1 |
spiga |
1.3 |
from Actor import *
|
2 |
afanfani |
1.1 |
import urllib
|
3 |
|
|
from xml.dom.minidom import parse
|
4 |
|
|
from crab_exceptions import *
|
5 |
spiga |
1.3 |
from WorkSpace import *
|
6 |
|
|
from urlparse import urlparse
|
7 |
|
|
from LFNBaseName import *
|
8 |
fanzago |
1.36 |
from crab_util import getUserName
|
9 |
afanfani |
1.1 |
|
10 |
|
|
class PhEDExDatasvcInfo:
|
11 |
fanzago |
1.28 |
def __init__( self , cfg_params=None, config=None ):
|
12 |
spiga |
1.3 |
|
13 |
|
|
## PhEDEx Data Service URL
|
14 |
belforte |
1.38 |
self.datasvc_url="https://cmsweb-testbed.cern.ch/phedex/datasvc/xml/prod"
|
15 |
spiga |
1.3 |
|
16 |
afanfani |
1.6 |
self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup'
|
17 |
spiga |
1.34 |
self.stage_out_faq='https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabHowTo#Stageout_and_publication'
|
18 |
spiga |
1.17 |
self.dataPub_faq = 'https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrabForPublication'
|
19 |
spiga |
1.3 |
|
20 |
fanzago |
1.28 |
self.usePhedex = True
|
21 |
spiga |
1.29 |
self.sched = common.scheduler.name().upper()
|
22 |
fanzago |
1.28 |
|
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 |
spiga |
1.3 |
self.srm_version = cfg_params.get("USER.srm_version",'srmv2')
|
43 |
|
|
self.node = cfg_params.get('USER.storage_element',None)
|
44 |
spiga |
1.17 |
|
45 |
spiga |
1.3 |
self.publish_data = cfg_params.get("USER.publish_data",0)
|
46 |
|
|
self.usenamespace = cfg_params.get("USER.usenamespace",0)
|
47 |
fanzago |
1.10 |
self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'')
|
48 |
fanzago |
1.7 |
if self.user_remote_dir:
|
49 |
|
|
if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/'
|
50 |
spiga |
1.17 |
|
51 |
spiga |
1.3 |
self.datasetpath = cfg_params.get("CMSSW.datasetpath")
|
52 |
|
|
self.publish_data_name = cfg_params.get('USER.publish_data_name','')
|
53 |
|
|
|
54 |
spiga |
1.8 |
self.user_port = cfg_params.get("USER.storage_port",'8443')
|
55 |
spiga |
1.3 |
self.user_se_path = cfg_params.get("USER.storage_path",'')
|
56 |
fanzago |
1.7 |
if self.user_se_path:
|
57 |
|
|
if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/'
|
58 |
|
|
|
59 |
spiga |
1.3 |
#check if using "private" Storage
|
60 |
spiga |
1.11 |
if not self.node :
|
61 |
|
|
msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n'
|
62 |
spiga |
1.34 |
msg +='\tFor further information please visit : %s'%self.stage_out_faq
|
63 |
spiga |
1.11 |
raise CrabException(msg)
|
64 |
spiga |
1.4 |
if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False
|
65 |
spiga |
1.15 |
|
66 |
|
|
if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ):
|
67 |
spiga |
1.3 |
msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n'
|
68 |
spiga |
1.15 |
msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n '
|
69 |
spiga |
1.34 |
msg += '\t For further information please visit : \n\t%s'%self.stage_out_faq
|
70 |
spiga |
1.3 |
raise CrabException(msg)
|
71 |
spiga |
1.22 |
|
72 |
|
|
self.forced_path = '/store/user/'
|
73 |
mcinquil |
1.32 |
if self.sched in ['CAF','LSF','PBS']:
|
74 |
spiga |
1.31 |
self.srm_version = 'direct'
|
75 |
mcinquil |
1.32 |
self.SE = {'CAF':'caf.cern.ch', 'LSF':'', 'PBS':''}
|
76 |
spiga |
1.22 |
if self.sched == 'CAF': self.forced_path = '/store/caf/user/'
|
77 |
|
|
|
78 |
spiga |
1.14 |
if not self.usePhedex:
|
79 |
spiga |
1.15 |
self.forced_path = self.user_remote_dir
|
80 |
spiga |
1.3 |
return
|
81 |
|
|
|
82 |
|
|
def getEndpoint(self):
|
83 |
|
|
'''
|
84 |
|
|
Return full SE endpoint and related infos
|
85 |
|
|
'''
|
86 |
|
|
self.lfn = self.getLFN()
|
87 |
|
|
|
88 |
|
|
#extract the PFN for the given node,LFN,protocol
|
89 |
|
|
endpoint = self.getStageoutPFN()
|
90 |
fanzago |
1.27 |
if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/'
|
91 |
spiga |
1.3 |
|
92 |
|
|
#extract SE name an SE_PATH (needed for publication)
|
93 |
|
|
SE, SE_PATH, User = self.splitEndpoint(endpoint)
|
94 |
|
|
|
95 |
|
|
return endpoint, self.lfn , SE, SE_PATH, User
|
96 |
|
|
|
97 |
|
|
def splitEndpoint(self, endpoint):
|
98 |
|
|
'''
|
99 |
|
|
Return relevant infos from endpoint
|
100 |
|
|
'''
|
101 |
|
|
SE = ''
|
102 |
|
|
SE_PATH = ''
|
103 |
spiga |
1.37 |
USER = getUserName()
|
104 |
spiga |
1.3 |
if self.usePhedex:
|
105 |
|
|
if self.protocol == 'direct':
|
106 |
|
|
query=endpoint
|
107 |
|
|
SE_PATH = endpoint
|
108 |
fanzago |
1.18 |
SE = self.SE[self.sched]
|
109 |
spiga |
1.3 |
else:
|
110 |
|
|
url = 'http://'+endpoint.split('://')[1]
|
111 |
|
|
scheme, host, path, params, query, fragment = urlparse(url)
|
112 |
spiga |
1.33 |
SE = self.getAuthoritativeSE()
|
113 |
spiga |
1.3 |
SE_PATH = endpoint.split(host)[1]
|
114 |
|
|
else:
|
115 |
fanzago |
1.36 |
#### to test #####
|
116 |
spiga |
1.21 |
# url = 'http://'+endpoint.split('://')[1]
|
117 |
|
|
# scheme, host, path, params, query, fragment = urlparse(url)
|
118 |
|
|
# SE = host.split(':')[0]
|
119 |
|
|
# SE_PATH = endpoint.split(host)[1]
|
120 |
spiga |
1.20 |
SE = self.node
|
121 |
|
|
SE_PATH = self.user_se_path + self.user_remote_dir
|
122 |
spiga |
1.37 |
if self.lfn.find('group') != -1:
|
123 |
|
|
try:
|
124 |
|
|
USER = (self.lfn.split('group')[1]).split('/')[1]
|
125 |
|
|
except:
|
126 |
|
|
pass
|
127 |
spiga |
1.3 |
return SE, SE_PATH, USER
|
128 |
|
|
|
129 |
|
|
def getLFN(self):
|
130 |
|
|
"""
|
131 |
|
|
define the LFN composing the needed pieces
|
132 |
|
|
"""
|
133 |
|
|
lfn = ''
|
134 |
|
|
l_User = False
|
135 |
|
|
if not self.usePhedex and (int(self.publish_data) == 0 and int(self.usenamespace) == 0) :
|
136 |
|
|
### add here check if user is trying to force a wrong LFN using a T2 TODO
|
137 |
|
|
## check if storage_name is a T2 (siteDB query)
|
138 |
|
|
## if yes :match self.user_lfn with LFNBaseName...
|
139 |
|
|
## if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn)
|
140 |
spiga |
1.15 |
lfn = self.user_remote_dir
|
141 |
spiga |
1.3 |
return lfn
|
142 |
|
|
if self.publish_data_name == '' and int(self.publish_data) == 1:
|
143 |
spiga |
1.17 |
msg = "Error. The [USER] section does not have 'publish_data_name'\n"
|
144 |
|
|
msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq
|
145 |
spiga |
1.3 |
raise CrabException(msg)
|
146 |
|
|
if self.publish_data_name == '' and int(self.usenamespace) == 1:
|
147 |
|
|
self.publish_data_name = "DefaultDataset"
|
148 |
fanzago |
1.26 |
if int(self.publish_data) == 1:
|
149 |
|
|
if self.sched in ['CAF']: l_User=True
|
150 |
|
|
primaryDataset = self.computePrimaryDataset()
|
151 |
|
|
### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True)
|
152 |
|
|
### for the publication in order to be able to check the lfn length
|
153 |
|
|
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) + '/${PSETHASH}/'
|
154 |
|
|
elif int(self.usenamespace) == 1:
|
155 |
spiga |
1.3 |
if self.sched in ['CAF']: l_User=True
|
156 |
|
|
primaryDataset = self.computePrimaryDataset()
|
157 |
spiga |
1.21 |
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name) + '/${PSETHASH}/'
|
158 |
spiga |
1.3 |
else:
|
159 |
spiga |
1.5 |
if self.sched in ['CAF','LSF']: l_User=True
|
160 |
spiga |
1.21 |
lfn = LFNBase(self.forced_path,self.user_remote_dir)
|
161 |
fanzago |
1.35 |
if ( lfn[-1] != '/' ) : lfn = lfn + '/'
|
162 |
spiga |
1.3 |
return lfn
|
163 |
|
|
|
164 |
|
|
def computePrimaryDataset(self):
|
165 |
|
|
"""
|
166 |
|
|
compute the last part for the LFN in case of publication
|
167 |
|
|
"""
|
168 |
|
|
if (self.datasetpath.upper() != 'NONE'):
|
169 |
|
|
primarydataset = self.datasetpath.split("/")[1]
|
170 |
|
|
else:
|
171 |
|
|
primarydataset = self.publish_data_name
|
172 |
|
|
return primarydataset
|
173 |
|
|
|
174 |
spiga |
1.33 |
def domPhedex(self,params,datasvc_baseUrl):
|
175 |
spiga |
1.3 |
"""
|
176 |
|
|
PhEDEx Data Service lfn2pfn call
|
177 |
|
|
|
178 |
spiga |
1.33 |
input: params,datasvc_baseUrl
|
179 |
spiga |
1.3 |
returns: DOM object with the content of the PhEDEx Data Service call
|
180 |
|
|
"""
|
181 |
|
|
params = urllib.urlencode(params)
|
182 |
|
|
try:
|
183 |
spiga |
1.33 |
urlresults = urllib.urlopen(datasvc_baseUrl, params)
|
184 |
spiga |
1.3 |
urlresults = parse(urlresults)
|
185 |
slacapra |
1.24 |
except IOError:
|
186 |
spiga |
1.33 |
msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl
|
187 |
slacapra |
1.24 |
raise CrabException(msg)
|
188 |
spiga |
1.3 |
except:
|
189 |
|
|
urlresults = None
|
190 |
|
|
|
191 |
|
|
return urlresults
|
192 |
afanfani |
1.1 |
|
193 |
spiga |
1.3 |
def parse_error(self,urlresults):
|
194 |
|
|
"""
|
195 |
|
|
look for errors in the DOM object returned by PhEDEx Data Service call
|
196 |
|
|
"""
|
197 |
|
|
errormsg = None
|
198 |
|
|
errors=urlresults.getElementsByTagName('error')
|
199 |
|
|
for error in errors:
|
200 |
|
|
errormsg=error.childNodes[0].data
|
201 |
|
|
if len(error.childNodes)>1:
|
202 |
|
|
errormsg+=error.childNodes[1].data
|
203 |
|
|
return errormsg
|
204 |
|
|
|
205 |
|
|
def parse_lfn2pfn(self,urlresults):
|
206 |
|
|
"""
|
207 |
|
|
Parse the content of the result of lfn2pfn PhEDEx Data Service call
|
208 |
|
|
|
209 |
|
|
input: DOM object with the content of the lfn2pfn call
|
210 |
|
|
returns: PFN
|
211 |
|
|
"""
|
212 |
|
|
result = urlresults.getElementsByTagName('phedex')
|
213 |
|
|
|
214 |
|
|
if not result:
|
215 |
|
|
return []
|
216 |
|
|
result = result[0]
|
217 |
|
|
pfn = None
|
218 |
|
|
mapping = result.getElementsByTagName('mapping')
|
219 |
|
|
for m in mapping:
|
220 |
|
|
pfn=m.getAttribute("pfn")
|
221 |
|
|
if pfn:
|
222 |
|
|
return pfn
|
223 |
|
|
|
224 |
|
|
def getStageoutPFN( self ):
|
225 |
|
|
"""
|
226 |
|
|
input: LFN,node name,protocol
|
227 |
|
|
returns: PFN
|
228 |
|
|
"""
|
229 |
|
|
if self.usePhedex:
|
230 |
spiga |
1.33 |
params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol}
|
231 |
|
|
datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url
|
232 |
spiga |
1.3 |
fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol)
|
233 |
spiga |
1.33 |
domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn)
|
234 |
spiga |
1.3 |
if not domlfn2pfn :
|
235 |
|
|
msg="Unable to get info from %s"%fullurl
|
236 |
|
|
raise CrabException(msg)
|
237 |
|
|
|
238 |
|
|
errormsg = self.parse_error(domlfn2pfn)
|
239 |
|
|
if errormsg:
|
240 |
|
|
msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
|
241 |
|
|
raise CrabException(msg)
|
242 |
|
|
|
243 |
|
|
stageoutpfn = self.parse_lfn2pfn(domlfn2pfn)
|
244 |
|
|
if not stageoutpfn:
|
245 |
afanfani |
1.6 |
msg ='Unable to get stageout path from TFC at Site %s \n'%self.node
|
246 |
|
|
msg+=' Please alert the CompInfraSup group through their savannah %s \n'%self.FacOps_savannah
|
247 |
|
|
msg+=' reporting: \n'
|
248 |
|
|
msg+=' Summary: Unable to get user stageout from TFC at Site %s \n'%self.node
|
249 |
|
|
msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl
|
250 |
spiga |
1.3 |
raise CrabException(msg)
|
251 |
|
|
else:
|
252 |
mcinquil |
1.32 |
if self.sched in ['CAF','LSF','PBS'] :
|
253 |
spiga |
1.12 |
stageoutpfn = self.user_se_path+self.lfn
|
254 |
spiga |
1.11 |
else:
|
255 |
|
|
stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn
|
256 |
spiga |
1.3 |
|
257 |
fanzago |
1.35 |
if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/'
|
258 |
spiga |
1.3 |
return stageoutpfn
|
259 |
afanfani |
1.6 |
|
260 |
spiga |
1.33 |
def getAuthoritativeSE(self):
|
261 |
|
|
"""
|
262 |
|
|
input: node name
|
263 |
|
|
returns: AuthoritativeSE
|
264 |
|
|
"""
|
265 |
|
|
params = {'node' : self.node }
|
266 |
|
|
datasvc_nodes="%s/nodes"%self.datasvc_url
|
267 |
|
|
fullurl="%s/nodes/?node=%s"%(self.datasvc_url,self.node)
|
268 |
|
|
domnodes = self.domPhedex(params,datasvc_nodes)
|
269 |
|
|
|
270 |
|
|
if not domnodes :
|
271 |
|
|
msg="Unable to get info from %s"%fullurl
|
272 |
|
|
raise CrabException(msg)
|
273 |
|
|
|
274 |
|
|
errormsg = self.parse_error(domnodes)
|
275 |
|
|
if errormsg:
|
276 |
|
|
msg="Error extracting info from %s due to: %s"%(fullurl,errormsg)
|
277 |
|
|
raise CrabException(msg)
|
278 |
|
|
result = domnodes.getElementsByTagName('phedex')
|
279 |
|
|
if not result:
|
280 |
|
|
return []
|
281 |
|
|
result = result[0]
|
282 |
|
|
se = None
|
283 |
|
|
node = result.getElementsByTagName('node')
|
284 |
|
|
for m in node:
|
285 |
|
|
se=m.getAttribute("se")
|
286 |
|
|
if se:
|
287 |
|
|
return se
|
288 |
afanfani |
1.6 |
|
289 |
|
|
|
290 |
|
|
if __name__ == '__main__':
|
291 |
|
|
"""
|
292 |
|
|
Sort of unit testing to check Phedex API for whatever site and/or lfn.
|
293 |
|
|
Usage:
|
294 |
|
|
python PhEDExDatasvcInfo.py --node T2_IT_Bari --lfn /store/maremma
|
295 |
|
|
|
296 |
|
|
"""
|
297 |
|
|
import getopt,sys
|
298 |
|
|
from crab_util import *
|
299 |
|
|
import common
|
300 |
|
|
klass_name = 'SchedulerGlite'
|
301 |
|
|
klass = importName(klass_name, klass_name)
|
302 |
|
|
common.scheduler = klass()
|
303 |
|
|
|
304 |
|
|
lfn="/store/user/"
|
305 |
|
|
node='T2_IT_Bari'
|
306 |
|
|
valid = ['node=','lfn=']
|
307 |
|
|
try:
|
308 |
|
|
opts, args = getopt.getopt(sys.argv[1:], "", valid)
|
309 |
|
|
except getopt.GetoptError, ex:
|
310 |
|
|
print str(ex)
|
311 |
|
|
sys.exit(1)
|
312 |
|
|
for o, a in opts:
|
313 |
|
|
if o == "--node":
|
314 |
|
|
node = a
|
315 |
|
|
if o == "--lfn":
|
316 |
|
|
lfn = a
|
317 |
|
|
|
318 |
|
|
mycfg_params = { 'USER.storage_element': node }
|
319 |
|
|
dsvc = PhEDExDatasvcInfo(mycfg_params)
|
320 |
|
|
dsvc.lfn = lfn
|
321 |
|
|
print dsvc.getStageoutPFN()
|
322 |
|
|
|