| 1 |
from Scheduler import Scheduler
|
| 2 |
from SchedulerLocal import SchedulerLocal
|
| 3 |
from crab_exceptions import *
|
| 4 |
from crab_util import *
|
| 5 |
from crab_logger import Logger
|
| 6 |
import common
|
| 7 |
|
| 8 |
import os,string
|
| 9 |
|
| 10 |
#
|
| 11 |
# Naming convention:
|
| 12 |
# methods starting with 'ws' are responsible to provide
|
| 13 |
# corresponding part of the job script ('ws' stands for 'write script').
|
| 14 |
#
|
| 15 |
# Author: Hartmut Stadie <stadie@mail.desy.de> Inst. f. Experimentalphysik; Universitaet Hamburg
|
| 16 |
#
|
| 17 |
|
| 18 |
class SchedulerSge(SchedulerLocal) :
|
| 19 |
|
| 20 |
def __init__(self):
|
| 21 |
Scheduler.__init__(self,"SGE")
|
| 22 |
self.datasetPath = None
|
| 23 |
self.selectNoInput = None
|
| 24 |
return
|
| 25 |
|
| 26 |
def configure(self, cfg_params):
|
| 27 |
SchedulerLocal.configure(self, cfg_params)
|
| 28 |
|
| 29 |
try:
|
| 30 |
tmp = cfg_params['CMSSW.datasetpath']
|
| 31 |
if tmp.lower() == 'none':
|
| 32 |
self.datasetPath = None
|
| 33 |
self.selectNoInput = 1
|
| 34 |
else:
|
| 35 |
self.datasetPath = tmp
|
| 36 |
self.selectNoInput = 0
|
| 37 |
except KeyError:
|
| 38 |
msg = "Error: datasetpath not defined "
|
| 39 |
raise CrabException(msg)
|
| 40 |
|
| 41 |
self.return_data = cfg_params.get('USER.return_data', 0)
|
| 42 |
self.copy_data = cfg_params.get("USER.copy_data", 0)
|
| 43 |
|
| 44 |
if ( int(self.return_data) == 0 and int(self.copy_data) == 0 ):
|
| 45 |
msg = 'Error: return_data and copy_data cannot be set both to 0\n'
|
| 46 |
msg = msg + 'Please modify your crab.cfg file\n'
|
| 47 |
raise CrabException(msg)
|
| 48 |
|
| 49 |
if ( int(self.return_data) == 1 and int(self.copy_data) == 1 ):
|
| 50 |
msg = 'Error: return_data and copy_data cannot be set both to 1\n'
|
| 51 |
msg = msg + 'Please modify your crab.cfg file\n'
|
| 52 |
raise CrabException(msg)
|
| 53 |
|
| 54 |
if ( int(self.copy_data) == 0 and int(self.publish_data) == 1 ):
|
| 55 |
msg = 'Warning: publish_data = 1 must be used with copy_data = 1\n'
|
| 56 |
msg = msg + 'Please modify copy_data value in your crab.cfg file\n'
|
| 57 |
common.logger.message(msg)
|
| 58 |
raise CrabException(msg)
|
| 59 |
|
| 60 |
if int(self.copy_data) == 1:
|
| 61 |
self.SE = cfg_params.get('USER.storage_element', None)
|
| 62 |
if not self.SE:
|
| 63 |
msg = "Error. The [USER] section has no 'storage_element'"
|
| 64 |
common.logger.message(msg)
|
| 65 |
raise CrabException(msg)
|
| 66 |
|
| 67 |
self.proxyValid = 0
|
| 68 |
self.dontCheckProxy = int(cfg_params.get("EDG.dont_check_proxy",0))
|
| 69 |
self.proxyServer = cfg_params.get("EDG.proxy_server",'myproxy.cern.ch')
|
| 70 |
common.logger.debug(5,'Setting myproxy server to ' + self.proxyServer)
|
| 71 |
|
| 72 |
self.group = cfg_params.get("EDG.group", None)
|
| 73 |
self.role = cfg_params.get("EDG.role", None)
|
| 74 |
self.VO = cfg_params.get('EDG.virtual_organization', 'cms')
|
| 75 |
|
| 76 |
self.checkProxy()
|
| 77 |
|
| 78 |
self.role = None
|
| 79 |
|
| 80 |
self.pool = cfg_params.get('USER.storage_pool',None)
|
| 81 |
# self.cpu = cfg_params.get('USER.cpu',172800)
|
| 82 |
# self.vmem = cfg_params.get('USER.vmem',2)
|
| 83 |
return
|
| 84 |
|
| 85 |
def envUniqueID(self):
|
| 86 |
id = "https://"+common.scheduler.name()+":/${JOB_ID}-"+ \
|
| 87 |
string.replace(common._db.queryTask('name'),"_","-")
|
| 88 |
return id
|
| 89 |
|
| 90 |
def realSchedParams(self,cfg_params):
|
| 91 |
"""
|
| 92 |
Return dictionary with specific parameters, to use
|
| 93 |
with real scheduler
|
| 94 |
"""
|
| 95 |
params = {}
|
| 96 |
return params
|
| 97 |
|
| 98 |
def sched_parameter(self,i,task):
|
| 99 |
"""
|
| 100 |
Returns parameter scheduler-specific, to use with BOSS .
