You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Problem: I run nvidia-ctk generate and put nvidia.yml to the custom directory being as a non-privileged user.
I add env XDG_CONFIG_HOME to my home dir and create nvidia-container-runtime there, where I re-define config.toml with custom spec-dirs values.
And if I run nvidia-ctk config - I see that the configuration at least can be read fine from my custom config file.
So I put nvidia.yml to /app/home/podman/cdi.
And nvidia-ctk cdi list show nothing.
But it shows devices if I run with flag --spec-dir, so it defenitely should work.
I expect nvidia-ctk cdi list to show devices without additional flag as soon as in the config I redefine spec-dirs.
What else am I missing here?
The text was updated successfully, but these errors were encountered:
Most nvidia-ctk commands do not currently load settings from the config.toml file. The one exception is nvidia-ctk config which processes the same config file that the nvidia-container-runtime would.
We can look into whether it makes sense to change this behaviour.
$ nvidia-ctk --version
NVIDIA Container Toolkit CLI version 1.17.4
commit: 9b69590
$ nvidia-ctk config
disable-require = false
supported-driver-capabilities = "compat32,compute,display,graphics,ngx,utility,video"
[nvidia-container-cli]
environment = []
ldconfig = "@/sbin/ldconfig"
load-kmods = true
no-cgroups = true
[nvidia-container-runtime]
debug = "/app/home/podman/.local/nvidia-container-runtime.log"
log-level = "info"
mode = "auto"
runtimes = ["docker-runc", "runc", "crun"]
[nvidia-container-runtime.modes]
[nvidia-container-runtime.modes.cdi]
annotation-prefixes = ["cdi.k8s.io/"]
default-kind = "nvidia.com/gpu"
spec-dirs = ["/etc/cdi", "/var/run/cdi", "/app/home/podman/cdi"]
[nvidia-container-runtime.modes.csv]
mount-spec-path = "/etc/nvidia-container-runtime/host-files-for-container.d"
[nvidia-container-runtime-hook]
path = "nvidia-container-runtime-hook"
skip-mode-detection = false
[nvidia-ctk]
path = "nvidia-ctk"
$ nvidia-ctk cdi list --spec-dir "/app/home/podman/cdi"
INFO[0000] Found 5 CDI devices
nvidia.com/gpu=0
nvidia.com/gpu=1
nvidia.com/gpu=GPU-a69be7f2-776f-034f-d202-ad700be58eac
nvidia.com/gpu=GPU-b6705994-caa0-dad7-095d-554686d20f12
nvidia.com/gpu=all
$ nvidia-ctk cdi list
INFO[0000] Found 0 CDI devices
Problem: I run nvidia-ctk generate and put nvidia.yml to the custom directory being as a non-privileged user.
I add env XDG_CONFIG_HOME to my home dir and create nvidia-container-runtime there, where I re-define config.toml with custom spec-dirs values.
And if I run nvidia-ctk config - I see that the configuration at least can be read fine from my custom config file.
So I put nvidia.yml to /app/home/podman/cdi.
And nvidia-ctk cdi list show nothing.
But it shows devices if I run with flag --spec-dir, so it defenitely should work.
I expect nvidia-ctk cdi list to show devices without additional flag as soon as in the config I redefine spec-dirs.
What else am I missing here?
The text was updated successfully, but these errors were encountered: