ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/COMP/CRAB/python/PhEDExDatasvcInfo.py

Comparing COMP/CRAB/python/PhEDExDatasvcInfo.py (file contents):
Revision 1.22 by spiga, Tue Mar 31 23:43:22 2009 UTC vs.
Revision 1.49 by fanzago, Fri Oct 12 13:36:22 2012 UTC

#	Line 2 | Line 2 | from Actor import *
2		import urllib
3		from xml.dom.minidom import parse
4		from crab_exceptions import *
5	–	from crab_logger import Logger
5		from WorkSpace import *
6		from urlparse import urlparse
7		from LFNBaseName import *
8	+	from crab_util import getUserName
9
10		class PhEDExDatasvcInfo:
11	<	def __init__( self , cfg_params ):
11	>	def __init__( self , cfg_params=None, config=None ):
12
13		## PhEDEx Data Service URL
14	<	url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
15	<	self.datasvc_url = cfg_params.get("USER.datasvc_url",url)
14	>	self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
15
16		self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
17	<	stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabFaq#How_to_store_output_with_CRAB_2'
17	>	self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMSPublic/SWGuideCrabFaq#Stageout_and_publication'
18		self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
19
20	+	self.usePhedex = True
21	+	self.sched = common.scheduler.name().upper()
22	+
23	+	if config!=None:
24	+	self.checkConfig(config)
25	+	else:
26	+	self.checkCfgConfig(cfg_params)
27	+
28	+	self.protocol = self.srm_version
29	+
30	+
31	+	def checkConfig(self,config):
32	+	"""
33	+	"""
34	+	self.srm_version = config.get("srm_version",'srmv2')
35	+	self.node = config.get('storage_element',None)
36	+	self.lfn='/store/'
37	+
38	+	def checkCfgConfig(self,cfg_params):
39	+	"""
40	+	"""
41	+	self.datasvc_url = cfg_params.get("USER.datasvc_url",self.datasvc_url)
42		self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
43		self.node = cfg_params.get('USER.storage_element',None)
23	–
44
25	–
26	–	self.user_lfn = cfg_params.get("USER.lfn",'')
45		self.publish_data = cfg_params.get("USER.publish_data",0)
46		self.usenamespace = cfg_params.get("USER.usenamespace",0)
47		self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
48		if self.user_remote_dir:
49		if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
32	–	if self.user_lfn:
33	–	msg  = 'Warning: lfn has been deprecated, CRAB will ignore it.\n'
34	–	msg += '\t Please use only user_remote_dir removing lfn from your crab.cfg\n'
35	–	msg += '\t For further information please visit : \n\t%s'%stage_out_faq
36	–	common.logger.message(msg)
50
51		self.datasetpath = cfg_params.get("CMSSW.datasetpath")
52		self.publish_data_name = cfg_params.get('USER.publish_data_name','')
53
54	+	self.pset = cfg_params.get('CMSSW.pset',None)
55	+
56		self.user_port = cfg_params.get("USER.storage_port",'8443')
57		self.user_se_path = cfg_params.get("USER.storage_path",'')
58		if self.user_se_path:
59		if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
60
61		#check if using "private" Storage
47	–	self.usePhedex = True
62		if not self.node :
63		msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
64	<	msg +='\tFor further information please visit : %s'%stage_out_faq
64	>	msg +='\tFor further information please visit : %s'%self.stage_out_faq
65		raise CrabException(msg)
66		if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
67
68		if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
69	<	msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
70	<	msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
71	<	msg += '\t For further information please visit : \n\t%s'%stage_out_faq
69	>	####### FEDE FOR BUG 73010 ############
70	>	msg =  'Error: task ' + common.work_space._top_dir + ' not correctly created. Please remove it. \n'
71	>	msg += '      You are asking to stage out without using CMS Storage Name convention. In this case you \n'
72	>	msg += '      must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n'
73	>	msg += '      For further information please visit : \n\t%s'%self.stage_out_faq
74	>	task = common._db.getTask()
75	>	#add = '\n\n'
76	>	#import shutil
77	>	#try:
78	>	#    add += '      Task not correctly created: removing the working_dir ' +  common.work_space._top_dir + ' \n'
79	>	#    shutil.rmtree(common.work_space._top_dir)
80	>	#except OSError:
81	>	#    add += '      Warning: problems removing the working_dir ' + common.work_space._top_dir + ' \n'
82	>	#    add += '      Please remove it by hand'
83	>	#msg += add
84		raise CrabException(msg)
59	–	self.sched = common.scheduler.name().upper()
60	–	self.protocol = self.srm_version
85
86		self.forced_path = '/store/user/'
87	<	if self.sched in ['CAF','LSF']:
88	<	self.protocol = 'direct'
89	<	self.SE = {'CAF':'caf.cern.ch', 'LSF':''}
66	<	if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
87	>	if self.sched in ['LSF','PBS']:
88	>	self.srm_version = 'direct'
89	>	self.SE = {'LSF':'', 'PBS':''}
90
91	+	if self.sched == 'CAF':
92	+	#### FEDE TEST FOR XROOTD
93	+	######### first solution ################
94	+	#eos = cfg_params.get("USER.caf_eos_area", 0)
95	+	#if eos == 0:
96	+	#    self.forced_path = '/store/caf/user/'
97	+	#else:
98	+	#    self.forced_path = '/store/eos/user'
99	+	#########################################
100	+	######### second solution ###############
101	+	self.forced_path = cfg_params.get("USER.caf_lfn", '/store/caf/user')
102	+	#########################################
103	+	#print "--->>> FORCING THE FIRST PART OF LFN WITH ", self.forced_path
104	+	self.SE = {'CAF':'caf.cern.ch'}
105	+	self.srm_version = 'stageout'
106	+	#print "--->>> query with 'stageout' "
107	+	#########################################
108	+
109		if not self.usePhedex:
110		self.forced_path = self.user_remote_dir
111		return
#	Line 77 | Line 118 | class PhEDExDatasvcInfo:
118
119		#extract the PFN for the given node,LFN,protocol
120		endpoint = self.getStageoutPFN()
121	+	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
122	+	### FEDE bug fix 93573
123	+	if ( self.lfn[-1] != '/' ) : self.lfn = self.lfn + '/'
124	+
125	+	if int(self.publish_data) == 1 or  int(self.usenamespace) == 1:
126	+	self.lfn = self.lfn  + '${PSETHASH}/'
127	+	endpoint = endpoint  + '${PSETHASH}/'
128
129		#extract SE name an SE_PATH (needed for publication)
130		SE, SE_PATH, User = self.splitEndpoint(endpoint)
131
132	+	#### FEDE FOR XROOTD #####
133	+	#print "in getEndpoint di PhEDExDatasvcInfo.py: "
134	+	#print "    SE = ", SE
135	+	#print "    SE_PATH = ", SE_PATH
136	+	#print "    User = ", User
137	+	#print "    endpoint = ", endpoint
138	+	##############################
139	+
140		return endpoint, self.lfn , SE, SE_PATH, User
141
142		def splitEndpoint(self, endpoint):
#	Line 89 | Line 145 | class PhEDExDatasvcInfo:
145		'''
146		SE = ''
147		SE_PATH = ''
148	<	USER = ''
148	>	USER = getUserName()
149		if self.usePhedex:
150	<	if self.protocol == 'direct':
151	<	query=endpoint
96	<	SE_PATH = endpoint
150	>	### FEDE PER TEST WITH XROOTD
151	>	if (self.protocol == 'direct' or self.protocol == 'stageout'):
152		SE = self.SE[self.sched]
153	+	SE_PATH = endpoint
154	+	#############################
155	+	#print "    SE_PATH = ", SE_PATH
156		else:
157		url = 'http://'+endpoint.split('://')[1]
100	–	# python > 2.4
101	–	# SE = urlparse(url).hostname
158		scheme, host, path, params, query, fragment = urlparse(url)
159	<	SE = host.split(':')[0]
159	>	SE = self.getAuthoritativeSE()
160		SE_PATH = endpoint.split(host)[1]
105	–	USER = (query.split('user')[1]).split('/')[1]
161		else:
107	–	#### to test #####
108	–	# url = 'http://'+endpoint.split('://')[1]
109	–	# scheme, host, path, params, query, fragment = urlparse(url)
110	–	# SE = host.split(':')[0]
111	–	# SE_PATH = endpoint.split(host)[1]
162		SE = self.node
163		SE_PATH = self.user_se_path + self.user_remote_dir
164	<	try:
165	<	USER = (self.lfn.split('user')[1]).split('/')[1]
166	<	except:
167	<	pass
168	<
164	>	if self.lfn.find('group') != -1:
165	>	try:
166	>	USER = (self.lfn.split('group')[1]).split('/')[1]
167	>	except:
168	>	pass
169		return SE, SE_PATH, USER
120	–
170
171		def getLFN(self):
172		"""
#	Line 138 | Line 187 | class PhEDExDatasvcInfo:
187		raise CrabException(msg)
188		if self.publish_data_name == '' and int(self.usenamespace) == 1:
189		self.publish_data_name = "DefaultDataset"
190	<	if int(self.publish_data) == 1 or int(self.usenamespace) == 1:
190	>	if int(self.publish_data) == 1:
191	>	if self.sched in ['CAF']: l_User=True
192	>	primaryDataset = self.computePrimaryDataset()
193	>	### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
194	>	### for the publication in order to be able to check the lfn length
195	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
196	>	elif int(self.usenamespace) == 1:
197		if self.sched in ['CAF']: l_User=True
198		primaryDataset = self.computePrimaryDataset()
199	<	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'
199	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)
200		else:
201		if self.sched in ['CAF','LSF']: l_User=True
202		lfn = LFNBase(self.forced_path,self.user_remote_dir)
203	+
204	+	if ( lfn[-1] != '/' ) : lfn = lfn + '/'
205	+
206		return lfn
207
208		def computePrimaryDataset(self):
#	Line 157 | Line 215 | class PhEDExDatasvcInfo:
215		primarydataset = self.publish_data_name
216		return primarydataset
217
218	<	def lfn2pfn(self):
218	>	def domPhedex(self,params,datasvc_baseUrl):
219		"""
220		PhEDEx Data Service lfn2pfn call
221
222	<	input:   LFN,node name,protocol
222	>	input:   params,datasvc_baseUrl
223		returns: DOM object with the content of the PhEDEx Data Service call
224		"""
167	–	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
225		params = urllib.urlencode(params)
169	–	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
170	–	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
226		try:
227	+	urlresults = urllib.urlopen(datasvc_baseUrl, params)
228		urlresults = parse(urlresults)
229	+	except IOError:
230	+	msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
231	+	raise CrabException(msg)
232		except:
233		urlresults = None
234
#	Line 212 | Line 271 | class PhEDExDatasvcInfo:
271		returns: PFN
272		"""
273		if self.usePhedex:
274	+	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
275	+	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
276		fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
277	<	domlfn2pfn = self.lfn2pfn()
277	>	#print "--->>> fullurl = ", fullurl
278	>	domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
279		if not domlfn2pfn :
280		msg="Unable to get info from %s"%fullurl
281		raise CrabException(msg)
#	Line 232 | Line 294 | class PhEDExDatasvcInfo:
294		msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
295		raise CrabException(msg)
296		else:
297	<	if self.sched in ['CAF','LSF'] :
298	<	stageoutpfn = self.user_se_path+self.lfn
297	>	if self.sched in ['CAF','LSF','PBS'] :
298	>	if (self.user_se_path[-1]=='/') and (self.lfn[0]=='/'):
299	>	stageoutpfn = self.user_se_path+(self.lfn).lstrip('/')
300	>	else:
301	>	stageoutpfn = self.user_se_path+self.lfn
302		else:
303		stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
304
305	+	if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
306		return stageoutpfn
307
308	+	def getAuthoritativeSE(self):
309	+	"""
310	+	input:   node name
311	+	returns: AuthoritativeSE
312	+	"""
313	+	params = {'node' : self.node }
314	+	datasvc_nodes="%s/nodes"%self.datasvc_url
315	+	fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
316	+	domnodes = self.domPhedex(params,datasvc_nodes)
317	+
318	+	if not domnodes :
319	+	msg="Unable to get info from %s"%fullurl
320	+	raise CrabException(msg)
321	+
322	+	errormsg = self.parse_error(domnodes)
323	+	if errormsg:
324	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
325	+	raise CrabException(msg)
326	+	result = domnodes.getElementsByTagName('phedex')
327	+	if not result:
328	+	return []
329	+	result = result[0]
330	+	se = None
331	+	node = result.getElementsByTagName('node')
332	+	for m in node:
333	+	se=m.getAttribute("se")
334	+	if se:
335	+	return se
336
337
338		if __name__ == '__main__':

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines