| Deploying charm cs:~containers/kubernetes-worker #90 on a machine with GPU and configured with cuda-version=9.0.176-1 (required to run TensorFlow version 1.5), the command: /var/lib/juju/agents/unit-kubernetes-worker-gpu-4/charm/reactive/cuda.sh --invoke install_cuda fails, apparently since the GPG key is missing, as it shows issuing the failing step by hand: sudo dpkg -i /tmp/cuda-repo-ubuntu1604_9.0.176-1_amd64.deb (Reading database ... 72620 files and directories currently installed.) Preparing to unpack .../cuda-repo-ubuntu1604_9.0.176-1_amd64.deb ... Unpacking cuda-repo-ubuntu1604 (9.0.176-1) over (9.0.176-1) ... Setting up cuda-repo-ubuntu1604 (9.0.176-1) ... The public CUDA GPG key does not appear to be installed. To install the key, run this command: sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub Indeed the installation instructions at https://developer.nvidia.com/cuda-90-download-archive?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1604&target_type=debnetwork say: Installation Instructions: • `sudo dpkg -i cuda-repo-ubuntu1604_9.0.176-1_amd64.deb` • `sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub` • `sudo apt-get update` • `sudo apt-get install cuda` The fix to the script should be to add: apt-key adv --fetch-keys $ROOT_URL/$UBUNTU_VERSION/x86_64/7fa2af80.pub One also needs to add --allow-unauthenticated to apt-get install. See the enclosed patch. — Beppe |
cuda.sh.patch
Description: Binary data
-- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju
