ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/COMP/CRAB/python/SchedulerGlite.py

Comparing COMP/CRAB/python/SchedulerGlite.py (file contents):
Revision 1.7 by slacapra, Tue Oct 17 09:52:57 2006 UTC vs.
Revision 1.67 by spiga, Tue May 26 10:23:01 2009 UTC

#	Line 1 | Line 1
1	<	from SchedulerEdg import SchedulerEdg
2	<	from crab_logger import Logger
1	>	"""
2	>	CRAB interface to BossLite gLite Scheduler
3	>	"""
4	>
5	>	__revision__ = "$Id$"
6	>	__version__ = "$Revision$"
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(SchedulerEdg):
19	<	def __init__(self):
20	<	SchedulerEdg.__init__(self)
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('EDG.rb')):
35	>	cfg_params['EDG.rb']='CERN'
36	>	self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("EDG.rb"))
37	>	self.wms_service=cfg_params.get("EDG.wms_service",'')
38	>	self.skipWMSAuth=cfg_params.get("EDG.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	<	self.glite_config = ''
48	<	self.rb_param_file = ''
47	>	if not RB: return None
48	>	glite_config = None
49	>	rb_param_file = None
50
51		gliteConfig = GliteConfig(RB)
52	<	self.glite_config = gliteConfig.config()
52	>	glite_config = gliteConfig.config()
53
54	<	if (self.glite_config != ''):
55	<	self.rb_param_file = 'WMSconfig = '+self.glite_config+';\n'
56	<	#print "rb_param_file = ", self.rb_param_file
57	<	return self.rb_param_file
58	<
59	<	def sched_parameter(self):
60	<	"""
61	<	Returns file with requirements and scheduler-specific parameters
62	<	"""
63	<	index = int(common.jobDB.nJobs()) - 1
31	<	job = common.job_list[index]
32	<	jbt = job.type()
33	<
34	<	lastDest=''
35	<	first = []
36	<	last  = []
37	<	for n in range(common.jobDB.nJobs()):
38	<	currDest=common.jobDB.destination(n)
39	<	if (currDest!=lastDest):
40	<	lastDest = currDest
41	<	first.append(n)
42	<	if n != 0:last.append(n-1)
43	<	if len(first)>len(last) :last.append(common.jobDB.nJobs())
44	<
54	>	if (glite_config ):
55	>	rb_param_file = glite_config
56	>	return rb_param_file
57	>
58	>	def ce_list(self):
59	>	"""
60	>	Returns string with requirement CE related
61	>	"""
62	>	ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
63	>	self.EDG_ce_black_list, common.logger)
64		req = ''
65	<	req = req + jbt.getRequirements()
66	<
48	<	if self.EDG_requirements:
49	<	if (req == ' '):
50	<	req = req + self.EDG_requirements
51	<	else:
52	<	req = req +  ' && ' + self.EDG_requirements
65	>	ce_white_list = []
66	>	ce_black_list = []
67		if self.EDG_ce_white_list:
68	<	ce_white_list = string.split(self.EDG_ce_white_list,',')
69	<	for i in range(len(ce_white_list)):
70	<	if i == 0:
71	<	if (req == ' '):
72	<	req = req + '((RegExp("' + ce_white_list[i] + '", other.GlueCEUniqueId))'
73	<	else:
74	<	req = req +  ' && ((RegExp("' + ce_white_list[i] + '", other.GlueCEUniqueId))'
75	<	pass
76	<	else:
77	<	req = req +  ' || (RegExp("' + ce_white_list[i] + '", other.GlueCEUniqueId))'
78	<	req = req + ')'
79	<
68	>	ce_white_list = ceParser.whiteList()
69	>	tmpCe=[]
70	>	concString = '&&'
71	>	for ce in ce_white_list:
72	>	tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
73	>	if len(tmpCe) == 1:
74	>	req +=  " && (" + concString.join(tmpCe) + ") "
75	>	elif len(tmpCe) > 1:
76	>	firstCE = 0
77	>	for reqTemp in tmpCe:
78	>	if firstCE == 0:
79	>	req += " && ( (" + reqTemp + ") "
80	>	firstCE = 1
81	>	elif firstCE > 0:
82	>	req += " || (" + reqTemp + ") "
83	>	if firstCE > 0:
84	>	req += ") "
85	>
86		if self.EDG_ce_black_list:
87	<	ce_black_list = string.split(self.EDG_ce_black_list,',')
87	>	ce_black_list = ceParser.blackList()
88	>	tmpCe=[]
89	>	concString = '&&'
90		for ce in ce_black_list:
91	<	if (req == ' '):
92	<	req = req + '(!RegExp("' + ce + '", other.GlueCEUniqueId))'
93	<	else:
94	<	req = req +  ' && (!RegExp("' + ce + '", other.GlueCEUniqueId))'
95	<	pass
91	>	tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
92	>	if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
93	>
94	>	# requirement added to skip gliteCE
95	>	req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
96	>	retWL = ','.join(ce_white_list)
97	>	retBL = ','.join(ce_black_list)
98	>	if not retWL:
99	>	retWL = None
100	>	if not retBL:
101	>	retBL = None
102	>
103	>	return req, retWL, retBL
104	>
105	>	def se_list(self, dest):
106	>	"""
107	>	Returns string with requirement SE related
108	>	"""
109	>	hostList=self.findSites_(dest)
110	>	req=''
111	>	reqtmp=[]
112	>	concString = '||'
113	>
114	>	for arg in hostList:
115	>	reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
116	>
117	>	if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "
118	>
119	>	return req
120	>
121	>	def jdlParam(self):
122	>	"""
123	>	Returns
124	>	"""
125	>	req=''
126	>	if self.EDG_addJdlParam:
127	>	if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
128	>	for p in self.EDG_addJdlParam:
129	>	req+=string.strip(p)+';\n'
130	>	return req
131	>
132	>	def specific_req(self):
133	>	"""
134	>	Returns string with specific requirements
135	>	"""
136	>	req=''
137		if self.EDG_clock_time:
138	<	if (req == ' '):
139	<	req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
77	<	else:
78	<	req = req + ' && other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
138	>	if (not req == ' '): req = req + ' && '
139	>	req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
140
141		if self.EDG_cpu_time:
142	<	if (req == ' '):
143	<	req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
83	<	else:
84	<	req = req + ' && other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
85	<
86	<	for i in range(len(first)): # Add loop DS
87	<	self.param='sched_param_'+str(i)+'.clad'
88	<	param_file = open(common.work_space.shareDir()+'/'+self.param, 'w')
89	<
90	<	itr4=self.findSites_(first[i])
91	<	reqSites=''
92	<	reqtmp=[]
93	<	j = 0
94	<	concString = '||'
95	<	for arg in itr4:
96	<	#############
97	<	# MC Changed matching syntax to avoid gang matching
98	<	#############
99	<	reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
100	<	if len(reqtmp): reqSites = reqSites + " && (" + concString.join(reqtmp) + ')'
101	<	reqSites = reqSites + ';\n'
102	<	param_file.write('Requirements = ' + req + reqSites )
103	<
104	<	if (self.rb_param_file != ''):
105	<	param_file.write(self.rb_param_file)
106	<
107	<	if len(self.EDG_addJdlParam):
108	<	for p in self.EDG_addJdlParam:
109	<	param_file.write(p)
110	<
111	<	param_file.close()
112	<
113	<	def wsSetupEnvironment(self):
114	<	"""
115	<	Returns part of a job script which does scheduler-specific work.
116	<	"""
117	<	txt = ''
118	<	txt += '# strip arguments\n'
119	<	txt += 'echo "strip arguments"\n'
120	<	txt += 'args=("$@")\n'
121	<	txt += 'nargs=$#\n'
122	<	txt += 'shift $nargs\n'
123	<	txt += "# job number (first parameter for job wrapper)\n"
124	<	txt += "NJob=${args[0]}\n"
125	<
126	<	txt += '# job identification to DashBoard \n'
127	<	txt += 'MonitorJobID=`echo ${NJob}_$GLITE_WMS_JOBID`\n'
128	<	txt += 'SyncGridJobId=`echo $GLITE_WMS_JOBID`\n'
129	<	txt += 'MonitorID=`echo ' + self._taskId + '`\n'
130	<	txt += 'echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
131	<	txt += 'echo "SyncGridJobId=`echo $SyncGridJobId`" | tee -a $RUNTIME_AREA/$repo \n'
132	<	txt += 'echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
133	<
134	<	txt += 'echo "middleware discovery " \n'
135	<	txt += 'if [ $VO_CMS_SW_DIR ]; then \n'
136	<	txt += '    middleware=LCG \n'
137	<	txt += '    echo "SyncCE=`glite-brokerinfo getCE`" | tee -a $RUNTIME_AREA/$repo \n'
138	<	txt += '    echo "GridFlavour=`echo $middleware`" | tee -a $RUNTIME_AREA/$repo \n'
139	<	txt += '    echo "middleware =$middleware" \n'
140	<	txt += 'elif [ $GRID3_APP_DIR ]; then\n'
141	<	txt += '    middleware=OSG \n'
142	<	txt += '    echo "SyncCE=`echo $EDG_WL_LOG_DESTINATION`" | tee -a $RUNTIME_AREA/$repo \n'
143	<	txt += '    echo "GridFlavour=`echo $middleware`" | tee -a $RUNTIME_AREA/$repo \n'
144	<	txt += '    echo "middleware =$middleware" \n'
145	<	txt += 'elif [ $OSG_APP ]; then \n'
146	<	txt += '    middleware=OSG \n'
147	<	txt += '    echo "SyncCE=`echo $EDG_WL_LOG_DESTINATION`" | tee -a $RUNTIME_AREA/$repo \n'
148	<	txt += '    echo "GridFlavour=`echo $middleware`" | tee -a $RUNTIME_AREA/$repo \n'
149	<	txt += '    echo "middleware =$middleware" \n'
150	<	txt += 'else \n'
151	<	txt += '    echo "SET_CMS_ENV 10030 ==> middleware not identified" \n'
152	<	txt += '    echo "JOB_EXIT_STATUS = 10030" \n'
153	<	txt += '    echo "JobExitCode=10030" | tee -a $RUNTIME_AREA/$repo \n'
154	<	txt += '    dumpStatus $RUNTIME_AREA/$repo \n'
155	<	txt += '    rm -f $RUNTIME_AREA/$repo \n'
156	<	txt += '    echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
157	<	txt += '    echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
158	<	txt += '    exit 1 \n'
159	<	txt += 'fi \n'
160	<
161	<	txt += '# report first time to DashBoard \n'
162	<	txt += 'dumpStatus $RUNTIME_AREA/$repo \n'
163	<	txt += 'rm -f $RUNTIME_AREA/$repo \n'
164	<	txt += 'echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
165	<	txt += 'echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
166	<
167	<	txt += '\n\n'
168	<
169	<	if int(self.copy_data) == 1:
170	<	if self.SE:
171	<	txt += 'export SE='+self.SE+'\n'
172	<	txt += 'echo "SE = $SE"\n'
173	<	if self.SE_PATH:
174	<	if ( self.SE_PATH[-1] != '/' ) : self.SE_PATH = self.SE_PATH + '/'
175	<	txt += 'export SE_PATH='+self.SE_PATH+'\n'
176	<	txt += 'echo "SE_PATH = $SE_PATH"\n'
177	<
178	<	txt += 'export VO='+self.VO+'\n'
179	<	### some line for LFC catalog setting
180	<	txt += 'if [ $middleware == LCG ]; then \n'