|
| 101 |
"""
|
| 102 |
index = int(common._db.nJobs()) - 1
|
| 103 |
sched_param= ''
|
| 104 |
|
| 105 |
for i in range(index): # Add loop DS
|
| 106 |
sched_param= ''
|
| 107 |
if (self.queue):
|
| 108 |
sched_param += '-q '+self.queue +' '
|
| 109 |
if (self.res): sched_param += ' -l '+self.res +' '
|
| 110 |
pass
|
| 111 |
|
| 112 |
#default is request 2G memory and 48 hours CPU time
|
| 113 |
#sched_param += ' -V -l h_vmem=2G -l h_cpu=172800 '
|
| 114 |
# sched_param += ' -V -l h_vmem='
|
| 115 |
# sched_param += self.vmem.__str__()
|
| 116 |
# sched_param += 'G -l h_cpu='
|
| 117 |
# sched_param += self.cpu.__str__()
|
| 118 |
# sched_param += ' '
|
| 119 |
|
| 120 |
return sched_param
|
| 121 |
|
| 122 |
def loggingInfo(self, id):
|
| 123 |
""" return logging info about job nj """
|
| 124 |
print "Warning: SchedulerSge::loggingInfo not implemented!"
|
| 125 |
return ""
|
| 126 |
|
| 127 |
def wsExitFunc(self):
|
| 128 |
"""
|
| 129 |
"""
|
| 130 |
txt = '\n'
|
| 131 |
|
| 132 |
txt += '#\n'
|
| 133 |
txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
|
| 134 |
txt += '#\n\n'
|
| 135 |
|
| 136 |
txt += 'func_exit() { \n'
|
| 137 |
txt += self.wsExitFunc_common()
|
| 138 |
|
| 139 |
txt += ' cp ${SGE_STDOUT_PATH} CMSSW_${NJob}.stdout \n'
|
| 140 |
txt += ' cp ${SGE_STDERR_PATH} CMSSW_${NJob}.stderr \n'
|
| 141 |
txt += ' tar zcvf ${out_files}.tgz ${filesToCheck}\n'
|
| 142 |
txt += ' exit $job_exit_code\n'
|
| 143 |
txt += '}\n'
|
| 144 |
|
| 145 |
return txt
|
| 146 |
|
| 147 |
def listMatch(self, dest, full):
|
| 148 |
"""
|
| 149 |
"""
|
| 150 |
#if len(dest)!=0:
|
| 151 |
sites = [self.blackWhiteListParser.cleanForBlackWhiteList(dest,'list')]
|
| 152 |
#else:
|
| 153 |
# sites = [str(getLocalDomain(self))]
|
| 154 |
return sites
|
| 155 |
|
| 156 |
def wsCopyOutput(self):
|
| 157 |
txt=self.wsCopyOutput_comm(self.pool)
|
| 158 |
return txt
|
| 159 |
|
| 160 |
def userName(self):
|
| 161 |
""" return the user name """
|
| 162 |
|
| 163 |
## hack for german naf
|
| 164 |
import pwd,getpass
|
| 165 |
tmp=pwd.getpwnam(getpass.getuser())[4]
|
| 166 |
tmp=tmp.rstrip(',')
|
| 167 |
tmp=tmp.rstrip(',')
|
| 168 |
tmp=tmp.rstrip(',')
|
| 169 |
|
| 170 |
|
| 171 |
return "/CN="+tmp.strip()
|
