def findGenericPilotCredentials( vo = False, group = False ):
if not group and not vo:
return S_ERROR( "Need a group or a VO to determine the Generic pilot credentials" )
if not vo:
vo = Registry.getVOForGroup( group )
if not vo:
return S_ERROR( "Group %s does not have a VO associated" % group )
opsHelper = Operations.Operations( vo = vo )
pilotGroup = opsHelper.getValue( "Pilot/GenericPilotGroup", "" )
pilotDN = opsHelper.getValue( "Pilot/GenericPilotDN", "" )
if pilotDN and pilotGroup:
gLogger.verbose( "Pilot credentials have been defined in the CS. Using %s@%s" % ( pilotDN, pilotGroup ) )
result = gProxyManager.userHasProxy( pilotDN, pilotGroup, 86400 )
if not result[ 'OK' ]:
return S_ERROR( "%s@%s has no proxy uploaded to the ProxyManager" )
return S_OK( ( pilotDN, pilotGroup ) )
#Auto discover
gLogger.verbose( "Pilot credentials are not defined. Autodiscovering..." )
if pilotGroup:
pilotGroups = [ pilotGroup ]
else:
result = Registry.getGroupsWithProperty( Properties.GENERIC_PILOT )
if not result[ 'OK' ]:
return result
pilotGroups = []
groups = result[ 'Value' ]
if not groups:
return S_ERROR( "No group with %s property defined" % Properties.GENERIC_PILOT )
result = Registry.getGroupsForVO( vo )
if not result[ 'OK' ]:
return result
for voGroup in result[ 'Value' ]:
if voGroup in groups:
pilotGroups.append( voGroup )
if not pilotGroups:
return S_ERROR( "No group for VO %s is a generic pilot group" % vo )
for pilotGroup in pilotGroups:
DNs = Registry.getDNsInGroup( pilotGroup )
if not DNs:
continue
if pilotDN:
if pilotDN not in DNs:
continue
result = gProxyManager.userHasProxy( pilotDN, pilotGroup, 86400 )
if result[ 'OK' ] and result[ 'Value' ]:
gLogger.verbose( "Discovered pilot credentials are %s@%s" % ( pilotDN, pilotGroup ) )
return S_OK( ( pilotDN, pilotGroup ) )
else:
for DN in DNs:
result = gProxyManager.userHasProxy( DN, pilotGroup, 86400 )
if result[ 'OK' ] and result[ 'Value' ]:
gLogger.verbose( "Discovered pilot credentials are %s@%s" % ( DN, pilotGroup ) )
return S_OK( ( DN, pilotGroup ) )
if pilotDN:
return S_ERROR( "DN %s does not have group %s" % ( pilotDN, pilotGroups ) )
return S_ERROR( "No generic proxy in the Proxy Manager with groups %s" % pilotGroups )
def findGenericPilotCredentials( vo = False, group = False ):
if not group and not vo:
return S_ERROR( "Need a group or a VO to determine the Generic pilot credentials" )
if not vo:
vo = Registry.getVOForGroup( group )
if not vo:
return S_ERROR( "Group %s does not have a VO associated" % group )
opsHelper = Operations.Operations( vo = vo )
pilotGroup = opsHelper.getValue( "Pilot/GenericPilotGroup", "" )
pilotDN = opsHelper.getValue( "Pilot/GenericPilotDN", "" )
if pilotDN and pilotGroup:
gLogger.verbose( "Pilot credentials from CS: %s@%s" % ( pilotDN, pilotGroup ) )
result = gProxyManager.userHasProxy( pilotDN, pilotGroup, 86400 )
if not result[ 'OK' ]:
return S_ERROR( "%s@%s has no proxy in ProxyManager" )
return S_OK( ( pilotDN, pilotGroup ) )
#Auto discover
gLogger.verbose( "Pilot credentials are not defined. Autodiscovering..." )
if pilotGroup:
pilotGroups = [ pilotGroup ]
else:
result = Registry.getGroupsWithProperty( Properties.GENERIC_PILOT )
if not result[ 'OK' ]:
return result
pilotGroups = []
groups = result[ 'Value' ]
if not groups:
return S_ERROR( "No group with %s property defined" % Properties.GENERIC_PILOT )
result = Registry.getGroupsForVO( vo )
if not result[ 'OK' ]:
return result
for voGroup in result[ 'Value' ]:
if voGroup in groups:
pilotGroups.append( voGroup )
if not pilotGroups:
return S_ERROR( "No generic pilot group for VO %s" % vo )
for pilotGroup in pilotGroups:
DNs = Registry.getDNsInGroup( pilotGroup )
if not DNs:
continue
if pilotDN:
