CRAB/python/Creator.py

from Actor import Actor
from WorkSpace import WorkSpace
from JobList import JobList
from JobDB import JobDB
from ScriptWriter import ScriptWriter
from Scheduler import Scheduler
from crab_logger import Logger
from crab_exceptions import *
from crab_util import *
import common

import os, string, math
# marco
class Creator(Actor):
    def __init__(self, job_type_name, cfg_params, ncjobs):
        self.job_type_name = job_type_name
        self.job_type = None
        self.cfg_params = cfg_params
        self.total_njobs = 0
        self.total_number_of_events = 0
        self.job_number_of_events = 0
        self.first_event = 0
        self.jobParamsList=[]

        self.createJobTypeObject(ncjobs)
        common.logger.debug(5, __name__+": JobType "+self.job_type.name()+" created")

        self.job_type.prepareSteeringCards()
        common.logger.debug(5, __name__+": Steering cards prepared")

        self.total_njobs = self.job_type.numberOfJobs();
        common.logger.debug(5, __name__+": total # of jobs = "+`self.total_njobs`)

        # Set number of jobs to be created

        self.ncjobs = ncjobs
        if ncjobs == 'all' : self.ncjobs = self.total_njobs
        if ncjobs > self.total_njobs : self.ncjobs = self.total_njobs
        
        # This is code for proto-monitoring
        code=common.taskDB.dict("CODE")
        if self.job_type.name() == 'CMSSW':
            try: 
                self.primaryDataset = cfg_params['CMSSW.datasetpath'].split("/")[1]
            except:
                self.primaryDataset = 'None'
            try:
                self.ProcessedDataset = cfg_params['CMSSW.datasetpath'].split("/")[3]
            except:
                self.ProcessedDataset = '  '
            common.taskDB.setDict("CODE",(code+'::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.primaryDataset)+'::'+str(self.ProcessedDataset)))
            pass

        # This is code for dashboard
          
        #First checkProxy
        common.scheduler.checkProxy()
        try:
            fl = open(common.work_space.shareDir() + '/' + self.cfg_params['apmon'].fName, 'w')
            self.cfg_params['GridName'] = runCommand("voms-proxy-info -identity")
            common.logger.debug(5, "GRIDNAME: "+self.cfg_params['GridName'])
            taskType = 'analysis'

            params = {'tool': common.prog_name, 'tool_ui': os.environ['HOSTNAME'], \
                      'scheduler': self.cfg_params['CRAB.scheduler'], 'GridName': self.cfg_params['GridName'].strip(), \
                      'taskType': taskType, 'vo': self.cfg_params['EDG.virtual_organization'], 'user': self.cfg_params['user']}
            jtParam = self.job_type.getParams()
            for i in jtParam.iterkeys():
                params[i] = string.strip(jtParam[i])
            for j, k in params.iteritems():
#                print "Values: %s %s"%(j, k)
                fl.write(j + ':' + k + '\n')
            fl.close()
        except:
            exctype, value = sys.exc_info()[:2]
            common.logger.message("Creator::run Exception raised in collection params for dashboard: %s %s"%(exctype, value))
            pass

        #TODO: deprecated code, not needed,
        # will be eliminated when WorkSpace.saveConfiguration()
        # will be improved.
        #
        # Set/Save Job Type name

        try:
            jt = common.taskDB.dict("JobType")
            if self.job_type_name:
                if ( jt != self.job_type_name+'\n' ):
                    msg = 'Job Type mismatch: requested <' + self.job_type_name
                    msg += '>, found <' + jt[:-1] + '>.'
                    raise CrabException(msg)
                pass
            else:
                self.job_type_name = jt[:-1]
                pass
            pass
        except CrabException:
            # Save job type name
            common.taskDB.setDict("JobType",self.job_type_name)
            pass
        #end of deprecated code

        common.logger.debug(5, "Creator constructor finished")
        return

    def writeJobsSpecsToDB(self):
        """
        Write firstEvent and maxEvents in the DB for future use
        """

        self.job_type.split(self.jobParamsList)
        return

    def nJobs(self):
        return self.total_njobs
    
    def createJobTypeObject(self,ncjobs):
        file_name = 'cms_'+ string.lower(self.job_type_name)
        klass_name = string.capitalize(self.job_type_name)

        try:
            klass = importName(file_name, klass_name)
        except KeyError:
            msg = 'No `class '+klass_name+'` found in file `'+file_name+'.py`'
            raise CrabException(msg)
        except ImportError, e:
            msg = 'Cannot create job type '+self.job_type_name
            msg += ' (file: '+file_name+', class '+klass_name+'):\n'
            msg += str(e)
            raise CrabException(msg)

        self.job_type = klass(self.cfg_params,ncjobs)
        return

    def jobType(self):
        return self.job_type

    def run(self):
        """
        The main method of the class.
        """

        common.logger.debug(5, "Creator::run() called")

        # Instantiate ScriptWriter

        script_writer = ScriptWriter('crab_template.sh')

        # Loop over jobs

        njc = 0
        for nj in range(self.total_njobs):
            if njc == self.ncjobs : break
            st = common.jobDB.status(nj)
            if st != 'X': continue

            common.logger.message("Creating job # "+`(nj+1)`)

            # Prepare configuration file

            self.job_type.modifySteeringCards(nj)
            # Create script (sh)
            script_writer.modifyTemplateScript()
            os.chmod(common.taskDB.dict("ScriptName"), 0744)

            # Create scheduler scripts (jdl)
            common.scheduler.createSchScript(nj)

            common.jobDB.setStatus(nj, 'C')
            # common: write input and output sandbox
            common.jobDB.setInputSandbox(nj, self.job_type.inputSandbox(nj))

            outputSandbox=self.job_type.outputSandbox(nj)
            stdout=common.job_list[nj].stdout()
            stderr=common.job_list[nj].stderr()
            outputSandbox.append(common.job_list[nj].stdout())
            # check if out!=err
            if stdout != stderr:
                outputSandbox.append(common.job_list[nj].stderr())
            common.jobDB.setOutputSandbox(nj, outputSandbox)
            common.jobDB.setTaskId(nj, self.cfg_params['taskId'])

            njc = njc + 1
            pass

        ####

        common.jobDB.save()
        common.taskDB.save()
        msg = '\nTotal of %d jobs created'%njc
        if njc != self.ncjobs: msg = msg + ' from %d requested'%self.ncjobs
        msg = msg + '.\n'
        common.logger.message(msg)
        return
Revision:	1.43
Committed:	Wed Sep 20 17:29:52 2006 UTC (18 years, 7 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
Changes since 1.42:	+10 -29 lines
Log Message:	BOSS4 + sub-file splitting + taskDB
#	User	Rev	Content
1	nsmirnov	1.1	from Actor import Actor
2			from WorkSpace import WorkSpace
3			from JobList import JobList
4			from JobDB import JobDB
5			from ScriptWriter import ScriptWriter
6			from Scheduler import Scheduler
7			from crab_logger import Logger
8			from crab_exceptions import *
9			from crab_util import *
10			import common
11
12	slacapra	1.8	import os, string, math
13	corvo	1.26	# marco
14	nsmirnov	1.1	class Creator(Actor):
15	corvo	1.30	def __init__(self, job_type_name, cfg_params, ncjobs):
16	nsmirnov	1.1	self.job_type_name = job_type_name
17	corvo	1.30	self.job_type = None
18	nsmirnov	1.1	self.cfg_params = cfg_params
19	nsmirnov	1.2	self.total_njobs = 0
20			self.total_number_of_events = 0
21			self.job_number_of_events = 0
22	slacapra	1.7	self.first_event = 0
23	gutsche	1.38	self.jobParamsList=[]
24	slacapra	1.14
25	gutsche	1.42	self.createJobTypeObject(ncjobs)
26	nsmirnov	1.3	common.logger.debug(5, __name__+": JobType "+self.job_type.name()+" created")
27	nsmirnov	1.1
28			self.job_type.prepareSteeringCards()
29	nsmirnov	1.3	common.logger.debug(5, __name__+": Steering cards prepared")
30	nsmirnov	1.1
31	gutsche	1.38	self.total_njobs = self.job_type.numberOfJobs();
32	nsmirnov	1.3	common.logger.debug(5, __name__+": total # of jobs = "+`self.total_njobs`)
33	nsmirnov	1.1
34			# Set number of jobs to be created
35
36			self.ncjobs = ncjobs
37			if ncjobs == 'all' : self.ncjobs = self.total_njobs
38			if ncjobs > self.total_njobs : self.ncjobs = self.total_njobs
39	spiga	1.11
40	slacapra	1.35	# This is code for proto-monitoring
41	slacapra	1.43	code=common.taskDB.dict("CODE")
42			if self.job_type.name() == 'CMSSW':
43	spiga	1.41	try:
44			self.primaryDataset = cfg_params['CMSSW.datasetpath'].split("/")[1]
45			except:
46			self.primaryDataset = 'None'
47			try:
48			self.ProcessedDataset = cfg_params['CMSSW.datasetpath'].split("/")[3]
49			except:
50			self.ProcessedDataset = ' '
51	slacapra	1.43	common.taskDB.setDict("CODE",(code+'::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.primaryDataset)+'::'+str(self.ProcessedDataset)))
52	spiga	1.37	pass
53
54	slacapra	1.35	# This is code for dashboard
55	spiga	1.36
56			#First checkProxy
57			common.scheduler.checkProxy()
58	corvo	1.33	try:
59			fl = open(common.work_space.shareDir() + '/' + self.cfg_params['apmon'].fName, 'w')
60			self.cfg_params['GridName'] = runCommand("voms-proxy-info -identity")
61			common.logger.debug(5, "GRIDNAME: "+self.cfg_params['GridName'])
62			taskType = 'analysis'
63
64			params = {'tool': common.prog_name, 'tool_ui': os.environ['HOSTNAME'], \
65			'scheduler': self.cfg_params['CRAB.scheduler'], 'GridName': self.cfg_params['GridName'].strip(), \
66			'taskType': taskType, 'vo': self.cfg_params['EDG.virtual_organization'], 'user': self.cfg_params['user']}
67			jtParam = self.job_type.getParams()
68			for i in jtParam.iterkeys():
69			params[i] = string.strip(jtParam[i])
70			for j, k in params.iteritems():
71			# print "Values: %s %s"%(j, k)
72			fl.write(j + ':' + k + '\n')
73			fl.close()
74			except:
75			exctype, value = sys.exc_info()[:2]
76	slacapra	1.35	common.logger.message("Creator::run Exception raised in collection params for dashboard: %s %s"%(exctype, value))
77	corvo	1.33	pass
78	nsmirnov	1.1
79			#TODO: deprecated code, not needed,
80			# will be eliminated when WorkSpace.saveConfiguration()
81			# will be improved.
82			#
83			# Set/Save Job Type name
84
85	slacapra	1.43	try:
86			jt = common.taskDB.dict("JobType")
87	nsmirnov	1.1	if self.job_type_name:
88			if ( jt != self.job_type_name+'\n' ):
89			msg = 'Job Type mismatch: requested <' + self.job_type_name
90			msg += '>, found <' + jt[:-1] + '>.'
91			raise CrabException(msg)
92			pass
93			else:
94			self.job_type_name = jt[:-1]
95			pass
96			pass
97	slacapra	1.43	except CrabException:
98	nsmirnov	1.1	# Save job type name
99	slacapra	1.43	common.taskDB.setDict("JobType",self.job_type_name)
100	nsmirnov	1.1	pass
101			#end of deprecated code
102
103	nsmirnov	1.3	common.logger.debug(5, "Creator constructor finished")
104	nsmirnov	1.1	return
105
106	slacapra	1.7	def writeJobsSpecsToDB(self):
107			"""
108			Write firstEvent and maxEvents in the DB for future use
109			"""
110
111	gutsche	1.38	self.job_type.split(self.jobParamsList)
112	nsmirnov	1.2	return
113
114	nsmirnov	1.1	def nJobs(self):
115			return self.total_njobs
116
117	gutsche	1.42	def createJobTypeObject(self,ncjobs):
118	nsmirnov	1.1	file_name = 'cms_'+ string.lower(self.job_type_name)
119			klass_name = string.capitalize(self.job_type_name)
120
121			try:
122			klass = importName(file_name, klass_name)
123			except KeyError:
124			msg = 'No `class '+klass_name+'` found in file `'+file_name+'.py`'
125			raise CrabException(msg)
126			except ImportError, e:
127			msg = 'Cannot create job type '+self.job_type_name
128			msg += ' (file: '+file_name+', class '+klass_name+'):\n'
129			msg += str(e)
130			raise CrabException(msg)
131
132	gutsche	1.42	self.job_type = klass(self.cfg_params,ncjobs)
133	nsmirnov	1.1	return
134	nsmirnov	1.5
135			def jobType(self):
136			return self.job_type
137	nsmirnov	1.1
138			def run(self):
139	nsmirnov	1.4	"""
140			The main method of the class.
141			"""
142	nsmirnov	1.1
143	nsmirnov	1.3	common.logger.debug(5, "Creator::run() called")
144
145	nsmirnov	1.1	# Instantiate ScriptWriter
146
147			script_writer = ScriptWriter('crab_template.sh')
148
149			# Loop over jobs
150
151			njc = 0
152	nsmirnov	1.4	for nj in range(self.total_njobs):
153	nsmirnov	1.1	if njc == self.ncjobs : break
154	nsmirnov	1.4	st = common.jobDB.status(nj)
155	nsmirnov	1.1	if st != 'X': continue
156
157	nsmirnov	1.6	common.logger.message("Creating job # "+`(nj+1)`)
158	nsmirnov	1.3
159	nsmirnov	1.1	# Prepare configuration file
160
161	nsmirnov	1.4	self.job_type.modifySteeringCards(nj)
162	fanzago	1.12	# Create script (sh)
163	slacapra	1.43	script_writer.modifyTemplateScript()
164			os.chmod(common.taskDB.dict("ScriptName"), 0744)
165	nsmirnov	1.1
166	fanzago	1.12	# Create scheduler scripts (jdl)
167			common.scheduler.createSchScript(nj)
168
169	nsmirnov	1.4	common.jobDB.setStatus(nj, 'C')
170	slacapra	1.9	# common: write input and output sandbox
171			common.jobDB.setInputSandbox(nj, self.job_type.inputSandbox(nj))
172
173			outputSandbox=self.job_type.outputSandbox(nj)
174	gutsche	1.32	stdout=common.job_list[nj].stdout()
175	slacapra	1.9	stderr=common.job_list[nj].stderr()
176			outputSandbox.append(common.job_list[nj].stdout())
177			# check if out!=err
178			if stdout != stderr:
179			outputSandbox.append(common.job_list[nj].stderr())
180			common.jobDB.setOutputSandbox(nj, outputSandbox)
181	corvo	1.18	common.jobDB.setTaskId(nj, self.cfg_params['taskId'])
182	slacapra	1.9
183	nsmirnov	1.1	njc = njc + 1
184			pass
185
186			####
187	fanzago	1.12
188	nsmirnov	1.1	common.jobDB.save()
189	slacapra	1.43	common.taskDB.save()
190	nsmirnov	1.1	msg = '\nTotal of %d jobs created'%njc
191			if njc != self.ncjobs: msg = msg + ' from %d requested'%self.ncjobs
192			msg = msg + '.\n'
193			common.logger.message(msg)
194			return