Subject: vkCreateInstance error (-9) when Installing NVIDIA ACE Animation Graph Microservice Docker Container
Hello NVIDIA Community,
I’m facing an issue while trying to install the NVIDIA ACE Animation Graph Microservice Docker container. I’m following the official NVIDIA documentation:
Animation Pipeline Local Docker Containers
When I execute the following command:
docker run -it --rm --gpus all --network=host --name anim-graph-ms -v /default-avatar-scene_v1.0.0:/home/ace/asset nvcr.io/eevaigoeixww/animation/ia-animation-graph-microservice:1.0.1
I receive the following output:
+ ldconfig -p
+ grep libGLX_nvidia.so.0
libGLX_nvidia.so.0 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libGLX_nvidia.so.0
+ [[ -v NOTFOUND ]]
+ export VK_ICD_FILENAMES=/tmp/nvidia_icd.json
+ VK_ICD_FILENAMES=/tmp/nvidia_icd.json
+ export LD_LIBRARY_PATH=:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/opt/nvidia/omniverse/kit-sdk-launcher/plugins/carb_gfx
+ LD_LIBRARY_PATH=:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/opt/nvidia/omniverse/kit-sdk-launcher/plugins/carb_gfx
+ /opt/nvidia/omniverse/vkapiversion/bin/vkapiversion /tmp/nvidia_icd.json
Writing disposable ICD file (/tmp/tmp_icd_HF6iIQ.json)...
vkCreateInstance error (-9)
Despite ensuring that my NVIDIA drivers and NVIDIA Container Toolkit are correctly configured, the issue persists. I’ve attempted this installation on:
- WSL Ubuntu 22.04
- WSL Ubuntu 24.04
- Dedicated Server with Ubuntu
In all cases, I encounter the same vkCreateInstance error (-9).
Here is the output of nvidia-smi on my system:
root@DECODED:~# nvidia-smi
Wed Oct 2 19:57:17 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.183.01 Driver Version: 561.09 CUDA Version: 12.6 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GeForce RTX 4070 On | 00000000:2B:00.0 On | N/A |
| 58% 36C P0 35W / 200W | 3695MiB / 12282MiB | 1% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
System Details:
- GPU: NVIDIA GeForce RTX 4070
- Driver Version: 561.09
- CUDA Version: 12.6
- Operating System: Ubuntu 22.04 and 24.04 (both on WSL and dedicated server)
- Docker Version: Latest
- NVIDIA Container Toolkit: Installed and configured
What I’ve Tried So Far:
- Verified that the NVIDIA drivers are up to date.
- Ensured the NVIDIA Container Toolkit is properly installed and configured.
- Used the
--gpus allflag with Docker to grant GPU access to the container. - Tested on both WSL and a dedicated Ubuntu server to rule out environment-specific issues.
Issue Description:
The error seems to occur during the Vulkan instance creation inside the container:
vkCreateInstance error (-9)
This suggests there might be an issue with Vulkan setup or GPU access within the container.
Request for Assistance:
Has anyone experienced a similar issue or can provide guidance on how to resolve this error? I’m wondering if there’s a compatibility issue with the RTX 4070 or a misconfiguration that I’m overlooking.
Any insights or suggestions would be greatly appreciated!
Thank you in advance for your help!
Best regards,
DECODED