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


        self.UseServer=0
        try:
            self.UseServer=int(self.cfg_params['CRAB.server_mode'])
        except KeyError:
            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...')
        # modified to support server mode -  DS
        # if crab is used in server mode the client must skyp 
        # boss declare step. In this case it is performed at server level  
        # just before the submission  step   
        if (self.UseServer== 0):
            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:	1.50
Committed:	Mon May 28 15:32:23 2007 UTC (17 years, 11 months ago) by spiga
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_0_0_pre9, CRAB_1_5_4, CRAB_1_5_4_pre2, CRAB_1_5_4_pre1, CRAB_2_0_0_pre7, CRAB_2_0_0_pre6, CRAB_1_5_3, CRAB_1_5_3_pre5, CRAB_1_5_3_pre4, CRAB_2_0_0_pre5, CRAB_1_5_3_pre3, configure, CRAB_2_0_0_pre4, CRAB_1_5_3_pre2, CRAB_1_5_3_pre1, CRAB_2_0_0_pre3, CRAB_2_0_0_pre2, CRAB_2_0_0_pre1
Branch point for:	CRAB_1_5_4_SLC3_start
Changes since 1.49:	+13 -2 lines
Log Message:	just a minor temporary workaround to support server_mode = 1 [if servermode =1--> move boss declaration at creation level to the submission one]
#	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	spiga	1.50
57			self.UseServer=0
58			try:
59			self.UseServer=int(self.cfg_params['CRAB.server_mode'])
60			except KeyError:
61			pass
62
63	slacapra	1.35	# This is code for dashboard
64	spiga	1.36	#First checkProxy
65			common.scheduler.checkProxy()
66	corvo	1.33	try:
67			fl = open(common.work_space.shareDir() + '/' + self.cfg_params['apmon'].fName, 'w')
68			self.cfg_params['GridName'] = runCommand("voms-proxy-info -identity")
69			common.logger.debug(5, "GRIDNAME: "+self.cfg_params['GridName'])
70			taskType = 'analysis'
71	slacapra	1.48	try: VO = cfg_params['EDG.virtual_organization']
72			except KeyError: VO = 'cms'
73	corvo	1.33
74	slacapra	1.47	params = {'tool': common.prog_name,\
75			'JSToolVersion': common.prog_version_str, \
76			'tool_ui': os.environ['HOSTNAME'], \
77			'scheduler': self.cfg_params['CRAB.scheduler'], \
78			'GridName': self.cfg_params['GridName'].strip(), \
79			'taskType': taskType, \
80	slacapra	1.48	'vo': VO, \
81	slacapra	1.47	'user': self.cfg_params['user']}
82	corvo	1.33	jtParam = self.job_type.getParams()
83			for i in jtParam.iterkeys():
84			params[i] = string.strip(jtParam[i])
85			for j, k in params.iteritems():
86			# print "Values: %s %s"%(j, k)
87			fl.write(j + ':' + k + '\n')
88			fl.close()
89			except:
90			exctype, value = sys.exc_info()[:2]
91	slacapra	1.35	common.logger.message("Creator::run Exception raised in collection params for dashboard: %s %s"%(exctype, value))
92	corvo	1.33	pass
93	nsmirnov	1.1
94			# Set/Save Job Type name
95	slacapra	1.49	self.job_type_name = common.taskDB.dict("jobtype")
96	nsmirnov	1.1
97	nsmirnov	1.3	common.logger.debug(5, "Creator constructor finished")
98	nsmirnov	1.1	return
99
100	slacapra	1.7	def writeJobsSpecsToDB(self):
101			"""
102			Write firstEvent and maxEvents in the DB for future use
103			"""
104
105	gutsche	1.38	self.job_type.split(self.jobParamsList)
106	nsmirnov	1.2	return
107
108	nsmirnov	1.1	def nJobs(self):
109			return self.total_njobs
110
111	gutsche	1.42	def createJobTypeObject(self,ncjobs):
112	nsmirnov	1.1	file_name = 'cms_'+ string.lower(self.job_type_name)
113			klass_name = string.capitalize(self.job_type_name)
114
115			try:
116			klass = importName(file_name, klass_name)
117			except KeyError:
118			msg = 'No `class '+klass_name+'` found in file `'+file_name+'.py`'
119			raise CrabException(msg)
120			except ImportError, e:
121			msg = 'Cannot create job type '+self.job_type_name
122			msg += ' (file: '+file_name+', class '+klass_name+'):\n'
123			msg += str(e)
124			raise CrabException(msg)
125
126	gutsche	1.42	self.job_type = klass(self.cfg_params,ncjobs)
127	nsmirnov	1.1	return
128	nsmirnov	1.5
129			def jobType(self):
130			return self.job_type
131	nsmirnov	1.1
132			def run(self):
133	nsmirnov	1.4	"""
134			The main method of the class.
135			"""
136	nsmirnov	1.1
137	nsmirnov	1.3	common.logger.debug(5, "Creator::run() called")
138	slacapra	1.45	start = time.time()
139	nsmirnov	1.3
140	nsmirnov	1.1	# Instantiate ScriptWriter
141
142			script_writer = ScriptWriter('crab_template.sh')
143
144			# Loop over jobs
145	slacapra	1.44	argsList = []
146	nsmirnov	1.1
147			njc = 0
148	slacapra	1.46	block = -1 # first block is 0
149			lastDest=''
150	nsmirnov	1.4	for nj in range(self.total_njobs):
151	nsmirnov	1.1	if njc == self.ncjobs : break
152	nsmirnov	1.4	st = common.jobDB.status(nj)
153	nsmirnov	1.1	if st != 'X': continue
154
155	slacapra	1.44	common.logger.debug(1,"Creating job # "+`(nj+1)`)
156	nsmirnov	1.3
157	nsmirnov	1.1	# Prepare configuration file
158
159	nsmirnov	1.4	self.job_type.modifySteeringCards(nj)
160	fanzago	1.12	# Create script (sh)
161	slacapra	1.43	script_writer.modifyTemplateScript()
162			os.chmod(common.taskDB.dict("ScriptName"), 0744)
163	nsmirnov	1.1
164	slacapra	1.44	# Create XML script and declare related jobs
165			argsList.append( str(nj+1)+' '+ self.job_type.getJobTypeArguments(nj, self.cfg_params['CRAB.scheduler']) )
166	fanzago	1.12
167	nsmirnov	1.4	common.jobDB.setStatus(nj, 'C')
168	slacapra	1.9	# common: write input and output sandbox
169			common.jobDB.setInputSandbox(nj, self.job_type.inputSandbox(nj))
170
171			outputSandbox=self.job_type.outputSandbox(nj)
172			# check if out!=err
173			common.jobDB.setOutputSandbox(nj, outputSandbox)
174	corvo	1.18	common.jobDB.setTaskId(nj, self.cfg_params['taskId'])
175	slacapra	1.9
176	slacapra	1.46	currDest=common.jobDB.destination(nj)
177			if (currDest!=lastDest):
178			block+=1
179			lastDest = currDest
180			pass
181			common.jobDB.setBlock(nj,block)
182	nsmirnov	1.1	njc = njc + 1
183			pass
184
185			####
186	slacapra	1.44	common.scheduler.createXMLSchScript(self.total_njobs, argsList)
187			common.logger.message('Creating '+str(self.total_njobs)+' jobs, please wait...')
188	spiga	1.50	# modified to support server mode - DS
189			# if crab is used in server mode the client must skyp
190			# boss declare step. In this case it is performed at server level
191			# just before the submission step
192			if (self.UseServer== 0):
193			common.scheduler.declareJob_() #Add for BOSS4
194	fanzago	1.12
195	slacapra	1.45	stop = time.time()
196	slacapra	1.47	common.logger.debug(2, "Creation Time: "+str(stop - start))
197	slacapra	1.45	common.logger.write("Creation Time: "+str(stop - start))
198
199	nsmirnov	1.1	common.jobDB.save()
200	slacapra	1.43	common.taskDB.save()
201	nsmirnov	1.1	msg = '\nTotal of %d jobs created'%njc
202			if njc != self.ncjobs: msg = msg + ' from %d requested'%self.ncjobs
203			msg = msg + '.\n'
204			common.logger.message(msg)
205	slacapra	1.45
206	nsmirnov	1.1	return