Skip to content

FAQ

Jump to bottom
AbdBarho edited this page Mar 8, 2023 · 36 revisions

General

Unfortunately, AMD GPUs #63 and Mac #35 are not supported, contributions to add support are very welcome.

Dockerfile parse error

Error response from daemon: dockerfile parse error line 33: unknown instruction: GIT
ERROR: Service 'model' failed to build : Build failed

Update docker to the latest version, and make sure you are using docker compose instead of docker-compose. #16, also, try setting the environment variable DOCKER_BUILDKIT=1

Unknown Flag --profile

Update docker to the latest version, and see this comment, try setting the environment variable mentioned in the previous point.

Output is a always green image

use --precision full --no-half. #9

Found no NVIDIA driver on your system even though the drivers are installed and nvidia-smi shows it

add NVIDIA_DRIVER_CAPABILITIES=compute,utility and NVIDIA_VISIBLE_DEVICES=all to container can resolve this problem #348

Linux

Error response from daemon: could not select device driver "nvidia" with capabilities: [[gpu]]

Install NVIDIA Container Toolkit and restart the docker service #81

Windows / WSL

Build fails at The Shell command, /bin/bash not found in WSL.

Edit the corresponding docker file, and change the SHELL from /bin/bash to //bin/bash #21, note: this is a hack and something in your wsl is messed up.

Build fails with credentials errors when logged in via SSH on WSL2/Windows

You can try forcing plain text auth creds storage by removing line with "credStore" from ~/.docker/config.json (in WSL). #56

unable to access 'https://github.com/...': Could not resolve host: github.com or any domain

Set the build/network of the service you are starting to host #114

Other build errors on windows

  • Make sure:
    • Windows 10 release >= 2021H2 (required for WSL to see the GPU)
    • WSL2 (check with wsl -l -v)
    • Latest Docker Desktop
  • You might need to create a .wslconfig and increase memory, if you have 16GB RAM, set the limit to something around 12GB, #34 #64
  • You might also need to force wsl to allow file permissions

AWS

You have to use one of AWS's GPU-enabled VMs and their Deep Learning OS images. These have the right divers, the toolkit and all the rest already installed and optimized. #70

Clone this wiki locally