nvidia-container-cli: mount error

[wangzhongliang]

after I run the commnad docker-compose up -d
the following error came out
[error ](nvidia-container-cli: mount error: file creation failed: /var/lib/docker/overlay2/72dc9ee5772968b2295c4827d10cc1ad8f74b2f241cae14396bf28f457537848/merged/usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1: file exists: unknown)

[successren]

Are you using Windows Docker? Our image is packaged on Linux, and when loaded on Windows, it encounters Nvidia-related errors for which there is currently no solution. We plan to package a Windows version of the Docker image.

[successren]

Are you using Windows Docker? Our image is packaged on Linux, and when loaded on Windows, it encounters Nvidia-related errors for which there is currently no solution. We plan to package a Windows version of the Docker image.

Or you can refer to NVIDIA/nvidia-container-toolkit#289. You need to delete some Nvidia-related .so files in the container and then repackage the image.

[wangzhongliang]

Are you using Windows Docker? Our image is packaged on Linux, and when loaded on Windows, it encounters Nvidia-related errors for which there is currently no solution. We plan to package a Windows version of the Docker image.

I use WSL2 on Windows, any solution for WSL2?

[wangzhongliang]

Are you using Windows Docker? Our image is packaged on Linux, and when loaded on Windows, it encounters Nvidia-related errors for which there is currently no solution. We plan to package a Windows version of the Docker image.

Or you can refer to NVIDIA/nvidia-container-toolkit#289 (comment). You need to delete some Nvidia-related .so files in the container and then repackage the image.

thx, I will try

[wangzhongliang]

Are you using Windows Docker? Our image is packaged on Linux, and when loaded on Windows, it encounters Nvidia-related errors for which there is currently no solution. We plan to package a Windows version of the Docker image.

Or you can refer to NVIDIA/nvidia-container-toolkit#289 (comment). You need to delete some Nvidia-related .so files in the container and then repackage the image.

I try the solution, and successfully enter the front end page
here is my modification:

1. rebuild a new image based on freeren/qanything:v1.0.7, rename it as freeren/qanything:v1.0.7-fix
2. modify the docker-compose.yaml as below (change the image name in qanything_local, and remove the runtime: nvidia)
   

But new problem came out, all the apis (eg. /api/local_doc_qa/list_knowledge_base) are not working and return 502 error

[xixihahaliu]

This situation is most likely due to the qanything service not starting correctly. You need to check the system logs to see the specific reason. You can paste the key parts from the QAnything/api.log file.

[wangzhongliang]

This situation is most likely due to the qanything service not starting correctly. You need to check the system logs to see the specific reason. You can paste the key parts from the QAnything/api.log file.

i use docker-compose logs -f qanything_local to get the logs, but there isn't any log
But I can see logs of other services, such as minio, as below

maybe the modification I made can't work?

[dj456119]

You can build a new Dockerfile and remove that files:

# 基于现有镜像
FROM freeren/qanything:v1.0.7

# 执行删除操作
RUN rm -rf /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 /usr/lib/x86_64-linux-gnu/libcuda.so.1 /usr/lib/x86_64-linux-gnu/libcudadebugger.so.1 /usr/lib/x86_64-linux-gnu/libnvidia-encode.so.1 /usr/lib/x86_64-linux-gnu/libnvidia-opticalflow.so.1 /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1

[xixihahaliu]

这种情况很可能是由于 qanything 服务未正确启动造成的。具体原因需要查看系统日志。您可以粘贴 QAnything/api.log 文件中的关键部分。

我用来docker-compose logs -f qanything_local获取日志，但没有任何日志 但是我可以看到其他服务的日志，例如minio，如下

也许我所做的修改不起作用？

The installation method for Windows has been updated. You can try again. For details, please refer to the readme file.