CRAB/python/SchedulerGlite.py

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

__revision__ = "$Id: SchedulerGlite.py,v 1.63 2008/10/17 10:52:33 slacapra Exp $"
__version__ = "$Revision: 1.63 $"

from SchedulerGrid import SchedulerGrid
from crab_logger import Logger
from crab_exceptions import *
from crab_util import *
from GliteConfig import *
import EdgLoggingInfo
import common

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('EDG.rb')):
            cfg_params['EDG.rb']='CERN'
        self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("EDG.rb"))
        self.wms_service=cfg_params.get("EDG.wms_service",'')
        self.skipWMSAuth=cfg_params.get("EDG.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
        """
        req = ''
        if self.EDG_ce_white_list:
            ce_white_list = self.EDG_ce_white_list
            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 = self.EDG_ce_black_list
            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))'

        return req,self.EDG_ce_white_list,self.EDG_ce_black_list

    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:	1.64
Committed:	Sat Nov 8 11:57:03 2008 UTC (16 years, 5 months ago) by spiga
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_4_3_pre7, CRAB_2_4_3_pre6, CRAB_2_4_3_pre5, CRAB_2_4_3_pre3, CRAB_2_4_3_pre2, CRAB_2_4_3_pre1, CRAB_2_4_2, CRAB_2_4_2_pre3
Changes since 1.63:	+3 -3 lines
Log Message:	skip wms Auth. is set to 1 by default.
#	User	Rev	Content
1	ewv	1.62	"""
2			CRAB interface to BossLite gLite Scheduler
3			"""
4
5	spiga	1.64	__revision__ = "$Id: SchedulerGlite.py,v 1.63 2008/10/17 10:52:33 slacapra Exp $"
6			__version__ = "$Revision: 1.63 $"
7	ewv	1.62
8	slacapra	1.31	from SchedulerGrid import SchedulerGrid
9	spiga	1.2	from crab_logger import Logger
10			from crab_exceptions import *
11			from crab_util import *
12	fanzago	1.4	from GliteConfig import *
13	spiga	1.47	import EdgLoggingInfo
14	spiga	1.2	import common
15
16			import os, sys, time
17
18	slacapra	1.31	class SchedulerGlite(SchedulerGrid):
19	slacapra	1.33	def __init__(self, name="GLITE"):
20			SchedulerGrid.__init__(self,name)
21	slacapra	1.31
22	fanzago	1.54	self.OSBsize = 55000000
23	spiga	1.51
24	slacapra	1.31	def configure(self,cfg_params):
25			SchedulerGrid.configure(self, cfg_params)
26	ewv	1.61	self.environment_unique_identifier = '$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	slacapra	1.63	if (not cfg_params.has_key('EDG.rb')):
35	slacapra	1.60	cfg_params['EDG.rb']='CERN'
36			self.rb_param_file=common.scheduler.rb_configure(cfg_params.get("EDG.rb"))
37	spiga	1.51	self.wms_service=cfg_params.get("EDG.wms_service",'')
38	spiga	1.64	self.skipWMSAuth=cfg_params.get("EDG.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	1.42	req = ''
63	spiga	1.2	if self.EDG_ce_white_list:
64	mcinquil	1.32	ce_white_list = self.EDG_ce_white_list
65	slacapra	1.13	tmpCe=[]
66			concString = '&&'
67			for ce in ce_white_list:
68			tmpCe.append('RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId)')
69	mcinquil	1.20	if len(tmpCe) == 1:
70			req += " && (" + concString.join(tmpCe) + ") "
71			elif len(tmpCe) > 1:
72			firstCE = 0
73			for reqTemp in tmpCe:
74			if firstCE == 0:
75			req += " && ( (" + reqTemp + ") "
76			firstCE = 1
77			elif firstCE > 0:
78			req += " \|\| (" + reqTemp + ") "
79			if firstCE > 0:
80			req += ") "
81	ewv	1.28
82	spiga	1.2	if self.EDG_ce_black_list:
83	mcinquil	1.32	ce_black_list = self.EDG_ce_black_list
84	slacapra	1.13	tmpCe=[]
85			concString = '&&'
86	spiga	1.2	for ce in ce_black_list:
87	slacapra	1.13	tmpCe.append('(!RegExp("' + string.strip(ce) + '", other.GlueCEUniqueId))')
88	spiga	1.36	if len(tmpCe): req += " && (" + concString.join(tmpCe) + ") "
89
90			# requirement added to skip gliteCE
91			req += '&& (!RegExp("blah", other.GlueCEUniqueId))'
92
93	spiga	1.41	return req,self.EDG_ce_white_list,self.EDG_ce_black_list
94	spiga	1.36
95	slacapra	1.60	def se_list(self, dest):
96	spiga	1.36	"""
97	ewv	1.42	Returns string with requirement SE related
98			"""
99	slacapra	1.60	hostList=self.findSites_(dest)
100	spiga	1.36	req=''
101			reqtmp=[]
102			concString = '\|\|'
103
104			for arg in hostList:
105			reqtmp.append(' Member("'+arg+'" , other.GlueCESEBindGroupSEUniqueID) ')
106
107			if len(reqtmp): req += " && (" + concString.join(reqtmp) + ") "
108	spiga	1.11
109	spiga	1.36	return req
110
111			def jdlParam(self):
112			"""
113	ewv	1.42	Returns
114			"""
115	spiga	1.36	req=''
116			if self.EDG_addJdlParam:
117	ewv	1.42	if self.EDG_addJdlParam[-1] == '': self.EDG_addJdlParam= self.EDG_addJdlParam[:-1]
118	spiga	1.36	for p in self.EDG_addJdlParam:
119	ewv	1.61	req+=string.strip(p)+';\n'
120	spiga	1.36	return req
121
122			def specific_req(self):
123			"""
124			Returns string with specific requirements
125	ewv	1.42	"""
126	spiga	1.36	req=''
127	spiga	1.2	if self.EDG_clock_time:
128	slacapra	1.13	if (not req == ' '): req = req + ' && '
129			req = req + 'other.GlueCEPolicyMaxWallClockTime>='+self.EDG_clock_time
130	spiga	1.2
131			if self.EDG_cpu_time:
132	slacapra	1.13	if (not req == ' '): req = req + ' && '
133			req = req + ' other.GlueCEPolicyMaxCPUTime>='+self.EDG_cpu_time
134	ewv	1.42
135	spiga	1.36	return req
136	ewv	1.28
137	spiga	1.39	def sched_parameter(self,i,task):
138			"""
139			Returns string with requirements and scheduler-specific parameters
140			"""
141	ewv	1.61	dest= task.jobs[i-1]['dlsDestination']
142	spiga	1.39
143			req=''
144			req +=task['jobType']
145
146			sched_param=''
147	slacapra	1.60	sched_param+='Requirements = ' + req +self.specific_req() + self.se_list(dest) +\
148	ewv	1.61	self.ce_list()[0] +';\n'
149			if self.EDG_addJdlParam: sched_param+=self.jdlParam()
150	spiga	1.39	sched_param+='MyProxyServer = "' + self.proxyServer + '";\n'
151			sched_param+='VirtualOrganisation = "' + self.VO + '";\n'
152			sched_param+='RetryCount = '+str(self.EDG_retry_count)+';\n'
153			sched_param+='ShallowRetryCount = '+str(self.EDG_shallow_retry_count)+';\n'
154
155			return sched_param
156	ewv	1.42
157	spiga	1.47	def decodeLogInfo(self, file):
158			"""
159			Parse logging info file and return main info
160			"""
161			loggingInfo = EdgLoggingInfo.EdgLoggingInfo()
162			reason = loggingInfo.decodeReason(file)
163			return reason
164
165	slacapra	1.60	def findSites_(self, sites):
166	spiga	1.39	itr4 =[]
167			if len(sites)>0 and sites[0]=="":
168			return itr4
169			if sites != [""]:
170	slacapra	1.60	replicas = self.blackWhiteListParser.checkBlackList(sites)
171	spiga	1.39	if len(replicas)!=0:
172	slacapra	1.60	replicas = self.blackWhiteListParser.checkWhiteList(replicas)
173	ewv	1.42
174	spiga	1.39	itr4 = replicas
175			return itr4
176
177	ewv	1.61
178	spiga	1.51	def wsExitFunc(self):
179			"""
180			"""
181			txt = '\n'
182
183			txt += '#\n'
184			txt += '# EXECUTE THIS FUNCTION BEFORE EXIT \n'
185			txt += '#\n\n'
186
187			txt += 'func_exit() { \n'
188			txt += self.wsExitFunc_common()
189			### specific Glite check for OSB
190			txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
191			txt += ' tmp_size=`ls -gGrta ${out_files}.tgz \| awk \'{ print $3 }\'`\n'
192	ewv	1.61	txt += ' rm ${out_files}.tgz\n'
193	spiga	1.51	txt += ' size=`expr $tmp_size`\n'
194			txt += ' echo "Total Output dimension: $size"\n'
195	ewv	1.61	txt += ' limit='+str(self.OSBsize) +' \n'
196	spiga	1.51	txt += ' echo "WARNING: output files size limit is set to: $limit"\n'
197	spiga	1.53	txt += ' if [ "$limit" -lt "$size" ]; then\n'
198	spiga	1.51	txt += ' exceed=1\n'
199			txt += ' job_exit_code=70000\n'
200			txt += ' echo "Output Sanbox too big. Produced output is lost "\n'
201			txt += ' else\n'
202			txt += ' exceed=0\n'
203	spiga	1.53	txt += ' echo "Total Output dimension $size is fine."\n'
204	spiga	1.51	txt += ' fi\n'
205
206			txt += ' echo "JOB_EXIT_STATUS = $job_exit_code"\n'
207			txt += ' echo "JobExitCode=$job_exit_code" >> $RUNTIME_AREA/$repo\n'
208			txt += ' dumpStatus $RUNTIME_AREA/$repo\n'
209			txt += ' if [ $exceed -ne 1 ]; then\n'
210			txt += ' tar zcvf ${out_files}.tgz ${final_list}\n'
211			txt += ' else\n'
212			txt += ' tar zcvf ${out_files}.tgz CMSSW_${NJob}.stdout CMSSW_${NJob}.stderr\n'
213			txt += ' fi\n'
214			txt += ' exit $job_exit_code\n'
215
216			txt += '}\n'
217			return txt