Merge branch 'master' into UpdatingForHI2024

dmwm · Oct 9, 2024 · a4bd257 · a4bd257
2 parents 0345e76 + 178a4c4
commit a4bd257
Show file tree

Hide file tree

Showing 18 changed files with 276 additions and 35 deletions.
diff --git a/bin/00_pypi_deploy_prod.sh b/bin/00_pypi_deploy_prod.sh
@@ -30,8 +30,8 @@ then
     exit
 fi
 
-WMAGENT_TAG=2.3.4
-TIER0_VERSION=3.2.3
+WMAGENT_TAG=2.3.5
+TIER0_VERSION=3.2.5
 COUCH_TAG=3.2.2
 
 BASE_DIR=/data/tier0
@@ -159,12 +159,11 @@ sed -i "s+config.RucioInjector.metaDIDProject.*+config.RucioInjector.metaDIDProj
 echo "config.RucioInjector.blockDeletionDelayHours = 168" >> $config/config.py
 #####
 ##### NOT IN REPLAY #####
-echo "config.RucioInjector.blockDeletionDelayHours = 168" >> $config/config.py
 echo 'config.BossAir.pluginNames = ["SimpleCondorPlugin"]' >> $config/config.py
 echo 'config.JobAccountant.maxAllowedRepackOutputSize = 24 * 1024 * 1024 * 1024' >> $config/config.py
 echo "config.AgentStatusWatcher.runningExpressPercent = 25" >> $config/config.py
 echo "config.AgentStatusWatcher.runningRepackPercent = 10" >> $config/config.py
-echo 'config.TaskArchiver.archiveDelayHours = 2190' >> $config.py
+echo 'config.TaskArchiver.archiveDelayHours = 720' >> $config.py
 ##### NOT IN REPLAY #####
 #####
 echo "config.DBS3Upload.datasetType = 'VALID'" >> $config/config.py

diff --git a/bin/00_pypi_deploy_replay.sh b/bin/00_pypi_deploy_replay.sh
@@ -15,8 +15,8 @@ then
     exit
 fi
 
-WMAGENT_TAG=2.3.4rc11
-TIER0_VERSION=3.2.4rc1
+WMAGENT_TAG=2.3.5
+TIER0_VERSION=3.2.5
 COUCH_TAG=3.2.2
 
 BASE_DIR=/data/tier0
@@ -150,7 +150,7 @@ echo "config.RucioInjector.blockDeletionDelayHours = 2" >> $config/config.py
 #echo 'config.JobAccountant.maxAllowedRepackOutputSize = 24 * 1024 * 1024 * 1024' >> ./config/tier0/config.py
 #echo "config.AgentStatusWatcher.runningExpressPercent = 25" >> ./config/tier0/config.py
 #echo "config.AgentStatusWatcher.runningRepackPercent = 10" >> ./config/tier0/config.py
-#echo 'config.TaskArchiver.archiveDelayHours = 2190' >> $config.py
+#echo 'config.TaskArchiver.archiveDelayHours = 720' >> $config.py
 ##### ONLY IN PROD #####
 #####
 echo "config.DBS3Upload.datasetType = 'VALID'" >> $config/config.py

diff --git a/bin/00_pypi_patches.sh b/bin/00_pypi_patches.sh
@@ -11,8 +11,8 @@ DEPLOY_DIR=$BASE_DIR/WMAgent.venv3
 ##################
 
 #Dummy demonstrative Patch for new alma 9 agent
-#curl https://patch-diff.githubusercontent.com/raw/dmwm/T0/pull/4961.patch | patch -d $WMA_DEPLOY_DIR/lib/python3.9/site-packages/ -p 3
-curl https://patch-diff.githubusercontent.com/raw/germanfgv/WMCore/pull/16.patch | patch -d $WMA_DEPLOY_DIR/lib/python3.9/site-packages/ -p 3
+#curl https://patch-diff.githubusercontent.com/raw/dmwm/T0/pull/4961.patch | patch -f -d $WMA_DEPLOY_DIR/lib/python3.9/site-packages/ -p 3
+curl https://patch-diff.githubusercontent.com/raw/germanfgv/WMCore/pull/16.patch | patch -f -d $WMA_DEPLOY_DIR/lib/python3.9/site-packages/ -p 3
 
 
 ######################

diff --git a/bin/00_pypi_tweak_prod_config.sh b/bin/00_pypi_tweak_prod_config.sh
@@ -8,7 +8,6 @@ sed -i "s+'team1,team2,cmsdataops'+'tier0production'+g" "$config/config.py"
 sed -i "s+config.RucioInjector.containerDiskRuleParams.*+config.RucioInjector.containerDiskRuleParams = {}+" "$config/config.py"
 echo "config.RucioInjector.blockRuleParams = {}" >> $config/config.py
 sed -i "s+config.RucioInjector.metaDIDProject.*+config.RucioInjector.metaDIDProject = 'Tier0'+" "$config/config.py"
-echo "config.RucioInjector.blockDeletionDelayHours = 168" >> $config/config.py
 ##### NOT IN REPLAY #####
 echo "config.RucioInjector.blockDeletionDelayHours = 168" >> $config/config.py
 echo 'config.BossAir.pluginNames = ["SimpleCondorPlugin"]' >> $config/config.py
@@ -37,4 +36,4 @@ if [ "x$DROPBOX_USER" == "x" ] || [ "x$DROPBOX_PASS" == "x" ]; then
 fi
 
 echo 'config.Tier0Feeder.dropboxuser = "'$DROPBOX_USER'"' >> $config/config.py
-echo 'config.Tier0Feeder.dropboxpass = "'$DROPBOX_PASS'"' >> $config/config.py
+echo 'config.Tier0Feeder.dropboxpass = "'$DROPBOX_PASS'"' >> $config/config.py
diff --git a/etc/HIProdOfflineConfiguration.py b/etc/HIProdOfflineConfiguration.py
@@ -25,6 +25,7 @@
 from T0.RunConfig.Tier0Config import setStreamerPNN
 from T0.RunConfig.Tier0Config import addSiteConfig
 from T0.RunConfig.Tier0Config import setStorageSite
+from T0.RunConfig.Tier0Config import setHelperAgentStreams
 
 # Create the Tier0 configuration object
 tier0Config = createTier0Config()
@@ -38,6 +39,11 @@
 # Set the max run number:
 setInjectMaxRun(tier0Config, 9999999)
 
+# Define streams to ignore. These wont be injected by the MainAgent
+setHelperAgentStreams(tier0Config, {'SecondAgent': [],
+                                    'ThirdAgent' : []
+                                    })
+
 # Settings up sites
 processingSite = "T2_CH_CERN"
 storageSite = "T0_CH_CERN_Disk"

diff --git a/etc/HIReplayOfflineConfiguration.py b/etc/HIReplayOfflineConfiguration.py
@@ -25,6 +25,7 @@
 from T0.RunConfig.Tier0Config import setEnableUniqueWorkflowName
 from T0.RunConfig.Tier0Config import addSiteConfig
 from T0.RunConfig.Tier0Config import setStorageSite
+from T0.RunConfig.Tier0Config import setHelperAgentStreams
 
 # Create the Tier0 configuration object
 tier0Config = createTier0Config()
@@ -41,6 +42,10 @@
 # 361694:361699,361779 - 2022 HI dry-run test runs
 setInjectRuns(tier0Config, [375820])
 
+# Define streams to ignore. These wont be injected by the MainAgent
+setHelperAgentStreams(tier0Config, {'SecondAgent': [],
+                                    'ThirdAgent' : []
+                                    })
 # Settings up sites
 processingSite = "T2_CH_CERN"
 storageSite = "T0_CH_CERN_Disk"

diff --git a/etc/ProdOfflineConfiguration.py b/etc/ProdOfflineConfiguration.py
@@ -28,7 +28,7 @@
 from T0.RunConfig.Tier0Config import addSiteConfig
 from T0.RunConfig.Tier0Config import setStorageSite
 from T0.RunConfig.Tier0Config import setExtraStreamDatasetMap
-
+from T0.RunConfig.Tier0Config import setHelperAgentStreams
 
 # Create the Tier0 configuration object
 tier0Config = createTier0Config()
@@ -42,6 +42,11 @@
 # Set the max run number:
 setInjectMaxRun(tier0Config, 9999999)
 
+# Set streams to ignore by agent. These will not be injected by the MainAgent
+setHelperAgentStreams(tier0Config, {"SecondAgent" : [],
+                                      "ThirdAgent" : []})
+
+
 # Settings up sites
 processingSite = "T2_CH_CERN"
 storageSite = "T0_CH_CERN_Disk"
@@ -60,7 +65,7 @@
 #  Data type
 #  Processing site (where jobs run)
 #  PhEDEx locations
-setAcquisitionEra(tier0Config, "Run2024G")
+setAcquisitionEra(tier0Config, "Run2024I")
 setEmulationAcquisitionEra(tier0Config, "Emulation2024")
 setBaseRequestPriority(tier0Config, 251000)
 setBackfill(tier0Config, None)
@@ -101,7 +106,7 @@
 # maxRunPreviousConfig = 999999 # Last run before era change 08/09/23
 # Defaults for CMSSW version
 defaultCMSSWVersion = {
-    'default': "CMSSW_14_0_14",
+    'default': "CMSSW_14_0_16",
     #'acqEra': {'Run2024F': "CMSSW_14_0_11"},
     #'maxRun': {maxRunPreviousConfig: "CMSSW_13_2_2"}
 }
@@ -123,6 +128,7 @@
 alcaPPSScenario = "AlCaPPS_Run3"
 hiTestppScenario = "ppEra_Run3_pp_on_PbPb_2023"
 hiRawPrimeScenario = "ppEra_Run3_pp_on_PbPb_approxSiStripClusters_2023"
+hltScoutingScenario = "hltScoutingEra_Run3_2024"
 
 # Defaults for processing version
 alcarawProcVersion = {
@@ -1109,13 +1115,13 @@
                dqm_sequences=["@common", "@ecal", "@jetmet", "@L1TMon", "@hcal", "@L1TEgamma"],
                alca_producers=["TkAlMinBias","LumiPixelsMinBias"],
                physics_skims=["LogError", "LogErrorMonitor"],
-	       disk_node="T2_CH_CERN",
+               disk_node="T2_CH_CERN",
                scenario=ppScenario)
 
 DATASETS = ["EphemeralHLTPhysics0","EphemeralHLTPhysics1", "EphemeralHLTPhysics2", "EphemeralHLTPhysics3",
-            "EphemeralHLTPhysics4", "EphemeralHLTPhysics5", "EphemeralHLTPhysics6","EphemeralHLTPhysics7"
-            "EphemeralHLTPhysics8", "EphemeralHLTPhysics9","EphemeralHLTPhysics10","EphemeralHLTPhysics11"
-            "EphemeralHLTPhysics12","EphemeralHLTPhysics13","EphemeralHLTPhysics14","EphemeralHLTPhysics15"
+            "EphemeralHLTPhysics4", "EphemeralHLTPhysics5", "EphemeralHLTPhysics6","EphemeralHLTPhysics7",
+            "EphemeralHLTPhysics8", "EphemeralHLTPhysics9","EphemeralHLTPhysics10","EphemeralHLTPhysics11",
+            "EphemeralHLTPhysics12","EphemeralHLTPhysics13","EphemeralHLTPhysics14","EphemeralHLTPhysics15",
             "EphemeralHLTPhysics16","EphemeralHLTPhysics17","EphemeralHLTPhysics18","EphemeralHLTPhysics19"]
 
 for dataset in DATASETS:
@@ -1130,6 +1136,29 @@
                disk_node="T2_CH_CERN",
                scenario=ppScenario)
 
+## DAQ TRANSFER TEST PDs (fall 2024)
+DATASETS_DAQ_TFTEST = ["TestHLTPhysics0","TestHLTPhysics1", "TestHLTPhysics2", "TestHLTPhysics3",
+            "TestHLTPhysics4", "TestHLTPhysics5", "TestHLTPhysics6","TestHLTPhysics7",
+            "TestHLTPhysics8", "TestHLTPhysics9","TestHLTPhysics10","TestHLTPhysics11",
+            "TestHLTPhysics12", "TestHLTPhysics13","TestHLTPhysics14","TestHLTPhysics15",
+            "TestHLTPhysics16", "TestHLTPhysics17","TestHLTPhysics18","TestHLTPhysics19",
+            "TestHLTPhysics20", "TestHLTPhysics21","TestHLTPhysics22","TestHLTPhysics23",
+            "TestHLTPhysics24", "TestHLTPhysics25","TestHLTPhysics26","TestHLTPhysics27",
+            "TestHLTPhysics28", "TestHLTPhysics29","TestHLTPhysics30","TestHLTPhysics31",
+            "TestHLTPhysics32", "TestHLTPhysics33","TestHLTPhysics34","TestHLTPhysics35",
+            "TestHLTPhysics36", "TestHLTPhysics37","TestHLTPhysics38","TestHLTPhysics39"]
+
+for dataset in DATASETS_DAQ_TFTEST:
+    addDataset(tier0Config, dataset,
+               do_reco=False,
+               raw_to_disk=False,
+               dqm_sequences=["@none"],
+               archival_node=None,
+               tape_node="T0_CH_CERN_MSS",
+               disk_node="T2_CH_CERN",
+               scenario=ppScenario)
+
+
 ########################################################
 ### SpecialRandom PDs                                  ###
 ########################################################
@@ -1291,10 +1320,12 @@
 DATASETS = ["ScoutingPFRun3"]
 for dataset in DATASETS:
     addDataset(tier0Config, dataset,
-               do_reco=False,
+               do_reco=True,
+               write_aod=False,
+               write_miniaod=False,
                tape_node="T1_US_FNAL_MSS",
                disk_node="T1_US_FNAL_Disk",
-               scenario=ppScenario)
+               scenario=hltScoutingScenario)
 
 DATASETS = ["ParkingDoubleElectronLowMass","ParkingDoubleElectronLowMass0"]
 for dataset in DATASETS:
@@ -1434,6 +1465,8 @@
 ignoreStream(tier0Config, "streamDQMRates")
 ignoreStream(tier0Config, "DQMPPSRandom")
 
+# Set streams to ignore by agent. These will not be injected
+setHelperAgentStreams(tier0Config, {"SecondAgent" : DATASETS_DAQ_TFTEST})
 
 ###################################
 ### currently inactive settings ###

diff --git a/etc/ReplayOfflineConfiguration.py b/etc/ReplayOfflineConfiguration.py
@@ -28,6 +28,7 @@
 from T0.RunConfig.Tier0Config import addSiteConfig
 from T0.RunConfig.Tier0Config import setStorageSite
 from T0.RunConfig.Tier0Config import setExtraStreamDatasetMap
+from T0.RunConfig.Tier0Config import setHelperAgentStreams
 
 # Create the Tier0 configuration object
 tier0Config = createTier0Config()
@@ -44,6 +45,11 @@
 # Use this in order to limit the number of lumisections to process
 #setInjectLimit(tier0Config, 10)
 
+# Define streams to ignore. These wont be injected by the MainAgent
+setHelperAgentStreams(tier0Config, {'SecondAgent': [],
+                                    'ThirdAgent' : []
+                                    })
+
 # Settings up sites
 processingSite = "T2_CH_CERN"
 storageSite = "T0_CH_CERN_Disk"
@@ -110,7 +116,7 @@
 
 # Defaults for CMSSW version
 defaultCMSSWVersion = {
-    'default': "CMSSW_14_0_15"
+    'default': "CMSSW_14_0_16"
 }
 
 # Configure ScramArch
@@ -130,6 +136,7 @@
 alcaPPSScenario = "AlCaPPS_Run3"
 hiTestppScenario = "ppEra_Run3_pp_on_PbPb_2023"
 hiRawPrimeScenario = "ppEra_Run3_pp_on_PbPb_approxSiStripClusters_2023"
+hltScoutingScenario = "hltScoutingEra_Run3_2024"
 
 # Procesing version number replays
 # Taking Replay processing ID from the last 8 digits of the DeploymentID
@@ -1017,9 +1024,9 @@
                scenario=ppScenario)
 
 DATASETS = ["EphemeralHLTPhysics0","EphemeralHLTPhysics1", "EphemeralHLTPhysics2", "EphemeralHLTPhysics3",
-            "EphemeralHLTPhysics4", "EphemeralHLTPhysics5", "EphemeralHLTPhysics6","EphemeralHLTPhysics7"
-            "EphemeralHLTPhysics8", "EphemeralHLTPhysics9","EphemeralHLTPhysics10","EphemeralHLTPhysics11"
-            "EphemeralHLTPhysics12","EphemeralHLTPhysics13","EphemeralHLTPhysics14","EphemeralHLTPhysics15"
+            "EphemeralHLTPhysics4", "EphemeralHLTPhysics5", "EphemeralHLTPhysics6","EphemeralHLTPhysics7",
+            "EphemeralHLTPhysics8", "EphemeralHLTPhysics9","EphemeralHLTPhysics10","EphemeralHLTPhysics11",
+            "EphemeralHLTPhysics12","EphemeralHLTPhysics13","EphemeralHLTPhysics14","EphemeralHLTPhysics15",
             "EphemeralHLTPhysics16","EphemeralHLTPhysics17","EphemeralHLTPhysics18","EphemeralHLTPhysics19"]
 
 for dataset in DATASETS:
@@ -1030,6 +1037,26 @@
                write_dqm=False,
                scenario=ppScenario)
 
+## DAQ TRANSFER TEST PDs (fall 2024)
+DATASETS_DAQ_TFTEST = ["TestHLTPhysics0","TestHLTPhysics1", "TestHLTPhysics2", "TestHLTPhysics3",
+            "TestHLTPhysics4", "TestHLTPhysics5", "TestHLTPhysics6","TestHLTPhysics7",
+            "TestHLTPhysics8", "TestHLTPhysics9","TestHLTPhysics10","TestHLTPhysics11",
+            "TestHLTPhysics12", "TestHLTPhysics13","TestHLTPhysics14","TestHLTPhysics15",
+            "TestHLTPhysics16", "TestHLTPhysics17","TestHLTPhysics18","TestHLTPhysics19",
+            "TestHLTPhysics20", "TestHLTPhysics21","TestHLTPhysics22","TestHLTPhysics23",
+            "TestHLTPhysics24", "TestHLTPhysics25","TestHLTPhysics26","TestHLTPhysics27",
+            "TestHLTPhysics28", "TestHLTPhysics29","TestHLTPhysics30","TestHLTPhysics31",
+            "TestHLTPhysics32", "TestHLTPhysics33","TestHLTPhysics34","TestHLTPhysics35",
+            "TestHLTPhysics36", "TestHLTPhysics37","TestHLTPhysics38","TestHLTPhysics39"]
+
+for dataset in DATASETS_DAQ_TFTEST:
+    addDataset(tier0Config, dataset,
+               do_reco=False,
+               raw_to_disk=False,
+               dqm_sequences=["@none"],
+               scenario=ppScenario)
+
+
 ########################################################
 ### SpecialRandom PDs                                ###
 ########################################################
@@ -1187,8 +1214,10 @@
 DATASETS = ["ScoutingPFRun3"]
 for dataset in DATASETS:
     addDataset(tier0Config, dataset,
-               do_reco=False,
-               scenario=ppScenario)
+               do_reco=True,
+               write_aod=False,
+               write_miniaod=False,
+               scenario=hltScoutingScenario)
 
