[GH-ISSUE #1346] [Bug]: flux1 kontext dev generates images, executes to the unet loader, and appears to be reconnecting and cannot continue execution-ComfyUI Setup 0.4.74 - x64 #374

New issue

Closed

opened 2026-10-31 22:12:12 +00:00 by bsakaguchi · 27 comments

bsakaguchi commented 2026-10-31 22:12:12 +00:00

Owner

Copy link

Originally created by @qlxxkj on GitHub (Sep 25, 2025).
Original GitHub issue: https://github.com/Comfy-Org/desktop/issues/1346

App Version

comfyui_destop_0.4.74

Expected Behavior

flux1 kontext dev successfully generates images

Actual Behavior

flux1 kontext dev generates images, executes to the unet loader, and appears to be reconnecting and cannot continue execution.

Steps to Reproduce

1. The comfyUI desktop version will prompt you that a new version is available.
2. Click Upgrade Now to complete the installation.
3. Open the template browser and find the flux_kontex_dev_basic template.
4. Load the reference image.
5. Execute.
6. Execute to the Unet loader and you will see a message saying "Reconnecting."

Debug Logs

Comfui.log
[2025-09-25 15:34:42.703] [info]  Adding extra search path custom_nodes C:\ComfyUI\custom_nodes
Adding extra search path download_model_base C:\ComfyUI\models
Adding extra search path custom_nodes C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes
Setting output directory to: C:\ComfyUI\output

[2025-09-25 15:34:42.709] [info]  Setting input directory to: C:\ComfyUI\input
Setting user directory to: C:\ComfyUI\user

[2025-09-25 15:34:45.805] [info]  [START] Security scan
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2025-09-25 15:34:45.805
[2025-09-25 15:34:45.809] [info]  
** Platform: Windows
** Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]
** Python executable: C:\ComfyUI\.venv\Scripts\python.exe
** ComfyUI Path: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI
** ComfyUI Base Folder Path: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI
** User directory: C:\ComfyUI\user
** ComfyUI-Manager config path: C:\ComfyUI\user\default\ComfyUI-Manager\config.ini
** Log path: C:\ComfyUI\user\comfyui.log

[2025-09-25 15:34:47.565] [info]  [ComfyUI-Manager] Skipped fixing the 'comfyui-frontend-package' dependency because the ComfyUI is outdated.

[2025-09-25 15:34:47.569] [info]  
Prestartup times for custom nodes:
   4.9 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager


[2025-09-25 15:34:50.568] [info]  Checkpoint files will always be loaded safely.

[2025-09-25 15:34:50.847] [info]  Total VRAM 16376 MB, total RAM 15874 MB
pytorch version: 2.8.0+cu129

[2025-09-25 15:34:50.851] [info]  Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA RTX A4000 : cudaMallocAsync

[2025-09-25 15:34:53.274] [info]  Using pytorch attention

[2025-09-25 15:34:55.910] [info]  Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]
ComfyUI version: 0.3.60

[2025-09-25 15:34:55.947] [info]  [Prompt Server] web root: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app

[2025-09-25 15:34:57.994] [info]  ### Loading: ComfyUI-Manager (V3.36)

[2025-09-25 15:34:57.998] [info]  [ComfyUI-Manager] network_mode: public
### ComfyUI Revision: UNKNOWN (The currently installed ComfyUI is not a Git repository)

[2025-09-25 15:34:58.027] [info]  
Import times for custom nodes:

[2025-09-25 15:34:58.030] [info]     0.0 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\websocket_image_save.py
   0.0 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager


[2025-09-25 15:34:58.260] [info]  [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json

[2025-09-25 15:34:58.296] [info]  [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json

[2025-09-25 15:34:58.368] [info]  [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json

[2025-09-25 15:34:58.459] [info]  [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json

[2025-09-25 15:34:58.553] [info]  [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json

[2025-09-25 15:34:59.131] [error] Failed to initialize database. Please ensure you have installed the latest requirements. If the error persists, please report this as in future the database will be required: (sqlite3.OperationalError) unable to open database file
(Background on this error at: https://sqlalche.me/e/20/e3q8)

[2025-09-25 15:34:59.155] [info]  Starting server


[2025-09-25 15:34:59.159] [info]  To see the GUI go to: http://127.0.0.1:8000

[2025-09-25 15:35:00.828] [error] comfyui-frontend-package not found in requirements.txt

[2025-09-25 15:35:01.767] [error] comfyui-frontend-package not found in requirements.txt

[2025-09-25 15:35:18.935] [info]  FETCH ComfyRegistry Data: 5/98

[2025-09-25 15:35:33.819] [info]  FETCH ComfyRegistry Data: 10/98

[2025-09-25 15:35:48.372] [info]  FETCH ComfyRegistry Data: 15/98

[2025-09-25 15:36:08.650] [info]  FETCH ComfyRegistry Data: 20/98

[2025-09-25 15:36:30.720] [info]  FETCH ComfyRegistry Data: 25/98

[2025-09-25 15:36:50.819] [info]  FETCH ComfyRegistry Data: 30/98

[2025-09-25 15:37:20.452] [info]  FETCH ComfyRegistry Data: 35/98

[2025-09-25 15:37:38.429] [info]  FETCH ComfyRegistry Data: 40/98

[2025-09-25 15:37:52.542] [info]  FETCH ComfyRegistry Data: 45/98

[2025-09-25 15:38:05.348] [info]  FETCH ComfyRegistry Data: 50/98

[2025-09-25 15:38:32.769] [info]  FETCH ComfyRegistry Data: 55/98

[2025-09-25 15:38:52.111] [info]  FETCH ComfyRegistry Data: 60/98

[2025-09-25 15:39:13.232] [info]  FETCH ComfyRegistry Data: 65/98

[2025-09-25 15:39:36.862] [info]  FETCH ComfyRegistry Data: 70/98

[2025-09-25 15:40:00.289] [info]  FETCH ComfyRegistry Data: 75/98

[2025-09-25 15:40:22.437] [info]  FETCH ComfyRegistry Data: 80/98

[2025-09-25 15:40:41.965] [info]  FETCH ComfyRegistry Data: 85/98

[2025-09-25 15:41:07.638] [info]  FETCH ComfyRegistry Data: 90/98

[2025-09-25 15:41:31.580] [info]  FETCH ComfyRegistry Data: 95/98

[2025-09-25 15:41:41.977] [info]  FETCH ComfyRegistry Data [DONE]

[2025-09-25 15:41:42.259] [info]  [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes

[2025-09-25 15:41:42.294] [info]  FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json

main.log
[2025-09-25 15:34:41.256] [info]  Using uv at C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe
[2025-09-25 15:34:41.257] [info]  Install state: installed
[2025-09-25 15:34:41.259] [info]  Validating installation. Recorded state: [installed]
[2025-09-25 15:34:41.264] [info]  Running uv command directly: pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt
[2025-09-25 15:34:41.265] [info]  Running uv child process: uv pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt
[2025-09-25 15:34:41.266] [info]  Running command: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt in C:\ComfyUI
[2025-09-25 15:34:41.362] [info]  Running uv command directly: pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt
[2025-09-25 15:34:41.363] [info]  Running uv child process: uv pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt
[2025-09-25 15:34:41.363] [info]  Running command: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt in C:\ComfyUI
[2025-09-25 15:34:41.529] [info]  Validation result: isValid:true, state:installed {"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}
[2025-09-25 15:34:42.345] [info]  Server start
[2025-09-25 15:34:42.378] [info]  Running command: C:\ComfyUI\.venv\Scripts\python.exe C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\main.py --user-directory C:\ComfyUI\user --input-directory C:\ComfyUI\input --output-directory C:\ComfyUI\output --front-end-root C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app --base-directory C:\ComfyUI --extra-model-paths-config C:\Users\Administrator\AppData\Roaming\ComfyUI\extra_models_config.yaml --log-stdout --listen 127.0.0.1 --port 8000 in C:\ComfyUI
[2025-09-25 15:34:42.412] [info]  Received renderer-ready message!
[2025-09-25 15:34:42.413] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.414] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.415] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.415] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.416] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.417] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.417] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.418] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.418] [info]  Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}}
[2025-09-25 15:34:42.419] [info]  Sending queued message {"channel":"loading-progress","data":{"status":"starting-server"}}
[2025-09-25 15:34:59.450] [info]  Python server is ready
[2025-09-25 15:35:12.341] [error] @todesktop/runtime: alwaysResolve: Promise timed out after 30000ms
[2025-09-25 15:35:12.973] [info]  @todesktop/runtime: Checking for update
[2025-09-25 15:35:12.974] [info]  @todesktop/runtime: AutoUpdater: checking-for-update
[2025-09-25 15:35:15.442] [info]  @todesktop/runtime: Update for version 0.4.74 is not available (latest version: 0.4.74, downgrade is allowed).
[2025-09-25 15:35:15.444] [info]  @todesktop/runtime: AutoUpdater: update-not-available {"version":"0.4.74","files":[{"url":"ComfyUI Setup 0.4.74 - Build 250924sn4vw8lfo-x64.exe","sha512":"EUB9h5eBMYCXAY1QeE84boGhZYF+lUSsFx4sZIGZnTCfU3qxiGkm7JS+BKTSoX8OU0gUBM9crSZHCANVd/o54A==","size":118282152}],"path":"ComfyUI Setup 0.4.74 - Build 250924sn4vw8lfo-x64.exe","sha512":"EUB9h5eBMYCXAY1QeE84boGhZYF+lUSsFx4sZIGZnTCfU3qxiGkm7JS+BKTSoX8OU0gUBM9crSZHCANVd/o54A==","releaseDate":"2025-09-24T02:25:54.886Z"}

Browser Logs

none

Settings JSON

comfy.settings.json

Other

No response

┆Issue is synchronized with this Notion page by Unito

Originally created by @qlxxkj on GitHub (Sep 25, 2025). Original GitHub issue: https://github.com/Comfy-Org/desktop/issues/1346 ### App Version comfyui_destop_0.4.74 <img width="698" height="103" alt="Image" src="https://github.com/user-attachments/assets/ec6cd010-7af0-4ca9-9105-5b055e2f4af1" /> ### Expected Behavior flux1 kontext dev successfully generates images ### Actual Behavior flux1 kontext dev generates images, executes to the unet loader, and appears to be reconnecting and cannot continue execution. ### Steps to Reproduce 1. The comfyUI desktop version will prompt you that a new version is available. 2. Click Upgrade Now to complete the installation. 3. Open the template browser and find the flux_kontex_dev_basic template. 4. Load the reference image. 5. Execute. 6. Execute to the Unet loader and you will see a message saying "Reconnecting." ### Debug Logs ```powershell Comfui.log [2025-09-25 15:34:42.703] [info] Adding extra search path custom_nodes C:\ComfyUI\custom_nodes Adding extra search path download_model_base C:\ComfyUI\models Adding extra search path custom_nodes C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes Setting output directory to: C:\ComfyUI\output [2025-09-25 15:34:42.709] [info] Setting input directory to: C:\ComfyUI\input Setting user directory to: C:\ComfyUI\user [2025-09-25 15:34:45.805] [info] [START] Security scan [DONE] Security scan ## ComfyUI-Manager: installing dependencies done. ** ComfyUI startup time: 2025-09-25 15:34:45.805 [2025-09-25 15:34:45.809] [info] ** Platform: Windows ** Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)] ** Python executable: C:\ComfyUI\.venv\Scripts\python.exe ** ComfyUI Path: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI ** ComfyUI Base Folder Path: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI ** User directory: C:\ComfyUI\user ** ComfyUI-Manager config path: C:\ComfyUI\user\default\ComfyUI-Manager\config.ini ** Log path: C:\ComfyUI\user\comfyui.log [2025-09-25 15:34:47.565] [info] [ComfyUI-Manager] Skipped fixing the 'comfyui-frontend-package' dependency because the ComfyUI is outdated. [2025-09-25 15:34:47.569] [info] Prestartup times for custom nodes: 4.9 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager [2025-09-25 15:34:50.568] [info] Checkpoint files will always be loaded safely. [2025-09-25 15:34:50.847] [info] Total VRAM 16376 MB, total RAM 15874 MB pytorch version: 2.8.0+cu129 [2025-09-25 15:34:50.851] [info] Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA RTX A4000 : cudaMallocAsync [2025-09-25 15:34:53.274] [info] Using pytorch attention [2025-09-25 15:34:55.910] [info] Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)] ComfyUI version: 0.3.60 [2025-09-25 15:34:55.947] [info] [Prompt Server] web root: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app [2025-09-25 15:34:57.994] [info] ### Loading: ComfyUI-Manager (V3.36) [2025-09-25 15:34:57.998] [info] [ComfyUI-Manager] network_mode: public ### ComfyUI Revision: UNKNOWN (The currently installed ComfyUI is not a Git repository) [2025-09-25 15:34:58.027] [info] Import times for custom nodes: [2025-09-25 15:34:58.030] [info] 0.0 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\websocket_image_save.py 0.0 seconds: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager [2025-09-25 15:34:58.260] [info] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json [2025-09-25 15:34:58.296] [info] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json [2025-09-25 15:34:58.368] [info] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json [2025-09-25 15:34:58.459] [info] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json [2025-09-25 15:34:58.553] [info] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [2025-09-25 15:34:59.131] [error] Failed to initialize database. Please ensure you have installed the latest requirements. If the error persists, please report this as in future the database will be required: (sqlite3.OperationalError) unable to open database file (Background on this error at: https://sqlalche.me/e/20/e3q8) [2025-09-25 15:34:59.155] [info] Starting server [2025-09-25 15:34:59.159] [info] To see the GUI go to: http://127.0.0.1:8000 [2025-09-25 15:35:00.828] [error] comfyui-frontend-package not found in requirements.txt [2025-09-25 15:35:01.767] [error] comfyui-frontend-package not found in requirements.txt [2025-09-25 15:35:18.935] [info] FETCH ComfyRegistry Data: 5/98 [2025-09-25 15:35:33.819] [info] FETCH ComfyRegistry Data: 10/98 [2025-09-25 15:35:48.372] [info] FETCH ComfyRegistry Data: 15/98 [2025-09-25 15:36:08.650] [info] FETCH ComfyRegistry Data: 20/98 [2025-09-25 15:36:30.720] [info] FETCH ComfyRegistry Data: 25/98 [2025-09-25 15:36:50.819] [info] FETCH ComfyRegistry Data: 30/98 [2025-09-25 15:37:20.452] [info] FETCH ComfyRegistry Data: 35/98 [2025-09-25 15:37:38.429] [info] FETCH ComfyRegistry Data: 40/98 [2025-09-25 15:37:52.542] [info] FETCH ComfyRegistry Data: 45/98 [2025-09-25 15:38:05.348] [info] FETCH ComfyRegistry Data: 50/98 [2025-09-25 15:38:32.769] [info] FETCH ComfyRegistry Data: 55/98 [2025-09-25 15:38:52.111] [info] FETCH ComfyRegistry Data: 60/98 [2025-09-25 15:39:13.232] [info] FETCH ComfyRegistry Data: 65/98 [2025-09-25 15:39:36.862] [info] FETCH ComfyRegistry Data: 70/98 [2025-09-25 15:40:00.289] [info] FETCH ComfyRegistry Data: 75/98 [2025-09-25 15:40:22.437] [info] FETCH ComfyRegistry Data: 80/98 [2025-09-25 15:40:41.965] [info] FETCH ComfyRegistry Data: 85/98 [2025-09-25 15:41:07.638] [info] FETCH ComfyRegistry Data: 90/98 [2025-09-25 15:41:31.580] [info] FETCH ComfyRegistry Data: 95/98 [2025-09-25 15:41:41.977] [info] FETCH ComfyRegistry Data [DONE] [2025-09-25 15:41:42.259] [info] [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes [2025-09-25 15:41:42.294] [info] FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json main.log [2025-09-25 15:34:41.256] [info] Using uv at C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe [2025-09-25 15:34:41.257] [info] Install state: installed [2025-09-25 15:34:41.259] [info] Validating installation. Recorded state: [installed] [2025-09-25 15:34:41.264] [info] Running uv command directly: pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt [2025-09-25 15:34:41.265] [info] Running uv child process: uv pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt [2025-09-25 15:34:41.266] [info] Running command: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt in C:\ComfyUI [2025-09-25 15:34:41.362] [info] Running uv command directly: pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt [2025-09-25 15:34:41.363] [info] Running uv child process: uv pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt [2025-09-25 15:34:41.363] [info] Running command: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\uv\win\uv.exe pip install --dry-run -r C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Manager\requirements.txt in C:\ComfyUI [2025-09-25 15:34:41.529] [info] Validation result: isValid:true, state:installed {"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"} [2025-09-25 15:34:42.345] [info] Server start [2025-09-25 15:34:42.378] [info] Running command: C:\ComfyUI\.venv\Scripts\python.exe C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\main.py --user-directory C:\ComfyUI\user --input-directory C:\ComfyUI\input --output-directory C:\ComfyUI\output --front-end-root C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app --base-directory C:\ComfyUI --extra-model-paths-config C:\Users\Administrator\AppData\Roaming\ComfyUI\extra_models_config.yaml --log-stdout --listen 127.0.0.1 --port 8000 in C:\ComfyUI [2025-09-25 15:34:42.412] [info] Received renderer-ready message! [2025-09-25 15:34:42.413] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.414] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.415] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.415] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.416] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.417] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.417] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.418] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.418] [info] Sending queued message {"channel":"validation-update","data":{"inProgress":false,"installState":"installed","basePath":"OK","venvDirectory":"OK","pythonInterpreter":"OK","uv":"OK","pythonPackages":"OK","git":"OK","vcRedist":"OK"}} [2025-09-25 15:34:42.419] [info] Sending queued message {"channel":"loading-progress","data":{"status":"starting-server"}} [2025-09-25 15:34:59.450] [info] Python server is ready [2025-09-25 15:35:12.341] [error] @todesktop/runtime: alwaysResolve: Promise timed out after 30000ms [2025-09-25 15:35:12.973] [info] @todesktop/runtime: Checking for update [2025-09-25 15:35:12.974] [info] @todesktop/runtime: AutoUpdater: checking-for-update [2025-09-25 15:35:15.442] [info] @todesktop/runtime: Update for version 0.4.74 is not available (latest version: 0.4.74, downgrade is allowed). [2025-09-25 15:35:15.444] [info] @todesktop/runtime: AutoUpdater: update-not-available {"version":"0.4.74","files":[{"url":"ComfyUI Setup 0.4.74 - Build 250924sn4vw8lfo-x64.exe","sha512":"EUB9h5eBMYCXAY1QeE84boGhZYF+lUSsFx4sZIGZnTCfU3qxiGkm7JS+BKTSoX8OU0gUBM9crSZHCANVd/o54A==","size":118282152}],"path":"ComfyUI Setup 0.4.74 - Build 250924sn4vw8lfo-x64.exe","sha512":"EUB9h5eBMYCXAY1QeE84boGhZYF+lUSsFx4sZIGZnTCfU3qxiGkm7JS+BKTSoX8OU0gUBM9crSZHCANVd/o54A==","releaseDate":"2025-09-24T02:25:54.886Z"} ``` ### Browser Logs none ### Settings JSON [comfy.settings.json](https://github.com/user-attachments/files/22531255/comfy.settings.json) ### Other _No response_ ┆Issue is synchronized with this [Notion page](https://www.notion.so/Issue-1346-Bug-flux1-kontext-dev-generates-images-executes-to-the-unet-loader-and-appears-to-2796d73d3650818eb9ffc38e65c1fd12) by [Unito](https://www.unito.io)

bsakaguchi 2026-10-31 22:12:12 +00:00

• closed this issue
• added the
  bug
  label

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 25, 2025):

The "Reconnecting" issue and execution hang at the Unet loader are most likely caused by missing or broken dependencies—especially the comfyui-frontend-package—in your Python virtual environment. This is a common problem after upgrading ComfyUI Desktop or installing/updating custom nodes, and is confirmed by the errors in your logs and recent community reports [docs].

To fix this, open the built-in terminal in the ComfyUI Desktop app and run:

pip install -r requirements.txt

Make sure you are running this inside the app's virtual environment, not your system Python. This will install all required dependencies, including the frontend package, and should resolve the "Reconnecting" state and allow workflows to execute past the Unet loader [docs].

If the issue persists, try these steps: disable all custom nodes, reset or delete the .venv folder to force a clean environment, back up and remove config files (like comfy.settings.json), and restart the app to let it regenerate fresh configs. These steps address corrupted environments and custom node conflicts, which are frequent causes of execution hangs [issue].

Database initialization errors (sqlite3.OperationalError) in your logs do not currently block startup but may affect future features; if you see persistent database errors, check for disk permission issues or clear the database files in your user directory [issue].

Let me know if you need more detailed instructions for any of these steps.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.

<!-- gh-comment-id:3332648734 --> @dosubot[bot] commented on GitHub (Sep 25, 2025): <!-- Answer --> The "Reconnecting" issue and execution hang at the Unet loader are most likely caused by missing or broken dependencies—especially the `comfyui-frontend-package`—in your Python virtual environment. This is a common problem after upgrading ComfyUI Desktop or installing/updating custom nodes, and is confirmed by the errors in your logs and recent community reports [[docs](https://docs.comfy.org/installation/update_comfyui)]. To fix this, open the built-in terminal in the ComfyUI Desktop app and run: ```bash pip install -r requirements.txt ``` Make sure you are running this inside the app's virtual environment, not your system Python. This will install all required dependencies, including the frontend package, and should resolve the "Reconnecting" state and allow workflows to execute past the Unet loader [[docs](https://docs.comfy.org/development/core-concepts/dependencies)]. If the issue persists, try these steps: disable all custom nodes, reset or delete the `.venv` folder to force a clean environment, back up and remove config files (like `comfy.settings.json`), and restart the app to let it regenerate fresh configs. These steps address corrupted environments and custom node conflicts, which are frequent causes of execution hangs [[issue](https://github.com/Comfy-Org/desktop/issues/1216)]. Database initialization errors (sqlite3.OperationalError) in your logs do not currently block startup but may affect future features; if you see persistent database errors, check for disk permission issues or clear the database files in your user directory [[issue](https://github.com/Comfy-Org/desktop/issues/1220#issuecomment-3062649469)]. Let me know if you need more detailed instructions for any of these steps. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Docs are dead. Just use [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=dcaf808d-ec45-4df0-a4d5-67ea7cb4f695) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-ask-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-share-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 25, 2025):

There is no requirements.txt in ComfyUI Desktop installation directory

<!-- gh-comment-id:3332891152 --> @qlxxkj commented on GitHub (Sep 25, 2025): There is no requirements.txt in ComfyUI Desktop installation directory

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 25, 2025):

