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 1.1 by afanfani, Wed Jun 25 12:03:43 2008 UTC vs.
Revision 1.40 by fanzago, Wed Jul 13 17:08:40 2011 UTC

#	Line 1 | Line 1
1	<	#!/usr/bin/env python
2	<
1	>	from Actor import *
2		import urllib
3		from xml.dom.minidom import parse
4		from crab_exceptions import *
5	+	from WorkSpace import *
6	+	from urlparse import urlparse
7	+	from LFNBaseName import *
8	+	from crab_util import getUserName
9
10		class PhEDExDatasvcInfo:
11	<	"""
12	<	provides information from PhEDEx Data Service
13	<	"""
14	<	# PhEDEx Data Service URL
15	<	datasvc_url="https://cmsweb.cern.ch/phedex/test/datasvc/xml/prod"
11	>	def __init__( self , cfg_params=None, config=None ):
12	>
13	>	## PhEDEx Data Service 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	>	self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#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)
44	>
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 + '/'
50	>
51	>	self.datasetpath = cfg_params.get("CMSSW.datasetpath")
52	>	self.publish_data_name = cfg_params.get('USER.publish_data_name','')
53	>
54	>	self.user_port = cfg_params.get("USER.storage_port",'8443')
55	>	self.user_se_path = cfg_params.get("USER.storage_path",'')
56	>	if self.user_se_path:
57	>	if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
58	>
59	>	#check if using "private" Storage
60	>	if not self.node :
61	>	msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
62	>	msg +='\tFor further information please visit : %s'%self.stage_out_faq
63	>	raise CrabException(msg)
64	>	if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
65	>
66	>	if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
67	>	##### remove
68	>	####### FEDE FOR BUG 73010 ############
69	>	print "--------->>>> self = ", self
70	>	task = common._db.getTask()
71	>	print "task = ", task
72	>	print "common.work_space = ", common.work_space._top_dir
73	>	print "removing common.work_space"
74	>	add = '\n'
75	>	import shutil
76	>	try:
77	>	shutil.rmtree(common.work_space._top_dir)
78	>	except OSError:
79	>	add += '\t Warning: problems removing the dir ' + common.work_space._top_dir + ' \n'
80	>	add += '\t Please remove it by hand'
81	>	#common.work_space.delete()
82	>
83	>	msg = 'Error: you are asking to stage out without using CMS Storage Name convention. In this case you \n'
84	>	msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
85	>	msg += '\t Otherwise this task can not be correctly created. \n'
86	>	msg += '\t For further information please visit : \n\t%s'%self.stage_out_faq
87	>	msg += add
88	>	raise CrabException(msg)
89	>	#########################################
90	>
91	>	self.forced_path = '/store/user/'
92	>	if self.sched in ['CAF','LSF','PBS']:
93	>	self.srm_version = 'direct'
94	>	self.SE = {'CAF':'caf.cern.ch', 'LSF':'', 'PBS':''}
95	>	if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
96	>
97	>	if not self.usePhedex:
98	>	self.forced_path = self.user_remote_dir
99	>	return
100	>
101	>	def getEndpoint(self):
102	>	'''
103	>	Return full SE endpoint and related infos
104	>	'''
105	>	self.lfn = self.getLFN()
106	>
107	>	#extract the PFN for the given node,LFN,protocol
108	>	endpoint = self.getStageoutPFN()
109	>	if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
110	>
111	>	#extract SE name an SE_PATH (needed for publication)
112	>	SE, SE_PATH, User = self.splitEndpoint(endpoint)
113	>
114	>	return endpoint, self.lfn , SE, SE_PATH, User
115	>
116	>	def splitEndpoint(self, endpoint):
117	>	'''
118	>	Return relevant infos from endpoint
119	>	'''
120	>	SE = ''
121	>	SE_PATH = ''
122	>	USER = getUserName()
123	>	if self.usePhedex:
124	>	if self.protocol == 'direct':
125	>	query=endpoint
126	>	SE_PATH = endpoint
127	>	SE = self.SE[self.sched]
128	>	else:
129	>	url = 'http://'+endpoint.split('://')[1]
130	>	scheme, host, path, params, query, fragment = urlparse(url)
131	>	SE = self.getAuthoritativeSE()
132	>	SE_PATH = endpoint.split(host)[1]
133	>	else:
134	>	#### to test #####
135	>	# url = 'http://'+endpoint.split('://')[1]
136	>	# scheme, host, path, params, query, fragment = urlparse(url)
137	>	# SE = host.split(':')[0]
138	>	# SE_PATH = endpoint.split(host)[1]
139	>	SE = self.node
140	>	SE_PATH = self.user_se_path + self.user_remote_dir
141	>	if self.lfn.find('group') != -1:
142	>	try:
143	>	USER = (self.lfn.split('group')[1]).split('/')[1]
144	>	except:
145	>	pass
146	>	return SE, SE_PATH, USER
147	>
148	>	def getLFN(self):
149	>	"""
150	>	define the LFN composing the needed pieces
151	>	"""
152	>	lfn = ''
153	>	l_User = False
154	>	if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
155	>	### add here check if user is trying to force a wrong LFN using a T2  TODO
156	>	## check if storage_name is a T2 (siteDB query)
157	>	## if yes :match self.user_lfn with LFNBaseName...
158	>	##     if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
159	>	lfn = self.user_remote_dir
160	>	return lfn
161	>	if self.publish_data_name == '' and int(self.publish_data) == 1:
162	>	msg = "Error. The [USER] section does not have 'publish_data_name'\n"
163	>	msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
164	>	raise CrabException(msg)
165	>	if self.publish_data_name == '' and int(self.usenamespace) == 1:
166	>	self.publish_data_name = "DefaultDataset"
167	>	if int(self.publish_data) == 1:
168	>	if self.sched in ['CAF']: l_User=True
169	>	primaryDataset = self.computePrimaryDataset()
170	>	### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
171	>	### for the publication in order to be able to check the lfn length
172	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)  + '/${PSETHASH}/'
173	>	elif int(self.usenamespace) == 1:
174	>	if self.sched in ['CAF']: l_User=True
175	>	primaryDataset = self.computePrimaryDataset()
176	>	lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name)  + '/${PSETHASH}/'
177	>	else:
178	>	if self.sched in ['CAF','LSF']: l_User=True
179	>	lfn = LFNBase(self.forced_path,self.user_remote_dir)
180	>	if ( lfn[-1] != '/' ) : lfn = lfn + '/'
181	>	return lfn
182	>
183	>	def computePrimaryDataset(self):
184	>	"""
185	>	compute the last part for the LFN in case of publication
186	>	"""
187	>	if (self.datasetpath.upper() != 'NONE'):
188	>	primarydataset = self.datasetpath.split("/")[1]
189	>	else:
190	>	primarydataset = self.publish_data_name
191	>	return primarydataset
192
193	<	def lfn2pfn(self,node,lfn,protocol):
194	<	"""
195	<	PhEDEx Data Service lfn2pfn call
17	<
18	<	input:   LFN,node name,protocol
19	<	returns: DOM object with the content of the PhEDEx Data Service call
20	<	"""
21	<	params = {'node' : node , 'lfn': lfn , 'protocol': protocol}
22	<	params = urllib.urlencode(params)
23	<	#print params
24	<	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
25	<	urlresults = urllib.urlopen(datasvc_lfn2pfn, params)
26	<	try:
27	<	urlresults = parse(urlresults)
28	<	except:
29	<	urlresults = None
30	<	return urlresults
31	<
32	<	def parse_error(self,urlresults):
33	<	"""
34	<	look for errors in the DOM object returned by PhEDEx Data Service call
35	<	"""
36	<	errormsg = None
37	<	errors=urlresults.getElementsByTagName('error')
38	<	for error in errors:
39	<	errormsg=error.childNodes[0].data
40	<	if len(error.childNodes)>1:
41	<	errormsg+=error.childNodes[1].data
42	<	return errormsg
43	<
44	<	def parse_lfn2pfn(self,urlresults):
45	<	"""
46	<	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
47	<
48	<	input:    DOM object with the content of the lfn2pfn call
49	<	returns:  PFN
50	<	"""
51	<	result = urlresults.getElementsByTagName('phedex')
52	<	if not result:
53	<	return []
54	<	result = result[0]
55	<	pfn = None
56	<	mapping = result.getElementsByTagName('mapping')
57	<	for m in mapping:
58	<	pfn=m.getAttribute("pfn")
59	<	if pfn:
60	<	return pfn
61	<
62	<	def getStageoutPFN(self,node,lfn,protocol):
63	<	"""
64	<	input:   LFN,node name,protocol
65	<	returns: PFN
66	<	"""
67	<	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,node,lfn,protocol)
68	<	domlfn2pfn = self.lfn2pfn(node,lfn,protocol)
69	<	if not domlfn2pfn :
70	<	msg="Unable to get info from %s"%fullurl
71	<	raise CrabException(msg)
72	<
73	<	errormsg = self.parse_error(domlfn2pfn)
74	<	if errormsg:
75	<	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
76	<	raise CrabException(msg)
77	<
78	<	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
79	<	if not stageoutpfn:
80	<	msg="Unable to get stageout path (PFN) from %s"%fullurl
81	<	raise CrabException(msg)
82	<	return stageoutpfn
83	<
84	<
85	<	if __name__ == '__main__' :
86	<	"""
87	<	"""
88	<	from crab_logger import Logger
89	<	from WorkSpace import *
90	<	continue_dir="/home/fanfani/CRAB"
91	<	cfg_params={'USER.logdir' : continue_dir }
92	<	common.work_space = WorkSpace(continue_dir, cfg_params)
93	<	log = Logger()
94	<	common.logger = log
95	<
96	<	from LFNBaseName import *
97	<	# test values
98	<	lfn = LFNBase("datasetstring")
99	<	node='T2_IT_Bari'
100	<	protocol="srmv2"
101	<
102	<	#create an instance of the PhEDExDatasvcInfo object
103	<	dsvc = PhEDExDatasvcInfo()
104	<	#extract the PFN for the given node,LFN,protocol
105	<	print "Stageout to %s"%dsvc.getStageoutPFN(node,lfn,protocol)
193	>	def domPhedex(self,params,datasvc_baseUrl):
194	>	"""
195	>	PhEDEx Data Service lfn2pfn call
196
197	+	input:   params,datasvc_baseUrl
198	+	returns: DOM object with the content of the PhEDEx Data Service call
199	+	"""
200	+	params = urllib.urlencode(params)
201	+	try:
202	+	urlresults = urllib.urlopen(datasvc_baseUrl, params)
203	+	urlresults = parse(urlresults)
204	+	except IOError:
205	+	msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
206	+	raise CrabException(msg)
207	+	except:
208	+	urlresults = None
209	+
210	+	return urlresults
211	+
212	+	def parse_error(self,urlresults):
213	+	"""
214	+	look for errors in the DOM object returned by PhEDEx Data Service call
215	+	"""
216	+	errormsg = None
217	+	errors=urlresults.getElementsByTagName('error')
218	+	for error in errors:
219	+	errormsg=error.childNodes[0].data
220	+	if len(error.childNodes)>1:
221	+	errormsg+=error.childNodes[1].data
222	+	return errormsg
223	+
224	+	def parse_lfn2pfn(self,urlresults):
225	+	"""
226	+	Parse the content of the result of lfn2pfn PhEDEx Data Service  call
227	+
228	+	input:    DOM object with the content of the lfn2pfn call
229	+	returns:  PFN
230	+	"""
231	+	result = urlresults.getElementsByTagName('phedex')
232	+
233	+	if not result:
234	+	return []
235	+	result = result[0]
236	+	pfn = None
237	+	mapping = result.getElementsByTagName('mapping')
238	+	for m in mapping:
239	+	pfn=m.getAttribute("pfn")
240	+	if pfn:
241	+	return pfn
242	+
243	+	def getStageoutPFN( self ):
244	+	"""
245	+	input:   LFN,node name,protocol
246	+	returns: PFN
247	+	"""
248	+	if self.usePhedex:
249	+	params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
250	+	datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
251	+	fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
252	+	domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
253	+	if not domlfn2pfn :
254	+	msg="Unable to get info from %s"%fullurl
255	+	raise CrabException(msg)
256	+
257	+	errormsg = self.parse_error(domlfn2pfn)
258	+	if errormsg:
259	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
260	+	raise CrabException(msg)
261	+
262	+	stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
263	+	if not stageoutpfn:
264	+	msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
265	+	msg+='      Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
266	+	msg+='      reporting: \n'
267	+	msg+='       Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
268	+	msg+='       OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
269	+	raise CrabException(msg)
270	+	else:
271	+	if self.sched in ['CAF','LSF','PBS'] :
272	+	stageoutpfn = self.user_se_path+self.lfn
273	+	else:
274	+	stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
275	+
276	+	if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
277	+	return stageoutpfn
278	+
279	+	def getAuthoritativeSE(self):
280	+	"""
281	+	input:   node name
282	+	returns: AuthoritativeSE
283	+	"""
284	+	params = {'node' : self.node }
285	+	datasvc_nodes="%s/nodes"%self.datasvc_url
286	+	fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
287	+	domnodes = self.domPhedex(params,datasvc_nodes)
288	+
289	+	if not domnodes :
290	+	msg="Unable to get info from %s"%fullurl
291	+	raise CrabException(msg)
292	+
293	+	errormsg = self.parse_error(domnodes)
294	+	if errormsg:
295	+	msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
296	+	raise CrabException(msg)
297	+	result = domnodes.getElementsByTagName('phedex')
298	+	if not result:
299	+	return []
300	+	result = result[0]
301	+	se = None
302	+	node = result.getElementsByTagName('node')
303	+	for m in node:
304	+	se=m.getAttribute("se")
305	+	if se:
306	+	return se
307	+
308	+
309	+	if __name__ == '__main__':
310	+	"""
311	+	Sort of unit testing to check Phedex API for whatever site and/or lfn.
312	+	Usage:
313	+	python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
314	+
315	+	"""
316	+	import getopt,sys
317	+	from crab_util import *
318	+	import common
319	+	klass_name = 'SchedulerGlite'
320	+	klass = importName(klass_name, klass_name)
321	+	common.scheduler = klass()
322	+
323	+	lfn="/store/user/"
324	+	node='T2_IT_Bari'
325	+	valid = ['node=','lfn=']
326	+	try:
327	+	opts, args = getopt.getopt(sys.argv[1:], "", valid)
328	+	except getopt.GetoptError, ex:
329	+	print str(ex)
330	+	sys.exit(1)
331	+	for o, a in opts:
332	+	if o == "--node":
333	+	node = a
334	+	if o == "--lfn":
335	+	lfn = a
336	+
337	+	mycfg_params = { 'USER.storage_element': node }
338	+	dsvc = PhEDExDatasvcInfo(mycfg_params)
339	+	dsvc.lfn = lfn
340	+	print dsvc.getStageoutPFN()
341	+

Diff Legend

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