def from_prod_id(prod_ids, file_type):
if not isinstance(prod_ids, list):
prod_ids = [prod_ids]
from LHCbDIRAC.BookkeepingSystem.Client.BookkeepingClient import BookkeepingClient
files = []
for prod_id in prod_ids:
bk_client = BookkeepingClient()
result = bk_client.getProductionFiles(prod_id, file_type)
if not result['OK']:
print('ERROR getting the files', result['Message'], file=sys.stderr)
sys.exit(1)
files.extend([DiracFile(lfn) for lfn in result['Value']])
if not files:
raise ValueError('No files found for BK query:', prod_id)
print(len(files), 'files found')
return files
Script.parseCommandLine()
args = Script.getPositionalArgs()
if not len(args) == 2:
Script.showHelp()
try:
prodID = int(args[0])
except:
Script.showHelp()
filetype = args[1]
from LHCbDIRAC.BookkeepingSystem.Client.BookkeepingClient import BookkeepingClient
client = BookkeepingClient()
res = client.getProductionFiles(prodID, filetype)
if not res['OK']:
print 'ERROR: Failed to retrieve production files: %s' % res['Message']
else:
if not res['Value']:
print 'No files found for production %s with type %s' % (prodID,
filetype)
else:
print '%s %s %s %s %s' % ('FileName'.ljust(100), 'Size'.ljust(10),
'GUID'.ljust(40), 'Replica'.ljust(8),
'Visible'.ljust(8))
for lfn in sorted(res['Value']):
size = res['Value'][lfn]['FileSize']
guid = res['Value'][lfn]['GUID']
hasReplica = res['Value'][lfn]['GotReplica']
visible = res['Value'][lfn]['Visible']
class TransformationCleaningAgent(DiracTCAgent):
""" .. class:: TransformationCleaningAgent
"""
def __init__(self, *args, **kwargs):
""" c'tor
"""
DiracTCAgent.__init__(self, *args, **kwargs)
self.directoryLocations = ['TransformationDB', 'StorageUsage']
self.archiveAfter = 7
self.fileTypesToKeep = ['GAUSSHIST']
self.bkClient = None
self.transClient = None
self.storageUsageClient = None
#############################################################################
def initialize(self):
""" Standard initialize method for agents
"""
DiracTCAgent.initialize(self)
self.directoryLocations = sorted(
self.am_getOption('DirectoryLocations', self.directoryLocations))
self.archiveAfter = self.am_getOption('ArchiveAfter',
self.archiveAfter) # days
self.fileTypesToKeep = Operations().getValue(
'Transformations/FileTypesToKeep', self.fileTypesToKeep)
self.bkClient = BookkeepingClient()
self.transClient = TransformationClient()
self.storageUsageClient = StorageUsageClient()
return S_OK()
def cleanMetadataCatalogFiles(self, transID):
""" clean the metadata using BKK and Data Manager. This method is a replacement of the one from base class
:param self: self reference
:param int transID: transformation ID
"""
res = self.bkClient.getProductionFiles(transID, 'ALL', 'Yes')
if not res['OK']:
return res
bkMetadata = res['Value']
fileToRemove = []
yesReplica = []
self.log.info(
"Found a total of %d files in the BK for transformation %d" %
(len(bkMetadata), transID))
for lfn, metadata in bkMetadata.iteritems():
if metadata['FileType'] != 'LOG':
fileToRemove.append(lfn)
if metadata['GotReplica'] == 'Yes':
yesReplica.append(lfn)
if fileToRemove:
self.log.info(
"Attempting to remove %d possible remnants from the catalog and storage"
% len(fileToRemove))
# Executing with shifter proxy
gConfigurationData.setOptionInCFG(
'/DIRAC/Security/UseServerCertificate', 'false')
res = DataManager().removeFile(fileToRemove, force=True)
gConfigurationData.setOptionInCFG(
'/DIRAC/Security/UseServerCertificate', 'true')
if not res['OK']:
return res
for lfn, reason in res['Value']['Failed'].iteritems():
self.log.error("Failed to remove file found in BK",
"%s %s" % (lfn, reason))
if res['Value']['Failed']:
return S_ERROR("Failed to remove all files found in the BK")
if yesReplica:
self.log.info(
"Ensuring that %d files are removed from the BK" %
(len(yesReplica)))
res = FileCatalog(
catalogs=['BookkeepingDB']).removeFile(yesReplica)
if not res['OK']:
return res
for lfn, reason in res['Value']['Failed'].iteritems():
self.log.error("Failed to remove file from BK",
"%s %s" % (lfn, reason))
if res['Value']['Failed']:
return S_ERROR("Failed to remove all files from the BK")
self.log.info("Successfully removed all files found in the BK")
return S_OK()
def getTransformationDirectories(self, transID):
""" get the directories for the supplied transformation from the transformation system
:param self: self reference
:param int transID: transformation ID
"""
res = DiracTCAgent.getTransformationDirectories(self, transID)
if not res['OK']:
return res
directories = res['Value']
if isinstance(directories, basestring): # Check for (stupid) formats
directories = ast.literal_eval(directories)
if not isinstance(directories, list):
return S_ERROR("Wrong format of output directories")
if 'StorageUsage' in self.directoryLocations:
res = self.storageUsageClient.getStorageDirectories(
'', '', transID, [])
if not res['OK']:
self.log.error("Failed to obtain storage usage directories",
res['Message'])
return res
transDirectories = res['Value']
directories = self._addDirs(transID, transDirectories, directories)
if not directories:
self.log.info("No output directories found")
# We should be removing from the list of directories
# those directories created for file types that are part of those:
# - uploaded (as output files)
# - not merged by subsequent steps
# but this is pretty difficult to identify at run time, so we better remove the "RemovingFiles" production status
# and replace it with a flush (this applies only to MC).
# So we just have a created list.
fileTypesToKeepDirs = []
for fileTypeToKeep in self.fileTypesToKeep:
fileTypesToKeepDirs.extend(
[x for x in directories if fileTypeToKeep in x])
directories = list(
set(directories).difference(set(fileTypesToKeepDirs)))
directories = sorted(directories)
return S_OK(directories)
