| 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 * |
| 14 |
|
self.datasvc_url = cfg_params.get("USER.datasvc_url",url) |
| 15 |
|
|
| 16 |
|
self.FacOps_savannah = 'https://savannah.cern.ch/support/?func=additem&group=cmscompinfrasup' |
| 17 |
< |
|
| 17 |
> |
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.srm_version = cfg_params.get("USER.srm_version",'srmv2') |
| 21 |
|
self.node = cfg_params.get('USER.storage_element',None) |
| 22 |
|
|
| 23 |
+ |
|
| 24 |
+ |
|
| 25 |
+ |
self.user_lfn = cfg_params.get("USER.lfn",'') |
| 26 |
|
self.publish_data = cfg_params.get("USER.publish_data",0) |
| 27 |
|
self.usenamespace = cfg_params.get("USER.usenamespace",0) |
| 28 |
< |
self.user_remote_dir = cfg_params.get("USER.remote_dir",'') |
| 28 |
> |
self.user_remote_dir = cfg_params.get("USER.user_remote_dir",'') |
| 29 |
|
if self.user_remote_dir: |
| 30 |
|
if ( self.user_remote_dir[-1] != '/' ) : self.user_remote_dir = self.user_remote_dir + '/' |
| 31 |
< |
|
| 31 |
> |
if self.user_lfn: |
| 32 |
> |
msg = 'Warning: lfn has been deprecated, CRAB will ignore it.\n' |
| 33 |
> |
msg += '\t Please use only user_remote_dir removing lfn from your crab.cfg\n' |
| 34 |
> |
msg += '\t For further information please visit : \n\t%s'%stage_out_faq |
| 35 |
> |
common.logger.info(msg) |
| 36 |
> |
|
| 37 |
|
self.datasetpath = cfg_params.get("CMSSW.datasetpath") |
| 38 |
|
self.publish_data_name = cfg_params.get('USER.publish_data_name','') |
| 39 |
|
|
| 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 |
– |
|
| 40 |
|
self.user_port = cfg_params.get("USER.storage_port",'8443') |
| 41 |
|
self.user_se_path = cfg_params.get("USER.storage_path",'') |
| 42 |
|
if self.user_se_path: |
| 43 |
|
if ( self.user_se_path[-1] != '/' ) : self.user_se_path = self.user_se_path + '/' |
| 44 |
|
|
| 41 |
– |
|
| 45 |
|
#check if using "private" Storage |
| 46 |
|
self.usePhedex = True |
| 47 |
+ |
if not self.node : |
| 48 |
+ |
msg = 'Please specify the storage_element name in your crab.cfg section [USER].\n' |
| 49 |
+ |
msg +='\tFor further information please visit : %s'%stage_out_faq |
| 50 |
+ |
raise CrabException(msg) |
| 51 |
|
if (self.node.find('T1_') + self.node.find('T2_')+self.node.find('T3_')) == -3: self.usePhedex = False |
| 52 |
< |
if not self.usePhedex and ( self.user_lfn == '' or self.user_se_path == '' ): |
| 52 |
> |
|
| 53 |
> |
if not self.usePhedex and ( self.user_remote_dir == '' or self.user_se_path == '' ): |
| 54 |
|
msg = 'You are asking to stage out without using CMS Storage Name convention. In this case you \n' |
| 55 |
< |
msg += ' must specify both lfn and storage_path in the crab.cfg section [USER].\n ' |
| 56 |
< |
msg += ' For further information please visit: ADD_TWIKI_LINK' |
| 55 |
> |
msg += '\t must specify both user_remote_dir and storage_path in the crab.cfg section [USER].\n ' |
| 56 |
> |
msg += '\t For further information please visit : \n\t%s'%stage_out_faq |
| 57 |
|
raise CrabException(msg) |
| 58 |
|
self.sched = common.scheduler.name().upper() |
| 59 |
|
self.protocol = self.srm_version |
| 52 |
– |
if self.sched in ['CAF','LSF']:self.protocol = 'direct' |
| 60 |
|
|
| 61 |
+ |
self.forced_path = '/store/user/' |
| 62 |
+ |
if self.sched in ['CAF','LSF']: |
| 63 |
+ |
self.protocol = 'direct' |
| 64 |
+ |
self.SE = {'CAF':'caf.cern.ch', 'LSF':''} |
| 65 |
+ |
if self.sched == 'CAF': self.forced_path = '/store/caf/user/' |
| 66 |
+ |
|
| 67 |
+ |
if not self.usePhedex: |
| 68 |
+ |
self.forced_path = self.user_remote_dir |
| 69 |
|
return |
| 70 |
|
|
| 71 |
|
def getEndpoint(self): |
| 93 |
|
if self.protocol == 'direct': |
| 94 |
|
query=endpoint |
| 95 |
|
SE_PATH = endpoint |
| 96 |
< |
### FEDE added SE ### |
| 82 |
< |
SE = self.sched |
| 96 |
> |
SE = self.SE[self.sched] |
| 97 |
|
else: |
| 98 |
|
url = 'http://'+endpoint.split('://')[1] |
| 99 |
|
# python > 2.4 |
| 103 |
|
SE_PATH = endpoint.split(host)[1] |
| 104 |
|
USER = (query.split('user')[1]).split('/')[1] |
| 105 |
|
else: |
| 106 |
+ |
#### to test ##### |
| 107 |
+ |
# url = 'http://'+endpoint.split('://')[1] |
| 108 |
+ |
# scheme, host, path, params, query, fragment = urlparse(url) |
| 109 |
+ |
# SE = host.split(':')[0] |
| 110 |
+ |
# SE_PATH = endpoint.split(host)[1] |
| 111 |
|
SE = self.node |
| 112 |
< |
SE_PATH = self.user_se_path + self.user_lfn |
| 112 |
> |
SE_PATH = self.user_se_path + self.user_remote_dir |
| 113 |
|
try: |
| 114 |
|
USER = (self.lfn.split('user')[1]).split('/')[1] |
| 115 |
|
except: |
| 129 |
|
## check if storage_name is a T2 (siteDB query) |
| 130 |
|
## if yes :match self.user_lfn with LFNBaseName... |
| 131 |
|
## if NOT : raise (you are using a T2. It's not allowed stage out into self.user_path+self.user_lfn) |
| 132 |
< |
lfn = self.user_lfn |
| 132 |
> |
lfn = self.user_remote_dir |
| 133 |
|
return lfn |
| 134 |
|
if self.publish_data_name == '' and int(self.publish_data) == 1: |
| 135 |
< |
msg = "Eeror. The [USER] section does not have 'publish_data_name'" |
| 135 |
> |
msg = "Error. The [USER] section does not have 'publish_data_name'\n" |
| 136 |
> |
msg += '\tFor further information please visit : \n\t%s'%self.dataPub_faq |
| 137 |
|
raise CrabException(msg) |
| 138 |
|
if self.publish_data_name == '' and int(self.usenamespace) == 1: |
| 139 |
|
self.publish_data_name = "DefaultDataset" |
| 140 |
< |
if int(self.publish_data) == 1 or int(self.usenamespace) == 1: |
| 140 |
> |
if int(self.publish_data) == 1: |
| 141 |
|
if self.sched in ['CAF']: l_User=True |
| 142 |
|
primaryDataset = self.computePrimaryDataset() |
| 143 |
< |
lfn = LFNBase(primaryDataset,self.publish_data_name,LocalUser=l_User) + '/${PSETHASH}/' |
| 143 |
> |
### added the case lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) |
| 144 |
> |
### for the publication in order to be able to check the lfn length |
| 145 |
> |
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name, publish=True) + '/${PSETHASH}/' |
| 146 |
> |
elif int(self.usenamespace) == 1: |
| 147 |
> |
if self.sched in ['CAF']: l_User=True |
| 148 |
> |
primaryDataset = self.computePrimaryDataset() |
| 149 |
> |
lfn = LFNBase(self.forced_path, primaryDataset, self.publish_data_name) + '/${PSETHASH}/' |
| 150 |
|
else: |
| 151 |
|
if self.sched in ['CAF','LSF']: l_User=True |
| 152 |
< |
lfn = LFNBase(self.user_remote_dir,LocalUser=l_User) |
| 152 |
> |
lfn = LFNBase(self.forced_path,self.user_remote_dir) |
| 153 |
|
return lfn |
| 154 |
|
|
| 155 |
|
def computePrimaryDataset(self): |
| 172 |
|
params = {'node' : self.node , 'lfn': self.lfn , 'protocol': self.protocol} |
| 173 |
|
params = urllib.urlencode(params) |
| 174 |
|
datasvc_lfn2pfn="%s/lfn2pfn"%self.datasvc_url |
| 149 |
– |
urlresults = urllib.urlopen(datasvc_lfn2pfn, params) |
| 175 |
|
try: |
| 176 |
+ |
urlresults = urllib.urlopen(datasvc_lfn2pfn, params) |
| 177 |
|
urlresults = parse(urlresults) |
| 178 |
+ |
except IOError: |
| 179 |
+ |
msg="Unable to access PhEDEx Data Service at %s"%datasvc_lfn2pfn |
| 180 |
+ |
raise CrabException(msg) |
| 181 |
|
except: |
| 182 |
|
urlresults = None |
| 183 |
|
|
| 240 |
|
msg+=' OriginalSubmission: stageout path is not retrieved from %s \n'%fullurl |
| 241 |
|
raise CrabException(msg) |
| 242 |
|
else: |
| 243 |
< |
stageoutpfn = 'srm://'+self.node+':'self.user_port+self.user_se_path+self.lfn |
| 243 |
> |
if self.sched in ['CAF','LSF'] : |
| 244 |
> |
stageoutpfn = self.user_se_path+self.lfn |
| 245 |
> |
else: |
| 246 |
> |
stageoutpfn = 'srm://'+self.node+':'+self.user_port+self.user_se_path+self.lfn |
| 247 |
|
|
| 248 |
|
return stageoutpfn |
| 249 |
|
|
