From 8d46fb2c11638367a7d285b31486c484240a7ed1 Mon Sep 17 00:00:00 2001
From: Kosyrev Serge <serge.kosyrev@iohk.io>
Date: Tue, 21 Dec 2021 01:54:38 +0300
Subject: [PATCH] workbench:  redo chain filtering specification

---
 .../base-k10.json}                            |  6 --
 .../base.json}                                |  0
 bench/chain-filters/no-ebnd.json              |  7 ++
 bench/chain-filters/no-rewards.json           | 13 +++
 bench/chain-filters/rewards.json              | 13 +++
 bench/chain-filters/size-full.json            |  7 ++
 bench/chain-filters/size-mid.json             | 13 +++
 bench/chain-filters/size-small.json           | 13 +++
 bench/flt-full.json                           | 25 -----
 bench/flt-midsize.json                        | 31 -------
 bench/flt-small.json                          | 31 -------
 nix/workbench/analyse.sh                      | 92 +++++++------------
 nix/workbench/profiles/defaults.jq            |  3 +-
 nix/workbench/profiles/node-services.nix      |  3 +-
 nix/workbench/run.sh                          | 12 +++
 nix/workbench/wb                              |  2 -
 16 files changed, 115 insertions(+), 156 deletions(-)
 rename bench/{flt-full-k10.json => chain-filters/base-k10.json} (74%)
 rename bench/{flt-range.json => chain-filters/base.json} (100%)
 create mode 100644 bench/chain-filters/no-ebnd.json
 create mode 100644 bench/chain-filters/no-rewards.json
 create mode 100644 bench/chain-filters/rewards.json
 create mode 100644 bench/chain-filters/size-full.json
 create mode 100644 bench/chain-filters/size-mid.json
 create mode 100644 bench/chain-filters/size-small.json
 delete mode 100644 bench/flt-full.json
 delete mode 100644 bench/flt-midsize.json
 delete mode 100644 bench/flt-small.json

diff --git a/bench/flt-full-k10.json b/bench/chain-filters/base-k10.json
similarity index 74%
rename from bench/flt-full-k10.json
rename to bench/chain-filters/base-k10.json
index 08aa821f92a..bdcdb339593 100644
--- a/bench/flt-full-k10.json
+++ b/bench/chain-filters/base-k10.json
@@ -16,10 +16,4 @@
     , "contents": true
     }
   }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessGEq"
-    , "contents": 0.9
-    }
-  }
 ]
diff --git a/bench/flt-range.json b/bench/chain-filters/base.json
similarity index 100%
rename from bench/flt-range.json
rename to bench/chain-filters/base.json
diff --git a/bench/chain-filters/no-ebnd.json b/bench/chain-filters/no-ebnd.json
new file mode 100644
index 00000000000..542f3b43d28
--- /dev/null
+++ b/bench/chain-filters/no-ebnd.json
@@ -0,0 +1,7 @@
+[ { "tag": "CSlot"
+  , "contents":
+    { "tag": "EpSlotGEq"
+    , "contents": 200
+    }
+  }
+]
diff --git a/bench/chain-filters/no-rewards.json b/bench/chain-filters/no-rewards.json
new file mode 100644
index 00000000000..4988007a02e
--- /dev/null
+++ b/bench/chain-filters/no-rewards.json
@@ -0,0 +1,13 @@
+[ { "tag": "CSlot"
+  , "contents":
+    { "tag": "EpochSafeIntLEq"
+    , "contents": 3
+    }
+  }
+, { "tag": "CSlot"
+  , "contents":
+    { "tag": "EpochSafeIntGEq"
+    , "contents": 8
+    }
+  }
+]
diff --git a/bench/chain-filters/rewards.json b/bench/chain-filters/rewards.json
new file mode 100644
index 00000000000..8c8507bd142
--- /dev/null
+++ b/bench/chain-filters/rewards.json
@@ -0,0 +1,13 @@
+[ { "tag": "CSlot"
+  , "contents":
+    { "tag": "EpochSafeIntGEq"
+    , "contents": 4
+    }
+  }
+, { "tag": "CSlot"
+  , "contents":
+    { "tag": "EpochSafeIntLEq"
+    , "contents": 7
+    }
+  }
+]
diff --git a/bench/chain-filters/size-full.json b/bench/chain-filters/size-full.json
new file mode 100644
index 00000000000..2fe94ab0b76
--- /dev/null
+++ b/bench/chain-filters/size-full.json
@@ -0,0 +1,7 @@
+[ { "tag": "CBlock"
+  , "contents":
+    { "tag": "BFullnessGEq"
+    , "contents": 0.9
+    }
+  }
+]
diff --git a/bench/chain-filters/size-mid.json b/bench/chain-filters/size-mid.json
new file mode 100644
index 00000000000..d48a8be529a
--- /dev/null
+++ b/bench/chain-filters/size-mid.json
@@ -0,0 +1,13 @@
+[ { "tag": "CBlock"
+  , "contents":
+    { "tag": "BFullnessGEq"
+    , "contents": 0.1
+    }
+  }
+, { "tag": "CBlock"
+  , "contents":
+    { "tag": "BFullnessLEq"
+    , "contents": 0.9
+    }
+  }
+]
diff --git a/bench/chain-filters/size-small.json b/bench/chain-filters/size-small.json
new file mode 100644
index 00000000000..96380c8526f
--- /dev/null
+++ b/bench/chain-filters/size-small.json
@@ -0,0 +1,13 @@
+[ { "tag": "CBlock"
+  , "contents":
+    { "tag": "BFullnessGEq"
+    , "contents": 0.03
+    }
+  }
+, { "tag": "CBlock"
+  , "contents":
+    { "tag": "BFullnessLEq"
+    , "contents": 0.1
+    }
+  }
+]
diff --git a/bench/flt-full.json b/bench/flt-full.json
deleted file mode 100644
index 297ded2bc76..00000000000
--- a/bench/flt-full.json
+++ /dev/null
@@ -1,25 +0,0 @@
-[ { "tag": "CSlot"
-  , "contents":
-    { "tag": "EpochGEq"
-    , "contents": 2
-    }
-  }
-, { "tag": "CSlot"
-  , "contents":
-    { "tag": "SlotLEq"
-    , "contents": 56000
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BUnitaryChainDelta"
-    , "contents": true
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessGEq"
-    , "contents": 0.9
-    }
-  }
-]
diff --git a/bench/flt-midsize.json b/bench/flt-midsize.json
deleted file mode 100644
index da612d78506..00000000000
--- a/bench/flt-midsize.json
+++ /dev/null
@@ -1,31 +0,0 @@
-[ { "tag": "CSlot"
-  , "contents":
-    { "tag": "EpochGEq"
-    , "contents": 2
-    }
-  }
-, { "tag": "CSlot"
-  , "contents":
-    { "tag": "SlotLEq"
-    , "contents": 56000
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BUnitaryChainDelta"
-    , "contents": true
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessGEq"
-    , "contents": 0.1
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessLEq"
-    , "contents": 0.9
-    }
-  }
-]
diff --git a/bench/flt-small.json b/bench/flt-small.json
deleted file mode 100644
index 3688d741670..00000000000
--- a/bench/flt-small.json
+++ /dev/null
@@ -1,31 +0,0 @@
-[ { "tag": "CSlot"
-  , "contents":
-    { "tag": "EpochGEq"
-    , "contents": 2
-    }
-  }
-, { "tag": "CSlot"
-  , "contents":
-    { "tag": "SlotLEq"
-    , "contents": 56000
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BUnitaryChainDelta"
-    , "contents": true
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessGEq"
-    , "contents": 0.03
-    }
-  }
-, { "tag": "CBlock"
-  , "contents":
-    { "tag": "BFullnessLEq"
-    , "contents": 0.1
-    }
-  }
-]
diff --git a/nix/workbench/analyse.sh b/nix/workbench/analyse.sh
index 481c4de7af3..f1cb6b3b021 100644
--- a/nix/workbench/analyse.sh
+++ b/nix/workbench/analyse.sh
@@ -1,32 +1,40 @@
 usage_analyse() {
      usage "analyse" "Analyse cluster runs" <<EOF
-    standard RUN-NAME     Standard batch of analyses: block-propagation, and
+    standard RUN-NAME..   Standard batch of analyses: block-propagation, and
                             machine-timeline
 
-    block-propagation RUN-NAME
+    block-propagation RUN-NAME..
                           Block propagation analysis for the entire cluster.
 
     machine-timeline RUN-NAME [MACH-NAME=node-1]
-                          Produce a general performance timeline for MACH-NAME
+                          Performance timeline for MACH-NAME.
 
-    chaininfo RUN-NAME    Print basic parameters of a run, as seen by locli
+    chaininfo RUN-NAME    Print basic parameters of a run, as seen by locli.
 
     Options of 'analyse' command:
 
-       --chain-filters F  Read chain filters to apply from the F JSON file
-       --reanalyse        Skip the preparatory steps and launch 'locli' directly
+       --filters F,F,F..  Comma-separated list of named chain filters:  see bench/chain-filters
+                            Note: filter names have no .json suffix
        --dump-logobjects  Dump the intermediate data: lifted log objects
 EOF
 }
 
 analyse() {
-local time= dump_logobjects= self_args=() locli_args=() prefilter='true' prefilter_jq='false'
+local dump_logobjects= force_prefilter= prefilter_jq= self_args=() locli_args=() filters=()
 while test $# -gt 0
 do case "$1" in
-       --reanalyse | --re ) prefilter='false';       self_args+=($1);;
-       --prefilter-jq )     prefilter_jq='true';     self_args+=($1);;
        --dump-logobjects )  dump_logobjects='true';  self_args+=($1);;
-       --chain-filters )    locli_args+=($1 $2);     self_args+=($1 $2); shift;;
+       --force-prefilter )  force_prefilter='true';  self_args+=($1);;
+       --prefilter-jq )     prefilter_jq='true';     self_args+=($1);;
+       --filters )          local filter_names=('base')
+                            filter_names+=($(echo $2 | sed 's_,_ _'))
+                            local filter_paths=(${filter_names[*]/#/"bench/chain-filters/"})
+                            local filter_files=(${filter_paths[*]/%/.json})
+                            for f in ${filter_files[*]}
+                            do test -f "$f" ||
+                                    fail "no such filter: $f"; done
+                            locli_args+=(${filter_files[*]/#/--filter })
+                            self_args+=($1 $2); shift;;
        * ) break;; esac; shift; done
 
 local op=${1:-$(usage_analyse)}; shift
@@ -45,8 +53,8 @@ case "$op" in
         ;;
     standard | std )
         for r in $*
-        do analyse ${self_args[*]}             block-propagation $r
-           analyse ${self_args[*]} --reanalyse machine-timeline  $r
+        do analyse ${self_args[*]} block-propagation $r
+           analyse ${self_args[*]} machine-timeline  $r
         done
         ;;
     block-propagation | bp )
@@ -56,8 +64,6 @@ case "$op" in
         local dir=$(run get "$name")
         test -n "$dir" || fail "malformed run: $name"
 
-        echo "{ \"run\": \"$(jq .meta.tag "$dir"/meta.json --raw-output)\" }"
-
         local adir=$dir/analysis
         mkdir -p "$adir"
 
@@ -66,11 +72,13 @@ case "$op" in
         locli analyse substring-keys > "$keyfile"
 
         ## 1. enumerate logs, filter by keyfile & consolidate
-        local logdirs=($(ls -d "$dir"/node-*/ 2>/dev/null) $(ls -d "$dir"/analysis/node-*/ 2>/dev/null))
-        # "$dir"/node-*/ "$dir"/analysis/node-*/
+        local logdirs=($(ls -d "$dir"/node-*/ 2>/dev/null))
+        local logfiles=($(ls "$adir"/logs-node-*.flt.json 2>/dev/null))
+        # echo logfiles: _${logfiles[*]}_
+        local prefilter=$(test "$force_prefilter" = 'true' -o -z "${logfiles[*]}" && echo 'true' || echo 'false')
 
-        echo "{ \"prefilter\": $prefilter, \"prefilter_jq\": $prefilter_jq }"
-        if test "$prefilter" = 'true' -o -z "$(ls "$adir"/logs-node-*.flt.json 2>/dev/null)"
+        echo "{ \"prefilter\": $prefilter }"
+        if test x$prefilter = xtrue
         then
             local jq_args=(
                 --sort-keys
@@ -94,16 +102,15 @@ case "$op" in
             wait
         fi
 
