CRAB/python/parseCrabFjr.py

#!/usr/bin/env python

import sys, getopt, string

from ProdCommon.FwkJobRep.ReportParser import readJobReport
from DashboardAPI import apmonSend, apmonFree

class parseFjr:
    def __init__(self, argv):
        """
        parseCrabFjr
 
        - parse CRAB FrameworkJobReport on WN: { 'protocol' : { 'action' : [attempted,succedeed,total-size,total-time,min-time,max-time] , ... } , ... }
        - report parameters to DashBoard using DashBoardApi.py: for all 'read' actions of all protocols, report MBPS
        - return ExitStatus and dump of DashBoard report separated by semi-colon to WN wrapper script
        """
        # defaults
        self.input = ''
        self.MonitorID = ''
        self.MonitorJobID = ''
        self.info2dash = False 
        self.exitCode = False 
        self.lfnList = False  
        self.debug = 0
        try:
            opts, args = getopt.getopt(argv, "", ["input=", "dashboard=", "exitcode", "lfn" , "debug", "help"])
        except getopt.GetoptError:
            print self.usage()
            sys.exit(2)
        self.check(opts)  

        return

    def check(self,opts): 
        # check command line parameter
        for opt, arg in opts :
            if opt  == "--help" :
                print self.usage()
                sys.exit()
            elif opt == "--input" :
                self.input = arg
            elif opt == "--exitcode":
                self.exitCode = True 
            elif opt == "--lfn":
                self.lfnList = True 
            elif opt == "--dashboard":
                self.info2dash = True 
                try: 
                   self.MonitorID = arg.split(",")[0]
                   self.MonitorJobID = arg.split(",")[1]
                except:
                   self.MonitorID = ''
                   self.MonitorJobID = ''
            elif opt == "--debug" :
                self.debug = 1
                
        if self.input == '' or (not self.info2dash and not self.lfnList and not self.exitCode)  :
            print self.usage()
            sys.exit()
         
        if self.info2dash: 
            if self.MonitorID == '' or self.MonitorJobID == '':
                print self.usage()
                sys.exit()
        return

    def run(self): 

        # load FwkJobRep
        jobReport = readJobReport(self.input)[0]
        if self.exitCode : 
            self.exitCodes(jobReport)
        if self.lfnList : 
           self.lfn_List(jobReport)
        if self.info2dash : 
           self.reportDash(jobReport)
        return

    def exitCodes(self, jobReport):

        exit_status = ''
        ##### temporary fix for FJR incomplete ####
        fjr = open (self.input)
        len_fjr = len(fjr.readlines())
        if (len_fjr <= 6):
           ### 50115 - cmsRun did not produce a valid/readable job report at runtime
           exit_status = str(50115)
        else: 
            # get ExitStatus of last error
            if len(jobReport.errors) != 0 :
                exit_status = str(jobReport.errors[-1]['ExitStatus'])
            else :
                exit_status = str(0)
        #check exit code 
        if string.strip(exit_status) == '': exit_status = -999
        print exit_status   
  
        return

    def lfn_List(self,jobReport):
        ''' 
        get list of analyzed files 
        '''
        lfnlist = [x['LFN'] for x in jobReport.inputFiles]
        for file in lfnlist: print file
        return 

    def storageStat(self,jobReport):
        ''' 
        get i/o statistics 
        '''
        storageStatistics = str(jobReport.storageStatistics)
        storage_report = {}
        # check if storageStatistics is valid
        if storageStatistics.find('Storage statistics:') != -1 :
            # report form: { 'protocol' : { 'action' : [attempted,succedeed,total-size,total-time,min-time,max-time] , ... } , ... }
            data = storageStatistics.split('Storage statistics:')[1]
            data_fields = data.split(';')
            for data_field in data_fields:
                # parse: format protocol/action = attepted/succedeed/total-size/total-time/min-time/max-time
                if data_field == ' ' or not data_field or data_field == '':
                   continue
                key = data_field.split('=')[0].strip()
                item = data_field.split('=')[1].strip()
                protocol = str(key.split('/')[0].strip())
                action = str(key.split('/')[1].strip())
                item_array = item.split('/')
                attempted = str(item_array[0].strip())
                succeeded = str(item_array[1].strip())
                total_size = str(item_array[2].strip().split('MB')[0])
                total_time = str(item_array[3].strip().split('ms')[0])
                min_time = str(item_array[4].strip().split('ms')[0])
                max_time = str(item_array[5].strip().split('ms')[0])
                # add to report
                if protocol in storage_report.keys() :
                    if action in storage_report[protocol].keys() :
                        print 'protocol/action:',protocol,'/',action,'listed twice in report, taking the first'
                    else :
                        storage_report[protocol][action] = [attempted,succeeded,total_size,total_time,min_time,max_time]
                else :
                    storage_report[protocol] = {action : [attempted,succeeded,total_size,total_time,min_time,max_time] }
 
            if self.debug :
                for protocol in storage_report.keys() :
                    print 'protocol:',protocol
                    for action in storage_report[protocol].keys() :
                        print 'action:',action,'measurement:',storage_report[protocol][action]
 
        if self.debug == 1 : print storage_report
        return storage_report

    def n_of_events(self,jobReport):
        '''
        #Brian's patch to sent number of events procedded to the Dashboard 
        # Add NoEventsPerRun to the Dashboard report
        '''   
        event_report = {}
        eventsPerRun = 0
        for inputFile in jobReport.inputFiles:
            try:
                eventsRead = str(inputFile.get('EventsRead', 0))
                eventsRead = int(eventsRead.strip())
            except:
                continue
            eventsPerRun += eventsRead
        event_report['NoEventsPerRun'] = eventsPerRun
        event_report['NbEvPerRun'] = eventsPerRun
        event_report['NEventsProcessed'] = eventsPerRun

        if self.debug == 1 : print event_report

        return event_report
       
    def reportDash(self,jobReport):
        '''
        dashboard report dictionary 
        '''
        event_report = self.n_of_events(jobReport)
        storage_report = self.storageStat(jobReport)
        dashboard_report = {}
        #
        for k,v in event_report.iteritems() :
            dashboard_report[k]=v

        # extract information to be sent to DashBoard
        # per protocol and for action=read, calculate MBPS
        # dashboard key is io_action
        dashboard_report['MonitorID'] = self.MonitorID
        dashboard_report['MonitorJobID'] = self.MonitorJobID
        for protocol in storage_report.keys() :
            for action in storage_report[protocol].keys() :
                try: size = float(storage_report[protocol][action][2])
                except: size = 'NULL'
                try: time = float(storage_report[protocol][action][3])/1000
                except: time = 'NULL'
                dashboard_report['io_'+protocol+'_'+action] = str(size)+'_'+str(time)
        if self.debug :
            ordered = dashboard_report.keys()
            ordered.sort()
            for key in ordered:
                print key,'=',dashboard_report[key]
 
        # send to DashBoard
        apmonSend(self.MonitorID, self.MonitorJobID, dashboard_report)
        apmonFree()

        if self.debug == 1 : print dashboard_report

        return

    def usage(self):
        
        msg=""" 
        required parameters:
        --input            :       input FJR xml file
 
        optional parameters:
        --dashboard        :       send info to the dashboard. require following args: "MonitorID,MonitorJobID"
            MonitorID        :       DashBoard MonitorID
            MonitorJobID     :       DashBoard MonitorJobID
        --exitcode         :       print executable exit code 
        --lfn              :       report list of files really analyzed
        --help             :       help
        --debug            :       debug statements
        """
        return msg 


if __name__ == '__main__' :
    try: 
        parseFjr_ = parseFjr(sys.argv[1:]) 
        parseFjr_.run()  
    except:
        pass
Revision:	1.10
Committed:	Thu May 29 21:01:14 2008 UTC (16 years, 11 months ago) by spiga
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_2_2_pre2, CRAB_2_2_2_pre1, CRAB_2_2_1, CRAB_2_2_1_pre6, CRAB_2_2_1_pre5, CRAB_2_2_1_pre4
Changes since 1.9:	+206 -144 lines
Log Message:	moved from debuggin_pset to more general debugging_wrapper. Plus major changes on parse framework job report script. Now it take arguments and return different output according to the requests.
#	User	Rev	Content
1	gutsche	1.1	#!/usr/bin/env python
2
3			import sys, getopt, string
4
5	fanzago	1.6	from ProdCommon.FwkJobRep.ReportParser import readJobReport
6	gutsche	1.2	from DashboardAPI import apmonSend, apmonFree
7
8	spiga	1.10	class parseFjr:
9			def __init__(self, argv):
10			"""
11			parseCrabFjr
12
13			- parse CRAB FrameworkJobReport on WN: { 'protocol' : { 'action' : [attempted,succedeed,total-size,total-time,min-time,max-time] , ... } , ... }
14			- report parameters to DashBoard using DashBoardApi.py: for all 'read' actions of all protocols, report MBPS
15			- return ExitStatus and dump of DashBoard report separated by semi-colon to WN wrapper script
16			"""
17			# defaults
18			self.input = ''
19			self.MonitorID = ''
20			self.MonitorJobID = ''
21			self.info2dash = False
22			self.exitCode = False
23			self.lfnList = False
24			self.debug = 0
25			try:
26			opts, args = getopt.getopt(argv, "", ["input=", "dashboard=", "exitcode", "lfn" , "debug", "help"])
27			except getopt.GetoptError:
28			print self.usage()
29			sys.exit(2)
30			self.check(opts)
31
32			return
33
34			def check(self,opts):
35			# check command line parameter
36			for opt, arg in opts :
37			if opt == "--help" :
38			print self.usage()
39			sys.exit()
40			elif opt == "--input" :
41			self.input = arg
42			elif opt == "--exitcode":
43			self.exitCode = True
44			elif opt == "--lfn":
45			self.lfnList = True
46			elif opt == "--dashboard":
47			self.info2dash = True
48			try:
49			self.MonitorID = arg.split(",")[0]
50			self.MonitorJobID = arg.split(",")[1]
51			except:
52			self.MonitorID = ''
53			self.MonitorJobID = ''
54			elif opt == "--debug" :
55			self.debug = 1
56
57			if self.input == '' or (not self.info2dash and not self.lfnList and not self.exitCode) :
58			print self.usage()
59	gutsche	1.1	sys.exit()
60	spiga	1.10
61			if self.info2dash:
62			if self.MonitorID == '' or self.MonitorJobID == '':
63			print self.usage()
64			sys.exit()
65			return
66
67			def run(self):
68
69			# load FwkJobRep
70			jobReport = readJobReport(self.input)[0]
71			if self.exitCode :
72			self.exitCodes(jobReport)
73			if self.lfnList :
74			self.lfn_List(jobReport)
75			if self.info2dash :
76			self.reportDash(jobReport)
77			return
78
79			def exitCodes(self, jobReport):
80
81			exit_status = ''
82			##### temporary fix for FJR incomplete ####
83			fjr = open (self.input)
84			len_fjr = len(fjr.readlines())
85			if (len_fjr <= 6):
86			### 50115 - cmsRun did not produce a valid/readable job report at runtime
87			exit_status = str(50115)
88			else:
89			# get ExitStatus of last error
90			if len(jobReport.errors) != 0 :
91			exit_status = str(jobReport.errors[-1]['ExitStatus'])
92			else :
93			exit_status = str(0)
94			#check exit code
95			if string.strip(exit_status) == '': exit_status = -999
96			print exit_status
97
98			return
99
100			def lfn_List(self,jobReport):
101			'''
102			get list of analyzed files
103			'''
104			lfnlist = [x['LFN'] for x in jobReport.inputFiles]
105			for file in lfnlist: print file
106			return
107
108			def storageStat(self,jobReport):
109			'''
110			get i/o statistics
111			'''
112			storageStatistics = str(jobReport.storageStatistics)
113			storage_report = {}
114			# check if storageStatistics is valid
115			if storageStatistics.find('Storage statistics:') != -1 :
116			# report form: { 'protocol' : { 'action' : [attempted,succedeed,total-size,total-time,min-time,max-time] , ... } , ... }
117			data = storageStatistics.split('Storage statistics:')[1]
118			data_fields = data.split(';')
119			for data_field in data_fields:
120			# parse: format protocol/action = attepted/succedeed/total-size/total-time/min-time/max-time
121			if data_field == ' ' or not data_field or data_field == '':
122			continue
123			key = data_field.split('=')[0].strip()
124			item = data_field.split('=')[1].strip()
125			protocol = str(key.split('/')[0].strip())
126			action = str(key.split('/')[1].strip())
127			item_array = item.split('/')
128			attempted = str(item_array[0].strip())
129			succeeded = str(item_array[1].strip())
130			total_size = str(item_array[2].strip().split('MB')[0])
131			total_time = str(item_array[3].strip().split('ms')[0])
132			min_time = str(item_array[4].strip().split('ms')[0])
133			max_time = str(item_array[5].strip().split('ms')[0])
134			# add to report
135			if protocol in storage_report.keys() :
136			if action in storage_report[protocol].keys() :
137			print 'protocol/action:',protocol,'/',action,'listed twice in report, taking the first'
138			else :
139			storage_report[protocol][action] = [attempted,succeeded,total_size,total_time,min_time,max_time]
140	gutsche	1.2	else :
141	spiga	1.10	storage_report[protocol] = {action : [attempted,succeeded,total_size,total_time,min_time,max_time] }
142
143			if self.debug :
144			for protocol in storage_report.keys() :
145			print 'protocol:',protocol
146			for action in storage_report[protocol].keys() :
147			print 'action:',action,'measurement:',storage_report[protocol][action]
148
149			if self.debug == 1 : print storage_report
150			return storage_report
151
152			def n_of_events(self,jobReport):
153			'''
154			#Brian's patch to sent number of events procedded to the Dashboard
155			# Add NoEventsPerRun to the Dashboard report
156			'''
157			event_report = {}
158			eventsPerRun = 0
159			for inputFile in jobReport.inputFiles:
160			try:
161			eventsRead = str(inputFile.get('EventsRead', 0))
162			eventsRead = int(eventsRead.strip())
163			except:
164			continue
165			eventsPerRun += eventsRead
166			event_report['NoEventsPerRun'] = eventsPerRun
167			event_report['NbEvPerRun'] = eventsPerRun
168			event_report['NEventsProcessed'] = eventsPerRun
169
170			if self.debug == 1 : print event_report
171
172			return event_report
173
174			def reportDash(self,jobReport):
175			'''
176			dashboard report dictionary
177			'''
178			event_report = self.n_of_events(jobReport)
179			storage_report = self.storageStat(jobReport)
180			dashboard_report = {}
181			#
182			for k,v in event_report.iteritems() :
183			dashboard_report[k]=v
184	gutsche	1.2
185			# extract information to be sent to DashBoard
186			# per protocol and for action=read, calculate MBPS
187			# dashboard key is io_action
188	spiga	1.10	dashboard_report['MonitorID'] = self.MonitorID
189			dashboard_report['MonitorJobID'] = self.MonitorJobID
190			for protocol in storage_report.keys() :
191			for action in storage_report[protocol].keys() :
192			try: size = float(storage_report[protocol][action][2])
193	gutsche	1.3	except: size = 'NULL'
194	spiga	1.10	try: time = float(storage_report[protocol][action][3])/1000
195	gutsche	1.3	except: time = 'NULL'
196			dashboard_report['io_'+protocol+'_'+action] = str(size)+'_'+str(time)
197	spiga	1.10	if self.debug :
198	gutsche	1.3	ordered = dashboard_report.keys()
199			ordered.sort()
200			for key in ordered:
201			print key,'=',dashboard_report[key]
202	spiga	1.10
203	gutsche	1.2	# send to DashBoard
204	spiga	1.10	apmonSend(self.MonitorID, self.MonitorJobID, dashboard_report)
205	gutsche	1.2	apmonFree()
206
207	spiga	1.10	if self.debug == 1 : print dashboard_report
208	gutsche	1.2
209	spiga	1.10	return
210	gutsche	1.5
211	spiga	1.10	def usage(self):
212
213			msg="""
214			required parameters:
215			--input : input FJR xml file
216
217			optional parameters:
218			--dashboard : send info to the dashboard. require following args: "MonitorID,MonitorJobID"
219			MonitorID : DashBoard MonitorID
220			MonitorJobID : DashBoard MonitorJobID
221			--exitcode : print executable exit code
222			--lfn : report list of files really analyzed
223			--help : help
224			--debug : debug statements
225			"""
226			return msg
227	gutsche	1.1
228
229			if __name__ == '__main__' :
230	spiga	1.10	try:
231			parseFjr_ = parseFjr(sys.argv[1:])
232			parseFjr_.run()
233			except:
234			pass