if pilotDN not in DNs:
continue
result = gProxyManager.userHasProxy( pilotDN, pilotGroup, 86400 )
if result[ 'OK' ] and result[ 'Value' ]:
gLogger.verbose( "Discovered pilot credentials: %s@%s" % ( pilotDN, pilotGroup ) )
return S_OK( ( pilotDN, pilotGroup ) )
else:
for DN in DNs:
result = gProxyManager.userHasProxy( DN, pilotGroup, 86400 )
if result[ 'OK' ] and result[ 'Value' ]:
gLogger.verbose( "Discovered pilot credentials: %s@%s" % ( DN, pilotGroup ) )
return S_OK( ( DN, pilotGroup ) )
if pilotDN:
return S_ERROR( "DN %s does not have group %s" % ( pilotDN, pilotGroups ) )
return S_ERROR( "No generic proxy in the Proxy Manager with groups %s" % pilotGroups )
def __checkIfProxyUploadIsRequired( self ):
result = gProxyManager.userHasProxy( self.ownerDN, self.ownerGroup, validSeconds = 18000 )
if not result[ 'OK' ]:
gLogger.error( "Can't check if the user has proxy uploaded:", result[ 'Message' ] )
return True
#Check if an upload is required
return result[ 'Value' ] == False
def __checkIfProxyUploadIsRequired(self):
result = gProxyManager.userHasProxy(self.ownerDN, self.ownerGroup, validSeconds=18000)
if not result['OK']:
gLogger.error("Can't check if the user has proxy uploaded:", result['Message'])
return True
# Check if an upload is required
return not result['Value']
def findGenericPilotCredentials(vo=False, group=False, pilotDN='', pilotGroup=''):
""" Looks into the Operations/<>/Pilot section of CS to find the pilot credentials.
Then check if the user has a registered proxy in ProxyManager.
if pilotDN or pilotGroup are specified, use them
"""
if not group and not vo:
return S_ERROR("Need a group or a VO to determine the Generic pilot credentials")
if not vo:
vo = Registry.getVOForGroup(group)
if not vo:
return S_ERROR("Group %s does not have a VO associated" % group)
opsHelper = Operations.Operations(vo=vo)
if not pilotGroup:
pilotGroup = opsHelper.getValue("Pilot/GenericPilotGroup", "")
if not pilotDN:
pilotDN = opsHelper.getValue("Pilot/GenericPilotDN", "")
if not pilotDN:
pilotUser = opsHelper.getValue("Pilot/GenericPilotUser", "")
if pilotUser:
result = Registry.getDNForUsername(pilotUser)
if result['OK']:
pilotDN = result['Value'][0]
if pilotDN and pilotGroup:
gLogger.verbose("Pilot credentials: %s@%s" % (pilotDN, pilotGroup))
result = gProxyManager.userHasProxy(pilotDN, pilotGroup, 86400)
if not result['OK']:
return S_ERROR("%s@%s has no proxy in ProxyManager")
return S_OK((pilotDN, pilotGroup))
if pilotDN:
return S_ERROR("DN %s does not have group %s" % (pilotDN, pilotGroup))
return S_ERROR("No generic proxy in the Proxy Manager with groups %s" % pilotGroup)
def findGenericCloudCredentials(vo=False, group=False):
if not group and not vo:
return S_ERROR(
"Need a group or a VO to determine the Generic cloud credentials")
if not vo:
vo = Registry.getVOForGroup(group)
if not vo:
return S_ERROR("Group %s does not have a VO associated" % group)
opsHelper = Operations.Operations(vo=vo)
cloudGroup = opsHelper.getValue("Cloud/GenericCloudGroup", "")
cloudDN = opsHelper.getValue("Cloud/GenericCloudDN", "")
if not cloudDN:
cloudUser = opsHelper.getValue("Cloud/GenericCloudUser", "")
if cloudUser:
result = Registry.getDNForUsername(cloudUser)
if result['OK']:
cloudDN = result['Value'][0]
if cloudDN and cloudGroup:
gLogger.verbose("Cloud credentials from CS: %s@%s" %
(cloudDN, cloudGroup))
result = gProxyManager.userHasProxy(cloudDN, cloudGroup, 86400)
if not result['OK']:
return S_ERROR("%s@%s has no proxy in ProxyManager")
return S_OK((cloudDN, cloudGroup))
return S_ERROR("Cloud credentials not found")
def findGenericCloudCredentials(vo=False, group=False):
"""Get the cloud credentials to use for a specific VO and/or group."""
