ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/COMP/CRAB/python/Publisher.py

Comparing COMP/CRAB/python/Publisher.py (file contents):
Revision 1.11.4.1 by fanzago, Fri Jul 4 14:01:01 2008 UTC vs.
Revision 1.41 by fanzago, Tue Sep 22 15:15:36 2009 UTC

#	Line 3 | Line 3 | import common
3		import time, glob
4		from Actor import *
5		from crab_util import *
6	–	from crab_logger import Logger
6		from crab_exceptions import *
7		from ProdCommon.FwkJobRep.ReportParser import readJobReport
8	+	from ProdCommon.FwkJobRep.ReportState import checkSuccess
9		from ProdCommon.MCPayloads.WorkflowSpec import WorkflowSpec
10		from ProdCommon.DataMgmt.DBS.DBSWriter import DBSWriter
11		from ProdCommon.DataMgmt.DBS.DBSErrors import DBSWriterError, formatEx,DBSReaderError
12		from ProdCommon.DataMgmt.DBS.DBSReader import DBSReader
13		from ProdCommon.DataMgmt.DBS.DBSWriter import DBSWriter,DBSWriterObjects
14		import sys
15	+	from DBSAPI.dbsApiException import DbsException
16	+	from DBSAPI.dbsApi import DbsApi
17
18		class Publisher(Actor):
19		def __init__(self, cfg_params):
#	Line 23 | Line 25 | class Publisher(Actor):
25		- publishes output data on DBS and DLS
26		"""
27
28	<	try:
29	<	self.processedData = cfg_params['USER.publish_data_name']
30	<	except KeyError:
28	>	self.cfg_params=cfg_params
29	>
30	>	if not cfg_params.has_key('USER.publish_data_name'):
31		raise CrabException('Cannot publish output data, because you did not specify USER.publish_data_name parameter in the crab.cfg file')
32	+	self.userprocessedData = cfg_params['USER.publish_data_name']
33	+	self.processedData = None
34
35	<	try:
36	<	if (int(cfg_params['USER.copy_data']) != 1): raise KeyError
37	<	except KeyError:
38	<	raise CrabException('You can not publish data because you did not selected *** copy_data = 1  *** in the crab.cfg file')
39	<	try:
40	<	self.pset = cfg_params['CMSSW.pset']
41	<	except KeyError:
35	>	if (not cfg_params.has_key('USER.copy_data') or int(cfg_params['USER.copy_data']) != 1 ) or \
36	>	(not cfg_params.has_key('USER.publish_data') or int(cfg_params['USER.publish_data']) != 1 ):
37	>	msg  = 'You can not publish data because you did not selected \n'
38	>	msg += '\t*** copy_data = 1 and publish_data = 1  *** in the crab.cfg file'
39	>	raise CrabException(msg)
40	>
41	>	if not cfg_params.has_key('CMSSW.pset'):
42		raise CrabException('Cannot publish output data, because you did not specify the psetname in [CMSSW] of your crab.cfg file')
43	<	try:
44	<	self.globalDBS=cfg_params['CMSSW.dbs_url']
45	<	except KeyError:
46	<	self.globalDBS="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
47	<	try:
44	<	self.DBSURL=cfg_params['USER.dbs_url_for_publication']
45	<	common.logger.message('<dbs_url_for_publication> = '+self.DBSURL)
46	<	if (self.DBSURL == "http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet") or (self.DBSURL == "https://cmsdbsprod.cern.ch:8443/cms_dbs_prod_global_writer/servlet/DBSServlet"):
47	<	msg = "You can not publish your data in the globalDBS = " + self.DBSURL + "\n"
48	<	msg = msg + "Please write your local one in the [USER] section 'dbs_url_for_publication'"
49	<	raise CrabException(msg)
50	<	except KeyError:
43	>	self.pset = cfg_params['CMSSW.pset']
44	>
45	>	self.globalDBS=cfg_params.get('CMSSW.dbs_url',"http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet")
46	>
47	>	if not cfg_params.has_key('USER.dbs_url_for_publication'):
48		msg = "Warning. The [USER] section does not have 'dbs_url_for_publication'"
49		msg = msg + " entry, necessary to publish the data.\n"
50		msg = msg + "Use the command **crab -publish -USER.dbs_url_for_publication=dbs_url_for_publication*** \nwhere dbs_url_for_publication is your local dbs instance."
51		raise CrabException(msg)
52	+
53	+	self.DBSURL=cfg_params['USER.dbs_url_for_publication']
54	+	common.logger.info('<dbs_url_for_publication> = '+self.DBSURL)
55	+	if (self.DBSURL == "http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet") or (self.DBSURL == "https://cmsdbsprod.cern.ch:8443/cms_dbs_prod_global_writer/servlet/DBSServlet"):
56	+	msg = "You can not publish your data in the globalDBS = " + self.DBSURL + "\n"
57	+	msg = msg + "Please write your local one in the [USER] section 'dbs_url_for_publication'"
58	+	raise CrabException(msg)
59
60		self.content=file(self.pset).read()
61		self.resDir = common.work_space.resDir()
#	Line 70 | Line 74 | class Publisher(Actor):
74		dataset=string.strip(dataset)
75		self.dataset_to_import.append(dataset)
76		###
77	<
77	>
78	>	self.import_all_parents = cfg_params.get('USER.publish_with_import_all_parents',1)
79	>	self.skipOcheck=cfg_params.get('CMSSW.publish_zero_event',0)
80	>
81		self.SEName=''
82		self.CMSSW_VERSION=''
83		self.exit_status=''
#	Line 81 | Line 88 | class Publisher(Actor):
88
89		def importParentDataset(self,globalDBS, datasetpath):
90		"""
91	<	"""
91	>	"""
92		dbsWriter = DBSWriter(self.DBSURL,level='ERROR')
93
94		try:
95	<	dbsWriter.importDatasetWithoutParentage(globalDBS, self.datasetpath, self.DBSURL)
95	>	if (self.import_all_parents==1):
96	>	common.logger.info("--->>> Importing all parents level")
97	>	### to skip the ProdCommon api exception in the case of block without location
98	>	### skipNoSiteError=True
99	>	dbsWriter.importDataset(globalDBS, datasetpath, self.DBSURL, skipNoSiteError=True)
100	>	else:
101	>	common.logger.info("--->>> Importing only the datasetpath " + datasetpath)
102	>	dbsWriter.importDatasetWithoutParentage(globalDBS, datasetpath, self.DBSURL, skipNoSiteError=True)
103		except DBSWriterError, ex:
104		msg = "Error importing dataset to be processed into local DBS\n"
105		msg += "Source Dataset: %s\n" % datasetpath
106		msg += "Source DBS: %s\n" % globalDBS
107		msg += "Destination DBS: %s\n" % self.DBSURL
108	<	common.logger.message(msg)
108	>	common.logger.info(msg)
109	>	common.logger.debug(str(ex))
110		return 1
111		return 0
112	<
112	>	"""
113	>	print " patch for importParentDataset: datasetpath = ", datasetpath
114	>	try:
115	>	args={}
116	>	args['url']=self.DBSURL
117	>	args['mode']='POST'
118	>	block = ""
119	>	api = DbsApi(args)
120	>	#api.migrateDatasetContents(srcURL, dstURL, path, block , False)
121	>	api.migrateDatasetContents(globalDBS, self.DBSURL, datasetpath, block , False)
122	>
123	>	except DbsException, ex:
124	>	print "Caught API Exception %s: %s "  % (ex.getClassName(), ex.getErrorMessage() )
125	>	if ex.getErrorCode() not in (None, ""):
126	>	print "DBS Exception Error Code: ", ex.getErrorCode()
127	>	return 1
128	>	print "Done"
129	>	return 0
130	>	"""
131		def publishDataset(self,file):
132		"""
133		"""
#	Line 104 | Line 137 | class Publisher(Actor):
137		except IndexError:
138		self.exit_status = '1'
139		msg = "Error: Problem with "+file+" file"
140	<	common.logger.message(msg)
140	>	common.logger.info(msg)
141		return self.exit_status
142
143		if (len(self.dataset_to_import) != 0):
144		for dataset in self.dataset_to_import:
145	<	common.logger.message("--->>> Importing parent dataset in the dbs: " +dataset)
145	>	common.logger.info("--->>> Importing parent dataset in the dbs: " +dataset)
146		status_import=self.importParentDataset(self.globalDBS, dataset)
147		if (status_import == 1):
148	<	common.logger.message('Problem with parent '+ dataset +' import from the global DBS '+self.globalDBS+ 'to the local one '+self.DBSURL)
148	>	common.logger.info('Problem with parent '+ dataset +' import from the global DBS '+self.globalDBS+ 'to the local one '+self.DBSURL)
149		self.exit_status='1'
150		return self.exit_status
151		else:
152	<	common.logger.message('Import ok of dataset '+dataset)
152	>	common.logger.info('Import ok of dataset '+dataset)
153
154		#// DBS to contact
155		dbswriter = DBSWriter(self.DBSURL)
#	Line 126 | Line 159 | class Publisher(Actor):
159		except IndexError:
160		self.exit_status = '1'
161		msg = "Error: No file to publish in xml file"+file+" file"
162	<	common.logger.message(msg)
162	>	common.logger.info(msg)
163		return self.exit_status
164
165		datasets=fileinfo.dataset
166	<	common.logger.debug(6,"FileInfo = " + str(fileinfo))
167	<	common.logger.debug(6,"DatasetInfo = " + str(datasets))
166	>	common.logger.log(10-1,"FileInfo = " + str(fileinfo))
167	>	common.logger.log(10-1,"DatasetInfo = " + str(datasets))
168	>	if len(datasets)<=0:
169	>	self.exit_status = '1'
170	>	msg = "Error: No info about dataset in the xml file "+file
171	>	common.logger.info(msg)
172	>	return self.exit_status
173		for dataset in datasets:
174		#### for production data
175	+	self.processedData = dataset['ProcessedDataset']
176		if (dataset['PrimaryDataset'] == 'null'):
177	<	dataset['PrimaryDataset'] = dataset['ProcessedDataset']
178	<
177	>	#dataset['PrimaryDataset'] = dataset['ProcessedDataset']
178	>	dataset['PrimaryDataset'] = self.userprocessedData
179	>	#else: # add parentage from input dataset
180	>	elif self.datasetpath.upper() != 'NONE':
181	>	dataset['ParentDataset']= self.datasetpath
182	>
183		dataset['PSetContent']=self.content
184		cfgMeta = {'name' : self.pset , 'Type' : 'user' , 'annotation': 'user cfg', 'version' : 'private version'} # add real name of user cfg
185	<	common.logger.message("PrimaryDataset = %s"%dataset['PrimaryDataset'])
186	<	common.logger.message("ProcessedDataset = %s"%dataset['ProcessedDataset'])
187	<	common.logger.message("<User Dataset Name> = /"+dataset['PrimaryDataset']+"/"+dataset['ProcessedDataset']+"/USER")
185	>	common.logger.info("PrimaryDataset = %s"%dataset['PrimaryDataset'])
186	>	common.logger.info("ProcessedDataset = %s"%dataset['ProcessedDataset'])
187	>	common.logger.info("<User Dataset Name> = /"+dataset['PrimaryDataset']+"/"+dataset['ProcessedDataset']+"/USER")
188	>	self.dataset_to_check="/"+dataset['PrimaryDataset']+"/"+dataset['ProcessedDataset']+"/USER"
189
190	<	common.logger.debug(6,"--->>> Inserting primary: %s processed : %s"%(dataset['PrimaryDataset'],dataset['ProcessedDataset']))
190	>	common.logger.log(10-1,"--->>> Inserting primary: %s processed : %s"%(dataset['PrimaryDataset'],dataset['ProcessedDataset']))
191
192		primary = DBSWriterObjects.createPrimaryDataset( dataset, dbswriter.dbs)
193	<	common.logger.debug(6,"Primary:  %s "%primary)
193	>	common.logger.log(10-1,"Primary:  %s "%primary)
194
195		algo = DBSWriterObjects.createAlgorithm(dataset, cfgMeta, dbswriter.dbs)
196	<	common.logger.debug(6,"Algo:  %s "%algo)
196	>	common.logger.log(10-1,"Algo:  %s "%algo)
197
198		processed = DBSWriterObjects.createProcessedDataset(primary, algo, dataset, dbswriter.dbs)
199	<	common.logger.debug(6,"Processed:  %s "%processed)
199	>	common.logger.log(10-1,"Processed:  %s "%processed)
200
201	<	common.logger.debug(6,"Inserted primary %s processed %s"%(primary,processed))
201	>	common.logger.log(10-1,"Inserted primary %s processed %s"%(primary,processed))
202
203	<	common.logger.debug(6,"exit_status = %s "%self.exit_status)
203	>	common.logger.log(10-1,"exit_status = %s "%self.exit_status)
204		return self.exit_status
205
206		def publishAJobReport(self,file,procdataset):
#	Line 181 | Line 225 | class Publisher(Actor):
225		elif (file['LFN'] == ''):
226		self.noLFN.append(file['PFN'])
227		else:
228	<	if int(file['TotalEvents']) != 0 :
229	<	file.lumisections = {}
228	>	if  self.skipOcheck==0:
229	>	if int(file['TotalEvents']) != 0:
230	>	#file.lumisections = {}
231	>	# lumi info are now in run hash
232	>	file.runs = {}
233	>	for ds in file.dataset:
234	>	### Fede for production
235	>	if (ds['PrimaryDataset'] == 'null'):
236	>	#ds['PrimaryDataset']=procdataset
237	>	ds['PrimaryDataset']=self.userprocessedData
238	>	filestopublish.append(file)
239	>	else:
240	>	self.noEventsFiles.append(file['LFN'])
241	>	else:
242	>	file.runs = {}
243		for ds in file.dataset:
187	–	ds['ProcessedDataset']=procdataset
244		### Fede for production
245		if (ds['PrimaryDataset'] == 'null'):
246	<	ds['PrimaryDataset']=procdataset
246	>	#ds['PrimaryDataset']=procdataset
247	>	ds['PrimaryDataset']=self.userprocessedData
248		filestopublish.append(file)
249	<	else:
193	<	self.noEventsFiles.append(file['LFN'])
249	>
250		jobReport.files = filestopublish
251		### if all files of FJR have number of events = 0
252		if (len(filestopublish) == 0):
#	Line 202 | Line 258 | class Publisher(Actor):
258		Blocks=None
259		try:
260		Blocks=dbswriter.insertFiles(jobReport)
261	<	common.logger.message("Blocks = %s"%Blocks)
261	>	common.logger.info("Inserting file in blocks = %s"%Blocks)
262		except DBSWriterError, ex:
263	<	common.logger.message("Insert file error: %s"%ex)
263	>	common.logger.info("Insert file error: %s"%ex)
264		return Blocks
265
266		def run(self):
#	Line 213 | Line 269 | class Publisher(Actor):
269		"""
270
271		file_list = glob.glob(self.resDir+"crab_fjr*.xml")
272	<	common.logger.debug(6, "file_list = "+str(file_list))
273	<	common.logger.debug(6, "len(file_list) = "+str(len(file_list)))
272	>	## Select only those fjr that are succesfull
273	>	if (len(file_list)==0):
274	>	common.logger.info("--->>> "+self.resDir+" empty: no file to publish on DBS")
275	>	self.exit_status = '1'
276	>	return self.exit_status
277	>
278	>	good_list=[]
279	>	for fjr in file_list:
280	>	reports = readJobReport(fjr)
281	>	if len(reports)>0:
282	>	if reports[0].status == "Success":
283	>	good_list.append(fjr)
284	>	file_list=good_list
285	>	##
286	>	common.logger.log(10-1, "file_list = "+str(file_list))
287	>	common.logger.log(10-1, "len(file_list) = "+str(len(file_list)))
288
289		if (len(file_list)>0):
290		BlocksList=[]
291	<	common.logger.message("--->>> Start dataset publication")
291	>	common.logger.info("--->>> Start dataset publication")
292		self.exit_status=self.publishDataset(file_list[0])
293		if (self.exit_status == '1'):
294		return self.exit_status
295	<	common.logger.message("--->>> End dataset publication")
295	>	common.logger.info("--->>> End dataset publication")
296
297
298	<	common.logger.message("--->>> Start files publication")
298	>	common.logger.info("--->>> Start files publication")
299		for file in file_list:
300	<	common.logger.message("file = "+file)
300	>	common.logger.debug( "file = "+file)
301		Blocks=self.publishAJobReport(file,self.processedData)
302		if Blocks:
303	<	[BlocksList.append(x) for x in Blocks]
303	>	for x in Blocks: # do not allow multiple entries of the same block
304	>	if x not in BlocksList:
305	>	BlocksList.append(x)
306
307		# close the blocks
308	<	common.logger.debug(6, "BlocksList = %s"%BlocksList)
308	>	common.logger.log(10-1, "BlocksList = %s"%BlocksList)
309		# dbswriter = DBSWriter(self.DBSURL,level='ERROR')
310		dbswriter = DBSWriter(self.DBSURL)
311
312		for BlockName in BlocksList:
313		try:
314		closeBlock=dbswriter.manageFileBlock(BlockName,maxFiles= 1)
315	<	common.logger.debug(6, "closeBlock %s"%closeBlock)
315	>	common.logger.log(10-1, "closeBlock %s"%closeBlock)
316		#dbswriter.dbs.closeBlock(BlockName)
317		except DBSWriterError, ex:
318	<	common.logger.message("Close block error %s"%ex)
318	>	common.logger.info("Close block error %s"%ex)
319
320		if (len(self.noEventsFiles)>0):
321	<	common.logger.message("--->>> WARNING: "+str(len(self.noEventsFiles))+" files not published because they contain 0 events are:")
321	>	common.logger.info("--->>> WARNING: "+str(len(self.noEventsFiles))+" files not published because they contain 0 events are:")
322		for lfn in self.noEventsFiles:
323	<	common.logger.message("------ LFN: %s"%lfn)
323	>	common.logger.info("------ LFN: %s"%lfn)
324		if (len(self.noLFN)>0):
325	<	common.logger.message("--->>> WARNING: there are "+str(len(self.noLFN))+" files not published because they have empty LFN")
325	>	common.logger.info("--->>> WARNING: there are "+str(len(self.noLFN))+" files not published because they have empty LFN")
326		for pfn in self.noLFN:
327	<	common.logger.message("------ pfn: %s"%pfn)
327	>	common.logger.info("------ pfn: %s"%pfn)
328		if (len(self.problemFiles)>0):
329	<	common.logger.message("--->>> WARNING: "+str(len(self.problemFiles))+" files not published because they had problem with copy to SE")
329	>	common.logger.info("--->>> WARNING: "+str(len(self.problemFiles))+" files not published because they had problem with copy to SE")
330		for lfn in self.problemFiles:
331	<	common.logger.message("------ LFN: %s"%lfn)
332	<	common.logger.message("--->>> End files publication")
333	<	common.logger.message("--->>> To check data publication please use: InspectDBS2.py --DBSURL=<dbs_url_for_publication> --datasetPath=<User Dataset Name>")
331	>	common.logger.info("------ LFN: %s"%lfn)
332	>	common.logger.info("--->>> End files publication")
333	>
334	>	self.cfg_params['USER.dataset_to_check']=self.dataset_to_check
335	>	from InspectDBS import InspectDBS
336	>	check=InspectDBS(self.cfg_params)
337	>	check.checkPublication()
338		return self.exit_status
339
340		else:
341	<	common.logger.message("--->>> "+self.resDir+" empty: no file to publish on DBS")
341	>	common.logger.info("--->>> No valid files to publish on DBS. Your jobs do not report exit codes = 0")
342		self.exit_status = '1'
343		return self.exit_status
344

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines