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
import time

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
        # --------------------------------->  Boss4: changed, "all" by default 

        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'
            try: VO = cfg_params['EDG.virtual_organization']
            except KeyError: VO = 'cms'

            params = {'tool': common.prog_name,\
                      'JSToolVersion': common.prog_version_str, \
                      'tool_ui': os.environ['HOSTNAME'], \
                      'scheduler': self.cfg_params['CRAB.scheduler'], \
                      'GridName': self.cfg_params['GridName'].strip(), \
                      'taskType': taskType, \
                      'vo': VO, \
                      '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

        # Set/Save Job Type name
        self.job_type_name = common.taskDB.dict("jobtype")

        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")
        start = time.time()

        # Instantiate ScriptWriter

        script_writer = ScriptWriter('crab_template.sh')

        # Loop over jobs
        argsList = []

        njc = 0
        block = -1 # first block is 0
        lastDest=''
        for nj in range(self.total_njobs):
            if njc == self.ncjobs : break
            st = common.jobDB.status(nj)
            if st != 'X': continue

            common.logger.debug(1,"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 XML script and declare related jobs 
            argsList.append( str(nj+1)+' '+ self.job_type.getJobTypeArguments(nj, self.cfg_params['CRAB.scheduler']) )

            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)
            # check if out!=err
            common.jobDB.setOutputSandbox(nj, outputSandbox)
            common.jobDB.setTaskId(nj, self.cfg_params['taskId'])

            currDest=common.jobDB.destination(nj)
            if (currDest!=lastDest):
                block+=1
                lastDest = currDest
                pass
            common.jobDB.setBlock(nj,block)
            njc = njc + 1
            pass

        ####
        common.scheduler.createXMLSchScript(self.total_njobs, argsList)
        common.logger.message('Creating '+str(self.total_njobs)+' jobs, please wait...')
        common.scheduler.declareJob_()   #Add for BOSS4

        stop = time.time()
        common.logger.debug(2, "Creation Time: "+str(stop - start))
        common.logger.write("Creation Time: "+str(stop - start))

