Merge pull request #6041 from IntersectMBO/bench-master

bench: improve genesis cache retrieval; maintenance for `locli` and others
IntersectMBO · Nov 22, 2024 · e37615c · e37615c
2 parents 6e969c6 + 186f0e7
commit e37615c
Show file tree

Hide file tree

Showing 19 changed files with 220 additions and 182 deletions.
diff --git a/bench/locli/CHANGELOG.md b/bench/locli/CHANGELOG.md
@@ -0,0 +1,11 @@
+# Revision history for locli
+
+## 1.36 -- Nov 2024
+
+* Add `CHANGELOG.md` for `locli`
+* Discern Plutus RIPEMD-160 workload in reports
+* Remove unused build-depends
+* Remove redundant fields from summary report: Perf analysis start/stop spread, Log text lines per host (NB. incompatible for comparison with `summary.org` files created with prior versions)
+* Remove unused CLI commands `list-logobject-keys-legacy` and `list-logobject-keys`
+* Remove unused `HostLogs` SHA256 checksums
+* Disable missing trace detection (temporarily), as raw data isn't properly evaluated to that end at the moment
diff --git a/bench/locli/locli.cabal b/bench/locli/locli.cabal
@@ -1,7 +1,7 @@
 cabal-version: 3.0
 
 name:                   locli
-version:                1.35
+version:                1.36
 synopsis:               Cardano log analysis CLI
 description:            Cardano log analysis CLI.
 category:               Cardano,
@@ -12,6 +12,7 @@ maintainer:             [email protected]
 license:                Apache-2.0
 license-files:          LICENSE
                         NOTICE
+extra-doc-files:        CHANGELOG.md
 build-type:             Simple
 
 common project-config
@@ -95,11 +96,8 @@ library
   autogen-modules:      Paths_locli
 
   build-depends:        aeson
-                      , Histogram
                       , aeson-pretty
                       , async
-                      , attoparsec
-                      , attoparsec-iso8601
                       , bytestring
                       , cardano-git-rev ^>= 0.2.2
                       , cardano-ledger-core
@@ -110,27 +108,16 @@ library
                       , directory
                       , ede
                       , extra
-                      , file-embed
                       , filepath
                       , fingertree
                       , hashable
-                      , gnuplot
-                      , iohk-monitoring
-                      , optparse-applicative-fork
-                      , optparse-generic
+                      , optparse-applicative-fork >= 0.18.1
                       , ouroboros-consensus
-                      -- for Data.SOP.Strict:
-                      , ouroboros-network ^>= 0.17
-                      , ouroboros-network-api
-                      , process
-                      , quiet
-                      , scientific
+                      , ouroboros-network-api ^>= 0.10
                       , sop-core
                       , split
                       , statistics
                       , strict-sop-core
-                      , system-filepath
-                      , template-haskell
                       , text
                       , text-short
                       , time
@@ -139,18 +126,15 @@ library
                       , transformers-except
                       , unix
                       , unordered-containers
-                      , utf8-string
                       , vector
                       , witherable
-                      , cardano-strict-containers ^>= 0.1
 
 executable locli
   import:               project-config
 
   hs-source-dirs:       app
   main-is:              locli.hs
   ghc-options:          -threaded
-                        -Wall
                         -rtsopts
                         "-with-rtsopts=-T -N7 -A2m -qb -H64m"
 

diff --git a/bench/locli/src/Cardano/Analysis/API/Ground.hs b/bench/locli/src/Cardano/Analysis/API/Ground.hs
@@ -10,14 +10,12 @@ module Cardano.Analysis.API.Ground
   )
 where
 
-import Prelude                          (show)
+import Prelude                          as P (show)
 import Cardano.Prelude                  hiding (head, toText)
 import Unsafe.Coerce                    qualified as Unsafe
 
 import Data.Aeson
 import Data.Aeson.Types                 (toJSONKeyText)
-import Data.Attoparsec.Text             qualified as Atto
-import Data.Attoparsec.Time             qualified as Iso8601
 import Data.ByteString.Lazy.Char8       qualified as LBS
 import Data.Map.Strict                  qualified as Map
 import Data.Text                        qualified as T
@@ -26,7 +24,6 @@ import Data.Text.Short                  (ShortText, fromText, toText)
 import Data.Time.Clock                  (UTCTime, NominalDiffTime)
 import Options.Applicative
 import Options.Applicative              qualified as Opt
-import Quiet                            (Quiet (..))
 import System.FilePath                  qualified as F
 
 import Cardano.Slotting.Slot            (EpochNo(..), SlotNo(..))
@@ -41,13 +38,17 @@ newtype FieldName = FieldName { unFieldName :: Text }
   deriving (Eq, Generic, Ord)
   deriving newtype (FromJSON, IsString, ToJSON)
   deriving anyclass NFData
-  deriving Show via Quiet FieldName
+
+instance Show FieldName where
+  show = ("FieldName " ++) . P.show . unFieldName
 
 newtype TId = TId { unTId :: ShortText }
   deriving (Eq, Generic, Ord)
   deriving newtype (FromJSON, ToJSON)
   deriving anyclass NFData
-  deriving Show via Quiet TId
+
+instance Show TId where
+  show = ("TId " ++) . P.show . unTId
 
 newtype Hash = Hash { unHash :: ShortText }
   deriving (Eq, Generic, Ord)
@@ -95,7 +96,9 @@ newtype Host = Host { unHost :: ShortText }
   deriving (Eq, Generic, Ord)
   deriving newtype (IsString, FromJSON, ToJSON)
   deriving anyclass NFData
-  deriving Show via Quiet Host
+
+instance Show Host where
+  show = ("Host " ++) . P.show . unHost
 
 newtype EpochSlot = EpochSlot { unEpochSlot :: Word64 }
   deriving stock (Eq, Generic, Ord, Show)
@@ -269,26 +272,6 @@ pSlotNo name desc =
      <> Opt.help desc
      )
 
-optUTCTime :: String -> String -> Parser UTCTime
-optUTCTime optname desc =
-  Opt.option (readerFromAttoParser Iso8601.utcTime)
-    $ long optname
-    <> metavar "ISO8601-TIME"
-    <> help desc
- where
-   -- Stolen from: cardano-cli/src/Cardano/CLI/Shelley/Parsers.hs
-   readerFromAttoParser :: Atto.Parser a -> Opt.ReadM a
-   readerFromAttoParser p =
-     Opt.eitherReader (Atto.parseOnly (p <* Atto.endOfInput) . T.pack)
-
-optDuration :: String -> String -> NominalDiffTime -> Parser NominalDiffTime
-optDuration optname desc def=
-  Opt.option ((realToFrac :: Double -> NominalDiffTime) <$> Opt.auto)
-    $ long optname
-    <> metavar "SEC"
-    <> help desc
-    <> value def
-
 optWord :: String -> String -> Word64 -> Parser Word64
 optWord optname desc def =
   Opt.option auto

diff --git a/bench/locli/src/Cardano/Analysis/API/Metrics.hs b/bench/locli/src/Cardano/Analysis/API/Metrics.hs
@@ -48,11 +48,10 @@ sumFieldsReport =
   , "plutusScript"
   , "sumHosts", "sumLogObjectsTotal"
   , "sumFilters"
-  , "cdfLogLinesEmitted", "cdfLogObjectsEmitted", "cdfLogObjects"
+  , "cdfLogObjectsEmitted", "cdfLogObjects"
   , "cdfRuntime", "cdfLogLineRate"
   , "ddRawCount.sumDomainTime", "ddFilteredCount.sumDomainTime", "dataDomainFilterRatio.sumDomainTime"
   , "ddRaw.sumStartSpread", "ddRaw.sumStopSpread"
-  , "ddFiltered.sumStartSpread", "ddFiltered.sumStopSpread"
   , "sumDomainSlots", "sumDomainBlocks", "sumBlocksRejected"]
 
 instance (KnownCDF f) => TimelineFields (Summary f)  where
@@ -95,8 +94,8 @@ instance (KnownCDF f) => TimelineFields (Summary f)  where
       ""
 
    <> fScalar "utxo"                   W12 Cnt (IWord64 $           utxo.sumGenesisSpec)
-      "Starting UTxO set size"
-      "Extra UTxO set size at the beginning of the benchmark"
+      "Stuffed UTxO size"
+      "Extra UTxO set entries (from genesis)"
 
    <> fScalar "dreps"                  W12 Cnt (IWord64 $          dreps.sumGenesisSpec)
       "DRep count"
@@ -136,10 +135,6 @@ instance (KnownCDF f) => TimelineFields (Summary f)  where
       "Number of filters applied"
       ""
 
-   <> fScalar "cdfLogLinesEmitted"     W12 Cnt (IFloat $ cdfAverageVal.cdfLogLinesEmitted)
-      "Log text lines emitted per host"
-      ""
-
    <> fScalar "cdfLogObjectsEmitted"   W12 Cnt (IFloat $ cdfAverageVal.cdfLogObjectsEmitted)
       "Log objects emitted per host"
       ""
@@ -180,14 +175,6 @@ instance (KnownCDF f) => TimelineFields (Summary f)  where
       "Node stop spread, s"
       ""
 
-   <> fScalar "ddFiltered.sumStartSpread" W9 Sec (IDeltaT$ maybe 0 (intvDurationSec.fmap (fromRUTCTime . arityProj cdfMedian)).ddFiltered.sumStartSpread)
-      "Perf analysis start spread, s"
-      ""
-
-   <> fScalar "ddFiltered.sumStopSpread"  W9 Sec (IDeltaT$ maybe 0 (intvDurationSec.fmap (fromRUTCTime . arityProj cdfMedian)).ddFiltered.sumStopSpread)
-      "Perf analysis stop spread, s"
-      ""
-
    <> fScalar "sumDomainSlots"         W12 Slo (IInt  $ floor.arityProj cdfMedian.cdfAverage.ddFilteredCount.sumDomainSlots)
       "Slots analysed"
       ""
@@ -211,7 +198,6 @@ instance (KnownCDF f) => TimelineFields (Summary f)  where
                        , manifestPackages <&> pkgCommit
                        ]
 
-
 -- fieldJSONOverlay f = (:[]) . tryOverlayFieldDescription f
 
 propSubsetFn :: PropSubset -> (Field DSelect p a -> Bool)

diff --git a/bench/locli/src/Cardano/Analysis/API/Types.hs b/bench/locli/src/Cardano/Analysis/API/Types.hs
@@ -50,7 +50,6 @@ data Summary f where
     , sumDomainBlocks        :: !(DataDomain f BlockNo)
     , sumProfilingData       :: !(Maybe (ProfilingData (CDF I)))
 
-    , cdfLogLinesEmitted     :: !(CDF f Int)
     , cdfLogObjectsEmitted   :: !(CDF f Int)
     , cdfLogObjects          :: !(CDF f Int)
     , cdfRuntime             :: !(CDF f NominalDiffTime)

diff --git a/bench/locli/src/Cardano/Analysis/Summary.hs b/bench/locli/src/Cardano/Analysis/Summary.hs
@@ -71,7 +71,6 @@ summariseMultiSummary sumAnalysisTime centiles xs@(headline:xss) = do
   sumMeta                <- summariseMetadata $ xs <&> sumMeta
   sumFilters             <- allEqOrElse (xs <&> sumFilters) SEIncoherentRunFilters
 
-  cdfLogLinesEmitted     <- sumCDF2 $ xs <&> cdfLogLinesEmitted
   cdfLogObjectsEmitted   <- sumCDF2 $ xs <&> cdfLogObjectsEmitted
   cdfLogObjects          <- sumCDF2 $ xs <&> cdfLogObjects
   cdfRuntime             <- sumCDF2 $ xs <&> cdfRuntime
@@ -172,7 +171,6 @@ computeSummary sumAnalysisTime
   --
   , cdfLogObjects        = cdf stdCentiles (objLists <&> length)
   , cdfLogObjectsEmitted = cdf stdCentiles logObjectsEmitted
-  , cdfLogLinesEmitted   = cdf stdCentiles textLinesEmitted
   , cdfRuntime           = cdf stdCentiles runtimes
   , ..
   }
@@ -183,7 +181,7 @@ computeSummary sumAnalysisTime
     rlHostLogs
     & Map.elems
 
-   (,) logObjectsEmitted textLinesEmitted =
+   (logObjectsEmitted, textLinesEmitted) =
      hostLogs
      & fmap (hlRawLogObjects &&& hlRawLines)
      & unzip
@@ -198,7 +196,7 @@ computeSummary sumAnalysisTime
    lineRates  = zipWith (/) (textLinesEmitted <&> fromIntegral)
                             (runtimes <&> fromIntegral @Int . truncate)
 
-   (,,) sumDomainTime sumStartSpread sumStopSpread =
+   (sumDomainTime, sumStartSpread, sumStopSpread) =
      slotDomains sumGenesis (losFirsts, losLasts) mpDomainSlots
 
    sumChainRejectionStats :: [(ChainFilter, Int)]

diff --git a/bench/locli/src/Cardano/Command.hs b/bench/locli/src/Cardano/Command.hs
@@ -47,10 +47,7 @@ newtype Command
   deriving Show
 
 data ChainCommand
-  = ListLogobjectKeys       TextOutputFile
-  | ListLogobjectKeysLegacy TextOutputFile
-
-  |        ReadMetaGenesis  (JsonInputFile  RunPartial) (JsonInputFile  Genesis)
+  =        ReadMetaGenesis  (JsonInputFile  RunPartial) (JsonInputFile  Genesis)
   |        WriteMetaGenesis TextOutputFile              TextOutputFile
 
   |        Unlog            (JsonInputFile (RunLogs ())) Bool (Maybe [LOAnyType])
@@ -104,12 +101,6 @@ data ChainCommand
 parseChainCommand :: Parser ChainCommand
 parseChainCommand =
   subparser (mconcat [ commandGroup "Common data:  logobject keys, run metafile & genesis"
-   , op "list-logobject-keys" "List logobject keys that analyses care about"
-     (ListLogobjectKeys
-       <$> optTextOutputFile  "keys"           "Text file to write logobject keys to")
-   , op "list-logobject-keys-legacy" "List legacy logobject keys that analyses care about"
-     (ListLogobjectKeysLegacy
-       <$> optTextOutputFile  "keys-legacy"     "Text file to write logobject keys to")
    , op "read-meta-genesis" "Read the run metadata:  meta.json and Shelley genesis"
      (ReadMetaGenesis
        <$> optJsonInputFile "run-metafile"    "The meta.json file from the benchmark run"
@@ -391,22 +382,9 @@ sAnchor State{sTags=[]} = error "sAnchor with no run or multi-summary."
 sAnchor s@State{sTags}
   = stateAnchor sTags s
 
-quote :: T.Text -> T.Text
-quote = (<> "\"") . ("\"" <>)
 
 runChainCommand :: State -> ChainCommand -> ExceptT CommandError IO State
 
-runChainCommand s
-  c@(ListLogobjectKeys f) = do
-  dumpText "logobject-keys" (quote . toText <$> logObjectStreamInterpreterKeys) f
-    & firstExceptT (CommandError c)
-  pure s
-runChainCommand s
-  c@(ListLogobjectKeysLegacy f) = do
-  dumpText "logobject-keys-legacy" (quote . toText <$> logObjectStreamInterpreterKeysLegacy) f
-    & firstExceptT (CommandError c)
-  pure s
-
 runChainCommand s
   c@(ReadMetaGenesis runMeta shelleyGenesis) = do
   progress "run" (Q $ printf "reading run metadata & Shelley genesis")
@@ -601,7 +579,7 @@ runChainCommand s@State{sRun=Just _run, sSlots=Just slots}
 runChainCommand _ c@TimelineSlots{} = missingCommandData c
   ["run metadata & genesis", "filtered slots"]
 
-runChainCommand s@State{sRun=Just run, sChain=Just chain@Chain{..}, sRunLogs}
+runChainCommand s@State{sRun=Just run, sChain=Just chain@Chain{..}}
   c@ComputePropagation = do
   progress "block-propagation" $ J (cDomBlocks, cDomSlots)
   prop <- pure (blockProp run chain)
@@ -615,9 +593,11 @@ runChainCommand s@State{sRun=Just run, sChain=Just chain@Chain{..}, sRunLogs}
        renderBlockPropError e
        <> maybe "" ((".\n\n  Missing traces in run logs (not all are fatal):\n\n    " <>)
                     . T.intercalate "\n    "
-                    . fmap toText
-                    . rlMissingTraces)
-                sRunLogs
+                    . const []
+                   )
+                Nothing
+                -- TODO: reactivate output once there's a proper evaluation of raw data
+
 runChainCommand _ c@ComputePropagation = missingCommandData c
   ["run metadata & genesis", "chain", "data domains for slots & blocks"]
 

diff --git a/bench/locli/src/Cardano/Report.hs b/bench/locli/src/Cardano/Report.hs
@@ -80,6 +80,7 @@ data Workload
   | WPlutusLoopCountdown
   | WPlutusLoopSECP
   | WPlutusLoopBLST
+  | WPlutusLoopRipemd
   | WPlutusUnknown
 
 instance ToJSON Workload where
@@ -88,13 +89,15 @@ instance ToJSON Workload where
     WPlutusLoopCountdown -> "Plutus countdown loop"
     WPlutusLoopSECP      -> "Plutus SECP loop"
     WPlutusLoopBLST      -> "Plutus BLST loop"
+    WPlutusLoopRipemd    -> "Plutus RIPEMD-160 loop"
     WPlutusUnknown       -> "Plutus (other)"
 
 filenameInfix :: Workload -> Text
 filenameInfix = \case
   WPlutusLoopCountdown  -> "plutus"
   WPlutusLoopSECP       -> "plutus-secp"
   WPlutusLoopBLST       -> "plutus-blst"
+  WPlutusLoopRipemd     -> "plutus-ripemd"
   WValue                -> "value-only"
   _                     -> "unknown"
 
@@ -150,6 +153,7 @@ liftTmplRun Summary{sumWorkload=generatorProfile
         | script == "EcdsaSecp256k1Loop"    -> WPlutusLoopSECP
         | script == "SchnorrSecp256k1Loop"  -> WPlutusLoopSECP
         | script == "HashOntoG2AndAdd"      -> WPlutusLoopBLST
+        | script == "Ripemd160"             -> WPlutusLoopRipemd
         | otherwise                         -> WPlutusUnknown
   }
 

diff --git a/bench/locli/src/Cardano/Unlog/LogObject.hs b/bench/locli/src/Cardano/Unlog/LogObject.hs
@@ -87,11 +87,8 @@ data HostLogs a
   = HostLogs
     { hlRawLogfiles    :: [FilePath]
     , hlRawLines       :: Int
-    , hlRawSha256      :: Hash
     , hlRawTraceFreqs  :: Map Text Int
-    , hlMissingTraces  :: [Text]
     , hlLogs           :: (JsonLogfile, a)
-    , hlFilteredSha256 :: Hash
     , hlProfile        :: [ProfileEntry I]
     , hlRawFirstAt     :: Maybe UTCTime
     , hlRawLastAt      :: Maybe UTCTime
@@ -107,8 +104,6 @@ hlRawLogObjects = sum . Map.elems . hlRawTraceFreqs
 data RunLogs a
   = RunLogs
     { rlHostLogs      :: Map.Map Host (HostLogs a)
-    , rlMissingTraces :: [Text]
-    , rlFilterKeys    :: [Text]
     , rlFilterDate    :: UTCTime
     }
   deriving (Generic, FromJSON, ToJSON)