修复双实例 owner-scoped 结果文件 job 被非归属实例误杀

线上现象：货源查询任务 26402/26400（归属 server-121）在归属机正常推进，
却被 server-110 判为「结果生成失败」，错误为「该任务已绑定到另一台服务实例处理」。

两个缺陷叠加：
1. 两台实例共用一个 RocketMQ consumer group，dispatch 消息被轮询投递；
   worker 对"PENDING + owner 不匹配"的 job 选择本地抢跑，模块层
   ensureTaskOwnedByCurrentInstance 直接抛 TaskOwnerMismatchException，
   异常落进通用 catch 被当成生成失败，5 次抢跑即耗尽 retryCount 并终态失败。
2. runPendingJobs 双重 claim：claimRunnableJobsForOwner 已把 job 翻成 RUNNING，
   随后又走 claimRunning（条件 PENDING/FAILED）必然失败 → job 被静默丢弃，
   归属机的定时兜底从未真正生效。

修复：
- worker：非归属实例一律跳过 owner-scoped job；抽出 processClaimed 供已持有
  claim 的调用方使用；processInternal 单独捕获 TaskOwnerMismatchException，
  只 deferRunning 退回 PENDING，不计入重试。
- TaskFileJobLocalDispatcher 新增 dispatchClaimed。
- dispatch 事件带上 owner，coordinator 按归属路由：本机 owner 直接本地派发不发 MQ，
  他机 owner 不发不跑交由归属机轮询接手，owner 为空维持原有 MQ 行为。
- 新增 owner 路由回归测试两组，回滚配合"本地抢跑"的旧断言。

同时提交此前工作区内已随 JAR 上线的 backend-java 改动：SimilarAsin 解析载荷
groups 内嵌 items（Python 旧链路兼容）与 chunk 类型不匹配跳过、
PermissionMenuSchemaInitializer 与 V103 两级菜单分组迁移、相关测试。

backend-java/src/main/java/com/nanri/aiimage/modules/permission/service/PermissionMenuSchemaInitializer.java

@@ -47,20 +47,34 @@ public class PermissionMenuSchemaInitializer {
             new DefaultAppChildMenu("ERP", "erp", "erp", 132, "brand_logistics_tools")
     );
 
+    /**
+     * 后台一级分组：只做权限树层级与「上级菜单」候选，不映射真实页面。
+     * route_path 使用 group- 前缀虚拟值以满足 (menu_type, route_path) 唯一约束；
+     * 前端按 column_key 前缀 admin_group_ 识别分组行，不会出现在页面选择器中。
+     */
+    private static final List<DefaultAdminGroup> DEFAULT_ADMIN_GROUPS = List.of(
+            new DefaultAdminGroup("账号与权限", "admin_group_account", "group-account", 5),
+            new DefaultAdminGroup("数据管理", "admin_group_data", "group-data", 15),
+            new DefaultAdminGroup("店铺管理", "admin_group_shop", "group-shop", 45),
+            new DefaultAdminGroup("记录与版本", "admin_group_record", "group-record", 55)
+    );
+
     private static final List<DefaultAdminMenu> DEFAULT_ADMIN_MENUS = List.of(
-            new DefaultAdminMenu("用户管理", "admin_users", "users", 10),
-            new DefaultAdminMenu("菜单权限配置", "admin_columns", "columns", 20),
-            new DefaultAdminMenu("分组管理", "admin_group_manage", "group-manage", 25),
-            new DefaultAdminMenu("去重数据汇总", "admin_dedupe_total_data", "dedupe-total-data", 30),
-            new DefaultAdminMenu("无效ASIN数据", "admin_invalid_asin_data", "invalid-asin-data", 35),
-            new DefaultAdminMenu("店铺密钥管理", "admin_shop_keys", "shop-keys", 40),
-            new DefaultAdminMenu("店铺管理", "admin_shop_manage", "shop-manage", 50),
-            new DefaultAdminMenu("跳过跟价ASIN", "admin_skip_price_asin", "skip-price-asin", 60),
-            new DefaultAdminMenu("查询ASIN", "admin_query_asin", "query-asin", 65),
-            new DefaultAdminMenu("商品类目", "admin_product_categories", "product-categories", 66),
-            new DefaultAdminMenu("生成记录", "admin_history", "history", 70),
-            new DefaultAdminMenu("软件版本管理", "admin_version", "version", 80),
-            new DefaultAdminMenu("数字人版本管理", "digital_human_version", "digital-human-version", 81)
+            new DefaultAdminMenu("用户管理", "admin_users", "users", 10, "admin_group_account"),
+            new DefaultAdminMenu("菜单权限配置", "admin_columns", "columns", 20, "admin_group_account"),
+            new DefaultAdminMenu("分组管理", "admin_group_manage", "group-manage", 25, "admin_group_account"),
+            new DefaultAdminMenu("去重数据汇总", "admin_dedupe_total_data", "dedupe-total-data", 30, "admin_group_data"),
+            new DefaultAdminMenu("品牌数据库", "admin_invalid_asin_data", "invalid-asin-data", 35, "admin_group_data"),
+            new DefaultAdminMenu("查询ASIN", "admin_query_asin", "query-asin", 65, "admin_group_data"),
+            new DefaultAdminMenu("商品类目", "admin_product_categories", "product-categories", 66, "admin_group_data"),
+            new DefaultAdminMenu("店铺密钥管理", "admin_shop_keys", "shop-keys", 40, "admin_group_shop"),
+            new DefaultAdminMenu("店铺管理", "admin_shop_manage", "shop-manage", 50, "admin_group_shop"),
+            new DefaultAdminMenu("最低价ASIN设置", "admin_skip_price_asin", "skip-price-asin", 60, "admin_group_shop"),
+            new DefaultAdminMenu("店铺数据记录", "admin_shop_data_crawl_tasks", "shop-data-crawl-tasks", 82, "admin_group_shop"),
+            new DefaultAdminMenu("生成记录", "admin_history", "history", 70, "admin_group_record"),
+            new DefaultAdminMenu("视频任务记录", "admin_image_video_tasks", "image-video-tasks", 75, "admin_group_record"),
+            new DefaultAdminMenu("软件版本管理", "admin_version", "version", 80, "admin_group_record"),
+            new DefaultAdminMenu("数字人版本管理", "digital_human_version", "digital-human-version", 81, "admin_group_record")
     );
 
     @EventListener(ApplicationReadyEvent.class)
@@ -86,6 +100,7 @@ public class PermissionMenuSchemaInitializer {
         executeQuietly("UPDATE columns SET menu_type = 'app' WHERE menu_type IS NULL OR menu_type = ''");
         executeQuietly("UPDATE columns SET sort_order = id WHERE sort_order IS NULL OR sort_order = 0");
         executeQuietly("ALTER TABLE columns ADD UNIQUE KEY uk_menu_type_route_path (menu_type, route_path)");
+        ensureDefaultAdminGroups();
         ensureDefaultAdminMenus();
         boolean shopDataCrawlDataPermissionCreated = ensureShopDataCrawlAdminDefaults();
         ensureDefaultAppMenus();
@@ -105,20 +120,37 @@ public class PermissionMenuSchemaInitializer {
         }
     }
 
-    private void ensureDefaultAdminMenus() {
-        for (DefaultAdminMenu menu : DEFAULT_ADMIN_MENUS) {
+    private void ensureDefaultAdminGroups() {
+        for (DefaultAdminGroup group : DEFAULT_ADMIN_GROUPS) {
             executeQuietly("""
-                    INSERT INTO columns (name, column_key, menu_type, route_path, sort_order)
-                    SELECT '%s', '%s', 'admin', '%s', %d
+                    INSERT INTO columns (name, column_key, menu_type, route_path, parent_id, sort_order)
+                    SELECT '%s', '%s', 'admin', '%s', NULL, %d
                     WHERE NOT EXISTS (
                         SELECT 1 FROM columns WHERE column_key = '%s'
                     )
-                    """.formatted(menu.name(), menu.columnKey(), menu.routePath(), menu.sortOrder(), menu.columnKey()));
+                    """.formatted(group.name(), group.columnKey(), group.routePath(), group.sortOrder(), group.columnKey()));
+        }
+    }
+
+    private void ensureDefaultAdminMenus() {
+        for (DefaultAdminMenu menu : DEFAULT_ADMIN_MENUS) {
             executeQuietly("""
-                    UPDATE columns
-                    SET sort_order = %d, parent_id = NULL
-                    WHERE column_key = '%s' AND (sort_order IS NULL OR sort_order = 0)
-                    """.formatted(menu.sortOrder(), menu.columnKey()));
+                    INSERT INTO columns (name, column_key, menu_type, route_path, sort_order, parent_id)
+                    SELECT '%s', '%s', 'admin', '%s', %d, parent.id
+                    FROM columns parent
+                    WHERE parent.column_key = '%s'
+                      AND NOT EXISTS (
+                          SELECT 1 FROM columns WHERE column_key = '%s'
+                      )
+                    """.formatted(menu.name(), menu.columnKey(), menu.routePath(), menu.sortOrder(),
+                    menu.parentKey(), menu.columnKey()));
+            executeQuietly("""
+                    UPDATE columns child
+                    JOIN columns parent ON parent.column_key = '%s'
+                    SET child.name = '%s', child.sort_order = %d,
+                        child.parent_id = COALESCE(child.parent_id, parent.id)
+                    WHERE child.column_key = '%s'
+                    """.formatted(menu.parentKey(), menu.name(), menu.sortOrder(), menu.columnKey()));
         }
     }
 
@@ -179,8 +211,9 @@ public class PermissionMenuSchemaInitializer {
 
     private boolean ensureShopDataCrawlAdminDefaults() {
         executeQuietly("""
-                INSERT INTO columns (name, column_key, menu_type, route_path, sort_order)
-                SELECT '店铺数据记录', 'admin_shop_data_crawl_tasks', 'admin', 'shop-data-crawl-tasks', 82
+                INSERT INTO columns (name, column_key, menu_type, route_path, sort_order, parent_id)
+                SELECT '店铺数据记录', 'admin_shop_data_crawl_tasks', 'admin', 'shop-data-crawl-tasks', 82,
+                       (SELECT id FROM columns WHERE column_key = 'admin_group_shop' LIMIT 1)
                 WHERE NOT EXISTS (
                     SELECT 1 FROM columns WHERE column_key = 'admin_shop_data_crawl_tasks'
                 )
@@ -226,7 +259,11 @@ public class PermissionMenuSchemaInitializer {
         }
     }
 
-    private record DefaultAdminMenu(String name, String columnKey, String routePath, int sortOrder) {
+    private record DefaultAdminMenu(String name, String columnKey, String routePath, int sortOrder,
+                                    String parentKey) {
+    }
+
+    private record DefaultAdminGroup(String name, String columnKey, String routePath, int sortOrder) {
     }
 
     private record DefaultAppMenu(String name, String columnKey, String routePath, int sortOrder) {

backend-java/src/main/java/com/nanri/aiimage/modules/similarasin/service/SimilarAsinTaskService.java

@@ -5127,10 +5127,47 @@ public class SimilarAsinTaskService {
         payload.setSourceFiles(sourceFiles == null ? List.of() : sourceFiles);
         payload.setHeaders(headers == null ? List.of() : headers);
         payload.setItems(allRows == null ? List.of() : new ArrayList<>(allRows));
-        payload.setGroups(groups == null ? List.of() : groups);
+        // 兼容 Python 旧链路：groups 需内嵌完整行（Python 端 normalize_groups
+        // 直接返回 groups 并按 gp["items"] 取行；仅有 startIndex/endIndex 引用时
+        // items 为空、任务空转）。startIndex/endIndex 保留，Java 内部 expandGroupRefs 继续使用。
+        payload.setGroups(fillGroupItems(groups, allRows));
         return writeJson(payload, "保存解析结果失败");
     }
 
+    /**
+     * 按 group.startIndex/endIndex 半开区间从全量行切片，内嵌回 group.items；
+     * 无索引或切片越界时退回按 group.items 原样（兼容旧 payload 反序列化场景）。
+     */
+    static List<SimilarAsinParsedGroupVo> fillGroupItems(List<SimilarAsinParsedGroupVo> groups, List<SimilarAsinParsedRowVo> allRows) {
+        if (groups == null || groups.isEmpty()) {
+            return groups == null ? List.of() : groups;
+        }
+        if (allRows == null || allRows.isEmpty()) {
+            return groups;
+        }
+        List<SimilarAsinParsedRowVo> rows = new ArrayList<>(allRows);
+        for (SimilarAsinParsedGroupVo group : groups) {
+            if (group == null) {
+                continue;
+            }
+            Integer start = group.getStartIndex();
+            Integer end = group.getEndIndex();
+            if (start != null && end != null) {
+                int from = Math.max(0, Math.min(start, rows.size()));
+                int to = Math.max(from, Math.min(end, rows.size()));
+                if (to > from) {
+                    group.setItems(new ArrayList<>(rows.subList(from, to)));
+                    continue;
+                }
+            }
+            // 索引缺失/无效：保留已有 items（兼容旧 payload），不覆盖
+            if (group.getItems() == null) {
+                group.setItems(new ArrayList<>());
+            }
+        }
+        return groups;
+    }
+
     private String buildTaskResultJson(String aiPrompt, String apiKey, Boolean imgSwitch, Boolean categorySwitch, List<SimilarAsinSourceFileDto> sourceFiles, String parsedPayloadPointer) {
         Map<String, Object> payload = new LinkedHashMap<>();
         payload.put("aiPrompt", normalize(aiPrompt));
@@ -5172,6 +5209,11 @@ public class SimilarAsinTaskService {
     }
 
     private SimilarAsinParsedPayloadDto hydrateParsedPayloadRows(SimilarAsinParsedPayloadDto payload) {
+        return hydrateForPythonCompat(payload);
+    }
+
+    /** 解析载荷 hydrate：回填 allItems/items/groups 内嵌 items（Python 旧链路兼容）。 */
+    static SimilarAsinParsedPayloadDto hydrateForPythonCompat(SimilarAsinParsedPayloadDto payload) {
         if (payload == null) {
             return new SimilarAsinParsedPayloadDto();
         }
@@ -5182,6 +5224,13 @@ public class SimilarAsinTaskService {
         if (payload.getItems() == null || payload.getItems().isEmpty()) {
             payload.setItems(rows);
         }
+        // Python 旧链路兼容：读取时也回填 groups 内嵌 items（存量任务 payload
+        // 只有引用），否则 Python normalize_groups 拿到空 items 任务空转。
+        if (payload.getGroups() != null && !payload.getGroups().isEmpty()) {
+            List<SimilarAsinParsedRowVo> filledRows = payload.getItems() == null || payload.getItems().isEmpty()
+                    ? rows : payload.getItems();
+            payload.setGroups(fillGroupItems(payload.getGroups(), filledRows));
+        }
         return payload;
     }
 
@@ -5205,6 +5254,16 @@ public class SimilarAsinTaskService {
             // P1-4：识别跨实例 local 指针读不到的场景，把"chunk 在另一实例"的元信息
             // 通过 task_scope_state.last_error 留痕，便于排查"为何 owner 切换后 chunk 读不到"。
             boolean crossInstance = msg.contains("only exists on instance=");
+            // 类型不兼容（如历史版本序列化数据含当前类加载器不认识的类型标记）：
+            // 该 chunk 永久不可读，跳过而不是把组装/收尾永久拖死，留痕便于排查。
+            boolean typeMismatch = ex instanceof TypeNotPresentException
+                    || msg.contains("not present");
+            if (typeMismatch) {
+                log.warn("[similar-asin] chunk payload type mismatch, skip chunk taskId={} chunk={} err={}",
+                        chunk.getTaskId(), chunk.getChunkIndex(), msg);
+                recordChunkReadFailure(chunk, false, msg);
+                return rows;
+            }
             log.warn("[similar-asin] read chunk payload failed taskId={} chunk={} crossInstance={} err={}",
                     chunk.getTaskId(), chunk.getChunkIndex(), crossInstance, msg);
             recordChunkReadFailure(chunk, crossInstance, msg);

backend-java/src/main/java/com/nanri/aiimage/modules/task/model/dto/TaskFileJobDispatchEvent.java

