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 1.13 by slacapra, Wed Feb 28 14:10:33 2007 UTC vs.
Revision 1.78 by spigafi, Wed Feb 3 14:59:30 2010 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 *
11	>	import EdgLoggingInfo
12		import common
13	+	from WMCore.SiteScreening.BlackWhiteListParser import CEBlackWhiteListParser
14
15		import os, sys, time
16
17	<	class SchedulerGlite(SchedulerEdg):
18	<	def __init__(self):
19	<	SchedulerEdg.__init__(self)
20	<
21	<	def rb_configure(self, RB):
22	<	self.glite_config = ''
23	<	self.rb_param_file = ''
17	>	class SchedulerGlite(SchedulerGrid):
18	>	def __init__(self, name="GLITE"):
19	>	SchedulerGrid.__init__(self,name)
20	>
21	>	self.OSBsize = 55000000
22	>
23	>	def configure(self,cfg_params):
24	>	SchedulerGrid.configure(self, cfg_params)
25	>	self.environment_unique_identifier = '$GLITE_WMS_JOBID'
26	>
27	>	def realSchedParams(self,cfg_params):
28	>	"""
29	>	Return dictionary with specific parameters, to use
30	>	with real scheduler
31	>	"""
32	>	self.rb_param_file=''
33	>	if (not cfg_params.has_key('GRID.rb')):
34	>	cfg_params['GRID.rb']='CERN'
35	>	self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("GRID.rb"))
36	>	self.wms_service=cfg_params.get("GRID.wms_service",'')
37	>	self.skipWMSAuth=cfg_params.get("GRID.skipwmsauth",1)
38	>	params = { 'service' : self.wms_service, \
39	>	'config' : self.rb_param_file, \
40	>	'skipWMSAuth' : self.skipWMSAuth
41	>	}
42	>	return  params
43
18	–	gliteConfig = GliteConfig(RB)
19	–	self.glite_config = gliteConfig.config()
44
45	<	if (self.glite_config != ''):
46	<	self.rb_param_file = 'WMSconfig = '+self.glite_config+';\n'
47	<	#print "rb_param_file = ", self.rb_param_file
48	<	return self.rb_param_file
45	>	def rb_configure(self, RB):
46	>	url ='http://cmsdoc.cern.ch/cms/LCG/crab/config/'
47	>	from Downloader import Downloader
48	>	common.logger.debug('Downloading config files for WMS: '+url)
49	>	## 25-Jun-2009 SL: patch to use Cream enabled WMS
50	>	if ( self.cfg_params.get('GRID.use_cream',None) ):
51	>	RB='CREAM'
52	>	if not RB: return None
53	>	rb_param_file = None
54	>	configFileName = 'glite_wms_'+str(RB)+'.conf'
55	>
56	>	results = Downloader(url)
57	>	gliteConfig  = results.filePath(configFileName)
58	>
59	>	if (gliteConfig ):
60	>	rb_param_file = gliteConfig
61	>	return rb_param_file
62
63	<	def sched_parameter(self):
63	>	def ce_list(self):
64		"""
65	<	Returns file with requirements and scheduler-specific parameters
65	>	Returns string with requirement CE related
66		"""
67	<	index = int(common.jobDB.nJobs()) - 1
68	<	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	<
67	>	ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
68	>	self.EDG_ce_black_list, common.logger())
69		req = ''
70	<	req = req + jbt.getRequirements()
71	<
48	<
49	<	if self.EDG_requirements:
50	<	if (not req == ' '): req = req +  ' && '
51	<	req = req + self.EDG_requirements
52	<
70	>	ce_white_list = []
71	>	ce_black_list = []
72		if self.EDG_ce_white_list:
73	<	ce_white_list = string.split(self.EDG_ce_white_list,',')
73	>	ce_white_list = ceParser.whiteList()
74		tmpCe=[]
75		concString = '&&'
76		for ce in ce_white_list:
77		tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
78	<	if len(tmpCe): req = req + " && (" + concString.join(tmpCe) + ") "
79	<
78	>	if len(tmpCe) == 1:
79	>	req +=  " && (" + concString.join(tmpCe) + ") "
80	>	elif len(tmpCe) > 1:
81	>	firstCE = 0
82	>	for reqTemp in tmpCe:
83	>	if firstCE == 0:
84	>	req += " && ( (" + reqTemp + ") "
85	>	firstCE = 1
86	>	elif firstCE > 0:
87	>	req += " || (" + reqTemp + ") "
88	>	if firstCE > 0:
89	>	req += ") "
90	>
91		if self.EDG_ce_black_list:
92	<	ce_black_list = string.split(self.EDG_ce_black_list,',')
92	>	ce_black_list = ceParser.blackList()
93		tmpCe=[]
94		concString = '&&'
95		for ce in ce_black_list:
96		tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
97	<	if len(tmpCe): req = req + " && (" + concString.join(tmpCe) + ") "
97	>	if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
98	>
99	>	# requirement added to skip gliteCE
100	>	# not more needed
101	>	#       req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
102	>	retWL = ','.join(ce_white_list)
103	>	retBL = ','.join(ce_black_list)
104	>	if not retWL:
105	>	retWL = None
106	>	if not retBL:
107	>	retBL = None
108	>
109	>	return req, retWL, retBL
110
111	+	def se_list(self, dest):
112	+	"""
113	+	Returns string with requirement SE related
114	+	"""
115	+	hostList=self.findSites_(dest)
116	+	req=''
117	+	reqtmp=[]
118	+	concString = '||'
119	+
120	+	for arg in hostList:
121	+	reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
122	+
123	+	if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "
124	+
125	+	return req
126	+
127	+	def jdlParam(self):
128	+	"""
129	+	Returns
130	+	"""
131	+	req=''
132	+	if self.EDG_addJdlParam:
133	+	if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
134	+	for p in self.EDG_addJdlParam:
135	+	req+=string.strip(p)+';\n'
136	+	return req
137	+
138	+	def specific_req(self):
139	+	"""
140	+	Returns string with specific requirements
141	+	"""
142	+	req=''
143		if self.EDG_clock_time:
144		if (not req == ' '): req = req + ' && '
145		req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
#	Line 73 | Line 147 | class SchedulerGlite(SchedulerEdg):
147		if self.EDG_cpu_time:
148		if (not req == ' '): req = req + ' && '
149		req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
76	–
77	–	for i in range(len(first)): # Add loop DS
78	–	self.param='sched_param_'+str(i)+'.clad'
79	–	param_file = open(common.work_space.shareDir()+'/'+self.param, 'w')
80	–
81	–	itr4=self.findSites_(first[i])
82	–	reqSites=''
83	–	reqtmp=[]
84	–	concString = '||'
85	–
86	–	#############
87	–	# MC Changed matching syntax to avoid gang matching
88	–	#############
89	–	for arg in itr4:
90	–	reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
91	–
92	–	if len(reqtmp): reqSites = reqSites + " && (" + concString.join(reqtmp) + ") "
93	–
94	–	# requirement added to skip gliteCE
95	–	reqSites = reqSites + '&& (!RegExp("blah", other.GlueCEUniqueId));\n'
96	–
97	–	param_file.write('Requirements = ' + req + reqSites )
98	–
99	–	if (self.rb_param_file != ''):
100	–	param_file.write(self.rb_param_file)
101	–
102	–	if len(self.EDG_addJdlParam):
103	–	for p in self.EDG_addJdlParam:
104	–	param_file.write(p)
105	–
106	–	param_file.close()
107	–
108	–	def wsSetupEnvironment(self):
109	–	"""
110	–	Returns part of a job script which does scheduler-specific work.