-        echo "{ \"dataSetSizeMB\": $(echo $(($(cat "$adir"/*.flt.json | wc -c) / 1000 / 1000))) }"
         locli_args+=(
             --genesis         "$dir"/genesis-shelley.json
             --run-metafile    "$dir"/meta.json
             ## ->
             --timeline-pretty "$adir"/block-propagation.txt
             --analysis-json   "$adir"/block-propagation.json
+            $(if test -n "$dump_logobjects"
+              then echo --logobjects-json "$adir"/logs-cluster.logobjects.json; fi)
         )
-        if test -n "$dump_logobjects"; then
-            locli_args+=(--logobjects-json "$adir"/logs-cluster.logobjects.json); fi
 
         time locli 'analyse' 'block-propagation' \
              "${locli_args[@]}" "$adir"/*.flt.json
@@ -112,38 +119,6 @@ case "$op" in
         if test $# -gt 0
         then analyse ${self_args[*]} block-propagation "$@"; fi;;
 
-    grep-filtered-logs | grep | g )
-        local usage="USAGE: wb analyse $op BLOCK [MACHSPEC=*] [RUN-NAME=current]"
-        local expr=$1
-        local mach=${2:-*}
-        local name=${3:-current}
-        local dir=$(run get "$name")
-        local adir=$dir/analysis
-
-        grep -h "$expr" "$adir"/logs-$mach.flt.json;;
-
-    list-blocks | blocks | bs )
-        local usage="USAGE: wb analyse $op [RUN-NAME=current]"
-        local name=${1:-current}
-        local dir=$(run get "$name")
-        local adir=$dir/analysis
-
-        fgrep -h "TraceForgedBlock" "$adir"/*.flt.json |
-            jq '{ at: .at, host: .host } * .data | del(.peer) | del(.slot)' -c |
-            sort | uniq;;
-
-    block-propagation-block | bpb )
-        local usage="USAGE: wb analyse $op BLOCK [RUN-NAME=current]"
-        local block=$1
-        local name=${2:-current}
-        local dir=$(run get "$name")
-        local adir=$dir/analysis
-
-        grep -h "$block" "$adir"/*.flt.json |
-            grep 'AddBlock\|TraceForgedBlock\|AddedToCurrentChain' |
-            jq '{ at: .at, host: .host } * .data | del(.peer) | del(.slot)' -c |
-            sort --stable | uniq;;
-
     machine-timeline | machine | mt )
         local usage="USAGE: wb analyse $op [RUN-NAME=current] [MACH-NAME=node-1]"
         local name=${1:-current}
@@ -151,14 +126,12 @@ case "$op" in
         local dir=$(run get "$name")
         test -n "$dir" || fail "malformed run: $name"
 
-        echo "{ \"run\": \"$(jq .meta.tag "$dir"/meta.json --raw-output)\" }"
-
         local adir=$dir/analysis
         mkdir -p "$adir"
 
         ## 0. subset what we care about into the keyfile
         local keyfile=$adir/substring-keys
-        locli analyse substring-keys | grep -v 'Temporary modify' > "$keyfile"
+        locli analyse substring-keys > "$keyfile"
 
         if test "$mach" = 'all'
         then local machs=($(run list-hosts $name))
@@ -168,13 +141,16 @@ case "$op" in
         do throttle_shell_job_spawns
            (
            ## 1. enumerate logs, filter by keyfile & consolidate
-           local logs=($(ls "$dir"/$mach/stdout* 2>/dev/null | tac) $(ls "$dir"/$mach/node-*.json 2>/dev/null) $(ls "$dir"/analysis/$mach/node-*.json 2>/dev/null)) consolidated="$adir"/logs-$mach.json
+               local logs=($(ls "$dir"/$mach/node-*.json 2>/dev/null))
+               local consolidated="$adir"/logs-$mach.json
 
            test -n "${logs[*]}" ||
                fail "no logs for $mach in run $name"
 
+           local prefilter=$(test "$force_prefilter" = 'true' -o -z "$(ls "$adir"/logs-node-*.flt.json 2>/dev/null)" && echo 'yes')
+
            echo "{ \"prefilter\": $prefilter }"
-           if test "$prefilter" = 'true' -o -z "$(ls "$adir"/logs-$mach.json 2>/dev/null)"
+           if test -n "$prefilter"
            then grep -hFf "$keyfile" "${logs[@]}"  > "$consolidated"; fi
 
            locli_args+=(
diff --git a/nix/workbench/profiles/defaults.jq b/nix/workbench/profiles/defaults.jq
index 323bfef5bec..3e174088960 100644
--- a/nix/workbench/profiles/defaults.jq
+++ b/nix/workbench/profiles/defaults.jq
@@ -9,8 +9,6 @@ def era_defaults($era):
   ## Choice of a cluster run scenario (wb scenario --help):
   , scenario:                         "default"
 
-  , tracing_backend:                  "iohk-monitoring"  ## or "trace-dispatcher"
-
   ## Cluster topology and composition:
   , composition:
     { locations:                      ["LO"]
@@ -73,6 +71,7 @@ def era_defaults($era):
 
   , node:
     { rts_flags_override:             []
+    , tracing_backend:                "iohk-monitoring"  ## or "trace-dispatcher"
     }
 
   , tolerances:
diff --git a/nix/workbench/profiles/node-services.nix b/nix/workbench/profiles/node-services.nix
index 47cb718c03b..74b474d356c 100644
--- a/nix/workbench/profiles/node-services.nix
+++ b/nix/workbench/profiles/node-services.nix
@@ -129,6 +129,7 @@ let
             TestAlonzoHardForkAtEpoch  = 0;
           };
       }.${profile.value.era};
+    };
     in
     backend.finaliseNodeService nodeSpec
     {
@@ -142,7 +143,7 @@ let
            (if __hasAttr "preset" profile.value
             then readJSONMay (./presets + "/${profile.value.preset}/config.json")
             else nodeConfigBits.era_setup_hardforks //
-                 nodeConfigBits.tracing.${profile.tracing_backend}));
+                 nodeConfigBits.tracing.${profile.value.node.tracing_backend}));
     };
 
   ## Given an env config, evaluate it and produce the node service.
diff --git a/nix/workbench/run.sh b/nix/workbench/run.sh
index ed3a50032bb..9f7205f80ed 100644
--- a/nix/workbench/run.sh
+++ b/nix/workbench/run.sh
@@ -351,6 +351,18 @@ case "$op" in
         then jq             'keys | .[]' -r "$dir"/node-specs.json
         else jq '.hostname | keys | .[]' -r "$dir"/meta.json; fi;;
 
+    fetch-analysis | fa )
+        local usage="USAGE: wb run $op ENV DEPL BATCH-OR-TAG.."
+        local   env=${1:?$usage}; shift
+        local  depl=${1:?$usage}; shift
+
+        for x in $*
+        do
+            ssh $env -- \
+                sh -c "'cd $depl/runs && tar c {*.$x.*,$x}/analysis/{block-propagation,logs-node-1.timeline}.txt --zstd --ignore-failed-read'" 2>/dev/null |
+                (cd run; tar x --zstd); done
+        ;;
+
     remote-machine-run-slice-list | rmrsl )
         local usage="USAGE: wb run $op ENV DEPL [HOST=DEPL]"
         local env=${1:?$usage}
diff --git a/nix/workbench/wb b/nix/workbench/wb
index 6021ad81ea8..5c930e557ae 100755
--- a/nix/workbench/wb
+++ b/nix/workbench/wb
@@ -49,7 +49,6 @@ usage_extra() {
   Other OPs:
 
     scenario     (s)      Run scenario control
-    reanalyse    (rea)    Re-analyse a cluster run (synonym for analyse --re)
     call ARGS..           Call internal functions with arguments
     mode                  Report mode set by --set-mode
 
@@ -191,7 +190,6 @@ main() {
         ## Internals:
         #
         scenario     | s )        scenario            "$@";;
-        reanalyse    | rea | re ) analyse --reanalyse "$@";;
         backend )                 backend             "$@";;
         call )                                        "$@";;
         mode )                    echo -n      $global_mode;;