 DATASETS = ["ParkingDoubleElectronLowMass0","ParkingDoubleElectronLowMass1","ParkingDoubleElectronLowMass2",
             "ParkingDoubleElectronLowMass3","ParkingDoubleElectronLowMass4","ParkingDoubleElectronLowMass5",

diff --git a/etc/Tier0Config.py b/etc/Tier0Config.py
@@ -39,7 +39,7 @@
 config.TaskArchiver.useReqMgrForCompletionCheck = False
 config.TaskArchiver.dashBoardUrl = "http://dashb-luminosity.cern.ch/dashboard/request.py/putluminositydata"
 config.TaskArchiver.logLevel = "DEBUG"
-config.TaskArchiver.archiveDelayHours = 2190
+config.TaskArchiver.archiveDelayHours = 720
 config.TaskArchiver.useWorkQueue = False
 
 config.DBS3Upload.primaryDatasetType = "data"
@@ -58,7 +58,6 @@
 config.AgentStatusWatcher.onlySSB = False
 
 config.RucioInjector.blockRuleParams = {}
-config.RucioInjector.blockDeletionDelayHours = 168
 config.RucioInjector.useDsetReplicaDeep = True
 
 config.BossAir.pluginNames = ["SimpleCondorPlugin"]

diff --git a/src/python/T0/RunConfig/Tier0Config.py b/src/python/T0/RunConfig/Tier0Config.py
@@ -286,6 +286,7 @@ def createTier0Config():
     tier0Config.Global.InjectMinRun = None
     tier0Config.Global.InjectMaxRun = None
 
+    tier0Config.Global.HelperAgentStreams = {}
     tier0Config.Global.SpecifiedStreamNames = None
 
     tier0Config.Global.ScramArches = {}
@@ -857,6 +858,14 @@ def setInjectLimit(config, injectLimit):
     config.Global.InjectLimit = injectLimit
     return
 
+def setHelperAgentStreams(config, helperAgentStreams):
+    """
+    _setSecondaryAgentStreams_
+
+    Define which streams should not be injected by the agent (for dual agents)
+    """
+    config.Global.HelperAgentStreams = helperAgentStreams
+    return
 
 def setEnableUniqueWorkflowName(config):
     """

diff --git a/src/python/T0/StorageManager/StorageManagerAPI.py b/src/python/T0/StorageManager/StorageManagerAPI.py
@@ -19,7 +19,8 @@ def injectNewData(dbInterfaceStorageManager,
                   minRun = None,
                   maxRun = None,
                   injectRun = None,
-                  injectLimit = None):
+                  injectLimit = None,
+                  agentType = None):
     """
     _injectNewData_
 
@@ -72,10 +73,17 @@ def injectNewData(dbInterfaceStorageManager,
                                     transaction = False)
 
     # remove already processed files
-    newData[:] = [newFile for newFile in newData if newFile['p5_id'] not in knownStreamers]
+    # Filtering streams for Main and Secondary agents before injecting
+    # Note that every agent is Main by default
+    # Note that the secondaryAgentStreams works as a specifyStreams if the agent is secondary
 
+    newData[:] = [newFile for newFile in newData if newFile['p5_id'] not in knownStreamers]
     logging.debug("StoragemanagerAPI: found %d new files", len(newData))
 
+    if agentType:
+        logging.info("Filtering streamer files for %s", agentType.name)
+        newData = agentType.filterStreamerFiles(streamerFiles = newData)
+
     newRuns = set()
     newRunStreams = {}
     for newFile in newData:
@@ -87,6 +95,7 @@ def injectNewData(dbInterfaceStorageManager,
 
         if run not in newRunStreams:
             newRunStreams[run] = set()
+
         if stream not in newRunStreams[run]:
             newRunStreams[run].add(stream)