Using LGSVL with Autoware : cant't build Docker containers leveraging NVIDIA GPUs

[Eolion]

Hello, I want to install LGSVL with autoware on my computer under ubuntu. But the command :
docker run --gpus all nvidia/cuda:10.0-base nvidia-smi doesn't work.

My terminal displays :
docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].
ERRO[0026] error waiting for container: context canceled

I try many things even reinstalling ubuntu. But nothing change i always have same error and I can't continue install Autoware.
You can see the result when I use nvdia-smi :
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.132 Driver Version: 390.132 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 770 Off | 00000000:01:00.0 N/A | N/A |
| 34% 34C P8 N/A / N/A | 341MiB / 1994MiB | N/A Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 Not Supported |
+-----------------------------------------------------------------------------+

I specify that I am obliged to use the nvidia-390 drivers because the others make display at lgsvl a white screen on which I can do nothing.

Does anyone have the solution to my problem?

[deepakramani]

First, did you install docker properly? Are you able to run other docker commands well? Is the docker version > 19.03?
Then, did you install nvidia docker, properly? -- https://github.com/NVIDIA/nvidia-docker#ubuntu-160418042004-debian-jessiestretchbuster

If you have, then can you try this? -- NVIDIA/nvidia-docker#1034 (comment)

[Eolion]

Yes I follow all steps : "Installation methods" to install docker.
Command "docker run hello-wold" works.

I try docker run --gpus all nvidia/cuda:9.0-base nvidia-smi and terminal don't displays error just the result of nvidia-smi
but after that when I try "Autoware/docker/generic/build.sh kinetic" my terminal displays this :

Use kinetic
./build.sh: 7: ./build.sh: nvidia-docker: not found

[deepakramani]

Okay docker works.

Leave Nvidia-440 for now. Have the driver version you wish installed.

These problems are nvidia-docker related. You need to have them properly installed before you go into autoware.

So,
Did you install nvidia-docker(https://github.com/NVIDIA/nvidia-docker#ubuntu-160418042004-debian-jessiestretchbuster) properly? Have you by chance installed nvidia-docker2? Which Ubuntu version are you using? In particular did you execute these commands?

sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit
sudo systemctl restart docker

After this, can you execute - docker run --gpus all nvidia/cuda:9.0-base nvidia-smi
Is the output of that command, exactly like when you execute nvidia-smi?

After that, make changes to run.sh as mentioned here --https://www.lgsvlsimulator.com/docs/autoware-instructions/#install-autoware and see if you can run autoware.

If you run into problems then also, try running the commands mentioned here -- NVIDIA/nvidia-docker#1034 (comment)

Then start over with autoware. Despite all these you get an error, you may need to try this -- https://github.com/NVIDIA/nvidia-container-runtime.

[Eolion]

Thank you ! To fix my problem I just install nvidia-docker2

Now I can execute "sudo build.sh kinetic" but I have another error :
sometimes red line appears with this message :

debconf: delaying package configuration, since apt-utils is not installed

W: GPG error: http://packages.ros.org/ros/ubuntu xenial InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY F42ED6FBAB17C654
W: The repository 'http://packages.ros.org/ros/ubuntu xenial InRelease' is not signed.

And at the end :

E: There were unauthenticated packages and -y was used without --allow-unauthenticated
The command '/bin/sh -c apt-get update && apt-get install -y ros-kinetic-desktop-full ros-kinetic-nmea-msgs ros-kinetic-nmea-navsat-driver ros-kinetic-sound-play ros-kinetic-jsk-visualization ros-kinetic-grid-map ros-kinetic-gps-common' returned a non-zero code: 100

[deepakramani]

The error you're getting is a common GPG error. You can overcome them by following "set up keys" given in the corresponding ros version page.

[Eolion]

Although I did add the key with the command:
curl -sSL 'http://keyserver.ubuntu.com/pks/lookup?op=get&search=0xC1CF6E31E6BADE8868B172B4F42ED6FBAB17C654' | sudo apt-key add -

I try : sudo apt-key adv --keyserver 'hkp://keyserver.ubuntu.com:80' --recv-key C1CF6E31E6BADE8868B172B4F42ED6FBAB17C654
as shown on this link : https://discourse.ros.org/t/new-gpg-keys-deployed-for-packages-ros-org/9454
too

I have exactly same error

[deepakramani]

After adding the key did you do the 'final step' mentioned here -- https://answers.ros.org/question/325039/apt-update-fails-cannot-install-pkgs-key-not-working/ ?

[Eolion]

I just added --allow-unauthenticated in dockerfile.kinetic after that no problem with key and install.
Thank you for your time and your answers. I close this issue