ComfyUI Desktop doesn't put requirements.txt in the main installation directory. Instead, you'll find it in the resources/ComfyUI subdirectory (for example: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt on Windows). To install or repair dependencies, use the built-in terminal (which opens in the app's virtual environment) and run:

pip install -r resources/ComfyUI/requirements.txt

If you don't see requirements.txt there, your installation might be incomplete or corrupted—reinstalling the app usually restores it. For custom nodes, each has its own requirements.txt in its folder under custom_nodes, and ComfyUI Manager typically handles those automatically. More details are in the official docs and related support threads.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3332901613 --> @dosubot[bot] commented on GitHub (Sep 25, 2025): <!-- Answer --> ComfyUI Desktop doesn't put requirements.txt in the main installation directory. Instead, you'll find it in the resources/ComfyUI subdirectory (for example: C:\Users\Administrator\AppData\Local\Programs\ComfyUI\resources\ComfyUI\requirements.txt on Windows). To install or repair dependencies, use the built-in terminal (which opens in the app's virtual environment) and run: ```bash pip install -r resources/ComfyUI/requirements.txt ``` If you don't see requirements.txt there, your installation might be incomplete or corrupted—reinstalling the app usually restores it. For custom nodes, each has its own requirements.txt in its folder under custom_nodes, and ComfyUI Manager typically handles those automatically. More details are in the [official docs](https://docs.comfy.org/installation/desktop/windows) and [related support threads](https://github.com/Comfy-Org/desktop/issues/560). <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=47e92b8d-2c76-49d5-b4eb-dc49ad9dca40) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 25, 2025):

Is it because of the upgrade of pytorch? The current hardware cuda version is 12.4

<!-- gh-comment-id:3333011094 --> @qlxxkj commented on GitHub (Sep 25, 2025): Is it because of the upgrade of pytorch? The current hardware cuda version is 12.4

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 25, 2025):

Yes, the upgrade to PyTorch 2.8.0 with CUDA 12.9 support could be causing compatibility issues if your system only has CUDA 12.4 drivers. PyTorch CUDA builds are typically only guaranteed to work with the same or newer driver versions, so a mismatch can lead to runtime errors or hangs during execution—like the "Reconnecting" state you’re seeing [docs].

To check and resolve this:

1. Update your NVIDIA GPU drivers to the latest version from NVIDIA’s website.
2. Verify CUDA support in PyTorch by running these commands in your ComfyUI environment:

   python -c "import torch; print('CUDA available:', torch.cuda.is_available()); print('CUDA version:', torch.version.cuda)"
   

   If CUDA is not available or the version doesn’t match, you may need to reinstall PyTorch with the correct CUDA version for your hardware and drivers.

If you need to match PyTorch to your system’s CUDA version, you can uninstall torch and reinstall it with:

pip uninstall torch torchvision torchaudio
pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu124

(Replace cu124 with the CUDA version your drivers support.)

If you continue to see issues, double-check your driver version with:

nvidia-smi --query-gpu=driver_version --format=csv

and make sure it’s compatible with the CUDA version required by your PyTorch build [docs].

Let me know if you need help with any of these steps!

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3333026481 --> @dosubot[bot] commented on GitHub (Sep 25, 2025): <!-- Answer --> Yes, the upgrade to PyTorch 2.8.0 with CUDA 12.9 support could be causing compatibility issues if your system only has CUDA 12.4 drivers. PyTorch CUDA builds are typically only guaranteed to work with the same or newer driver versions, so a mismatch can lead to runtime errors or hangs during execution—like the "Reconnecting" state you’re seeing [[docs](https://docs.comfy.org/troubleshooting/overview)]. To check and resolve this: 1. Update your NVIDIA GPU drivers to the latest version from NVIDIA’s website. 2. Verify CUDA support in PyTorch by running these commands in your ComfyUI environment: ```bash python -c "import torch; print('CUDA available:', torch.cuda.is_available()); print('CUDA version:', torch.version.cuda)" ``` If CUDA is not available or the version doesn’t match, you may need to reinstall PyTorch with the correct CUDA version for your hardware and drivers. If you need to match PyTorch to your system’s CUDA version, you can uninstall torch and reinstall it with: ```bash pip uninstall torch torchvision torchaudio pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu124 ``` (Replace `cu124` with the CUDA version your drivers support.) If you continue to see issues, double-check your driver version with: ```bash nvidia-smi --query-gpu=driver_version --format=csv ``` and make sure it’s compatible with the CUDA version required by your PyTorch build [[docs](https://docs.comfy.org/troubleshooting/overview)]. Let me know if you need help with any of these steps! <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=6869fb07-9d55-4227-abe7-5f3cc4a121d4) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 26, 2025):

Comfy Destop automatically installs PyTorch 2.8.0+cu129 with the latest version. I updated my hardware's CUDA to 12.8, but that didn't fix the problem. It can be started normally, but it says "[ComfyUI-Manger] skipped fixing the 'comfyui-fronted-package' dependency because the comfyui is outdated."

<!-- gh-comment-id:3336457975 --> @qlxxkj commented on GitHub (Sep 26, 2025): Comfy Destop automatically installs PyTorch 2.8.0+cu129 with the latest version. I updated my hardware's CUDA to 12.8, but that didn't fix the problem. It can be started normally, but it says "[ComfyUI-Manger] skipped fixing the 'comfyui-fronted-package' dependency because the comfyui is outdated."

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 26, 2025):

The message about skipping the 'comfyui-frontend-package' dependency because "comfyui is outdated" means the Desktop app's validation logic detected that your ComfyUI core version or environment doesn't meet the requirements for safe dependency updates. This can happen if the core code, virtual environment, or requirements files are mismatched, incomplete, or not fully upgraded. As a result, ComfyUI-Manager won't attempt to fix or install frontend dependencies, which can break workflows and cause issues like the "Reconnecting" hang you described [docs].

To resolve this, make sure your core code and dependencies are fully updated and consistent. In the Desktop app's built-in terminal, run:

pip install -r resources/ComfyUI/requirements.txt

If that doesn't fix it, try deleting the .venv folder in your ComfyUI directory and restarting the app to force a clean environment and reinstall dependencies. If you still see the "outdated" message, your Desktop version may be lagging behind the latest core and dependency updates—this is a known limitation of the stable Desktop builds [docs]. For the latest features and fixes, consider using the portable or manual install versions, which track development updates more closely.

If you need to install a specific dependency manually (like comfyui-frontend-package), check the required version in resources/ComfyUI/requirements.txt and run:

pip install comfyui-frontend-package==<version>

Replace <version> with the version listed in the requirements file. This should help restore missing frontend functionality and allow your workflows to run past the Unet loader [docs].

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3336463395 --> @dosubot[bot] commented on GitHub (Sep 26, 2025): <!-- Answer --> The message about skipping the 'comfyui-frontend-package' dependency because "comfyui is outdated" means the Desktop app's validation logic detected that your ComfyUI core version or environment doesn't meet the requirements for safe dependency updates. This can happen if the core code, virtual environment, or requirements files are mismatched, incomplete, or not fully upgraded. As a result, ComfyUI-Manager won't attempt to fix or install frontend dependencies, which can break workflows and cause issues like the "Reconnecting" hang you described [[docs](https://docs.comfy.org/installation/update_comfyui)]. To resolve this, make sure your core code and dependencies are fully updated and consistent. In the Desktop app's built-in terminal, run: ```bash pip install -r resources/ComfyUI/requirements.txt ``` If that doesn't fix it, try deleting the `.venv` folder in your ComfyUI directory and restarting the app to force a clean environment and reinstall dependencies. If you still see the "outdated" message, your Desktop version may be lagging behind the latest core and dependency updates—this is a known limitation of the stable Desktop builds [[docs](https://docs.comfy.org/installation/update_comfyui)]. For the latest features and fixes, consider using the portable or manual install versions, which track development updates more closely. If you need to install a specific dependency manually (like comfyui-frontend-package), check the required version in `resources/ComfyUI/requirements.txt` and run: ```bash pip install comfyui-frontend-package==<version> ``` Replace `<version>` with the version listed in the requirements file. This should help restore missing frontend functionality and allow your workflows to run past the Unet loader [[docs](https://docs.comfy.org/installation/update_comfyui)]. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=ce4f00a8-d8d0-4beb-8c94-b40096d19736) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:13 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 26, 2025):

During the execution of the flux_kontex_dev_basic workflow, the log displays:

got prompt

Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
Requested to load AutoencodingEngine
loaded completely 11011.0 159.87335777282715 True
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4777.53759765625 True
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
clip missing: ['text_projection.weight']
……
Execution stops, and the interface displays "Reconnecting"

<!-- gh-comment-id:3336464797 --> @qlxxkj commented on GitHub (Sep 26, 2025): During the execution of the flux_kontex_dev_basic workflow, the log displays: got prompt Using pytorch attention in VAE Using pytorch attention in VAE VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16 Requested to load AutoencodingEngine loaded completely 11011.0 159.87335777282715 True Requested to load FluxClipModel_ loaded completely 9.5367431640625e+25 4777.53759765625 True CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16 clip missing: ['text_projection.weight'] …… Execution stops, and the interface displays "Reconnecting"

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 26, 2025):

got prompt
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
Requested to load AutoencodingEngine
loaded completely 11011.0 159.87335777282715 True
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4777.53759765625 True
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
clip missing: ['text_projection.weight']

D:\SD\ComfyUI_windows_portable>pause

<!-- gh-comment-id:3336917465 --> @qlxxkj commented on GitHub (Sep 26, 2025): got prompt Using pytorch attention in VAE Using pytorch attention in VAE VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16 Requested to load AutoencodingEngine loaded completely 11011.0 159.87335777282715 True Requested to load FluxClipModel_ loaded completely 9.5367431640625e+25 4777.53759765625 True CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16 clip missing: ['text_projection.weight'] D:\SD\ComfyUI_windows_portable>pause

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 26, 2025):

Now it cannot be repaired, I don't know why, whether I install comfyUI desktop version or ComfyUI_windows_portable.The terminal prints the same error message:

got prompt
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
Requested to load AutoencodingEngine
loaded completely 11011.0 159.87335777282715 True
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4777.53759765625 True
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
clip missing: ['text_projection.weight']

<!-- gh-comment-id:3337091015 --> @qlxxkj commented on GitHub (Sep 26, 2025): Now it cannot be repaired, I don't know why, whether I install comfyUI desktop version or ComfyUI_windows_portable.The terminal prints the same error message: got prompt Using pytorch attention in VAE Using pytorch attention in VAE VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16 Requested to load AutoencodingEngine loaded completely 11011.0 159.87335777282715 True Requested to load FluxClipModel_ loaded completely 9.5367431640625e+25 4777.53759765625 True CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16 clip missing: ['text_projection.weight']

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 26, 2025):

The error clip missing: ['text_projection.weight'] means the CLIP/text encoder model required by your workflow is either incomplete or incompatible. This isn't caused by your ComfyUI Desktop or portable installation, but by the model files themselves.

For Flux workflows, you must use the correct combination of models: Flux checkpoints require a specific VAE and two text encoders (CLIP-L and T5-XXL). If you use the wrong CLIP model, or if the file is corrupted or from a different architecture, you'll get missing weight errors like this one.

To fix it:

1. Check which CLIP/text encoder model your workflow is loading (look in the node or template settings).
2. Make sure you have the correct CLIP model files, downloaded completely from a trusted source (such as the official Flux Hugging Face release).
3. Place the models in the correct folders (models/checkpoints/, models/vae/, models/embeddings/).
4. If you still see the error, re-download the models to ensure they're not corrupted, and double-check that your workflow is referencing the right files.
5. If you use custom nodes, try disabling them and running a default workflow to isolate the issue.

More details and troubleshooting steps are in the official documentation and support threads. This is a common issue when mixing models from different series or using incomplete downloads.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3337109223 --> @dosubot[bot] commented on GitHub (Sep 26, 2025): <!-- Answer --> The error `clip missing: ['text_projection.weight']` means the CLIP/text encoder model required by your workflow is either incomplete or incompatible. This isn't caused by your ComfyUI Desktop or portable installation, but by the model files themselves. For Flux workflows, you must use the correct combination of models: Flux checkpoints require a specific VAE and two text encoders (CLIP-L and T5-XXL). If you use the wrong CLIP model, or if the file is corrupted or from a different architecture, you'll get missing weight errors like this one. To fix it: 1. Check which CLIP/text encoder model your workflow is loading (look in the node or template settings). 2. Make sure you have the correct CLIP model files, downloaded completely from a trusted source (such as the official Flux Hugging Face release). 3. Place the models in the correct folders (`models/checkpoints/`, `models/vae/`, `models/embeddings/`). 4. If you still see the error, re-download the models to ensure they're not corrupted, and double-check that your workflow is referencing the right files. 5. If you use custom nodes, try disabling them and running a default workflow to isolate the issue. More details and troubleshooting steps are in the [official documentation](https://docs.comfy.org/zh-CN/troubleshooting/model-issues) and [support threads](https://github.com/Comfy-Org/desktop/issues/698). This is a common issue when mixing models from different series or using incomplete downloads. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=a9f16da5-26cc-4d6a-8b24-4a24e4c5b79c) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 26, 2025):

I switched to flux_dev_fp8 workflow and got the same error. I also changed the Model Loader node to fp8_e4m3fn but it didn't work.

<!-- gh-comment-id:3337555994 --> @qlxxkj commented on GitHub (Sep 26, 2025): I switched to flux_dev_fp8 workflow and got the same error. I also changed the Model Loader node to fp8_e4m3fn but it didn't work.

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 26, 2025):

The persistent clip missing: ['text_projection.weight'] error in fp8 workflows (even after switching to flux_dev_fp8 and setting the Model Loader node to fp8_e4m3fn) means the CLIP/text encoder model file is incomplete, corrupted, or incompatible with the workflow. This is not caused by the workflow or node settings, but by the actual model files being loaded.

For fp8 Flux workflows, you need these exact models:

• clip_l.safetensors and t5xxl_fp8_e4m3fn.safetensors in models/text_encoders/
• ae.safetensors in models/vae/
• The correct fp8 diffusion model (like flux1-schnell.safetensors or flux1-krea-dev_fp8_scaled.safetensors) in models/diffusion_models/

Make sure you download all model files directly from trusted sources like Hugging Face, and verify their file sizes and hashes if possible. Place them in the correct folders and double-check that your workflow nodes reference the right files. If you still see the error, re-download the models to rule out corruption, and avoid mixing models from different series or using incomplete downloads [docs].

If you use custom nodes, disable them and run only the default workflow with the required models to isolate the issue. More troubleshooting details are in the official documentation and model setup guide.

Let me know if you need direct download links or help verifying your model files.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.

<!-- gh-comment-id:3337569473 --> @dosubot[bot] commented on GitHub (Sep 26, 2025): <!-- Answer --> The persistent `clip missing: ['text_projection.weight']` error in fp8 workflows (even after switching to flux_dev_fp8 and setting the Model Loader node to fp8_e4m3fn) means the CLIP/text encoder model file is incomplete, corrupted, or incompatible with the workflow. This is not caused by the workflow or node settings, but by the actual model files being loaded. For fp8 Flux workflows, you need these exact models: - `clip_l.safetensors` and `t5xxl_fp8_e4m3fn.safetensors` in `models/text_encoders/` - `ae.safetensors` in `models/vae/` - The correct fp8 diffusion model (like `flux1-schnell.safetensors` or `flux1-krea-dev_fp8_scaled.safetensors`) in `models/diffusion_models/` Make sure you download all model files directly from trusted sources like Hugging Face, and verify their file sizes and hashes if possible. Place them in the correct folders and double-check that your workflow nodes reference the right files. If you still see the error, re-download the models to rule out corruption, and avoid mixing models from different series or using incomplete downloads [[docs](https://docs.comfy.org/zh-CN/tutorials/flux/flux-1-text-to-image)]. If you use custom nodes, disable them and run only the default workflow with the required models to isolate the issue. More troubleshooting details are in the [official documentation](https://docs.comfy.org/zh-CN/tutorials/flux/flux1-krea-dev) and [model setup guide](https://docs.comfy.org/tutorials/flux/flux-1-kontext-dev). Let me know if you need direct download links or help verifying your model files. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Docs are dead. Just use [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=f8554f8d-e239-462b-8a21-e7cd9e91fad5) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-ask-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-share-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 27, 2025):

The previous version worked normally, it was not a problem with the model

<!-- gh-comment-id:3341213928 --> @qlxxkj commented on GitHub (Sep 27, 2025): The previous version worked normally, it was not a problem with the model

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 27, 2025):

You're right—the problem isn't with your model files if they worked in the previous version. This kind of regression after an upgrade is usually caused by a mismatch between the ComfyUI Desktop (stable) version and the requirements of advanced workflows like Flux.1 Kontext Dev. The Desktop version often lags behind the Nightly/development branch, missing critical updates or nodes needed for these workflows, which leads to errors like clip missing: ['text_projection.weight'] even when your models are correct [docs].

To fix this, update to the latest Nightly version of ComfyUI (not just the Desktop stable build), reset or recreate your virtual environment, and reinstall dependencies. If you use custom nodes, disable them to rule out conflicts. These steps resolve most upgrade-related issues where the environment or dependencies become inconsistent [troubleshooting].

This is a known limitation of the Desktop app—advanced workflows may require features or fixes only available in the Nightly branch [model issues]. For future upgrades, back up your configs and consider tracking Nightly releases if you rely on cutting-edge workflows.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3341215831 --> @dosubot[bot] commented on GitHub (Sep 27, 2025): <!-- Answer --> You're right—the problem isn't with your model files if they worked in the previous version. This kind of regression after an upgrade is usually caused by a mismatch between the ComfyUI Desktop (stable) version and the requirements of advanced workflows like Flux.1 Kontext Dev. The Desktop version often lags behind the Nightly/development branch, missing critical updates or nodes needed for these workflows, which leads to errors like `clip missing: ['text_projection.weight']` even when your models are correct [[docs](https://docs.comfy.org/tutorials/flux/flux-1-kontext-dev)]. To fix this, update to the latest Nightly version of ComfyUI (not just the Desktop stable build), reset or recreate your virtual environment, and reinstall dependencies. If you use custom nodes, disable them to rule out conflicts. These steps resolve most upgrade-related issues where the environment or dependencies become inconsistent [[troubleshooting](https://docs.comfy.org/troubleshooting/overview)]. This is a known limitation of the Desktop app—advanced workflows may require features or fixes only available in the Nightly branch [[model issues](https://docs.comfy.org/troubleshooting/model-issues)]. For future upgrades, back up your configs and consider tracking Nightly releases if you rely on cutting-edge workflows. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=12b4b161-bc27-4fc0-80bd-0596654c1384) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 27, 2025):

Where can I download the latest Nightly version? @dosu

<!-- gh-comment-id:3341307419 --> @qlxxkj commented on GitHub (Sep 27, 2025): Where can I download the latest Nightly version? @dosu

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 27, 2025):

The latest Nightly (development) version of ComfyUI is not available as a Desktop installer. To get Nightly, use either the Portable version (Windows only) or Manual installation:

• For Windows Portable: Download the Portable package (see the official Portable guide), then run the update script update/update_comfyui.bat in the extracted folder to switch to the latest Nightly build. This keeps your install up to date with the newest features and fixes details.

• For Manual install (all platforms): Clone the repo with

  git clone https://github.com/comfyanonymous/ComfyUI.git
  

  then run

  pip install -r requirements.txt
  python main.py
  

  in the ComfyUI directory. This always gives you the latest Nightly code manual install steps.

The Desktop app only tracks stable releases, so it may lag behind Nightly and miss critical updates needed for advanced workflows see docs.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

<!-- gh-comment-id:3341308122 --> @dosubot[bot] commented on GitHub (Sep 27, 2025): <!-- Answer --> The latest Nightly (development) version of ComfyUI is not available as a Desktop installer. To get Nightly, use either the Portable version (Windows only) or Manual installation: - For Windows Portable: Download the Portable package (see the official [Portable guide](https://docs.comfy.org/installation/comfyui_portable_windows)), then run the update script `update/update_comfyui.bat` in the extracted folder to switch to the latest Nightly build. This keeps your install up to date with the newest features and fixes [details](https://docs.comfy.org/installation/update_comfyui). - For Manual install (all platforms): Clone the repo with ``` git clone https://github.com/comfyanonymous/ComfyUI.git ``` then run ``` pip install -r requirements.txt python main.py ``` in the ComfyUI directory. This always gives you the latest Nightly code [manual install steps](https://docs.comfy.org/installation/manual_install). The Desktop app only tracks stable releases, so it may lag behind Nightly and miss critical updates needed for advanced workflows [see docs](https://docs.comfy.org/installation/update_comfyui). <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Share context across your team and agents. Try [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=afa83da4-b234-445a-9817-be34a34c44ec) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-learn-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=knowledge-infrastructure-add-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:14 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 27, 2025):

The latest Nightly (development) version of ComfyUI is also have the error. @dosu
Here is logs:

got prompt
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
Requested to load AutoencodingEngine
loaded completely 11011.0 159.87335777282715 True
Using scaled fp8: fp8 matrix mult: False, scale input: False
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4903.231597900391 True
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
clip missing: ['text_projection.weight']

<!-- gh-comment-id:3341337227 --> @qlxxkj commented on GitHub (Sep 27, 2025): The latest Nightly (development) version of ComfyUI is also have the error. @dosu Here is logs: -------------------------------------- got prompt Using pytorch attention in VAE Using pytorch attention in VAE VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16 Requested to load AutoencodingEngine loaded completely 11011.0 159.87335777282715 True Using scaled fp8: fp8 matrix mult: False, scale input: False Requested to load FluxClipModel_ loaded completely 9.5367431640625e+25 4903.231597900391 True CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16 clip missing: ['text_projection.weight']

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Sep 27, 2025):

The clip missing: ['text_projection.weight'] error means the CLIP/text encoder model loaded by your workflow is missing a required parameter, usually because of a model architecture mismatch or an incorrect model file—even if the same files worked before. For Flux workflows, you must use both clip_l.safetensors and t5xxl_fp8_e4m3fn.safetensors (or t5xxl_fp16.safetensors) in the DualCLIP node, with each in its correct slot. If you accidentally load the same model in both slots, or use a model from a different series, you’ll get this error. Double-check your node configuration and model paths, and re-download the models from trusted sources if needed. Also, make sure you’re not mixing models from different architecture families in the same workflow details setup guide.

If you’re sure the models and workflow are correct, try disabling all custom nodes and running only the default workflow to rule out conflicts. If the error persists after a Nightly upgrade, it’s likely a configuration or model compatibility issue, not a bug in the ComfyUI core.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.

<!-- gh-comment-id:3341338161 --> @dosubot[bot] commented on GitHub (Sep 27, 2025): <!-- Answer --> The `clip missing: ['text_projection.weight']` error means the CLIP/text encoder model loaded by your workflow is missing a required parameter, usually because of a model architecture mismatch or an incorrect model file—even if the same files worked before. For Flux workflows, you must use both `clip_l.safetensors` and `t5xxl_fp8_e4m3fn.safetensors` (or `t5xxl_fp16.safetensors`) in the DualCLIP node, with each in its correct slot. If you accidentally load the same model in both slots, or use a model from a different series, you’ll get this error. Double-check your node configuration and model paths, and re-download the models from trusted sources if needed. Also, make sure you’re not mixing models from different architecture families in the same workflow [details](https://docs.comfy.org/zh-CN/troubleshooting/model-issues) [setup guide](https://docs.comfy.org/tutorials/flux/flux-1-kontext-dev). If you’re sure the models and workflow are correct, try disabling all custom nodes and running only the default workflow to rule out conflicts. If the error persists after a Nightly upgrade, it’s likely a configuration or model compatibility issue, not a bug in the ComfyUI core. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=reply-with-mention&utm_term=Comfy-Org%2Fdesktop).* --- Docs are dead. Just use [Dosu](https://dosu.dev?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-tagline&utm_term=Comfy-Org%2Fdesktop). [](https://app.dosu.dev/response-feedback?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-feedback&utm_term=Comfy-Org%2Fdesktop#message_id=94dca576-f848-4242-b028-fb83fdc25e81) [](https://github.dosu.com/Comfy-Org/desktop?utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-ask-repo&utm_term=Comfy-Org%2Fdesktop) [](https://app.dosu.dev/signup?referrer=openSource&source=github-footer&utm_source=github&utm_medium=bot-comment&utm_campaign=github-comment-footer-20260415&utm_content=static-docs-share-team&utm_term=Comfy-Org%2Fdesktop)

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@webfiltered commented on GitHub (Sep 28, 2025):

@qlxxkj What GPU do you have? You said you updated to CUDA 12.8; was that because you CAN'T update to 12.9? If you're on a 10 series card, you just need to use an older version of pytorch:

# PyTorch 2.8 CUDA 12.8
uv pip install torch==2.8.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128
# or 2.7 CUDA 12.8
uv pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

<!-- gh-comment-id:3344096703 --> @webfiltered commented on GitHub (Sep 28, 2025): @qlxxkj What GPU do you have? You said you updated to CUDA 12.8; was that because you CAN'T update to 12.9? If you're on a 10 series card, you just need to use an older version of pytorch: ```bash # PyTorch 2.8 CUDA 12.8 uv pip install torch==2.8.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128 # or 2.7 CUDA 12.8 uv pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128 ```

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Sep 30, 2025):

1. When the desktop version prompted a new version, I clicked "Upgrade Now" to the latest version (0.4.74).
2. The flux_kontext_fp8_basic workflow executed to the unet loader, and the interface prompted "Reconnecting."
3. The log displayed "clip missing: ['text_projection.weight']."
4. I checked the desktop version of PyTorch (2.8.0+cu129) and the graphics driver version (CUDA 12.4). I think there might be a mismatch.
5. I upgraded the RTX A4000 graphics driver to the latest version (CUDA 12.8).
6. I executed the flux_kontext_fp8_basic workflow again, and the same issue occurred (reconnecting).
7. I rolled back the desktop version from 0.4.74 to 0.4.45, and the same issue occurred.
8. I rolled back the graphics driver version (CUDA 12.4), and the same issue occurred.

<!-- gh-comment-id:3349763414 --> @qlxxkj commented on GitHub (Sep 30, 2025): 1. When the desktop version prompted a new version, I clicked "Upgrade Now" to the latest version (0.4.74). 2. The flux_kontext_fp8_basic workflow executed to the unet loader, and the interface prompted "Reconnecting." 3. The log displayed "clip missing: ['text_projection.weight']." 4. I checked the desktop version of PyTorch (2.8.0+cu129) and the graphics driver version (CUDA 12.4). I think there might be a mismatch. 5. I upgraded the RTX A4000 graphics driver to the latest version (CUDA 12.8). 6. I executed the flux_kontext_fp8_basic workflow again, and the same issue occurred (reconnecting). 7. I rolled back the desktop version from 0.4.74 to 0.4.45, and the same issue occurred. 8. I rolled back the graphics driver version (CUDA 12.4), and the same issue occurred. <img width="976" height="409" alt="Image" src="https://github.com/user-attachments/assets/52bc8194-fbf4-46e5-8fbf-5f2805eed7ea" />

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@webfiltered commented on GitHub (Sep 30, 2025):

I would definitely try one of the commands I suggested. You can run either one from inside Comfy, then restart the app once done. The issue is unlikely to be related your graphics driver.

<!-- gh-comment-id:3349819291 --> @webfiltered commented on GitHub (Sep 30, 2025): I would definitely try one of the commands I suggested. You can run either one from inside Comfy, then restart the app once done. The issue is unlikely to be related your graphics driver.

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Oct 9, 2025):

@qlxxkj What GPU do you have? You said you updated to CUDA 12.8; was that because you CAN'T update to 12.9? If you're on a 10 series card, you just need to use an older version of pytorch:

PyTorch 2.8 CUDA 12.8

uv pip install torch==2.8.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

or 2.7 CUDA 12.8

uv pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

---

I can't fix it after running either one.

<!-- gh-comment-id:3384323327 --> @qlxxkj commented on GitHub (Oct 9, 2025): > [@qlxxkj](https://github.com/qlxxkj) What GPU do you have? You said you updated to CUDA 12.8; was that because you CAN'T update to 12.9? If you're on a 10 series card, you just need to use an older version of pytorch: > > # PyTorch 2.8 CUDA 12.8 > uv pip install torch==2.8.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128 > # or 2.7 CUDA 12.8 > uv pip install torch==2.7.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128 ----------------------------------------------------- I can't fix it after running either one.

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Oct 10, 2025):

How to use pytorch (2.5.1+cu121) in the latest version of ComfyUI desktop? After installing pytorch (2.5.1+cu121) in the comfyui virtual environment, the desktop application still uses pytorch (2.8.0+cu129) when opening it. @webfiltered

<!-- gh-comment-id:3388350777 --> @qlxxkj commented on GitHub (Oct 10, 2025): How to use pytorch (2.5.1+cu121) in the latest version of ComfyUI desktop? After installing pytorch (2.5.1+cu121) in the comfyui virtual environment, the desktop application still uses pytorch (2.8.0+cu129) when opening it. @webfiltered

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Oct 10, 2025):

Each time a workflow is executed and a reconnect occurs, the following Windows system log appears:
System event log: Windows successfully diagnosed a low virtual memory situation. The following programs used the most virtual memory: python.exe (11896) used 25301602304 bytes; ComfyUI.exe (6964) used 358973440 bytes; MsMpEng.exe (4412) used 330895360 bytes.

<!-- gh-comment-id:3388534148 --> @qlxxkj commented on GitHub (Oct 10, 2025): Each time a workflow is executed and a reconnect occurs, the following Windows system log appears: System event log: Windows successfully diagnosed a low virtual memory situation. The following programs used the most virtual memory: python.exe (11896) used 25301602304 bytes; ComfyUI.exe (6964) used 358973440 bytes; MsMpEng.exe (4412) used 330895360 bytes.

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@qlxxkj commented on GitHub (Oct 10, 2025):

I've fixed this issue. According to the Windows system event log, I found that the "Reconnecting" error was caused by insufficient system memory. Perhaps a workflow execution caused a memory overflow, leading to a memory failure. Reinserting the two 16GB memory sticks fixed the issue. @webfiltered

<!-- gh-comment-id:3389046175 --> @qlxxkj commented on GitHub (Oct 10, 2025): I've fixed this issue. According to the Windows system event log, I found that the "Reconnecting" error was caused by insufficient system memory. Perhaps a workflow execution caused a memory overflow, leading to a memory failure. Reinserting the two 16GB memory sticks fixed the issue. @webfiltered

bsakaguchi commented 2026-10-31 22:12:15 +00:00

Author

Owner

Copy link

@dosubot[bot] commented on GitHub (Jan 9, 2026):

Hi, @qlxxkj. I'm Dosu (https://dosu.dev), and I'm helping the desktop team manage their backlog and am marking this issue as stale.

