CRAB/python/SchedulerGlite.py

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

__revision__ = "$Id: SchedulerGlite.py,v 1.68 2009/05/26 16:53:23 spiga Exp $"
__version__ = "$Revision: 1.68 $"

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):
        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 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:	1.69
Committed:	Tue Jun 9 13:12:06 2009 UTC (15 years, 10 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_6_0_pre13, CRAB_2_6_0_pre12, CRAB_2_6_0_pre11, CRAB_2_6_0_pre10, CRAB_2_6_0_pre9, CRAB_2_6_0_pre8
Changes since 1.68:	+3 -3 lines
Log Message:	Introduce a CrabLogger class which hanlde the std python logging
#	User	Rev	Content
1	ewv	1.62	"""
2			CRAB interface to BossLite gLite Scheduler
3			"""
4
5	slacapra	1.69	__revision__ = "$Id: SchedulerGlite.py,v 1.68 2009/05/26 16:53:23 spiga Exp $"
6			__version__ = "$Revision: 1.68 $"
7	ewv	1.62
8	slacapra	1.31	from SchedulerGrid import SchedulerGrid
9	spiga	1.2	from crab_exceptions import *
10			from crab_util import *
11	fanzago	1.4	from GliteConfig import *
12	spiga	1.47	import EdgLoggingInfo
13	spiga	1.2	import common
14	ewv	1.65	from WMCore.SiteScreening.BlackWhiteListParser import CEBlackWhiteListParser
15	spiga	1.2
16			import os, sys, time
17
18	slacapra	1.31	class SchedulerGlite(SchedulerGrid):
19	slacapra	1.33	def __init__(self, name="GLITE"):
20			SchedulerGrid.__init__(self,name)
21	slacapra	1.31
22	fanzago	1.54	self.OSBsize = 55000000
23	spiga	1.51
24	slacapra	1.31	def configure(self,cfg_params):
25			SchedulerGrid.configure(self, cfg_params)
26	ewv	1.61	self.environment_unique_identifier = '$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	1.31	if not RB: return None
48			glite_config = None
49			rb_param_file = None
50	fanzago	1.4
51			gliteConfig = GliteConfig(RB)
52	slacapra	1.31	glite_config = gliteConfig.config()
53	fanzago	1.4
54	slacapra	1.31	if (glite_config ):
55	spiga	1.40	rb_param_file = glite_config
56	slacapra	1.31	return rb_param_file
57	fanzago	1.4
58	spiga	1.36	def ce_list(self):
59	spiga	1.2	"""
60	ewv	1.42	Returns string with requirement CE related
61	spiga	1.2	"""
62	ewv	1.65	ceParser = CEBlackWhiteListParser(self.EDG_ce_white_list,
63	slacapra	1.69	self.EDG_ce_black_list, common.logger())
64	ewv	1.42	req = ''
65	ewv	1.65	ce_white_list = []
66			ce_black_list = []
67	spiga	1.2	if self.EDG_ce_white_list:
68	ewv	1.65	ce_white_list = ceParser.whiteList()
69	slacapra	1.13	tmpCe=[]
70			concString = '&&'
71			for ce in ce_white_list:
72			tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
73	mcinquil	1.20	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	ewv	1.28
86	spiga	1.2	if self.EDG_ce_black_list:
87	ewv	1.65	ce_black_list = ceParser.blackList()
88	slacapra	1.13	tmpCe=[]
89			concString = '&&'
90	spiga	1.2	for ce in ce_black_list:
91	slacapra	1.13	tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
92	spiga	1.36	if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
93
94			# requirement added to skip gliteCE
95			req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
96	ewv	1.66	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	spiga	1.36
103	ewv	1.66	return req, retWL, retBL
104	spiga	1.36
105	slacapra	1.60	def se_list(self, dest):
106	spiga	1.36	"""
107	ewv	1.42	Returns string with requirement SE related
108			"""
109	slacapra	1.60	hostList=self.findSites_(dest)
110	spiga	1.36	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	spiga	1.11
119	spiga	1.36	return req
120
121			def jdlParam(self):
122			"""
123	ewv	1.42	Returns
124			"""
125	spiga	1.36	req=''
126			if self.EDG_addJdlParam:
127	ewv	1.42	if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
128	spiga	1.36	for p in self.EDG_addJdlParam:
129	ewv	1.61	req+=string.strip(p)+';\n'
130	spiga	1.36	return req
131
132			def specific_req(self):
133			"""
134			Returns string with specific requirements
135	ewv	1.42	"""
136	spiga	1.36	req=''
137	spiga	1.2	if self.EDG_clock_time:
138	slacapra	1.13	if (not req == ' '): req = req + ' && '
139			req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
140	spiga	1.2
141			if self.EDG_cpu_time:
142	slacapra	1.13	if (not req == ' '): req = req + ' && '
143			req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
144	ewv	1.42
145	spiga	1.36	return req
146	ewv	1.28
147	spiga	1.39	def sched_parameter(self,i,task):
148			"""
149			Returns string with requirements and scheduler-specific parameters
150			"""
151	ewv	1.61	dest= task.jobs[i-1]['dlsDestination']
152	spiga	1.39
153			req=''
154			req +=task['jobType']
155
156			sched_param=''
157	slacapra	1.60	sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
158	ewv	1.61	self.ce_list()[0] +';\n'
159			if self.EDG_addJdlParam: sched_param+=self.jdlParam()
160	spiga	1.39	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	ewv	1.42
167	spiga	1.47	def decodeLogInfo(self, file):
168			"""
169			Parse logging info file and return main info
170			"""
171			loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
172			reason = loggingInfo.decodeReason(file)
173			return reason
174
175	slacapra	1.60	def findSites_(self, sites):
176	spiga	1.39	itr4 =[]
177			if len(sites)>0 and sites[0]=="":
178			return itr4
179			if sites != [""]:
180	slacapra	1.60	replicas = self.blackWhiteListParser.checkBlackList(sites)
181	spiga	1.39	if len(replicas)!=0:
182	slacapra	1.60	replicas = self.blackWhiteListParser.checkWhiteList(replicas)
183	ewv	1.42
184	spiga	1.39	itr4 = replicas
185			return itr4
186
187	ewv	1.61
188	spiga	1.51	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	ewv	1.61	txt += ' rm ${out_files}.tgz\n'
203	spiga	1.51	txt += ' size=`expr $tmp_size`\n'
204			txt += ' echo "Total Output dimension: $size"\n'
205	ewv	1.61	txt += ' limit='+str(self.OSBsize) +' \n'
206	spiga	1.51	txt += ' echo "WARNING: output files size limit is set to: $limit"\n'
207	spiga	1.53	txt += ' if [ "$limit" -lt "$size" ]; then\n'
208	spiga	1.51	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	spiga	1.53	txt += ' echo "Total Output dimension $size is fine."\n'
214	spiga	1.51	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