| 1 |
"""
|
| 2 |
CRAB interface to BossLite gLite Scheduler
|
| 3 |
"""
|
| 4 |
|
| 5 |
__revision__ = "$Id: SchedulerGlite.py,v 1.72 2010/01/13 18:40:41 spiga Exp $"
|
| 6 |
__version__ = "$Revision: 1.72 $"
|
| 7 |
|
| 8 |
from SchedulerGrid import SchedulerGrid
|
| 9 |
from crab_exceptions import *
|
| 10 |
from crab_util import *
|
| 11 |
from GliteConfig import *
|
| 12 |
import EdgLoggingInfo
|
| 13 |
import common
|
| 14 |
from WMCore.SiteScreening.BlackWhiteListParser import CEBlackWhiteListParser
|
| 15 |
|
| 16 |
import os, sys, time
|
| 17 |
|
| 18 |
class SchedulerGlite(SchedulerGrid):
|
| 19 |
def __init__(self, name="GLITE"):
|
| 20 |
SchedulerGrid.__init__(self,name)
|
| 21 |
|
| 22 |
self.OSBsize = 55000000
|
| 23 |
|
| 24 |
def configure(self,cfg_params):
|
| 25 |
SchedulerGrid.configure(self, cfg_params)
|
| 26 |
self.environment_unique_identifier = '$GLITE_WMS_JOBID'
|
| 27 |
|
| 28 |
def realSchedParams(self,cfg_params):
|
| 29 |
"""
|
| 30 |
Return dictionary with specific parameters, to use
|
| 31 |
with real scheduler
|
| 32 |
"""
|
| 33 |
self.rb_param_file=''
|
| 34 |
if (not cfg_params.has_key('GRID.rb')):
|
| 35 |
cfg_params['GRID.rb']='CERN'
|
| 36 |
self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("GRID.rb"))
|
| 37 |
self.wms_service=cfg_params.get("GRID.wms_service",'')
|
| 38 |
self.skipWMSAuth=cfg_params.get("GRID.skipwmsauth",1)
|
| 39 |
params = { 'service' : self.wms_service, \
|
| 40 |
'config' : self.rb_param_file, \
|
| 41 |
'skipWMSAuth' : self.skipWMSAuth
|
| 42 |
}
|
| 43 |
return params
|
| 44 |
|
| 45 |
|
| 46 |
def rb_configure(self, RB):
|
| 47 |
## 25-Jun-2009 SL: patch to use Cream enabled WMS
|
| 48 |
if ( self.cfg_params.get('GRID.use_cream',None) ):
|
| 49 |
RB='CREAM'
|
| 50 |
if not RB: return None
|
| 51 |
glite_config = None
|
| 52 |
rb_param_file = None
|
| 53 |
|
| 54 |
gliteConfig = GliteConfig(RB)
|
| 55 |
glite_config = gliteConfig.config()
|
| 56 |
|
| 57 |
if (glite_config ):
|
| 58 |
rb_param_file = glite_config
|
| 59 |
return rb_param_file
|
| 60 |
|
| 61 |
def ce_list(self):
|
| 62 |
"""
|
| 63 |
Returns string with requirement CE related
|
| 64 |
"""
|
| 65 |
ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
|
| 66 |
self.EDG_ce_black_list, common.logger())
|
| 67 |
req = ''
|
| 68 |
ce_white_list = []
|
| 69 |
ce_black_list = []
|
| 70 |
if self.EDG_ce_white_list:
|
| 71 |
ce_white_list = ceParser.whiteList()
|
| 72 |
tmpCe=[]
|
| 73 |
concString = '&&'
|
| 74 |
for ce in ce_white_list:
|
| 75 |
tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
|
| 76 |
if len(tmpCe) == 1:
|
| 77 |
req += " && (" + concString.join(tmpCe) + ") "
|
| 78 |
elif len(tmpCe) > 1:
|
| 79 |
firstCE = 0
|
| 80 |
for reqTemp in tmpCe:
|
| 81 |
if firstCE == 0:
|
| 82 |
req += " && ( (" + reqTemp + ") "
|
| 83 |
firstCE = 1
|
| 84 |
elif firstCE > 0:
|
| 85 |
req += " || (" + reqTemp + ") "
|
| 86 |
if firstCE > 0:
|
| 87 |
req += ") "
|
| 88 |
|
| 89 |
if self.EDG_ce_black_list:
|
| 90 |
ce_black_list = ceParser.blackList()
|
| 91 |
tmpCe=[]
|
| 92 |
concString = '&&'
|
| 93 |
for ce in ce_black_list:
|
| 94 |
tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
|
| 95 |
if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
|
| 96 |
|
| 97 |
# requirement added to skip gliteCE
|
| 98 |
# not more needed
|
| 99 |
# req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
|
| 100 |
retWL = ','.join(ce_white_list)
|
| 101 |
retBL = ','.join(ce_black_list)
|
| 102 |
if not retWL:
|
| 103 |
retWL = None
|
| 104 |
if not retBL:
|
| 105 |
retBL = None
|
| 106 |
|
| 107 |
return req, retWL, retBL
|
| 108 |
|
| 109 |
def se_list(self, dest):
|
| 110 |
"""
|
| 111 |
Returns string with requirement SE related
|
| 112 |
"""
|
| 113 |
hostList=self.findSites_(dest)
|
| 114 |
req=''
|
| 115 |
reqtmp=[]
|
| 116 |
concString = '||'
|
| 117 |
|
| 118 |
for arg in hostList:
|
| 119 |
reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
|
| 120 |
|
| 121 |
if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "
|
| 122 |
|
| 123 |
return req
|
| 124 |
|
| 125 |
def jdlParam(self):
|
| 126 |
"""
|
| 127 |
Returns
|
| 128 |
"""
|
| 129 |
req=''
|
| 130 |
if self.EDG_addJdlParam:
|
| 131 |
if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
|
| 132 |
for p in self.EDG_addJdlParam:
|
| 133 |
req+=string.strip(p)+';\n'
|
| 134 |
return req
|
| 135 |
|
| 136 |
def specific_req(self):
|
| 137 |
"""
|
| 138 |
Returns string with specific requirements
|
| 139 |
"""
|
| 140 |
req=''
|
| 141 |
if self.EDG_clock_time:
|
| 142 |
if (not req == ' '): req = req + ' && '
|
| 143 |
req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
|
| 144 |
|
| 145 |
if self.EDG_cpu_time:
|
| 146 |
if (not req == ' '): req = req + ' && '
|
| 147 |
req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
|
| 148 |
|
| 149 |
return req
|
| 150 |
|
| 151 |
def sched_parameter(self,i,task):
|
| 152 |
"""
|
| 153 |
Returns string with requirements and scheduler-specific parameters
|
| 154 |
"""
|
| 155 |
dest= task.jobs[i-1]['dlsDestination']
|
| 156 |
|
| 157 |
req=''
|
| 158 |
req +=task['jobType']
|
| 159 |
|
| 160 |
sched_param=''
|
| 161 |
sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
|
| 162 |
self.ce_list()[0] +';\n'
|
| 163 |
if self.EDG_addJdlParam: sched_param+=self.jdlParam()
|
| 164 |
sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
|
| 165 |
sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
|
| 166 |
sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
|
| 167 |
sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'
|
| 168 |
|
| 169 |
return sched_param
|
| 170 |
|
| 171 |
def decodeLogInfo(self, file):
|
| 172 |
"""
|
| 173 |
Parse logging info file and return main info
|
| 174 |
"""
|
| 175 |
loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
|
| 176 |
reason = loggingInfo.decodeReason(file)
|
| 177 |
return reason
|
| 178 |
|
| 179 |
def findSites_(self, sites):
|
| 180 |
itr4 =[]
|
| 181 |
if len(sites)>0 and sites[0]=="":
|
| 182 |
return itr4
|
| 183 |
if sites != [""]:
|
| 184 |
replicas = self.blackWhiteListParser.checkBlackList(sites)
|
| 185 |
if len(replicas)!=0:
|
| 186 |
replicas = self.blackWhiteListParser.checkWhiteList(replicas)
|
| 187 |
|
| 188 |
itr4 = replicas
|
| 189 |
return itr4
|
| 190 |
|
| 191 |
def delegateProxy(self):
|
| 192 |
self.boss().delegateProxy()
|
| 193 |
return
|
| 194 |
|
| 195 |
def wsExitFunc(self):
|
| 196 |
"""
|
| 197 |
"""
|
| 198 |
txt = '\n'
|
| 199 |
|
| 200 |
txt += '#\n'
|
| 201 |
txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
|
| 202 |
txt += '#\n\n'
|
| 203 |
|
| 204 |
txt += 'func_exit() { \n'
|
| 205 |
txt += self.wsExitFunc_common()
|
| 206 |
### specific Glite check for OSB
|
| 207 |
txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
|
| 208 |
txt += ' tmp_size=`ls -gGrta ${out_files}.tgz | awk \'{ print $3 }\'`\n'
|
| 209 |
txt += ' rm ${out_files}.tgz\n'
|
| 210 |
txt += ' size=`expr $tmp_size`\n'
|
| 211 |
txt += ' echo "Total Output dimension: $size"\n'
|
| 212 |
txt += ' limit='+str(self.OSBsize) +' \n'
|
| 213 |
txt += ' echo "WARNING: output files size limit is set to: $limit"\n'
|
| 214 |
txt += ' if [ "$limit" -lt "$size" ]; then\n'
|
| 215 |
txt += ' exceed=1\n'
|
| 216 |
txt += ' job_exit_code=70000\n'
|
| 217 |
txt += ' echo "Output Sanbox too big. Produced output is lost "\n'
|
| 218 |
txt += ' else\n'
|
| 219 |
txt += ' exceed=0\n'
|
| 220 |
txt += ' echo "Total Output dimension $size is fine."\n'
|
| 221 |
txt += ' fi\n'
|
| 222 |
|
| 223 |
txt += ' echo "JOB_EXIT_STATUS = $job_exit_code"\n'
|
| 224 |
txt += ' echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
|
| 225 |
txt += ' dumpStatus $RUNTIME_AREA/$repo\n'
|
| 226 |
txt += ' if [ $exceed -ne 1 ]; then\n'
|
| 227 |
txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
|
| 228 |
txt += ' else\n'
|
| 229 |
txt += ' tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
|
| 230 |
txt += ' fi\n'
|
| 231 |
txt += ' python $RUNTIME_AREA/fillCrabFjr.py $RUNTIME_AREA/crab_fjr_$NJob.xml --errorcode $job_exit_code \n'
|
| 232 |
txt += ' exit $job_exit_code\n'
|
| 233 |
|
| 234 |
txt += '}\n'
|
| 235 |
return txt
|