111	–	"""
112	–	txt = ''
113	–	txt += '# strip arguments\n'
114	–	txt += 'echo "strip arguments"\n'
115	–	txt += 'args=("$@")\n'
116	–	txt += 'nargs=$#\n'
117	–	txt += 'shift $nargs\n'
118	–	txt += "# job number (first parameter for job wrapper)\n"
119	–	txt += "NJob=${args[0]}\n"
120	–
121	–	txt += '# job identification to DashBoard \n'
122	–	txt += 'MonitorJobID=`echo ${NJob}_$GLITE_WMS_JOBID`\n'
123	–	txt += 'SyncGridJobId=`echo $GLITE_WMS_JOBID`\n'
124	–	txt += 'MonitorID=`echo ' + self._taskId + '`\n'
125	–	txt += 'echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
126	–	txt += 'echo "SyncGridJobId=`echo $SyncGridJobId`" | tee -a $RUNTIME_AREA/$repo \n'
127	–	txt += 'echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
128	–
129	–	txt += 'echo "middleware discovery " \n'
130	–	txt += 'if [ $VO_CMS_SW_DIR ]; then \n'
131	–	txt += '    middleware=LCG \n'
132	–	txt += '    echo "SyncCE=`glite-brokerinfo getCE`" | tee -a $RUNTIME_AREA/$repo \n'
133	–	txt += '    echo "GridFlavour=`echo $middleware`" | tee -a $RUNTIME_AREA/$repo \n'
134	–	txt += '    echo "middleware =$middleware" \n'
135	–	txt += 'elif [ $GRID3_APP_DIR ]; then\n'
136	–	txt += '    middleware=OSG \n'
137	–	txt += '    echo "SyncCE=`echo $EDG_WL_LOG_DESTINATION`" | 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 [ $OSG_APP ]; 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 += 'else \n'
146	–	txt += '    echo "SET_CMS_ENV 10030 ==> middleware not identified" \n'
147	–	txt += '    echo "JOB_EXIT_STATUS = 10030" \n'
148	–	txt += '    echo "JobExitCode=10030" | tee -a $RUNTIME_AREA/$repo \n'
149	–	txt += '    dumpStatus $RUNTIME_AREA/$repo \n'
150	–	txt += '    rm -f $RUNTIME_AREA/$repo \n'
151	–	txt += '    echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
152	–	txt += '    echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
153	–	txt += '    exit 1 \n'
154	–	txt += 'fi \n'
155	–
156	–	txt += '# report first time to DashBoard \n'
157	–	txt += 'dumpStatus $RUNTIME_AREA/$repo \n'
158	–	txt += 'rm -f $RUNTIME_AREA/$repo \n'
159	–	txt += 'echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
160	–	txt += 'echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
161	–
162	–	txt += '\n\n'
150
151	<	if int(self.copy_data) == 1:
165	<	if self.SE:
166	<	txt += 'export SE='+self.SE+'\n'
167	<	txt += 'echo "SE = $SE"\n'
168	<	if self.SE_PATH:
169	<	if ( self.SE_PATH[-1] != '/' ) : self.SE_PATH = self.SE_PATH + '/'
170	<	txt += 'export SE_PATH='+self.SE_PATH+'\n'
171	<	txt += 'echo "SE_PATH = $SE_PATH"\n'
172	<
173	<	txt += 'export VO='+self.VO+'\n'
174	<	### some line for LFC catalog setting
175	<	txt += 'if [ $middleware == LCG ]; then \n'
176	<	txt += '    if [[ $LCG_CATALOG_TYPE != \''+self.lcg_catalog_type+'\' ]]; then\n'
177	<	txt += '        export LCG_CATALOG_TYPE='+self.lcg_catalog_type+'\n'
178	<	txt += '    fi\n'
179	<	txt += '    if [[ $LFC_HOST != \''+self.lfc_host+'\' ]]; then\n'
180	<	txt += '        export LFC_HOST='+self.lfc_host+'\n'
181	<	txt += '    fi\n'
182	<	txt += '    if [[ $LFC_HOME != \''+self.lfc_home+'\' ]]; then\n'
183	<	txt += '        export LFC_HOME='+self.lfc_home+'\n'
184	<	txt += '    fi\n'
185	<	txt += 'elif [ $middleware == OSG ]; then\n'
186	<	txt += '    echo "LFC catalog setting to be implemented for OSG"\n'
187	<	txt += 'fi\n'
188	<	#####
189	<	if int(self.register_data) == 1:
190	<	txt += 'if [ $middleware == LCG ]; then \n'
191	<	txt += '    export LFN='+self.LFN+'\n'
192	<	txt += '    lfc-ls $LFN\n'
193	<	txt += '    result=$?\n'
194	<	txt += '    echo $result\n'
195	<	### creation of LFN dir in LFC catalog, under /grid/cms dir
196	<	txt += '    if [ $result != 0 ]; then\n'
197	<	txt += '       lfc-mkdir $LFN\n'
198	<	txt += '       result=$?\n'
199	<	txt += '       echo $result\n'
200	<	txt += '    fi\n'
201	<	txt += 'elif [ $middleware == OSG ]; then\n'
202	<	txt += '    echo " Files registration to be implemented for OSG"\n'
203	<	txt += 'fi\n'
204	<	txt += '\n'
205	<
206	<	if self.VO:
207	<	txt += 'export VO='+self.VO+'\n'
208	<	if self.LFN:
209	<	txt += 'if [ $middleware == LCG ]; then \n'
210	<	txt += '    export LFN='+self.LFN+'\n'
211	<	txt += 'fi\n'
212	<	txt += '\n'
213	<
214	<	txt += 'if [ $middleware == LCG ]; then\n'
215	<	txt += '    CloseCEs=`glite-brokerinfo getCE`\n'
216	<	txt += '    echo "CloseCEs = $CloseCEs"\n'
217	<	txt += '    CE=`echo $CloseCEs | sed -e "s/:.*//"`\n'
218	<	txt += '    echo "CE = $CE"\n'
219	<	txt += 'elif [ $middleware == OSG ]; then \n'
220	<	txt += '    if [ $OSG_JOB_CONTACT ]; then \n'
221	<	txt += '        CE=`echo $OSG_JOB_CONTACT | /usr/bin/awk -F\/ \'{print $1}\'` \n'
222	<	txt += '    else \n'
223	<	txt += '        echo "SET_CMS_ENV 10099 ==> OSG mode: ERROR in setting CE name from OSG_JOB_CONTACT" \n'
224	<	txt += '        echo "JOB_EXIT_STATUS = 10099" \n'
225	<	txt += '        echo "JobExitCode=10099" | tee -a $RUNTIME_AREA/$repo \n'
226	<	txt += '        dumpStatus $RUNTIME_AREA/$repo \n'
227	<	txt += '        rm -f $RUNTIME_AREA/$repo \n'
228	<	txt += '        echo "MonitorJobID=`echo $MonitorJobID`" | tee -a $RUNTIME_AREA/$repo \n'
229	<	txt += '        echo "MonitorID=`echo $MonitorID`" | tee -a $RUNTIME_AREA/$repo\n'
230	<	txt += '        exit 1 \n'
231	<	txt += '    fi \n'
232	<	txt += 'fi \n'
151	>	return req
152
153	<	return txt
235	<
236	<	def loggingInfo(self, id):
153	>	def sched_parameter(self,i,task):
154		"""
155	<	retrieve the logging info from logging and bookkeeping and return it
155	>	Returns string with requirements and scheduler-specific parameters
156		"""
157	<	self.checkProxy()
158	<	cmd = 'glite-job-logging-info -v 3 ' + id
159	<	cmd_out = runCommand(cmd)
160	<	return cmd_out
161	<
162	<	def queryDetailedStatus(self, id):
163	<	""" Query a detailed status of the job with id """
164	<	cmd = 'glite-job-status '+id
165	<	cmd_out = runCommand(cmd)
166	<	return cmd_out
157	>	dest=  task.jobs[i-1]['dlsDestination']
158	>
159	>	req=''
160	>	req +=task['jobType']
161	>
162	>	sched_param=''
163	>	sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
164	>	self.ce_list()[0] +';\n'
165	>	if self.EDG_addJdlParam: sched_param+=self.jdlParam()
166	>	sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
167	>	sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
168	>	sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