181	<	txt += '    if [[ $LCG_CATALOG_TYPE != \''+self.lcg_catalog_type+'\' ]]; then\n'
182	<	txt += '        export LCG_CATALOG_TYPE='+self.lcg_catalog_type+'\n'
183	<	txt += '    fi\n'
184	<	txt += '    if [[ $LFC_HOST != \''+self.lfc_host+'\' ]]; then\n'
185	<	txt += '        export LFC_HOST='+self.lfc_host+'\n'
186	<	txt += '    fi\n'
187	<	txt += '    if [[ $LFC_HOME != \''+self.lfc_home+'\' ]]; then\n'
188	<	txt += '        export LFC_HOME='+self.lfc_home+'\n'
189	<	txt += '    fi\n'
190	<	txt += 'elif [ $middleware == OSG ]; then\n'
191	<	txt += '    echo "LFC catalog setting to be implemented for OSG"\n'
192	<	txt += 'fi\n'
193	<	#####
194	<	if int(self.register_data) == 1:
195	<	txt += 'if [ $middleware == LCG ]; then \n'
196	<	txt += '    export LFN='+self.LFN+'\n'
197	<	txt += '    lfc-ls $LFN\n'
198	<	txt += '    result=$?\n'
199	<	txt += '    echo $result\n'
200	<	### creation of LFN dir in LFC catalog, under /grid/cms dir
201	<	txt += '    if [ $result != 0 ]; then\n'
202	<	txt += '       lfc-mkdir $LFN\n'
203	<	txt += '       result=$?\n'
204	<	txt += '       echo $result\n'
205	<	txt += '    fi\n'
206	<	txt += 'elif [ $middleware == OSG ]; then\n'
207	<	txt += '    echo " Files registration to be implemented for OSG"\n'
208	<	txt += 'fi\n'
209	<	txt += '\n'
210	<
211	<	if self.VO:
212	<	txt += 'export VO='+self.VO+'\n'
213	<	if self.LFN:
214	<	txt += 'if [ $middleware == LCG ]; then \n'
215	<	txt += '    export LFN='+self.LFN+'\n'
216	<	txt += 'fi\n'
217	<	txt += '\n'
218	<
219	<	txt += 'if [ $middleware == LCG ]; then\n'
220	<	txt += '    CloseCEs=`glite-brokerinfo getCE`\n'
221	<	txt += '    echo "CloseCEs = $CloseCEs"\n'
222	<	txt += '    CE=`echo $CloseCEs | sed -e "s/:.*//"`\n'
223	<	txt += '    echo "CE = $CE"\n'
224	<	txt += 'elif [ $middleware == OSG ]; then \n'
225	<	txt += '    if [ $OSG_JOB_CONTACT ]; then \n'
226	<	txt += '        CE=`echo $OSG_JOB_CONTACT | /usr/bin/awk -F\/ \'{print $1}\'` \n'
227	<	txt += '    else \n'
228	<	txt += '        echo "SET_CMS_ENV 10099 ==> OSG mode: ERROR in setting CE name from OSG_JOB_CONTACT" \n'
229	<	txt += '        echo "JOB_EXIT_STATUS = 10099" \n'
230	<	txt += '        echo "JobExitCode=10099" | tee -a $RUNTIME_AREA/$repo \n'
231	<	txt += '        dumpStatus $RUNTIME_AREA/$repo \n'
232	<	txt += '        rm -f $RUNTIME_AREA/$repo \n'
233	<	txt += '        echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
234	<	txt += '        echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
235	<	txt += '        exit 1 \n'
236	<	txt += '    fi \n'
237	<	txt += 'fi \n'
142	>	if (not req == ' '): req = req + ' && '
143	>	req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
144
145	<	return txt
146	<
147	<	def loggingInfo(self, id):
145	>	return req
146	>
147	>	def sched_parameter(self,i,task):
148	>	"""
149	>	Returns string with requirements and scheduler-specific parameters
150	>	"""
151	>	dest=  task.jobs[i-1]['dlsDestination']
152	>
153	>	req=''
154	>	req +=task['jobType']
155	>
156	>	sched_param=''
157	>	sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
158	>	self.ce_list()[0] +';\n'
159	>	if self.EDG_addJdlParam: sched_param+=self.jdlParam()
160	>	sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
161	>	sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
162	>	sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
163	>	sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'
164	>
165	>	return sched_param
166	>
167	>	def decodeLogInfo(self, file):
168		"""
169	<	retrieve the logging info from logging and bookkeeping and return it
169	>	Parse logging info file and return main info
170		"""
171	<	self.checkProxy()
172	<	cmd = 'glite-job-logging-info -v 2 ' + id
173	<	cmd_out = runCommand(cmd)
248	<	return cmd_out
249	<
250	<	def queryDetailedStatus(self, id):
251	<	""" Query a detailed status of the job with id """
252	<	cmd = 'glite-job-status '+id
253	<	cmd_out = runCommand(cmd)
254	<	return cmd_out
171	>	loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
172	>	reason = loggingInfo.decodeReason(file)
173	>	return reason
174
175	<	def findSites_(self, n):
176	<	sites = common.jobDB.destination(n)
175	>	def findSites_(self, sites):
176	>	itr4 =[]
177		if len(sites)>0 and sites[0]=="":
178	<	return []
179	<	return sites
178	>	return itr4
179	>	if sites != [""]:
180	>	replicas = self.blackWhiteListParser.checkBlackList(sites)
181	>	if len(replicas)!=0:
182	>	replicas = self.blackWhiteListParser.checkWhiteList(replicas)
183	>
184	>	itr4 = replicas
185	>	return itr4
186	>
187	>
188	>	def wsExitFunc(self):
189	>	"""
190	>	"""
191	>	txt = '\n'
192	>
193	>	txt += '#\n'
194	>	txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
195	>	txt += '#\n\n'
196	>
197	>	txt += 'func_exit() { \n'
198	>	txt += self.wsExitFunc_common()
199	>	### specific Glite check for OSB
200	>	txt += '    tar zcvf ${out_files}.tgz  ${final_list}\n'
201	>	txt += '    tmp_size=`ls -gGrta ${out_files}.tgz | awk \'{ print $3 }\'`\n'
202	>	txt += '    rm ${out_files}.tgz\n'
203	>	txt += '    size=`expr $tmp_size`\n'
204	>	txt += '    echo "Total Output dimension: $size"\n'
205	>	txt += '    limit='+str(self.OSBsize) +' \n'
206	>	txt += '    echo "WARNING: output files size limit is set to: $limit"\n'
207	>	txt += '    if [ "$limit" -lt "$size" ]; then\n'
208	>	txt += '        exceed=1\n'
209	>	txt += '        job_exit_code=70000\n'
210	>	txt += '        echo "Output Sanbox too big. Produced output is lost "\n'
211	>	txt += '    else\n'
212	>	txt += '        exceed=0\n'
213	>	txt += '        echo "Total Output dimension $size is fine."\n'
214	>	txt += '    fi\n'
215	>
216	>	txt += '    echo "JOB_EXIT_STATUS = $job_exit_code"\n'
217	>	txt += '    echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
218	>	txt += '    dumpStatus $RUNTIME_AREA/$repo\n'
219	>	txt += '    if [ $exceed -ne 1 ]; then\n'
220	>	txt += '        tar zcvf ${out_files}.tgz  ${final_list}\n'
221	>	txt += '    else\n'
222	>	txt += '        tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
223	>	txt += '    fi\n'
224	>	txt += '    exit $job_exit_code\n'
225	>
226	>	txt += '}\n'
227	>	return txt

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines