docker compose up error in label-studio-ml-backend\label_studio_ml\examples\yolo

[dvdface]

Anybody can help me ?

D:\Documents\Github\label-studio-ml-backend\label_studio_ml\examples\yolo>docker compose up --build
time="2025-06-07T10:20:36+08:00" level=warning msg="The \"TEST_ENV\" variable is not set. Defaulting to a blank string."
time="2025-06-07T10:20:36+08:00" level=warning msg="D:\\Documents\\Github\\label-studio-ml-backend\\label_studio_ml\\examples\\yolo\\docker-compose.yml: the attribute `version` is obsolete, it will be ignored, please remove it to avoid potential confusion"
Compose can now delegate builds to bake for better performance.
 To do so, set COMPOSE_BAKE=true.
[+] Building 21.7s (3/3) FINISHED                                                                                                                                                                                                                        docker:desktop-linux
 => [yolo internal] load build definition from Dockerfile                                                                                                                                                                                                                0.1s
 => => transferring dockerfile: 1.98kB                                                                                                                                                                                                                                   0.0s
 => ERROR [yolo internal] load metadata for docker.io/pytorch/pytorch:2.7.1-cuda12.8-cudnn9-runtime                                                                                                                                                                     21.5s
 => [yolo auth] pytorch/pytorch:pull token for registry-1.docker.io                                                                                                                                                                                                      0.0s
------
 > [yolo internal] load metadata for docker.io/pytorch/pytorch:2.7.1-cuda12.8-cudnn9-runtime:
------
failed to solve: failed to fetch oauth token: Post "https://auth.docker.io/token": dial tcp [2a03:2880:f134:183:face:b00c:0:25de]:443: connectex: A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

[heidi-humansignal]

Hello,

Thank you for contacting Label Studio,

Based on the error message, it looks like Docker is attempting to pull the PyTorch image from Docker Hub but is failing to connect. Notably, Docker is trying to connect via an IPv6 address, which can sometimes cause connectivity issues depending on your network configuration. : Please find some suggestions that could help you resolve this issue:

Try Manually Pulling the Image
Run this command to see if the issue persists outside of Compose:

docker pull pytorch/pytorch:2.7.1-cuda12.8-cudnn9-runtime

2. Temporarily Disable VPN/Proxy (if applicable)
   If you're using a VPN or corporate proxy, try disabling it to rule out interference.
3. Force Docker to Use IPv4
   Docker may be defaulting to IPv6. To switch to IPv4:

• Open (or create) this file:
  C:\Users<your_username>\.docker\config.json

Add or modify it like this:

{ "dns": ["8.8.8.8"], "experimental": false }

• Then restart Docker Desktop.

Flush DNS Cache (on Windows)
Open Command Prompt and run:

ipconfig /flushdns

Let me know if the issue persists after trying these steps, and I’ll be happy to continue helping you debug!

Comment by Oussama Assili
Workflow Run