Fork of the NVIDIA device plugin for Kubernetes with support for shared GPUs by declaring GPUs multiple times

This project is deprecated: the upstream NVIDIA device plugin now supports "time-slicing"

Since v0.12.0, upstream NVIDIA device plugin for Kubernetes supports sharing access to one GPU from multiple Pods, the feature is called "time-slicing".

This project will thus not be maintained anymore.

Table of Contents

• About this fork
• About
• Prerequisites
• Quick Start
  • Preparing your GPU Nodes
  • Enabling GPU Support in Kubernetes
  • Running GPU Jobs
• Docs
• Changelog
• Issues and Contributing

About this fork

This fork advertises multiple fake GPU for each real GPU, allowing to share a GPU between multiple pods using the Kubernetes device plugin api.

The goal is to schedule pods on GPUs until the GPU memory is full (GPU memory bin-packing).

Limits

This is a big workaround given the current situation. It has many drawbacks: The kubernetes scheduler doesn't know how the underlying real GPUs are shared between the deepomatic.com/shared-gpu resources it allocates among Pods.

• there is no way to control/guarantee spreading the pods among real GPUs: the current workaround is to limit to one real GPU per node and to indirectly schedule via other resources such as memory (assuming there is a correlation between memory and GPU (memory) usage.
• in the case of multiple real GPUs per node, asking for multiple shared GPUs for one Pod doesn't make sense as there is no guarantee the pod will be allocated shared GPUs from different real GPUs

Roadmap

For proper scheduling, this device plugin will advertise SharedGPUMemory as Kubernetes Extended Resources. Since the SharedGPUMemory resource is at the Node level (instead of at the Device level), we effectively support only one GPU per node.

Configuration

You can control the number of fake GPU device declared by this device plugin by changing the value of the DP_NUMBER_CONTAINERS_PER_GPU environment variable in the DaemonSet definition (default: 100).

About

The NVIDIA device plugin for Kubernetes is a Daemonset that allows you to automatically:

• Expose the number of GPUs on each nodes of your cluster
• Keep track of the health of your GPUs
• Run GPU enabled containers in your Kubernetes cluster.

This repository contains NVIDIA's official implementation of the Kubernetes device plugin.

Prerequisites

The list of prerequisites for running the NVIDIA device plugin is described below:

• NVIDIA drivers ~= 361.93
• nvidia-docker version > 2.0 (see how to install and it's prerequisites)
• docker configured with nvidia as the default runtime.
• Kubernetes version = 1.10
• The DevicePlugins feature gate enabled

Quick Start

Preparing your GPU Nodes

The following steps need to be executed on all your GPU nodes. Additionally, this README assumes that the NVIDIA drivers and nvidia-docker has been installed.

First you will need to check and/or enable the nvidia runtime as your default runtime on your node. We will be editing the docker daemon config file which is usually present at /etc/docker/daemon.json:

{
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

if runtimes is not already present, head to the install page of nvidia-docker

The second step is to enable the DevicePlugins feature gate on all your GPU nodes.

If your Kubernetes cluster is deployed using kubeadm and your nodes are running systemd you will have to open the kubeadm systemd unit file at /etc/systemd/system/kubelet.service.d/10-kubeadm.conf and add the following environment argument:

Environment="KUBELET_EXTRA_ARGS=--feature-gates=DevicePlugins=true"

If you spot the Accelerators feature gate you should remove it as it might interfere with the DevicePlugins feature gate

Reload and restart the kubelet to pick up the config change:

$ sudo systemctl daemon-reload
$ sudo systemctl restart kubelet

In this guide we used kubeadm and kubectl as the method for setting up and administering the Kubernetes cluster, but there are many ways to deploy a Kubernetes cluster. To enable the DevicePlugins feature gate if you are not using the kubeadm + systemd configuration, you will need to make sure that the arguments that are passed to Kubelet include the following --feature-gates=DevicePlugins=true.

Enabling GPU Support in Kubernetes

Once you have enabled this option on all the GPU nodes you wish to use, you can then enable GPU support in your cluster by deploying the following Daemonset:

$ kubectl create -f https://raw.githubusercontent.com/Deepomatic/shared-gpu-nvidia-k8s-device-plugin/v1.10/deepomatic-shared-gpu-nvidia-device-plugin.yml

Running GPU Jobs

NVIDIA GPUs can now be consumed via container level resource requirements using the resource name deepomatic.com/shared-gpu:

apiVersion: v1
kind: Pod
metadata:
  name: shared-gpu-pod
spec:
  containers:
    - name: cuda-container
      image: nvidia/cuda:9.0-devel
      resources:
        limits:
          deepomatic.com/shared-gpu: 1 # requesting 1 shared GPU
    - name: digits-container
      image: nvidia/digits:6.0
      resources:
        limits:
          deepomatic.com/shared-gpu: 1 # requesting 1 shared GPU

WARNING: if you don't request GPUs when using the device plugin with NVIDIA images all the GPUs on the machine will be exposed inside your container.

Docs

Please note that:

• the device plugin feature is still alpha which is why it requires the feature gate to be enabled.
• the NVIDIA device plugin is still considered alpha and is missing
  • Security features
  • More comprehensive GPU health checking features
  • GPU cleanup features
  • ...
• support will only be provided for the official NVIDIA device plugin.

The next sections are focused on building the device plugin and running it.

With Docker

Build

Option 1, pull the prebuilt image from Docker Hub:

$ docker pull deepomatic/shared-gpu-nvidia-k8s-device-plugin:1.10

Option 2, build without cloning the repository:

$ docker build -t deepomatic/shared-gpu-nvidia-k8s-device-plugin:1.10 https://github.com/deepomatic/shared-gpu-nvidia-k8s-device-plugin.git#v1.10

Option 3, if you want to modify the code:

$ git clone https://github.com/deepomatic/shared-gpu-nvidia-k8s-device-plugin.git && cd shared-gpu-nvidia-k8s-device-plugin
$ docker build -t deepomatic/shared-gpu-nvidia-k8s-device-plugin:1.10 .

Run locally

$ docker run --security-opt=no-new-privileges --cap-drop=ALL --network=none -it -v /var/lib/kubelet/device-plugins:/var/lib/kubelet/device-plugins deepomatic/shared-gpu-nvidia-k8s-device-plugin:1.10

Deploy as Daemon Set:

$ kubectl create -f deepomatic-shared-gpu-nvidia-device-plugin.yml

Without Docker

Build

$ C_INCLUDE_PATH=/usr/local/cuda/include LIBRARY_PATH=/usr/local/cuda/lib64 go build

Run locally

$ ./k8s-device-plugin

Changelog

Version 1.10

• The device Plugin API is now v1beta1

Version 1.9

• The device Plugin API changed and is no longer compatible with 1.8
• Error messages were added

Issues and Contributing

• You can report a bug by filing a new issue
• You can contribute by opening a pull request