@@ -9,6 +9,7 @@ public record TaskFileJobDispatchEvent(
         Long resultId,
         String scopeKey,
         String jobType,
+        String owner,
         LocalDateTime createdAt
 ) {
 }

backend-java/src/main/java/com/nanri/aiimage/modules/task/service/TaskFileJobDispatchCoordinator.java

@@ -1,5 +1,6 @@
 package com.nanri.aiimage.modules.task.service;
 
+import com.nanri.aiimage.config.InstanceMetadata;
 import com.nanri.aiimage.modules.task.model.dto.TaskFileJobDispatchEvent;
 import com.nanri.aiimage.modules.task.model.dto.TaskFileJobMessage;
 import lombok.RequiredArgsConstructor;
@@ -17,6 +18,7 @@ public class TaskFileJobDispatchCoordinator {
 
     private final TaskFileJobPublisher taskFileJobPublisher;
     private final TaskFileJobLocalDispatcher taskFileJobLocalDispatcher;
+    private final InstanceMetadata instanceMetadata;
     @Qualifier("taskFileJobDispatchExecutor")
     private final TaskExecutor taskFileJobDispatchExecutor;
 
@@ -43,6 +45,28 @@ public class TaskFileJobDispatchCoordinator {
     }
 
     private void dispatchAfterCommit(TaskFileJobDispatchEvent event, TaskFileJobMessage message) {
+        // owner 非空 ⇔ owner-scoped job（只有 SIMILAR_ASIN/APPEARANCE_PATENT/PUBLISH/
+        // SHOP_DATA_CRAWL 会写 "task:{id}:owner:{instance}" 形式的 scopeKey）。
+        // 这类 job 必须在归属实例上执行，走共享 consumer group 的 MQ 会被轮询投递到
+        // 另一台，那边只能跳过，白白多绕一圈。
+        String owner = event.owner() == null ? "" : event.owner().trim();
+        if (!owner.isBlank()) {
+            if (owner.equals(currentInstanceId())) {
+                boolean dispatched = taskFileJobLocalDispatcher.dispatch(
+                        event.jobId(), event.taskId(), event.moduleType(), true);
+                if (dispatched) {
+                    return;
+                }
+                log.warn("[task-file-job] owner-scoped local dispatch declined, fallback to mq jobId={} taskId={} moduleType={} owner={}",
+                        event.jobId(), event.taskId(), event.moduleType(), owner);
+            } else {
+                // 归属实例每 15 秒扫一次自己 owner 的 PENDING/FAILED job，这里不发 MQ、
+                // 不本地跑，避免消息落到非归属实例后被丢弃。
+                log.info("[task-file-job] skip dispatch because job belongs to another instance jobId={} taskId={} moduleType={} owner={} current={}",
+                        event.jobId(), event.taskId(), event.moduleType(), owner, currentInstanceId());
+                return;
+            }
+        }
         boolean published = taskFileJobPublisher.publish(message);
         if (published) {
             return;
@@ -56,4 +80,9 @@ public class TaskFileJobDispatchCoordinator {
                     event.jobId(), event.taskId(), event.moduleType());
         }
     }
+
+    private String currentInstanceId() {
+        String instanceId = instanceMetadata == null ? null : instanceMetadata.getInstanceId();
+        return instanceId == null || instanceId.isBlank() ? "unknown-instance" : instanceId;
+    }
 }

backend-java/src/main/java/com/nanri/aiimage/modules/task/service/TaskFileJobLocalDispatcher.java

@@ -42,7 +42,19 @@ public class TaskFileJobLocalDispatcher {
         return dispatch(jobId, taskId, moduleType, false);
     }
 
+    /**
+     * 调用方已把 job claim 成 RUNNING 时使用：复用同样的背压/去重/异步执行，
+     * 但落到 worker.processClaimed，避免二次 claimRunning 失败导致 job 被静默丢弃。
+     */
+    public boolean dispatchClaimed(Long jobId, Long taskId, String moduleType) {
+        return submit(jobId, taskId, moduleType, false, true);
+    }
+
     public boolean dispatch(Long jobId, Long taskId, String moduleType, boolean force) {
+        return submit(jobId, taskId, moduleType, force, false);
+    }
+
+    private boolean submit(Long jobId, Long taskId, String moduleType, boolean force, boolean alreadyClaimed) {
         if (jobId == null || jobId <= 0) {
             return false;
         }
@@ -63,7 +75,7 @@ public class TaskFileJobLocalDispatcher {
             return true;
         }
         try {
-            taskFileJobDispatchExecutor.execute(() -> processLocally(jobId, taskId, moduleType));
+            taskFileJobDispatchExecutor.execute(() -> processLocally(jobId, taskId, moduleType, alreadyClaimed));
             return true;
         } catch (RuntimeException ex) {
             inflightJobIds.remove(jobId);
@@ -71,7 +83,7 @@ public class TaskFileJobLocalDispatcher {
                     jobId, taskId, moduleType, ex.getMessage(), ex);
             if (force) {
                 try {
-                    processLocally(jobId, taskId, moduleType);
+                    processLocally(jobId, taskId, moduleType, alreadyClaimed);
                     return true;
                 } finally {
                     inflightJobIds.remove(jobId);
@@ -81,7 +93,7 @@ public class TaskFileJobLocalDispatcher {
         }
     }
 
-    private void processLocally(Long jobId, Long taskId, String moduleType) {
+    private void processLocally(Long jobId, Long taskId, String moduleType, boolean alreadyClaimed) {
         try {
             TaskFileJobEntity job = taskFileJobMapper.selectById(jobId);
             if (job == null) {
@@ -95,7 +107,11 @@ public class TaskFileJobLocalDispatcher {
                         jobId, taskId, moduleType);
                 return;
             }
-            taskResultFileJobWorker.process(job);
+            if (alreadyClaimed) {
+                taskResultFileJobWorker.processClaimed(job);
+            } else {
+                taskResultFileJobWorker.process(job);
+            }
         } catch (Exception ex) {
             log.warn("[task-file-job] local dispatch failed jobId={} taskId={} moduleType={} msg={}",
                     jobId, taskId, moduleType, ex.getMessage(), ex);

backend-java/src/main/java/com/nanri/aiimage/modules/task/service/TaskFileJobService.java

@@ -57,7 +57,14 @@ public class TaskFileJobService {
         if (existing == null) {
             return null;
         }
-        if ("SUCCESS".equals(existing.getStatus()) || "RUNNING".equals(existing.getStatus()) || "PENDING".equals(existing.getStatus())) {
+        if ("SUCCESS".equals(existing.getStatus()) || "RUNNING".equals(existing.getStatus())) {
+            return existing;
+        }
+        if ("PENDING".equals(existing.getStatus())) {
+            // PENDING 僵尸兜底：job 入队时若 MQ 发送失败且本地兜底未生效（或消息被
+            // 非 owner 实例静默吞掉），job 会永远停留在 PENDING、无人处理。
+            // 补发一次 dispatch 事件，让调度链重试（幂等：consumer 侧 claim 有状态翻转保护）。
+            publishDispatchEvent(existing);
             return existing;
         }
         // 终态失败：retryCount 已达上限的 FAILED job 不再重置为 PENDING，
@@ -237,6 +244,13 @@ public class TaskFileJobService {
                                         .lt(TaskFileJobEntity::getUpdatedAt, threshold)
                                         .or()
                                         .isNull(TaskFileJobEntity::getUpdatedAt)))
+                        .or(stuckPending -> stuckPending
+                                .eq(TaskFileJobEntity::getStatus, "PENDING")
+                                .ne(TaskFileJobEntity::getRetryCount, MAX_RETRY_COUNT)
+                                .and(age -> age
+                                        .lt(TaskFileJobEntity::getUpdatedAt, threshold)
+                                        .or()
+                                        .isNull(TaskFileJobEntity::getUpdatedAt)))
                         .or(exhausted -> exhausted
                                 .eq(TaskFileJobEntity::getStatus, "FAILED")
                                 .ge(TaskFileJobEntity::getRetryCount, MAX_RETRY_COUNT)
@@ -251,6 +265,24 @@ public class TaskFileJobService {
                 exhaustedJobs.add(job);
                 continue;
             }
+            if ("PENDING".equals(job.getStatus())) {
+                // PENDING 僵尸兜底：补发 dispatch 事件（worker claim 幂等），
+                // 并累加 retryCount 防无限重发；注意不翻转状态，
+                // 避免与真实 PENDING 竞争 worker 的 claim 条件更新冲突。
+                int retryCount = job.getRetryCount() == null ? 0 : job.getRetryCount();
+                int nextRetryCount = Math.min(MAX_RETRY_COUNT - 1, retryCount + 1);
+                int updated = taskFileJobMapper.update(null, new LambdaUpdateWrapper<TaskFileJobEntity>()
+                        .eq(TaskFileJobEntity::getId, job.getId())
+                        .eq(TaskFileJobEntity::getStatus, "PENDING")
+                        .eq(job.getUpdatedAt() != null, TaskFileJobEntity::getUpdatedAt, job.getUpdatedAt())
+                        .isNull(job.getUpdatedAt() == null, TaskFileJobEntity::getUpdatedAt)
+                        .set(TaskFileJobEntity::getRetryCount, nextRetryCount)
+                        .set(TaskFileJobEntity::getUpdatedAt, LocalDateTime.now()));
+                if (updated > 0) {
+                    publishDispatchEvent(taskFileJobMapper.selectById(job.getId()));
+                }
+                continue;
+            }
             int retryCount = job.getRetryCount() == null ? 0 : job.getRetryCount();
             int nextRetryCount = Math.min(MAX_RETRY_COUNT, retryCount + 1);
             LocalDateTime now = LocalDateTime.now();
@@ -642,6 +674,7 @@ public class TaskFileJobService {
                 entity.getResultId(),
                 entity.getScopeKey(),
                 entity.getJobType(),
+                entity.getOwner(),
                 LocalDateTime.now()
         ));
     }

backend-java/src/main/java/com/nanri/aiimage/modules/task/service/TaskResultFileJobWorker.java

@@ -1,6 +1,7 @@
 package com.nanri.aiimage.modules.task.service;
 
 import com.nanri.aiimage.common.exception.BusinessException;
+import com.nanri.aiimage.common.exception.TaskOwnerMismatchException;
 import com.nanri.aiimage.modules.appearancepatent.service.AppearancePatentTaskService;
 import com.nanri.aiimage.config.InstanceMetadata;
 import com.nanri.aiimage.modules.brand.service.BrandTaskService;
@@ -76,13 +77,16 @@ public class TaskResultFileJobWorker {
         }
         log.info("[task-file-job] scheduled worker claimed jobs count={}", jobs.size());
         for (TaskFileJobEntity job : jobs) {
-            boolean dispatched = taskFileJobLocalDispatcher.dispatch(
+            // claimRunnableJobsForOwner 已把 job 原子翻转成 RUNNING，这里必须走"已 claim"
+            // 通道：若再走 process() 的 claimRunning，条件 UPDATE 会因状态已是 RUNNING 而
+            // 失败，job 被静默丢弃并挂到 stuck 扫描才回收，定时兜底等于失效。
+            boolean dispatched = taskFileJobLocalDispatcher.dispatchClaimed(
                     job.getId(),
                     job.getTaskId(),
                     job.getModuleType()
             );
             if (!dispatched) {
-                process(job);
+                processClaimed(job);
             }
         }
     }
@@ -104,18 +108,39 @@ public class TaskResultFileJobWorker {
     }
 
     public void process(TaskFileJobEntity job) {
-        if (isOwnerScopedJob(job) && !isOwnedByCurrentInstance(job)) {
-            log.debug("[task-file-job] skip owner-scoped job because owner is another instance jobId={} taskId={} moduleType={} owner={} current={}",
-                    job.getId(), job.getTaskId(), job.getModuleType(), ownerFromScopeKey(job.getScopeKey()), currentInstanceId());
+        if (job == null || job.getId() == null) {
             return;
         }
-        if (job == null || job.getId() == null) {
+        if (isOwnerScopedJob(job) && !isOwnedByCurrentInstance(job)) {
+            // 非归属实例一律不碰 owner-scoped job：模块层 ensureTaskOwnedByCurrentInstance
+            // 会直接抛 TaskOwnerMismatchException，抢跑只会白白消耗 job 的重试次数，
+            // 重试耗尽后把归属机上跑得好好的任务判成失败。
+            // job 不会因此卡死：归属实例的 runPendingJobs 每 15 秒扫一次自己 owner 的
+            // PENDING/FAILED job，enqueueAssembleResult / stuck 扫描也会补发 dispatch。
+            log.info("[task-file-job] skip owner-scoped job because owner is another instance jobId={} taskId={} moduleType={} status={} owner={} current={}",
+                    job.getId(), job.getTaskId(), job.getModuleType(), job.getStatus(),
+                    ownerFromScopeKey(job.getScopeKey()), currentInstanceId());
             return;
         }
         TaskFileJobEntity claim = taskFileJobService.claimRunning(job.getId());
         if (claim == null) {
             return;
         }
+        offloadOrRun(job, claim);
+    }
+
+    /**
+     * 供"调用方已持有 RUNNING claim"的路径使用（定时 worker 的 claimRunnableJobsForOwner）：
+     * 跳过 claimRunning，直接按 handler 能力异步/内联执行。
+     */
+    public void processClaimed(TaskFileJobEntity claim) {
+        if (claim == null || claim.getId() == null) {
+            return;
+        }
+        offloadOrRun(claim, claim);
+    }
+
+    private void offloadOrRun(TaskFileJobEntity job, TaskFileJobEntity claim) {
         ResultFileJobHandler handler = handlerRegistry.asMap().get(job.getModuleType());
         if (handler != null && handler.supportsAsyncOffload()) {
             try {
@@ -206,6 +231,12 @@ public class TaskResultFileJobWorker {
             if (finalizeWithdraw) {
                 withdrawTaskService.tryFinalizeTask(job.getTaskId(), false);
             }
+        } catch (TaskOwnerMismatchException ex) {
+            // 归属校验失败不是"生成失败"：不能计入 retryCount，否则任务会在归属机
+            // 一切正常的情况下被本机重试耗尽判死。退回 PENDING 交还归属实例。
+            log.warn("[task-file-job] process rejected because task belongs to another instance jobId={} taskId={} moduleType={} owner={} current={}",
+                    job.getId(), job.getTaskId(), job.getModuleType(), ex.getOwnerInstanceId(), ex.getCurrentInstanceId());
+            taskFileJobService.deferRunning(job.getId(), ex.getMessage());
         } catch (Exception ex) {
             TaskFileJobEntity latest = taskFileJobService.findById(job.getId());
             if (latest != null && "SUCCESS".equals(latest.getStatus())) {

backend-java/src/main/resources/db/V103__admin_menu_two_level_groups.sql

@@ -0,0 +1,63 @@
+-- V103: 后台管理菜单改为两级结构（一级分组 + 子页面）
+-- 新增 4 个后台一级分组（不映射真实页面，route_path 为分组虚拟值），
+-- 并把这 16 个后台页面菜单挂到对应分组下。
+-- 分组只用于权限树层级与「上级菜单」候选；column_key / route_path 校验逻辑不变。
+-- 对应 Java 端清单见 PermissionMenuSchemaInitializer.DEFAULT_ADMIN_GROUPS / DEFAULT_ADMIN_MENUS。
+
+-- ---------- 1. 新增一级分组 ----------
+
+INSERT INTO `columns` (`name`, `column_key`, `menu_type`, `route_path`, `parent_id`, `sort_order`)
+SELECT '账号与权限', 'admin_group_account', 'admin', 'group-account', NULL, 5
+WHERE NOT EXISTS (SELECT 1 FROM `columns` WHERE `column_key` = 'admin_group_account');
+
+INSERT INTO `columns` (`name`, `column_key`, `menu_type`, `route_path`, `parent_id`, `sort_order`)
+SELECT '数据管理', 'admin_group_data', 'admin', 'group-data', NULL, 15
+WHERE NOT EXISTS (SELECT 1 FROM `columns` WHERE `column_key` = 'admin_group_data');
+
+INSERT INTO `columns` (`name`, `column_key`, `menu_type`, `route_path`, `parent_id`, `sort_order`)
+SELECT '店铺管理', 'admin_group_shop', 'admin', 'group-shop', NULL, 45
+WHERE NOT EXISTS (SELECT 1 FROM `columns` WHERE `column_key` = 'admin_group_shop');
+
+INSERT INTO `columns` (`name`, `column_key`, `menu_type`, `route_path`, `parent_id`, `sort_order`)
+SELECT '记录与版本', 'admin_group_record', 'admin', 'group-record', NULL, 55
+WHERE NOT EXISTS (SELECT 1 FROM `columns` WHERE `column_key` = 'admin_group_record');
+
+-- ---------- 2. 页面菜单挂到分组下，并同步显示名 ----------
+-- 只更新当前没有父级的页面（保留管理员手工调整过的层级），
+-- COALESCE 保证已挂接的叶子不重复指向。
+
+UPDATE `columns` child
+JOIN `columns` parent ON parent.`column_key` = 'admin_group_account'
+SET child.`parent_id` = COALESCE(child.`parent_id`, parent.`id`)
+WHERE child.`column_key` IN ('admin_users', 'admin_columns', 'admin_group_manage');
+
+UPDATE `columns` child
+JOIN `columns` parent ON parent.`column_key` = 'admin_group_data'
+SET child.`parent_id` = COALESCE(child.`parent_id`, parent.`id`)
+WHERE child.`column_key` IN ('admin_dedupe_total_data', 'admin_invalid_asin_data',
+                             'admin_query_asin', 'admin_product_categories');
+
+UPDATE `columns` child
+JOIN `columns` parent ON parent.`column_key` = 'admin_group_shop'
+SET child.`parent_id` = COALESCE(child.`parent_id`, parent.`id`)
+WHERE child.`column_key` IN ('admin_shop_keys', 'admin_shop_manage', 'admin_skip_price_asin',
+                             'admin_shop_data_crawl_tasks');
+
+UPDATE `columns` child
+JOIN `columns` parent ON parent.`column_key` = 'admin_group_record'
+SET child.`parent_id` = COALESCE(child.`parent_id`, parent.`id`)
+WHERE child.`column_key` IN ('admin_history', 'admin_version', 'digital_human_version',
+                             'admin_image_video_tasks');
+
+-- ---------- 3. 显示名对齐前端界面 ---
+UPDATE `columns` SET `name` = '品牌数据库'
+WHERE `menu_type` = 'admin' AND `column_key` = 'admin_invalid_asin_data';
+
+UPDATE `columns` SET `name` = '最低价ASIN设置'
+WHERE `menu_type` = 'admin' AND `column_key` = 'admin_skip_price_asin';
+
+UPDATE `columns` SET `name` = '视频任务记录'
+WHERE `menu_type` = 'admin' AND `column_key` = 'admin_image_video_tasks';
+
+UPDATE `columns` SET `name` = '店铺数据记录'
+WHERE `menu_type` = 'admin' AND `column_key` = 'admin_shop_data_crawl_tasks';

backend-java/src/test/java/com/nanri/aiimage/config/ExplainIndexAuditDocTest.java

@@ -106,8 +106,17 @@ class ExplainIndexAuditDocTest {
         assertTrue(audit.contains("不做 DDL") || audit.contains("本任务无 DDL")
                         || audit.contains("不新增索引") || audit.contains("只审计"),
                 "审计文档明确本任务不做 DDL");
-        assertFalse(audit.contains("ALTER TABLE"),
-                "审计文档不包含 ALTER TABLE（DDL 属于后续任务 116）");
+        // 任务 116 落地 V101 索引后，审计文档会把 ALTER TABLE 作为候选索引的
+        // 回滚说明引用。审计任务自身依然不执行 DDL——判定口径改为「ALTER TABLE
+        // 只能出现在候选/回滚等说明性上下文」，而不是文档里不得出现该字符串。
+        for (String line : audit.split("\r?\n")) {
+            if (!line.contains("ALTER TABLE")) {
+                continue;
+            }
+            assertTrue(line.contains("回滚") || line.contains("候选") || line.contains("建议")
+                            || line.contains("风险") || line.trim().startsWith(">"),
+                    "审计文档中的 ALTER TABLE 必须是候选/回滚说明，实际行：" + line);
+        }
         assertTrue(spec.contains("只读") || spec.contains("巡检"),
                 "12 spec 巡检报表模式与审计文档对应");
     }

backend-java/src/test/java/com/nanri/aiimage/metrics/ExternalCallMetricsRecorderTest.java

@@ -27,6 +27,7 @@ import java.util.List;
 import java.util.Map;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
+import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.concurrent.atomic.AtomicInteger;
 
@@ -268,6 +269,7 @@ class ExternalCallMetricsRecorderTest {
             }
         } finally {
             pool.shutdown();
+            assertTrue(pool.awaitTermination(60, TimeUnit.SECONDS), "并发任务在超时前全部结束");
         }
 
         awaitMetric("aiimage.external-call.total", "client", "llm", "result", "success");
@@ -275,6 +277,12 @@ class ExternalCallMetricsRecorderTest {
             Thread.sleep(10);
         }
         assertEquals(20, llmSubmitCount.get(), "并发 20 请求全部真实发出");
+        // 指标在响应处理完成之后才落，submit 计数到 20 不代表指标已收敛：
+        // 直接断言会读到中间值（实测 18.0）。等指标计数自身收敛再断言。
+        for (int i = 0; i < 2000 && counterCount("aiimage.external-call.total",
+                "client", "llm", "result", "success") < 20.0; i++) {
+            Thread.sleep(10);
+        }
         assertEquals(20.0, counterCount("aiimage.external-call.total",
                 "client", "llm", "result", "success"), "20 次成功全部记录，无重复");
     }

backend-java/src/test/java/com/nanri/aiimage/modules/collectdata/service/CollectDataExcelCleanupTest.java

@@ -7,6 +7,7 @@ import com.nanri.aiimage.modules.collectdata.model.vo.CollectDataResultRowVo;
 import org.apache.poi.ss.usermodel.Sheet;
 import org.apache.poi.ss.usermodel.Workbook;
 import org.apache.poi.ss.usermodel.WorkbookFactory;
+import org.apache.poi.util.DefaultTempFileCreationStrategy;
 import org.apache.poi.util.TempFile;
 import org.apache.poi.util.TempFileCreationStrategy;
 import org.junit.jupiter.api.AfterEach;
@@ -39,20 +40,36 @@ class CollectDataExcelCleanupTest {
 
     private final List<File> trackedTempFiles = new ArrayList<>();
 
+    private Path sxssfTempDir;
+
     @AfterEach
+    void restoreDefaultTempFileStrategy() {
+        // 还原 POI 的进程级全局策略：trackSxssfTempFiles 改写的是单例，
+        // 不还原会泄漏到同一 JVM 内后续所有用到 SXSSF 的测试。
+        // （原来这个 @AfterEach 挂在 private 的 trackSxssfTempFiles 上，
+        //  既没有还原策略，方法本身还是被各用例手工调用的 setup 辅助。）
+        TempFile.setTempFileCreationStrategy(new DefaultTempFileCreationStrategy());
+    }
 
     private void trackSxssfTempFiles() {
+        // 把 SXSSF 滚动窗口临时文件重定向到本用例独占目录，使残留探测只覆盖
+        // 本用例自己产生的文件。
+        try {
+            sxssfTempDir = Files.createDirectory(tempDir.resolve("sxssf-probe"));
+        } catch (IOException ex) {
+            throw new IllegalStateException(ex);
+        }
         TempFile.setTempFileCreationStrategy(new TempFileCreationStrategy() {
             @Override
             public File createTempFile(String prefix, String suffix) throws IOException {
-                File file = File.createTempFile(prefix, suffix);
+                File file = Files.createTempFile(sxssfTempDir, prefix, suffix).toFile();
                 trackedTempFiles.add(file);
                 return file;
             }
 
             @Override
             public File createTempDirectory(String prefix) throws IOException {
-                File dir = Files.createTempDirectory(prefix).toFile();
+                File dir = Files.createTempDirectory(sxssfTempDir, prefix).toFile();
                 trackedTempFiles.add(dir);
                 return dir;
             }
@@ -62,10 +79,13 @@ class CollectDataExcelCleanupTest {
     private void assertSxssfTempFilesCleaned() {
         assertThat(trackedTempFiles).as("SXSSF 滚动窗口临时文件均已删除")
                 .allSatisfy(f -> assertThat(f).doesNotExist());
-        // POI 5.2.5 无法读取当前策略，改用系统临时目录中 SXSSF 固定前缀的残留探测。
-        try (Stream<Path> paths = Files.list(Path.of(System.getProperty("java.io.tmpdir")))) {
+        // 只探测本用例独占目录。原实现扫描系统共享临时目录并要求 poi-sxssf*
+        // 前缀零命中，会把其他进程、其他用例和历史运行遗留的文件算成本用例
+        // 的泄漏——本机共享临时目录里就躺着十几个历史 poi-sxssf-sheet*.xml，
+        // 导致实现完全正确时这 7 个用例照样红。
+        try (Stream<Path> paths = Files.list(sxssfTempDir)) {
             assertThat(paths.filter(p -> p.getFileName().toString().startsWith("poi-sxssf")))
-                    .as("系统临时目录无 SXSSF 滚动窗口残留")
+                    .as("本用例 SXSSF 临时目录无滚动窗口残留")
                     .isEmpty();
         } catch (IOException ex) {
             throw new IllegalStateException(ex);

backend-java/src/test/java/com/nanri/aiimage/modules/shopdatacrawl/service/ShopDataCrawlCleanupTest.java

@@ -149,6 +149,7 @@ class ShopDataCrawlCleanupTest {
     private final AtomicLong memberIdSeq = new AtomicLong(1000);
     private final AtomicLong fileSeq = new AtomicLong(5000);
     private Long lastJobTaskId;
+    private Set<Path> preExistingWorkRoots = Set.of();
 
     @BeforeEach
     void configureStorage() {
@@ -408,6 +409,12 @@ class ShopDataCrawlCleanupTest {
                 .deleteTaskItems(anyLong(), eq(MODULE_TYPE));
         lenient().doNothing().when(taskProgressSnapshotService)
                 .delete(anyLong(), eq(MODULE_TYPE));
+        // 共享临时目录里的既有 daily-* 目录不属于本用例，先快照、断言时做差集
+        try {
+            preExistingWorkRoots = Set.copyOf(scanWorkRoots());
+        } catch (Exception ex) {
+            throw new IllegalStateException(ex);
+        }
     }
 
     // ---- 1. 删除路径 ----
@@ -1261,6 +1268,16 @@ class ShopDataCrawlCleanupTest {
     }
 
     private List<Path> workRoots() throws Exception {
+        // 只统计本用例新产生的工作目录。shop-data-crawl-result 落在机器共享的
+        // 系统临时目录下，其他用例、历史运行、本机跑过的应用都会往里留 daily-*
+        // 目录（本机实测已堆了 200+ 个），原实现把这些全算作本用例的泄漏，
+        // 于是实现完全正确时这些用例照样红。改为与用例开始时的快照做差集。
+        List<Path> current = new ArrayList<>(scanWorkRoots());
+        current.removeAll(preExistingWorkRoots);
+        return current;
+    }
+
+    private List<Path> scanWorkRoots() throws Exception {
         Path root = Path.of(System.getProperty("java.io.tmpdir"), "shop-data-crawl-result");
         if (!Files.exists(root)) {
             return List.of();

backend-java/src/test/java/com/nanri/aiimage/modules/similarasin/service/SimilarAsinTaskServiceGroupRefTest.java

@@ -336,6 +336,56 @@ class SimilarAsinTaskServiceGroupRefTest {
         assertEquals(80, coverage);
     }
 
+    @Test
+    void test_task_006_group_payload_embeds_items_for_python() throws Exception {
+        // Python 旧链路（normalize_groups 直接返回 groups 并按 gp["items"] 取行）：
+        // 写入存储的 payload 中每组必须内嵌完整行，startIndex/endIndex 保留供 Java 内部展开。
+        java.util.concurrent.atomic.AtomicReference<String> captured = new java.util.concurrent.atomic.AtomicReference<>();
+        when(transientPayloadStorageService.storeParsedPayloadFast(
+                        eq(SimilarAsinTaskService.MODULE_TYPE), any(), anyString(), anyString(), eq(false)))
+                .thenAnswer(invocation -> {
+                    captured.set(invocation.getArgument(3));
+                    return "rustfs:task-parsed/similar-asin/30000/payload.json";
+                });
+        File workbook = buildWorkbook(12);
+        SimilarAsinParseVo vo = parse(workbook, "uploads/20260829/gr-python.xlsx");
+        assertEquals(12, vo.getAcceptedRows());
+        SimilarAsinParsedPayloadDto stored = objectMapper.readValue(captured.get(), SimilarAsinParsedPayloadDto.class);
+        assertNotNull(stored.getGroups());
+        assertEquals(12, stored.getGroups().size());
+        // 每组内嵌 items 与引用区间宽度一致，且行内容对应 items[startIndex]
+        for (SimilarAsinParsedGroupVo group : stored.getGroups()) {
+            assertNotNull(group.getItems());
+            assertEquals(group.getEndIndex() - group.getStartIndex(), group.getItems().size(),
+                    "Python 兼容：组内嵌 items 必须与引用区间宽度一致");
+            int start = group.getStartIndex();
+            assertEquals(stored.getItems().get(start).getRowToken(), group.getItems().get(0).getRowToken(),
+                    "组 items 首行必须是 items[startIndex]");
+        }
+        // Java 内部展开路径不受内嵌 items 影响：仍按引用切片
+        List<SimilarAsinParsedRowVo> expanded = SimilarAsinTaskService.expandGroupRefs(stored);
+        assertEquals(12, expanded.size(), "Java 内部展开仍按引用，不受内嵌 items 影响");
+    }
+
+    @Test
+    void test_task_006_group_hydrate_embeds_items_for_legacy_payload() throws Exception {
+        // 存量任务 payload 只有 startIndex/endIndex 引用（无内嵌 items）：
+        // parsedPayload 读取时 hydrate 必须回填组内嵌 items，Python 才能取到行。
+        String legacy = "{\"items\":["
+                + "{\"rowToken\":\"t1\",\"asin\":\"B0GRP00001\",\"sourceFileKey\":\"f.xlsx\",\"country\":\"英国\"},"
+                + "{\"rowToken\":\"t2\",\"asin\":\"B0GRP00002\",\"sourceFileKey\":\"f.xlsx\",\"country\":\"英国\"}"
+                + "],\"groups\":[{\"groupKey\":\"g1\",\"startIndex\":0,\"endIndex\":2,\"itemCount\":2}]}";
+        SimilarAsinParsedPayloadDto payload = objectMapper.readValue(legacy, SimilarAsinParsedPayloadDto.class);
+        SimilarAsinParsedPayloadDto hydrated = SimilarAsinTaskService.hydrateForPythonCompat(payload);
+        assertEquals(1, hydrated.getGroups().size());
+        assertNotNull(hydrated.getGroups().get(0).getItems());
+        assertEquals(2, hydrated.getGroups().get(0).getItems().size(), "hydrate 后组内嵌 items 必须为 2 行");
+        assertEquals("B0GRP00001", hydrated.getGroups().get(0).getItems().get(0).getAsin(),
+                "hydrate 后组内嵌 items 首行必须对应 items[0]");
+        assertEquals("B0GRP00002", hydrated.getGroups().get(0).getItems().get(1).getAsin(),
+                "hydrate 后组内嵌 items 第二行必须对应 items[1]");
+    }
+
     // ---- helpers ----
 
     private static String groupRefJson(int groupCount, int[][] ranges) {

backend-java/src/test/java/com/nanri/aiimage/modules/task/service/TaskFileJobDispatchCoordinatorOwnerRoutingTest.java

@@ -0,0 +1,85 @@
+package com.nanri.aiimage.modules.task.service;
+
+import com.nanri.aiimage.config.InstanceMetadata;
+import com.nanri.aiimage.modules.task.model.dto.TaskFileJobDispatchEvent;
+import com.nanri.aiimage.modules.task.model.dto.TaskFileJobMessage;
+import org.junit.jupiter.api.Test;
+import org.springframework.core.task.TaskExecutor;
+
+import java.time.LocalDateTime;
+
+import static org.mockito.ArgumentMatchers.any;
+import static org.mockito.ArgumentMatchers.anyBoolean;
+import static org.mockito.ArgumentMatchers.anyLong;
+import static org.mockito.ArgumentMatchers.eq;
+import static org.mockito.Mockito.mock;
+import static org.mockito.Mockito.never;
+import static org.mockito.Mockito.verify;
+import static org.mockito.Mockito.verifyNoInteractions;
+import static org.mockito.Mockito.when;
+
+/**
+ * owner-scoped job 的派发路由（2026-09-01 事故）：两台实例共用一个 RocketMQ consumer group，
+ * 消息被轮询投递到非归属实例后只能丢弃。归属实例改为直接本地派发，非归属实例不再发消息。
+ */
+class TaskFileJobDispatchCoordinatorOwnerRoutingTest {
+
+    private final TaskFileJobPublisher publisher = mock(TaskFileJobPublisher.class);
+    private final TaskFileJobLocalDispatcher localDispatcher = mock(TaskFileJobLocalDispatcher.class);
+    private final InstanceMetadata instanceMetadata = mock(InstanceMetadata.class);
+    private final TaskExecutor inlineExecutor = Runnable::run;
+
+    private TaskFileJobDispatchCoordinator buildCoordinator() {
+        return new TaskFileJobDispatchCoordinator(publisher, localDispatcher, instanceMetadata, inlineExecutor);
+    }
+
+    private static TaskFileJobDispatchEvent event(String owner) {
+        return new TaskFileJobDispatchEvent(18524L, 26402L, "SIMILAR_ASIN", 29405L,
+                owner == null ? "task:26402" : "task:26402:owner:" + owner,
+                TaskFileJobService.JOB_TYPE_ASSEMBLE_RESULT, owner, LocalDateTime.now());
+    }
+
+    @Test
+    void ownerJobDispatchedLocallyWithoutMq() {
+        when(instanceMetadata.getInstanceId()).thenReturn("server-121");
+        when(localDispatcher.dispatch(18524L, 26402L, "SIMILAR_ASIN", true)).thenReturn(true);
+
+        buildCoordinator().onTaskFileJobDispatch(event("server-121"));
+
+        verify(localDispatcher).dispatch(18524L, 26402L, "SIMILAR_ASIN", true);
+        verifyNoInteractions(publisher);
+    }
+
+    @Test
+    void ownerJobFallsBackToMqWhenLocalDispatchDeclined() {
+        when(instanceMetadata.getInstanceId()).thenReturn("server-121");
+        when(localDispatcher.dispatch(18524L, 26402L, "SIMILAR_ASIN", true)).thenReturn(false, true);
+        when(publisher.publish(any(TaskFileJobMessage.class))).thenReturn(false);
+
+        buildCoordinator().onTaskFileJobDispatch(event("server-121"));
+
+        verify(publisher).publish(any(TaskFileJobMessage.class));
+    }
+
+    @Test
+    void otherInstanceJobIsNeitherPublishedNorRun() {
+        when(instanceMetadata.getInstanceId()).thenReturn("server-110");
+
+        buildCoordinator().onTaskFileJobDispatch(event("server-121"));
+
+        verifyNoInteractions(publisher);
+        verify(localDispatcher, never()).dispatch(anyLong(), anyLong(), any(), anyBoolean());
+        verify(localDispatcher, never()).dispatch(anyLong(), anyLong(), any());
+    }
+
+    @Test
+    void jobWithoutOwnerStillGoesThroughMq() {
+        when(instanceMetadata.getInstanceId()).thenReturn("server-110");
+        when(publisher.publish(any(TaskFileJobMessage.class))).thenReturn(true);
+
+        buildCoordinator().onTaskFileJobDispatch(event(null));
+
+        verify(publisher).publish(any(TaskFileJobMessage.class));
+        verify(localDispatcher, never()).dispatch(anyLong(), anyLong(), eq("SIMILAR_ASIN"), anyBoolean());
+    }
+}

backend-java/src/test/java/com/nanri/aiimage/modules/task/service/TaskResultFileJobWorkerOwnerRoutingTest.java

@@ -0,0 +1,163 @@
+package com.nanri.aiimage.modules.task.service;
+
+import com.nanri.aiimage.common.exception.TaskOwnerMismatchException;
+import com.nanri.aiimage.config.InstanceMetadata;
+import com.nanri.aiimage.modules.appearancepatent.service.AppearancePatentTaskService;
+import com.nanri.aiimage.modules.brand.service.BrandTaskService;
+import com.nanri.aiimage.modules.collectdata.service.CollectDataService;
+import com.nanri.aiimage.modules.deletebrand.service.DeleteBrandRunService;
+import com.nanri.aiimage.modules.patroldelete.service.PatrolDeleteTaskService;
+import com.nanri.aiimage.modules.pricetrack.service.PriceTrackTaskService;
+import com.nanri.aiimage.modules.productrisk.service.ProductRiskTaskService;
+import com.nanri.aiimage.modules.publish.service.PublishTaskService;
+import com.nanri.aiimage.modules.queryasin.service.QueryAsinTaskService;
+import com.nanri.aiimage.modules.shopdatacrawl.service.ShopDataCrawlTaskService;
+import com.nanri.aiimage.modules.shopmatch.service.ShopMatchTaskService;
+import com.nanri.aiimage.modules.similarasin.service.SimilarAsinTaskService;
+import com.nanri.aiimage.modules.task.mapper.FileResultMapper;
+import com.nanri.aiimage.modules.task.model.entity.TaskFileJobEntity;
+import com.nanri.aiimage.modules.withdraw.service.WithdrawTaskService;
+import org.junit.jupiter.api.Test;
+
+import java.lang.reflect.Field;
+import java.time.LocalDateTime;
+import java.util.List;
+
+import static org.mockito.ArgumentMatchers.any;
+import static org.mockito.ArgumentMatchers.anyLong;
+import static org.mockito.ArgumentMatchers.eq;
+import static org.mockito.Mockito.doThrow;
+import static org.mockito.Mockito.mock;
+import static org.mockito.Mockito.never;
+import static org.mockito.Mockito.verify;
+import static org.mockito.Mockito.when;
+
+/**
+ * 双实例归属路由回归测试（2026-09-01 货源查询任务被非归属实例误杀）：
+ * 1) 定时 worker 已 claim 的 job 不能再走 claimRunning，否则条件 UPDATE 失败被静默丢弃；
+ * 2) 处理中撞上归属校验时只退回 PENDING，不得计入 retryCount（markFailed）。
+ */
+class TaskResultFileJobWorkerOwnerRoutingTest {
+
+    private final ShopMatchTaskService shopMatch = mock(ShopMatchTaskService.class);
+    private final PriceTrackTaskService priceTrack = mock(PriceTrackTaskService.class);
+    private final ProductRiskTaskService productRisk = mock(ProductRiskTaskService.class);
+    private final PublishTaskService publish = mock(PublishTaskService.class);
+    private final QueryAsinTaskService queryAsin = mock(QueryAsinTaskService.class);
+    private final ShopDataCrawlTaskService shopDataCrawl = mock(ShopDataCrawlTaskService.class);
+    private final WithdrawTaskService withdraw = mock(WithdrawTaskService.class);
+    private final PatrolDeleteTaskService patrolDelete = mock(PatrolDeleteTaskService.class);
+    private final AppearancePatentTaskService appearance = mock(AppearancePatentTaskService.class);
+    private final SimilarAsinTaskService similar = mock(SimilarAsinTaskService.class);
+    private final DeleteBrandRunService deleteBrand = mock(DeleteBrandRunService.class);
+    private final BrandTaskService brand = mock(BrandTaskService.class);
+    private final CollectDataService collectData = mock(CollectDataService.class);
+    private final TaskResultPayloadService payload = mock(TaskResultPayloadService.class);
+    private final TaskFileJobService taskFileJobService = mock(TaskFileJobService.class);
+    private final TaskDistributedLockService taskDistributedLockService = mock(TaskDistributedLockService.class);
+    private final TaskDistributedLockService.LockHandle lock = mock(TaskDistributedLockService.LockHandle.class);
+    private final TaskFileJobLocalDispatcher localDispatcher = mock(TaskFileJobLocalDispatcher.class);
+    private final InstanceMetadata instanceMetadata = mock(InstanceMetadata.class);
+
+    private TaskResultFileJobWorker buildWorker() throws Exception {
+        List<ResultFileJobHandler> handlers = List.of(
+                new ShopMatchResultFileJobHandler(shopMatch, payload),
+                new PriceTrackResultFileJobHandler(priceTrack, payload),
+                new ProductRiskResultFileJobHandler(productRisk, payload),
+                new PublishResultFileJobHandler(publish),
+                new QueryAsinResultFileJobHandler(queryAsin, payload),
+                new ShopDataCrawlResultFileJobHandler(shopDataCrawl, payload),
+                new WithdrawResultFileJobHandler(withdraw, payload),
+                new PatrolDeleteResultFileJobHandler(patrolDelete, payload),
+                new AppearancePatentResultFileJobHandler(appearance),
+                new SimilarAsinResultFileJobHandler(similar),
+                new DeleteBrandResultFileJobHandler(deleteBrand),
+                new BrandResultFileJobHandler(brand, payload),
+                new CollectDataResultFileJobHandler(collectData));
+        TaskResultFileJobWorker worker = new TaskResultFileJobWorker(
+                taskFileJobService,
+                taskDistributedLockService,
+                mock(FileResultMapper.class),
+                localDispatcher,
+                instanceMetadata,
+                withdraw, brand,
+                new ResultFileJobHandlerRegistry(handlers));
+        set(worker, "taskQueueExecutor", (org.springframework.core.task.TaskExecutor) Runnable::run);
+        set(worker, "localWorkerEnabled", true);
+        set(worker, "batchSize", 20);
+        return worker;
+    }
+
+    private static void set(TaskResultFileJobWorker worker, String fieldName, Object value) throws Exception {
+        Field field = TaskResultFileJobWorker.class.getDeclaredField(fieldName);
+        field.setAccessible(true);
+        field.set(worker, value);
+    }
+
+    private static TaskFileJobEntity claimedJob(String moduleType, long jobId, long taskId, String scopeKey) {
+        TaskFileJobEntity entity = new TaskFileJobEntity();
+        entity.setId(jobId);
+        entity.setTaskId(taskId);
+        entity.setResultId(taskId + 1000L);
+        entity.setModuleType(moduleType);
+        entity.setScopeKey(scopeKey);
+        entity.setStatus("RUNNING");
+        entity.setUpdatedAt(LocalDateTime.now());
+        return entity;
+    }
+
+    @Test
+    void scheduledWorkerProcessesAlreadyClaimedJobWithoutReclaiming() throws Exception {
+        TaskResultFileJobWorker worker = buildWorker();
+        when(instanceMetadata.getInstanceId()).thenReturn("instance-a");
+        TaskFileJobEntity claim = claimedJob(PublishTaskService.MODULE_TYPE, 18524L, 26402L, "task:26402:owner:instance-a");
+        when(taskFileJobService.claimRunnableJobsForOwner(20, "instance-a")).thenReturn(List.of(claim));
+        when(taskFileJobService.activateRunningClaim(claim)).thenReturn(true);
+        when(localDispatcher.dispatchClaimed(claim.getId(), claim.getTaskId(), claim.getModuleType())).thenReturn(false);
+        when(taskDistributedLockService.acquire(PublishTaskService.MODULE_TYPE, claim.getTaskId(),
+                TaskDistributedLockService.DEFAULT_WAIT_MILLIS)).thenReturn(lock);
+
+        worker.runPendingJobs();
+
+        verify(taskFileJobService, never()).claimRunning(anyLong());
+        verify(publish).processResultFileJob(claim);
+        verify(taskFileJobService).markSuccess(eq(claim), any());
+    }
+
+    @Test
+    void scheduledWorkerPrefersClaimedDispatch() throws Exception {
+        TaskResultFileJobWorker worker = buildWorker();
+        when(instanceMetadata.getInstanceId()).thenReturn("instance-a");
+        TaskFileJobEntity claim = claimedJob(PublishTaskService.MODULE_TYPE, 18525L, 26403L, "task:26403:owner:instance-a");
+        when(taskFileJobService.claimRunnableJobsForOwner(20, "instance-a")).thenReturn(List.of(claim));
+        when(localDispatcher.dispatchClaimed(claim.getId(), claim.getTaskId(), claim.getModuleType())).thenReturn(true);
+
+        worker.runPendingJobs();
+
+        verify(localDispatcher).dispatchClaimed(claim.getId(), claim.getTaskId(), claim.getModuleType());
+        verify(localDispatcher, never()).dispatch(anyLong(), anyLong(), any());
+        verify(taskFileJobService, never()).claimRunning(anyLong());
+        verify(publish, never()).processResultFileJob(any());
+    }
+
+    @Test
+    void ownerMismatchDuringProcessingDefersInsteadOfFailing() throws Exception {
+        TaskResultFileJobWorker worker = buildWorker();
+        when(instanceMetadata.getInstanceId()).thenReturn("instance-a");
+        // scopeKey 归属本机（通过 process 的前置校验），但模块层读到的 task owner 已经变成他机
+        TaskFileJobEntity job = claimedJob(PublishTaskService.MODULE_TYPE, 18526L, 26404L, "task:26404:owner:instance-a");
+        when(taskFileJobService.claimRunning(job.getId())).thenReturn(job);
+        when(taskFileJobService.activateRunningClaim(job)).thenReturn(true);
+        when(taskDistributedLockService.acquire(PublishTaskService.MODULE_TYPE, job.getTaskId(),
+                TaskDistributedLockService.DEFAULT_WAIT_MILLIS)).thenReturn(lock);
+        doThrow(new TaskOwnerMismatchException(job.getTaskId(), "assemble result file", "instance-b", "instance-a"))
+                .when(publish).processResultFileJob(job);
+
+        worker.process(job);
+
+        verify(taskFileJobService).deferRunning(eq(job.getId()), any());
+        verify(taskFileJobService, never()).markFailed(any(), any());
+        verify(taskFileJobService, never()).markFailedPermanent(any(), any());
+        verify(taskFileJobService, never()).isRetryExhausted(anyLong());
+    }
+}

backend-java/src/test/java/com/nanri/aiimage/modules/task/service/TaskResultFileJobWorkerOwnerScopedTest.java

@@ -128,6 +128,20 @@ class TaskResultFileJobWorkerOwnerScopedTest {
         TaskResultFileJobWorker worker = buildWorker();
         when(instanceMetadata.getInstanceId()).thenReturn("instance-a");
         TaskFileJobEntity job = job("PUBLISH", 2L, 12L, "task:12:owner:instance-b");
+        job.setStatus("RUNNING");
+        worker.process(job);
+        verifyNoInteractions(taskFileJobService, taskDistributedLockService, publish);
+    }
+
+    @Test
+    void ownerScopedOtherOwnerPendingSkip() {
+        // PENDING + owner 不匹配也必须跳过：模块层 ensureTaskOwnedByCurrentInstance 会抛
+        // TaskOwnerMismatchException，本机抢跑只会耗尽 job 重试次数并误杀归属机上的任务。
+        // job 由归属实例的 runPendingJobs（每 15 秒扫自己 owner 的 PENDING/FAILED）接手。
+        TaskResultFileJobWorker worker = buildWorker();
+        when(instanceMetadata.getInstanceId()).thenReturn("instance-a");
+        TaskFileJobEntity job = job("PUBLISH", 2L, 12L, "task:12:owner:instance-b");
+        job.setStatus("PENDING");
         worker.process(job);
         verifyNoInteractions(taskFileJobService, taskDistributedLockService, publish);
     }