169	>	sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'
170
171	<	def findSites_(self, n):
172	<	sites = common.jobDB.destination(n)
171	>	return sched_param
172	>
173	>	def decodeLogInfo(self, file):
174	>	"""
175	>	Parse logging info file and return main info
176	>	"""
177	>	loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
178	>	reason = loggingInfo.decodeReason(file)
179	>	return reason
180	>
181	>	def findSites_(self, sites):
182	>	itr4 =[]
183		if len(sites)>0 and sites[0]=="":
184	<	return []
184	>	return itr4
185	>	if sites != [""]:
186	>	replicas = self.blackWhiteListParser.checkBlackList(sites)
187	>	if len(replicas)!=0:
188	>	replicas = self.blackWhiteListParser.checkWhiteList(replicas)
189	>
190	>	itr4 = replicas
191	>	return itr4
192	>
193	>	def delegateProxy(self):
194	>	self.boss().delegateProxy()
195	>	return
196	>
197	>	def wsExitFunc(self):
198	>	"""
199	>	"""
200	>	txt = '\n'
201	>
202	>	txt += '#\n'
203	>	txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
204	>	txt += '#\n\n'
205	>
206	>	txt += 'func_exit() { \n'
207	>	txt += self.wsExitFunc_common()
208	>	### specific Glite check for OSB
209	>	txt += '    tar zcvf ${out_files}.tgz  ${final_list}\n'
210	>	txt += '    tmp_size=`ls -gGrta ${out_files}.tgz | awk \'{ print $3 }\'`\n'
211	>	txt += '    rm ${out_files}.tgz\n'
212	>	txt += '    size=`expr $tmp_size`\n'
213	>	txt += '    echo "Total Output dimension: $size"\n'
214	>	txt += '    limit='+str(self.OSBsize) +' \n'
215	>	txt += '    echo "WARNING: output files size limit is set to: $limit"\n'
216	>	txt += '    if [ "$limit" -lt "$size" ]; then\n'
217	>	txt += '        exceed=1\n'
218	>	txt += '        job_exit_code=70000\n'
219	>	txt += '        echo "Output Sanbox too big. Produced output is lost "\n'
220	>	txt += '    else\n'
221	>	txt += '        exceed=0\n'
222	>	txt += '        echo "Total Output dimension $size is fine."\n'
223	>	txt += '    fi\n'
224	>
225	>	txt += '    echo "JOB_EXIT_STATUS = $job_exit_code"\n'
226	>	txt += '    echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
227	>	txt += '    dumpStatus $RUNTIME_AREA/$repo\n'
228	>	txt += '    if [ $exceed -ne 1 ]; then\n'
229	>	txt += '        tar zcvf ${out_files}.tgz  ${final_list}\n'
230	>	txt += '    else\n'
231	>	txt += '        tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
232	>	txt += '    fi\n'
233	>	txt += '    python $RUNTIME_AREA/fillCrabFjr.py $RUNTIME_AREA/crab_fjr_$NJob.xml --errorcode $job_exit_code \n'
234	>	txt += '    exit $job_exit_code\n'
235	>
236	>	txt += '}\n'
237	>	return txt
238	>
239	>	def listMatch(self, dest, full):
240	>	matching='fast'
241	>
242	>	if self.boss().schedulerConfig['name'] == 'SchedulerGLite' :
243	>	taskId=common._db.getTask()
244	>	req=str(self.sched_parameter(1,taskId))
245	>	sites = self.boss().schedSession().matchResources(taskId, requirements=req)
246	>	else :
247	>	sites = SchedulerGrid.listMatch(self, dest, full)
248	>
249	>	if full == True: matching='full'
250	>	common.logger.debug("list of available site ( "+str(matching) +" matching ) : "+str(sites))
251	>
252		return sites

Diff Legend

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