong with GlobalTag `%s' " \
                          "used by dataset `%s'!" % \
                          (globaltag, dataset_name)
                    if self.use_ref_hists:
                        msg += "\n(Either it does not exist or it " \
                               "does not contain the required key to " \
                               "be used with reference histograms.)"
                    else:
                        msg += "\n(It probably just does not exist.)"
                    self.logger.fatal(msg)
                    raise Usage(msg)

            

            # Require that each run is available at least somewhere.
            runs_without_sites = [i for (i, j) in \
                                  self.datasets_information[dataset_name] \
                                  ["sites"].items() \
                                  if len(j) < 1 and \
                                  i in self.datasets_to_use[dataset_name]]
            if len(runs_without_sites) > 0:
                for run_without_sites in runs_without_sites:
                    try:
                        dataset_names_after_checks[dataset_name].remove(run_without_sites)
                    except KeyError:
                        pass
                self.logger.warning("  removed %d unavailable run(s) " \
                                    "from dataset `%s'" % \
                                    (len(runs_without_sites), dataset_name))
                self.logger.debug("    (%s)" % \
                                  ", ".join([str(i) for i in \
                                             runs_without_sites]))

            

            # Unless we're running two-step harvesting: only allow
            # samples located on a single site.
            if not self.harvesting_mode == "two-step":
                for run_number in self.datasets_to_use[dataset_name]:
                    # DEBUG DEBUG DEBUG

check_dbs()

def cmsHarvester.check_dbs

(

self

)

Check if DBS is setup.

Definition at line 2356 of file cmsHarvester.py.

References debug.

    def check_dbs(self):
        """Check if DBS is setup.

        """

        # Try to access the DBSCMD_HOME environment variable. If this
        # looks useful we consider DBS to be set up
        # properly. Otherwise we raise an error.
        dbs_home = os.getenv("DBSCMD_HOME")
        if dbs_home is None:
            self.logger.fatal("It seems DBS is not setup...")
            self.logger.fatal("  $DBSCMD_HOME is empty")
            raise Error("ERROR: DBS needs to be setup first!")

check_globaltag()

def cmsHarvester.check_globaltag	(		self,
			globaltag = None
	)

Check if globaltag exists.

Check if globaltag exists as GlobalTag in the database given
by self.frontier_connection_name['globaltag']. If globaltag is
None, self.globaltag is used instead.

If we're going to use reference histograms this method also
checks for the existence of the required key in the GlobalTag.

Definition at line 4500 of file cmsHarvester.py.

    def check_globaltag(self, globaltag=None):
        """Check if globaltag exists.

        Check if globaltag exists as GlobalTag in the database given
        by self.frontier_connection_name['globaltag']. If globaltag is
        None, self.globaltag is used instead.

        If we're going to use reference histograms this method also
        checks for the existence of the required key in the GlobalTag.

        """

        if globaltag is None:
            globaltag = self.globaltag

        # All GlobalTags should end in `::All', right?
        if globaltag.endswith("::All"):
            globaltag = globaltag[:-5]

        connect_name = self.frontier_connection_name["globaltag"]
        # BUG BUG BUG
        # There is a bug in cmscond_tagtree_list: some magic is
        # missing from the implementation requiring one to specify
        # explicitly the name of the squid to connect to. Since the
        # cmsHarvester can only be run from the CERN network anyway,
        # cmsfrontier:8000 is hard-coded in here. Not nice but it
        # works.
        connect_name = connect_name.replace("frontier://",
                                            "frontier://cmsfrontier:8000/")
        # BUG BUG BUG end
        connect_name += self.db_account_name_cms_cond_globaltag()

        tag_exists = self.check_globaltag_exists(globaltag, connect_name)

        #----------

        tag_contains_ref_hist_key = False
        if self.use_ref_hists and tag_exists:
            # Check for the key required to use reference histograms.
            tag_contains_ref_hist_key = self.check_globaltag_contains_ref_hist_key(globaltag, connect_name)

        #----------

        if self.use_ref_hists:
            ret_val = tag_exists and tag_contains_ref_hist_key
        else:
            ret_val = tag_exists

        #----------

        # End of check_globaltag.
        return ret_val

check_globaltag_contains_ref_hist_key()

def cmsHarvester.check_globaltag_contains_ref_hist_key	(		self,
			globaltag,
			connect_name
	)

Check if globaltag contains the required RefHistos key.

Definition at line 4597 of file cmsHarvester.py.

References debug, and info().

    def check_globaltag_contains_ref_hist_key(self, globaltag, connect_name):
        """Check if globaltag contains the required RefHistos key.

        """

        # Check for the key required to use reference histograms.
        tag_contains_key = None
        ref_hist_key = "RefHistos"
        self.logger.info("Checking existence of reference " \
                         "histogram key `%s' in GlobalTag `%s'" % \
                         (ref_hist_key, globaltag))
        self.logger.debug("  (Using database connection `%s')" % \
                              connect_name)
        cmd = "cmscond_tagtree_list -c %s -T %s -n %s" % \
              (connect_name, globaltag, ref_hist_key)
        (status, output) = subprocess.getstatusoutput(cmd)
        if status != 0 or \
               output.find("error") > -1:
            msg = "Could not check existence of key `%s'" % \
                  (ref_hist_key, connect_name)
            self.logger.fatal(msg)
            self.logger.debug("Command used:")
            self.logger.debug("  %s" % cmd)
            self.logger.debug("Output received:")
            self.logger.debug("  %s" % output)
            raise Error(msg)
        if len(output) < 1:
            self.logger.debug("Required key for use of reference " \
                              "histograms `%s' does not exist " \
                              "in GlobalTag `%s':" % \
                              (ref_hist_key, globaltag))
            self.logger.debug("Output received:")
            self.logger.debug(output)
            tag_contains_key = False
        else:
            tag_contains_key = True

        self.logger.info("  GlobalTag contains `%s' key? -> %s" % \
                         (ref_hist_key, tag_contains_key))

        # End of check_globaltag_contains_ref_hist_key.
        return tag_contains_key

check_globaltag_exists()

def cmsHarvester.check_globaltag_exists	(		self,
			globaltag,
			connect_name
	)

Check if globaltag exists.

Definition at line 4555 of file cmsHarvester.py.

References debug, info(), and submitPVValidationJobs.split().

    def check_globaltag_exists(self, globaltag, connect_name):
        """Check if globaltag exists.

        """

        self.logger.info("Checking existence of GlobalTag `%s'" % \
                         globaltag)
        self.logger.debug("  (Using database connection `%s')" % \
                          connect_name)

        cmd = "cmscond_tagtree_list -c %s -T %s" % \
              (connect_name, globaltag)
        (status, output) = subprocess.getstatusoutput(cmd)
        if status != 0 or \
               output.find("error") > -1:
            msg = "Could not check existence of GlobalTag `%s' in `%s'" % \
                  (globaltag, connect_name)
            if output.find(".ALL_TABLES not found") > -1:
                msg = "%s\n" \
                      "Missing database account `%s'" % \
                      (msg, output.split(".ALL_TABLES")[0].split()[-1])
            self.logger.fatal(msg)
            self.logger.debug("Command used:")
            self.logger.debug("  %s" % cmd)
            self.logger.debug("Output received:")
            self.logger.debug(output)
            raise Error(msg)
        if output.find("does not exist") > -1:
            self.logger.debug("GlobalTag `%s' does not exist in `%s':" % \
                              (globaltag, connect_name))
            self.logger.debug("Output received:")
            self.logger.debug(output)
            tag_exists = False
        else:
            tag_exists = True
        self.logger.info("  GlobalTag exists? -> %s" % tag_exists)

        # End of check_globaltag_exists.
        return tag_exists

check_input_status()

def cmsHarvester.check_input_status

(

self

)

Check completeness and correctness of input information.

Check that all required information has been specified and
that, at least as far as can be easily checked, it makes
sense.

NOTE: This is also where any default values are applied.

Definition at line 2192 of file cmsHarvester.py.

References info(), and join().

    def check_input_status(self):
        """Check completeness and correctness of input information.

        Check that all required information has been specified and
        that, at least as far as can be easily checked, it makes
        sense.

        NOTE: This is also where any default values are applied.

        """

        self.logger.info("Checking completeness/correctness of input...")

        # The cmsHarvester does not take (i.e. understand) any
        # arguments so there should not be any.
        if len(self.args) > 0:
            msg = "Sorry but I don't understand `%s'" % \
                  (" ".join(self.args))
            self.logger.fatal(msg)
            raise Usage(msg)

        # BUG BUG BUG
        # While we wait for some bugs left and right to get fixed, we
        # disable two-step.
        if self.harvesting_mode == "two-step":
            msg = "--------------------\n" \
                  "  Sorry, but for the moment (well, till it works)" \
                  "  the two-step mode has been disabled.\n" \
                  "--------------------\n"
            self.logger.fatal(msg)
            raise Error(msg)
        # BUG BUG BUG end

        # We need a harvesting method to be specified
        if self.harvesting_type is None:
            msg = "Please specify a harvesting type"
            self.logger.fatal(msg)
            raise Usage(msg)
        # as well as a harvesting mode.
        if self.harvesting_mode is None:
            self.harvesting_mode = self.harvesting_mode_default
            msg = "No harvesting mode specified --> using default `%s'" % \
                  self.harvesting_mode
            self.logger.warning(msg)
            #raise Usage(msg)

        

        # We need an input method so we can find the dataset name(s).
        if self.input_method["datasets"]["use"] is None:
            msg = "Please specify an input dataset name " \
                  "or a list file name"
            self.logger.fatal(msg)
            raise Usage(msg)

        # DEBUG DEBUG DEBUG
        # If we get here, we should also have an input name.
        assert not self.input_name["datasets"]["use"] is None
        # DEBUG DEBUG DEBUG end

        

        # The same holds for the reference histogram mapping file (if
        # we're using references).
        if self.use_ref_hists:
            if self.ref_hist_mappings_file_name is None:
                self.ref_hist_mappings_file_name = self.ref_hist_mappings_file_name_default
                msg = "No reference histogram mapping file specified --> " \
                      "using default `%s'" % \
                      self.ref_hist_mappings_file_name
                self.logger.warning(msg)

        

        # We need to know where to put the stuff (okay, the results)
        # on CASTOR.
        if self.castor_base_dir is None:
            self.castor_base_dir = self.castor_base_dir_default
            msg = "No CASTOR area specified -> using default `%s'" % \
                  self.castor_base_dir
            self.logger.warning(msg)
            #raise Usage(msg)

        # Only the CERN CASTOR area is supported.
        if not self.castor_base_dir.startswith(self.castor_prefix):
            msg = "CASTOR area does not start with `%s'" % \
                  self.castor_prefix
            self.logger.fatal(msg)
            if self.castor_base_dir.find("castor") > -1 and \
               not self.castor_base_dir.find("cern.ch") > -1:
                self.logger.fatal("Only CERN CASTOR is supported")
            raise Usage(msg)

        

        # TODO TODO TODO
        # This should be removed in the future, once I find out how to
        # get the config file used to create a given dataset from DBS.

        # For data we need to have a GlobalTag. (For MC we can figure
        # it out by ourselves.)
        if self.globaltag is None:
            self.logger.warning("No GlobalTag specified. This means I cannot")
            self.logger.warning("run on data, only on MC.")
            self.logger.warning("I will skip all data datasets.")

        # TODO TODO TODO end

        # Make sure the GlobalTag ends with `::All'.
        if not self.globaltag is None:
            if not self.globaltag.endswith("::All"):
                self.logger.warning("Specified GlobalTag `%s' does " \
                                    "not end in `::All' --> " \
                                    "appending this missing piece" % \
                                    self.globaltag)
                self.globaltag = "%s::All" % self.globaltag

        

        # Dump some info about the Frontier connections used.
        for (key, value) in self.frontier_connection_name.items():
            frontier_type_str = "unknown"
            if key == "globaltag":
                frontier_type_str = "the GlobalTag"
            elif key == "refhists":
                frontier_type_str = "the reference histograms"
            non_str = None
            if self.frontier_connection_overridden[key] == True:
                non_str = "non-"
            else:
                non_str = ""
            self.logger.info("Using %sdefault Frontier " \
                             "connection for %s: `%s'" % \
                             (non_str, frontier_type_str, value))

        

        # End of check_input_status.

check_ref_hist_mappings()

def cmsHarvester.check_ref_hist_mappings

(

self

)

Make sure all necessary reference histograms exist.

Check that for each of the datasets to be processed a
reference histogram is specified and that that histogram
exists in the database.

NOTE: There's a little complication here. Since this whole
thing was designed to allow (in principle) harvesting of both
data and MC datasets in one go, we need to be careful to check
the availability fof reference mappings only for those
datasets that need it.

Definition at line 5245 of file cmsHarvester.py.

References info().

    def check_ref_hist_mappings(self):
        """Make sure all necessary reference histograms exist.

        Check that for each of the datasets to be processed a
        reference histogram is specified and that that histogram
        exists in the database.

        NOTE: There's a little complication here. Since this whole
        thing was designed to allow (in principle) harvesting of both
        data and MC datasets in one go, we need to be careful to check
        the availability fof reference mappings only for those
        datasets that need it.

        """

        self.logger.info("Checking reference histogram mappings")

        for dataset_name in self.datasets_to_use:
            try:
                ref_hist_name = self.ref_hist_mappings[dataset_name]
            except KeyError:
                msg = "ERROR: No reference histogram mapping found " \
                      "for dataset `%s'" % \
                      dataset_name
                self.logger.fatal(msg)
                raise Error(msg)

            if not self.check_ref_hist_tag(ref_hist_name):
                msg = "Reference histogram tag `%s' " \
                      "(used for dataset `%s') does not exist!" % \
                      (ref_hist_name, dataset_name)
                self.logger.fatal(msg)
                raise Usage(msg)

        self.logger.info("  Done checking reference histogram mappings.")

        # End of check_ref_hist_mappings.

check_ref_hist_tag()

def cmsHarvester.check_ref_hist_tag	(		self,
			tag_name
	)

Check the existence of tag_name in database connect_name.

Check if tag_name exists as a reference histogram tag in the
database given by self.frontier_connection_name['refhists'].

Definition at line 4642 of file cmsHarvester.py.

References debug, and join().

    def check_ref_hist_tag(self, tag_name):
        """Check the existence of tag_name in database connect_name.

        Check if tag_name exists as a reference histogram tag in the
        database given by self.frontier_connection_name['refhists'].

        """

        connect_name = self.frontier_connection_name["refhists"]
        connect_name += self.db_account_name_cms_cond_dqm_summary()

        self.logger.debug("Checking existence of reference " \
                          "histogram tag `%s'" % \
                          tag_name)
        self.logger.debug("  (Using database connection `%s')" % \
                          connect_name)

        cmd = "cmscond_list_iov -c %s" % \
              connect_name
        (status, output) = subprocess.getstatusoutput(cmd)
        if status != 0:
            msg = "Could not check existence of tag `%s' in `%s'" % \
                  (tag_name, connect_name)
            self.logger.fatal(msg)
            self.logger.debug("Command used:")
            self.logger.debug("  %s" % cmd)
            self.logger.debug("Output received:")
            self.logger.debug(output)
            raise Error(msg)
        if not tag_name in output.split():
            self.logger.debug("Reference histogram tag `%s' " \
                              "does not exist in `%s'" % \
                              (tag_name, connect_name))
            self.logger.debug(" Existing tags: `%s'" % \
                              "', `".join(output.split()))
            tag_exists = False
        else:
            tag_exists = True
        self.logger.debug("  Reference histogram tag exists? " \
                          "-> %s" % tag_exists)

        # End of check_ref_hist_tag.
        return tag_exists

create_and_check_castor_dir()

def cmsHarvester.create_and_check_castor_dir	(		self,
			castor_dir
	)

Check existence of the give CASTOR dir, if necessary create
it.

Some special care has to be taken with several things like
setting the correct permissions such that CRAB can store the
output results. Of course this means that things like
/castor/cern.ch/ and user/j/ have to be recognised and treated
properly.

NOTE: Only CERN CASTOR area (/castor/cern.ch/) supported for
the moment.

NOTE: This method uses some slightly tricky caching to make
sure we don't keep over and over checking the same base paths.

Definition at line 1490 of file cmsHarvester.py.

References debug, and spr.find().

    def create_and_check_castor_dir(self, castor_dir):
        """Check existence of the give CASTOR dir, if necessary create
        it.

        Some special care has to be taken with several things like
        setting the correct permissions such that CRAB can store the
        output results. Of course this means that things like
        /castor/cern.ch/ and user/j/ have to be recognised and treated
        properly.

        NOTE: Only CERN CASTOR area (/castor/cern.ch/) supported for
        the moment.

        NOTE: This method uses some slightly tricky caching to make
        sure we don't keep over and over checking the same base paths.

        """

        

        # Local helper function to fully split a path into pieces.
        def split_completely(path):
            (parent_path, name) = os.path.split(path)
            if name == "":
                return (parent_path, )
            else:
                return split_completely(parent_path) + (name, )

        

        # Local helper function to check rfio (i.e. CASTOR)
        # directories.
        def extract_permissions(rfstat_output):
            """Parse the output from rfstat and return the
            5-digit permissions string."""

            permissions_line = [i for i in output.split("\n") \
                                if i.lower().find("protection") > -1]
            regexp = re.compile(".*\(([0123456789]{5})\).*")
            match = regexp.search(rfstat_output)
            if not match or len(match.groups()) != 1:
                msg = "Could not extract permissions " \
                      "from output: %s" % rfstat_output
                self.logger.fatal(msg)
                raise Error(msg)
            permissions = match.group(1)

            # End of extract_permissions.
            return permissions

        

        # These are the pieces of CASTOR directories that we do not
        # want to touch when modifying permissions.

        # NOTE: This is all a bit involved, basically driven by the
        # fact that one wants to treat the `j' directory of
        # `/castor/cern.ch/user/j/jhegeman/' specially.
        # BUG BUG BUG
        # This should be simplified, for example by comparing to the
        # CASTOR prefix or something like that.
        # BUG BUG BUG end
        castor_paths_dont_touch = {
            0: ["/", "castor", "cern.ch", "cms", "store", "temp",
                "dqm", "offline", "user"],
            -1: ["user", "store"]
            }

        self.logger.debug("Checking CASTOR path `%s'" % castor_dir)

        

        # First we take the full CASTOR path apart.
        castor_path_pieces = split_completely(castor_dir)

        # Now slowly rebuild the CASTOR path and see if a) all
        # permissions are set correctly and b) the final destination
        # exists.
        path = ""
        check_sizes = sorted(castor_paths_dont_touch.keys())
        len_castor_path_pieces = len(castor_path_pieces)
        for piece_index in range (len_castor_path_pieces):
            skip_this_path_piece = False
            piece = castor_path_pieces[piece_index]

create_and_check_castor_dirs()

def cmsHarvester.create_and_check_castor_dirs

(

self

)

Make sure all required CASTOR output dirs exist.

This checks the CASTOR base dir specified by the user as well
as all the subdirs required by the current set of jobs.

Definition at line 1431 of file cmsHarvester.py.

References debug, info(), mps_monitormerge.items, WZElectronSkims53X_cff.max, and MessageLogger_cfi.warning.

    def create_and_check_castor_dirs(self):
        """Make sure all required CASTOR output dirs exist.

        This checks the CASTOR base dir specified by the user as well
        as all the subdirs required by the current set of jobs.

        """

        self.logger.info("Checking (and if necessary creating) CASTOR " \
                         "output area(s)...")

        # Call the real checker method for the base dir.
        self.create_and_check_castor_dir(self.castor_base_dir)

        # Now call the checker for all (unique) subdirs.
        castor_dirs = []
        for (dataset_name, runs) in self.datasets_to_use.items():

            for run in runs:
                castor_dirs.append(self.datasets_information[dataset_name] \
                                   ["castor_path"][run])
        castor_dirs_unique = sorted(set(castor_dirs))
        # This can take some time. E.g. CRAFT08 has > 300 runs, each
        # of which will get a new directory. So we show some (rough)
        # info in between.
        ndirs = len(castor_dirs_unique)
        step = max(ndirs / 10, 1)
        for (i, castor_dir) in enumerate(castor_dirs_unique):
            if (i + 1) % step == 0 or \
                   (i + 1) == ndirs:
                self.logger.info("  %d/%d" % \
                                 (i + 1, ndirs))
            self.create_and_check_castor_dir(castor_dir)

            # Now check if the directory is empty. If (an old version
            # of) the output file already exists CRAB will run new
            # jobs but never copy the results back. We assume the user
            # knows what they are doing and only issue a warning in
            # case the directory is not empty.
            self.logger.debug("Checking if path `%s' is empty" % \
                              castor_dir)
            cmd = "rfdir %s" % castor_dir
            (status, output) = subprocess.getstatusoutput(cmd)
            if status != 0:
                msg = "Could not access directory `%s'" \
                      " !!! This is bad since I should have just" \
                      " created it !!!" % castor_dir
                self.logger.fatal(msg)
                raise Error(msg)
            if len(output) > 0:
                self.logger.warning("Output directory `%s' is not empty:" \
                                    " new jobs will fail to" \
                                    " copy back output" % \
                                    castor_dir)

        # End of create_and_check_castor_dirs.

create_castor_path_name_common()

def cmsHarvester.create_castor_path_name_common	(		self,
			dataset_name
	)

Build the common part of the output path to be used on
CASTOR.

This consists of the CASTOR area base path specified by the
user and a piece depending on the data type (data vs. MC), the
harvesting type and the dataset name followed by a piece
containing the run number and event count. (See comments in
create_castor_path_name_special for details.) This method
creates the common part, without run number and event count.

Definition at line 1327 of file cmsHarvester.py.

References create_castor_path_name_special(), python.rootplot.root2matplotlib.replace(), and nano_mu_digi_cff.strip.

    def create_castor_path_name_common(self, dataset_name):
        """Build the common part of the output path to be used on
        CASTOR.

        This consists of the CASTOR area base path specified by the
        user and a piece depending on the data type (data vs. MC), the
        harvesting type and the dataset name followed by a piece
        containing the run number and event count. (See comments in
        create_castor_path_name_special for details.) This method
        creates the common part, without run number and event count.

        """

        castor_path = self.castor_base_dir

        

        # The data type: data vs. mc.
        datatype = self.datasets_information[dataset_name]["datatype"]
        datatype = datatype.lower()
        castor_path = os.path.join(castor_path, datatype)

        # The harvesting type.
        harvesting_type = self.harvesting_type
        harvesting_type = harvesting_type.lower()
        castor_path = os.path.join(castor_path, harvesting_type)

        # The CMSSW release version (only the `digits'). Note that the
        # CMSSW version used here is the version used for harvesting,
        # not the one from the dataset. This does make the results
        # slightly harder to find. On the other hand it solves
        # problems in case one re-harvests a given dataset with a
        # different CMSSW version, which would lead to ambiguous path
        # names. (Of course for many cases the harvesting is done with
        # the same CMSSW version the dataset was created with.)
        release_version = self.cmssw_version
        release_version = release_version.lower(). \
                          replace("cmssw", ""). \
                          strip("_")
        castor_path = os.path.join(castor_path, release_version)

        # The dataset name.
        dataset_name_escaped = self.escape_dataset_name(dataset_name)
        castor_path = os.path.join(castor_path, dataset_name_escaped)

        

        castor_path = os.path.normpath(castor_path)

        # End of create_castor_path_name_common.
        return castor_path

create_castor_path_name_special()

def cmsHarvester.create_castor_path_name_special	(		self,
			dataset_name,
			run_number,
			castor_path_common
	)

Create the specialised part of the CASTOR output dir name.

NOTE: To avoid clashes with `incremental harvesting'
(re-harvesting when a dataset grows) we have to include the
event count in the path name. The underlying `problem' is that
CRAB does not overwrite existing output files so if the output
file already exists CRAB will fail to copy back the output.

NOTE: It's not possible to create different kinds of
harvesting jobs in a single call to this tool. However, in
principle it could be possible to create both data and MC jobs
in a single go.

NOTE: The number of events used in the path name is the
_total_ number of events in the dataset/run at the time of
harvesting. If we're doing partial harvesting the final
results will reflect lower statistics. This is a) the easiest
to code and b) the least likely to lead to confusion if
someone ever decides to swap/copy around file blocks between
sites.

Definition at line 1383 of file cmsHarvester.py.

Referenced by create_castor_path_name_common().

                                        castor_path_common):
        """Create the specialised part of the CASTOR output dir name.

        NOTE: To avoid clashes with `incremental harvesting'
        (re-harvesting when a dataset grows) we have to include the
        event count in the path name. The underlying `problem' is that
        CRAB does not overwrite existing output files so if the output
        file already exists CRAB will fail to copy back the output.

        NOTE: It's not possible to create different kinds of
        harvesting jobs in a single call to this tool. However, in
        principle it could be possible to create both data and MC jobs
        in a single go.

        NOTE: The number of events used in the path name is the
        _total_ number of events in the dataset/run at the time of
        harvesting. If we're doing partial harvesting the final
        results will reflect lower statistics. This is a) the easiest
        to code and b) the least likely to lead to confusion if
        someone ever decides to swap/copy around file blocks between
        sites.

        """

        castor_path = castor_path_common

        

        # The run number part.
        castor_path = os.path.join(castor_path, "run_%d" % run_number)

        

        # The event count (i.e. the number of events we currently see
        # for this dataset).
        #nevents = self.datasets_information[dataset_name] \
        #          ["num_events"][run_number]
        castor_path = os.path.join(castor_path, "nevents")

        

        castor_path = os.path.normpath(castor_path)

        # End of create_castor_path_name_special.
        return castor_path

create_config_file_name()

def cmsHarvester.create_config_file_name	(		self,
			dataset_name,
			run_number
	)

Generate the name of the configuration file to be run by
CRAB.

Depending on the harvesting mode (single-step or two-step)
this is the name of the real harvesting configuration or the
name of the first-step ME summary extraction configuration.

Definition at line 4064 of file cmsHarvester.py.

Referenced by create_multicrab_config().

    def create_config_file_name(self, dataset_name, run_number):
        """Generate the name of the configuration file to be run by
        CRAB.

        Depending on the harvesting mode (single-step or two-step)
        this is the name of the real harvesting configuration or the
        name of the first-step ME summary extraction configuration.

        """

        if self.harvesting_mode == "single-step":
            config_file_name = self.create_harvesting_config_file_name(dataset_name)
        elif self.harvesting_mode == "single-step-allow-partial":
            config_file_name = self.create_harvesting_config_file_name(dataset_name)

create_crab_config()

def cmsHarvester.create_crab_config

(

self

)

Create a CRAB configuration for a given job.

NOTE: This is _not_ a complete (as in: submittable) CRAB
configuration. It is used to store the common settings for the
multicrab configuration.

NOTE: Only CERN CASTOR area (/castor/cern.ch/) is supported.

NOTE: According to CRAB, you `Must define exactly two of
total_number_of_events, events_per_job, or
number_of_jobs.'. For single-step harvesting we force one job,
for the rest we don't really care.

# BUG BUG BUG
# With the current version of CRAB (2.6.1), in which Daniele
# fixed the behaviour of no_block_boundary for me, one _has to
# specify_ the total_number_of_events and one single site in
# the se_white_list.
# BUG BUG BUG end

Definition at line 4232 of file cmsHarvester.py.

References spr.find(), and join().

    def create_crab_config(self):
        """Create a CRAB configuration for a given job.

        NOTE: This is _not_ a complete (as in: submittable) CRAB
        configuration. It is used to store the common settings for the
        multicrab configuration.

        NOTE: Only CERN CASTOR area (/castor/cern.ch/) is supported.

        NOTE: According to CRAB, you `Must define exactly two of
        total_number_of_events, events_per_job, or
        number_of_jobs.'. For single-step harvesting we force one job,
        for the rest we don't really care.

        # BUG BUG BUG
        # With the current version of CRAB (2.6.1), in which Daniele
        # fixed the behaviour of no_block_boundary for me, one _has to
        # specify_ the total_number_of_events and one single site in
        # the se_white_list.
        # BUG BUG BUG end

        """

        tmp = []

        # This is the stuff we will need to fill in.
        castor_prefix = self.castor_prefix

        tmp.append(self.config_file_header())
        tmp.append("")

        
        tmp.append("[CRAB]")
        tmp.append("jobtype = cmssw")
        tmp.append("")

        
        tmp.append("[GRID]")
        tmp.append("virtual_organization=cms")
        tmp.append("")

        
        tmp.append("[USER]")
        tmp.append("copy_data = 1")
        tmp.append("")

        
        tmp.append("[CMSSW]")
        tmp.append("# This reveals data hosted on T1 sites,")
        tmp.append("# which is normally hidden by CRAB.")
        tmp.append("show_prod = 1")
        tmp.append("number_of_jobs = 1")
        if self.Jsonlumi == True:
            tmp.append("lumi_mask = %s" % self.Jsonfilename)
            tmp.append("total_number_of_lumis = -1")
        else:
            if self.harvesting_type == "DQMOffline":
                tmp.append("total_number_of_lumis = -1")
            else:
                tmp.append("total_number_of_events = -1")
        if self.harvesting_mode.find("single-step") > -1:
            tmp.append("# Force everything to run in one job.")
            tmp.append("no_block_boundary = 1")
        tmp.append("")

        
        tmp.append("[CAF]")

        crab_config = "\n".join(tmp)

        # End of create_crab_config.
        return crab_config

create_harvesting_config()

def cmsHarvester.create_harvesting_config	(		self,
			dataset_name
	)

Create the Python harvesting configuration for harvesting.

The basic configuration is created by
Configuration.PyReleaseValidation.ConfigBuilder. (This mimics
what cmsDriver.py does.) After that we add some specials
ourselves.

NOTE: On one hand it may not be nice to circumvent
cmsDriver.py, on the other hand cmsDriver.py does not really
do anything itself. All the real work is done by the
ConfigBuilder so there is not much risk that we miss out on
essential developments of cmsDriver in the future.

Definition at line 4688 of file cmsHarvester.py.

References join().

    def create_harvesting_config(self, dataset_name):
        """Create the Python harvesting configuration for harvesting.

        The basic configuration is created by
        Configuration.PyReleaseValidation.ConfigBuilder. (This mimics
        what cmsDriver.py does.) After that we add some specials
        ourselves.

        NOTE: On one hand it may not be nice to circumvent
        cmsDriver.py, on the other hand cmsDriver.py does not really
        do anything itself. All the real work is done by the
        ConfigBuilder so there is not much risk that we miss out on
        essential developments of cmsDriver in the future.

        """

        # Setup some options needed by the ConfigBuilder.
        config_options = defaultOptions

        # These are fixed for all kinds of harvesting jobs. Some of
        # them are not needed for the harvesting config, but to keep
        # the ConfigBuilder happy.
        config_options.name = "harvesting"
        config_options.scenario = "pp"
        config_options.number = 1
        config_options.arguments = self.ident_string()
        config_options.evt_type = config_options.name
        config_options.customisation_file = None
        config_options.filein = "dummy_value"
        config_options.filetype = "EDM"
        # This seems to be new in CMSSW 3.3.X, no clue what it does.
        config_options.gflash = "dummy_value"
        # This seems to be new in CMSSW 3.3.0.pre6, no clue what it
        # does.
        #config_options.himix = "dummy_value"
        config_options.dbsquery = ""

        

        # These options depend on the type of harvesting we're doing
        # and are stored in self.harvesting_info.

        config_options.step = "HARVESTING:%s" % \
                              self.harvesting_info[self.harvesting_type] \
                              ["step_string"]
        config_options.beamspot = self.harvesting_info[self.harvesting_type] \
                                  ["beamspot"]
        config_options.eventcontent = self.harvesting_info \
                                      [self.harvesting_type] \
                                      ["eventcontent"]
        config_options.harvesting = self.harvesting_info \
                                    [self.harvesting_type] \
                                    ["harvesting"]

        

        # This one is required (see also above) for each dataset.

        datatype = self.datasets_information[dataset_name]["datatype"]
        config_options.isMC = (datatype.lower() == "mc")
        config_options.isData = (datatype.lower() == "data")
        globaltag = self.datasets_information[dataset_name]["globaltag"]

        config_options.conditions = self.format_conditions_string(globaltag)

        

        if "with_input" in getargspec(ConfigBuilder.__init__)[0]:
            # This is the case for 3.3.X.
            config_builder = ConfigBuilder(config_options, with_input=True)
        else:
            # This is the case in older CMSSW versions.
            config_builder = ConfigBuilder(config_options)
        config_builder.prepare(True)
        config_contents = config_builder.pythonCfgCode

        

        # Add our signature to the top of the configuration.  and add
        # some markers to the head and the tail of the Python code
        # generated by the ConfigBuilder.
        marker_lines = []
        sep = "#" * 30
        marker_lines.append(sep)
        marker_lines.append("# Code between these markers was generated by")
        marker_lines.append("# Configuration.PyReleaseValidation." \
                            "ConfigBuilder")

        marker_lines.append(sep)
        marker = "\n".join(marker_lines)

        tmp = [self.config_file_header()]
        tmp.append("")
        tmp.append(marker)
        tmp.append("")
        tmp.append(config_contents)
        tmp.append("")
        tmp.append(marker)
        tmp.append("")
        config_contents = "\n".join(tmp)

        

        # Now we add some stuff of our own.
        customisations = [""]

        customisations.append("# Now follow some customisations")
        customisations.append("")
        connect_name = self.frontier_connection_name["globaltag"]
        connect_name += self.db_account_name_cms_cond_globaltag()
        customisations.append("process.GlobalTag.connect = \"%s\"" % \
                              connect_name)


        if self.saveByLumiSection == True:
            customisations.append("process.dqmSaver.saveByLumiSection = 1")
        

        customisations.append("")

        # About the reference histograms... For data there is only one
        # set of references and those are picked up automatically
        # based on the GlobalTag. For MC we have to do some more work
        # since the reference histograms to be used depend on the MC
        # sample at hand. In this case we glue in an es_prefer snippet
        # to pick up the references. We do this only for RelVals since
        # for MC there are no meaningful references so far.

        # NOTE: Due to the lack of meaningful references for
        # MC samples reference histograms are explicitly
        # switched off in this case.

        use_es_prefer = (self.harvesting_type == "RelVal")
        use_refs = use_es_prefer or \
                   (not self.harvesting_type == "MC")
        # Allow global override.
        use_refs = use_refs and self.use_ref_hists

        if not use_refs:
            # Disable reference histograms explicitly. The histograms
            # are loaded by the dqmRefHistoRootFileGetter
            # EDAnalyzer. This analyzer can be run from several
            # sequences. Here we remove it from each sequence that
            # exists.
            customisations.append("print \"Not using reference histograms\"")
            customisations.append("if hasattr(process, \"dqmRefHistoRootFileGetter\"):")
            customisations.append("    for (sequence_name, sequence) in process.sequences.items():")
            customisations.append("        if sequence.remove(process.dqmRefHistoRootFileGetter):")
            customisations.append("            print \"Removed process.dqmRefHistoRootFileGetter from sequence `%s'\" % \\")
            customisations.append("                  sequence_name")
            customisations.append("process.dqmSaver.referenceHandling = \"skip\"")
        else:
            # This makes sure all reference histograms are saved to
            # the output ROOT file.
            customisations.append("process.dqmSaver.referenceHandling = \"all\"")
            if use_es_prefer:
                es_prefer_snippet = self.create_es_prefer_snippet(dataset_name)
                customisations.append(es_prefer_snippet)

        # Make sure we get the `workflow' correct. As far as I can see
        # this is only important for the output file name.
        workflow_name = dataset_name
        if self.harvesting_mode == "single-step-allow-partial":
            workflow_name += "_partial"
        customisations.append("process.dqmSaver.workflow = \"%s\"" % \
                              workflow_name)

        # BUG BUG BUG
        # This still does not work. The current two-step harvesting
        # efforts are on hold waiting for the solution to come from
        # elsewhere. (In this case the elsewhere is Daniele Spiga.)

create_harvesting_config_file_name()

def cmsHarvester.create_harvesting_config_file_name	(		self,
			dataset_name
	)

Definition at line 4096 of file cmsHarvester.py.

Referenced by write_harvesting_config().

    def create_harvesting_config_file_name(self, dataset_name):
        "Generate the name to be used for the harvesting config file."

        file_name_base = "harvesting.py"
        dataset_name_escaped = self.escape_dataset_name(dataset_name)
        config_file_name = file_name_base.replace(".py",
                                                  "_%s.py" % \
                                                  dataset_name_escaped)

        # End of create_harvesting_config_file_name.
        return config_file_name

create_harvesting_output_file_name()

def cmsHarvester.create_harvesting_output_file_name	(		self,
			dataset_name,
			run_number
	)

Generate the name to be used for the harvesting output file.

This harvesting output file is the _final_ ROOT output file
containing the harvesting results. In case of two-step
harvesting there is an intermediate ME output file as well.

Definition at line 4168 of file cmsHarvester.py.

References spr.find().

    def create_harvesting_output_file_name(self, dataset_name, run_number):
        """Generate the name to be used for the harvesting output file.

        This harvesting output file is the _final_ ROOT output file
        containing the harvesting results. In case of two-step
        harvesting there is an intermediate ME output file as well.

        """

        dataset_name_escaped = self.escape_dataset_name(dataset_name)

        # Hmmm, looking at the code for the DQMFileSaver this might
        # actually be the place where the first part of this file
        # naming scheme comes from.
        # NOTE: It looks like the `V0001' comes from the DQM
        # version. This is something that cannot be looked up from
        # here, so let's hope it does not change too often.
        output_file_name = "DQM_V0001_R%09d__%s.root" % \
                           (run_number, dataset_name_escaped)
        if self.harvesting_mode.find("partial") > -1:
            # Only add the alarming piece to the file name if this is
            # a spread-out dataset.
            if self.datasets_information[dataset_name] \
                   ["mirrored"][run_number] == False:
                output_file_name = output_file_name.replace(".root", \
                                                            "_partial.root")

        # End of create_harvesting_output_file_name.
        return output_file_name

create_me_extraction_config()

def cmsHarvester.create_me_extraction_config	(		self,
			dataset_name
	)

Definition at line 4914 of file cmsHarvester.py.

References create_output_file_name(), and join().

    def create_me_extraction_config(self, dataset_name):
        """

        """

        # Big chunk of hard-coded Python. Not such a big deal since
        # this does not do much and is not likely to break.
        tmp = []
        tmp.append(self.config_file_header())
        tmp.append("")
        tmp.append("import FWCore.ParameterSet.Config as cms")
        tmp.append("")
        tmp.append("process = cms.Process(\"ME2EDM\")")
        tmp.append("")
        tmp.append("# Import of standard configurations")
        tmp.append("process.load(\"Configuration/EventContent/EventContent_cff\")")
        tmp.append("")
        tmp.append("# We don't really process any events, just keep this set to one to")
        tmp.append("# make sure things work.")
        tmp.append("process.maxEvents = cms.untracked.PSet(")
        tmp.append("    input = cms.untracked.int32(1)")
        tmp.append("    )")
        tmp.append("")
        tmp.append("process.options = cms.untracked.PSet(")
        tmp.append("    Rethrow = cms.untracked.vstring(\"ProductNotFound\")")
        tmp.append("    )")
        tmp.append("")
        tmp.append("process.source = cms.Source(\"PoolSource\",")
        tmp.append("                            processingMode = \\")
        tmp.append("                            cms.untracked.string(\"RunsAndLumis\"),")
        tmp.append("                            fileNames = \\")
        tmp.append("                            cms.untracked.vstring(\"no_file_specified\")")
        tmp.append("                            )")
        tmp.append("")
        tmp.append("# Output definition: drop everything except for the monitoring.")
        tmp.append("process.output = cms.OutputModule(")
        tmp.append("    \"PoolOutputModule\",")
        tmp.append("    outputCommands = \\")
        tmp.append("    cms.untracked.vstring(\"drop *\", \\")
        tmp.append("                          \"keep *_MEtoEDMConverter_*_*\"),")
        output_file_name = self. \
                           create_output_file_name(dataset_name)
        tmp.append("    fileName = \\")
        tmp.append("    cms.untracked.string(\"%s\")," % output_file_name)
        tmp.append("    dataset = cms.untracked.PSet(")
        tmp.append("    dataTier = cms.untracked.string(\"RECO\"),")
        tmp.append("    filterName = cms.untracked.string(\"\")")
        tmp.append("    )")
        tmp.append("    )")
        tmp.append("")
        tmp.append("# Additional output definition")
        tmp.append("process.out_step = cms.EndPath(process.output)")
        tmp.append("")
        tmp.append("# Schedule definition")
        tmp.append("process.schedule = cms.Schedule(process.out_step)")
        tmp.append("")

        config_contents = "\n".join(tmp)

        # End of create_me_extraction_config.
        return config_contents

create_me_summary_config_file_name()

def cmsHarvester.create_me_summary_config_file_name	(		self,
			dataset_name
	)

Definition at line 4110 of file cmsHarvester.py.

Referenced by write_me_extraction_config().

    def create_me_summary_config_file_name(self, dataset_name):
        "Generate the name of the ME summary extraction config file."

        file_name_base = "me_extraction.py"
        dataset_name_escaped = self.escape_dataset_name(dataset_name)
        config_file_name = file_name_base.replace(".py",
                                                  "_%s.py" % \
                                                  dataset_name_escaped)

        # End of create_me_summary_config_file_name.
        return config_file_name

create_me_summary_output_file_name()

def cmsHarvester.create_me_summary_output_file_name	(		self,
			dataset_name
	)

Generate the name of the intermediate ME file name to be
used in two-step harvesting.

Definition at line 4200 of file cmsHarvester.py.

    def create_me_summary_output_file_name(self, dataset_name):
        """Generate the name of the intermediate ME file name to be
        used in two-step harvesting.

        """

        dataset_name_escaped = self.escape_dataset_name(dataset_name)
        output_file_name = "me_summary_%s.root" % \
                           dataset_name_escaped

        # End of create_me_summary_output_file_name.
        return output_file_name

create_multicrab_block_name()

def cmsHarvester.create_multicrab_block_name	(		self,
			dataset_name,
			run_number,
			index
	)

Create the block name to use for this dataset/run number.

This is what appears in the brackets `[]' in multicrab.cfg. It
is used as the name of the job and to create output
directories.

Definition at line 4215 of file cmsHarvester.py.

    def create_multicrab_block_name(self, dataset_name, run_number, index):
        """Create the block name to use for this dataset/run number.

        This is what appears in the brackets `[]' in multicrab.cfg. It
        is used as the name of the job and to create output
        directories.

        """

        dataset_name_escaped = self.escape_dataset_name(dataset_name)
        block_name = "%s_%09d_%s" % (dataset_name_escaped, run_number, index)

        # End of create_multicrab_block_name.
        return block_name

create_multicrab_config()

def cmsHarvester.create_multicrab_config

(

self

)

Create a multicrab.cfg file for all samples.

This creates the contents for a multicrab.cfg file that uses
the crab.cfg file (generated elsewhere) for the basic settings
and contains blocks for each run of each dataset.

# BUG BUG BUG
# The fact that it's necessary to specify the se_white_list
# and the total_number_of_events is due to our use of CRAB
# version 2.6.1. This should no longer be necessary in the
# future.
# BUG BUG BUG end

Definition at line 4312 of file cmsHarvester.py.

References create_config_file_name(), create_output_file_name(), info(), join(), relativeConstraints.keys, print(), and isotrackApplyRegressor.range.

    def create_multicrab_config(self):
        """Create a multicrab.cfg file for all samples.

        This creates the contents for a multicrab.cfg file that uses
        the crab.cfg file (generated elsewhere) for the basic settings
        and contains blocks for each run of each dataset.

        # BUG BUG BUG
        # The fact that it's necessary to specify the se_white_list
        # and the total_number_of_events is due to our use of CRAB
        # version 2.6.1. This should no longer be necessary in the
        # future.
        # BUG BUG BUG end

        """

        cmd="who i am | cut -f1 -d' '"
        (status, output)=subprocess.getstatusoutput(cmd)
        UserName = output

        if self.caf_access == True:
            print("Extracting %s as user name" %UserName) 

        number_max_sites = self.nr_max_sites + 1

        multicrab_config_lines = []
        multicrab_config_lines.append(self.config_file_header())
        multicrab_config_lines.append("")
        multicrab_config_lines.append("[MULTICRAB]")
        multicrab_config_lines.append("cfg = crab.cfg")
        multicrab_config_lines.append("")

        dataset_names = sorted(self.datasets_to_use.keys())

        for dataset_name in dataset_names:
            runs = self.datasets_to_use[dataset_name]
            dataset_name_escaped = self.escape_dataset_name(dataset_name)
            castor_prefix = self.castor_prefix

            for run in runs:

                # CASTOR output dir.
                castor_dir = self.datasets_information[dataset_name] \
                                 ["castor_path"][run]

                cmd = "rfdir %s" % castor_dir
                (status, output) = subprocess.getstatusoutput(cmd)

                if len(output) <= 0:

                    # DEBUG DEBUG DEBUG
                    # We should only get here if we're treating a
                    # dataset/run that is fully contained at a single
                    # site.
                    assert (len(self.datasets_information[dataset_name] \
                            ["sites"][run]) == 1) or \
                            self.datasets_information[dataset_name]["mirrored"]
                    # DEBUG DEBUG DEBUG end

                    site_names = self.datasets_information[dataset_name] \
                             ["sites"][run].keys()

                    for i in range(1, number_max_sites, 1):
                        if len(site_names) > 0: 
                            index = "site_%02d" % (i)

                            config_file_name = self. \
                                       create_config_file_name(dataset_name, run)
                            output_file_name = self. \
                                       create_output_file_name(dataset_name, run)


                            # If we're looking at a mirrored dataset we just pick
                            # one of the sites. Otherwise there is nothing to
                            # choose.

