diff --git a/app/modules/mediavault/mediavault.py b/app/modules/mediavault/mediavault.py
index 9888593d4..199df08ca 100644
--- a/app/modules/mediavault/mediavault.py
+++ b/app/modules/mediavault/mediavault.py
@@ -244,10 +244,18 @@ class MediaVault:
         def rows() -> Generator[Dict[str, Any], Any, None]:
             result: Dict[str, Any] = first
             page = 1
+            seen = 0
             while True:
                 batch = result.get("items") or []
+                seen += len(batch)
                 yield from batch
-                if len(batch) < self.PAGE_LIMIT:
+                total = result.get("total")
+                # 有可靠总数时以它为准：末页恰好满额也不再多发一次请求，
+                # 那次多余请求一旦失败会把完整结果误报成服务不可达
+                if isinstance(total, int):
+                    if seen >= total:
+                        return
+                elif len(batch) < self.PAGE_LIMIT:
                     return
                 page += 1
                 following = self.__query_items(
@@ -355,18 +363,19 @@ class MediaVault:
         if rows is None:
             return None
         try:
-            search_rows = list(rows)
+            # 逐行检查、命中即返回：预取全部页会让第一页已命中的目标
+            # 因后续页请求失败被误报成服务不可达
+            for row in rows:
+                item = self.__format_item_info(row)
+                if not item or item.title != title:
+                    continue
+                if year and str(item.year) != str(year):
+                    continue
+                if not MediaServerIdentityHelper.is_compatible(item, media_source, media_id):
+                    continue
+                return str(item.item_id)
         except _SearchInterrupted:
             return None
-        for row in search_rows:
-            item = self.__format_item_info(row)
-            if not item or item.title != title:
-                continue
-            if year and str(item.year) != str(year):
-                continue
-            if not MediaServerIdentityHelper.is_compatible(item, media_source, media_id):
-                continue
-            return str(item.item_id)
         return ""
 
     def get_season_episode_ids(self, item_id: str, season: int) -> Dict[int, str]:
diff --git a/tests/test_mediavault_module.py b/tests/test_mediavault_module.py
index 1112e214a..232a63c3f 100644
--- a/tests/test_mediavault_module.py
+++ b/tests/test_mediavault_module.py
@@ -549,7 +549,8 @@ def _failing_second_page(total_rows: list):
 
 def test_get_movies_reports_unreachable_when_a_later_page_fails():
     """翻页中途断连必须返回 None，不能把残缺结果当成「不在库中」。"""
-    rows = [_item_row(i, title=f"沙丘{i}") for i in range(MediaVault.PAGE_LIMIT)]
+    # total 必须超过一页，否则第一页就判定取完，构造不出「中途失败」
+    rows = [_item_row(i, title=f"沙丘{i}") for i in range(MediaVault.PAGE_LIMIT + 1)]
     client = _client({"/items": _failing_second_page(rows)})
 
     assert client.get_movies(title="沙丘") is None
@@ -557,7 +558,33 @@ def test_get_movies_reports_unreachable_when_a_later_page_fails():
 
 def test_find_series_reports_unreachable_when_a_later_page_fails():
     """同上：剧集定位不能把断连误判成整部剧未入库。"""
-    rows = [_item_row(i, kind="Series", title=f"剧{i}") for i in range(MediaVault.PAGE_LIMIT)]
+    rows = [_item_row(i, kind="Series", title=f"剧{i}") for i in range(MediaVault.PAGE_LIMIT + 1)]
     client = _client({"/items": _failing_second_page(rows)})
 
     assert client.get_tv_episodes(title="剧A") == (None, None)
+
+
+def test_search_stops_at_reported_total_without_an_extra_request():
+    """末页恰好满额时用 total 判定取完，不再多发一次可能失败的请求。"""
+    rows = [_item_row(i, title="沙丘", year=2021) for i in range(MediaVault.PAGE_LIMIT)]
+    client = _client({"/items": _paged_items(rows)})
+
+    matched = client.get_movies(title="沙丘")
+
+    assert len(matched) == MediaVault.PAGE_LIMIT
+    assert [call["params"]["page"] for call in client._api.calls] == [1]
+
+
+def test_find_series_returns_first_page_hit_even_if_a_later_page_fails():
+    """第一页已命中就该直接返回，不因后续页失败被误报成服务不可达。"""
+    rows = [_item_row(0, kind="Series", title="剧A", year=2020)]
+    rows += [_item_row(i, kind="Series", title=f"其它{i}") for i in range(1, MediaVault.PAGE_LIMIT + 1)]
+    routes = {
+        "/items": _failing_second_page(rows),
+        "/items/id-0/episodes": Result(True, {"seasons": {"1": [1]}}),
+    }
+
+    item_id, seasons = _client(routes).get_tv_episodes(title="剧A", year="2020")
+
+    assert item_id == "id-0"
+    assert seasons == {1: [1]}