It look a little bit to run this down, but after an upgrade four days ago my containers using the nivida runtime get:
docker: Error response from daemon: failed to create shim task: OCI runtime create failed: nvidia-container-runtime did not terminate successfully: exit status 1: unknown.
And none of them would run (3 out of my 30 containers are using the nvidia runtime). I'm using the centos8 libnvidia-container repo on Fedora 35.
I'm not sure if this is the right place to report this. I also reported this over at https://github.com/NVIDIA/nvidia-container-toolkit issue # 34
It look a little bit to run this down, but after an upgrade four days ago my containers using the nivida runtime get:
docker: Error response from daemon: failed to create shim task: OCI runtime create failed: nvidia-container-runtime did not terminate successfully: exit status 1: unknown.
And none of them would run (3 out of my 30 containers are using the nvidia runtime). I'm using the centos8 libnvidia-container repo on Fedora 35.
I found these in my dnf history:
Rolling back from 1.11.0 to 1.10.0 and reinstalling nvidia-docker2 fixed this issue.
Some other info that may be helpful:
And...