You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
环境
症状(与 #10515 相同)
打开应用退回"选择工作区"界面,选择/创建工作区报:
应用无任何报错日志、无崩溃;host 进程 CPU 0%,永久挂起。
根因(已通过 10 轮 DSH_HOME 沙盒二分实验实证)
~/.dsh/profiles/desktop)安装过一个实验性 bundle(0.2.2-rc.1,版本高于当时应用 0.2.1-rc.5)。pnpm 的 hoisted 布局把它的传递依赖@deepseek-ai/dsh-storage-domain@0.2.2-rc.1和@deepseek-ai/dsh-typert-protocol@0.2.2-rc.1提升到了 node_modules 顶层。fiberPhase=pending,workspaceController 服务不注册 → 网关对该服务的所有调用返回 service-unavailable。最小复现条件(二分矩阵结论):
node_modules 顶层的 dsh-storage-domain(版本≠应用内置)×存在历史会话数据×存在工作区存储—— 三者同时满足即挂起。给其他用户的排查方法(无需任何工具)
从终端启动应用拿到网关地址和 token(stdout 会打印
dsh web: http://127.0.0.1:PORT/?token=...),然后:看返回里的
fiberPhase:如果 workspace-controller 等 6 个插件是pending而其余 160+ 都是active,八成就是本问题。修复方法(用户侧,已验证有效)
退出应用后,把
~/.dsh/profiles/desktop/node_modules/@deepseek-ai/目录整个移走(改名备份),或连同pnpm-lock.yaml一起删除后用pnpm install按 package.json 重建。实验性 bundle 不需要外装:应用已内置同版本实现,保留 package.json 的 bundles 引用即可。修复后 189 个插件全部 active,工作区与历史会话完整恢复。给官方的两条改进建议
@deepseek-ai/*基础库(storage-domain、typert-protocol 等)时应优先使用内置版本,或至少校验版本匹配、不匹配时快速报错,而不是静默加载错配版本。另附带一个小问题:应用内更新器下载新版本时曾卡死(进度 0.3% 停滞 9 分钟以上),同一网络环境下 curl 直连官方 CDN 可达 61MB/s,疑似更新器的连接管理问题。
—— 完整二分实验矩阵与时间线存档于本地,如需更多细节可联系发帖人。
All reactions