CRAB/python/SchedulerGlite.py

"""
CRAB interface to BossLite gLite Scheduler
"""

__revision__ = "$Id: SchedulerGlite.py,v 1.70 2009/06/24 21:13:45 spiga Exp $"
__version__ = "$Revision: 1.70 $"

from SchedulerGrid import SchedulerGrid
from crab_exceptions import *
from crab_util import *
from GliteConfig import *
import EdgLoggingInfo
import common
from WMCore.SiteScreening.BlackWhiteListParser import CEBlackWhiteListParser

import os, sys, time

class SchedulerGlite(SchedulerGrid):
    def __init__(self, name="GLITE"):
        SchedulerGrid.__init__(self,name)

        self.OSBsize = 55000000

    def configure(self,cfg_params):
        SchedulerGrid.configure(self, cfg_params)
        self.environment_unique_identifier = '$GLITE_WMS_JOBID'

    def realSchedParams(self,cfg_params):
        """
        Return dictionary with specific parameters, to use
        with real scheduler
        """
        self.rb_param_file=''
        if (not cfg_params.has_key('GRID.rb')):
            cfg_params['GRID.rb']='CERN'
        self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("GRID.rb"))
        self.wms_service=cfg_params.get("GRID.wms_service",'')
        self.skipWMSAuth=cfg_params.get("GRID.skipwmsauth",1)
        params = { 'service' : self.wms_service, \
                   'config' : self.rb_param_file, \
                   'skipWMSAuth' : self.skipWMSAuth
                 }
        return  params


    def rb_configure(self, RB):
        ## 25-Jun-2009 SL: patch to use Cream enabled WMS
        if ( self.cfg_params.get('GRID.use_cream',None) ):
            RB='CREAM'
        if not RB: return None
        glite_config = None
        rb_param_file = None

        gliteConfig = GliteConfig(RB)
        glite_config = gliteConfig.config()

        if (glite_config ):
            rb_param_file = glite_config
        return rb_param_file

    def ce_list(self):
        """
        Returns string with requirement CE related
        """
        ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
                                          self.EDG_ce_black_list, common.logger())
        req = ''
        ce_white_list = []
        ce_black_list = []
        if self.EDG_ce_white_list:
            ce_white_list = ceParser.whiteList()
            tmpCe=[]
            concString = '&&'
            for ce in ce_white_list:
                tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
            if len(tmpCe) == 1:
                req +=  " && (" + concString.join(tmpCe) + ") "
            elif len(tmpCe) > 1:
                firstCE = 0
                for reqTemp in tmpCe:
                    if firstCE == 0:
                        req += " && ( (" + reqTemp + ") "
                        firstCE = 1
                    elif firstCE > 0:
                        req += " || (" + reqTemp + ") "
                if firstCE > 0:
                    req += ") "

        if self.EDG_ce_black_list:
            ce_black_list = ceParser.blackList()
            tmpCe=[]
            concString = '&&'
            for ce in ce_black_list:
                tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
            if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "

        # requirement added to skip gliteCE
        req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
        retWL = ','.join(ce_white_list)
        retBL = ','.join(ce_black_list)
        if not retWL:
            retWL = None
        if not retBL:
            retBL = None

        return req, retWL, retBL

    def se_list(self, dest):
        """
        Returns string with requirement SE related
        """
        hostList=self.findSites_(dest)
        req=''
        reqtmp=[]
        concString = '||'

        for arg in hostList:
            reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')

        if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "

        return req

    def jdlParam(self):
        """
        Returns
        """
        req=''
        if self.EDG_addJdlParam:
            if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
            for p in self.EDG_addJdlParam:
                req+=string.strip(p)+';\n'
        return req

    def specific_req(self):
        """
        Returns string with specific requirements
        """
        req=''
        if self.EDG_clock_time:
            if (not req == ' '): req = req + ' && '
            req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time

        if self.EDG_cpu_time:
            if (not req == ' '): req = req + ' && '
            req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time

        return req

    def sched_parameter(self,i,task):
        """
        Returns string with requirements and scheduler-specific parameters
        """
        dest=  task.jobs[i-1]['dlsDestination']

        req=''
        req +=task['jobType']

        sched_param=''
        sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
                                        self.ce_list()[0] +';\n'
        if self.EDG_addJdlParam: sched_param+=self.jdlParam()
        sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
        sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
        sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
        sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'

        return sched_param

    def decodeLogInfo(self, file):
        """
        Parse logging info file and return main info
        """
        loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
        reason = loggingInfo.decodeReason(file)
        return reason

    def findSites_(self, sites):
        itr4 =[]
        if len(sites)>0 and sites[0]=="":
            return itr4
        if sites != [""]:
            replicas = self.blackWhiteListParser.checkBlackList(sites)
            if len(replicas)!=0:
                replicas = self.blackWhiteListParser.checkWhiteList(replicas)

            itr4 = replicas
        return itr4

    def delegateProxy(self):
        self.boss().delegateProxy()   
        return

    def wsExitFunc(self):
        """
        """
        txt = '\n'

        txt += '#\n'
        txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
        txt += '#\n\n'

        txt += 'func_exit() { \n'
        txt += self.wsExitFunc_common()
        ### specific Glite check for OSB
        txt += '    tar zcvf ${out_files}.tgz  ${final_list}\n'
        txt += '    tmp_size=`ls -gGrta ${out_files}.tgz | awk \'{ print $3 }\'`\n'
        txt += '    rm ${out_files}.tgz\n'
        txt += '    size=`expr $tmp_size`\n'
        txt += '    echo "Total Output dimension: $size"\n'
        txt += '    limit='+str(self.OSBsize) +' \n'
        txt += '    echo "WARNING: output files size limit is set to: $limit"\n'
        txt += '    if [ "$limit" -lt "$size" ]; then\n'
        txt += '        exceed=1\n'
        txt += '        job_exit_code=70000\n'
        txt += '        echo "Output Sanbox too big. Produced output is lost "\n'
        txt += '    else\n'
        txt += '        exceed=0\n'
        txt += '        echo "Total Output dimension $size is fine."\n'
        txt += '    fi\n'

        txt += '    echo "JOB_EXIT_STATUS = $job_exit_code"\n'
        txt += '    echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
        txt += '    dumpStatus $RUNTIME_AREA/$repo\n'
        txt += '    if [ $exceed -ne 1 ]; then\n'
        txt += '        tar zcvf ${out_files}.tgz  ${final_list}\n'
        txt += '    else\n'
        txt += '        tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
        txt += '    fi\n'
        txt += '    exit $job_exit_code\n'

        txt += '}\n'
        return txt
Revision:	1.71
Committed:	Thu Jun 25 10:40:26 2009 UTC (15 years, 10 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_6_6_pre6, CRAB_2_7_1_pre2, CRAB_2_6_6_pre5, CRAB_2_7_1_pre1, CRAB_2_6_6_pre4, CRAB_2_6_6_pre3, CRAB_2_6_6_pre2, CRAB_2_6_6_check, CRAB_2_6_6, CRAB_2_6_6_pre1, CRAB_2_7_0, CRAB_2_6_5, CRAB_2_7_0_pre8, CRAB_2_6_5_pre1, CRAB_2_7_0_pre7, CRAB_2_6_4, CRAB_2_7_0_pre6, CRAB_2_6_4_pre1, CRAB_2_7_0_pre5, CRAB_2_6_3_patch_2, CRAB_2_6_3_patch_2_pre2, CRAB_2_6_3_patch_2_pre1, CRAB_2_6_3_patch_1, CRAB_2_7_0_pre4, CRAB_2_7_0_pre3, CRAB_2_6_3, CRAB_2_6_3_pre5, CRAB_2_6_3_pre4, CRAB_2_6_3_pre3, CRAB_2_6_3_pre2, CRAB_2_7_0_pre2, CRAB_2_6_3_pre1, test_1, CRAB_2_7_0_pre1, CRAB_2_6_2, CRAB_2_6_2_pre2, CRAB_2_6_2_pre1, CRAB_2_6_1_pre4, CRAB_2_6_1_pre3, CRAB_2_6_1_pre2, CRAB_2_6_1_pre1, CRAB_2_6_1, CRAB_2_6_0
Branch point for:	Lumi2_8, CRAB_2_6_X_br
Changes since 1.70:	+5 -2 lines
Log Message:	changes to allows use of CREAM ce by setting GRID.use_cream = 1
#	User	Rev	Content
1	ewv	1.62	"""
2			CRAB interface to BossLite gLite Scheduler
3			"""
4
5	slacapra	1.71	__revision__ = "$Id: SchedulerGlite.py,v 1.70 2009/06/24 21:13:45 spiga Exp $"
6			__version__ = "$Revision: 1.70 $"
7	ewv	1.62
8	slacapra	1.31	from SchedulerGrid import SchedulerGrid
9	spiga	1.2	from crab_exceptions import *
10			from crab_util import *
11	fanzago	1.4	from GliteConfig import *
12	spiga	1.47	import EdgLoggingInfo
13	spiga	1.2	import common
14	ewv	1.65	from WMCore.SiteScreening.BlackWhiteListParser import CEBlackWhiteListParser
15	spiga	1.2
16			import os, sys, time
17
18	slacapra	1.31	class SchedulerGlite(SchedulerGrid):
19	slacapra	1.33	def __init__(self, name="GLITE"):
20			SchedulerGrid.__init__(self,name)
21	slacapra	1.31
22	fanzago	1.54	self.OSBsize = 55000000
23	spiga	1.51
24	slacapra	1.31	def configure(self,cfg_params):
25			SchedulerGrid.configure(self, cfg_params)
26	ewv	1.61	self.environment_unique_identifier = '$GLITE_WMS_JOBID'
27	slacapra	1.3
28	spiga	1.51	def realSchedParams(self,cfg_params):
29			"""
30	ewv	1.61	Return dictionary with specific parameters, to use
31			with real scheduler
32	spiga	1.51	"""
33			self.rb_param_file=''
34	spiga	1.68	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	spiga	1.51	params = { 'service' : self.wms_service, \
40	afanfani	1.55	'config' : self.rb_param_file, \
41	ewv	1.61	'skipWMSAuth' : self.skipWMSAuth
42	spiga	1.51	}
43			return params
44	ewv	1.61
45	spiga	1.51
46	fanzago	1.4	def rb_configure(self, RB):
47	slacapra	1.71	## 25-Jun-2009 SL: patch to use Cream enabled WMS
48			if ( self.cfg_params.get('GRID.use_cream',None) ):
49			RB='CREAM'
50	slacapra	1.31	if not RB: return None
51			glite_config = None
52			rb_param_file = None
53	fanzago	1.4
54			gliteConfig = GliteConfig(RB)
55	slacapra	1.31	glite_config = gliteConfig.config()
56	fanzago	1.4
57	slacapra	1.31	if (glite_config ):
58	spiga	1.40	rb_param_file = glite_config
59	slacapra	1.31	return rb_param_file
60	fanzago	1.4
61	spiga	1.36	def ce_list(self):
62	spiga	1.2	"""
63	ewv	1.42	Returns string with requirement CE related
64	spiga	1.2	"""
65	ewv	1.65	ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
66	slacapra	1.69	self.EDG_ce_black_list, common.logger())
67	ewv	1.42	req = ''
68	ewv	1.65	ce_white_list = []
69			ce_black_list = []
70	spiga	1.2	if self.EDG_ce_white_list:
71	ewv	1.65	ce_white_list = ceParser.whiteList()
72	slacapra	1.13	tmpCe=[]
73			concString = '&&'
74			for ce in ce_white_list:
75			tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
76	mcinquil	1.20	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	ewv	1.28
89	spiga	1.2	if self.EDG_ce_black_list:
90	ewv	1.65	ce_black_list = ceParser.blackList()
91	slacapra	1.13	tmpCe=[]
92			concString = '&&'
93	spiga	1.2	for ce in ce_black_list:
94	slacapra	1.13	tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
95	spiga	1.36	if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
96
97			# requirement added to skip gliteCE
98			req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
99	ewv	1.66	retWL = ','.join(ce_white_list)
100			retBL = ','.join(ce_black_list)
101			if not retWL:
102			retWL = None
103			if not retBL:
104			retBL = None
105	spiga	1.36
106	ewv	1.66	return req, retWL, retBL
107	spiga	1.36
108	slacapra	1.60	def se_list(self, dest):
109	spiga	1.36	"""
110	ewv	1.42	Returns string with requirement SE related
111			"""
112	slacapra	1.60	hostList=self.findSites_(dest)
113	spiga	1.36	req=''
114			reqtmp=[]
115			concString = '\|\|'
116
117			for arg in hostList:
118			reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
119
120			if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "
121	spiga	1.11
122	spiga	1.36	return req
123
124			def jdlParam(self):
125			"""
126	ewv	1.42	Returns
127			"""
128	spiga	1.36	req=''
129			if self.EDG_addJdlParam:
130	ewv	1.42	if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
131	spiga	1.36	for p in self.EDG_addJdlParam:
132	ewv	1.61	req+=string.strip(p)+';\n'
133	spiga	1.36	return req
134
135			def specific_req(self):
136			"""
137			Returns string with specific requirements
138	ewv	1.42	"""
139	spiga	1.36	req=''
140	spiga	1.2	if self.EDG_clock_time:
141	slacapra	1.13	if (not req == ' '): req = req + ' && '
142			req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
143	spiga	1.2
144			if self.EDG_cpu_time:
145	slacapra	1.13	if (not req == ' '): req = req + ' && '
146			req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
147	ewv	1.42
148	spiga	1.36	return req
149	ewv	1.28
150	spiga	1.39	def sched_parameter(self,i,task):
151			"""
152			Returns string with requirements and scheduler-specific parameters
153			"""
154	ewv	1.61	dest= task.jobs[i-1]['dlsDestination']
155	spiga	1.39
156			req=''
157			req +=task['jobType']
158
159			sched_param=''
160	slacapra	1.60	sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
161	ewv	1.61	self.ce_list()[0] +';\n'
162			if self.EDG_addJdlParam: sched_param+=self.jdlParam()
163	spiga	1.39	sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
164			sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
165			sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
166			sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'
167
168			return sched_param
169	ewv	1.42
170	spiga	1.47	def decodeLogInfo(self, file):
171			"""
172			Parse logging info file and return main info
173			"""
174			loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
175			reason = loggingInfo.decodeReason(file)
176			return reason
177
178	slacapra	1.60	def findSites_(self, sites):
179	spiga	1.39	itr4 =[]
180			if len(sites)>0 and sites[0]=="":
181			return itr4
182			if sites != [""]:
183	slacapra	1.60	replicas = self.blackWhiteListParser.checkBlackList(sites)
184	spiga	1.39	if len(replicas)!=0:
185	slacapra	1.60	replicas = self.blackWhiteListParser.checkWhiteList(replicas)
186	ewv	1.42
187	spiga	1.39	itr4 = replicas
188			return itr4
189
190	spiga	1.70	def delegateProxy(self):
191			self.boss().delegateProxy()
192			return
193	ewv	1.61
194	spiga	1.51	def wsExitFunc(self):
195			"""
196			"""
197			txt = '\n'
198
199			txt += '#\n'
200			txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
201			txt += '#\n\n'
202
203			txt += 'func_exit() { \n'
204			txt += self.wsExitFunc_common()
205			### specific Glite check for OSB
206			txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
207			txt += ' tmp_size=`ls -gGrta ${out_files}.tgz \| awk \'{ print $3 }\'`\n'
208	ewv	1.61	txt += ' rm ${out_files}.tgz\n'
209	spiga	1.51	txt += ' size=`expr $tmp_size`\n'
210			txt += ' echo "Total Output dimension: $size"\n'
211	ewv	1.61	txt += ' limit='+str(self.OSBsize) +' \n'
212	spiga	1.51	txt += ' echo "WARNING: output files size limit is set to: $limit"\n'
213	spiga	1.53	txt += ' if [ "$limit" -lt "$size" ]; then\n'
214	spiga	1.51	txt += ' exceed=1\n'
215			txt += ' job_exit_code=70000\n'
216			txt += ' echo "Output Sanbox too big. Produced output is lost "\n'
217			txt += ' else\n'
218			txt += ' exceed=0\n'
219	spiga	1.53	txt += ' echo "Total Output dimension $size is fine."\n'
220	spiga	1.51	txt += ' fi\n'
221
222			txt += ' echo "JOB_EXIT_STATUS = $job_exit_code"\n'
223			txt += ' echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
224			txt += ' dumpStatus $RUNTIME_AREA/$repo\n'
225			txt += ' if [ $exceed -ne 1 ]; then\n'
226			txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
227			txt += ' else\n'
228			txt += ' tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
229			txt += ' fi\n'
230			txt += ' exit $job_exit_code\n'
231
232			txt += '}\n'
233			return txt