Fix random DQM path order from ConfigBuilder in case of duplicate DQM…

… sequences

PromptReco for Heavy Ions was generating different RECO configurations
for each job of the same Primary Dataset (and Run). This caused
problems during merging as the ParameterSet registry grew.

The difference between the RECO configurations was in how the DQM
sequences were assigned to DQM Paths. One source for random shuffling
was found to be in ConfigBuilder in case the expansion of the DQM
sequence specifier leads to duplicate sequences.

This commit fixes the path ordering, and adds a test for it.

Co-authored-by: Christopher Jones <[email protected]>

Configuration/Applications/python/ConfigBuilder.py

@@ -16,6 +16,7 @@
 import collections
 from subprocess import Popen,PIPE
 import FWCore.ParameterSet.DictTypes as DictTypes
+from FWCore.ParameterSet.OrderedSet import OrderedSet
 class Options:
     pass
 
@@ -2052,7 +2053,7 @@ def prepare_DQM(self, stepSpec = 'DQMOffline'):
         self.expandMapping(postSequenceList,autoDQM,index=1)
 
         if len(set(sequenceList))!=len(sequenceList):
-            sequenceList=list(set(sequenceList))
+            sequenceList=list(OrderedSet(sequenceList))
             print("Duplicate entries for DQM:, using",sequenceList)
 
         pathName='dqmoffline_step'
@@ -2099,7 +2100,7 @@ def prepare_HARVESTING(self, stepSpec = None):
         self.expandMapping(harvestingList,combined_mapping,index=-1)
 
         if len(set(harvestingList))!=len(harvestingList):
-            harvestingList=list(set(harvestingList))
+            harvestingList=list(OrderedSet(harvestingList))
             print("Duplicate entries for HARVESTING, using",harvestingList)
 
         for name in harvestingList:

Configuration/Applications/test/BuildFile.xml

@@ -0,0 +1 @@
+<test name="TestConfigurationApplicationsConfigBuilder" command="python3 ${LOCALTOP}/src/Configuration/Applications/test/ConfigBuilderTest.py"/>

Configuration/Applications/test/ConfigBuilderTest.py

@@ -0,0 +1,80 @@
+import FWCore.ParameterSet.Config as cms
+from Configuration.Applications.ConfigBuilder import ConfigBuilder, defaultOptions
+
+import copy
+
+def prepareDQMSequenceOrder():
+    options = copy.deepcopy(defaultOptions)
+    options.scenario = "Test"
+
+    process = cms.Process("Test")
+    process.a1 = cms.EDAnalyzer("A1")
+    process.a2 = cms.EDAnalyzer("A2")
+    process.a3 = cms.EDAnalyzer("A3")
+    process.a4 = cms.EDAnalyzer("A4")
+    process.seq1 = cms.Sequence(process.a1)
+    process.seq2 = cms.Sequence(process.a2)
+    process.seq3 = cms.Sequence(process.a3)
+    process.seq4 = cms.Sequence(process.a4)
+    process.ps1 = cms.Sequence()
+    process.ps2 = cms.Sequence()
+    process.ps3 = cms.Sequence()
+    process.ps3 = cms.Sequence()
+    process.ps4 = cms.Sequence()
+    return (options, process)
+
+if __name__=="__main__":
+    import unittest
+
+    class TestModuleCommand(unittest.TestCase):
+        def setUp(self):
+            """Nothing to do """
+            None
+
+        def testDQMSequenceOrder(self):
+            def extract(process, count):
+                if count == 0:
+                    return []
+                ret = list(process.dqmoffline_step.moduleNames())
+                for i in range(1, count):
+                    ret.extend(list(getattr(process, f"dqmoffline_{i}_step").moduleNames()))
+                return ret
+
+            # DQM sequences are unique
+            (options, process) = prepareDQMSequenceOrder()
+            order = [3, 1, 2]
+            cb = ConfigBuilder(options, process)
+            cb.prepare_DQM("+".join(f"seq{o}" for o in order))
+            self.assertEqual([f"a{o}" for o in order], extract(process, len(order)))
+
+            # Code in prepare_DQM() call assumes the 'sequenceList`
+            # has at least as many elements as `postSequenceList`. We
+            # can't fake 'postSequenceList' as it is also derived from
+            # the prepare_DQM() argument, but duplicates are not
+            # removed. The only handle we have (besides code changes,
+            # that are beyond this bug fix), is to modify the autoDQM.
+            from DQMOffline.Configuration.autoDQM import autoDQM
+            autoDQM_orig = copy.deepcopy(autoDQM)
+            autoDQM.clear()
+            autoDQM["alias1"] = ["seq1", "ps1", "not needed"]
+            autoDQM["alias2"] = ["seq2", "ps2", "not needed"]
+            autoDQM["alias3"] = ["seq3", "ps3", "not needed"]
+            # seq4 is used only to have the expanded and uniquified
+            # 'sequenceList' to have at least as many elements as
+            # 'postSequenceList'
+            autoDQM["alias4"] = ["seq2+seq4", "ps4", "not needed"]
+
+            order = [2, 1, 3]
+            cb = ConfigBuilder(options, process)
+            cb.prepare_DQM("+".join(f"@alias{o}" for o in order))
+            self.assertEqual([f"a{o}" for o in order], extract(process, len(order)))
+
+            cb = ConfigBuilder(options, process)
+            order = [2, 1, 4, 3]
+            cb.prepare_DQM("+".join(f"@alias{o}" for o in order))
+            self.assertEqual([f"a{o}" for o in order], extract(process, len(order)))
+
+            autoDQM.clear()
+            autoDQM.update(autoDQM_orig)
+
+    unittest.main()