        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:	1.49
Committed:	Tue Feb 27 17:46:47 2007 UTC (18 years, 2 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_1_5_2, CRAB_1_5_1, CRAB_1_5_1_pre4, CRAB_1_5_1_pre3, CRAB_1_5_1_pre2, CRAB_1_5_1_pre1, CRAB_1_5_0, CRAB_1_5_0_pre9, CRAB_1_5_0_pre8, CRAB_1_5_0_pre7
Changes since 1.48:	+1 -22 lines
Log Message:	cleanup of deprecated code + use unique tag for jobtype from 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	slacapra	1.45	import time
14
15	nsmirnov	1.1	class Creator(Actor):
16	corvo	1.30	def __init__(self, job_type_name, cfg_params, ncjobs):
17	nsmirnov	1.1	self.job_type_name = job_type_name
18	corvo	1.30	self.job_type = None
19	nsmirnov	1.1	self.cfg_params = cfg_params
20	nsmirnov	1.2	self.total_njobs = 0
21			self.total_number_of_events = 0
22			self.job_number_of_events = 0
23	slacapra	1.7	self.first_event = 0
24	gutsche	1.38	self.jobParamsList=[]
25	slacapra	1.14
26	gutsche	1.42	self.createJobTypeObject(ncjobs)
27	nsmirnov	1.3	common.logger.debug(5, __name__+": JobType "+self.job_type.name()+" created")
28	nsmirnov	1.1
29			self.job_type.prepareSteeringCards()
30	nsmirnov	1.3	common.logger.debug(5, __name__+": Steering cards prepared")
31	nsmirnov	1.1
32	gutsche	1.38	self.total_njobs = self.job_type.numberOfJobs();
33	nsmirnov	1.3	common.logger.debug(5, __name__+": total # of jobs = "+`self.total_njobs`)
34	nsmirnov	1.1
35			# Set number of jobs to be created
36	slacapra	1.44	# ---------------------------------> Boss4: changed, "all" by default
37	nsmirnov	1.1
38			self.ncjobs = ncjobs
39			if ncjobs == 'all' : self.ncjobs = self.total_njobs
40			if ncjobs > self.total_njobs : self.ncjobs = self.total_njobs
41	spiga	1.11
42	slacapra	1.35	# This is code for proto-monitoring
43	slacapra	1.43	code=common.taskDB.dict("CODE")
44			if self.job_type.name() == 'CMSSW':
45	spiga	1.41	try:
46			self.primaryDataset = cfg_params['CMSSW.datasetpath'].split("/")[1]
47			except:
48			self.primaryDataset = 'None'
49			try:
50			self.ProcessedDataset = cfg_params['CMSSW.datasetpath'].split("/")[3]
51			except:
52			self.ProcessedDataset = ' '
53	slacapra	1.43	common.taskDB.setDict("CODE",(code+'::'+str(self.job_type.name())+'::'+str(self.ncjobs)+'::'+str(self.primaryDataset)+'::'+str(self.ProcessedDataset)))
54	spiga	1.37	pass
55
56	slacapra	1.35	# This is code for dashboard
57	spiga	1.36
58			#First checkProxy
59			common.scheduler.checkProxy()
60	corvo	1.33	try:
61			fl = open(common.work_space.shareDir() + '/' + self.cfg_params['apmon'].fName, 'w')
62			self.cfg_params['GridName'] = runCommand("voms-proxy-info -identity")
63			common.logger.debug(5, "GRIDNAME: "+self.cfg_params['GridName'])
64			taskType = 'analysis'
65	slacapra	1.48	try: VO = cfg_params['EDG.virtual_organization']
66			except KeyError: VO = 'cms'
67	corvo	1.33
68	slacapra	1.47	params = {'tool': common.prog_name,\
69			'JSToolVersion': common.prog_version_str, \
70			'tool_ui': os.environ['HOSTNAME'], \
71			'scheduler': self.cfg_params['CRAB.scheduler'], \
72			'GridName': self.cfg_params['GridName'].strip(), \
73			'taskType': taskType, \
74	slacapra	1.48	'vo': VO, \
75	slacapra	1.47	'user': self.cfg_params['user']}
76	corvo	1.33	jtParam = self.job_type.getParams()
77			for i in jtParam.iterkeys():
78			params[i] = string.strip(jtParam[i])
79			for j, k in params.iteritems():
80			# print "Values: %s %s"%(j, k)
81			fl.write(j + ':' + k + '\n')
82			fl.close()
83			except:
84			exctype, value = sys.exc_info()[:2]
85	slacapra	1.35	common.logger.message("Creator::run Exception raised in collection params for dashboard: %s %s"%(exctype, value))
86	corvo	1.33	pass
87	nsmirnov	1.1
88			# Set/Save Job Type name
89	slacapra	1.49	self.job_type_name = common.taskDB.dict("jobtype")
90	nsmirnov	1.1
91	nsmirnov	1.3	common.logger.debug(5, "Creator constructor finished")
92	nsmirnov	1.1	return
93
94	slacapra	1.7	def writeJobsSpecsToDB(self):
95			"""
96			Write firstEvent and maxEvents in the DB for future use
97			"""
98
99	gutsche	1.38	self.job_type.split(self.jobParamsList)
100	nsmirnov	1.2	return
101
102	nsmirnov	1.1	def nJobs(self):
103			return self.total_njobs
104
105	gutsche	1.42	def createJobTypeObject(self,ncjobs):
106	nsmirnov	1.1	file_name = 'cms_'+ string.lower(self.job_type_name)
107			klass_name = string.capitalize(self.job_type_name)
108
109			try:
110			klass = importName(file_name, klass_name)
111			except KeyError:
112			msg = 'No `class '+klass_name+'` found in file `'+file_name+'.py`'
113			raise CrabException(msg)
114			except ImportError, e:
115			msg = 'Cannot create job type '+self.job_type_name
116			msg += ' (file: '+file_name+', class '+klass_name+'):\n'
117			msg += str(e)
118			raise CrabException(msg)
119
120	gutsche	1.42	self.job_type = klass(self.cfg_params,ncjobs)
121	nsmirnov	1.1	return
122	nsmirnov	1.5
123			def jobType(self):
124			return self.job_type
125	nsmirnov	1.1
126			def run(self):
127	nsmirnov	1.4	"""
128			The main method of the class.
129			"""
130	nsmirnov	1.1
131	nsmirnov	1.3	common.logger.debug(5, "Creator::run() called")
132	slacapra	1.45	start = time.time()
133	nsmirnov	1.3
134	nsmirnov	1.1	# Instantiate ScriptWriter
135
136			script_writer = ScriptWriter('crab_template.sh')
137
138			# Loop over jobs
139	slacapra	1.44	argsList = []
140	nsmirnov	1.1
141			njc = 0
142	slacapra	1.46	block = -1 # first block is 0
143			lastDest=''
144	nsmirnov	1.4	for nj in range(self.total_njobs):
145	nsmirnov	1.1	if njc == self.ncjobs : break
146	nsmirnov	1.4	st = common.jobDB.status(nj)
147	nsmirnov	1.1	if st != 'X': continue
148
149	slacapra	1.44	common.logger.debug(1,"Creating job # "+`(nj+1)`)
150	nsmirnov	1.3
151	nsmirnov	1.1	# Prepare configuration file
152
153	nsmirnov	1.4	self.job_type.modifySteeringCards(nj)
154	fanzago	1.12	# Create script (sh)
155	slacapra	1.43	script_writer.modifyTemplateScript()
156			os.chmod(common.taskDB.dict("ScriptName"), 0744)
157	nsmirnov	1.1
158	slacapra	1.44	# Create XML script and declare related jobs
159			argsList.append( str(nj+1)+' '+ self.job_type.getJobTypeArguments(nj, self.cfg_params['CRAB.scheduler']) )
160	fanzago	1.12
161	nsmirnov	1.4	common.jobDB.setStatus(nj, 'C')
162	slacapra	1.9	# common: write input and output sandbox
163			common.jobDB.setInputSandbox(nj, self.job_type.inputSandbox(nj))
164
165			outputSandbox=self.job_type.outputSandbox(nj)
166			# check if out!=err
167			common.jobDB.setOutputSandbox(nj, outputSandbox)
168	corvo	1.18	common.jobDB.setTaskId(nj, self.cfg_params['taskId'])
169	slacapra	1.9
170	slacapra	1.46	currDest=common.jobDB.destination(nj)
171			if (currDest!=lastDest):
172			block+=1
173			lastDest = currDest
174			pass
175			common.jobDB.setBlock(nj,block)
176	nsmirnov	1.1	njc = njc + 1
177			pass
178
179			####
180	slacapra	1.44	common.scheduler.createXMLSchScript(self.total_njobs, argsList)
181			common.logger.message('Creating '+str(self.total_njobs)+' jobs, please wait...')
182			common.scheduler.declareJob_() #Add for BOSS4
183	fanzago	1.12
184	slacapra	1.45	stop = time.time()
185	slacapra	1.47	common.logger.debug(2, "Creation Time: "+str(stop - start))
186	slacapra	1.45	common.logger.write("Creation Time: "+str(stop - start))
187
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	slacapra	1.45
195	nsmirnov	1.1	return