eng-spfresh: include refresh-reference.sh helper on merge branch

Author: eng-spfresh-worker

evaluation/distributed/refresh-reference.sh

@@ -0,0 +1,97 @@
+#!/usr/bin/env bash
+# refresh-reference.sh — re-run the merge-base reference perf at <scale>
+# and cache the result under
+#   <workspace-repo>/tasks/distributed-index-scale-2k/results/_reference/<SCALE>/.
+#
+# This is a DELIBERATE HUMAN ACTION. Subagents must NOT run it on their
+# own — the protocol caches the reference once and refreshes only when
+# the merge-base advances by more than a trivial commit set. Running
+# this rotates the perf gate's reference for every fault case.
+#
+# Usage (on sdr):
+#   evaluation/distributed/refresh-reference.sh --scale 1m|10m
+#                                                [--workspace-repo PATH]
+#                                                [--cluster-conf PATH]
+#                                                [--mergebase-ref REF]
+#                                                [--yes]
+#
+# Defaults:
+#   --scale            (required)
+#   --workspace-repo   $HOME/workspace/openclaw-tasks
+#   --cluster-conf     evaluation/distributed/configs/sdr_single.conf
+#                      (auto-generated by run_perf_validation.sh on first run)
+#   --mergebase-ref    users/zhangt/merge-distributed-to-tikv
+#   --yes              skip the interactive confirmation
+#
+# Hard guards (same as run_perf_validation.sh): no spfresh-dev-amd-005,
+# no scale != {1m, 10m}.
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+SPTAG_REPO_ROOT="$(cd "${SCRIPT_DIR}/../.." && pwd)"
+
+HOSTSHORT="$(hostname -s 2>/dev/null || hostname)"
+if [[ "$HOSTSHORT" == *"spfresh-dev-amd-005"* || "$HOSTSHORT" == *"msccl-dev-000005"* ]]; then
+    echo "REFUSING: never run refresh-reference on spfresh-dev-amd-005" >&2
+    exit 10
+fi
+
+SCALE_LOW=""
+WS_REPO="${HOME}/workspace/openclaw-tasks"
+CLUSTER_CONF="${SCRIPT_DIR}/configs/sdr_single.conf"
+MERGEBASE_REF="users/zhangt/merge-distributed-to-tikv"
+ASSUME_YES=0
+
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --scale)            SCALE_LOW="$2"; shift 2 ;;
+        --workspace-repo)   WS_REPO="$2"; shift 2 ;;
+        --cluster-conf)     CLUSTER_CONF="$2"; shift 2 ;;
+        --mergebase-ref)    MERGEBASE_REF="$2"; shift 2 ;;
+        --yes|-y)           ASSUME_YES=1; shift ;;
+        *) echo "unknown arg: $1" >&2; exit 2 ;;
+    esac
+done
+
+[[ -n "$SCALE_LOW" ]] || { echo "Usage: $0 --scale 1m|10m [--yes]" >&2; exit 2; }
+case "$SCALE_LOW" in 1m|10m) ;; *) echo "REFUSING: scale '$SCALE_LOW'" >&2; exit 11 ;; esac
+SCALE_HI="${SCALE_LOW^^}"
+
+REF_DIR="$WS_REPO/tasks/distributed-index-scale-2k/results/_reference/$SCALE_HI"
+mkdir -p "$REF_DIR"
+
+# Choose a clean worktree at MERGEBASE_REF.
+REF_WT="$HOME/workspace/sptag-ft/wt/_reference"
+if [[ ! -d "$REF_WT" ]]; then
+    git -C "$SPTAG_REPO_ROOT" worktree add "$REF_WT" "$MERGEBASE_REF"
+fi
+git -C "$REF_WT" fetch origin "$MERGEBASE_REF" --quiet || true
+git -C "$REF_WT" checkout "$MERGEBASE_REF" --quiet
+COMMIT="$(git -C "$REF_WT" rev-parse HEAD)"
+
+if [[ "$ASSUME_YES" != "1" ]]; then
+    cat <<EOF
+About to refresh perf reference at $SCALE_HI.
+  branch:   $MERGEBASE_REF @ $COMMIT
+  worktree: $REF_WT
+  output:   $REF_DIR/{metrics.json,run.log,COMMIT}
+This rotates the gate for every fault case. Continue? [y/N]
+EOF
+    read -r ans; [[ "$ans" =~ ^[Yy] ]] || { echo "aborted"; exit 0; }
+fi
+
+[[ -x "$REF_WT/Release/SPTAGTest" ]] || { echo "Build SPTAGTest in $REF_WT first." >&2; exit 3; }
+
+# Run via the existing single-node harness in the reference worktree.
+pushd "$REF_WT" >/dev/null
+"$REF_WT/evaluation/distributed/run_distributed.sh" run "$CLUSTER_CONF" "$SCALE_LOW" 1 \
+    2>&1 | tee "$REF_DIR/run.log"
+popd >/dev/null
+
+EMITTED="$REF_WT/output_${SCALE_LOW}_1node.json"
+[[ -f "$EMITTED" ]] || { echo "ERROR: $EMITTED missing"; exit 4; }
+cp "$EMITTED" "$REF_DIR/metrics.json"
+echo "$COMMIT" > "$REF_DIR/COMMIT"
+date -u +%Y%m%dT%H%M%SZ > "$REF_DIR/REFRESHED_AT"
+echo "[refresh-reference] cached reference at $REF_DIR (commit $COMMIT)"