diff --git a/src/core/command.rs b/src/core/command.rs
index 5b2ea85..6cd7e0a 100644
--- a/src/core/command.rs
+++ b/src/core/command.rs
@@ -3,7 +3,6 @@ use std::path::{Path, PathBuf};
 use std::pin::Pin;
 
 use anyhow::{bail, Result};
-use dirs::config_dir;
 use entity::{favorite, page, video};
 use filenamify::filenamify;
 use futures::stream::{FuturesOrdered, FuturesUnordered};
@@ -17,8 +16,11 @@ use tokio::fs;
 use tokio::sync::{Mutex, Semaphore};
 
 use super::status::{PageStatus, VideoStatus};
-use super::utils::{unhandled_videos_pages, ModelWrapper, NFOMode, NFOSerializer, TEMPLATE};
+use super::utils::{
+    unhandled_videos_pages, update_pages_model, update_videos_model, ModelWrapper, NFOMode, NFOSerializer, TEMPLATE,
+};
 use crate::bilibili::{BestStream, BiliClient, BiliError, FavoriteList, FilterOption, PageInfo, Video};
+use crate::config::CONFIG;
 use crate::core::utils::{
     create_video_pages, create_videos, exist_labels, filter_unfilled_videos, handle_favorite_info, total_video_count,
 };
@@ -135,10 +137,15 @@ pub async fn download_unprocessed_videos(
     // 对于视频，允许五个同时下载（视频内还有分页、不同分页还有多种下载任务）
     let semaphore = Semaphore::new(5);
     let downloader = Downloader::default();
-    let mut uppers_mutex: HashMap<i32, (Mutex<()>, Mutex<()>)> = HashMap::new();
+    let mut uppers_mutex: HashMap<i64, (Mutex<()>, Mutex<()>)> = HashMap::new();
     for (video_model, _) in &unhandled_videos_pages {
         uppers_mutex.insert(video_model.upper_id, (Mutex::new(()), Mutex::new(())));
     }
+    let upper_path = {
+        let config = CONFIG.lock().unwrap();
+        config.upper_path.clone()
+    };
+    let upper_path = Path::new(&upper_path);
     let mut tasks = unhandled_videos_pages
         .into_iter()
         .map(|(video_model, pages_model)| {
@@ -150,6 +157,7 @@ pub async fn download_unprocessed_videos(
                 connection,
                 &semaphore,
                 &downloader,
+                upper_path,
                 upper_mutex,
             )
         })
@@ -169,18 +177,18 @@ pub async fn download_unprocessed_videos(
         }
         // 满十个就写入数据库
         if models.len() == 10 {
-            video::Entity::insert_many(std::mem::replace(&mut models, Vec::with_capacity(10)))
-                .exec(connection)
-                .await?;
+            update_videos_model(std::mem::replace(&mut models, Vec::with_capacity(10)), connection).await?;
         }
     }
     if !models.is_empty() {
-        video::Entity::insert_many(models).exec(connection).await?;
+        update_videos_model(models, connection).await?;
     }
     info!("download videos in favorite: {} done.", favorite_model.f_id);
     Ok(())
 }
 
+/// 暂时这样做，后面提取成上下文
+#[allow(clippy::too_many_arguments)]
 pub async fn download_video_pages(
     bili_client: &BiliClient,
     video_model: video::Model,
@@ -188,6 +196,7 @@ pub async fn download_video_pages(
     connection: &DatabaseConnection,
     semaphore: &Semaphore,
     downloader: &Downloader,
+    upper_path: &Path,
     upper_mutex: &(Mutex<()>, Mutex<()>),
 ) -> Result<video::ActiveModel> {
     let permit = semaphore.acquire().await;
@@ -199,10 +208,7 @@ pub async fn download_video_pages(
 
     let base_path = Path::new(&video_model.path);
     let upper_id = video_model.upper_id.to_string();
-    let base_upper_path = config_dir()
-        .unwrap()
-        .join("bili-sync")
-        .join("upper")
+    let base_upper_path = upper_path
         .join(upper_id.chars().next().unwrap().to_string())
         .join(upper_id);
 
@@ -292,12 +298,18 @@ pub async fn dispatch_download_page(
         .into_iter()
         .map(|page_model| download_page(bili_client, video_model, page_model, &child_semaphore, downloader))
         .collect::<FuturesUnordered<_>>();
-    // 任务结束会返回 Result<PageModel>
-
     let mut models = Vec::with_capacity(10);
+    let mut should_error = false;
     while let Some(res) = tasks.next().await {
         match res {
             Ok(model) => {
+                if let Set(status) = model.download_status {
+                    let status = PageStatus::new(status);
+                    if status.should_run().iter().any(|v| *v) {
+                        // 有一个分页没下载完成，就应该将视频本身标记为未完成
+                        should_error = true;
+                    }
+                }
                 models.push(model);
             }
             Err(e) => {
@@ -308,13 +320,14 @@ pub async fn dispatch_download_page(
             }
         }
         if models.len() == 10 {
-            page::Entity::insert_many(std::mem::replace(&mut models, Vec::with_capacity(10)))
-                .exec(connection)
-                .await?;
+            update_pages_model(std::mem::replace(&mut models, Vec::with_capacity(10)), connection).await?;
         }
     }
     if !models.is_empty() {
-        page::Entity::insert_many(models).exec(connection).await?;
+        update_pages_model(models, connection).await?;
+    }
+    if should_error {
+        bail!("Some pages failed to download");
     }
     Ok(())
 }
diff --git a/src/core/utils.rs b/src/core/utils.rs
index 65e3bd1..c0fda00 100644
--- a/src/core/utils.rs
+++ b/src/core/utils.rs
@@ -225,6 +225,29 @@ pub async fn unhandled_videos_pages(
         .all(connection)
         .await?)
 }
+/// 更新视频 model 的下载状态
+pub async fn update_videos_model(videos: Vec<video::ActiveModel>, connection: &DatabaseConnection) -> Result<()> {
+    video::Entity::insert_many(videos)
+        .on_conflict(
+            OnConflict::column(video::Column::Id)
+                .update_column(video::Column::DownloadStatus)
+                .to_owned(),
+        )
+        .exec(connection)
+        .await?;
+    Ok(())
+}
+
+/// 更新视频页 model 的下载状态
+pub async fn update_pages_model(pages: Vec<page::ActiveModel>, connection: &DatabaseConnection) -> Result<()> {
+    let query = page::Entity::insert_many(pages).on_conflict(
+        OnConflict::column(page::Column::Id)
+            .update_columns([page::Column::DownloadStatus, page::Column::Path])
+            .to_owned(),
+    );
+    query.exec(connection).await?;
+    Ok(())
+}
 
 /// serde xml 似乎不太好用，先这么裸着写
 /// （真是又臭又长啊