| 1 |
#!/usr/bin/env python
|
| 2 |
|
| 3 |
import sys, os, getopt
|
| 4 |
|
| 5 |
#import dbsCgiApi, dbsApi
|
| 6 |
import DBSAPI.dbsApi
|
| 7 |
|
| 8 |
import dlsApi
|
| 9 |
import dlsClient
|
| 10 |
from dlsDataObjects import DlsLocation, DlsFileBlock, DlsEntry
|
| 11 |
import xml.dom.ext.reader
|
| 12 |
|
| 13 |
import exceptions
|
| 14 |
from DBSAPI.dbsApiException import *
|
| 15 |
import common
|
| 16 |
|
| 17 |
|
| 18 |
# #######################################
|
| 19 |
class DBSError_DBS2(exceptions.Exception):
|
| 20 |
def __init__(self, errorName, errorMessage):
|
| 21 |
args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
|
| 22 |
exceptions.Exception.__init__(self, args)
|
| 23 |
pass
|
| 24 |
|
| 25 |
def getErrorMessage(self):
|
| 26 |
""" Return error message """
|
| 27 |
return "%s" % (self.args)
|
| 28 |
|
| 29 |
# ####################################
|
| 30 |
class DataDiscoveryError_DBS2(exceptions.Exception):
|
| 31 |
def __init__(self, errorMessage):
|
| 32 |
self.args=errorMessage
|
| 33 |
exceptions.Exception.__init__(self, self.args)
|
| 34 |
pass
|
| 35 |
|
| 36 |
def getErrorMessage(self):
|
| 37 |
""" Return exception error """
|
| 38 |
return "%s" % (self.args)
|
| 39 |
|
| 40 |
# ####################################
|
| 41 |
class NotExistingDatasetError_DBS2(exceptions.Exception):
|
| 42 |
def __init__(self, errorMessage):
|
| 43 |
self.args=errorMessage
|
| 44 |
exceptions.Exception.__init__(self, self.args)
|
| 45 |
pass
|
| 46 |
|
| 47 |
def getErrorMessage(self):
|
| 48 |
""" Return exception error """
|
| 49 |
return "%s" % (self.args)
|
| 50 |
|
| 51 |
# #######################################
|
| 52 |
def main(argv) :
|
| 53 |
"""
|
| 54 |
|
| 55 |
DBSDLSQuery
|
| 56 |
|
| 57 |
query DBS and DLS to discovery datasets
|
| 58 |
|
| 59 |
required parameters
|
| 60 |
--pattern <pattern> : pattern to restrict discovery
|
| 61 |
|
| 62 |
optional parameters :
|
| 63 |
--help (-h) : help
|
| 64 |
--debug (-d) : debug statements
|
| 65 |
|
| 66 |
"""
|
| 67 |
|
| 68 |
# default
|
| 69 |
pattern = ''
|
| 70 |
debug = 0
|
| 71 |
|
| 72 |
try:
|
| 73 |
opts, args = getopt.getopt(argv, "", ["help", "debug", "pattern="])
|
| 74 |
except getopt.GetoptError:
|
| 75 |
print main.__doc__
|
| 76 |
sys.exit(2)
|
| 77 |
|
| 78 |
# check command line parameter
|
| 79 |
for opt, arg in opts :
|
| 80 |
if opt == "--help" :
|
| 81 |
print main.__doc__
|
| 82 |
sys.exit()
|
| 83 |
elif opt == "--debug" :
|
| 84 |
debug = 1
|
| 85 |
elif opt == "--pattern" :
|
| 86 |
pattern = arg
|
| 87 |
|
| 88 |
# guarantee that pattern has "*" at beginning and end
|
| 89 |
if pattern != '' :
|
| 90 |
if pattern[0] != '*' :
|
| 91 |
pattern = '*'+pattern
|
| 92 |
if pattern[-1] != '*' :
|
| 93 |
pattern = pattern+'*'
|
| 94 |
else :
|
| 95 |
pattern = '*'
|
| 96 |
|
| 97 |
|
| 98 |
# DBS initialization
|
| 99 |
dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
|
| 100 |
dbs_version="v00_00_06"
|
| 101 |
args = {}
|
| 102 |
args['url'] = dbs_url
|
| 103 |
args['version'] = dbs_version
|
| 104 |
args['level'] = 'CRITICAL'
|
| 105 |
|
| 106 |
dbs = DBSAPI.dbsApi.DbsApi(args)
|
| 107 |
|
| 108 |
# DLS initialization
|
| 109 |
type = "DLS_TYPE_DBS"
|
| 110 |
endpoint= "http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
|
| 111 |
dls = dlsClient.getDlsApi(dls_type=type,dls_endpoint=endpoint)
|
| 112 |
|
| 113 |
|
| 114 |
# DBS query for datasets
|
| 115 |
dbsDatasets = dbs.listProcessedDatasets(pattern)
|
| 116 |
|
| 117 |
for dbsDataset in dbsDatasets:
|
| 118 |
|
| 119 |
dataset = dbsDataset.get('PathList')[0]
|
| 120 |
|
| 121 |
# DBS query for fileblocks
|
| 122 |
eventsPerBlock = {}
|
| 123 |
|
| 124 |
try:
|
| 125 |
files = dbs.listFiles(dataset)
|
| 126 |
except DbsBadRequest, msg:
|
| 127 |
raise DataDiscoveryError_DBS2(msg)
|
| 128 |
except DBSError_DBS2, msg:
|
| 129 |
raise DataDiscoveryError_DBS2(msg)
|
| 130 |
|
| 131 |
|
| 132 |
for file in files :
|
| 133 |
|
| 134 |
filename = file['LogicalFileName']
|
| 135 |
events = file['NumberOfEvents']
|
| 136 |
fileblock = file['Block']['Name']
|
| 137 |
|
| 138 |
# number of events per block
|
| 139 |
if fileblock in eventsPerBlock.keys() :
|
| 140 |
eventsPerBlock[fileblock] += events
|
| 141 |
else :
|
| 142 |
eventsPerBlock[fileblock] = events
|
| 143 |
|
| 144 |
|
| 145 |
if len(eventsPerBlock) <= 0:
|
| 146 |
raise NotExistingDatasetError_DBS2 (("\nNo data for %s in DBS\nPlease check" + " dataset path variables in crab.cfg") % self.datasetPath)
|
| 147 |
|
| 148 |
siteDictionary = {}
|
| 149 |
for fileBlock in eventsPerBlock.keys() :
|
| 150 |
if eventsPerBlock[fileBlock] > 0 :
|
| 151 |
# query DLS
|
| 152 |
try:
|
| 153 |
locationList = dls.getLocations([fileBlock])
|
| 154 |
except dlsApi.DlsApiError, inst:
|
| 155 |
# create empty list
|
| 156 |
locationList = {}
|
| 157 |
msg = "Error in the DLS query: %s." % str(inst)
|
| 158 |
if debug :
|
| 159 |
print msg
|
| 160 |
pass
|
| 161 |
|
| 162 |
for locationItem in locationList:
|
| 163 |
for location in locationItem.locations:
|
| 164 |
host = str(location.host)
|
| 165 |
if host in siteDictionary.keys() :
|
| 166 |
siteDictionary[host].append(fileBlock)
|
| 167 |
else :
|
| 168 |
siteDictionary[host] = [fileBlock]
|
| 169 |
|
| 170 |
for site in siteDictionary.keys() :
|
| 171 |
events = 0
|
| 172 |
for block in siteDictionary[site] :
|
| 173 |
events += eventsPerBlock[block]
|
| 174 |
print dataset,site,events
|
| 175 |
|
| 176 |
if __name__ == '__main__' :
|
| 177 |
main(sys.argv[1:])
|
