Merge branch 'nvidia_ppc64'

steps/data/setup/puppet/modules/env/files/big/nvidia/cuda-9.0.conf

-/usr/local/cuda/lib64

steps/data/setup/puppet/modules/env/files/big/nvidia/cuda.conf

-/usr/local/cuda-7.0/lib
-/usr/local/cuda-7.0/lib64
+/usr/local/cuda/lib64

steps/data/setup/puppet/modules/env/files/big/nvidia/dcgm-exporter.service

 [Unit]
 Description=NVIDIA DCGM prometheus exporter service
 After=network.target
+# Ensure that /dev/nvidia0 is created by first calling nvidia-smi.
+# If no GPU is found, nvidia-smi will not create /dev/nvidia0 and we will not run.
+Wants=nvidia-smi.service
+After=nvidia-smi.service
 ConditionPathExists=/dev/nvidia0
 
 [Service]

steps/data/setup/puppet/modules/env/files/big/nvidia/nvidia-persistenced-9.0.service → steps/data/setup/puppet/modules/env/files/big/nvidia/nvidia-persistenced.service

 [Unit]
 Description=NVIDIA Persistence Daemon
 Wants=syslog.target
+# Ensure that /dev/nvidia0 is created by first calling nvidia-smi.
+# If no GPU is found, nvidia-smi will not create /dev/nvidia0 and we will not run.
+Wants=nvidia-smi.service
+After=nvidia-smi.service
+ConditionPathExists=/dev/nvidia0
 
 [Service]
 Type=forking

steps/data/setup/puppet/modules/env/files/big/nvidia/nvidia-smi.service

+[Unit]
+Description=Call nvidia-smi once to create /dev/nvidiaX
+
+[Service]
+Type=oneshot
+# Ignore the exit code: the command fails when no GPU is found
+ExecStart=-/usr/bin/nvidia-smi
+
+[Install]
+WantedBy=multi-user.target

steps/data/setup/puppet/modules/env/files/big/nvidia/profile

-# /etc/profile: system-wide .profile file for the Bourne shell (sh(1))
-# and Bourne compatible shells (bash(1), ksh(1), ash(1), ...).
-
-if [ "`id -u`" -eq 0 ]; then
-  PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/cuda-7.0/bin"
-else
-  PATH="/usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games:/usr/local/cuda-7.0/bin"
-fi
-
-export PATH
-
-if [ "$PS1" ]; then
-  if [ "$BASH" ] && [ "$BASH" != "/bin/sh" ]; then
-    # The file bash.bashrc already sets the default PS1.
-    # PS1='\h:\w\$ '
-    if [ -f /etc/bash.bashrc ]; then
-      . /etc/bash.bashrc
-    fi
-  else
-    if [ "`id -u`" -eq 0 ]; then
-      PS1='# '
-    else
-      PS1='$ '
-    fi
-  fi
-fi
-
-if [ -d /etc/profile.d ]; then
-  for i in /etc/profile.d/*.sh; do
-    if [ -r $i ]; then
-      . $i
-    fi
-  done
-  unset i
-fi
-

steps/data/setup/puppet/modules/env/manifests/big.pp

@@ -17,10 +17,12 @@ class env::big ( $variant = "big", $parent_parameters = {} ){
   class { 'env::big::configure_postfix': }
   # kvm
   class { 'env::big::configure_kvm': }
-  if $env::deb_arch == 'amd64' {
-    # nvidia
+  # nvidia
+  if $env::deb_arch == 'amd64' or $env::deb_arch == 'ppc64el' {
     class { 'env::big::configure_nvidia_gpu': }
-    # beegfs install
+  }
+  # beegfs install
+  if $env::deb_arch == 'amd64' {
     class { 'env::big::install_beegfs': }
   }
   #Allow sshfs

steps/data/setup/puppet/modules/env/manifests/big/configure_nvidia_gpu.pp

@@ -6,6 +6,8 @@ class env::big::configure_nvidia_gpu () {
   include 'env::big::configure_nvidia_gpu::modules'
   # Install nvidia drivers
   include 'env::big::configure_nvidia_gpu::drivers'
+  # Install additional services (currently nvidia-smi, needed by cuda and prometheus)
+  include 'env::big::configure_nvidia_gpu::services'
   # Install cuda
   include 'env::big::configure_nvidia_gpu::cuda'
   # Install nvidia ganglia plugins

steps/data/setup/puppet/modules/env/manifests/big/configure_nvidia_gpu/cuda.pp

@@ -2,7 +2,20 @@ class env::big::configure_nvidia_gpu::cuda () {
 
   case "${::lsbdistcodename}" {
     "buster" : {
-      $driver_source = 'http://packages.grid5000.fr/other/cuda/cuda_10.1.168_418.67_linux.run'
+      case "$env::deb_arch" {
+        "amd64": {
+          $driver_source = 'http://packages.grid5000.fr/other/cuda/cuda_10.1.243_418.87.00_linux.run'
+          $libcuda = '/usr/lib/x86_64-linux-gnu/libcuda.so'
+        }
+        "ppc64el": {
+          $driver_source = 'http://packages.grid5000.fr/other/cuda/cuda_10.1.243_418.87.00_linux_ppc64le.run'
+          $libcuda = '/usr/lib/powerpc64le-linux-gnu/libcuda.so'
+        }
+        default: {
+          err "${env::deb_arch} not supported"
+        }
+      }
+
       $opengl_packages = ['ocl-icd-libopencl1', 'opencl-headers']
 
       exec{
@@ -26,27 +39,7 @@ class env::big::configure_nvidia_gpu::cuda () {
     "stretch" : {
       $driver_source = 'http://packages.grid5000.fr/other/cuda/cuda_9.0.176_384.81_linux-run'
       $opengl_packages = ['ocl-icd-libopencl1', 'opencl-headers']
-
-      exec{
-        'retrieve_nvidia_cuda':
-          command   => "/usr/bin/wget -q $driver_source -O /tmp/NVIDIA-Linux_cuda.run && chmod u+x /tmp/NVIDIA-Linux_cuda.run",
-          timeout   => 1200, # 20 min
-          creates   => "/tmp/NVIDIA-Linux_cuda.run";
-        'install_nvidia_cuda':
-          command     => "/tmp/NVIDIA-Linux_cuda.run --silent --toolkit --samples && /bin/rm /tmp/NVIDIA-Linux_cuda.run",
-          timeout     => 2400, # 20 min
-          user        => root,
-          require     =>  File['/tmp/NVIDIA-Linux_cuda.run'];
-        'update_ld_conf':
-          command   => "/sbin/ldconfig",
-          user      => root,
-          refreshonly => true;
-      }
-    }
-
-    "jessie" : {
-      $driver_source = 'http://packages.grid5000.fr/other/cuda/cuda_9.0.176_384.81_linux-run'
-      $opengl_packages = ['ocl-icd-libopencl1', 'opencl-headers', 'amd-opencl-icd']
+      $libcuda = '/usr/lib/x86_64-linux-gnu/libcuda.so'
 
       exec{
         'retrieve_nvidia_cuda':
@@ -74,7 +67,7 @@ class env::big::configure_nvidia_gpu::cuda () {
           require   => Exec['retrieve_nvidia_cuda'];
         '/usr/local/cuda/lib64/libcuda.so':
           ensure    => 'link',
-          target    => '/usr/lib/x86_64-linux-gnu/libcuda.so',
+          target    => $libcuda,
           require   => Exec['install_nvidia_cuda'],
           notify    => Exec['update_ld_conf'];
         '/etc/ld.so.conf.d/cuda.conf':
@@ -82,31 +75,17 @@ class env::big::configure_nvidia_gpu::cuda () {
           owner     => root,
           group     => root,
           mode      => '0644',
-          source    => 'puppet:///modules/env/big/nvidia/cuda-9.0.conf',
+          source    => 'puppet:///modules/env/big/nvidia/cuda.conf',
           notify    => Exec['update_ld_conf'];
         '/etc/systemd/system/nvidia-persistenced.service':
           ensure    => file,
           owner     => root,
           group     => root,
           mode      => '0644',
-          source    => 'puppet:///modules/env/big/nvidia/nvidia-persistenced-9.0.service';
-      }
-    }
-    "jessie" : {
-      file{
-        '/tmp/NVIDIA-Linux_cuda.run':
-          ensure    => file,
-          require   => Exec['retrieve_nvidia_cuda'];
-        '/etc/ld.so.conf.d/cuda.conf':
-          ensure    => file,
-          owner     => root,
-          group     => root,
-          mode      => '0644',
-          source    => 'puppet:///modules/env/big/nvidia/cuda.conf',
-          notify    => Exec['update_ld_conf'];
-        '/usr/local/cuda/lib64/libcuda.so':
-          ensure    => 'link',
-          target    => '/usr/lib/libcuda.so';
+          source    => 'puppet:///modules/env/big/nvidia/nvidia-persistenced.service';
+        '/etc/systemd/system/multi-user.target.wants/nvidia-persistenced.service':
+          ensure => link,
+          target => '/etc/systemd/system/nvidia-persistenced.service';
       }
     }
   }

steps/data/setup/puppet/modules/env/manifests/big/configure_nvidia_gpu/drivers.pp

@@ -2,8 +2,17 @@ class env::big::configure_nvidia_gpu::drivers () {
 
   ### This class exists for gpuclus cluster, that require a recent version of nvidia driver
 
-  # May be changed to a link inside g5k if required
-  $driver_source = 'http://packages.grid5000.fr/other/nvidia//NVIDIA-Linux-x86_64-450.51.05.run'
+  case "$env::deb_arch" {
+    "amd64": {
+      $driver_source = 'http://packages.grid5000.fr/other/nvidia/NVIDIA-Linux-x86_64-450.80.02.run'
+    }
+    "ppc64el": {
+      $driver_source = 'http://packages.grid5000.fr/other/nvidia/NVIDIA-Linux-ppc64le-450.80.02.run'
+    }
+    default: {
+      err "${env::deb_arch} not supported"
+    }
+  }
 
   package {
     ['module-assistant', 'dkms']:

steps/data/setup/puppet/modules/env/manifests/big/configure_nvidia_gpu/services.pp

+class env::big::configure_nvidia_gpu::services () {
+
+  # We only install the service but do not enable it.
+  # Services that depend on it can add "Wants=nvidia-smi.service"
+  # and "After=nvidia-smi.service", and this will automatically start
+  # this service.
+  file{
+    '/etc/systemd/system/nvidia-smi.service':
+      ensure    => file,
+      owner     => root,
+      group     => root,
+      mode      => '0644',
+      source    => 'puppet:///modules/env/big/nvidia/nvidia-smi.service';
+  }
+}