| 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 |
+ |
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 |
< |
|
| 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.remote_dir",'') |
| 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 |
< |
|
| 50 |
> |
|
| 51 |
|
self.datasetpath = cfg_params.get("CMSSW.datasetpath") |
| 52 |
|
self.publish_data_name = cfg_params.get('USER.publish_data_name','') |
| 53 |
|
|
| 32 |
– |
self.user_lfn = cfg_params.get("USER.lfn",'') |
| 33 |
– |
if self.user_lfn: |
| 34 |
– |
if ( self.user_lfn[-1] != '/' ) : self.user_lfn = self.user_lfn + '/' |
| 35 |
– |
|
| 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 |
|
|
| 41 |
– |
|
| 59 |
|
#check if using "private" Storage |
| 60 |
< |
self.usePhedex = True |
| 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 |
< |
if not self.usePhedex and ( self.user_lfn == '' or self.user_se_path == '' ): |
| 65 |
> |
|
| 66 |
> |
if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ): |
| 67 |
|
msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n' |
| 68 |
< |
msg += ' must specify both lfn and storage_path in the crab.cfg section [USER].\n ' |
| 69 |
< |
msg += ' For further information please visit: ADD_TWIKI_LINK' |
| 68 |
> |
msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n ' |
| 69 |
> |
msg += '\t For further information please visit : \n\t%s'%self.stage_out_faq |
| 70 |
|
raise CrabException(msg) |
| 50 |
– |
self.sched = common.scheduler.name().upper() |
| 51 |
– |
self.protocol = self.srm_version |
| 52 |
– |
if self.sched in ['CAF','LSF']:self.protocol = 'direct' |
| 71 |
|
|
| 72 |
+ |
self.forced_path = '/store/user/' |
| 73 |
+ |
if self.sched in ['CAF','LSF','PBS']: |
| 74 |
+ |
self.srm_version = 'direct' |
| 75 |
+ |
self.SE = {'CAF':'caf.cern.ch', 'LSF':'', 'PBS':''} |
| 76 |
+ |
if self.sched == 'CAF': self.forced_path = '/store/caf/user/' |
| 77 |
+ |
|
| 78 |
+ |
if not self.usePhedex: |
| 79 |
+ |
self.forced_path = self.user_remote_dir |
| 80 |
|
return |
| 81 |
|
|
| 82 |
|
def getEndpoint(self): |
| 87 |
|
|
| 88 |
|
#extract the PFN for the given node,LFN,protocol |
| 89 |
|
endpoint = self.getStageoutPFN() |
| 90 |
+ |
if ( endpoint[-1] != '/' ) : endpoint = endpoint + '/' |
| 91 |
|
|
| 92 |
|
#extract SE name an SE_PATH (needed for publication) |
| 93 |
|
SE, SE_PATH, User = self.splitEndpoint(endpoint) |
| 100 |
|
''' |
| 101 |
|
SE = '' |
| 102 |
|
SE_PATH = '' |
| 103 |
< |
USER = '' |
| 103 |
> |
USER = getUserName() |
| 104 |
|
if self.usePhedex: |
| 105 |
|
if self.protocol == 'direct': |
| 106 |
|
query=endpoint |
| 107 |
|
SE_PATH = endpoint |
| 108 |
< |
### FEDE added SE ### |
| 82 |
< |
SE = self.sched |
| 108 |
> |
SE = self.SE[self.sched] |
| 109 |
|
else: |
| 110 |
|
url = 'http://'+endpoint.split('://')[1] |
| 85 |
– |
# python > 2.4 |
| 86 |
– |
# SE = urlparse(url).hostname |
| 111 |
|
scheme, host, path, params, query, fragment = urlparse(url) |
| 112 |
< |
SE = host.split(':')[0] |
| 112 |
> |
SE = self.getAuthoritativeSE() |
| 113 |
|
SE_PATH = endpoint.split(host)[1] |
| 90 |
– |
USER = (query.split('user')[1]).split('/')[1] |
| 114 |
|
else: |
| 115 |
+ |
#### to test ##### |
| 116 |
+ |
# 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 |
|
SE = self.node |
| 121 |
< |
SE_PATH = self.user_se_path + self.user_lfn |
| 122 |
< |
try: |
| 123 |
< |
USER = (self.lfn.split('user')[1]).split('/')[1] |
| 124 |
< |
except: |
| 125 |
< |
pass |
| 126 |
< |
|
| 121 |
> |
SE_PATH = self.user_se_path + self.user_remote_dir |
| 122 |
> |
if self.lfn.find('group') != -1: |
| 123 |
> |
try: |
| 124 |
> |
USER = (self.lfn.split('group')[1]).split('/')[1] |
| 125 |
> |
except: |
| 126 |
> |
pass |
| 127 |
|
return SE, SE_PATH, USER |
| 100 |
– |
|
| 128 |
|
|
| 129 |
|
def getLFN(self): |
| 130 |
|
""" |
| 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 |
< |
lfn = self.user_lfn |
| 140 |
> |
lfn = self.user_remote_dir |
| 141 |
|
return lfn |
| 142 |
|
if self.publish_data_name == '' and int(self.publish_data) == 1: |
| 143 |
< |
msg = "Eeror. The [USER] section does not have 'publish_data_name'" |
| 143 |
> |
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 |
|
raise CrabException(msg) |
| 146 |
|
if self.publish_data_name == '' and int(self.usenamespace) == 1: |
| 147 |
|
self.publish_data_name = "DefaultDataset" |
| 148 |
< |
if int(self.publish_data) == 1 or int(self.usenamespace) == 1: |
| 148 |
> |
if int(self.publish_data) == 1: |
| 149 |
|
if self.sched in ['CAF']: l_User=True |
| 150 |
|
primaryDataset = self.computePrimaryDataset() |
| 151 |
< |
lfn = LFNBase(primaryDataset,self.publish_data_name,LocalUser=l_User) + '/${PSETHASH}/' |
| 151 |
> |
### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) |
| 152 |
> |
### for the publication in order to be able to check the lfn length |
| 153 |
> |
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) + '/${PSETHASH}/' |
| 154 |
> |
elif int(self.usenamespace) == 1: |
| 155 |
> |
if self.sched in ['CAF']: l_User=True |
| 156 |
> |
primaryDataset = self.computePrimaryDataset() |
| 157 |
> |
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name) + '/${PSETHASH}/' |
| 158 |
|
else: |
| 159 |
|
if self.sched in ['CAF','LSF']: l_User=True |
| 160 |
< |
lfn = LFNBase(self.user_remote_dir,LocalUser=l_User) |
| 160 |
> |
lfn = LFNBase(self.forced_path,self.user_remote_dir) |
| 161 |
> |
if ( lfn[-1] != '/' ) : lfn = lfn + '/' |
| 162 |
|
return lfn |
| 163 |
|
|
| 164 |
|
def computePrimaryDataset(self): |
| 171 |
|
primarydataset = self.publish_data_name |
| 172 |
|
return primarydataset |
| 173 |
|
|
| 174 |
< |
def lfn2pfn(self): |
| 174 |
> |
def domPhedex(self,params,datasvc_baseUrl): |
| 175 |
|
""" |
| 176 |
|
PhEDEx Data Service lfn2pfn call |
| 177 |
|
|
| 178 |
< |
input: LFN,node name,protocol |
| 178 |
> |
input: params,datasvc_baseUrl |
| 179 |
|
returns: DOM object with the content of the PhEDEx Data Service call |
| 180 |
|
""" |
| 146 |
– |
params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol} |
| 181 |
|
params = urllib.urlencode(params) |
| 148 |
– |
datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url |
| 149 |
– |
urlresults = urllib.urlopen(datasvc_lfn2pfn, params) |
| 182 |
|
try: |
| 183 |
+ |
urlresults = urllib.urlopen(datasvc_baseUrl, params) |
| 184 |
|
urlresults = parse(urlresults) |
| 185 |
+ |
except IOError: |
| 186 |
+ |
msg="Unable to access PhEDEx Data Service at %s"%datasvc_baseUrl |
| 187 |
+ |
raise CrabException(msg) |
| 188 |
|
except: |
| 189 |
|
urlresults = None |
| 190 |
|
|
| 227 |
|
returns: PFN |
| 228 |
|
""" |
| 229 |
|
if self.usePhedex: |
| 230 |
+ |
params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol} |
| 231 |
+ |
datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url |
| 232 |
|
fullurl="%s/lfn2pfn?node=%s&lfn=%s&protocol=%s"%(self.datasvc_url,self.node,self.lfn,self.protocol) |
| 233 |
< |
domlfn2pfn = self.lfn2pfn() |
| 233 |
> |
domlfn2pfn = self.domPhedex(params,datasvc_lfn2pfn) |
| 234 |
|
if not domlfn2pfn : |
| 235 |
|
msg="Unable to get info from %s"%fullurl |
| 236 |
|
raise CrabException(msg) |
| 249 |
|
msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl |
| 250 |
|
raise CrabException(msg) |
| 251 |
|
else: |
| 252 |
< |
stageoutpfn = 'srm://'+self.node+':'self.user_port+self.user_se_path+self.lfn |
| 252 |
> |
if self.sched in ['CAF','LSF','PBS'] : |
| 253 |
> |
stageoutpfn = self.user_se_path+self.lfn |
| 254 |
> |
else: |
| 255 |
> |
stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn |
| 256 |
|
|
| 257 |
+ |
if ( stageoutpfn[-1] != '/' ) : stageoutpfn = stageoutpfn + '/' |
| 258 |
|
return stageoutpfn |
| 259 |
|
|
| 260 |
+ |
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 |
|
|
| 289 |
|
|
| 290 |
|
if __name__ == '__main__': |
