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()
        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
        fileCODE1 = open(common.work_space.shareDir()+"/.code","a")
        if self.job_type.name() == 'ORCA' or self.job_type.name() == 'ORCA_PUBDB' :
            self.owner = cfg_params['ORCA.owner']
            self.dataset = cfg_params['ORCA.dataset']
            fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.dataset)+'::'+str(self.owner))
            pass
        elif self.job_type.name() == 'FAMOS':
            self.inputFile = cfg_params['FAMOS.input_lfn']
            self.executable = cfg_params['FAMOS.executable']
            fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.inputFile)+'::'+str(self.executable))
            pass
        elif self.job_type.name() == 'CMSSW':
            self.primaryDataset = cfg_params['CMSSW.datasetpath'].split("/")[1]
            self.ProcessedDataset = cfg_params['CMSSW.datasetpath'].split("/")[3]
            fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.primaryDataset)+'::'+str(self.ProcessedDataset))
            pass

        fileCODE1.close()

        # 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

        jt_fname = common.work_space.shareDir() + 'jobtype'
        if os.path.exists(jt_fname):
            # Read stored job type name
            jt_file = open(jt_fname, 'r')
            jt = jt_file.read()
            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
            jt_file.close()
            pass
        else:
            # Save job type name
            jt_file = open(jt_fname, 'w')
            jt_file.write(self.job_type_name+'\n')
            jt_file.close()
            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):
        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)
        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)
            #print "nel for prima del modifyTemplateScript, nj vale", nj
            script_writer.modifyTemplateScript(nj)
            os.chmod(common.job_list[nj].scriptFilename(), 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()

        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:	1.40
Committed:	Fri Jul 7 14:35:17 2006 UTC (18 years, 9 months ago) by spiga
Content type:	text/x-python
Branch:	MAIN
Changes since 1.39:	+3 -3 lines
Log Message:	updated for cmssw jobtype
#	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	corvo	1.30	self.createJobTypeObject()
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	spiga	1.20	fileCODE1 = open(common.work_space.shareDir()+"/.code","a")
42	slacapra	1.39	if self.job_type.name() == 'ORCA' or self.job_type.name() == 'ORCA_PUBDB' :
43	spiga	1.20	self.owner = cfg_params['ORCA.owner']
44			self.dataset = cfg_params['ORCA.dataset']
45			fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.dataset)+'::'+str(self.owner))
46			pass
47			elif self.job_type.name() == 'FAMOS':
48	spiga	1.27	self.inputFile = cfg_params['FAMOS.input_lfn']
49			self.executable = cfg_params['FAMOS.executable']
50			fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.inputFile)+'::'+str(self.executable))
51	spiga	1.20	pass
52	spiga	1.37	elif self.job_type.name() == 'CMSSW':
53	spiga	1.40	self.primaryDataset = cfg_params['CMSSW.datasetpath'].split("/")[1]
54			self.ProcessedDataset = cfg_params['CMSSW.datasetpath'].split("/")[3]
55			fileCODE1.write('::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.primaryDataset)+'::'+str(self.ProcessedDataset))
56	spiga	1.37	pass
57
58	spiga	1.11	fileCODE1.close()
59
60	slacapra	1.35	# This is code for dashboard
61	spiga	1.36
62			#First checkProxy
63			common.scheduler.checkProxy()
64	corvo	1.33	try:
65			fl = open(common.work_space.shareDir() + '/' + self.cfg_params['apmon'].fName, 'w')
66			self.cfg_params['GridName'] = runCommand("voms-proxy-info -identity")
67			common.logger.debug(5, "GRIDNAME: "+self.cfg_params['GridName'])
68			taskType = 'analysis'
69
70			params = {'tool': common.prog_name, 'tool_ui': os.environ['HOSTNAME'], \
71			'scheduler': self.cfg_params['CRAB.scheduler'], 'GridName': self.cfg_params['GridName'].strip(), \
72			'taskType': taskType, 'vo': self.cfg_params['EDG.virtual_organization'], 'user': self.cfg_params['user']}
73			jtParam = self.job_type.getParams()
74			for i in jtParam.iterkeys():
75			params[i] = string.strip(jtParam[i])
76			for j, k in params.iteritems():
77			# print "Values: %s %s"%(j, k)
78			fl.write(j + ':' + k + '\n')
79			fl.close()
80			except:
81			exctype, value = sys.exc_info()[:2]
82	slacapra	1.35	common.logger.message("Creator::run Exception raised in collection params for dashboard: %s %s"%(exctype, value))
83	corvo	1.33	pass
84	nsmirnov	1.1
85			#TODO: deprecated code, not needed,
86			# will be eliminated when WorkSpace.saveConfiguration()
87			# will be improved.
88			#
89			# Set/Save Job Type name
90
91			jt_fname = common.work_space.shareDir() + 'jobtype'
92			if os.path.exists(jt_fname):
93			# Read stored job type name
94			jt_file = open(jt_fname, 'r')
95			jt = jt_file.read()
96			if self.job_type_name:
97			if ( jt != self.job_type_name+'\n' ):
98			msg = 'Job Type mismatch: requested <' + self.job_type_name
99			msg += '>, found <' + jt[:-1] + '>.'
100			raise CrabException(msg)
101			pass
102			else:
103			self.job_type_name = jt[:-1]
104			pass
105			jt_file.close()
106			pass
107			else:
108			# Save job type name
109			jt_file = open(jt_fname, 'w')
110			jt_file.write(self.job_type_name+'\n')
111			jt_file.close()
112			pass
113			#end of deprecated code
114
115	nsmirnov	1.3	common.logger.debug(5, "Creator constructor finished")
116	nsmirnov	1.1	return
117
118	slacapra	1.7	def writeJobsSpecsToDB(self):
119			"""
120			Write firstEvent and maxEvents in the DB for future use
121			"""
122
123	gutsche	1.38	self.job_type.split(self.jobParamsList)
124	nsmirnov	1.2	return
125
126	nsmirnov	1.1	def nJobs(self):
127			return self.total_njobs
128
129			def createJobTypeObject(self):
130			file_name = 'cms_'+ string.lower(self.job_type_name)
131			klass_name = string.capitalize(self.job_type_name)
132
133			try:
134			klass = importName(file_name, klass_name)
135			except KeyError:
136			msg = 'No `class '+klass_name+'` found in file `'+file_name+'.py`'
137			raise CrabException(msg)
138			except ImportError, e:
139			msg = 'Cannot create job type '+self.job_type_name
140			msg += ' (file: '+file_name+', class '+klass_name+'):\n'
141			msg += str(e)
142			raise CrabException(msg)
143
144			self.job_type = klass(self.cfg_params)
145			return
146	nsmirnov	1.5
147			def jobType(self):
148			return self.job_type
149	nsmirnov	1.1
150			def run(self):
151	nsmirnov	1.4	"""
152			The main method of the class.
153			"""
154	nsmirnov	1.1
155	nsmirnov	1.3	common.logger.debug(5, "Creator::run() called")
156
157	nsmirnov	1.1	# Instantiate ScriptWriter
158
159			script_writer = ScriptWriter('crab_template.sh')
160
161			# Loop over jobs
162
163			njc = 0
164	nsmirnov	1.4	for nj in range(self.total_njobs):
165	nsmirnov	1.1	if njc == self.ncjobs : break
166	nsmirnov	1.4	st = common.jobDB.status(nj)
167	nsmirnov	1.1	if st != 'X': continue
168
169	nsmirnov	1.6	common.logger.message("Creating job # "+`(nj+1)`)
170	nsmirnov	1.3
171	nsmirnov	1.1	# Prepare configuration file
172
173	nsmirnov	1.4	self.job_type.modifySteeringCards(nj)
174	fanzago	1.12	# Create script (sh)
175	fanzago	1.13	#print "nel for prima del modifyTemplateScript, nj vale", nj
176	nsmirnov	1.4	script_writer.modifyTemplateScript(nj)
177			os.chmod(common.job_list[nj].scriptFilename(), 0744)
178	nsmirnov	1.1
179	fanzago	1.12	# Create scheduler scripts (jdl)
180			common.scheduler.createSchScript(nj)
181
182	nsmirnov	1.4	common.jobDB.setStatus(nj, 'C')
183	slacapra	1.9	# common: write input and output sandbox
184			common.jobDB.setInputSandbox(nj, self.job_type.inputSandbox(nj))
185
186			outputSandbox=self.job_type.outputSandbox(nj)
187	gutsche	1.32	stdout=common.job_list[nj].stdout()
188	slacapra	1.9	stderr=common.job_list[nj].stderr()
189			outputSandbox.append(common.job_list[nj].stdout())
190			# check if out!=err
191			if stdout != stderr:
192			outputSandbox.append(common.job_list[nj].stderr())
193			common.jobDB.setOutputSandbox(nj, outputSandbox)
194	corvo	1.18	common.jobDB.setTaskId(nj, self.cfg_params['taskId'])
195	slacapra	1.9
196	nsmirnov	1.1	njc = njc + 1
197			pass
198
199			####
200	fanzago	1.12
201	nsmirnov	1.1	common.jobDB.save()
202
203			msg = '\nTotal of %d jobs created'%njc
204			if njc != self.ncjobs: msg = msg + ' from %d requested'%self.ncjobs
205			msg = msg + '.\n'
206			common.logger.message(msg)
207			return