if not group and not vo:
return S_ERROR(
"Need a group or a VO to determine the Generic cloud credentials")
if not vo:
vo = Registry.getVOForGroup(group)
if not vo:
return S_ERROR("Group %s does not have a VO associated" % group)
opsHelper = Operations.Operations(vo=vo)
cloudGroup = opsHelper.getValue("Cloud/GenericCloudGroup", "")
cloudDN = opsHelper.getValue("Cloud/GenericCloudDN", "")
if not cloudDN:
cloudUser = opsHelper.getValue("Cloud/GenericCloudUser", "")
if cloudUser:
result = Registry.getDNForUsername(cloudUser)
if result["OK"]:
cloudDN = result["Value"][0]
else:
return S_ERROR("Failed to find suitable CloudDN")
if cloudDN and cloudGroup:
gLogger.verbose("Cloud credentials from CS: %s@%s" %
(cloudDN, cloudGroup))
result = gProxyManager.userHasProxy(cloudDN, cloudGroup, 86400)
if not result["OK"]:
return result
return S_OK((cloudDN, cloudGroup))
return S_ERROR("Cloud credentials not found")
def __checkIfProxyUploadIsRequired(self):
"""Check if an upload is required
:return: bool
"""
result = gProxyManager.userHasProxy(self.ownerDN,
self.ownerGroup,
validSeconds=18000)
if not result["OK"]:
self.log.error("Can't check if the user has proxy uploaded",
result["Message"])
return True
# Check if an upload is required
return not result["Value"]
def checkProxyUploaded(self, userDN, userGroup, requiredTime):
"""Set the persistence of a proxy in the Proxy Manager
Example usage:
>>> print diracAdmin.setProxyPersistency( 'some DN', 'dirac group', True )
{'OK': True, 'Value' : True/False }
:param userDN: User DN
:type userDN: string
:param userGroup: DIRAC Group
:type userGroup: string
:param requiredTime: Required life time of the uploaded proxy
:type requiredTime: boolean
:return: S_OK,S_ERROR
"""
return gProxyManager.userHasProxy(userDN, userGroup, requiredTime)
def checkProxyUploaded( self, userDN, userGroup, requiredTime ):
"""Set the persistence of a proxy in the Proxy Manager
Example usage:
>>> print diracAdmin.setProxyPersistency( 'some DN', 'dirac group', True )
{'OK': True, 'Value' : True/False }
@param userDN: User DN
@type userDN: string
@param userGroup: DIRAC Group
@type userGroup: string
@param requiredTime: Required life time of the uploaded proxy
@type requiredTime: boolean
@return: S_OK,S_ERROR
"""
return gProxyManager.userHasProxy( userDN, userGroup, requiredTime )
def findGenericPilotCredentials(vo=False,
group=False,
pilotDN="",
pilotGroup=""):
"""Looks into the Operations/<>/Pilot section of CS to find the pilot credentials.
Then check if the user has a registered proxy in ProxyManager.
if pilotDN or pilotGroup are specified, use them
:param str vo: VO name
:param str group: group name
:param str pilotDN: pilot DN
:param str pilotGroup: pilot group
:return: S_OK(tuple)/S_ERROR()
"""
if not group and not vo:
return S_ERROR(
"Need a group or a VO to determine the Generic pilot credentials")
if not vo:
vo = Registry.getVOForGroup(group)
if not vo:
return S_ERROR("Group %s does not have a VO associated" % group)
opsHelper = Operations.Operations(vo=vo)
if not pilotGroup:
pilotGroup = opsHelper.getValue("Pilot/GenericPilotGroup", "")
if not pilotDN:
pilotDN = opsHelper.getValue("Pilot/GenericPilotDN", "")
if not pilotDN:
pilotUser = opsHelper.getValue("Pilot/GenericPilotUser", "")
if pilotUser:
result = Registry.getDNForUsername(pilotUser)
if result["OK"]:
pilotDN = result["Value"][0]
if pilotDN and pilotGroup:
gLogger.verbose("Pilot credentials: %s@%s" % (pilotDN, pilotGroup))
result = gProxyManager.userHasProxy(pilotDN, pilotGroup, 86400)
if not result["OK"]:
return S_ERROR("%s@%s has no proxy in ProxyManager")
return S_OK((pilotDN, pilotGroup))
if pilotDN:
return S_ERROR("DN %s does not have group %s" % (pilotDN, pilotGroup))
return S_ERROR("No generic proxy in the Proxy Manager with groups %s" %
pilotGroup)
def submitNewBigJob( self ):
result = jobDB.getJobJDL( str( self.__jobID ) , True )
classAdJob = ClassAd( result['Value'] )
executableFile = ""
if classAdJob.lookupAttribute( 'Executable' ):
executableFile = classAdJob.getAttributeString( 'Executable' )
tempPath = self.__tmpSandBoxDir
dirac = Dirac()
if not os.path.exists( tempPath ):
os.makedirs( tempPath )
settingJobSandBoxDir = dirac.getInputSandbox( self.__jobID, tempPath )
self.log.info( 'Writting temporal SandboxDir in Server', settingJobSandBoxDir )
moveData = self.__tmpSandBoxDir + "/InputSandbox" + str( self.__jobID )
HiveV1Cli = HiveV1Client( self.__User , self.__publicIP )
returned = HiveV1Cli.dataCopy( moveData, self.__tmpSandBoxDir )
self.log.info( 'Copy the job contain to the Hadoop Master with HIVE: ', returned )
jobInfo = jobDB.getJobAttributes( self.__jobID )
if not jobInfo['OK']:
return S_ERROR( jobInfo['Value'] )
proxy = ""
jobInfo = jobInfo['Value']
if gProxyManager.userHasProxy( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] ):
proxy = gProxyManager.downloadProxyToFile( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
else:
proxy = self.__requestProxyFromProxyManager( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
HiveJob = "InputSandbox" + str( self.__jobID ) + "/" + executableFile
HiveJobOutput = str( self.__jobID ) + "_" + executableFile + "_out"
returned = HiveV1Cli.jobSubmit( tempPath, HiveJob, proxy['chain'], HiveJobOutput )
self.log.info( 'Launch Hadoop-Hive job to the Master: ', returned )
if not returned['OK']:
return S_ERROR( returned['Message'] )
else:
self.log.info( 'Hadoop-Hive Job ID: ', returned['Value'] )
return S_OK( returned['Value'] )
def _getPilotProxyFromDIRACGroup( self, ownerDN, ownerGroup, requiredTimeLeft ):
"""
Download a limited pilot proxy with VOMS extensions depending on the group
"""
if gProxyManager.userHasProxy( ownerDN, ownerGroup ):
return gProxyManager.downloadProxyToFile( ownerDN, ownerGroup )
#Assign VOMS attribute
vomsAttr = CS.getVOMSAttributeForGroup( ownerGroup )
if not vomsAttr:
self.log.info( "Downloading a proxy without VOMS extensions for %s@%s" % ( ownerDN, ownerGroup ) )
return gProxyManager.downloadProxy( ownerDN, ownerGroup, limited = True,
requiredTimeLeft = requiredTimeLeft )
else:
self.log.info( "Downloading a proxy with '%s' VOMS extension for %s@%s" % ( vomsAttr, ownerDN, ownerGroup ) )
return gProxyManager.downloadVOMSProxy( ownerDN,
ownerGroup,
limited = True,
requiredTimeLeft = requiredTimeLeft,
requiredVOMSAttribute = vomsAttr )
def submitNewBigPilot(self):
tempPath = self.__tmpSandBoxDir + str(self.__jobID)
dirac = Dirac()
if not os.path.exists(tempPath):
os.makedirs(tempPath)
settingJobSandBoxDir = dirac.getInputSandbox(self.__jobID, tempPath)
self.log.info("Writting temporal SandboxDir in Server", settingJobSandBoxDir)
jobXMLName = "job:" + str(self.__jobID) + ".xml"
with open(os.path.join(tempPath, jobXMLName), "wb") as temp_file:
temp_file.write(self.jobWrapper())
self.log.info("Writting temporal Hadoop Job.xml")
HadoopV1cli = HadoopV1Client(self.__User, self.__publicIP, self.__Port)
# returned = HadoopV1cli.dataCopy( tempPath, self.__tmpSandBoxDir )
# self.log.info( 'Copy the job contain to the Hadoop Master: ', returned )
jobInfo = jobDB.getJobAttributes(self.__jobID)
if not jobInfo["OK"]:
return S_ERROR(jobInfo["Value"])
proxy = ""
jobInfo = jobInfo["Value"]
if gProxyManager.userHasProxy(jobInfo["OwnerDN"], jobInfo["OwnerGroup"]):
proxy = gProxyManager.downloadProxyToFile(jobInfo["OwnerDN"], jobInfo["OwnerGroup"])
else:
proxy = self.__requestProxyFromProxyManager(jobInfo["OwnerDN"], jobInfo["OwnerGroup"])
returned = HadoopV1cli.submitPilotJob(tempPath, jobXMLName, proxy["chain"])
self.log.info("Launch Hadoop pilot to the Hadoop Master: ", returned)
if not returned["OK"]:
return S_ERROR(returned["Value"])
else:
self.log.info("Hadoop Job ID: ", returned["Value"])
return S_OK(returned["Value"])
def submitNewBigJob( self ):
tempPath = self.__tmpSandBoxDir + str( self.__jobID )
dirac = Dirac()
if not os.path.exists( tempPath ):
os.makedirs( tempPath )
settingJobSandBoxDir = dirac.getInputSandbox( self.__jobID, tempPath )
self.log.info( 'Writting temporal SandboxDir in Server', settingJobSandBoxDir )
jobXMLName = "job:" + str( self.__jobID ) + '.xml'
with open( os.path.join( tempPath, jobXMLName ), 'wb' ) as temp_file:
temp_file.write( self.jobWrapper() )
self.log.info( 'Writting temporal Hadoop Job.xml' )
HadoopV1cli = HadoopV2Client( self.__User , self.__publicIP )
returned = HadoopV1cli.dataCopy( tempPath, self.__tmpSandBoxDir )
self.log.info( 'Copy the job contain to the Hadoop Master: ', returned )
jobInfo = jobDB.getJobAttributes( self.__jobID )
if not jobInfo['OK']:
return S_ERROR( jobInfo['Value'] )
proxy = ""
jobInfo = jobInfo['Value']
if gProxyManager.userHasProxy( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] ):
proxy = gProxyManager.downloadProxyToFile( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
else:
proxy = self.__requestProxyFromProxyManager( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
returned = HadoopV1cli.jobSubmit( tempPath, jobXMLName, proxy['chain'] )
self.log.info( 'Launch Hadoop job to the Hadoop Master: ', returned )
if not returned['OK']:
return S_ERROR( returned['Message'] )
else:
self.log.info( 'Hadoop Job ID: ', returned['Value'] )
return S_OK( returned['Value'] )
def submitNewBigJob( self ):
#1.- Creamos carpeta temporal
self.log.debug( 'Step1::: mkdir temp folder' )
tempPath = self.__tmpSandBoxDir + str( self.__jobID ) + "/"
dirac = Dirac()
if not os.path.exists( tempPath ):
os.makedirs( tempPath )
#2.- Introducimos el contenido del inputsandbox en la carpeta temporal
