| 1 |
#!/usr/bin/env python
|
| 2 |
|
| 3 |
import sys, os, getopt
|
| 4 |
|
| 5 |
import dbsCgiApi
|
| 6 |
import dbsApi
|
| 7 |
|
| 8 |
import dlsApi
|
| 9 |
import dlsClient
|
| 10 |
from dlsDataObjects import DlsLocation, DlsFileBlock, DlsEntry
|
| 11 |
import xml.dom.ext.reader
|
| 12 |
|
| 13 |
|
| 14 |
def main(argv) :
|
| 15 |
"""
|
| 16 |
|
| 17 |
DBSDLSQuery
|
| 18 |
|
| 19 |
query DBS and DLS to discovery datasets
|
| 20 |
|
| 21 |
required parameters
|
| 22 |
--pattern <pattern> : pattern to restrict discovery
|
| 23 |
|
| 24 |
optional parameters :
|
| 25 |
--help (-h) : help
|
| 26 |
--debug (-d) : debug statements
|
| 27 |
|
| 28 |
"""
|
| 29 |
|
| 30 |
# default
|
| 31 |
pattern = ''
|
| 32 |
debug = 0
|
| 33 |
|
| 34 |
try:
|
| 35 |
opts, args = getopt.getopt(argv, "", ["help", "debug", "pattern="])
|
| 36 |
except getopt.GetoptError:
|
| 37 |
print main.__doc__
|
| 38 |
sys.exit(2)
|
| 39 |
|
| 40 |
# check command line parameter
|
| 41 |
for opt, arg in opts :
|
| 42 |
if opt == "--help" :
|
| 43 |
print main.__doc__
|
| 44 |
sys.exit()
|
| 45 |
elif opt == "--debug" :
|
| 46 |
debug = 1
|
| 47 |
elif opt == "--pattern" :
|
| 48 |
pattern = arg
|
| 49 |
|
| 50 |
# guarantee that pattern has "*" at beginning and end
|
| 51 |
if pattern != '' :
|
| 52 |
if pattern[0] != '*' :
|
| 53 |
pattern = '*'+pattern
|
| 54 |
if pattern[-1] != '*' :
|
| 55 |
pattern = pattern+'*'
|
| 56 |
else :
|
| 57 |
pattern = '*'
|
| 58 |
|
| 59 |
# DBS initialization
|
| 60 |
dbs_url="http://cmsdoc.cern.ch/cms/test/aprom/DBS/CGIServer/prodquery"
|
| 61 |
dbs_instance = {'instance' : 'MCGlobal/Writer'}
|
| 62 |
dbs = dbsCgiApi.DbsCgiApi(dbs_url, dbs_instance)
|
| 63 |
|
| 64 |
# DLS initialization
|
| 65 |
type = "DLS_TYPE_DLI"
|
| 66 |
endpoint="prod-lfc-cms-central.cern.ch/grid/cms/DLS/LFC"
|
| 67 |
dls = dlsClient.getDlsApi(dls_type=type,dls_endpoint=endpoint)
|
| 68 |
|
| 69 |
# DBS query for datasets
|
| 70 |
try:
|
| 71 |
dbsDatasets = dbs.listProcessedDatasets(pattern)
|
| 72 |
except dbsApi.InvalidDataTier, ex:
|
| 73 |
raise DBSInvalidDataTierError(ex.getClassName(),ex.getErrorMessage())
|
| 74 |
except dbsApi.DbsApiException, ex:
|
| 75 |
raise DBSError(ex.getClassName(),ex.getErrorMessage())
|
| 76 |
except dbsCgiApi.DbsCgiToolError , ex:
|
| 77 |
raise DBSError(ex.getClassName(),ex.getErrorMessage())
|
| 78 |
except dbsCgiApi.DbsCgiBadResponse , ex:
|
| 79 |
raise DBSError(ex.getClassName(),ex.getErrorMessage())
|
| 80 |
|
| 81 |
for dbsDataset in dbsDatasets:
|
| 82 |
dataset = dbsDataset.get('datasetPathName')
|
| 83 |
|
| 84 |
# DBS query for fileblocks
|
| 85 |
eventsPerBlock = {}
|
| 86 |
try:
|
| 87 |
fileBlocks = dbs.getDatasetContents(dataset)
|
| 88 |
except dbsApi.DbsApiException, ex:
|
| 89 |
raise DBSError(ex.getClassName(),ex.getErrorMessage())
|
| 90 |
except dbsCgiApi.DbsCgiBadResponse, ex:
|
| 91 |
raise DBSError(ex.getClassName(),ex.getErrorMessage())
|
| 92 |
for fileBlock in fileBlocks:
|
| 93 |
## get the event collections for each block
|
| 94 |
nevts = 0
|
| 95 |
eventCollectionList = fileBlock.get('eventCollectionList')
|
| 96 |
for eventCollection in eventCollectionList:
|
| 97 |
nevts = nevts + eventCollection.get('numberOfEvents')
|
| 98 |
|
| 99 |
eventsPerBlock[fileBlock.get('blockName')]=nevts
|
| 100 |
if debug :
|
| 101 |
print 'DBS query result in block:',fileBlock.get('blockName'),'with',nevts,'events'
|
| 102 |
|
| 103 |
siteDictionary = {}
|
| 104 |
for fileBlock in eventsPerBlock.keys() :
|
| 105 |
if eventsPerBlock[fileBlock] > 0 :
|
| 106 |
# query DLS
|
| 107 |
try:
|
| 108 |
locationList = dls.getLocations([fileBlock])
|
| 109 |
except dlsApi.DlsApiError, inst:
|
| 110 |
# create empty list
|
| 111 |
locationList = {}
|
| 112 |
msg = "Error in the DLS query: %s." % str(inst)
|
| 113 |
if debug :
|
| 114 |
print msg
|
| 115 |
pass
|
| 116 |
|
| 117 |
for locationItem in locationList:
|
| 118 |
for location in locationItem.locations:
|
| 119 |
host = str(location.host)
|
| 120 |
if host in siteDictionary.keys() :
|
| 121 |
siteDictionary[host].append(fileBlock)
|
| 122 |
else :
|
| 123 |
siteDictionary[host] = [fileBlock]
|
| 124 |
|
| 125 |
for site in siteDictionary.keys() :
|
| 126 |
events = 0
|
| 127 |
for block in siteDictionary[site] :
|
| 128 |
events += eventsPerBlock[block]
|
| 129 |
print dataset,site,events
|
| 130 |
|
| 131 |
if __name__ == '__main__' :
|
| 132 |
main(sys.argv[1:])
|
