| 1 |
eulisse |
1.1 |
import os
|
| 2 |
|
|
import sys
|
| 3 |
|
|
import getopt
|
| 4 |
|
|
import popen2
|
| 5 |
|
|
import time
|
| 6 |
|
|
|
| 7 |
eulisse |
1.2 |
from ProdCommon.MCPayloads.WorkflowSpec import WorkflowSpec
|
| 8 |
|
|
from ProdCommon.MCPayloads.LFNAlgorithm import unmergedLFNBase, mergedLFNBase
|
| 9 |
|
|
from ProdCommon.CMSConfigTools.CfgInterface import CfgInterface
|
| 10 |
|
|
from ProdCommon.MCPayloads.DatasetExpander import splitMultiTier
|
| 11 |
|
|
import ProdCommon.MCPayloads.WorkflowTools as WorkflowTools
|
| 12 |
|
|
import ProdCommon.MCPayloads.UUID as MCPayloadsUUID
|
| 13 |
|
|
from ProdCommon.CMSConfigTools.CfgInterface import CfgInterface
|
| 14 |
|
|
|
| 15 |
|
|
def checkPSetType (pycfg):
|
| 16 |
|
|
cfgInt = CfgInterface (pycfg, True)
|
| 17 |
|
|
for outModName, val in cfgInt.outputModules.items():
|
| 18 |
|
|
datasets = val.datasets()
|
| 19 |
|
|
for outDataset in datasets:
|
| 20 |
|
|
if outDataset.has_key ('dataTier'):
|
| 21 |
|
|
return "NEW"
|
| 22 |
|
|
else:
|
| 23 |
|
|
return "OLD"
|
| 24 |
|
|
return "OLD"
|
| 25 |
eulisse |
1.1 |
|
| 26 |
|
|
# Fixme: do the same for createProdWorkflow
|
| 27 |
eulisse |
1.2 |
def createPreProdWorkflow (PyCfg = None,
|
| 28 |
|
|
PSetHash = None,
|
| 29 |
eulisse |
1.1 |
prodName = None,
|
| 30 |
eulisse |
1.2 |
cmsswVersion = None,
|
| 31 |
eulisse |
1.1 |
category = "PreProd",
|
| 32 |
|
|
timestamp = int(time.time()),
|
| 33 |
|
|
fakeHash = False,
|
| 34 |
|
|
pileupDS = None,
|
| 35 |
|
|
pileupFilesPerJob = 1,
|
| 36 |
|
|
dbsAddress = None,
|
| 37 |
|
|
dbsUrl = None,
|
| 38 |
|
|
dlsAddress = None,
|
| 39 |
|
|
dlsType = None,
|
| 40 |
|
|
pileupSkipLocation = False):
|
| 41 |
|
|
spec = WorkflowSpec()
|
| 42 |
|
|
spec.setWorkflowName(prodName)
|
| 43 |
|
|
spec.setRequestCategory(category)
|
| 44 |
|
|
spec.setRequestTimestamp(timestamp)
|
| 45 |
|
|
|
| 46 |
|
|
# //
|
| 47 |
|
|
# // This value was created by running the EdmConfigHash tool
|
| 48 |
|
|
# // on the original cfg file.
|
| 49 |
|
|
|
| 50 |
|
|
|
| 51 |
|
|
cmsRun = spec.payload
|
| 52 |
|
|
cmsRun.name = "cmsRun1" # every node in the workflow needs a unique name
|
| 53 |
|
|
cmsRun.type = "CMSSW" # Nodes that are CMSSW based should set the name
|
| 54 |
|
|
cmsRun.application["Project"] = "CMSSW" # project
|
| 55 |
eulisse |
1.2 |
cmsRun.application["Version"] = cmsswVersion # version
|
| 56 |
eulisse |
1.1 |
cmsRun.application["Architecture"] = "slc3_ia32_gcc323" # arch (not needed)
|
| 57 |
|
|
cmsRun.application["Executable"] = "cmsRun" # binary name
|
| 58 |
eulisse |
1.2 |
cmsRun.configuration = PyCfg # Python PSet file
|
| 59 |
eulisse |
1.1 |
|
| 60 |
|
|
|
| 61 |
|
|
# //
|
| 62 |
|
|
# // Pileup sample?
|
| 63 |
|
|
# //
|
| 64 |
|
|
if pileupDS != None:
|
| 65 |
|
|
puPrimary = pileupDS.split("/")[1]
|
| 66 |
|
|
puTier = pileupDS.split("/")[2]
|
| 67 |
|
|
puProc = pileupDS.split("/")[3]
|
| 68 |
|
|
puDataset = cmsRun.addPileupDataset(puPrimary, puTier, puProc)
|
| 69 |
|
|
puDataset['FilesPerJob'] = pileupFilesPerJob
|
| 70 |
|
|
if customDBSDLS:
|
| 71 |
|
|
puDataset['DBSAddress'] = dbsAddress
|
| 72 |
|
|
puDataset['DBSURL'] = dbsUrl
|
| 73 |
|
|
puDataset['DLSType'] = dlsType
|
| 74 |
|
|
puDataset['DLSAddress'] = dlsAddress
|
| 75 |
|
|
if pileupSkipLocation:
|
| 76 |
|
|
puDataset['SkipLocation'] = pileupSkipLocation
|
| 77 |
|
|
|
| 78 |
|
|
# //
|
| 79 |
|
|
# // Pull all the output modules from the configuration file,
|
| 80 |
|
|
#// treat the output module name as DataTier and AppFamily,
|
| 81 |
|
|
# //since there isnt a standard way to derive these things yet.
|
| 82 |
|
|
# //
|
| 83 |
|
|
#// For each module a dataset declaration is created in the spec
|
| 84 |
|
|
cfgInt = CfgInterface(cmsRun.configuration, True)
|
| 85 |
|
|
datasetList = []
|
| 86 |
|
|
for outModName, val in cfgInt.outputModules.items():
|
| 87 |
|
|
# //
|
| 88 |
|
|
# // Check for Multi Tiers.
|
| 89 |
|
|
#// If Output module contains - characters, we split based on it
|
| 90 |
|
|
# //And create a different tier for each basic tier
|
| 91 |
|
|
# //
|
| 92 |
|
|
#//
|
| 93 |
|
|
|
| 94 |
|
|
tierList = splitMultiTier(outModName)
|
| 95 |
|
|
for dataTier in tierList:
|
| 96 |
|
|
processedDS = "%s-%s-%s-unmerged" % (
|
| 97 |
|
|
cmsRun.application['Version'], outModName, timestamp)
|
| 98 |
|
|
outDS = cmsRun.addOutputDataset(prodName,
|
| 99 |
|
|
processedDS,
|
| 100 |
|
|
outModName)
|
| 101 |
|
|
|
| 102 |
|
|
outDS['DataTier'] = dataTier
|
| 103 |
|
|
outDS["ApplicationName"] = cmsRun.application["Executable"]
|
| 104 |
|
|
outDS["ApplicationProject"] = cmsRun.application["Project"]
|
| 105 |
|
|
outDS["ApplicationVersion"] = cmsRun.application["Version"]
|
| 106 |
|
|
outDS["ApplicationFamily"] = outModName
|
| 107 |
|
|
if fakeHash:
|
| 108 |
|
|
guid = MCPayloadsUUID.uuidgen()
|
| 109 |
|
|
if guid == None:
|
| 110 |
|
|
guid = MCPayloadsUUID.uuid()
|
| 111 |
eulisse |
1.2 |
hashValue = "hash=%s;guid=%s" % (PSetHash, guid)
|
| 112 |
eulisse |
1.1 |
outDS['PSetHash'] = hashValue
|
| 113 |
|
|
else:
|
| 114 |
eulisse |
1.2 |
outDS['PSetHash'] = PSetHash
|
| 115 |
eulisse |
1.1 |
datasetList.append(outDS.name())
|
| 116 |
|
|
|
| 117 |
|
|
stageOut = cmsRun.newNode("stageOut1")
|
| 118 |
|
|
stageOut.type = "StageOut"
|
| 119 |
|
|
stageOut.application["Project"] = ""
|
| 120 |
|
|
stageOut.application["Version"] = ""
|
| 121 |
|
|
stageOut.application["Architecture"] = ""
|
| 122 |
|
|
stageOut.application["Executable"] = "RuntimeStageOut.py" # binary name
|
| 123 |
|
|
stageOut.configuration = ""
|
| 124 |
|
|
|
| 125 |
|
|
mergedLFNBase(spec)
|
| 126 |
|
|
unmergedLFNBase(spec)
|
| 127 |
|
|
|
| 128 |
eulisse |
1.2 |
return spec
|