self.log.debug( 'Step2::: download inputsand to temp folder' )
settingJobSandBoxDir = dirac.getInputSandbox( self.__jobID, tempPath )
self.log.info( 'Writting temporal SandboxDir in Server', settingJobSandBoxDir )
moveData = tempPath + "/InputSandbox" + str( self.__jobID )
#3.- Move the data to client
self.log.debug( 'Step2::: download inputsandbox to temp folder' )
HadoopV1InteractiveCli = HadoopV1InteractiveClient( self.__User , self.__publicIP, self.__Port )
returned = HadoopV1InteractiveCli.dataCopy( tempPath, self.__tmpSandBoxDir )
self.log.debug( 'Returned of copy the job contain to the Hadoop Master with HadoopInteractive::: ', returned )
#3.- Get executable file
result = jobDB.getJobJDL( str( self.__jobID ) , True )
classAdJob = ClassAd( result['Value'] )
executableFile = ""
if classAdJob.lookupAttribute( 'Executable' ):
executableFile = classAdJob.getAttributeString( 'Executable' )
self.log.debug( 'Step3::: Get executable file: ', executableFile )
jobInfo = jobDB.getJobAttributes( self.__jobID )
if not jobInfo['OK']:
return S_ERROR( jobInfo['Value'] )
proxy = ""
jobInfo = jobInfo['Value']
if gProxyManager.userHasProxy( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] ):
proxy = gProxyManager.downloadProxyToFile( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
else:
proxy = self.__requestProxyFromProxyManager( jobInfo["OwnerDN"], jobInfo["OwnerGroup"] )
HadoopInteractiveJob = "InputSandbox" + str( self.__jobID ) + "/" + executableFile
HadoopInteractiveJobCommand = "InputSandbox" + str( self.__jobID ) + "/" + executableFile + " " + self.__JobName
HadoopInteractiveJobOutput = tempPath + str( self.__jobID ) + "_" + executableFile + "_out"
#4.- Creating second part of the job name
if ( len( re.split( " ", self.__JobName ) ) > 1 ):
#(name for random writter with -D)name_job = re.split( " ", self.__JobName )[0] + " " + re.split( " ", self.__JobName )[1] + " " + re.split( " ", self.__JobName )[2]
name_job = re.split( " ", self.__JobName )[0] + " " + re.split( " ", self.__JobName )[1]
#(name for random writter with -D)output_job = moveData + "/" + re.split( " ", self.__JobName )[3]
#(name for random writter with -D)cfg_job = ""
#(name for random writter with -D)if ( len( re.split( " ", self.__JobName ) ) > 4 ):
#(name for random writter with -D) cfg_job = moveData + "/" + re.split( " ", self.__JobName )[4]
#5.- Parsing execution command
#cmd = "hadoop jar " + tempPath + HadoopInteractiveJob + " " + name_job + " " + output_job + " " + cfg_job
cmd = "hadoop jar " + tempPath + HadoopInteractiveJob + " " + name_job + " " + tempPath + "/InputSandbox" + str( self.__jobID ) + "/" + "/dataset-USC-a-grep '[and]+'"
else:
dataset = re.split( "/", self.__Dataset )
count = 0
datasetname = ""
for dir in dataset:
count = count + 1
if ( count > 2 ):
datasetname = datasetname + "/" + dir
cmd = "hadoop jar " + tempPath + HadoopInteractiveJob + " " + self.__JobName + " " + datasetname + " " + tempPath + "/" + self.__JobName.replace( " ", "" ) + "_" + str( self.__jobID )
self.log.debug( 'Step4::: Making CMD for submission: ', cmd )
self.log.debug( 'Step5::: Submit file to hadoop: ' )
returned = HadoopV1InteractiveCli.jobSubmit( tempPath, HadoopInteractiveJob, proxy['chain'],
HadoopInteractiveJobOutput, cmd )
self.log.info( 'Launch Hadoop-HadoopInteractive job to the Master: ', returned )
if not returned['OK']:
return S_ERROR( returned['Message'] )
else:
self.log.info( 'Hadoop-HadoopInteractive Job ID: ', returned['Value'] )
return S_OK( returned['Value'] )
