ViewVC Help
View File | Revision Log | Show Annotations | Root Listing
root/cvsroot/COMP/CRAB/python/PhEDExDatasvcInfo.py
Revision: 1.35
Committed: Fri Dec 4 12:08:21 2009 UTC (15 years, 4 months ago) by fanzago
Content type: text/x-python
Branch: MAIN
CVS Tags: fede_170310, CRAB_2_7_1_pre12, CRAB_2_7_1_pre11, CRAB_2_7_1_pre10, CRAB_2_7_1_pre9, CRAB_LumiMask, CRAB_2_7_lumi, from_LimiMask, CRAB_2_7_1_pre8, CRAB_2_7_1_pre6, CRAB_2_7_1_pre5, CRAB_2_7_1_wmbs_pre4, CRAB_2_7_1_pre4, CRAB_2_7_1_pre3, CRAB_2_7_1_pre2, CRAB_2_7_1_pre1
Branch point for: CRAB_multiout, CRAB_2_7_1_branch
Changes since 1.34: +2 -2 lines
Log Message:
changed for backup copy to close SE. adding temp in the path dir and wrapper exit code 60308

File Contents

# Content
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
9 class PhEDExDatasvcInfo:
10 def __init__( self , cfg_params=None, config=None ):
11
12 ## PhEDEx Data Service URL
13 self.datasvc_url="https://cmsweb.cern.ch/phedex/datasvc/xml/prod"
14
15 self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
16 self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication'
17 self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
18
19 self.usePhedex = True
20 self.sched = common.scheduler.name().upper()
21
22 if config!=None:
23 self.checkConfig(config)
24 else:
25 self.checkCfgConfig(cfg_params)
26
27 self.protocol = self.srm_version
28
29
30 def checkConfig(self,config):
31 """
32 """
33 self.srm_version = config.get("srm_version",'srmv2')
34 self.node = config.get('storage_element',None)
35 self.lfn='/store/'
36
37 def checkCfgConfig(self,cfg_params):
38 """
39 """
40 self.datasvc_url = cfg_params.get("USER.datasvc_url",self.datasvc_url)
41 self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
42 self.node = cfg_params.get('USER.storage_element',None)
43
44 self.publish_data = cfg_params.get("USER.publish_data",0)
45 self.usenamespace = cfg_params.get("USER.usenamespace",0)
46 self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
47 if self.user_remote_dir:
48 if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
49
50 self.datasetpath = cfg_params.get("CMSSW.datasetpath")
51 self.publish_data_name = cfg_params.get('USER.publish_data_name','')
52
53 self.user_port = cfg_params.get("USER.storage_port",'8443')
54 self.user_se_path = cfg_params.get("USER.storage_path",'')
55 if self.user_se_path:
56 if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
57
58 #check if using "private" Storage
59 if not self.node :
60 msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
61 msg +='\tFor further information please visit : %s'%self.stage_out_faq
62 raise CrabException(msg)
63 if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
64
65 if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
66 msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
67 msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
68 msg += '\t For further information please visit : \n\t%s'%self.stage_out_faq
69 raise CrabException(msg)
70
71 self.forced_path = '/store/user/'
72 if self.sched in ['CAF','LSF','PBS']:
73 self.srm_version = 'direct'
74 self.SE = {'CAF':'caf.cern.ch', 'LSF':'', 'PBS':''}
75 if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
76
77 if not self.usePhedex:
78 self.forced_path = self.user_remote_dir
79 return
80
81 def getEndpoint(self):
82 '''
83 Return full SE endpoint and related infos
84 '''
85 self.lfn = self.getLFN()
86
87 #extract the PFN for the given node,LFN,protocol
88 endpoint = self.getStageoutPFN()
89 if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
90
91 #extract SE name an SE_PATH (needed for publication)
92 SE, SE_PATH, User = self.splitEndpoint(endpoint)
93
94 return endpoint, self.lfn , SE, SE_PATH, User
95
96 def splitEndpoint(self, endpoint):
97 '''
98 Return relevant infos from endpoint
99 '''
100 SE = ''
101 SE_PATH = ''
102 USER = ''
103 if self.usePhedex:
104 if self.protocol == 'direct':
105 query=endpoint
106 SE_PATH = endpoint
107 SE = self.SE[self.sched]
108 else:
109 url = 'http://'+endpoint.split('://')[1]
110 # python > 2.4
111 # SE = urlparse(url).hostname
112 scheme, host, path, params, query, fragment = urlparse(url)
113 # SE = host.split(':')[0]
114 SE = self.getAuthoritativeSE()
115 SE_PATH = endpoint.split(host)[1]
116 USER = (query.split('user')[1]).split('/')[1]
117 else:
118 #### to test #####
119 # url = 'http://'+endpoint.split('://')[1]
120 # scheme, host, path, params, query, fragment = urlparse(url)
121 # SE = host.split(':')[0]
122 # SE_PATH = endpoint.split(host)[1]
123 SE = self.node
124 SE_PATH = self.user_se_path + self.user_remote_dir
125 if self.lfn.find('user'):
126 try:
127 USER = (self.lfn.split('user')[1]).split('/')[1]
128 except:
129 pass
130 if self.lfn.find('group'):
131 try:
132 USER = (self.lfn.split('group')[1]).split('/')[1]
133 except:
134 pass
135 return SE, SE_PATH, USER
136
137
138 def getLFN(self):
139 """
140 define the LFN composing the needed pieces
141 """
142 lfn = ''
143 l_User = False
144 if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
145 ### add here check if user is trying to force a wrong LFN using a T2 TODO
146 ## check if storage_name is a T2 (siteDB query)
147 ## if yes :match self.user_lfn with LFNBaseName...
148 ## if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
149 lfn = self.user_remote_dir
150 return lfn
151 if self.publish_data_name == '' and int(self.publish_data) == 1:
152 msg = "Error. The [USER] section does not have 'publish_data_name'\n"
153 msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
154 raise CrabException(msg)
155 if self.publish_data_name == '' and int(self.usenamespace) == 1:
156 self.publish_data_name = "DefaultDataset"
157 if int(self.publish_data) == 1:
158 if self.sched in ['CAF']: l_User=True
159 primaryDataset = self.computePrimaryDataset()
160 ### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
161 ### for the publication in order to be able to check the lfn length
162 lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) + '/${PSETHASH}/'
163 elif int(self.usenamespace) == 1:
164 if self.sched in ['CAF']: l_User=True
165 primaryDataset = self.computePrimaryDataset()
166 lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name) + '/${PSETHASH}/'
167 else:
168 if self.sched in ['CAF','LSF']: l_User=True
169 lfn = LFNBase(self.forced_path,self.user_remote_dir)
170 if ( lfn[-1] != '/' ) : lfn = lfn + '/'
171 return lfn
172
173 def computePrimaryDataset(self):
174 """
175 compute the last part for the LFN in case of publication
176 """
177 if (self.datasetpath.upper() != 'NONE'):
178 primarydataset = self.datasetpath.split("/")[1]
179 else:
180 primarydataset = self.publish_data_name
181 return primarydataset
182
183 def domPhedex(self,params,datasvc_baseUrl):
184 """
185 PhEDEx Data Service lfn2pfn call
186
187 input: params,datasvc_baseUrl
188 returns: DOM object with the content of the PhEDEx Data Service call
189 """
190 params = urllib.urlencode(params)
191 try:
192 urlresults = urllib.urlopen(datasvc_baseUrl, params)
193 urlresults = parse(urlresults)
194 except IOError:
195 msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
196 raise CrabException(msg)
197 except:
198 urlresults = None
199
200 return urlresults
201
202 def parse_error(self,urlresults):
203 """
204 look for errors in the DOM object returned by PhEDEx Data Service call
205 """
206 errormsg = None
207 errors=urlresults.getElementsByTagName('error')
208 for error in errors:
209 errormsg=error.childNodes[0].data
210 if len(error.childNodes)>1:
211 errormsg+=error.childNodes[1].data
212 return errormsg
213
214 def parse_lfn2pfn(self,urlresults):
215 """
216 Parse the content of the result of lfn2pfn PhEDEx Data Service call
217
218 input: DOM object with the content of the lfn2pfn call
219 returns: PFN
220 """
221 result = urlresults.getElementsByTagName('phedex')
222
223 if not result:
224 return []
225 result = result[0]
226 pfn = None
227 mapping = result.getElementsByTagName('mapping')
228 for m in mapping:
229 pfn=m.getAttribute("pfn")
230 if pfn:
231 return pfn
232
233 def getStageoutPFN( self ):
234 """
235 input: LFN,node name,protocol
236 returns: PFN
237 """
238 if self.usePhedex:
239 params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
240 datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
241 fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
242 domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
243 if not domlfn2pfn :
244 msg="Unable to get info from %s"%fullurl
245 raise CrabException(msg)
246
247 errormsg = self.parse_error(domlfn2pfn)
248 if errormsg:
249 msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
250 raise CrabException(msg)
251
252 stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
253 if not stageoutpfn:
254 msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
255 msg+=' Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
256 msg+=' reporting: \n'
257 msg+=' Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
258 msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
259 raise CrabException(msg)
260 else:
261 if self.sched in ['CAF','LSF','PBS'] :
262 stageoutpfn = self.user_se_path+self.lfn
263 else:
264 stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
265
266 if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
267 return stageoutpfn
268
269 def getAuthoritativeSE(self):
270 """
271 input: node name
272 returns: AuthoritativeSE
273 """
274 params = {'node' : self.node }
275 datasvc_nodes="%s/nodes"%self.datasvc_url
276 fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
277 domnodes = self.domPhedex(params,datasvc_nodes)
278
279 if not domnodes :
280 msg="Unable to get info from %s"%fullurl
281 raise CrabException(msg)
282
283 errormsg = self.parse_error(domnodes)
284 if errormsg:
285 msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
286 raise CrabException(msg)
287 result = domnodes.getElementsByTagName('phedex')
288 if not result:
289 return []
290 result = result[0]
291 se = None
292 node = result.getElementsByTagName('node')
293 for m in node:
294 se=m.getAttribute("se")
295 if se:
296 return se
297
298
299 if __name__ == '__main__':
300 """
301 Sort of unit testing to check Phedex API for whatever site and/or lfn.
302 Usage:
303 python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
304
305 """
306 import getopt,sys
307 from crab_util import *
308 import common
309 klass_name = 'SchedulerGlite'
310 klass = importName(klass_name, klass_name)
311 common.scheduler = klass()
312
313 lfn="/store/user/"
314 node='T2_IT_Bari'
315 valid = ['node=','lfn=']
316 try:
317 opts, args = getopt.getopt(sys.argv[1:], "", valid)
318 except getopt.GetoptError, ex:
319 print str(ex)
320 sys.exit(1)
321 for o, a in opts:
322 if o == "--node":
323 node = a
324 if o == "--lfn":
325 lfn = a
326
327 mycfg_params = { 'USER.storage_element': node }
328 dsvc = PhEDExDatasvcInfo(mycfg_params)
329 dsvc.lfn = lfn
330 print dsvc.getStageoutPFN()
331