Issue Summary:

• You reported that after upgrading to ComfyUI desktop 0.4.74, running flux_kontex_dev_basic stalled at the UNet loader with a "Reconnecting" message.
• Logs indicated a missing "clip" component: ['text_projection.weight'], suggesting possible model or dependency issues.
• Despite attempts to fix dependencies, PyTorch versions, and driver updates, the problem persisted even on Nightly builds.
• You ultimately identified insufficient system memory as the root cause, and reseating the RAM sticks resolved the issue.

Next Steps:

• Please confirm if this issue is still relevant with the latest version of the ComfyUI desktop.
• If it is, feel free to comment to keep the discussion open; otherwise, I will automatically close this issue in 7 days.

Thanks for your understanding and contribution!

<!-- gh-comment-id:3729544725 --> @dosubot[bot] commented on GitHub (Jan 9, 2026): Hi, @qlxxkj. I'm Dosu (https://dosu.dev), and I'm helping the desktop team manage their backlog and am marking this issue as stale. **Issue Summary:** - You reported that after upgrading to ComfyUI desktop 0.4.74, running flux_kontex_dev_basic stalled at the UNet loader with a "Reconnecting" message. - Logs indicated a missing "clip" component: ['text_projection.weight'], suggesting possible model or dependency issues. - Despite attempts to fix dependencies, PyTorch versions, and driver updates, the problem persisted even on Nightly builds. - You ultimately identified insufficient system memory as the root cause, and reseating the RAM sticks resolved the issue. **Next Steps:** - Please confirm if this issue is still relevant with the latest version of the ComfyUI desktop. - If it is, feel free to comment to keep the discussion open; otherwise, I will automatically close this issue in 7 days. Thanks for your understanding and contribution!

bsakaguchi referenced this issue 2026-10-31 22:17:15 +00:00

[PR #374] [MERGED] Validate hardware requirements during install time. #828

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

main

chore/archive-notice

deepme987/desktop/deprecation-banner

bl/bump-comfyui-0-22-3

add-comfy-environment-marker

bl/missing-model-download-contract

bl/bump-comfyui-0.20.2

sno-proxysettings

bl/missing-model-download-ui-consume-fix

bl/missing-model-download-ui-test

feat/deep-link-protocol-handler

ben/oem-machine-scope-install-state

bl/fix-missing-model-download-path-e2e

bump-workflow-templates

chore/example-pr-close

chore/bump-frontend-1.39.19-r2

fix/frontend-optional-branch-core-1-39

fwra

chore/frontend-override-backport-9185-core-1.39

bump-comfyui-v0.15.0-package-json-only

ben/auto-bump-comfyui-release-optional-main-commit

bump-uv-v0.10.4-and-macos-resolution

ben/auto-bump-comfyui-release

fix/sentry-gpu-driver-reporting

fix/close-confirmation-ipc

fix/desktop-frontend-sourcemaps

revert-1565-chore/pytorch-2-10

chore/deps-rollup-2026-01-29

fix/macos-torch-index

fix/disable-forced-pytorch-upgrade

fix/rename-upgrade-packages-to-needs-package-update

fix/venv-troubleshooting-e2e

update-comfyui-core-0.9.2

torch-update-prompt

issue-1534-nvidia-driver-warning

revert-core-comfyui-0-9-1

bl-canary-release

bl-flaky-version-mask

chore/pytorch-2.9.1-cu130

bl-fix-amd-aarch-warning

bl-shaggy-urial

fix-todesktop-workflow

fix/self-heal-missing-dirs

bl-aggregate-ape

bl-interesting-warbler

bl-dramatic-opossum

bl-northern-whippet

bl-update-deps

copilot/fix-1228

build-head-20251001

custom-dialog

parse-uv-v

api-clean-up

parse-uv

release-action

uninstaller-options

full-installer

bundle-git-installer

vcpp-installer

uninstall-cleanup

update-bump-stable-command

frontend-optional-branch-v3-interface-menu

system-pop-up

feat-custom-launch

publish/private

macos-install-path

0.4.40

model_file_import

rh-fix3

robinjhuang-patch-1

rh-manager2

remove-redundant-ci

e2e-bad-start

version-bump-2025-02-20-1740010810152

test/beforeunload-confirm

e2e2-2

debug-log-events

rh-nodes

cb-store

clean-comfy-server

lumina-release

release-2025-feb06

rh-tsx

fix_setting_ids

use_npm

default-no-reload-page

first-time-user-setting

window-style-change

readme-dev-linux

virtual-validation-testing

app-state

enh/Add-Check-For-Updates-button

git-installer

enh/Save-user-preference-for-crashes-after-confirmation-dialog

enh/include-vcredist

error-handling

dev-ken

dev-kendal-todesktop-securit

dev-kendal-startup-test

dev-squirrel-uninstall

dev-kendal

v0.9.4

v0.9.3

v0.9.2

v0.9.1

v0.9.0

v0.8.36

v0.8.35

v0.8.34

v0.8.33

v0.8.32

v0.8.31

v0.8.30

v0.8.29

v0.8.28

v0.8.27

v0.8.26

v0.8.25

v0.8.24

v0.8.23

v0.8.22

v0.8.21

v0.8.20

v0.8.19

v0.8.18

v0.8.16

v0.8.15

v0.8.14

v0.8.13

v0.8.12

v0.8.11

v0.8.10

v0.8.9

v0.8.8

v0.8.7

v0.8.6

v0.8.5

v0.8.4

v0.8.3

v0.8.2

v0.8.1

v0.8.0

v0.7.5

v0.7.4

v0.7.3

v0.7.2

v0.7.1

v0.7.0

v0.6.2

v0.6.1

v0.6.0

v0.5.12

v0.5.11

v0.5.10

v0.5.9

v0.5.8

v0.5.7

v0.5.6

v0.5.5

v0.5.4

v0.5.3

v0.5.2

v0.5.1

v0.5.0

v0.4.76

v0.4.75

v0.4.74

v0.4.73

v0.4.72

v0.4.71

v0.4.70

v0.4.69

v0.4.68

v0.4.67

v0.4.66

v0.4.65

v0.4.64

v0.4.63

v0.4.62

v0.4.61

v0.4.60

v0.4.59

v0.4.58

v0.4.57

v0.4.56

v0.4.55

v0.4.54

v0.4.53

v0.4.52

v0.4.51

v0.4.50

v0.4.49

v0.4.48

v0.4.47

v0.4.46

v0.4.45

v0.4.44

v0.4.41

v0.4.40

v0.4.39

v0.4.36

v0.4.35

v0.4.34

v0.4.33

v0.4.32

v0.4.30

v0.4.29

v0.4.28

v0.4.27

v0.4.26

v0.4.25

v0.4.24

v0.4.23

v0.4.22

v0.4.21

v0.4.20

v0.4.19

v0.4.18

v0.4.17

v0.4.16

v0.4.13

v0.4.12

v0.4.7

v0.4.6

v0.4.5

v0.4.4

v0.4.3

v0.3.18

v0.4.2

Labels

Clear labels   

area: desktop-app

  

Bounty

  

bug

  

bug-cop:answered

  

bug-cop:response-received

  

component: installer

  

component: preload

  

custom node

  

dependencies

  

Design

  

documentation

  

duplicate

  

enhancement

  

expected

  

frontend

  

good first issue

  

installation

  

Low-Priority

  

performance

  

Planned

  

platform: cross-platform

  

platform: windows

  

Potential Bug

  

Potential Bug

  

pull-request

Mirrored from GitHub Pull Request

  

question

  

refactor

  

Release

  

stale

  

User Support

  

V1

No labels

area: desktop-app

Bounty

bug

bug-cop:answered

bug-cop:response-received

component: installer

component: preload

custom node

dependencies

Design

documentation

duplicate

enhancement

expected

frontend

good first issue

installation

Low-Priority

performance

Planned

platform: cross-platform

platform: windows

Potential Bug

Potential Bug

pull-request

question

refactor

Release

stale

User Support

V1

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

Comfy-Org/desktop#374

No description provided.