Docker: Error response from daemon: failed to create shim: OCI runtime create failed: container_linux.go:380: starting container process caused: process_linux.go:545: container init caused: Running hook #0:: error running hook: signal: segmentation fault

General
1

Hey everyone, I am getting the above error when running the docker run --rm --gpus all command with at least 1 argument. I am not familiar enough with docker to know what other information to provide. I have heard answers referencing the NVIDIA driver or ubuntu/linux but couldn’t find anything specific enough.

Example to reproduce

docker run --rm --gpus all nvcr.io/nvidia/k8s/cuda-sample:vectoradd-cuda10.2

Anyone have an idea? I can provide further information as well.

Windows 10
Cloud integration: v1.0.24
Version: 20.10.14
API version: 1.41
Go version: go1.16.15

OS/Arch: windows/amd64
Context: default
Experimental: true

Server: Docker Desktop 4.8.0
Engine:

Version: 20.10.14
API version: 1.41 (minimum version 1.12)
Go version: go1.16.15
Built: Thu Mar 24 01:46:14 2022
OS/Arch: linux/amd64
Experimental: false
containerd:
Version: 1.5.11
runc:
Version: 1.0.3

docker-init:
Version: 0.19.0

2

Also asked on python - process_linux.go:545: container init caused: Running hook #0:: error running hook: signal: segmentation fault, stdout: , stderr:: unknown - Stack Overflow and then python - process_linux.go:545: container init caused: Running hook #0:: error running hook: signal: segmentation fault, stdout: , stderr:: unknown - Stack Overflow

3

I got the same error in my environment WSL2 on Windows 10, version 21H1.
I updated the version of Windows 21H2 and worked well.
I’m using the WSL2 backend for Docker Desktop.

FYI, I found the following in What’s new in Windows 10, version 21H2:

GPU compute support for the Windows Subsystem for Linux

Starting with Windows 10 version 21H2, the Windows Subsystem for Linux has full graphics processing unit (GPU) compute support. It was available to Windows Insiders, and is now available to everyone. The Linux binaries can use your Windows GPU, and run different workloads, including artificial intelligence (AI) and machine learning (ML) development workflows.

For more information, and what GPU compute support means for you, see the GPU accelerated ML training inside the Windows Subsystem for Linux blog post.

1 Like