                            # Loop variable
                            loop = 0

                            if len(site_names) > 1:
                                cmssw_version = self.datasets_information[dataset_name] \
                                            ["cmssw_version"]
                                self.logger.info("Picking site for mirrored dataset " \
                                             "`%s', run %d" % \
                                             (dataset_name, run))
                                site_name = self.pick_a_site(site_names, cmssw_version)
                                if site_name in site_names:
                                    site_names.remove(site_name)

                            else:
                                site_name = site_names[0]
                                site_names.remove(site_name)

                            if site_name is self.no_matching_site_found_str:
                                if loop < 1:
                                    break

                            nevents = self.datasets_information[dataset_name]["num_events"][run]

                            # The block name.
                            multicrab_block_name = self.create_multicrab_block_name( \
                                dataset_name, run, index)
                            multicrab_config_lines.append("[%s]" % \
                                                      multicrab_block_name)

                            
                            if site_name == "caf.cern.ch":
                                multicrab_config_lines.append("CRAB.use_server=0")
                                multicrab_config_lines.append("CRAB.scheduler=caf")
                            else:
                                multicrab_config_lines.append("scheduler = glite")

                            
                            if site_name == "caf.cern.ch":
                                pass
                            else:
                                multicrab_config_lines.append("GRID.se_white_list = %s" % \
                                                      site_name)
                                multicrab_config_lines.append("# This removes the default blacklisting of T1 sites.")
                                multicrab_config_lines.append("GRID.remove_default_blacklist = 1")
                                multicrab_config_lines.append("GRID.rb = CERN")
                                if not self.non_t1access:
                                    multicrab_config_lines.append("GRID.role = t1access")

                            

                            castor_dir = castor_dir.replace(castor_prefix, "")
                            multicrab_config_lines.append("USER.storage_element=srm-cms.cern.ch")
                            multicrab_config_lines.append("USER.user_remote_dir = %s" % \
                                                      castor_dir)
                            multicrab_config_lines.append("USER.check_user_remote_dir=0")

                            if site_name == "caf.cern.ch":
                                multicrab_config_lines.append("USER.storage_path=%s" % castor_prefix)
                                #multicrab_config_lines.append("USER.storage_element=T2_CH_CAF")
                                #castor_dir = castor_dir.replace("/cms/store/caf/user/%s" %UserName, "")
                                #multicrab_config_lines.append("USER.user_remote_dir = %s" % \
                                #             castor_dir)
                            else:
                                multicrab_config_lines.append("USER.storage_path=/srm/managerv2?SFN=%s" % castor_prefix)
                                #multicrab_config_lines.append("USER.user_remote_dir = %s" % \
                                #             castor_dir)
                                #multicrab_config_lines.append("USER.storage_element=srm-cms.cern.ch")

                            
                            multicrab_config_lines.append("CMSSW.pset = %s" % \
                                                       config_file_name)
                            multicrab_config_lines.append("CMSSW.datasetpath = %s" % \
                                                      dataset_name)
                            multicrab_config_lines.append("CMSSW.runselection = %d" % \
                                                  run)

                            if self.Jsonlumi == True:
                                pass
                            else:
                                if self.harvesting_type == "DQMOffline":
                                    pass
                                else:
                                    multicrab_config_lines.append("CMSSW.total_number_of_events = %d" % \
                                                                  nevents)
                            # The output file name.
                            multicrab_config_lines.append("CMSSW.output_file = %s" % \
                                                      output_file_name)

                            
                            if site_name == "caf.cern.ch":
                                multicrab_config_lines.append("CAF.queue=cmscaf1nd")


                            # End of block.
                            multicrab_config_lines.append("")

                            loop = loop + 1

                            self.all_sites_found = True

        multicrab_config = "\n".join(multicrab_config_lines)

        # End of create_multicrab_config.
        return multicrab_config

create_output_file_name()

def cmsHarvester.create_output_file_name	(		self,
			dataset_name,
			run_number = None
	)

Create the name of the output file name to be used.

This is the name of the output file of the `first step'. In
the case of single-step harvesting this is already the final
harvesting output ROOT file. In the case of two-step
harvesting it is the name of the intermediary ME summary
file.

Definition at line 4124 of file cmsHarvester.py.

Referenced by create_me_extraction_config(), and create_multicrab_config().

    def create_output_file_name(self, dataset_name, run_number=None):
        """Create the name of the output file name to be used.

        This is the name of the output file of the `first step'. In
        the case of single-step harvesting this is already the final
        harvesting output ROOT file. In the case of two-step
        harvesting it is the name of the intermediary ME summary
        file.

        """

        # BUG BUG BUG
        # This method has become a bit of a mess. Originally it was
        # nice to have one entry point for both single- and two-step
        # output file names. However, now the former needs the run
        # number, while the latter does not even know about run
        # numbers. This should be fixed up a bit.
        # BUG BUG BUG end

        if self.harvesting_mode == "single-step":
            # DEBUG DEBUG DEBUG
            assert not run_number is None
            # DEBUG DEBUG DEBUG end
            output_file_name = self.create_harvesting_output_file_name(dataset_name, run_number)
        elif self.harvesting_mode == "single-step-allow-partial":
            # DEBUG DEBUG DEBUG
            assert not run_number is None
            # DEBUG DEBUG DEBUG end
            output_file_name = self.create_harvesting_output_file_name(dataset_name, run_number)
        elif self.harvesting_mode == "two-step":
            # DEBUG DEBUG DEBUG
            assert run_number is None
            # DEBUG DEBUG DEBUG end
            output_file_name = self.create_me_summary_output_file_name(dataset_name)
        else:
            # This should not be possible, but hey...
            assert False, "ERROR Unknown harvesting mode `%s'" % \
                   self.harvesting_mode

        # End of create_harvesting_output_file_name.
        return output_file_name

dbs_check_dataset_spread()

def cmsHarvester.dbs_check_dataset_spread	(		self,
			dataset_name
	)

Figure out the number of events in each run of this dataset.

This is a more efficient way of doing this than calling
dbs_resolve_number_of_events for each run.

Definition at line 3076 of file cmsHarvester.py.

References cms::cuda.assert(), and debug.

    def dbs_check_dataset_spread(self, dataset_name):
        """Figure out the number of events in each run of this dataset.

        This is a more efficient way of doing this than calling
        dbs_resolve_number_of_events for each run.

        """

        self.logger.debug("Checking spread of dataset `%s'" % dataset_name)

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find run.number, site, file.name, file.numevents " \
                    "where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["run.number", "site", "file.name", "file.numevents"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        try:
            # OBSOLETE OBSOLETE OBSOLETE

dbs_resolve_cmssw_version()

def cmsHarvester.dbs_resolve_cmssw_version	(		self,
			dataset_name
	)

Ask DBS for the CMSSW version used to create this dataset.

Definition at line 2475 of file cmsHarvester.py.

References cms::cuda.assert().

    def dbs_resolve_cmssw_version(self, dataset_name):
        """Ask DBS for the CMSSW version used to create this dataset.

        """

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find algo.version where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["algo.version"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        cmssw_version = handler.results.values()[0]

        # DEBUG DEBUG DEBUG
        assert len(cmssw_version) == 1
        # DEBUG DEBUG DEBUG end

        cmssw_version = cmssw_version[0]

        # End of dbs_resolve_cmssw_version.
        return cmssw_version

dbs_resolve_dataset_name()

def cmsHarvester.dbs_resolve_dataset_name	(		self,
			dataset_name
	)

Use DBS to resolve a wildcarded dataset name.

Definition at line 2419 of file cmsHarvester.py.

References cms::cuda.assert(), and MessageLogger_cfi.warning.

Referenced by build_dataset_list().

    def dbs_resolve_dataset_name(self, dataset_name):
        """Use DBS to resolve a wildcarded dataset name.

        """

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        # Some minor checking to make sure that whatever we've been
        # given as dataset name actually sounds like a dataset name.
        if not (dataset_name.startswith("/") and \
                dataset_name.endswith("RECO")):
            self.logger.warning("Dataset name `%s' does not sound " \
                                 "like a valid dataset name!" % \
                                dataset_name)

        #----------

        api = self.dbs_api
        dbs_query = "find dataset where dataset like %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        # Setup parsing.
        handler = DBSXMLHandler(["dataset"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        # Parse.
        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        # Extract the results.
        datasets = handler.results.values()[0]

        # End of dbs_resolve_dataset_name.
        return datasets

dbs_resolve_datatype()

def cmsHarvester.dbs_resolve_datatype	(		self,
			dataset_name
	)

Ask DBS for the the data type (data or mc) of a given
dataset.

Definition at line 2682 of file cmsHarvester.py.

References cms::cuda.assert().

    def dbs_resolve_datatype(self, dataset_name):
        """Ask DBS for the the data type (data or mc) of a given
        dataset.

        """

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find datatype.type where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["datatype.type"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        datatype = handler.results.values()[0]

        # DEBUG DEBUG DEBUG
        assert len(datatype) == 1
        # DEBUG DEBUG DEBUG end

        datatype = datatype[0]

        # End of dbs_resolve_datatype.
        return datatype

dbs_resolve_globaltag()

def cmsHarvester.dbs_resolve_globaltag	(		self,
			dataset_name
	)

Ask DBS for the globaltag corresponding to a given dataset.

# BUG BUG BUG
# This does not seem to work for data datasets? E.g. for
# /Cosmics/Commissioning08_CRAFT0831X_V1_311_ReReco_FromSuperPointing_v1/RAW-RECO
# Probaly due to the fact that the GlobalTag changed during
# datataking...
BUG BUG BUG end

Definition at line 2626 of file cmsHarvester.py.

References cms::cuda.assert().

    def dbs_resolve_globaltag(self, dataset_name):
        """Ask DBS for the globaltag corresponding to a given dataset.

        # BUG BUG BUG
        # This does not seem to work for data datasets? E.g. for
        # /Cosmics/Commissioning08_CRAFT0831X_V1_311_ReReco_FromSuperPointing_v1/RAW-RECO
        # Probaly due to the fact that the GlobalTag changed during
        # datataking...
        BUG BUG BUG end

        """

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find dataset.tag where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["dataset.tag"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(parser)

        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        globaltag = handler.results.values()[0]

        # DEBUG DEBUG DEBUG
        assert len(globaltag) == 1
        # DEBUG DEBUG DEBUG end

        globaltag = globaltag[0]

        # End of dbs_resolve_globaltag.
        return globaltag

dbs_resolve_number_of_events()

def cmsHarvester.dbs_resolve_number_of_events	(		self,
			dataset_name,
			run_number = None
	)

Determine the number of events in a given dataset (and run).

Ask DBS for the number of events in a dataset. If a run number
is specified the number of events returned is that in that run
of that dataset. If problems occur we throw an exception.

# BUG BUG BUG
# Since DBS does not return the number of events correctly,
# neither for runs nor for whole datasets, we have to work
# around that a bit...
# BUG BUG BUG end

Definition at line 2735 of file cmsHarvester.py.

References cms::cuda.assert().

    def dbs_resolve_number_of_events(self, dataset_name, run_number=None):
        """Determine the number of events in a given dataset (and run).

        Ask DBS for the number of events in a dataset. If a run number
        is specified the number of events returned is that in that run
        of that dataset. If problems occur we throw an exception.

        # BUG BUG BUG
        # Since DBS does not return the number of events correctly,
        # neither for runs nor for whole datasets, we have to work
        # around that a bit...
        # BUG BUG BUG end

        """

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find file.name, file.numevents where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        if not run_number is None:
            dbs_query = dbq_query + (" and run = %d" % run_number)
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["file.name", "file.numevents"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        num_events = sum(handler.results["file.numevents"])

        # End of dbs_resolve_number_of_events.
        return num_events

dbs_resolve_runs()

def cmsHarvester.dbs_resolve_runs	(		self,
			dataset_name
	)

Ask DBS for the list of runs in a given dataset.

# NOTE: This does not (yet?) skip/remove empty runs. There is
# a bug in the DBS entry run.numevents (i.e. it always returns
# zero) which should be fixed in the `next DBS release'.
# See also:
#   https://savannah.cern.ch/bugs/?53452
#   https://savannah.cern.ch/bugs/?53711

Definition at line 2569 of file cmsHarvester.py.

References cms::cuda.assert(), and createfilelist.int.

    def dbs_resolve_runs(self, dataset_name):
        """Ask DBS for the list of runs in a given dataset.

        # NOTE: This does not (yet?) skip/remove empty runs. There is
        # a bug in the DBS entry run.numevents (i.e. it always returns
        # zero) which should be fixed in the `next DBS release'.
        # See also:
        #   https://savannah.cern.ch/bugs/?53452
        #   https://savannah.cern.ch/bugs/?53711

        """

        # TODO TODO TODO
        # We should remove empty runs as soon as the above mentioned
        # bug is fixed.
        # TODO TODO TODO end

        # DEBUG DEBUG DEBUG
        # If we get here DBS should have been set up already.
        assert not self.dbs_api is None
        # DEBUG DEBUG DEBUG end

        api = self.dbs_api
        dbs_query = "find run where dataset = %s " \
                    "and dataset.status = VALID" % \
                    dataset_name
        try:
            api_result = api.executeQuery(dbs_query)
        except DBSAPI.dbsApiException.DbsApiException:
            msg = "ERROR: Could not execute DBS query"
            self.logger.fatal(msg)
            raise Error(msg)

        handler = DBSXMLHandler(["run"])
        parser = xml.sax.make_parser()
        parser.setContentHandler(handler)

        try:
            xml.sax.parseString(api_result, handler)
        except SAXParseException:
            msg = "ERROR: Could not parse DBS server output"
            self.logger.fatal(msg)
            raise Error(msg)

        # DEBUG DEBUG DEBUG
        assert(handler.check_results_validity()), "ERROR The DBSXMLHandler screwed something up!"
        # DEBUG DEBUG DEBUG end

        runs = handler.results.values()[0]
        # Turn strings into integers.
        runs = sorted([int(i) for i in runs])

        # End of dbs_resolve_runs.
        return runs

escape_dataset_name()

def cmsHarvester.escape_dataset_name	(		self,
			dataset_name
	)

Escape a DBS dataset name.

Escape a DBS dataset name such that it does not cause trouble
with the file system. This means turning each `/' into `__',
except for the first one which is just removed.

Definition at line 4045 of file cmsHarvester.py.

    def escape_dataset_name(self, dataset_name):
        """Escape a DBS dataset name.

        Escape a DBS dataset name such that it does not cause trouble
        with the file system. This means turning each `/' into `__',
        except for the first one which is just removed.

        """

        escaped_dataset_name = dataset_name
        escaped_dataset_name = escaped_dataset_name.strip("/")
        escaped_dataset_name = escaped_dataset_name.replace("/", "__")

        return escaped_dataset_name

load_ref_hist_mappings()

def cmsHarvester.load_ref_hist_mappings

(

self

)

Load the reference histogram mappings from file.

The dataset name to reference histogram name mappings are read
from a text file specified in self.ref_hist_mappings_file_name.

Definition at line 5169 of file cmsHarvester.py.

References geometryDiff.file, info(), mps_monitormerge.items, relativeConstraints.keys, WZElectronSkims53X_cff.max, and nano_mu_digi_cff.strip.

    def load_ref_hist_mappings(self):
        """Load the reference histogram mappings from file.

        The dataset name to reference histogram name mappings are read
        from a text file specified in self.ref_hist_mappings_file_name.

        """

        # DEBUG DEBUG DEBUG
        assert len(self.ref_hist_mappings) < 1, \
               "ERROR Should not be RE-loading " \
               "reference histogram mappings!"
        # DEBUG DEBUG DEBUG end

        self.logger.info("Loading reference histogram mappings " \
                         "from file `%s'" % \
                         self.ref_hist_mappings_file_name)

        mappings_lines = None
        try:
            mappings_file = file(self.ref_hist_mappings_file_name, "r")            mappings_lines = mappings_file.readlines()
            mappings_file.close()
        except IOError:
            msg = "ERROR: Could not open reference histogram mapping "\
                  "file `%s'" % self.ref_hist_mappings_file_name
            self.logger.fatal(msg)
            raise Error(msg)

        

        # The format we expect is: two white-space separated pieces
        # per line. The first the dataset name for which the reference
        # should be used, the second one the name of the reference
        # histogram in the database.

        for mapping in mappings_lines:
            # Skip comment lines.
            if not mapping.startswith("#"):
                mapping = mapping.strip()
                if len(mapping) > 0:
                    mapping_pieces = mapping.split()
                    if len(mapping_pieces) != 2:
                        msg = "ERROR: The reference histogram mapping " \
                              "file contains a line I don't " \
                              "understand:\n  %s" % mapping
                        self.logger.fatal(msg)
                        raise Error(msg)
                    dataset_name = mapping_pieces[0].strip()
                    ref_hist_name = mapping_pieces[1].strip()
                    # We don't want people to accidentally specify
                    # multiple mappings for the same dataset. Just
                    # don't accept those cases.
                    if dataset_name in self.ref_hist_mappings:
                        msg = "ERROR: The reference histogram mapping " \
                              "file contains multiple mappings for " \
                              "dataset `%s'."
                        self.logger.fatal(msg)
                        raise Error(msg)

                    # All is well that ends well.
                    self.ref_hist_mappings[dataset_name] = ref_hist_name

        

        self.logger.info("  Successfully loaded %d mapping(s)" % \
                         len(self.ref_hist_mappings))
        max_len = max([len(i) for i in self.ref_hist_mappings.keys()])
        for (map_from, map_to) in self.ref_hist_mappings.items():
            self.logger.info("    %-*s -> %s" % \
                              (max_len, map_from, map_to))

        # End of load_ref_hist_mappings.

option_handler_caf_access()

def cmsHarvester.option_handler_caf_access	(		self,
			option,
			opt_str,
			value,
			parser
	)

Set the self.caf_access flag to try and create jobs that
run on the CAF.

Definition at line 1103 of file cmsHarvester.py.

    def option_handler_caf_access(self, option, opt_str, value, parser):
        """Set the self.caf_access flag to try and create jobs that
        run on the CAF.

        """
        self.caf_access = True

        self.logger.warning("Running in `caf_access' mode. " \
                            "Will try to create jobs that run " \
                            "on CAF but no" \
                            "further promises...")

        # End of option_handler_caf_access.

option_handler_castor_dir()

def cmsHarvester.option_handler_castor_dir	(		self,
			option,
			opt_str,
			value,
			parser
	)

Specify where on CASTOR the output should go.

At the moment only output to CERN CASTOR is
supported. Eventually the harvested results should go into the
central place for DQM on CASTOR anyway.

Definition at line 1061 of file cmsHarvester.py.

    def option_handler_castor_dir(self, option, opt_str, value, parser):
        """Specify where on CASTOR the output should go.

        At the moment only output to CERN CASTOR is
        supported. Eventually the harvested results should go into the
        central place for DQM on CASTOR anyway.

        """

        # Check format of specified CASTOR area.
        castor_dir = value
        #castor_dir = castor_dir.lstrip(os.path.sep)
        castor_prefix = self.castor_prefix

        # Add a leading slash if necessary and clean up the path.
        castor_dir = os.path.join(os.path.sep, castor_dir)
        self.castor_base_dir = os.path.normpath(castor_dir)

        self.logger.info("CASTOR (base) area to be used: `%s'" % \
                         self.castor_base_dir)

        # End of option_handler_castor_dir.

option_handler_crab_submission()

def cmsHarvester.option_handler_crab_submission	(		self,
			option,
			opt_str,
			value,
			parser
	)

Crab jobs are not created and
"submitted automatically",

Definition at line 1131 of file cmsHarvester.py.

    def option_handler_crab_submission(self, option, opt_str, value, parser):
        """Crab jobs are not created and
        "submitted automatically",
        """
        self.crab_submission = True

        # End of option_handler_crab_submission.

option_handler_list_types()

def cmsHarvester.option_handler_list_types	(		self,
			option,
			opt_str,
			value,
			parser
	)

List all harvesting types and their mappings.

This lists all implemented harvesting types with their
corresponding mappings to sequence names. This had to be
separated out from the help since it depends on the CMSSW
version and was making things a bit of a mess.

NOTE: There is no way (at least not that I could come up with)
to code this in a neat generic way that can be read both by
this method and by setup_harvesting_info(). Please try hard to
keep these two methods in sync!

Definition at line 1153 of file cmsHarvester.py.

References print().

    def option_handler_list_types(self, option, opt_str, value, parser):
        """List all harvesting types and their mappings.

        This lists all implemented harvesting types with their
        corresponding mappings to sequence names. This had to be
        separated out from the help since it depends on the CMSSW
        version and was making things a bit of a mess.

        NOTE: There is no way (at least not that I could come up with)
        to code this in a neat generic way that can be read both by
        this method and by setup_harvesting_info(). Please try hard to
        keep these two methods in sync!

        """

        sep_line = "-" * 50
        sep_line_short = "-" * 20

        print(sep_line)
        print("The following harvesting types are available:")
        print(sep_line)

        print("`RelVal' maps to:")
        print("  pre-3_3_0           : HARVESTING:validationHarvesting")
        print("  3_4_0_pre2 and later: HARVESTING:validationHarvesting+dqmHarvesting")
        print("  Exceptions:")
        print("    3_3_0_pre1-4        : HARVESTING:validationHarvesting")
        print("    3_3_0_pre6          : HARVESTING:validationHarvesting")
        print("    3_4_0_pre1          : HARVESTING:validationHarvesting")

        print(sep_line_short)

        print("`RelValFS' maps to:")
        print("  always              : HARVESTING:validationHarvestingFS")

        print(sep_line_short)

        print("`MC' maps to:")
        print("    always          : HARVESTING:validationprodHarvesting")

        print(sep_line_short)

        print("`DQMOffline' maps to:")
        print("  always              : HARVESTING:dqmHarvesting")

        print(sep_line)

        # We're done, let's quit. (This is the same thing optparse
        # does after printing the help.)
        raise SystemExit

        # End of option_handler_list_types.

option_handler_no_t1access()

def cmsHarvester.option_handler_no_t1access	(		self,
			option,
			opt_str,
			value,
			parser
	)

Set the self.no_t1access flag to try and create jobs that
run without special `t1access' role.

Definition at line 1086 of file cmsHarvester.py.

    def option_handler_no_t1access(self, option, opt_str, value, parser):
        """Set the self.no_t1access flag to try and create jobs that
        run without special `t1access' role.

        """

        self.non_t1access = True

        self.logger.warning("Running in `non-t1access' mode. " \
                            "Will try to create jobs that run " \
                            "without special rights but no " \
                            "further promises...")

        # End of option_handler_no_t1access.

option_handler_preferred_site()

def cmsHarvester.option_handler_preferred_site	(		self,
			option,
			opt_str,
			value,
			parser
	)

Definition at line 1147 of file cmsHarvester.py.

    def option_handler_preferred_site(self, option, opt_str, value, parser):

        self.preferred_site = value

option_handler_saveByLumiSection()

def cmsHarvester.option_handler_saveByLumiSection	(		self,
			option,
			opt_str,
			value,
			parser
	)

Set process.dqmSaver.saveByLumiSectiont=1 in cfg harvesting file

Definition at line 1119 of file cmsHarvester.py.

    def option_handler_saveByLumiSection(self, option, opt_str, value, parser):
        """Set process.dqmSaver.saveByLumiSectiont=1 in cfg harvesting file
        """
        self.saveByLumiSection = True

        self.logger.warning("waning concerning saveByLumiSection option")

        # End of option_handler_saveByLumiSection.

option_handler_sites()

def cmsHarvester.option_handler_sites	(		self,
			option,
			opt_str,
			value,
			parser
	)

Definition at line 1141 of file cmsHarvester.py.

    def option_handler_sites(self, option, opt_str, value, parser):

        self.nr_max_sites = value

parse_cmd_line_options()

def cmsHarvester.parse_cmd_line_options

(

self

)

Definition at line 1870 of file cmsHarvester.py.

    def parse_cmd_line_options(self):

        # Set up the command line parser. Note that we fix up the help
        # formatter so that we can add some text pointing people to
        # the Twiki etc.
        parser = optparse.OptionParser(version="%s %s" % \
                                       ("%prog", self.version),
                                       formatter=CMSHarvesterHelpFormatter())

        self.option_parser = parser

        # The debug switch.
        parser.add_option("-d", "--debug",
                          help="Switch on debug mode",
                          action="callback",
                          callback=self.option_handler_debug)

        # The quiet switch.
        parser.add_option("-q", "--quiet",
                          help="Be less verbose",
                          action="callback",
                          callback=self.option_handler_quiet)

        # The force switch. If this switch is used sanity checks are
        # performed but failures do not lead to aborts. Use with care.
        parser.add_option("", "--force",
                          help="Force mode. Do not abort on sanity check "
                          "failures",
                          action="callback",
                          callback=self.option_handler_force)

        # Choose between the different kinds of harvesting.
        parser.add_option("", "--harvesting_type",
                          help="Harvesting type: %s" % \
                          ", ".join(self.harvesting_types),
                          action="callback",
                          callback=self.option_handler_harvesting_type,
                          type="string",
                          metavar="HARVESTING_TYPE")

        # Choose between single-step and two-step mode.
        parser.add_option("", "--harvesting_mode",
                          help="Harvesting mode: %s (default = %s)" % \
                          (", ".join(self.harvesting_modes),
                          self.harvesting_mode_default),
                          action="callback",
                          callback=self.option_handler_harvesting_mode,
                          type="string",
                          metavar="HARVESTING_MODE")

        # Override the GlobalTag chosen by the cmsHarvester.
        parser.add_option("", "--globaltag",
                          help="GlobalTag to use. Default is the ones " \
                          "the dataset was created with for MC, for data" \
                          "a GlobalTag has to be specified.",
                          action="callback",
                          callback=self.option_handler_globaltag,
                          type="string",
                          metavar="GLOBALTAG")

        # Allow switching off of reference histograms.
        parser.add_option("", "--no-ref-hists",
                          help="Don't use any reference histograms",
                          action="callback",
                          callback=self.option_handler_no_ref_hists)

        # Allow the default (i.e. the one that should be used)
        # Frontier connection to be overridden.
        parser.add_option("", "--frontier-connection",
                          help="Use this Frontier connection to find " \
                          "GlobalTags and LocalTags (for reference " \
                          "histograms).\nPlease only use this for " \
                          "testing.",
                          action="callback",
                          callback=self.option_handler_frontier_connection,
                          type="string",
                          metavar="FRONTIER")

        # Similar to the above but specific to the Frontier connection
        # to be used for the GlobalTag.
        parser.add_option("", "--frontier-connection-for-globaltag",
                          help="Use this Frontier connection to find " \
                          "GlobalTags.\nPlease only use this for " \
                          "testing.",
                          action="callback",
                          callback=self.option_handler_frontier_connection,
                          type="string",
                          metavar="FRONTIER")

        # Similar to the above but specific to the Frontier connection
        # to be used for the reference histograms.
        parser.add_option("", "--frontier-connection-for-refhists",
                          help="Use this Frontier connection to find " \
                          "LocalTags (for reference " \
                          "histograms).\nPlease only use this for " \
                          "testing.",
                          action="callback",
                          callback=self.option_handler_frontier_connection,
                          type="string",
                          metavar="FRONTIER")

        # Option to specify the name (or a regexp) of the dataset(s)
        # to be used.
        parser.add_option("", "--dataset",
                          help="Name (or regexp) of dataset(s) to process",
                          action="callback",
                          #callback=self.option_handler_dataset_name,
                          callback=self.option_handler_input_spec,
                          type="string",
                          #dest="self.input_name",
                          metavar="DATASET")

        # Option to specify the name (or a regexp) of the dataset(s)
        # to be ignored.
        parser.add_option("", "--dataset-ignore",
                          help="Name (or regexp) of dataset(s) to ignore",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="DATASET-IGNORE")

        # Option to specify the name (or a regexp) of the run(s)
        # to be used.
        parser.add_option("", "--runs",
                          help="Run number(s) to process",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="RUNS")                   

        # Option to specify the name (or a regexp) of the run(s)
        # to be ignored.
        parser.add_option("", "--runs-ignore",
                          help="Run number(s) to ignore",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="RUNS-IGNORE")

        # Option to specify a file containing a list of dataset names
        # (or regexps) to be used.
        parser.add_option("", "--datasetfile",
                          help="File containing list of dataset names " \
                          "(or regexps) to process",
                          action="callback",
                          #callback=self.option_handler_listfile_name,
                          callback=self.option_handler_input_spec,
                          type="string",
                          #dest="self.input_name",
                          metavar="DATASETFILE")

        # Option to specify a file containing a list of dataset names
        # (or regexps) to be ignored.
        parser.add_option("", "--datasetfile-ignore",
                          help="File containing list of dataset names " \
                          "(or regexps) to ignore",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="DATASETFILE-IGNORE")

        # Option to specify a file containing a list of runs to be
        # used.
        parser.add_option("", "--runslistfile",
                          help="File containing list of run numbers " \
                          "to process",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="RUNSLISTFILE")

        # Option to specify a file containing a list of runs
        # to be ignored.
        parser.add_option("", "--runslistfile-ignore",
                          help="File containing list of run numbers " \
                          "to ignore",
                          action="callback",
                          callback=self.option_handler_input_spec,
                          type="string",
                          metavar="RUNSLISTFILE-IGNORE")

        # Option to specify a Jsonfile contaning a list of runs
        # to be used.
        parser.add_option("", "--Jsonrunfile",
                          help="Jsonfile containing dictionary of run/lumisections pairs. " \
                          "All lumisections of runs contained in dictionary are processed.",
                          action="callback",
                          callback=self.option_handler_input_Jsonrunfile,
                          type="string",
                          metavar="JSONRUNFILE")

        # Option to specify a Jsonfile contaning a dictionary of run/lumisections pairs
        # to be used.
        parser.add_option("", "--Jsonfile",
                          help="Jsonfile containing dictionary of run/lumisections pairs. " \
                          "Only specified lumisections of runs contained in dictionary are processed.",
                          action="callback",
                          callback=self.option_handler_input_Jsonfile,
                          type="string",
                          metavar="JSONFILE")

        # Option to specify a ToDo file contaning a list of runs
        # to be used.
        parser.add_option("", "--todo-file",
                          help="Todo file containing a list of runs to process.",
                          action="callback",
                          callback=self.option_handler_input_todofile,
                          type="string",
                          metavar="TODO-FILE")

        # Option to specify which file to use for the dataset name to
        # reference histogram name mappings.
        parser.add_option("", "--refhistmappingfile",
                          help="File to be use for the reference " \
                          "histogram mappings. Default: `%s'." % \
                          self.ref_hist_mappings_file_name_default,
                          action="callback",
                          callback=self.option_handler_ref_hist_mapping_file,
                          type="string",
                          metavar="REFHISTMAPPING-FILE")

        # Specify the place in CASTOR where the output should go.
        # NOTE: Only output to CASTOR is supported for the moment,
        # since the central DQM results place is on CASTOR anyway.
        parser.add_option("", "--castordir",
                          help="Place on CASTOR to store results. " \
                          "Default: `%s'." % \
                          self.castor_base_dir_default,
                          action="callback",
                          callback=self.option_handler_castor_dir,
                          type="string",
                          metavar="CASTORDIR")

        # Use this to try and create jobs that will run without
        # special `t1access' role.
        parser.add_option("", "--no-t1access",
                          help="Try to create jobs that will run " \
                          "without special `t1access' role",
                          action="callback",
                          callback=self.option_handler_no_t1access)

        # Use this to create jobs that may run on CAF
        parser.add_option("", "--caf-access",
                          help="Crab jobs may run " \
                          "on CAF",
                          action="callback",
                          callback=self.option_handler_caf_access)

        # set process.dqmSaver.saveByLumiSection=1 in harvesting cfg file
        parser.add_option("", "--saveByLumiSection",
                          help="set saveByLumiSection=1 in harvesting cfg file",
                          action="callback",
                          callback=self.option_handler_saveByLumiSection)

        # Use this to enable automatic creation and submission of crab jobs
        parser.add_option("", "--automatic-crab-submission",
                          help="Crab jobs are created and " \
                          "submitted automatically",
                          action="callback",
                          callback=self.option_handler_crab_submission)

        # Option to set the max number of sites, each
        #job is submitted to 
        parser.add_option("", "--max-sites",
                          help="Max. number of sites each job is submitted to",
                          action="callback",
                          callback=self.option_handler_sites,
                          type="int") 

        # Option to set the preferred site
        parser.add_option("", "--site",
                          help="Crab jobs are submitted to specified site. T1 sites may be shortened by the following (country) codes: \
                          srm-cms.cern.ch : CH \
                          ccsrm.in2p3.fr : FR \
                          cmssrm-fzk.gridka.de : DE \
                          cmssrm.fnal.gov : GOV \
                          gridka-dCache.fzk.de : DE2 \
                          rm-cms.gridpp.rl.ac.uk : UK \
                          srm.grid.sinica.edu.tw : TW \
                          srm2.grid.sinica.edu.tw : TW2 \
                          srmcms.pic.es : ES \
                          storm-fe-cms.cr.cnaf.infn.it : IT",
                          action="callback",
                          callback=self.option_handler_preferred_site,
                          type="string") 

        # This is the command line flag to list all harvesting
        # type-to-sequence mappings.
        parser.add_option("-l", "--list",
                          help="List all harvesting types and their" \
                          "corresponding sequence names",
                          action="callback",
                          callback=self.option_handler_list_types)

        # If nothing was specified: tell the user how to do things the
        # next time and exit.
        # NOTE: We just use the OptParse standard way of doing this by
        #       acting as if a '--help' was specified.
        if len(self.cmd_line_opts) < 1:
            self.cmd_line_opts = ["--help"]

        # Some trickery with the options. Why? Well, since these
        # options change the output level immediately from the option
        # handlers, the results differ depending on where they are on
        # the command line. Let's just make sure they are at the
        # front.
        # NOTE: Not very efficient or sophisticated, but it works and
        # it only has to be done once anyway.
        for i in ["-d", "--debug",
                  "-q", "--quiet"]:
            if i in self.cmd_line_opts:
                self.cmd_line_opts.remove(i)
                self.cmd_line_opts.insert(0, i)

        # Everything is set up, now parse what we were given.
        parser.set_defaults()
        (self.options, self.args) = parser.parse_args(self.cmd_line_opts)

        # End of parse_cmd_line_options.

pick_a_site()

def cmsHarvester.pick_a_site	(		self,
			sites,
			cmssw_version
	)

Definition at line 1706 of file cmsHarvester.py.

References debug, relativeConstraints.error, and info().

    def pick_a_site(self, sites, cmssw_version):

        # Create list of forbidden sites
        sites_forbidden = []

        if (self.preferred_site == "CAF") or (self.preferred_site == "caf.cern.ch"):
            self.caf_access = True

        if self.caf_access == False:
            sites_forbidden.append("caf.cern.ch")

        # These are the T1 sites. These are only forbidden if we're
        # running in non-T1 mode.
        # Source:
        # https://cmsweb.cern.ch/sitedb/sitelist/?naming_scheme=ce
        # Hard-coded, yes. Not nice, no.

        all_t1 = [
                "srm-cms.cern.ch",
                "ccsrm.in2p3.fr",
                "cmssrm-fzk.gridka.de",
                "cmssrm.fnal.gov",
                "gridka-dCache.fzk.de",
                "srm-cms.gridpp.rl.ac.uk",
                "srm.grid.sinica.edu.tw",
                "srm2.grid.sinica.edu.tw",
                "srmcms.pic.es",
                "storm-fe-cms.cr.cnaf.infn.it"
                ]

        country_codes = {
              "CAF" : "caf.cern.ch",
              "CH" : "srm-cms.cern.ch",
              "FR" : "ccsrm.in2p3.fr",
              "DE" : "cmssrm-fzk.gridka.de",
              "GOV" : "cmssrm.fnal.gov",
              "DE2" : "gridka-dCache.fzk.de",
              "UK" : "srm-cms.gridpp.rl.ac.uk",
              "TW" : "srm.grid.sinica.edu.tw",
              "TW2" : "srm2.grid.sinica.edu.tw",
              "ES" : "srmcms.pic.es",
              "IT" : "storm-fe-cms.cr.cnaf.infn.it"
                }

        if self.non_t1access: 
            sites_forbidden.extend(all_t1)

        for site in sites_forbidden:
            if site in sites:
                sites.remove(site)

        if self.preferred_site in country_codes:
            self.preferred_site = country_codes[self.preferred_site]

        if self.preferred_site != "no preference":
            if self.preferred_site in sites:
                sites = [self.preferred_site]
            else:
                sites= []

        #print sites

        # Looks like we have to do some caching here, otherwise things
        # become waaaay toooo sloooooow. So that's what the
        # sites_and_versions_cache does.

        # NOTE: Keep this set to None!
        site_name = None
        cmd = None
        while len(sites) > 0 and \
              site_name is None:

            # Create list of t1_sites
            t1_sites = []
            for site in sites:
                if site in all_t1:
                    t1_sites.append(site)
                if site == "caf.cern.ch":
                    t1_sites.append(site)

            # If avilable pick preferred site
            #if self.preferred_site in sites:
            #  se_name = self.preferred_site
            # Else, if available pick t1 site

            if len(t1_sites) > 0:
                se_name = choice(t1_sites)
            # Else pick any site
            else:
                se_name = choice(sites)

            # But check that it hosts the CMSSW version we want.

            if se_name in self.sites_and_versions_cache and \
                   cmssw_version in self.sites_and_versions_cache[se_name]:
                if self.sites_and_versions_cache[se_name][cmssw_version]:
                    site_name = se_name
                    break
                else:
                    self.logger.debug("  --> rejecting site `%s'" % se_name)
                    sites.remove(se_name)

            else:
                self.logger.info("Checking if site `%s' " \
                                 "has CMSSW version `%s'" % \
                                 (se_name, cmssw_version))
                self.sites_and_versions_cache[se_name] = {}

                # TODO TODO TODO
                # Test for SCRAM architecture removed as per request
                # from Andreas.
                # scram_arch = os.getenv("SCRAM_ARCH")
                # cmd = "lcg-info --list-ce " \
                #       "--query '" \
                #       "Tag=VO-cms-%s," \
                #       "Tag=VO-cms-%s," \
                #       "CEStatus=Production," \
                #       "CloseSE=%s'" % \
                #       (cmssw_version, scram_arch, se_name)
                # TODO TODO TODO end

                cmd = "lcg-info --list-ce " \
                      "--query '" \
                      "Tag=VO-cms-%s," \
                      "CEStatus=Production," \
                      "CloseSE=%s'" % \
                      (cmssw_version, se_name)
                (status, output) = subprocess.getstatusoutput(cmd)
                if status != 0:
                    self.logger.error("Could not check site information " \
                                      "for site `%s'" % se_name)
                else:
                    if (len(output) > 0) or (se_name == "caf.cern.ch"):
                        self.sites_and_versions_cache[se_name][cmssw_version] = True
                        site_name = se_name
                        break
                    else:
                        self.sites_and_versions_cache[se_name][cmssw_version] = False
                        self.logger.debug("  --> rejecting site `%s'" % se_name)
                        sites.remove(se_name)

        if site_name is self.no_matching_site_found_str:
            self.logger.error("  --> no matching site found")
            self.logger.error("    --> Your release or SCRAM " \
                              "architecture may not be available" \
                              "anywhere on the (LCG) grid.")
            if not cmd is None:
                self.logger.debug("      (command used: `%s')" % cmd)
        else:
            self.logger.debug("  --> selected site `%s'" % site_name)

        # Return something more descriptive (than `None') in case we
        # found nothing.
        if site_name is None:
            site_name = self.no_matching_site_found_str
            # Keep track of our global flag signifying that this
            # happened.
            self.all_sites_found = False

        # End of pick_a_site.
        return site_name

process_dataset_ignore_list()

def cmsHarvester.process_dataset_ignore_list

(

self

)

Update the list of datasets taking into account the ones to
ignore.

Both lists have been generated before from DBS and both are
assumed to be unique.

NOTE: The advantage of creating the ignore list from DBS (in
case a regexp is given) and matching that instead of directly
matching the ignore criterion against the list of datasets (to
consider) built from DBS is that in the former case we're sure
that all regexps are treated exactly as DBS would have done
without the cmsHarvester.

NOTE: This only removes complete samples. Exclusion of single
runs is done by the book keeping. So the assumption is that a
user never wants to harvest just part (i.e. n out of N runs)
of a sample.

Definition at line 3565 of file cmsHarvester.py.

References debug, info(), and relativeConstraints.keys.

    def process_dataset_ignore_list(self):
        """Update the list of datasets taking into account the ones to
        ignore.

        Both lists have been generated before from DBS and both are
        assumed to be unique.

        NOTE: The advantage of creating the ignore list from DBS (in
        case a regexp is given) and matching that instead of directly
        matching the ignore criterion against the list of datasets (to
        consider) built from DBS is that in the former case we're sure
        that all regexps are treated exactly as DBS would have done
        without the cmsHarvester.

        NOTE: This only removes complete samples. Exclusion of single
        runs is done by the book keeping. So the assumption is that a
        user never wants to harvest just part (i.e. n out of N runs)
        of a sample.

        """

        self.logger.info("Processing list of datasets to ignore...")

        self.logger.debug("Before processing ignore list there are %d " \
                          "datasets in the list to be processed" % \
                          len(self.datasets_to_use))

        # Simple approach: just loop and search.
        dataset_names_filtered = copy.deepcopy(self.datasets_to_use)
        for dataset_name in self.datasets_to_use.keys():
            if dataset_name in self.datasets_to_ignore.keys():
                del dataset_names_filtered[dataset_name]

        self.logger.info("  --> Removed %d dataset(s)" % \
                         (len(self.datasets_to_use) -
                          len(dataset_names_filtered)))

        self.datasets_to_use = dataset_names_filtered

        self.logger.debug("After processing ignore list there are %d " \
                          "datasets in the list to be processed" % \
                          len(self.datasets_to_use))

process_runs_use_and_ignore_lists()

def cmsHarvester.process_runs_use_and_ignore_lists

(

self

)

Definition at line 3612 of file cmsHarvester.py.

References info(), print(), and MessageLogger_cfi.warning.

    def process_runs_use_and_ignore_lists(self):

        self.logger.info("Processing list of runs to use and ignore...")

        # This basically adds all runs in a dataset to be processed,
        # except for any runs that are not specified in the `to use'
        # list and any runs that are specified in the `to ignore'
        # list.

        # NOTE: It is assumed that those lists make sense. The input
        # should be checked against e.g. overlapping `use' and
        # `ignore' lists.

        runs_to_use = self.runs_to_use
        runs_to_ignore = self.runs_to_ignore

        for dataset_name in self.datasets_to_use:
            runs_in_dataset = self.datasets_information[dataset_name]["runs"]

            # First some sanity checks.
            runs_to_use_tmp = []
            for run in runs_to_use:
                if not run in runs_in_dataset:
                    self.logger.warning("Dataset `%s' does not contain " \
                                        "requested run %d " \
                                        "--> ignoring `use' of this run" % \
                                        (dataset_name, run))
                else:
                    runs_to_use_tmp.append(run)

            if len(runs_to_use) > 0:
                runs = runs_to_use_tmp
                self.logger.info("Using %d out of %d runs " \
                                 "of dataset `%s'" % \
                                 (len(runs), len(runs_in_dataset),
                                  dataset_name))
            else:
                runs = runs_in_dataset

            if len(runs_to_ignore) > 0:
                runs_tmp = []
                for run in runs:
                    if not run in runs_to_ignore:
                        runs_tmp.append(run)
                self.logger.info("Ignoring %d out of %d runs " \
                                 "of dataset `%s'" % \
                                 (len(runs)- len(runs_tmp),
                                  len(runs_in_dataset),
                                  dataset_name))
                runs = runs_tmp

            if self.todofile != "YourToDofile.txt":
                runs_todo = []
                print("Reading runs from file /afs/cern.ch/cms/CAF/CMSCOMM/COMM_DQM/harvesting/%s" %self.todofile)
                cmd="grep %s /afs/cern.ch/cms/CAF/CMSCOMM/COMM_DQM/harvesting/%s | cut -f5 -d' '" %(dataset_name,self.todofile)
                (status, output)=subprocess.getstatusoutput(cmd)
                for run in runs:
                    run_str="%s" %run
                    if run_str in output:
                        runs_todo.append(run)
                self.logger.info("Using %d runs " \
                                 "of dataset `%s'" % \
                                 (len(runs_todo),
                                  dataset_name))
                runs=runs_todo

            Json_runs = []
            if self.Jsonfilename != "YourJSON.txt":
                good_runs = []
                self.Jsonlumi = True
                # We were passed a Jsonfile containing a dictionary of
                # run/lunisection-pairs
                self.logger.info("Reading runs and lumisections from file `%s'" % \
                                self.Jsonfilename)
                try:
                    Jsonfile = open(self.Jsonfilename, "r")                    for names in Jsonfile:
                        dictNames= eval(str(names))
                        for key in dictNames:
                            intkey=int(key)
                            Json_runs.append(intkey)
                    Jsonfile.close()
                except IOError:
                    msg = "ERROR: Could not open Jsonfile `%s'" % \
                          input_name
                    self.logger.fatal(msg)
                    raise Error(msg)
                for run in runs:
                    if run in Json_runs:
                        good_runs.append(run)
                self.logger.info("Using %d runs " \
                                 "of dataset `%s'" % \
                                 (len(good_runs),
                                  dataset_name))
                runs=good_runs
            if (self.Jsonrunfilename != "YourJSON.txt") and (self.Jsonfilename == "YourJSON.txt"):
                good_runs = []
                # We were passed a Jsonfile containing a dictionary of
                # run/lunisection-pairs
                self.logger.info("Reading runs from file `%s'" % \
                                self.Jsonrunfilename)
                try:
                    Jsonfile = open(self.Jsonrunfilename, "r")
                    for names in Jsonfile:
                        dictNames= eval(str(names))
                        for key in dictNames:
                            intkey=int(key)
                            Json_runs.append(intkey)
                    Jsonfile.close()
                except IOError:
                    msg = "ERROR: Could not open Jsonfile `%s'" % \
                          input_name
                    self.logger.fatal(msg)
                    raise Error(msg)
                for run in runs:
                    if run in Json_runs: 
                        good_runs.append(run)
                self.logger.info("Using %d runs " \
                                 "of dataset `%s'" % \
                                 (len(good_runs),
                                  dataset_name))
                runs=good_runs

            self.datasets_to_use[dataset_name] = runs

        # End of process_runs_use_and_ignore_lists().

ref_hist_mappings_needed()

def cmsHarvester.ref_hist_mappings_needed	(		self,
			dataset_name = None
	)

Check if we need to load and check the reference mappings.

For data the reference histograms should be taken
automatically from the GlobalTag, so we don't need any
mappings. For RelVals we need to know a mapping to be used in
the es_prefer code snippet (different references for each of
the datasets.)

WARNING: This implementation is a bit convoluted.

Definition at line 5135 of file cmsHarvester.py.

References relativeConstraints.keys.

    def ref_hist_mappings_needed(self, dataset_name=None):
        """Check if we need to load and check the reference mappings.

        For data the reference histograms should be taken
        automatically from the GlobalTag, so we don't need any
        mappings. For RelVals we need to know a mapping to be used in
        the es_prefer code snippet (different references for each of
        the datasets.)

        WARNING: This implementation is a bit convoluted.

        """

        # If no dataset name given, do everything, otherwise check
        # only this one dataset.
        if not dataset_name is None:
            data_type = self.datasets_information[dataset_name] \
                        ["datatype"]
            mappings_needed = (data_type == "mc")
            # DEBUG DEBUG DEBUG
            if not mappings_needed:
                assert data_type == "data"
            # DEBUG DEBUG DEBUG end
        else:
            tmp = [self.ref_hist_mappings_needed(dataset_name) \
                   for dataset_name in \
                   self.datasets_information.keys()]
            mappings_needed = (True in tmp)

        # End of ref_hist_mappings_needed.
        return mappings_needed

run()

def cmsHarvester.run

(

self

)

Definition at line 5486 of file cmsHarvester.py.

References info(), relativeConstraints.keys, update, and contentValuesCheck.values.

    def run(self):
        "Main entry point of the CMS harvester."

        # Start with a positive thought.
        exit_code = 0

        try:

            try:

                # Parse all command line options and arguments
                self.parse_cmd_line_options()
                # and check that they make sense.
                self.check_input_status()

                # Check if CMSSW is setup.
                self.check_cmssw()

                # Check if DBS is setup,
                self.check_dbs()
                # and if all is fine setup the Python side.
                self.setup_dbs()

                # Fill our dictionary with all the required info we
                # need to understand harvesting jobs. This needs to be
                # done after the CMSSW version is known.
                self.setup_harvesting_info()

                # Obtain list of dataset names to consider
                self.build_dataset_use_list()
                # and the list of dataset names to ignore.
                self.build_dataset_ignore_list()

                # The same for the runs lists (if specified).
                self.build_runs_use_list()
                self.build_runs_ignore_list()

                # Process the list of datasets to ignore and fold that
                # into the list of datasets to consider.
                # NOTE: The run-based selection is done later since
                # right now we don't know yet which runs a dataset
                # contains.
                self.process_dataset_ignore_list()

                # Obtain all required information on the datasets,
                # like run numbers and GlobalTags.
                self.build_datasets_information()

                if self.use_ref_hists and \
                       self.ref_hist_mappings_needed():
                    # Load the dataset name to reference histogram
                    # name mappings from file.
                    self.load_ref_hist_mappings()
                    # Now make sure that for all datasets we want to
                    # process there is a reference defined. Otherwise
                    # just bomb out before wasting any more time.
                    self.check_ref_hist_mappings()
                else:
                    self.logger.info("No need to load reference " \
                                     "histogram mappings file")

                # OBSOLETE OBSOLETE OBSOLETE

setup_dbs()

def cmsHarvester.setup_dbs

(

self

)

Setup the Python side of DBS.

For more information see the DBS Python API documentation:
https://twiki.cern.ch/twiki/bin/view/CMS/DBSApiDocumentation

Definition at line 2393 of file cmsHarvester.py.

    def setup_dbs(self):
        """Setup the Python side of DBS.

        For more information see the DBS Python API documentation:
        https://twiki.cern.ch/twiki/bin/view/CMS/DBSApiDocumentation

        """

        try:
            args={}
            args["url"]= "http://cmsdbsprod.cern.ch/cms_dbs_prod_global/" \
                         "servlet/DBSServlet"
            api = DbsApi(args)
            self.dbs_api = api

        except DBSAPI.dbsApiException.DbsApiException as ex:
            self.logger.fatal("Caught DBS API exception %s: %s "  % \
                              (ex.getClassName(), ex.getErrorMessage()))
            if ex.getErrorCode() not in (None, ""):
                logger.debug("DBS exception error code: ", ex.getErrorCode())
            raise

        # End of setup_dbs.

setup_harvesting_info()

def cmsHarvester.setup_harvesting_info

(

self

)

Fill our dictionary with all info needed to understand
harvesting.

This depends on the CMSSW version since at some point the
names and sequences were modified.

NOTE: There is no way (at least not that I could come up with)
to code this in a neat generic way that can be read both by
this method and by option_handler_list_types(). Please try
hard to keep these two methods in sync!

Definition at line 1208 of file cmsHarvester.py.

    def setup_harvesting_info(self):
        """Fill our dictionary with all info needed to understand
        harvesting.

        This depends on the CMSSW version since at some point the
        names and sequences were modified.

        NOTE: There is no way (at least not that I could come up with)
        to code this in a neat generic way that can be read both by
        this method and by option_handler_list_types(). Please try
        hard to keep these two methods in sync!

        """

        assert not self.cmssw_version is None, \
               "ERROR setup_harvesting() requires " \
               "self.cmssw_version to be set!!!"

        harvesting_info = {}

        # This is the version-independent part.
        harvesting_info["DQMOffline"] = {}
        harvesting_info["DQMOffline"]["beamspot"] = None
        harvesting_info["DQMOffline"]["eventcontent"] = None
        harvesting_info["DQMOffline"]["harvesting"] = "AtRunEnd"

        harvesting_info["RelVal"] = {}
        harvesting_info["RelVal"]["beamspot"] = None
        harvesting_info["RelVal"]["eventcontent"] = None
        harvesting_info["RelVal"]["harvesting"] = "AtRunEnd"

        harvesting_info["RelValFS"] = {}
        harvesting_info["RelValFS"]["beamspot"] = None
        harvesting_info["RelValFS"]["eventcontent"] = None
        harvesting_info["RelValFS"]["harvesting"] = "AtRunEnd"

        harvesting_info["MC"] = {}
        harvesting_info["MC"]["beamspot"] = None
        harvesting_info["MC"]["eventcontent"] = None
        harvesting_info["MC"]["harvesting"] = "AtRunEnd"

        # This is the version-dependent part. And I know, strictly
        # speaking it's not necessary to fill in all three types since
        # in a single run we'll only use one type anyway. This does
        # look more readable, however, and required less thought from
        # my side when I put this together.

        # DEBUG DEBUG DEBUG
        # Check that we understand our own version naming.
        assert self.cmssw_version.startswith("CMSSW_")
        # DEBUG DEBUG DEBUG end

        version = self.cmssw_version[6:]

        #----------

        # RelVal
        step_string = None
        if version < "3_3_0":
            step_string = "validationHarvesting"
        elif version in ["3_3_0_pre1", "3_3_0_pre2",
                         "3_3_0_pre3", "3_3_0_pre4",
                         "3_3_0_pre6", "3_4_0_pre1"]:
            step_string = "validationHarvesting"
        else:
            step_string = "validationHarvesting+dqmHarvesting"

        harvesting_info["RelVal"]["step_string"] = step_string

        # DEBUG DEBUG DEBUG
        # Let's make sure we found something.
        assert not step_string is None, \
               "ERROR Could not decide a RelVal harvesting sequence " \
               "for CMSSW version %s" % self.cmssw_version
        # DEBUG DEBUG DEBUG end

        #----------

        # RelVal
        step_string = "validationHarvestingFS"

        harvesting_info["RelValFS"]["step_string"] = step_string

        #----------

        # MC
        step_string = "validationprodHarvesting"

        harvesting_info["MC"]["step_string"] = step_string

        # DEBUG DEBUG DEBUG
        # Let's make sure we found something.
        assert not step_string is None, \
               "ERROR Could not decide a MC harvesting " \
               "sequence for CMSSW version %s" % self.cmssw_version
        # DEBUG DEBUG DEBUG end

        #----------

        # DQMOffline
        step_string = "dqmHarvesting"

        harvesting_info["DQMOffline"]["step_string"] = step_string

        #----------

        self.harvesting_info = harvesting_info

        self.logger.info("Based on the CMSSW version (%s) " \
                         "I decided to use the `HARVESTING:%s' " \
                         "sequence for %s harvesting" % \
                         (self.cmssw_version,
                          self.harvesting_info[self.harvesting_type]["step_string"],
                          self.harvesting_type))

        # End of setup_harvesting_info.

show_exit_message()

def cmsHarvester.show_exit_message

(

self

)

Tell the user what to do now, after this part is done.

This should provide the user with some (preferably
copy-pasteable) instructions on what to do now with the setups
and files that have been created.

Definition at line 5433 of file cmsHarvester.py.

References info(), and MessageLogger_cfi.warning.

    def show_exit_message(self):
        """Tell the user what to do now, after this part is done.

        This should provide the user with some (preferably
        copy-pasteable) instructions on what to do now with the setups
        and files that have been created.

        """

        # TODO TODO TODO
        # This could be improved a bit.
        # TODO TODO TODO end

        sep_line = "-" * 60

        self.logger.info("")
        self.logger.info(sep_line)
        self.logger.info("  Configuration files have been created.")
        self.logger.info("  From here on please follow the usual CRAB instructions.")
        self.logger.info("  Quick copy-paste instructions are shown below.")
        self.logger.info(sep_line)

        self.logger.info("")
        self.logger.info("    Create all CRAB jobs:")
        self.logger.info("      multicrab -create")
        self.logger.info("")
        self.logger.info("    Submit all CRAB jobs:")
        self.logger.info("      multicrab -submit")
        self.logger.info("")
        self.logger.info("    Check CRAB status:")
        self.logger.info("      multicrab -status")
        self.logger.info("")

        self.logger.info("")
        self.logger.info("  For more information please see the CMS Twiki:")
        self.logger.info("    %s" % twiki_url)
        self.logger.info(sep_line)

        # If there were any jobs for which we could not find a
        # matching site show a warning message about that.
        if not self.all_sites_found:
            self.logger.warning("  For some of the jobs no matching " \
                                "site could be found")
            self.logger.warning("  --> please scan your multicrab.cfg" \
                                "for occurrences of `%s'." % \
                                self.no_matching_site_found_str)
            self.logger.warning("      You will have to fix those " \
                                "by hand, sorry.")

        # End of show_exit_message.

singlify_datasets()

def cmsHarvester.singlify_datasets

(

self

)

Remove all but the largest part of all datasets.

This allows us to harvest at least part of these datasets
using single-step harvesting until the two-step approach
works.

Definition at line 3741 of file cmsHarvester.py.

References mps_monitormerge.items, WZElectronSkims53X_cff.max, contentValuesCheck.values, and MessageLogger_cfi.warning.

    def singlify_datasets(self):
        """Remove all but the largest part of all datasets.

        This allows us to harvest at least part of these datasets
        using single-step harvesting until the two-step approach
        works.

        """

        # DEBUG DEBUG DEBUG
        assert self.harvesting_mode == "single-step-allow-partial"
        # DEBUG DEBUG DEBUG end

        for dataset_name in self.datasets_to_use:
            for run_number in self.datasets_information[dataset_name]["runs"]:
                max_events = max(self.datasets_information[dataset_name]["sites"][run_number].values())
                sites_with_max_events = [i[0] for i in self.datasets_information[dataset_name]["sites"][run_number].items() if i[1] == max_events]
                self.logger.warning("Singlifying dataset `%s', " \
                                    "run %d" % \
                                    (dataset_name, run_number))
                cmssw_version = self.datasets_information[dataset_name] \
                                ["cmssw_version"]
                selected_site = self.pick_a_site(sites_with_max_events,
                                                 cmssw_version)

                # Let's tell the user that we're manhandling this dataset.
                nevents_old = self.datasets_information[dataset_name]["num_events"][run_number]
                self.logger.warning("  --> " \
                                    "only harvesting partial statistics: " \
                                    "%d out of %d events (5.1%f%%) " \
                                    "at site `%s'" % \
                                    (max_events,
                                     nevents_old,
                                     100. * max_events / nevents_old,
                                     selected_site))
                self.logger.warning("!!! Please note that the number of " \
                                    "events in the output path name will " \
                                    "NOT reflect the actual statistics in " \
                                    "the harvested results !!!")

                # We found the site with the highest statistics and
                # the corresponding number of events. (CRAB gets upset
                # if we ask for more events than there are at a given
                # site.) Now update this information in our main
                # datasets_information variable.
                self.datasets_information[dataset_name]["sites"][run_number] = {selected_site: max_events}
                self.datasets_information[dataset_name]["num_events"][run_number] = max_events
                #self.datasets_information[dataset_name]["sites"][run_number] = [selected_site]

        # End of singlify_datasets.

write_crab_config()

def cmsHarvester.write_crab_config

(

self

)

Write a CRAB job configuration Python file.

Definition at line 5011 of file cmsHarvester.py.

References geometryDiff.file, and info().

    def write_crab_config(self):
        """Write a CRAB job configuration Python file.

        """

        self.logger.info("Writing CRAB configuration...")

        file_name_base = "crab.cfg"

        # Create CRAB configuration.
        crab_contents = self.create_crab_config()

        # Write configuration to file.
        crab_file_name = file_name_base
        try:
            crab_file = file(crab_file_name, "w")
            crab_file.write(crab_contents)
            crab_file.close()
        except IOError:
            self.logger.fatal("Could not write " \
                              "CRAB configuration to file `%s'" % \
                              crab_file_name)
            raise Error("ERROR: Could not write to file `%s'!" % \
                        crab_file_name)

        # End of write_crab_config.

write_harvesting_config()

def cmsHarvester.write_harvesting_config	(		self,
			dataset_name
	)

Write a harvesting job configuration Python file.

NOTE: This knows nothing about single-step or two-step
harvesting. That's all taken care of by
create_harvesting_config.

Definition at line 5069 of file cmsHarvester.py.

References create_harvesting_config_file_name(), debug, and geometryDiff.file.

    def write_harvesting_config(self, dataset_name):
        """Write a harvesting job configuration Python file.

        NOTE: This knows nothing about single-step or two-step
        harvesting. That's all taken care of by
        create_harvesting_config.

        """

        self.logger.debug("Writing harvesting configuration for `%s'..." % \
                          dataset_name)

        # Create Python configuration.
        config_contents = self.create_harvesting_config(dataset_name)

        # Write configuration to file.
        config_file_name = self. \
                           create_harvesting_config_file_name(dataset_name)
        try:
            config_file = file(config_file_name, "w")
            config_file.write(config_contents)
            config_file.close()
        except IOError:
            self.logger.fatal("Could not write " \
                              "harvesting configuration to file `%s'" % \
                              config_file_name)
            raise Error("ERROR: Could not write to file `%s'!" % \
                        config_file_name)

        # End of write_harvesting_config.

write_me_extraction_config()

def cmsHarvester.write_me_extraction_config	(		self,
			dataset_name
	)

Write an ME-extraction configuration Python file.

This `ME-extraction' (ME = Monitoring Element) is the first
step of the two-step harvesting.

Definition at line 5102 of file cmsHarvester.py.

References create_me_summary_config_file_name(), debug, and geometryDiff.file.

    def write_me_extraction_config(self, dataset_name):
        """Write an ME-extraction configuration Python file.

        This `ME-extraction' (ME = Monitoring Element) is the first
        step of the two-step harvesting.

        """

        self.logger.debug("Writing ME-extraction configuration for `%s'..." % \
                          dataset_name)

        # Create Python configuration.
        config_contents = self.create_me_extraction_config(dataset_name)

        # Write configuration to file.
        config_file_name = self. \
                           create_me_summary_config_file_name(dataset_name)
        try:
            config_file = file(config_file_name, "w")
            config_file.write(config_contents)
            config_file.close()
        except IOError:
            self.logger.fatal("Could not write " \
                              "ME-extraction configuration to file `%s'" % \
                              config_file_name)
            raise Error("ERROR: Could not write to file `%s'!" % \
                        config_file_name)

        # End of write_me_extraction_config.

write_multicrab_config()

def cmsHarvester.write_multicrab_config

(

self

)

Write a multi-CRAB job configuration Python file.

Definition at line 5040 of file cmsHarvester.py.

References geometryDiff.file, and info().

    def write_multicrab_config(self):
        """Write a multi-CRAB job configuration Python file.

        """

        self.logger.info("Writing multi-CRAB configuration...")

        file_name_base = "multicrab.cfg"

        # Create multi-CRAB configuration.
        multicrab_contents = self.create_multicrab_config()

        # Write configuration to file.
        multicrab_file_name = file_name_base
        try:
            multicrab_file = file(multicrab_file_name, "w")
            multicrab_file.write(multicrab_contents)
            multicrab_file.close()
        except IOError:
            self.logger.fatal("Could not write " \
                              "multi-CRAB configuration to file `%s'" % \
                              multicrab_file_name)
            raise Error("ERROR: Could not write to file `%s'!" % \
                        multicrab_file_name)

        # End of write_multicrab_config.

Variable Documentation

__author__

cmsHarvester.__author__

private

Definition at line 40 of file cmsHarvester.py.

__version__

cmsHarvester.__version__

private

Definition at line 39 of file cmsHarvester.py.

all_file_names

cmsHarvester.all_file_names

Definition at line 3231 of file cmsHarvester.py.

all_sites_found

cmsHarvester.all_sites_found

CAF

Definition at line 1863 of file cmsHarvester.py.

caf_access

cmsHarvester.caf_access

Definition at line 1108 of file cmsHarvester.py.

castor_base_dir

cmsHarvester.castor_base_dir

Definition at line 1077 of file cmsHarvester.py.

castor_path_checks_cache

cmsHarvester.castor_path_checks_cache

self.logger.debug("Path is now `%s'" % \ path)

Definition at line 1603 of file cmsHarvester.py.

castor_path_common

cmsHarvester.castor_path_common

DEBUG DEBUG DEBUG

This is probably only useful to make sure we don't muck

things up, right?

Figure out across how many sites this sample has been spread.

if num_sites == 1: self.logger.info(" sample is contained at a single site") else: self.logger.info(" sample is spread across %d sites" % \ num_sites) if num_sites < 1:

NOTE: This should not happen with any valid dataset.

self.logger.warning(" --> skipping dataset which is not " \ "hosted anywhere")

DEBUG DEBUG DEBUG end

Definition at line 5417 of file cmsHarvester.py.

castor_paths

cmsHarvester.castor_paths

Definition at line 5421 of file cmsHarvester.py.

cmd

cmsHarvester.cmd

Definition at line 1632 of file cmsHarvester.py.

cmd_line_opts

cmsHarvester.cmd_line_opts

Definition at line 2169 of file cmsHarvester.py.

cmssw_version

cmsHarvester.cmssw_version

Definition at line 2347 of file cmsHarvester.py.

complete_sites

cmsHarvester.complete_sites

site_names_ref = set(files_info[run_number].values()[0][1]) for site_names_tmp in files_info[run_number].values()[1:]: if set(site_names_tmp[1]) != site_names_ref: mirrored = False break

Definition at line 3276 of file cmsHarvester.py.

config_contents

cmsHarvester.config_contents

In case this file is the second step (the real harvesting

step) of the two-step harvesting we have to tell it to use

our local files.

if self.harvesting_mode == "two-step": castor_dir = self.datasets_information[dataset_name] \ ["castor_path"][run] customisations.append("") customisations.append("# This is the second step (the real") customisations.append("# harvesting step) of a two-step") customisations.append("# harvesting procedure.")

BUG BUG BUG

To be removed in production version.

customisations.append("import pdb")

BUG BUG BUG end

customisations.append("import subprocess") customisations.append("import os") customisations.append("castor_dir = \"s"" % castor_dir) customisations.append("cmd = "rfdir s" % castor_dir") customisations.append("(status, output) = subprocess.getstatusoutput(cmd)") customisations.append("if status != 0:") customisations.append(" print "ERROR"") customisations.append(" raise Exception, "ERROR"") customisations.append("file_names = [os.path.join("rfio:s" % path, i) for i in output.split() if i.startswith("EDM_summary") and i.endswith(".root")]") #customisations.append("pdb.set_trace()") customisations.append("process.source.fileNames = cms.untracked.vstring(*file_names)") customisations.append("")

Definition at line 4890 of file cmsHarvester.py.

config_file_name

cmsHarvester.config_file_name

Only add the alarming piece to the file name if this is

a spread-out dataset.

pdb.set_trace() if self.datasets_information[dataset_name] \ ["mirrored"][run_number] == False: config_file_name = config_file_name.replace(".py", "_partial.py")

Definition at line 4085 of file cmsHarvester.py.

crab_submission

cmsHarvester.crab_submission

Definition at line 1135 of file cmsHarvester.py.

dataset_names_after_checks

cmsHarvester.dataset_names_after_checks

Definition at line 4030 of file cmsHarvester.py.

dataset_names_after_checks_tmp

cmsHarvester.dataset_names_after_checks_tmp

Definition at line 4023 of file cmsHarvester.py.

datasets_information

cmsHarvester.datasets_information

Definition at line 5305 of file cmsHarvester.py.

datasets_to_ignore

cmsHarvester.datasets_to_ignore

Definition at line 3457 of file cmsHarvester.py.

datasets_to_use

cmsHarvester.datasets_to_use

Definition at line 4039 of file cmsHarvester.py.

dbs_api

cmsHarvester.dbs_api

Definition at line 2406 of file cmsHarvester.py.

empty_runs

cmsHarvester.empty_runs

Definition at line 4007 of file cmsHarvester.py.

exit_code

cmsHarvester.exit_code

Definition at line 5654 of file cmsHarvester.py.

file_name

cmsHarvester.file_name

Definition at line 3175 of file cmsHarvester.py.

files_at_site

cmsHarvester.files_at_site

Definition at line 3235 of file cmsHarvester.py.

files_info

cmsHarvester.files_info

Definition at line 3161 of file cmsHarvester.py.

files_without_sites

cmsHarvester.files_without_sites

Definition at line 3201 of file cmsHarvester.py.

globaltag

cmsHarvester.globaltag

Definition at line 2307 of file cmsHarvester.py.

harvesting_info

cmsHarvester.harvesting_info

Definition at line 1314 of file cmsHarvester.py.

harvesting_mode

cmsHarvester.harvesting_mode

Definition at line 2216 of file cmsHarvester.py.

harvesting_type

cmsHarvester.harvesting_type

Definition at line 3858 of file cmsHarvester.py.

Jsonfilename

cmsHarvester.Jsonfilename

Definition at line 3707 of file cmsHarvester.py.

Jsonlumi

cmsHarvester.Jsonlumi

CRAB

GRID

USER

CMSSW

Definition at line 3681 of file cmsHarvester.py.

mirrored

cmsHarvester.mirrored

Definition at line 3222 of file cmsHarvester.py.

msg

cmsHarvester.msg

class Handler(xml.sax.handler.ContentHandler): def startElement(self, name, attrs): if name == "result": site_name = str(attrs["STORAGEELEMENT_SENAME"])

TODO TODO TODO

Ugly hack to get around cases like this:

$ dbs search –query="find dataset, site, file.count where dataset=/RelValQCD_Pt_3000_3500/CMSSW_3_3_0_pre1-STARTUP31X_V4-v1/GEN-SIM-RECO"

Using DBS instance at: http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet

Processing ...

\

PATH STORAGEELEMENT_SENAME COUNT_FILES

_________________________________________________________________________________

/RelValQCD_Pt_3000_3500/CMSSW_3_3_0_pre1-STARTUP31X_V4-v1/GEN-SIM-RECO 1

/RelValQCD_Pt_3000_3500/CMSSW_3_3_0_pre1-STARTUP31X_V4-v1/GEN-SIM-RECO cmssrm.fnal.gov 12

/RelValQCD_Pt_3000_3500/CMSSW_3_3_0_pre1-STARTUP31X_V4-v1/GEN-SIM-RECO srm-cms.cern.ch 12

if len(site_name) < 1: return

TODO TODO TODO end

run_number = int(attrs["RUNS_RUNNUMBER"]) file_name = str(attrs["FILES_LOGICALFILENAME"]) nevents = int(attrs["FILES_NUMBEROFEVENTS"])

I know, this is a bit of a kludge.

if not files_info.has_key(run_number):

New run.

files_info[run_number] = {} files_info[run_number][file_name] = (nevents, [site_name]) elif not files_info[run_number].has_key(file_name):

New file for a known run.

files_info[run_number][file_name] = (nevents, [site_name]) else:

New entry for a known file for a known run.

DEBUG DEBUG DEBUG

Each file should have the same number of

events independent of the site it's at.

assert nevents == files_info[run_number][file_name][0]

DEBUG DEBUG DEBUG end

files_info[run_number][file_name][1].append(site_name) OBSOLETE OBSOLETE OBSOLETE end

Definition at line 1640 of file cmsHarvester.py.

nevents

cmsHarvester.nevents

Definition at line 3176 of file cmsHarvester.py.

non_t1access

cmsHarvester.non_t1access

Definition at line 1092 of file cmsHarvester.py.

nr_max_sites

cmsHarvester.nr_max_sites

Definition at line 1143 of file cmsHarvester.py.

num_events_catalog

cmsHarvester.num_events_catalog

Definition at line 3215 of file cmsHarvester.py.

num_events_dataset

cmsHarvester.num_events_dataset

Definition at line 3985 of file cmsHarvester.py.

num_sites

cmsHarvester.num_sites

if self.datasets_information[dataset_name]["num_events"][run_number] != 0: pdb.set_trace() DEBUG DEBUG DEBUG end

Definition at line 3955 of file cmsHarvester.py.

option_parser

cmsHarvester.option_parser

Definition at line 1879 of file cmsHarvester.py.

output

cmsHarvester.output

Definition at line 1633 of file cmsHarvester.py.

path

cmsHarvester.path

else:

Piece not in the list, fine.

self.logger.debug(" accepting") Add piece to the path we're building. self.logger.debug("!!! Skip path piece `%s'? %s" % \ (piece, str(skip_this_path_piece))) self.logger.debug("Adding piece to path...")

Definition at line 1592 of file cmsHarvester.py.

permissions

cmsHarvester.permissions

Definition at line 1649 of file cmsHarvester.py.

Referenced by cond::CredentialStore.updatePrincipal().

permissions_new

cmsHarvester.permissions_new

Definition at line 1679 of file cmsHarvester.py.

permissions_target

cmsHarvester.permissions_target

Definition at line 1673 of file cmsHarvester.py.

preferred_site

cmsHarvester.preferred_site

Definition at line 1149 of file cmsHarvester.py.

ref_hist_mappings_file_name

cmsHarvester.ref_hist_mappings_file_name

Definition at line 2258 of file cmsHarvester.py.

run_number

cmsHarvester.run_number

Definition at line 3174 of file cmsHarvester.py.

runs_to_ignore

cmsHarvester.runs_to_ignore

Definition at line 3554 of file cmsHarvester.py.

runs_to_use

cmsHarvester.runs_to_use

Definition at line 3530 of file cmsHarvester.py.

saveByLumiSection

cmsHarvester.saveByLumiSection

Definition at line 1122 of file cmsHarvester.py.

site_names

cmsHarvester.site_names

Definition at line 3217 of file cmsHarvester.py.

sites_with_complete_copies

cmsHarvester.sites_with_complete_copies

Definition at line 3233 of file cmsHarvester.py.

skip_this_path_piece

cmsHarvester.skip_this_path_piece

self.logger.debug("Checking CASTOR path piece `%s'" % \ piece)

self.logger.debug("Checking `%s' against `%s'" % \ (castor_path_pieces[piece_index + check_size], castor_paths_dont_touch[check_size])) self.logger.debug(" skipping")

Definition at line 1584 of file cmsHarvester.py.

status

cmsHarvester.status

Definition at line 1633 of file cmsHarvester.py.

tmp

cmsHarvester.tmp

TODO TODO TODO

Need to think about where this should go, but

somewhere we have to move over the fact that we want

to process all runs for each dataset that we're

considering.

This basically means copying over the

information from self.datasets_information[]["runs"]

to self.datasets_to_use[].

for dataset_name in self.datasets_to_use.keys(): self.datasets_to_use[dataset_name] = self.datasets_information[dataset_name]["runs"]

TODO TODO TODO end

OBSOLETE OBSOLETE OBSOLETE end tmp = self.datasets_information[dataset_name] \ ["num_events"]

Definition at line 3982 of file cmsHarvester.py.

traceback_string

cmsHarvester.traceback_string

Definition at line 5679 of file cmsHarvester.py.

twiki_url

cmsHarvester.twiki_url

Definition at line 43 of file cmsHarvester.py.

---

Generated for CMSSW Reference Manual by  1.8.14