Nvidia-installation problem on GCP

Sometimes when you reboot a VM with Nvidia drivers installed, it shows : E: Package 'nvidia-cuda-toolkit' has no installation candidate

Now you can do two things : 

But, there is an easier option if you are using Deep Learning VM (which I always recommend) : 

This will first remove all nvidia-drivers, then install all the necessary dependencies.

Incase the VM is not a deep learning VM, you can download the below file and run the install-driver.sh the zip file, here


#!/bin/bash -eu


# Copyright 2020 Google Inc.


# Licensed under the Apache License, Version 2.0 (the "License");

# you may not use this file except in compliance with the License.

# You may obtain a copy of the License at


#     http://www.apache.org/licenses/LICENSE-2.0


# Unless required by applicable law or agreed to in writing, software

# distributed under the License is distributed on an "AS IS" BASIS,


# See the License for the specific language governing permissions and

# limitations under the License.

# Purpose: This script installs NVIDIA Drivers for GPU


# Refer the following links for NVIDIA driver installation.

# https://developer.nvidia.com/cuda-toolkit-archive

# https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/"

# https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html

export ENV_FILE="/etc/profile.d/env.sh"

# shellcheck source=/etc/profile.d/env.sh disable=SC1091

source "${ENV_FILE}" || exit 1

set -x

function get_metadata_value() {

  curl --retry 5 \

    -s \

    -f \

    -H "Metadata-Flavor: Google" \



function get_attribute_value() {

  get_metadata_value "instance/attributes/$1"


function install_linux_headers() {

  # Install linux headers. Note that the kernel version might be changed after

  # installing gvnic version. For example: 4.19.0-8-cloud-amd64 ->

  # 4.19.0-9-cloud-amd64. So we install the kernel headers for each driver

  # installation.

  echo "install linux headers: linux-headers-$(uname -r)"

  sudo apt-get -o DPkg::Lock::Timeout=120 install -y linux-headers-"$(uname -r)"


# Try to download driver via Web if GCS failed (Example: VPC-SC/GCS failure)

function download_driver_via_http() {

  local driver_url_path=$1

  local downloaded_file=$2

  echo "Could not use Google Cloud Storage APIs to download driver. Attempting to download them directly from Nvidia."

  echo "Downloading driver from URL: ${driver_url_path}"

  wget -nv "${driver_url_path}" -O "${downloaded_file}" || {

    echo 'Download driver via Web failed!' &&

    rm -f "${downloaded_file}" &&

    echo "${downloaded_file} deleted"



function wait_apt_locks_released() {

  # Wait for apt lock to be released

  # Source: https://askubuntu.com/a/373478

  echo "wait apt locks released"

  while sudo fuser /var/{lib/{dpkg,apt/lists},cache/apt/archives}/lock >/dev/null 2>&1 ||

     sudo fuser /var/lib/dpkg/lock-frontend >/dev/null 2>&1 ; do

     sleep 1



function install_nvidia_linux_drivers() {


  local driver_installer_file_name="driver_installer.run"

  local nvidia_driver_file_name="NVIDIA-Linux-x86_64-${DRIVER_VERSION}.run"

  if [[ -z "${DRIVER_GCS_PATH}" ]]; then



  local driver_gcs_file_path=${DRIVER_GCS_PATH}/${nvidia_driver_file_name}

  echo "Downloading driver from GCS location and install: ${driver_gcs_file_path}"

  set +e

  gsutil -q cp "${driver_gcs_file_path}" "${driver_installer_file_name}"

  set -e

  # open kernel modules argument, blank if installing proprietary kernel modules

  # "-m=kernel-open" if installing open kernel modules. Provided to the Runfile.

  local open_kernel_module_arg="-m=kernel-open"

  IFS=. read -r major minor patch <<< "${DRIVER_VERSION}"

  # full machine type string, example: projects/1098636466100/machineTypes/a2-highgpu-1g

  # -r for readonly as we will not be modifying this var, and this satisfies

  # https://github.com/koalaman/shellcheck/wiki/SC2155

  local -r machine_type_full=$(get_metadata_value instance/machine-type)

  # get last element of string split by '/' to get machine type

  # https://stackoverflow.com/questions/3162385/how-to-split-a-string-in-shell-and-get-the-last-field

  local machine_type=${machine_type_full##*/}

  # Install proprietary kernel modules if the instance is on an N-series machine

  # or if the driver version is on a major release older than R525.

  # Nvidia R515 is the first major release that supports open kernel modules,

  # and R525 is the first major release that will be compatible with an upcoming

  # required kernel module needed for Kyubi.

  if [[ "${major}" -lt 525 ]] || [[ $machine_type =~ ^n ]]; then



  # Download driver via http if GCS failed.

  if [[ ! -f "${driver_installer_file_name}" ]]; then


    download_driver_via_http "${driver_url_path}" "${driver_installer_file_name}"


  if [[ ! -f "${driver_installer_file_name}" ]]; then

    echo "Failed to find drivers!"

    exit 1


  chmod +x ${driver_installer_file_name}

  sudo ./${driver_installer_file_name} --dkms -a -s --no-drm --install-libglvnd "${open_kernel_module_arg}"

  rm -rf ${driver_installer_file_name}


main() {



  # shellcheck source=/opt/deeplearning/driver-version.sh disable=SC1091

  source "${DL_PATH}/driver-version.sh"


  # Custom GCS driver location via instance metadata.

  DRIVER_GCS_PATH=$(get_attribute_value nvidia-driver-gcs-path)


  exit 0