def __init__(self, url):
self._reader_url = '%s/%s' % (url, 'DBSReader')
self._writer_url = '%s/%s' % (url, 'DBSWriter')
self._migrate_url = '%s/%s' % (url, 'DBSMigrate')
self._proxy_path = os.environ.get('X509_USER_PROXY', '')
if not os.path.exists(self._proxy_path):
raise UserError(
'VOMS proxy needed to query DBS3! Environment variable X509_USER_PROXY is "%s"'
% self._proxy_path)
self._jrc = JSONRestClient(cert=self._proxy_path)
def __init__(self,
config,
datasource_name,
dataset_expr,
dataset_nick=None,
dataset_proc=None):
dataset_config = config.change_view(
default_on_change=TriggerResync(['datasets', 'parameters']))
self._lumi_filter = dataset_config.get_lookup(
['lumi filter', '%s lumi filter' % datasource_name],
default={},
parser=parse_lumi_filter,
strfun=str_lumi)
if not self._lumi_filter.empty():
config.set('%s processor' % datasource_name, 'LumiDataProcessor',
'+=')
DataProvider.__init__(self, config, datasource_name, dataset_expr,
dataset_nick, dataset_proc)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = dataset_config.get_bool(
['lumi metadata',
'%s lumi metadata' % datasource_name],
default=not self._lumi_filter.empty())
config.set('phedex sites matcher mode', 'ShellStyleMatcher', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedex_filter = dataset_config.get_filter(
'phedex sites',
'-* T1_*_Disk T2_* T3_*',
default_matcher='BlackWhiteMatcher',
default_filter='StrictListFilter')
self._only_complete = dataset_config.get_bool('only complete sites',
True)
self._only_valid = dataset_config.get_bool('only valid', True)
self._allow_phedex = dataset_config.get_bool('allow phedex', True)
self._location_format = dataset_config.get_enum(
'location format', CMSLocationFormat, CMSLocationFormat.hostname)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
self._sitedb = SiteDB()
dataset_expr_parts = split_opt(dataset_expr, '@#')
(self._dataset_path, self._dataset_instance,
self._dataset_block_selector) = dataset_expr_parts
instance_default = dataset_config.get('dbs instance', '')
self._dataset_instance = self._dataset_instance or instance_default
if not self._dataset_instance:
self._dataset_instance = 'prod/global'
elif '/' not in self._dataset_instance:
self._dataset_instance = 'prod/%s' % self._dataset_instance
self._dataset_block_selector = self._dataset_block_selector or 'all'
def __init__(self, config, datasource_name, dataset_expr, dataset_nick=None, dataset_proc=None):
dataset_config = config.change_view(default_on_change=TriggerResync(['datasets', 'parameters']))
self._lumi_filter = dataset_config.get_lookup(['lumi filter', '%s lumi filter' % datasource_name],
default={}, parser=parse_lumi_filter, strfun=str_lumi)
if not self._lumi_filter.empty():
config.set('%s processor' % datasource_name, 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasource_name, dataset_expr, dataset_nick, dataset_proc)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = dataset_config.get_bool(
['lumi metadata', '%s lumi metadata' % datasource_name], default=not self._lumi_filter.empty())
config.set('phedex sites matcher mode', 'ShellStyleMatcher', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedex_filter = dataset_config.get_filter('phedex sites', '-* T1_*_Disk T2_* T3_*',
default_matcher='BlackWhiteMatcher', default_filter='StrictListFilter')
self._only_complete = dataset_config.get_bool('only complete sites', True)
self._only_valid = dataset_config.get_bool('only valid', True)
self._location_format = dataset_config.get_enum('location format',
CMSLocationFormat, CMSLocationFormat.hostname)
self._pjrc = JSONRestClient(url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
self._sitedb = SiteDB()
dataset_expr_parts = split_opt(dataset_expr, '@#')
(self._dataset_path, self._dataset_instance, self._dataset_block_selector) = dataset_expr_parts
instance_default = dataset_config.get('dbs instance', '')
self._dataset_instance = self._dataset_instance or instance_default
if not self._dataset_instance:
self._dataset_instance = 'prod/global'
elif '/' not in self._dataset_instance:
self._dataset_instance = 'prod/%s' % self._dataset_instance
self._dataset_block_selector = self._dataset_block_selector or 'all'
def __init__(self, config, datasetExpr, datasetNick = None):
self._changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {}, parser = parseLumiFilter, strfun = strLumi, onChange = self._changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata', not self._lumi_filter.empty(), onChange = self._changeTrigger)
config.set('phedex sites matcher mode', 'shell', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites', '-* T1_*_Disk T2_* T3_*',
defaultMatcher = 'blackwhite', defaultFilter = 'strict', onChange = self._changeTrigger)
self._onlyComplete = config.getBool('only complete sites', True, onChange = self._changeTrigger)
self._locationFormat = config.getEnum('location format', CMSLocationFormat, CMSLocationFormat.hostname, onChange = self._changeTrigger)
self._pjrc = JSONRestClient(url = 'https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
self._sitedb = SiteDB()
(self._datasetPath, self._datasetInstance, self._datasetBlock) = optSplit(datasetExpr, '@#')
instance_default = config.get('dbs instance', '', onChange = self._changeTrigger)
self._datasetInstance = self._datasetInstance or instance_default
if not self._datasetInstance:
self._datasetInstance = 'prod/global'
elif '/' not in self._datasetInstance:
self._datasetInstance = 'prod/%s' % self._datasetInstance
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid', True, onChange = self._changeTrigger)
def __init__(self, config, datasetExpr, datasetNick=None, datasetID=0):
changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {},
parser=parseLumiFilter,
strfun=strLumi,
onChange=changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick,
datasetID)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata',
not self._lumi_filter.empty(),
onChange=changeTrigger)
# PhEDex blacklist: 'T1_DE_KIT', 'T1_US_FNAL' and '*_Disk' allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites',
'-T3_US_FNALLPC',
defaultMatcher='blackwhite',
defaultFilter='weak',
onChange=changeTrigger)
self._phedexT1Filter = config.getFilter('phedex t1 accept',
'T1_DE_KIT T1_US_FNAL',
defaultMatcher='blackwhite',
defaultFilter='weak',
onChange=changeTrigger)
self._phedexT1Mode = config.getEnum('phedex t1 mode',
PhedexT1Mode,
PhedexT1Mode.disk,
onChange=changeTrigger)
self.onlyComplete = config.getBool('only complete sites',
True,
onChange=changeTrigger)
self._locationFormat = config.getEnum('location format',
CMSLocationFormat,
CMSLocationFormat.hostname,
onChange=changeTrigger)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
(self._datasetPath, self._url,
self._datasetBlock) = optSplit(datasetExpr, '@#')
self._url = self._url or config.get('dbs instance', '')
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid',
True,
onChange=changeTrigger)
def __init__(self, config, datasetExpr, datasetNick=None):
self._changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {},
parser=parseLumiFilter,
strfun=strLumi,
onChange=self._changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata',
not self._lumi_filter.empty(),
onChange=self._changeTrigger)
config.set('phedex sites matcher mode', 'shell', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites',
'-* T1_*_Disk T2_* T3_*',
defaultMatcher='blackwhite',
defaultFilter='strict',
onChange=self._changeTrigger)
self._onlyComplete = config.getBool('only complete sites',
True,
onChange=self._changeTrigger)
self._locationFormat = config.getEnum('location format',
CMSLocationFormat,
CMSLocationFormat.hostname,
onChange=self._changeTrigger)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
self._sitedb = SiteDB()
(self._datasetPath, self._datasetInstance,
self._datasetBlock) = optSplit(datasetExpr, '@#')
instance_default = config.get('dbs instance',
'',
onChange=self._changeTrigger)
self._datasetInstance = self._datasetInstance or instance_default
if not self._datasetInstance:
self._datasetInstance = 'prod/global'
elif '/' not in self._datasetInstance:
self._datasetInstance = 'prod/%s' % self._datasetInstance
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid',
True,
onChange=self._changeTrigger)
def _lfn2pfn(node, lfn, prot='srmv2'):
return JSONRestClient().get(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/lfn2pfn',
params={
'node': node,
'protocol': prot,
'lfn': lfn
})['phedex']['mapping']
class DBS3LiteClient(object):
def __init__(self, url):
self._reader_url = '%s/%s' % (url, 'DBSReader')
self._writer_url = '%s/%s' % (url, 'DBSWriter')
self._migrate_url = '%s/%s' % (url, 'DBSMigrate')
self._proxy_path = os.environ.get('X509_USER_PROXY', '')
if not os.path.exists(self._proxy_path):
raise UserError(
'VOMS proxy needed to query DBS3! Environment variable X509_USER_PROXY is "%s"'
% self._proxy_path)
self._jrc = JSONRestClient(cert=self._proxy_path)
def listBlocks(self, **kwargs):
return self._jrc.get(url=self._reader_url, api='blocks', params=kwargs)
def listFiles(self, **kwargs):
return self._jrc.get(url=self._reader_url, api='files', params=kwargs)
def listFileParents(self, **kwargs):
return self._jrc.get(url=self._reader_url,
api='fileparents',
params=kwargs)
def insertBulkBlock(self, data):
return self._jrc.post(url=self._writer_url,
api='bulkblocks',
data=data)
def migrateSubmit(self, data):
return self._jrc.post(url=self._migrate_url, api='submit', data=data)
def migrateStatus(self, **kwargs):
return self._jrc.get(url=self._migrate_url,
api='status',
params=kwargs)
def __init__(self, config, datasetExpr, datasetNick = None, datasetID = 0):
changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {}, parser = parseLumiFilter, strfun = strLumi, onChange = changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick, datasetID)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata', not self._lumi_filter.empty(), onChange = changeTrigger)
# PhEDex blacklist: 'T1_DE_KIT', 'T1_US_FNAL' and '*_Disk' allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites', '-T3_US_FNALLPC',
defaultMatcher = 'blackwhite', defaultFilter = 'weak', onChange = changeTrigger)
self._phedexT1Filter = config.getFilter('phedex t1 accept', 'T1_DE_KIT T1_US_FNAL',
defaultMatcher = 'blackwhite', defaultFilter = 'weak', onChange = changeTrigger)
self._phedexT1Mode = config.getEnum('phedex t1 mode', PhedexT1Mode, PhedexT1Mode.disk, onChange = changeTrigger)
self.onlyComplete = config.getBool('only complete sites', True, onChange = changeTrigger)
self._locationFormat = config.getEnum('location format', CMSLocationFormat, CMSLocationFormat.hostname, onChange = changeTrigger)
self._pjrc = JSONRestClient(url = 'https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
(self._datasetPath, self._url, self._datasetBlock) = optSplit(datasetExpr, '@#')
self._url = self._url or config.get('dbs instance', '')
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid', True, onChange = changeTrigger)
class CMSBaseProvider(DataProvider):
def __init__(self, config, datasetExpr, datasetNick = None):
self._changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {}, parser = parseLumiFilter, strfun = strLumi, onChange = self._changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata', not self._lumi_filter.empty(), onChange = self._changeTrigger)
config.set('phedex sites matcher mode', 'shell', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites', '-* T1_*_Disk T2_* T3_*',
defaultMatcher = 'blackwhite', defaultFilter = 'strict', onChange = self._changeTrigger)
self._onlyComplete = config.getBool('only complete sites', True, onChange = self._changeTrigger)
self._locationFormat = config.getEnum('location format', CMSLocationFormat, CMSLocationFormat.hostname, onChange = self._changeTrigger)
self._pjrc = JSONRestClient(url = 'https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
self._sitedb = SiteDB()
(self._datasetPath, self._datasetInstance, self._datasetBlock) = optSplit(datasetExpr, '@#')
instance_default = config.get('dbs instance', '', onChange = self._changeTrigger)
self._datasetInstance = self._datasetInstance or instance_default
if not self._datasetInstance:
self._datasetInstance = 'prod/global'
elif '/' not in self._datasetInstance:
self._datasetInstance = 'prod/%s' % self._datasetInstance
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid', True, onChange = self._changeTrigger)
# Define how often the dataprovider can be queried automatically
def queryLimit(self):
return 2 * 60 * 60 # 2 hour delay minimum
# Check if splitterClass is valid
def checkSplitter(self, splitterClass):
if (DataSplitter.Skipped in splitterClass.neededEnums()) and not self._lumi_filter.empty():
self._log.debug('Selected splitter %s is not compatible with active lumi filter!', splitterClass.__name__)
self._log.warning('Active lumi section filter forced selection of HybridSplitter')
return HybridSplitter
return splitterClass
def _replicaLocation(self, replica_info):
(name_node, name_hostname, _) = replica_info
if self._locationFormat == CMSLocationFormat.siteDB:
yield name_node
else:
if name_hostname is not None:
name_hostnames = [name_hostname]
else:
name_hostnames = self._sitedb.cms_name_to_se(name_node)
for name_hostname in name_hostnames:
if self._locationFormat == CMSLocationFormat.hostname:
yield name_hostname
else:
yield '%s/%s' % (name_node, name_hostname)
def _fmtLocations(self, replica_infos):
for replica_info in replica_infos:
(_, _, completed) = replica_info
if completed:
for entry in self._replicaLocation(replica_info):
yield entry
else:
for entry in self._replicaLocation(replica_info):
yield '(%s)' % entry
def _processReplicas(self, blockPath, replica_infos):
def empty_with_warning(*args):
self._log.warning(*args)
return []
def expanded_replica_locations(replica_infos):
for replica_info in replica_infos:
for entry in self._replicaLocation(replica_info):
yield entry
if not replica_infos:
return empty_with_warning('Dataset block %r has no replica information!', blockPath)
replica_infos_selected = self._phedexFilter.filterList(replica_infos, key = itemgetter(0))
if not replica_infos_selected:
return empty_with_warning('Dataset block %r is not available at the selected locations!\nAvailable locations: %s', blockPath,
str.join(', ', self._fmtLocations(replica_infos)))
if not self._onlyComplete:
return list(expanded_replica_locations(replica_infos_selected))
replica_infos_complete = lfilter(lambda nn_nh_c: nn_nh_c[2], replica_infos_selected)
if not replica_infos_complete:
return empty_with_warning('Dataset block %r is not completely available at the selected locations!\nAvailable locations: %s', blockPath,
str.join(', ', self._fmtLocations(replica_infos)))
return list(expanded_replica_locations(replica_infos_complete))
# Get dataset se list from PhEDex (perhaps concurrent with listFiles)
def _getPhedexReplicas(self, blockPath, dictReplicas):
dictReplicas[blockPath] = []
for phedexBlock in self._pjrc.get(params = {'block': blockPath})['phedex']['block']:
for replica in phedexBlock['replica']:
dictReplicas[blockPath].append((replica['node'], replica.get('se'), replica['complete'] == 'y'))
def getDatasets(self):
if self._cache_dataset is None:
self._cache_dataset = [self._datasetPath]
if '*' in self._datasetPath:
self._cache_dataset = list(self._getCMSDatasets(self._datasetPath))
if not self._cache_dataset:
raise DatasetError('No datasets selected by DBS wildcard %s !' % self._datasetPath)
return self._cache_dataset
def _getCMSBlocks(self, datasetPath, getSites):
iter_blockname_selist = self._getCMSBlocksImpl(datasetPath, getSites)
n_blocks = 0
selected_blocks = False
for (blockname, selist) in iter_blockname_selist:
n_blocks += 1
if (self._datasetBlock != 'all') and (str.split(blockname, '#')[1] != self._datasetBlock):
continue
selected_blocks = True
yield (blockname, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError('Dataset %r contains %d blocks, but none were selected by %r' % (datasetPath, n_blocks, self._datasetBlock))
def _fillCMSFiles(self, block, blockPath):
lumi_used = False
lumiDict = {}
if self._lumi_query: # central lumi query
lumiDict = self._getCMSLumisImpl(blockPath)
fileList = []
for (fileInfo, listLumi) in self._getCMSFilesImpl(blockPath, self.onlyValid, self._lumi_query):
if lumiDict and not listLumi:
listLumi = lumiDict.get(fileInfo[DataProvider.URL], [])
if listLumi:
(listLumiExt_Run, listLumiExt_Lumi) = ([], [])
for (run, lumi_list) in sorted(listLumi):
listLumiExt_Run.extend([run] * len(lumi_list))
listLumiExt_Lumi.extend(lumi_list)
fileInfo[DataProvider.Metadata] = [listLumiExt_Run, listLumiExt_Lumi]
lumi_used = True
fileList.append(fileInfo)
if lumi_used:
block.setdefault(DataProvider.Metadata, []).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fileList
def _getCMSLumisImpl(self, blockPath):
return None
def _getGCBlocks(self, usePhedex):
for datasetPath in self.getDatasets():
counter = 0
for (blockPath, replica_infos) in self._getCMSBlocks(datasetPath, getSites = not usePhedex):
result = {}
result[DataProvider.Dataset] = blockPath.split('#')[0]
result[DataProvider.BlockName] = blockPath.split('#')[1]
if usePhedex: # Start parallel phedex query
dictReplicas = {}
tPhedex = start_thread('Query phedex site info for %s' % blockPath, self._getPhedexReplicas, blockPath, dictReplicas)
self._fillCMSFiles(result, blockPath)
tPhedex.join()
replica_infos = dictReplicas.get(blockPath)
else:
self._fillCMSFiles(result, blockPath)
result[DataProvider.Locations] = self._processReplicas(blockPath, replica_infos)
if len(result[DataProvider.FileList]):
counter += 1
yield result
if counter == 0:
raise DatasetError('Dataset %s does not contain any valid blocks!' % datasetPath)
tau-cream.hep.tau.ac.il
tech-crm.hep.technion.ac.il
top.ucr.edu
umiss001.hep.olemiss.edu
uosaf0008.sscc.uos.ac.kr
uscms1.fltech-grid3.fit.edu
v6ce00.grid.hep.ph.ic.ac.uk
vserv13.hep.phy.cam.ac.uk
wipp-crm.weizmann.ac.il
"""
import sys, time
from grid_control.utils.webservice import JSONRestClient
from python_compat import set, imap, lmap, lfilter, sorted
jrc = JSONRestClient(
url='http://maps.googleapis.com/maps/api/geocode/json')
def geocode(loc):
result = jrc.get(params={
'address': str.join('.',
loc.split('.')[2:]),
'sensor': 'false'
})
if 'Placemark' in result:
result = lmap(
lambda x:
(x['address'], tuple(reversed(x['Point']['coordinates'][:2]))),
result['Placemark'])
return result
counter = 0
class CMSBaseProvider(DataProvider):
# required format: <dataset path>[@<instance>][#<block>]
def __init__(self,
config,
datasource_name,
dataset_expr,
dataset_nick=None,
dataset_proc=None):
dataset_config = config.change_view(
default_on_change=TriggerResync(['datasets', 'parameters']))
self._lumi_filter = dataset_config.get_lookup(
['lumi filter', '%s lumi filter' % datasource_name],
default={},
parser=parse_lumi_filter,
strfun=str_lumi)
if not self._lumi_filter.empty():
config.set('%s processor' % datasource_name, 'LumiDataProcessor',
'+=')
DataProvider.__init__(self, config, datasource_name, dataset_expr,
dataset_nick, dataset_proc)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = dataset_config.get_bool(
['lumi metadata',
'%s lumi metadata' % datasource_name],
default=not self._lumi_filter.empty())
config.set('phedex sites matcher mode', 'ShellStyleMatcher', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedex_filter = dataset_config.get_filter(
'phedex sites',
'-* T1_*_Disk T2_* T3_*',
default_matcher='BlackWhiteMatcher',
default_filter='StrictListFilter')
self._only_complete = dataset_config.get_bool('only complete sites',
True)
self._only_valid = dataset_config.get_bool('only valid', True)
self._allow_phedex = dataset_config.get_bool('allow phedex', True)
self._location_format = dataset_config.get_enum(
'location format', CMSLocationFormat, CMSLocationFormat.hostname)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
self._sitedb = SiteDB()
dataset_expr_parts = split_opt(dataset_expr, '@#')
(self._dataset_path, self._dataset_instance,
self._dataset_block_selector) = dataset_expr_parts
instance_default = dataset_config.get('dbs instance', '')
self._dataset_instance = self._dataset_instance or instance_default
if not self._dataset_instance:
self._dataset_instance = 'prod/global'
elif '/' not in self._dataset_instance:
self._dataset_instance = 'prod/%s' % self._dataset_instance
self._dataset_block_selector = self._dataset_block_selector or 'all'
def check_splitter(self, splitter):
# Check if splitter is valid
if (DataSplitter.Skipped in splitter.get_needed_enums()
) and not self._lumi_filter.empty():
self._log.debug(
'Selected splitter %s is not compatible with active lumi filter!',
splitter.__name__)
self._log.warning(
'Active lumi section filter forced selection of HybridSplitter'
)
return HybridSplitter
return splitter
def get_dataset_name_list(self):
if self._cache_dataset is None:
self._cache_dataset = [self._dataset_path]
if '*' in self._dataset_path:
activity = Activity('Getting dataset list for %s' %
self._dataset_path)
self._cache_dataset = list(
self._get_cms_dataset_list(self._dataset_path))
if not self._cache_dataset:
raise DatasetError(
'No datasets selected by DBS wildcard %s !' %
self._dataset_path)
activity.finish()
return self._cache_dataset
def get_query_interval(self):
# Define how often the dataprovider can be queried automatically
return 2 * 60 * 60 # 2 hour delay minimum
def _fill_cms_fi_list(self, block, block_path):
activity_fi = Activity('Getting file information')
lumi_used = False
lumi_info_dict = {}
if self._lumi_query: # central lumi query
lumi_info_dict = self._get_cms_lumi_dict(block_path)
fi_list = []
for (fi,
lumi_info_list) in self._iter_cms_files(block_path,
self._only_valid,
self._lumi_query):
self._raise_on_abort()
if lumi_info_dict and not lumi_info_list:
lumi_info_list = lumi_info_dict.get(fi[DataProvider.URL], [])
if lumi_info_list:
(run_list_result, lumi_list_result) = ([], [])
for (run, lumi_list) in sorted(lumi_info_list):
run_list_result.extend([run] * len(lumi_list))
lumi_list_result.extend(lumi_list)
assert len(run_list_result) == len(lumi_list_result)
fi[DataProvider.Metadata] = [run_list_result, lumi_list_result]
lumi_used = True
fi_list.append(fi)
if lumi_used:
block.setdefault(DataProvider.Metadata,
[]).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fi_list
activity_fi.finish()
def _filter_cms_blockinfo_list(self, dataset_path, do_query_sites):
iter_dataset_block_name_selist = self._iter_cms_blocks(
dataset_path, do_query_sites)
n_blocks = 0
selected_blocks = False
for (dataset_block_name, selist) in iter_dataset_block_name_selist:
n_blocks += 1
block_name = str.split(dataset_block_name, '#')[1]
if (self._dataset_block_selector !=
'all') and (block_name != self._dataset_block_selector):
continue
selected_blocks = True
yield (dataset_block_name, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError(
'Dataset %r contains %d blocks, but none were selected by %r' %
(dataset_path, n_blocks, self._dataset_block_selector))
def _get_cms_dataset_list(self, dataset_path):
raise AbstractError
def _get_cms_lumi_dict(self, block_path):
return None
def _get_gc_block_list(self, use_phedex):
dataset_name_list = self.get_dataset_name_list()
progress_ds = ProgressActivity('Getting dataset',
len(dataset_name_list))
for dataset_idx, dataset_path in enumerate(dataset_name_list):
progress_ds.update_progress(dataset_idx,
msg='Getting dataset %s' %
dataset_path)
counter = 0
blockinfo_list = list(
self._filter_cms_blockinfo_list(dataset_path, not use_phedex))
progress_block = ProgressActivity('Getting block information',
len(blockinfo_list))
for (block_path, replica_infos) in blockinfo_list:
result = {}
result[DataProvider.Dataset] = block_path.split('#')[0]
result[DataProvider.BlockName] = block_path.split('#')[1]
progress_block.update_progress(
counter,
msg='Getting block information for ' +
result[DataProvider.BlockName])
if use_phedex and self._allow_phedex: # Start parallel phedex query
replicas_dict = {}
phedex_thread = start_thread(
'Query phedex site info for %s' % block_path,
self._get_phedex_replica_list, block_path,
replicas_dict)
self._fill_cms_fi_list(result, block_path)
phedex_thread.join()
replica_infos = replicas_dict.get(block_path)
else:
self._fill_cms_fi_list(result, block_path)
result[DataProvider.Locations] = self._process_replica_list(
block_path, replica_infos)
if len(result[DataProvider.FileList]):
counter += 1
yield result
progress_block.finish()
if counter == 0:
raise DatasetError(
'Dataset %s does not contain any valid blocks!' %
dataset_path)
progress_ds.finish()
def _get_phedex_replica_list(self, block_path, replicas_dict):
activity_fi = Activity('Getting file replica information from PhEDex')
# Get dataset se list from PhEDex (perhaps concurrent with get_dbs_file_list)
replicas_dict[block_path] = []
for phedex_block in self._pjrc.get(
params={'block': block_path})['phedex']['block']:
for replica in phedex_block['replica']:
replica_info = (replica['node'], replica.get('se'),
replica['complete'] == 'y')
replicas_dict[block_path].append(replica_info)
activity_fi.finish()
def _iter_cms_blocks(self, dataset_path, do_query_sites):
raise AbstractError
def _iter_cms_files(self, block_path, query_only_valid, query_lumi):
raise AbstractError
def _iter_formatted_locations(self, replica_infos):
for replica_info in replica_infos:
(_, _, completed) = replica_info
if completed:
for entry in self._iter_replica_locations(replica_info):
yield entry
else:
for entry in self._iter_replica_locations(replica_info):
yield '(%s)' % entry
def _iter_replica_locations(self, replica_info):
(name_node, name_hostname, _) = replica_info
if self._location_format == CMSLocationFormat.siteDB:
yield name_node
else:
if name_hostname is not None:
name_hostnames = [name_hostname]
else:
name_hostnames = self._sitedb.cms_name_to_se(name_node)
for name_hostname in name_hostnames:
if self._location_format == CMSLocationFormat.hostname:
yield name_hostname
else:
yield '%s/%s' % (name_node, name_hostname)
def _process_replica_list(self, block_path, replica_infos):
def _empty_with_warning(error_msg, *args):
self._log.warning('Dataset block %r ' + error_msg, block_path,
*args)
return []
def _expanded_replica_locations(replica_infos):
for replica_info in replica_infos:
for entry in self._iter_replica_locations(replica_info):
yield entry
if not replica_infos:
return _empty_with_warning('has no replica information!')
replica_infos_selected = self._phedex_filter.filter_list(
replica_infos, key=itemgetter(0))
if not replica_infos_selected:
return _empty_with_warning(
'is not available at the selected locations!\n' +
'Available locations: %s',
str.join(', ', self._iter_formatted_locations(replica_infos)))
if not self._only_complete:
return list(_expanded_replica_locations(replica_infos_selected))
replica_infos_complete = lfilter(lambda nn_nh_c: nn_nh_c[2],
replica_infos_selected)
if not replica_infos_complete:
return _empty_with_warning(
'is not completely available at the selected locations!\n' +
'Available locations: %s',
str.join(', ', self._iter_formatted_locations(replica_infos)))
return list(_expanded_replica_locations(replica_infos_complete))
def __init__(self):
from grid_control.utils.webservice import JSONRestClient
self._jrc = JSONRestClient(
url='http://maps.googleapis.com/maps/api/geocode/json')
class GeoResolver(object):
def __init__(self):
from grid_control.utils.webservice import JSONRestClient
self._jrc = JSONRestClient(
url='http://maps.googleapis.com/maps/api/geocode/json')
def run(self):
# output of lcg-infosites ce | while read X X X X X CE; do echo $CE; done
# | cut -d "/" -f 1 | cut -d ":" -f 1 | sort | uniq
ce_list_str = """alcyone-cms.grid.helsinki.fi
alice23.spbu.ru
arc-ce01.gridpp.rl.ac.uk
arc-ce02.gridpp.rl.ac.uk
arc-ce03.gridpp.rl.ac.uk
argoce01.na.infn.it
atlasce1.lnf.infn.it
atlasce2.lnf.infn.it
atlasce3.lnf.infn.it
atlas-cream01.na.infn.it
atlas-cream02.na.infn.it
boce.bo.infn.it
bonner-grid.rice.edu
brux3.hep.brown.edu
cale.uniandes.edu.co
carter-osg.rcac.purdue.edu
cccreamceli07.in2p3.fr
cccreamceli08.in2p3.fr
cce.ihep.ac.cn
ce0002.m45.ihep.su
ce0004.m45.ihep.su
ce01.cmsaf.mit.edu
ce01.jinr-t1.ru
ce01-lcg.cr.cnaf.infn.it
ce-01.roma3.infn.it
ce01.tier2.hep.manchester.ac.uk
ce02.cmsaf.mit.edu
ce02.jinr-t1.ru
ce02.ngcc.acad.bg
ce02.tier2.hep.manchester.ac.uk
ce04-lcg.cr.cnaf.infn.it
ce05.esc.qmul.ac.uk
ce05-lcg.cr.cnaf.infn.it
ce05.ncg.ingrid.pt
ce06.esc.qmul.ac.uk
ce06-lcg.cr.cnaf.infn.it
ce07.esc.qmul.ac.uk
ce07-lcg.cr.cnaf.infn.it
ce08-lcg.cr.cnaf.infn.it
ce101.grid.ucy.ac.cy
ce1.accre.vanderbilt.edu
ce1.dur.scotgrid.ac.uk
ce1.grid.lebedev.ru
ce1.ts.infn.it
ce201.cern.ch
ce202.cern.ch
ce203.cern.ch
ce204.cern.ch
ce205.cern.ch
ce206.cern.ch
ce207.cern.ch
ce208.cern.ch
ce2.accre.vanderbilt.edu
ce2.particles.ipm.ac.ir
ce301.cern.ch
ce302.cern.ch
ce3.ppgrid1.rhul.ac.uk
ce401.cern.ch
ce402.cern.ch
ce403.cern.ch
ce404.cern.ch
ce405.cern.ch
ce406.cern.ch
ce407.cern.ch
ce408.cern.ch
ce64.ipb.ac.rs
ce6.grid.icm.edu.pl
ce7.glite.ecdf.ed.ac.uk
ce9.grid.icm.edu.pl
cebo-t3-01.cr.cnaf.infn.it
cebo-t3-02.cr.cnaf.infn.it
ce.cis.gov.pl
cecream.ca.infn.it
ce.fesb.egi.cro-ngi.hr
ce.grid.unesp.br
ce.irb.egi.cro-ngi.hr
ceprod05.grid.hep.ph.ic.ac.uk
ceprod06.grid.hep.ph.ic.ac.uk
ceprod07.grid.hep.ph.ic.ac.uk
ceprod08.grid.hep.ph.ic.ac.uk
cert-37.pd.infn.it
ce.scope.unina.it
ce.srce.egi.cro-ngi.hr
cetest01.grid.hep.ph.ic.ac.uk
cetest02.grid.hep.ph.ic.ac.uk
ce.ulakbim.gov.tr
cit-gatekeeper2.ultralight.org
cit-gatekeeper.ultralight.org
cluster118.knu.ac.kr
cluster50.knu.ac.kr
cms-0.mps.ohio-state.edu
cmsce01.na.infn.it
cmsgrid01.hep.wisc.edu
cmsgrid02.hep.wisc.edu
cms-grid0.hep.uprm.edu
cmsosgce2.fnal.gov
cmsosgce4.fnal.gov
cmsosgce.fnal.gov
cmsrm-cream01.roma1.infn.it
cmsrm-cream02.roma1.infn.it
cmsrm-cream03.roma1.infn.it
cmstest1.rcac.purdue.edu
cms.tier3.ucdavis.edu
conte-osg.rcac.purdue.edu
cox01.grid.metu.edu.tr
cr1.ipp.acad.bg
cream01.grid.auth.gr
cream01.grid.sinica.edu.tw
cream01.grid.uoi.gr
cream01.kallisto.hellasgrid.gr
cream01.lcg.cscs.ch
cream02.grid.cyf-kr.edu.pl
cream02.iihe.ac.be
cream02.lcg.cscs.ch
cream03.lcg.cscs.ch
cream04.grid.sinica.edu.tw
cream04.lcg.cscs.ch
cream05.grid.sinica.edu.tw
cream2.ppgrid1.rhul.ac.uk
cream3.hep.kbfi.ee
cream4.hep.kbfi.ee
cream.afroditi.hellasgrid.gr
cream-ce01.ariagni.hellasgrid.gr
cream-ce01.indiacms.res.in
cream-ce01.marie.hellasgrid.gr
cream-ce02.cat.cbpf.br
creamce02.ciemat.es
cream-ce02.marie.hellasgrid.gr
creamce03.ciemat.es
creamce1.itep.ru
cream-ce-2.ba.infn.it
cream-ce-4.ba.infn.it
cream-ce.cat.cbpf.br
cream-ce.grid.atomki.hu
creamce.hephy.oeaw.ac.at
creamce.inula.man.poznan.pl
cream-ce.kipt.kharkov.ua
cream-ce.pg.infn.it
creamce.reef.man.poznan.pl
cream.grid.cyf-kr.edu.pl
cream.ipb.ac.rs
dc2-grid-66.brunel.ac.uk
dc2-grid-68.brunel.ac.uk
dc2-grid-70.brunel.ac.uk
dwarf.wcss.wroc.pl
earth.crc.nd.edu
epgr02.ph.bham.ac.uk
erbium.lsr.nectec.or.th
f-cream01.grid.sinica.edu.tw
f-cream04.grid.sinica.edu.tw
fiupg.hep.fiu.edu
foam.grid.kiae.ru
fornax-ce2.itwm.fhg.de
fornax-ce.itwm.fhg.de
grcreamce01.inr.troitsk.ru
grid001.ics.forth.gr
grid002.jet.efda.org
grid012.ct.infn.it
grid01.physics.uoi.gr
grid0.fe.infn.it
grid106.kfki.hu
grid107.kfki.hu
grid109.kfki.hu
grid129.sinp.msu.ru
grid36.lal.in2p3.fr
grid72.phy.ncu.edu.tw
gridce01.ifca.es
gridce03.ifca.es
gridce0.pi.infn.it
gridce1.pi.infn.it
grid-ce2.physik.rwth-aachen.de
gridce2.pi.infn.it
gridce3.pi.infn.it
gridce4.pi.infn.it
gridce.ilc.cnr.it
grid-ce.physik.rwth-aachen.de
grid-cr0.desy.de
grid-cr1.desy.de
grid-cr2.desy.de
grid-cr3.desy.de
grid-cr4.desy.de
gridgk01.racf.bnl.gov
gridgk02.racf.bnl.gov
gridgk03.racf.bnl.gov
gridgk04.racf.bnl.gov
gridgk05.racf.bnl.gov
gridgk06.racf.bnl.gov
gridgk08.racf.bnl.gov
gridtest02.racf.bnl.gov
gridvm03.roma2.infn.it
grisuce.scope.unina.it
gt3.pnpi.nw.ru
hansen-osg.rcac.purdue.edu
hepcms-0.umd.edu
hepgrid10.ph.liv.ac.uk
hepgrid5.ph.liv.ac.uk
hepgrid6.ph.liv.ac.uk
hepgrid97.ph.liv.ac.uk
hephygr.oeaw.ac.at
heposg01.colorado.edu
hurr.tamu.edu
ingrid.cism.ucl.ac.be
jade-cms.hip.fi
juk.nikhef.nl
kalkan1.ulakbim.gov.tr
khaldun.biruni.upm.my
klomp.nikhef.nl
kodiak-ce.baylor.edu
lcg18.sinp.msu.ru
lcg52.sinp.msu.ru
lcgce01.phy.bris.ac.uk
lcgce03.phy.bris.ac.uk
lcgce04.phy.bris.ac.uk
lcgce12.jinr.ru
lcgce1.shef.ac.uk
lcgce21.jinr.ru
lcgce2.shef.ac.uk
lcg-cream.ifh.de
llrcream.in2p3.fr
lpnhe-cream.in2p3.fr
lyogrid07.in2p3.fr
magic.cse.buffalo.edu
mwt2-gk.campuscluster.illinois.edu
ndcms.crc.nd.edu
node01-03.usm.renam.md
node01-04.grid.renam.md
node05-02.imi.renam.md
node74.datagrid.cea.fr
nodeslab-0002.nlab.tb.hiit.fi
ntugrid2.phys.ntu.edu.tw
ntugrid5.phys.ntu.edu.tw
nys1.cac.cornell.edu
osgce.hepgrid.uerj.br
osg-ce.sprace.org.br
osg-gk.mwt2.org
osg-gw-6.t2.ucsd.edu
osg-gw-7.t2.ucsd.edu
osg.hpc.ufl.edu
osg-nemo-ce.phys.uwm.edu
osg.rcac.purdue.edu
osgserv01.slac.stanford.edu
osgserv02.slac.stanford.edu
ouhep0.nhn.ou.edu
pamelace01.na.infn.it
pcncp04.ncp.edu.pk
pcncp05.ncp.edu.pk
pre7230.datagrid.cea.fr
prod-ce-01.pd.infn.it
razi.biruni.upm.my
recasce01.na.infn.it
red-gw1.unl.edu
red-gw2.unl.edu
red.unl.edu
rossmann-osg.rcac.purdue.edu
sbgce2.in2p3.fr
snf-189278.vm.okeanos.grnet.gr
snf-458754.vm.okeanos.grnet.gr
spacina-ce.scope.unina.it
svr009.gla.scotgrid.ac.uk
svr010.gla.scotgrid.ac.uk
svr011.gla.scotgrid.ac.uk
svr014.gla.scotgrid.ac.uk
t2arc01.physics.ox.ac.uk
t2-ce-01.lnl.infn.it
t2-ce-01.to.infn.it
t2-ce-02.lnl.infn.it
t2ce02.physics.ox.ac.uk
t2-ce-03.lnl.infn.it
t2-ce-04.lnl.infn.it
t2-ce-04.mi.infn.it
t2ce04.physics.ox.ac.uk
t2-ce-05.mi.infn.it
t2-ce-06.lnl.infn.it
t2ce06.physics.ox.ac.uk
t3serv007.mit.edu
tau-cream.hep.tau.ac.il
tech-crm.hep.technion.ac.il
top.ucr.edu
umiss001.hep.olemiss.edu
uosaf0008.sscc.uos.ac.kr
uscms1.fltech-grid3.fit.edu
v6ce00.grid.hep.ph.ic.ac.uk
vserv13.hep.phy.cam.ac.uk
wipp-crm.weizmann.ac.il
"""
import sys, time
from python_compat import set, imap, lfilter, sorted
counter = 0
used = set()
for line in imap(str.strip, ce_list_str.splitlines()):
time.sleep(0.2)
match = get_geo_match(line)
if not match:
counter += 1
sys.stderr.write('\t%r: %r\n' % (line, self._geocode(line)))
else:
used.add(match)
sys.stderr.write('%s unmatched entries\n' % counter)
sys.stderr.write('unused entries:\n%s\n' %
repr(lfilter(lambda x: x not in used, _GEO_DICT)))
sys.stdout.write('_GEO_DICT = {\n')
geo_dict_key_list = sorted(
_GEO_DICT.keys(),
key=lambda x: str.join('.', reversed(x.split('.'))))
for entry in geo_dict_key_list:
sys.stdout.write('\t%r: (%.6f, %.6f),\n' %
(entry, _GEO_DICT[entry][0], _GEO_DICT[entry][1]))
sys.stdout.write('}\n')
def _geocode(self, loc):
result = self._jrc.get(params={
'address': str.join('.',
loc.split('.')[2:]),
'sensor': 'false'
})
if 'Placemark' in result: # unfold placemake entries
place_list = []
for entry in result['Placemark']:
place_list.append(
(entry['address'],
tuple(reversed(entry['Point']['coordinates'][:2]))))
return place_list
return result
class GeoResolver(object):
def __init__(self):
from grid_control.utils.webservice import JSONRestClient
self._jrc = JSONRestClient(url='http://maps.googleapis.com/maps/api/geocode/json')
def run(self):
# output of lcg-infosites ce | while read X X X X X CE; do echo $CE; done
# | cut -d "/" -f 1 | cut -d ":" -f 1 | sort | uniq
ce_list_str = """alcyone-cms.grid.helsinki.fi
alice23.spbu.ru
arc-ce01.gridpp.rl.ac.uk
arc-ce02.gridpp.rl.ac.uk
arc-ce03.gridpp.rl.ac.uk
argoce01.na.infn.it
atlasce1.lnf.infn.it
atlasce2.lnf.infn.it
atlasce3.lnf.infn.it
atlas-cream01.na.infn.it
atlas-cream02.na.infn.it
boce.bo.infn.it
bonner-grid.rice.edu
brux3.hep.brown.edu
cale.uniandes.edu.co
carter-osg.rcac.purdue.edu
cccreamceli07.in2p3.fr
cccreamceli08.in2p3.fr
cce.ihep.ac.cn
ce0002.m45.ihep.su
ce0004.m45.ihep.su
ce01.cmsaf.mit.edu
ce01.jinr-t1.ru
ce01-lcg.cr.cnaf.infn.it
ce-01.roma3.infn.it
ce01.tier2.hep.manchester.ac.uk
ce02.cmsaf.mit.edu
ce02.jinr-t1.ru
ce02.ngcc.acad.bg
ce02.tier2.hep.manchester.ac.uk
ce04-lcg.cr.cnaf.infn.it
ce05.esc.qmul.ac.uk
ce05-lcg.cr.cnaf.infn.it
ce05.ncg.ingrid.pt
ce06.esc.qmul.ac.uk
ce06-lcg.cr.cnaf.infn.it
ce07.esc.qmul.ac.uk
ce07-lcg.cr.cnaf.infn.it
ce08-lcg.cr.cnaf.infn.it
ce101.grid.ucy.ac.cy
ce1.accre.vanderbilt.edu
ce1.dur.scotgrid.ac.uk
ce1.grid.lebedev.ru
ce1.ts.infn.it
ce201.cern.ch
ce202.cern.ch
ce203.cern.ch
ce204.cern.ch
ce205.cern.ch
ce206.cern.ch
ce207.cern.ch
ce208.cern.ch
ce2.accre.vanderbilt.edu
ce2.particles.ipm.ac.ir
ce301.cern.ch
ce302.cern.ch
ce3.ppgrid1.rhul.ac.uk
ce401.cern.ch
ce402.cern.ch
ce403.cern.ch
ce404.cern.ch
ce405.cern.ch
ce406.cern.ch
ce407.cern.ch
ce408.cern.ch
ce64.ipb.ac.rs
ce6.grid.icm.edu.pl
ce7.glite.ecdf.ed.ac.uk
ce9.grid.icm.edu.pl
cebo-t3-01.cr.cnaf.infn.it
cebo-t3-02.cr.cnaf.infn.it
ce.cis.gov.pl
cecream.ca.infn.it
ce.fesb.egi.cro-ngi.hr
ce.grid.unesp.br
ce.irb.egi.cro-ngi.hr
ceprod05.grid.hep.ph.ic.ac.uk
ceprod06.grid.hep.ph.ic.ac.uk
ceprod07.grid.hep.ph.ic.ac.uk
ceprod08.grid.hep.ph.ic.ac.uk
cert-37.pd.infn.it
ce.scope.unina.it
ce.srce.egi.cro-ngi.hr
cetest01.grid.hep.ph.ic.ac.uk
cetest02.grid.hep.ph.ic.ac.uk
ce.ulakbim.gov.tr
cit-gatekeeper2.ultralight.org
cit-gatekeeper.ultralight.org
cluster118.knu.ac.kr
cluster50.knu.ac.kr
cms-0.mps.ohio-state.edu
cmsce01.na.infn.it
cmsgrid01.hep.wisc.edu
cmsgrid02.hep.wisc.edu
cms-grid0.hep.uprm.edu
cmsosgce2.fnal.gov
cmsosgce4.fnal.gov
cmsosgce.fnal.gov
cmsrm-cream01.roma1.infn.it
cmsrm-cream02.roma1.infn.it
cmsrm-cream03.roma1.infn.it
cmstest1.rcac.purdue.edu
cms.tier3.ucdavis.edu
conte-osg.rcac.purdue.edu
cox01.grid.metu.edu.tr
cr1.ipp.acad.bg
cream01.grid.auth.gr
cream01.grid.sinica.edu.tw
cream01.grid.uoi.gr
cream01.kallisto.hellasgrid.gr
cream01.lcg.cscs.ch
cream02.grid.cyf-kr.edu.pl
cream02.iihe.ac.be
cream02.lcg.cscs.ch
cream03.lcg.cscs.ch
cream04.grid.sinica.edu.tw
cream04.lcg.cscs.ch
cream05.grid.sinica.edu.tw
cream2.ppgrid1.rhul.ac.uk
cream3.hep.kbfi.ee
cream4.hep.kbfi.ee
cream.afroditi.hellasgrid.gr
cream-ce01.ariagni.hellasgrid.gr
cream-ce01.indiacms.res.in
cream-ce01.marie.hellasgrid.gr
cream-ce02.cat.cbpf.br
creamce02.ciemat.es
cream-ce02.marie.hellasgrid.gr
creamce03.ciemat.es
creamce1.itep.ru
cream-ce-2.ba.infn.it
cream-ce-4.ba.infn.it
cream-ce.cat.cbpf.br
cream-ce.grid.atomki.hu
creamce.hephy.oeaw.ac.at
creamce.inula.man.poznan.pl
cream-ce.kipt.kharkov.ua
cream-ce.pg.infn.it
creamce.reef.man.poznan.pl
cream.grid.cyf-kr.edu.pl
cream.ipb.ac.rs
dc2-grid-66.brunel.ac.uk
dc2-grid-68.brunel.ac.uk
dc2-grid-70.brunel.ac.uk
dwarf.wcss.wroc.pl
earth.crc.nd.edu
epgr02.ph.bham.ac.uk
erbium.lsr.nectec.or.th
f-cream01.grid.sinica.edu.tw
f-cream04.grid.sinica.edu.tw
fiupg.hep.fiu.edu
foam.grid.kiae.ru
fornax-ce2.itwm.fhg.de
fornax-ce.itwm.fhg.de
grcreamce01.inr.troitsk.ru
grid001.ics.forth.gr
grid002.jet.efda.org
grid012.ct.infn.it
grid01.physics.uoi.gr
grid0.fe.infn.it
grid106.kfki.hu
grid107.kfki.hu
grid109.kfki.hu
grid129.sinp.msu.ru
grid36.lal.in2p3.fr
grid72.phy.ncu.edu.tw
gridce01.ifca.es
gridce03.ifca.es
gridce0.pi.infn.it
gridce1.pi.infn.it
grid-ce2.physik.rwth-aachen.de
gridce2.pi.infn.it
gridce3.pi.infn.it
gridce4.pi.infn.it
gridce.ilc.cnr.it
grid-ce.physik.rwth-aachen.de
grid-cr0.desy.de
grid-cr1.desy.de
grid-cr2.desy.de
grid-cr3.desy.de
grid-cr4.desy.de
gridgk01.racf.bnl.gov
gridgk02.racf.bnl.gov
gridgk03.racf.bnl.gov
gridgk04.racf.bnl.gov
gridgk05.racf.bnl.gov
gridgk06.racf.bnl.gov
gridgk08.racf.bnl.gov
gridtest02.racf.bnl.gov
gridvm03.roma2.infn.it
grisuce.scope.unina.it
gt3.pnpi.nw.ru
hansen-osg.rcac.purdue.edu
hepcms-0.umd.edu
hepgrid10.ph.liv.ac.uk
hepgrid5.ph.liv.ac.uk
hepgrid6.ph.liv.ac.uk
hepgrid97.ph.liv.ac.uk
hephygr.oeaw.ac.at
heposg01.colorado.edu
hurr.tamu.edu
ingrid.cism.ucl.ac.be
jade-cms.hip.fi
juk.nikhef.nl
kalkan1.ulakbim.gov.tr
khaldun.biruni.upm.my
klomp.nikhef.nl
kodiak-ce.baylor.edu
lcg18.sinp.msu.ru
lcg52.sinp.msu.ru
lcgce01.phy.bris.ac.uk
lcgce03.phy.bris.ac.uk
lcgce04.phy.bris.ac.uk
lcgce12.jinr.ru
lcgce1.shef.ac.uk
lcgce21.jinr.ru
lcgce2.shef.ac.uk
lcg-cream.ifh.de
llrcream.in2p3.fr
lpnhe-cream.in2p3.fr
lyogrid07.in2p3.fr
magic.cse.buffalo.edu
mwt2-gk.campuscluster.illinois.edu
ndcms.crc.nd.edu
node01-03.usm.renam.md
node01-04.grid.renam.md
node05-02.imi.renam.md
node74.datagrid.cea.fr
nodeslab-0002.nlab.tb.hiit.fi
ntugrid2.phys.ntu.edu.tw
ntugrid5.phys.ntu.edu.tw
nys1.cac.cornell.edu
osgce.hepgrid.uerj.br
osg-ce.sprace.org.br
osg-gk.mwt2.org
osg-gw-6.t2.ucsd.edu
osg-gw-7.t2.ucsd.edu
osg.hpc.ufl.edu
osg-nemo-ce.phys.uwm.edu
osg.rcac.purdue.edu
osgserv01.slac.stanford.edu
osgserv02.slac.stanford.edu
ouhep0.nhn.ou.edu
pamelace01.na.infn.it
pcncp04.ncp.edu.pk
pcncp05.ncp.edu.pk
pre7230.datagrid.cea.fr
prod-ce-01.pd.infn.it
razi.biruni.upm.my
recasce01.na.infn.it
red-gw1.unl.edu
red-gw2.unl.edu
red.unl.edu
rossmann-osg.rcac.purdue.edu
sbgce2.in2p3.fr
snf-189278.vm.okeanos.grnet.gr
snf-458754.vm.okeanos.grnet.gr
spacina-ce.scope.unina.it
svr009.gla.scotgrid.ac.uk
svr010.gla.scotgrid.ac.uk
svr011.gla.scotgrid.ac.uk
svr014.gla.scotgrid.ac.uk
t2arc01.physics.ox.ac.uk
t2-ce-01.lnl.infn.it
t2-ce-01.to.infn.it
t2-ce-02.lnl.infn.it
t2ce02.physics.ox.ac.uk
t2-ce-03.lnl.infn.it
t2-ce-04.lnl.infn.it
t2-ce-04.mi.infn.it
t2ce04.physics.ox.ac.uk
t2-ce-05.mi.infn.it
t2-ce-06.lnl.infn.it
t2ce06.physics.ox.ac.uk
t3serv007.mit.edu
tau-cream.hep.tau.ac.il
tech-crm.hep.technion.ac.il
top.ucr.edu
umiss001.hep.olemiss.edu
uosaf0008.sscc.uos.ac.kr
uscms1.fltech-grid3.fit.edu
v6ce00.grid.hep.ph.ic.ac.uk
vserv13.hep.phy.cam.ac.uk
wipp-crm.weizmann.ac.il
"""
import sys, time
from python_compat import set, imap, lfilter, sorted
counter = 0
used = set()
for line in imap(str.strip, ce_list_str.splitlines()):
time.sleep(0.2)
match = get_geo_match(line)
if not match:
counter += 1
sys.stderr.write('\t%r: %r\n' % (line, self._geocode(line)))
else:
used.add(match)
sys.stderr.write('%s unmatched entries\n' % counter)
sys.stderr.write('unused entries:\n%s\n' % repr(lfilter(lambda x: x not in used, _GEO_DICT)))
sys.stdout.write('_GEO_DICT = {\n')
geo_dict_key_list = sorted(_GEO_DICT.keys(), key=lambda x: str.join('.', reversed(x.split('.'))))
for entry in geo_dict_key_list:
sys.stdout.write('\t%r: (%.6f, %.6f),\n' % (entry, _GEO_DICT[entry][0], _GEO_DICT[entry][1]))
sys.stdout.write('}\n')
def _geocode(self, loc):
result = self._jrc.get(params={'address': str.join('.', loc.split('.')[2:]), 'sensor': 'false'})
if 'Placemark' in result: # unfold placemake entries
place_list = []
for entry in result['Placemark']:
place_list.append((entry['address'], tuple(reversed(entry['Point']['coordinates'][:2]))))
return place_list
return result
class CMSBaseProvider(DataProvider):
# required format: <dataset path>[@<instance>][#<block>]
def __init__(self, config, datasource_name, dataset_expr, dataset_nick=None, dataset_proc=None):
dataset_config = config.change_view(default_on_change=TriggerResync(['datasets', 'parameters']))
self._lumi_filter = dataset_config.get_lookup(['lumi filter', '%s lumi filter' % datasource_name],
default={}, parser=parse_lumi_filter, strfun=str_lumi)
if not self._lumi_filter.empty():
config.set('%s processor' % datasource_name, 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasource_name, dataset_expr, dataset_nick, dataset_proc)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = dataset_config.get_bool(
['lumi metadata', '%s lumi metadata' % datasource_name], default=not self._lumi_filter.empty())
config.set('phedex sites matcher mode', 'ShellStyleMatcher', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedex_filter = dataset_config.get_filter('phedex sites', '-* T1_*_Disk T2_* T3_*',
default_matcher='BlackWhiteMatcher', default_filter='StrictListFilter')
self._only_complete = dataset_config.get_bool('only complete sites', True)
self._only_valid = dataset_config.get_bool('only valid', True)
self._location_format = dataset_config.get_enum('location format',
CMSLocationFormat, CMSLocationFormat.hostname)
self._pjrc = JSONRestClient(url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
self._sitedb = SiteDB()
dataset_expr_parts = split_opt(dataset_expr, '@#')
(self._dataset_path, self._dataset_instance, self._dataset_block_selector) = dataset_expr_parts
instance_default = dataset_config.get('dbs instance', '')
self._dataset_instance = self._dataset_instance or instance_default
if not self._dataset_instance:
self._dataset_instance = 'prod/global'
elif '/' not in self._dataset_instance:
self._dataset_instance = 'prod/%s' % self._dataset_instance
self._dataset_block_selector = self._dataset_block_selector or 'all'
def check_splitter(self, splitter):
# Check if splitter is valid
if (DataSplitter.Skipped in splitter.get_needed_enums()) and not self._lumi_filter.empty():
self._log.debug('Selected splitter %s is not compatible with active lumi filter!',
splitter.__name__)
self._log.warning('Active lumi section filter forced selection of HybridSplitter')
return HybridSplitter
return splitter
def get_dataset_name_list(self):
if self._cache_dataset is None:
self._cache_dataset = [self._dataset_path]
if '*' in self._dataset_path:
activity = Activity('Getting dataset list for %s' % self._dataset_path)
self._cache_dataset = list(self._get_cms_dataset_list(self._dataset_path))
if not self._cache_dataset:
raise DatasetError('No datasets selected by DBS wildcard %s !' % self._dataset_path)
activity.finish()
return self._cache_dataset
def get_query_interval(self):
# Define how often the dataprovider can be queried automatically
return 2 * 60 * 60 # 2 hour delay minimum
def _fill_cms_fi_list(self, block, block_path):
activity_fi = Activity('Getting file information')
lumi_used = False
lumi_info_dict = {}
if self._lumi_query: # central lumi query
lumi_info_dict = self._get_cms_lumi_dict(block_path)
fi_list = []
for (fi, lumi_info_list) in self._iter_cms_files(block_path, self._only_valid, self._lumi_query):
self._raise_on_abort()
if lumi_info_dict and not lumi_info_list:
lumi_info_list = lumi_info_dict.get(fi[DataProvider.URL], [])
if lumi_info_list:
(run_list_result, lumi_list_result) = ([], [])
for (run, lumi_list) in sorted(lumi_info_list):
run_list_result.extend([run] * len(lumi_list))
lumi_list_result.extend(lumi_list)
assert len(run_list_result) == len(lumi_list_result)
fi[DataProvider.Metadata] = [run_list_result, lumi_list_result]
lumi_used = True
fi_list.append(fi)
if lumi_used:
block.setdefault(DataProvider.Metadata, []).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fi_list
activity_fi.finish()
def _filter_cms_blockinfo_list(self, dataset_path, do_query_sites):
iter_dataset_block_name_selist = self._iter_cms_blocks(dataset_path, do_query_sites)
n_blocks = 0
selected_blocks = False
for (dataset_block_name, selist) in iter_dataset_block_name_selist:
n_blocks += 1
block_name = str.split(dataset_block_name, '#')[1]
if (self._dataset_block_selector != 'all') and (block_name != self._dataset_block_selector):
continue
selected_blocks = True
yield (dataset_block_name, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError('Dataset %r contains %d blocks, but none were selected by %r' % (
dataset_path, n_blocks, self._dataset_block_selector))
def _get_cms_dataset_list(self, dataset_path):
raise AbstractError
def _get_cms_lumi_dict(self, block_path):
return None
def _get_gc_block_list(self, use_phedex):
dataset_name_list = self.get_dataset_name_list()
progress_ds = ProgressActivity('Getting dataset', len(dataset_name_list))
for dataset_idx, dataset_path in enumerate(dataset_name_list):
progress_ds.update_progress(dataset_idx, msg='Getting dataset %s' % dataset_path)
counter = 0
blockinfo_list = list(self._filter_cms_blockinfo_list(dataset_path, not use_phedex))
progress_block = ProgressActivity('Getting block information', len(blockinfo_list))
for (block_path, replica_infos) in blockinfo_list:
result = {}
result[DataProvider.Dataset] = block_path.split('#')[0]
result[DataProvider.BlockName] = block_path.split('#')[1]
progress_block.update_progress(counter,
msg='Getting block information for ' + result[DataProvider.BlockName])
if use_phedex: # Start parallel phedex query
replicas_dict = {}
phedex_thread = start_thread('Query phedex site info for %s' % block_path,
self._get_phedex_replica_list, block_path, replicas_dict)
self._fill_cms_fi_list(result, block_path)
phedex_thread.join()
replica_infos = replicas_dict.get(block_path)
else:
self._fill_cms_fi_list(result, block_path)
result[DataProvider.Locations] = self._process_replica_list(block_path, replica_infos)
if len(result[DataProvider.FileList]):
counter += 1
yield result
progress_block.finish()
if counter == 0:
raise DatasetError('Dataset %s does not contain any valid blocks!' % dataset_path)
progress_ds.finish()
def _get_phedex_replica_list(self, block_path, replicas_dict):
activity_fi = Activity('Getting file replica information from PhEDex')
# Get dataset se list from PhEDex (perhaps concurrent with get_dbs_file_list)
replicas_dict[block_path] = []
for phedex_block in self._pjrc.get(params={'block': block_path})['phedex']['block']:
for replica in phedex_block['replica']:
replica_info = (replica['node'], replica.get('se'), replica['complete'] == 'y')
replicas_dict[block_path].append(replica_info)
activity_fi.finish()
def _iter_cms_blocks(self, dataset_path, do_query_sites):
raise AbstractError
def _iter_cms_files(self, block_path, query_only_valid, query_lumi):
raise AbstractError
def _iter_formatted_locations(self, replica_infos):
for replica_info in replica_infos:
(_, _, completed) = replica_info
if completed:
for entry in self._iter_replica_locations(replica_info):
yield entry
else:
for entry in self._iter_replica_locations(replica_info):
yield '(%s)' % entry
def _iter_replica_locations(self, replica_info):
(name_node, name_hostname, _) = replica_info
if self._location_format == CMSLocationFormat.siteDB:
yield name_node
else:
if name_hostname is not None:
name_hostnames = [name_hostname]
else:
name_hostnames = self._sitedb.cms_name_to_se(name_node)
for name_hostname in name_hostnames:
if self._location_format == CMSLocationFormat.hostname:
yield name_hostname
else:
yield '%s/%s' % (name_node, name_hostname)
def _process_replica_list(self, block_path, replica_infos):
def _empty_with_warning(error_msg, *args):
self._log.warning('Dataset block %r ' + error_msg, block_path, *args)
return []
def _expanded_replica_locations(replica_infos):
for replica_info in replica_infos:
for entry in self._iter_replica_locations(replica_info):
yield entry
if not replica_infos:
return _empty_with_warning('has no replica information!')
replica_infos_selected = self._phedex_filter.filter_list(replica_infos, key=itemgetter(0))
if not replica_infos_selected:
return _empty_with_warning('is not available at the selected locations!\n' +
'Available locations: %s', str.join(', ', self._iter_formatted_locations(replica_infos)))
if not self._only_complete:
return list(_expanded_replica_locations(replica_infos_selected))
replica_infos_complete = lfilter(lambda nn_nh_c: nn_nh_c[2], replica_infos_selected)
if not replica_infos_complete:
return _empty_with_warning('is not completely available at the selected locations!\n' +
'Available locations: %s', str.join(', ', self._iter_formatted_locations(replica_infos)))
return list(_expanded_replica_locations(replica_infos_complete))
def __init__(self): from grid_control.utils.webservice import JSONRestClient self._jrc = JSONRestClient(url='http://maps.googleapis.com/maps/api/geocode/json')
class CMSBaseProvider(DataProvider):
def __init__(self, config, datasetExpr, datasetNick=None, datasetID=0):
changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {},
parser=parseLumiFilter,
strfun=strLumi,
onChange=changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick,
datasetID)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata',
not self._lumi_filter.empty(),
onChange=changeTrigger)
# PhEDex blacklist: 'T1_DE_KIT', 'T1_US_FNAL' and '*_Disk' allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites',
'-T3_US_FNALLPC',
defaultMatcher='blackwhite',
defaultFilter='weak',
onChange=changeTrigger)
self._phedexT1Filter = config.getFilter('phedex t1 accept',
'T1_DE_KIT T1_US_FNAL',
defaultMatcher='blackwhite',
defaultFilter='weak',
onChange=changeTrigger)
self._phedexT1Mode = config.getEnum('phedex t1 mode',
PhedexT1Mode,
PhedexT1Mode.disk,
onChange=changeTrigger)
self.onlyComplete = config.getBool('only complete sites',
True,
onChange=changeTrigger)
self._locationFormat = config.getEnum('location format',
CMSLocationFormat,
CMSLocationFormat.hostname,
onChange=changeTrigger)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
(self._datasetPath, self._url,
self._datasetBlock) = optSplit(datasetExpr, '@#')
self._url = self._url or config.get('dbs instance', '')
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid',
True,
onChange=changeTrigger)
# Define how often the dataprovider can be queried automatically
def queryLimit(self):
return 2 * 60 * 60 # 2 hour delay minimum
# Check if splitterClass is valid
def checkSplitter(self, splitterClass):
if (DataSplitter.Skipped in splitterClass.neededEnums()
) and not self._lumi_filter.empty():
self._log.debug(
'Selected splitter %s is not compatible with active lumi filter!',
splitterClass.__name__)
self._log.warning(
'Active lumi section filter forced selection of HybridSplitter'
)
return HybridSplitter
return splitterClass
def _nodeFilter(self, nameSiteDB, complete):
# Remove T0 and T1 by default
result = not (nameSiteDB.startswith('T0_')
or nameSiteDB.startswith('T1_'))
# check if listed on the accepted list
if self._phedexT1Mode in [PhedexT1Mode.disk, PhedexT1Mode.accept]:
result = result or (self._phedexT1Filter.filterList([nameSiteDB])
== [nameSiteDB])
if self._phedexT1Mode == PhedexT1Mode.disk:
result = result or nameSiteDB.lower().endswith('_disk')
# apply phedex blacklist
result = result and (self._phedexFilter.filterList([nameSiteDB])
== [nameSiteDB])
# check for completeness at the site
result = result and (complete or not self.onlyComplete)
return result
# Get dataset se list from PhEDex (perhaps concurrent with listFiles)
def _getPhedexSEList(self, blockPath, dictSE):
dictSE[blockPath] = []
for phedexBlock in self._pjrc.get(
params={'block': blockPath})['phedex']['block']:
for replica in phedexBlock['replica']:
if self._nodeFilter(replica['node'],
replica['complete'] == 'y'):
location = None
if self._locationFormat == CMSLocationFormat.hostname:
location = replica.get('se')
elif self._locationFormat == CMSLocationFormat.siteDB:
location = replica.get('node')
elif (self._locationFormat == CMSLocationFormat.both) and (
replica.get('node') or replica.get('se')):
location = '%s/%s' % (replica.get('node'),
replica.get('se'))
if location:
dictSE[blockPath].append(location)
else:
self._log.warning(
'Dataset block %s replica at %s / %s is skipped!',
blockPath, replica.get('node'), replica.get('se'))
def getDatasets(self):
if self._cache_dataset is None:
self._cache_dataset = [self._datasetPath]
if '*' in self._datasetPath:
self._cache_dataset = list(
self.getCMSDatasets(self._datasetPath))
if not self._cache_dataset:
raise DatasetError(
'No datasets selected by DBS wildcard %s !' %
self._datasetPath)
return self._cache_dataset
def getCMSBlocks(self, datasetPath, getSites):
iter_blockname_selist = self.getCMSBlocksImpl(datasetPath, getSites)
n_blocks = 0
selected_blocks = False
for (blockname, selist) in iter_blockname_selist:
n_blocks += 1
if (self._datasetBlock != 'all') and (str.split(blockname, '#')[1]
!= self._datasetBlock):
continue
selected_blocks = True
yield (blockname, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError(
'Dataset %r contains %d blocks, but none were selected by %r' %
(datasetPath, n_blocks, self._datasetBlock))
def fillCMSFiles(self, block, blockPath):
lumi_used = False
lumiDict = {}
if self._lumi_query: # central lumi query
lumiDict = self.getCMSLumisImpl(blockPath)
fileList = []
for (fileInfo,
listLumi) in self.getCMSFilesImpl(blockPath, self.onlyValid,
self._lumi_query):
if lumiDict and not listLumi:
listLumi = lumiDict.get(fileInfo[DataProvider.URL], [])
if listLumi:
(listLumiExt_Run, listLumiExt_Lumi) = ([], [])
for (run, lumi_list) in sorted(listLumi):
listLumiExt_Run.extend([run] * len(lumi_list))
listLumiExt_Lumi.extend(lumi_list)
fileInfo[DataProvider.Metadata] = [
listLumiExt_Run, listLumiExt_Lumi
]
lumi_used = True
fileList.append(fileInfo)
if lumi_used:
block.setdefault(DataProvider.Metadata,
[]).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fileList
def getCMSLumisImpl(self, blockPath):
return None
def getGCBlocks(self, usePhedex):
for datasetPath in self.getDatasets():
counter = 0
for (blockPath,
listSE) in self.getCMSBlocks(datasetPath,
getSites=not usePhedex):
result = {}
result[DataProvider.Dataset] = blockPath.split('#')[0]
result[DataProvider.BlockName] = blockPath.split('#')[1]
if usePhedex: # Start parallel phedex query
dictSE = {}
tPhedex = start_thread(
'Query phedex site info for %s' % blockPath,
self._getPhedexSEList, blockPath, dictSE)
self.fillCMSFiles(result, blockPath)
tPhedex.join()
listSE = dictSE.get(blockPath)
else:
self.fillCMSFiles(result, blockPath)
result[DataProvider.Locations] = listSE
if len(result[DataProvider.FileList]):
counter += 1
yield result
if counter == 0:
raise DatasetError(
'Dataset %s does not contain any valid blocks!' %
datasetPath)
class CMSBaseProvider(DataProvider):
def __init__(self, config, datasetExpr, datasetNick=None):
self._changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {},
parser=parseLumiFilter,
strfun=strLumi,
onChange=self._changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata',
not self._lumi_filter.empty(),
onChange=self._changeTrigger)
config.set('phedex sites matcher mode', 'shell', '?=')
# PhEDex blacklist: 'T1_*_Disk nodes allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites',
'-* T1_*_Disk T2_* T3_*',
defaultMatcher='blackwhite',
defaultFilter='strict',
onChange=self._changeTrigger)
self._onlyComplete = config.getBool('only complete sites',
True,
onChange=self._changeTrigger)
self._locationFormat = config.getEnum('location format',
CMSLocationFormat,
CMSLocationFormat.hostname,
onChange=self._changeTrigger)
self._pjrc = JSONRestClient(
url='https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas'
)
self._sitedb = SiteDB()
(self._datasetPath, self._datasetInstance,
self._datasetBlock) = optSplit(datasetExpr, '@#')
instance_default = config.get('dbs instance',
'',
onChange=self._changeTrigger)
self._datasetInstance = self._datasetInstance or instance_default
if not self._datasetInstance:
self._datasetInstance = 'prod/global'
elif '/' not in self._datasetInstance:
self._datasetInstance = 'prod/%s' % self._datasetInstance
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid',
True,
onChange=self._changeTrigger)
# Define how often the dataprovider can be queried automatically
def queryLimit(self):
return 2 * 60 * 60 # 2 hour delay minimum
# Check if splitterClass is valid
def checkSplitter(self, splitterClass):
if (DataSplitter.Skipped in splitterClass.neededEnums()
) and not self._lumi_filter.empty():
self._log.debug(
'Selected splitter %s is not compatible with active lumi filter!',
splitterClass.__name__)
self._log.warning(
'Active lumi section filter forced selection of HybridSplitter'
)
return HybridSplitter
return splitterClass
def _replicaLocation(self, replica_info):
(name_node, name_hostname, _) = replica_info
if self._locationFormat == CMSLocationFormat.siteDB:
yield name_node
else:
if name_hostname is not None:
name_hostnames = [name_hostname]
else:
name_hostnames = self._sitedb.cms_name_to_se(name_node)
for name_hostname in name_hostnames:
if self._locationFormat == CMSLocationFormat.hostname:
yield name_hostname
else:
yield '%s/%s' % (name_node, name_hostname)
def _fmtLocations(self, replica_infos):
for replica_info in replica_infos:
(_, _, completed) = replica_info
if completed:
for entry in self._replicaLocation(replica_info):
yield entry
else:
for entry in self._replicaLocation(replica_info):
yield '(%s)' % entry
def _processReplicas(self, blockPath, replica_infos):
def empty_with_warning(*args):
self._log.warning(*args)
return []
def expanded_replica_locations(replica_infos):
for replica_info in replica_infos:
for entry in self._replicaLocation(replica_info):
yield entry
if not replica_infos:
return empty_with_warning(
'Dataset block %r has no replica information!', blockPath)
replica_infos_selected = self._phedexFilter.filterList(
replica_infos, key=itemgetter(0))
if not replica_infos_selected:
return empty_with_warning(
'Dataset block %r is not available at the selected locations!\nAvailable locations: %s',
blockPath, str.join(', ', self._fmtLocations(replica_infos)))
if not self._onlyComplete:
return list(expanded_replica_locations(replica_infos_selected))
replica_infos_complete = lfilter(lambda nn_nh_c: nn_nh_c[2],
replica_infos_selected)
if not replica_infos_complete:
return empty_with_warning(
'Dataset block %r is not completely available at the selected locations!\nAvailable locations: %s',
blockPath, str.join(', ', self._fmtLocations(replica_infos)))
return list(expanded_replica_locations(replica_infos_complete))
# Get dataset se list from PhEDex (perhaps concurrent with listFiles)
def _getPhedexReplicas(self, blockPath, dictReplicas):
dictReplicas[blockPath] = []
for phedexBlock in self._pjrc.get(
params={'block': blockPath})['phedex']['block']:
for replica in phedexBlock['replica']:
dictReplicas[blockPath].append(
(replica['node'], replica.get('se'),
replica['complete'] == 'y'))
def getDatasets(self):
if self._cache_dataset is None:
self._cache_dataset = [self._datasetPath]
if '*' in self._datasetPath:
self._cache_dataset = list(
self._getCMSDatasets(self._datasetPath))
if not self._cache_dataset:
raise DatasetError(
'No datasets selected by DBS wildcard %s !' %
self._datasetPath)
return self._cache_dataset
def _getCMSBlocks(self, datasetPath, getSites):
iter_blockname_selist = self._getCMSBlocksImpl(datasetPath, getSites)
n_blocks = 0
selected_blocks = False
for (blockname, selist) in iter_blockname_selist:
n_blocks += 1
if (self._datasetBlock != 'all') and (str.split(blockname, '#')[1]
!= self._datasetBlock):
continue
selected_blocks = True
yield (blockname, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError(
'Dataset %r contains %d blocks, but none were selected by %r' %
(datasetPath, n_blocks, self._datasetBlock))
def _fillCMSFiles(self, block, blockPath):
lumi_used = False
lumiDict = {}
if self._lumi_query: # central lumi query
lumiDict = self._getCMSLumisImpl(blockPath)
fileList = []
for (fileInfo,
listLumi) in self._getCMSFilesImpl(blockPath, self.onlyValid,
self._lumi_query):
if lumiDict and not listLumi:
listLumi = lumiDict.get(fileInfo[DataProvider.URL], [])
if listLumi:
(listLumiExt_Run, listLumiExt_Lumi) = ([], [])
for (run, lumi_list) in sorted(listLumi):
listLumiExt_Run.extend([run] * len(lumi_list))
listLumiExt_Lumi.extend(lumi_list)
fileInfo[DataProvider.Metadata] = [
listLumiExt_Run, listLumiExt_Lumi
]
lumi_used = True
fileList.append(fileInfo)
if lumi_used:
block.setdefault(DataProvider.Metadata,
[]).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fileList
def _getCMSLumisImpl(self, blockPath):
return None
def _getGCBlocks(self, usePhedex):
for datasetPath in self.getDatasets():
counter = 0
for (blockPath,
replica_infos) in self._getCMSBlocks(datasetPath,
getSites=not usePhedex):
result = {}
result[DataProvider.Dataset] = blockPath.split('#')[0]
result[DataProvider.BlockName] = blockPath.split('#')[1]
if usePhedex: # Start parallel phedex query
dictReplicas = {}
tPhedex = start_thread(
'Query phedex site info for %s' % blockPath,
self._getPhedexReplicas, blockPath, dictReplicas)
self._fillCMSFiles(result, blockPath)
tPhedex.join()
replica_infos = dictReplicas.get(blockPath)
else:
self._fillCMSFiles(result, blockPath)
result[DataProvider.Locations] = self._processReplicas(
blockPath, replica_infos)
if len(result[DataProvider.FileList]):
counter += 1
yield result
if counter == 0:
raise DatasetError(
'Dataset %s does not contain any valid blocks!' %
datasetPath)
class CMSBaseProvider(DataProvider):
def __init__(self, config, datasetExpr, datasetNick = None, datasetID = 0):
changeTrigger = triggerResync(['datasets', 'parameters'])
self._lumi_filter = config.getLookup('lumi filter', {}, parser = parseLumiFilter, strfun = strLumi, onChange = changeTrigger)
if not self._lumi_filter.empty():
config.set('dataset processor', 'LumiDataProcessor', '+=')
DataProvider.__init__(self, config, datasetExpr, datasetNick, datasetID)
# LumiDataProcessor instantiated in DataProcessor.__ini__ will set lumi metadata as well
self._lumi_query = config.getBool('lumi metadata', not self._lumi_filter.empty(), onChange = changeTrigger)
# PhEDex blacklist: 'T1_DE_KIT', 'T1_US_FNAL' and '*_Disk' allow user jobs - other T1's dont!
self._phedexFilter = config.getFilter('phedex sites', '-T3_US_FNALLPC',
defaultMatcher = 'blackwhite', defaultFilter = 'weak', onChange = changeTrigger)
self._phedexT1Filter = config.getFilter('phedex t1 accept', 'T1_DE_KIT T1_US_FNAL',
defaultMatcher = 'blackwhite', defaultFilter = 'weak', onChange = changeTrigger)
self._phedexT1Mode = config.getEnum('phedex t1 mode', PhedexT1Mode, PhedexT1Mode.disk, onChange = changeTrigger)
self.onlyComplete = config.getBool('only complete sites', True, onChange = changeTrigger)
self._locationFormat = config.getEnum('location format', CMSLocationFormat, CMSLocationFormat.hostname, onChange = changeTrigger)
self._pjrc = JSONRestClient(url = 'https://cmsweb.cern.ch/phedex/datasvc/json/prod/blockreplicas')
(self._datasetPath, self._url, self._datasetBlock) = optSplit(datasetExpr, '@#')
self._url = self._url or config.get('dbs instance', '')
self._datasetBlock = self._datasetBlock or 'all'
self.onlyValid = config.getBool('only valid', True, onChange = changeTrigger)
# Define how often the dataprovider can be queried automatically
def queryLimit(self):
return 2 * 60 * 60 # 2 hour delay minimum
# Check if splitterClass is valid
def checkSplitter(self, splitterClass):
if (DataSplitter.Skipped in splitterClass.neededEnums()) and not self._lumi_filter.empty():
self._log.debug('Selected splitter %s is not compatible with active lumi filter!', splitterClass.__name__)
self._log.warning('Active lumi section filter forced selection of HybridSplitter')
return HybridSplitter
return splitterClass
def _nodeFilter(self, nameSiteDB, complete):
# Remove T0 and T1 by default
result = not (nameSiteDB.startswith('T0_') or nameSiteDB.startswith('T1_'))
# check if listed on the accepted list
if self._phedexT1Mode in [PhedexT1Mode.disk, PhedexT1Mode.accept]:
result = result or (self._phedexT1Filter.filterList([nameSiteDB]) == [nameSiteDB])
if self._phedexT1Mode == PhedexT1Mode.disk:
result = result or nameSiteDB.lower().endswith('_disk')
# apply phedex blacklist
result = result and (self._phedexFilter.filterList([nameSiteDB]) == [nameSiteDB])
# check for completeness at the site
result = result and (complete or not self.onlyComplete)
return result
# Get dataset se list from PhEDex (perhaps concurrent with listFiles)
def _getPhedexSEList(self, blockPath, dictSE):
dictSE[blockPath] = []
for phedexBlock in self._pjrc.get(params = {'block': blockPath})['phedex']['block']:
for replica in phedexBlock['replica']:
if self._nodeFilter(replica['node'], replica['complete'] == 'y'):
location = None
if self._locationFormat == CMSLocationFormat.hostname:
location = replica.get('se')
elif self._locationFormat == CMSLocationFormat.siteDB:
location = replica.get('node')
elif (self._locationFormat == CMSLocationFormat.both) and (replica.get('node') or replica.get('se')):
location = '%s/%s' % (replica.get('node'), replica.get('se'))
if location:
dictSE[blockPath].append(location)
else:
self._log.warning('Dataset block %s replica at %s / %s is skipped!',
blockPath, replica.get('node'), replica.get('se'))
def getDatasets(self):
if self._cache_dataset is None:
self._cache_dataset = [self._datasetPath]
if '*' in self._datasetPath:
self._cache_dataset = list(self.getCMSDatasets(self._datasetPath))
if not self._cache_dataset:
raise DatasetError('No datasets selected by DBS wildcard %s !' % self._datasetPath)
return self._cache_dataset
def getCMSBlocks(self, datasetPath, getSites):
iter_blockname_selist = self.getCMSBlocksImpl(datasetPath, getSites)
n_blocks = 0
selected_blocks = False
for (blockname, selist) in iter_blockname_selist:
n_blocks += 1
if (self._datasetBlock != 'all') and (str.split(blockname, '#')[1] != self._datasetBlock):
continue
selected_blocks = True
yield (blockname, selist)
if (n_blocks > 0) and not selected_blocks:
raise DatasetError('Dataset %r contains %d blocks, but none were selected by %r' % (datasetPath, n_blocks, self._datasetBlock))
def fillCMSFiles(self, block, blockPath):
lumi_used = False
lumiDict = {}
if self._lumi_query: # central lumi query
lumiDict = self.getCMSLumisImpl(blockPath)
fileList = []
for (fileInfo, listLumi) in self.getCMSFilesImpl(blockPath, self.onlyValid, self._lumi_query):
if lumiDict and not listLumi:
listLumi = lumiDict.get(fileInfo[DataProvider.URL], [])
if listLumi:
(listLumiExt_Run, listLumiExt_Lumi) = ([], [])
for (run, lumi_list) in sorted(listLumi):
listLumiExt_Run.extend([run] * len(lumi_list))
listLumiExt_Lumi.extend(lumi_list)
fileInfo[DataProvider.Metadata] = [listLumiExt_Run, listLumiExt_Lumi]
lumi_used = True
fileList.append(fileInfo)
if lumi_used:
block.setdefault(DataProvider.Metadata, []).extend(['Runs', 'Lumi'])
block[DataProvider.FileList] = fileList
def getCMSLumisImpl(self, blockPath):
return None
def getGCBlocks(self, usePhedex):
for datasetPath in self.getDatasets():
counter = 0
for (blockPath, listSE) in self.getCMSBlocks(datasetPath, getSites = not usePhedex):
result = {}
result[DataProvider.Dataset] = blockPath.split('#')[0]
result[DataProvider.BlockName] = blockPath.split('#')[1]
if usePhedex: # Start parallel phedex query
dictSE = {}
tPhedex = start_thread('Query phedex site info for %s' % blockPath, self._getPhedexSEList, blockPath, dictSE)
self.fillCMSFiles(result, blockPath)
tPhedex.join()
listSE = dictSE.get(blockPath)
else:
self.fillCMSFiles(result, blockPath)
result[DataProvider.Locations] = listSE
if len(result[DataProvider.FileList]):
counter += 1
yield result
if counter == 0:
raise DatasetError('Dataset %s does not contain any valid blocks!' % datasetPath)
