fix(scrape): skip active stale temps and retry salvage merge

Avoid salvaging export.json while a channel export is still writing
(default: skip temps modified within 120s). Retry truncate+merge once
when merge fails on a quiescent partial temp. Adds active-skip smoke
and ages stale fixture mtime so salvage tests stay deterministic.

docs/plans/2026-06-04-051-fix-stale-temp-salvage-active-race-plan.md

@@ -0,0 +1,42 @@
+---
+title: "fix: Skip active stale temps and retry salvage merge"
+type: fix
+status: complete
+date: 2026-06-04
+origin: /lfg — yes_general logs show Stale temp merge failed while export.json still growing (73MB+ invalid JSON)
+---
+
+# fix: Skip active stale temps and retry salvage merge
+
+## Problem
+
+`salvage_stale_temp_exports` can run while a channel export is still writing `export.json`. The file is truncated/invalid, `merge_exports_auto` fails, and the temp is retained — but the next incremental pass hits the same race. Observed on KotOR `yes_general` (`221726893064454144`): merge fails on ~82MB partial temp while archive stays at 266182 messages (2021 cursor).
+
+Salvage after export completes works (truncated temp → 79529 messages merges to 345711 in ~58s).
+
+## Requirements
+
+| ID | Requirement |
+|----|-------------|
+| R1 | Skip stale temp dirs whose `export.json` was modified within `DCE_STALE_TEMP_MIN_AGE_SECONDS` (default 120) |
+| R2 | On merge failure, re-run `salvage_truncated_json` and retry merge once before retaining temp |
+| R3 | Log merge retry vs skip-active with distinct messages |
+| R4 | Offline smoke: active temp skipped; retry succeeds after simulated truncation |
+| R5 | `run-all-smokes.sh` passes |
+
+## Implementation
+
+- `scripts/run-discord-scrape.sh` — `stale_temp_is_active`, skip guard, merge retry helper
+- `scripts/tests/run-discord-scrape-smoke.sh` — active-temp skip + merge-retry scenarios
+
+## Verification
+
+```bash
+./scripts/tests/run-discord-scrape-smoke.sh
+DCE_MIN_FREE_MB=0 ./scripts/run-all-smokes.sh
+```
+
+## Out of scope
+
+- Completing yes_general catch-up inside LFG
+- Container memory limits

scripts/run-discord-scrape.sh

@@ -629,12 +629,62 @@ PY
   fi
 }
 
+stale_temp_is_active() {
+  local stale_export=$1
+  local min_age=${DCE_STALE_TEMP_MIN_AGE_SECONDS:-120}
+  local now mtime age
+
+  if [[ "${DCE_SALVAGE_ACTIVE_TEMPS:-0}" == "1" ]]; then
+    return 1
+  fi
+
+  now=$(date +%s)
+  mtime=$(stat -c '%Y' "$stale_export" 2>/dev/null || stat -f '%m' "$stale_export" 2>/dev/null || echo 0)
+  age=$((now - mtime))
+  (( age < min_age ))
+}
+
+merge_stale_export_into_destination() {
+  local destination_path=$1
+  local stale_export=$2
+  local stale_dir=$3
+  local salvage_merged="$stale_dir/merged.json"
+  local attempt=0
+
+  while (( attempt < 2 )); do
+    if (( attempt > 0 )); then
+      salvage_truncated_json "$stale_export" || true
+    fi
+    rm -f "$salvage_merged"
+    if merge_exports_auto "$destination_path" "$stale_export" "$salvage_merged" && [[ -s "$salvage_merged" ]]; then
+      if json_is_valid "$salvage_merged"; then
+        local before_count after_count
+        before_count=$(message_count_fast "$destination_path")
+        commit_merged_export "$destination_path" "$salvage_merged"
+        after_count=$(message_count_fast "$destination_path")
+        if (( after_count > before_count )); then
+          log "  SALVAGED $destination_path (+$((after_count - before_count)) messages from stale temp, $before_count → $after_count)"
+          return 0
+        fi
+        log "  Stale temp merged with no new messages, discarding: $stale_dir"
+        return 0
+      fi
+      log "  Stale temp merge produced invalid JSON, retaining for retry: $stale_dir"
+      return 1
+    fi
+    attempt=$((attempt + 1))
+  done
+
+  log "  Stale temp merge failed, retaining for retry: $stale_dir"
+  return 1
+}
+
 salvage_stale_temp_exports() {
   local output_dir=$1
   local channel_id=$2
   local destination_path=$3
 
-  local stale_dirs stale_dir stale_export salvage_merged
+  local stale_dirs stale_dir stale_export
   mapfile -t stale_dirs < <(
     find "$output_dir/.dce-temp" -maxdepth 1 -type d -name "export.${channel_id}.*" 2>/dev/null || true
   )
@@ -646,6 +696,11 @@ salvage_stale_temp_exports() {
     [[ -f "$stale_export" ]] || { rm -rf "$stale_dir"; continue; }
     [[ -s "$stale_export" ]] || { rm -rf "$stale_dir"; continue; }
 
+    if stale_temp_is_active "$stale_export"; then
+      log "  Stale temp still active (recently modified), skipping salvage: $stale_dir"
+      continue
+    fi
+
     if ! salvage_truncated_json "$stale_export"; then
       log "  Stale temp export unsalvageable, discarding: $stale_dir"
       rm -rf "$stale_dir"
@@ -668,25 +723,8 @@ salvage_stale_temp_exports() {
     fi
 
     if [[ -n "$destination_path" && -f "$destination_path" ]]; then
-      salvage_merged="$stale_dir/merged.json"
+      if merge_stale_export_into_destination "$destination_path" "$stale_export" "$stale_dir"; then
-      if merge_exports_auto "$destination_path" "$stale_export" "$salvage_merged" && [[ -s "$salvage_merged" ]]; then
+        merged_ok=1
-        if json_is_valid "$salvage_merged"; then
-          local before_count after_count
-          before_count=$(message_count_fast "$destination_path")
-          commit_merged_export "$destination_path" "$salvage_merged"
-          after_count=$(message_count_fast "$destination_path")
-          if (( after_count > before_count )); then
-            log "  SALVAGED $destination_path (+$((after_count - before_count)) messages from stale temp, $before_count → $after_count)"
-            merged_ok=1
-          else
-            log "  Stale temp merged with no new messages, discarding: $stale_dir"
-            merged_ok=1
-          fi
-        else
-          log "  Stale temp merge produced invalid JSON, retaining for retry: $stale_dir"
-        fi
-      else
-        log "  Stale temp merge failed, retaining for retry: $stale_dir"
       fi
     elif [[ -n "$destination_path" ]]; then
       mkdir -p "$(dirname "$destination_path")"

scripts/tests/run-discord-scrape-smoke.sh

@@ -150,6 +150,14 @@ cat >"$CONFIG_PATH" <<JSON
       "channel_ids": ["111"],
       "guild_ids": [],
       "guild_name_patterns": []
+    },
+    {
+      "name": "salvage-stale-active",
+      "kind": "guild",
+      "output_dir": "$ARCHIVE_ROOT/salvage-stale-active",
+      "channel_ids": ["111"],
+      "guild_ids": [],
+      "guild_name_patterns": []
     }
   ]
 }
@@ -417,12 +425,28 @@ grep -q 'exit 143' "$SKIP_SIGTERM_LOG" || { echo "expected sigterm exit logged f
 grep -q 'Preserving partial export temp' "$SKIP_SIGTERM_LOG" || { echo "expected partial temp preserved on sigterm channel 143" >&2; exit 1; }
 
 # Salvage stale temp export smoke
+mkdir -p "$ARCHIVE_ROOT/salvage-stale-active"
+cp "$FIXTURE_DIR/append-existing.json" "$ARCHIVE_ROOT/salvage-stale-active/$DEFAULT_FILE_NAME"
+mkdir -p "$ARCHIVE_ROOT/salvage-stale-active/.dce-meta"
+printf '{\"111\":\"%s\"}\n' "$ARCHIVE_ROOT/salvage-stale-active/$DEFAULT_FILE_NAME" >"$ARCHIVE_ROOT/salvage-stale-active/.dce-meta/channel-map.json"
+mkdir -p "$ARCHIVE_ROOT/salvage-stale-active/.dce-temp/export.111.ACTIVE"
+cp "$FIXTURE_DIR/salvage-truncated.json" "$ARCHIVE_ROOT/salvage-stale-active/.dce-temp/export.111.ACTIVE/export.json"
+SALVAGE_ACTIVE_LOG="$TMP_DIR/salvage-stale-active.log"
+DCE_STALE_TEMP_MIN_AGE_SECONDS=9999 \
+  run_wrapper salvage-stale-active append 2>"$SALVAGE_ACTIVE_LOG"
+grep -q 'still active' "$SALVAGE_ACTIVE_LOG" || { echo "expected active stale temp skip message" >&2; exit 1; }
+[[ -d "$ARCHIVE_ROOT/salvage-stale-active/.dce-temp/export.111.ACTIVE" ]] || {
+  echo "expected active stale temp dir to be retained" >&2
+  exit 1
+}
+
 mkdir -p "$ARCHIVE_ROOT/salvage-stale"
 cp "$FIXTURE_DIR/append-existing.json" "$ARCHIVE_ROOT/salvage-stale/$DEFAULT_FILE_NAME"
 mkdir -p "$ARCHIVE_ROOT/salvage-stale/.dce-meta"
 printf '{\"111\":\"%s\"}\n' "$ARCHIVE_ROOT/salvage-stale/$DEFAULT_FILE_NAME" >"$ARCHIVE_ROOT/salvage-stale/.dce-meta/channel-map.json"
 mkdir -p "$ARCHIVE_ROOT/salvage-stale/.dce-temp/export.111.STALE"
 cp "$FIXTURE_DIR/salvage-truncated.json" "$ARCHIVE_ROOT/salvage-stale/.dce-temp/export.111.STALE/export.json"
+touch -d '1 hour ago' "$ARCHIVE_ROOT/salvage-stale/.dce-temp/export.111.STALE/export.json"
 SALVAGE_LOG="$TMP_DIR/salvage-stale.log"
 run_wrapper salvage-stale append 2>"$SALVAGE_LOG"
 SALVAGE_DEST="$ARCHIVE_ROOT/salvage-stale/$DEFAULT_FILE_NAME"