Open szhang99-bu opened 7 months ago
I have seen on there thread that this issue can be solved by editing /etc/nvidia-container-runtime/config.toml
and changing:
[nvidia-container-cli]
no-cgroups = true
[nvidia-container-runtime]
debug = "/tmp/nvidia-container-runtime.log"
Is this correct today? Because the thread is nearly 4 years ago. And I do not have a file located in "/tmp/nvidia-container-runtime.log"
And this is the current setting for /etc/nvidia-container-runtime/config.toml
#accept-nvidia-visible-devices-as-volume-mounts = false
#accept-nvidia-visible-devices-envvar-when-unprivileged = true
disable-require = false
supported-driver-capabilities = "compat32,compute,display,graphics,ngx,utility,video"
#swarm-resource = "DOCKER_RESOURCE_GPU"
[nvidia-container-cli]
#debug = "/var/log/nvidia-container-toolkit.log"
environment = []
#ldcache = "/etc/ld.so.cache"
ldconfig = "@/sbin/ldconfig.real"
load-kmods = true
#no-cgroups = false
#path = "/usr/bin/nvidia-container-cli"
#root = "/run/nvidia/driver"
#user = "root:video"
[nvidia-container-runtime]
#debug = "/var/log/nvidia-container-runtime.log"
log-level = "info"
mode = "auto"
runtimes = ["docker-runc", "runc"]
[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"]
[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"
@szhang99-bu for completeness, how is Docker installed? Is this Docker Desktop?
@elezar Docker is installed following Docker Desktop installation guide for Ubuntu from docker website using DEB package. And Daemon has been configured to NVIDIA in the setting.
{
"runtimes": {
"nvidia": {
"path": "nvidia-container-runtime",
"runtimeArgs": []
}
},
"default-runtime": "nvidia"
}
@szhang99-bu the toolkit currently only supports docker-ce
and not Docker Desktop on Linux.
Is this support coming?
1. Issue or feature description
I am currently trying to install a version of Alphafold 2 on a desktop with a 3090. By following the installation instruction, i run into the issue of unable to run docker with NVIDIA container toolkit without
sudo
I have correctly installed docker desktop & NVIDIA container toolkit, and followed the steps to add docker to userAnd currently, docker can run the verfication hello world step with no issue.
However, when running docker with --gpu flag
sudo
is requiredIt seems other people have run into the same issue: https://github.com/google-deepmind/alphafold/issues/865#issue-2007089233
2. Steps to reproduce the issue
docker run --rm --gpus all nvidia/cuda:11.0.3-base-ubuntu20.04 nvidia-smi
3. Information to attach (optional if deemed irrelevant)
[ ] Some nvidia-container information:
nvidia-container-cli -k -d /dev/tty info
[ ] Kernel version from
uname -a
Linux shiyuzhang-X570-AORUS-ULTRA 6.2.0-37-generic NVIDIA/nvidia-docker#38~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Nov 2 18:01:13 UTC 2 x86_64 x86_64 x86_64 GNU/Linux
[ ] Any relevant kernel output lines from
dmesg
[ ] Driver information from
nvidia-smi -a
docker version
dpkg -l '*nvidia*'
orrpm -qa '*nvidia*'
nvidia-container-cli -V