Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

AllTalk V2 Docker builds broken? #510

Open
icsy7867 opened this issue Feb 5, 2025 · 6 comments
Open

AllTalk V2 Docker builds broken? #510

icsy7867 opened this issue Feb 5, 2025 · 6 comments

Comments

@icsy7867
Copy link

icsy7867 commented Feb 5, 2025

Describe the bug
The Docker build for v2 doesn't seem to work. It seems to run, but then seems to hang with a conda message to either start or activate the environment during the build.

To Reproduce
I am using Fedora CoreOs, and installed the buildx docker plug-in. Then I run the build bash script.

Screenshots
Will provide shortly!

Text/logs
I will add the exact text shortly!
Ultimately it seems to run for several minutes until it creates a conda environment, then the build seems to hang on a few lines about conda and how to activate the environment. It seems to hang here indefinitely.

Desktop (please complete the following information):
N/A - But i did a git clone today and tried this morning.

Additional context
It would be great if the build didn't rely on buildx. There are many build tools (podman, buildah, kaniko) that would be nice to be able to use.

@erew123
Copy link
Owner

erew123 commented Feb 10, 2025

Hi this is in process the moment and is being worked on by someone external to myself. Discussion is here #378 and I know a new build is in the offing soon.

Thanks

@Demannu
Copy link

Demannu commented Feb 10, 2025

The docker build step for Conda is failing due to the success check logic. There appears to be a new dependency that contains the word error, likely introduced as a 3rd party dependency.

The current logic in /docker/conda/Dockerfile is:

    if echo $RESULT | grep -izq "error" ; then
      echo "Failed to install conda dependencies: $RESULT"
      exit 1
    fi

Which will now fail due to libgpg-error being in the package list.

I'm validating a different way to verify the conda success, after that I'll put up an MR/PR fixing this.

If you are comfortable editing code, you can can comment out that logic entirely and skip the validation for now to get the docker image building.

@RenNagasaki
Copy link

@Demannu is the problem still around?
I was trying the build on my wsl and got this:

`120.1 Failed to install conda dependencies:

Dockerfile:38

37 | ##############################################################################
38 | >>> RUN <<EOR
39 | >>> CUDA_SHORT_VERSION=${CUDA_VERSION%.*}
40 | >>> conda create -y -n "alltalk" -c conda-forge python=${PYTHON_VERSION}
41 | >>> conda activate alltalk
42 | >>> conda install -y
43 | >>> gcc_linux-64
44 | >>> gxx_linux-64
45 | >>> libaio
46 | >>> faiss-gpu=1.9.0
47 | >>> cuda=${CUDA_VERSION} cuda-tools=${CUDA_VERSION} cuda-toolkit=${CUDA_VERSION} cuda-version=${CUDA_SHORT_VERSION}
48 | >>> cuda-command-line-tools=${CUDA_VERSION} cuda-compiler=${CUDA_VERSION} cuda-runtime=${CUDA_VERSION}
49 | >>> cuda-libraries=${CUDA_VERSION} cuda-libraries-dev=${CUDA_VERSION} cuda-libraries-static=${CUDA_VERSION}
50 | >>> cuda-cudart=${CUDA_SHORT_VERSION} cuda-cudart_linux-64=${CUDA_SHORT_VERSION}
51 | >>> cudnn=9.3
52 | >>> py-cpuinfo=9
53 | >>> conda-forge::ffmpeg=7.1.0
54 | >>> conda-forge::portaudio=19.7.0
55 | >>> -c pytorch
56 | >>> -c nvidia
57 | >>> -c anaconda
58 | >>>
59 | >>> if [ $? -ne 0 ] ; then
60 | >>> echo "Failed to install conda dependencies: $RESULT"
61 | >>> exit 1
62 | >>> fi
63 | >>>
64 | >>> conda clean -a && pip cache purge
65 | >>> EOR
66 |

ERROR: failed to solve: process "/bin/bash -l -c CUDA_SHORT_VERSION=${CUDA_VERSION%.*}\n conda create -y -n "alltalk" -c conda-forge python=${PYTHON_VERSION}\n conda activate alltalk\n conda install -y \\n gcc_linux-64 \\n gxx_linux-64 \\n libaio \\n faiss-gpu=1.9.0 \\n cuda=${CUDA_VERSION} cuda-tools=${CUDA_VERSION} cuda-toolkit=${CUDA_VERSION} cuda-version=${CUDA_SHORT_VERSION} \\n cuda-command-line-tools=${CUDA_VERSION} cuda-compiler=${CUDA_VERSION} cuda-runtime=${CUDA_VERSION} \\n cuda-libraries=${CUDA_VERSION} cuda-libraries-dev=${CUDA_VERSION} cuda-libraries-static=${CUDA_VERSION} \\n cuda-cudart=${CUDA_SHORT_VERSION} cuda-cudart_linux-64=${CUDA_SHORT_VERSION} \\n cudnn=9.3 \\n py-cpuinfo=9 \\n conda-forge::ffmpeg=7.1.0 \\n conda-forge::portaudio=19.7.0 \\n -c pytorch \\n -c nvidia \\n -c anaconda\n\n if [ $? -ne 0 ] ; then\n echo "Failed to install conda dependencies: $RESULT"\n exit 1\n fi\n\n conda clean -a && pip cache purge\n" did not complete successfully: exit code: 1`

@Demannu
Copy link

Demannu commented Feb 24, 2025

I believe the initial error I reported is resolved, your error suggests conda did not finish installing successfully and exited as a result.

#514 was merged which should've resolved the incorrect check logic.

@Paladinium
Copy link
Collaborator

Paladinium commented Feb 28, 2025

@icsy7867 : can pull the latest changes please and retry?

In addition, the images are now being pushed to Docker hub, so local builds are not needed anymore. See #378 (comment)

Please close this issue if it works for you.

@icsy7867
Copy link
Author

Might be Monday before I can. But I'll give it a shot when I can!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants