mirror of
https://github.com/intel/intel-device-plugins-for-kubernetes.git
synced 2025-06-03 03:59:37 +00:00
docs: gpu: Add more details, re-arrange section order
Re-arrange the section order a little (such as putting the use of the DaemonSet before the sudo hand-deploy), and add a lot more detail of what to expect, and how to check if the pod has launched correctly. Signed-off-by: Graham Whaley <graham.whaley@intel.com>
This commit is contained in:
Graham Whaley
parent
6705a8e461
commit
79a86c10e8
@ -2,19 +2,18 @@
|
||||
|
||||
# Table of Contents
|
||||
|
||||
|
||||
* [Introduction](#introduction)
|
||||
* [Build and test](#build-and-test)
|
||||
* [Getting the source code:](#getting-the-source-code)
|
||||
* [Verify kubelet socket exists in /var/lib/kubelet/device-plugins/ directory:](#verify-kubelet-socket-exists-in-varlibkubeletdevice-plugins-directory)
|
||||
* [Deploy GPU device plugin as host process for development purposes](#deploy-gpu-device-plugin-as-host-process-for-development-purposes)
|
||||
* [Build GPU device plugin:](#build-gpu-device-plugin)
|
||||
* [Run GPU device plugin as administrator:](#run-gpu-device-plugin-as-administrator)
|
||||
* [Deploy GPU device plugin as a DaemonSet:](#deploy-gpu-device-plugin-as-a-daemonset)
|
||||
* [Build plugin image](#build-plugin-image)
|
||||
* [Create plugin DaemonSet](#create-plugin-daemonset)
|
||||
* [Verify GPU device plugin is registered on master:](#verify-gpu-device-plugin-is-registered-on-master)
|
||||
* [Test GPU device plugin:](#test-gpu-device-plugin)
|
||||
* [Installation](#installation)
|
||||
* [Getting the source code](#getting-the-source-code)
|
||||
* [Verify node kubelet config](#verify-node-kubelet-config)
|
||||
* [Deploying as a DaemonSet](#deploying-as-a-daemonset)
|
||||
* [Build the plugin image](#build-the-plugin-image)
|
||||
* [Deploy plugin DaemonSet](#deploy-plugin-daemonset)
|
||||
* [Deploy by hand](#deploy-by-hand)
|
||||
* [Build the plugin](#build-the-plugin)
|
||||
* [Run the plugin as administrator](#run-the-plugin-as-administrator)
|
||||
* [Verify plugin registration](#verify-plugin-registration)
|
||||
* [Testing the plugin](#testing-the-plugin)
|
||||
|
||||
# Introduction
|
||||
|
||||
@ -26,90 +25,166 @@ and acceleration, supporting GPUs of the following hardware families:
|
||||
- Integrated GPUs within Intel Xeon processors
|
||||
- Intel Visual Compute Accelerator (Intel VCA)
|
||||
|
||||
The GPU plugin offloads the processing of computation intensive workloads to GPU hardware.
|
||||
The GPU plugin facilitates offloading the processing of computation intensive workloads to GPU hardware.
|
||||
There are two primary use cases:
|
||||
|
||||
- hardware vendor-independent acceleration using the [Intel Media SDK](https://github.com/Intel-Media-SDK/MediaSDK)
|
||||
- OpenCL code tuned for high end Intel devices.
|
||||
|
||||
For example, the Intel Media SDK can offload video transcoding operations, and the OpenCL™ libraries can provide computation acceleration for Intel GPUs
|
||||
For example, the Intel Media SDK can offload video transcoding operations, and the OpenCL libraries can provide computation acceleration for Intel GPUs
|
||||
|
||||
For information on Intel GVT-g virtual GPU device passthrough, see
|
||||
For information on Intel GVT-g virtual GPU device passthrough (as opposed to full device passthrough), see
|
||||
[this site](https://github.com/intel/gvt-linux/wiki/GVTg_Setup_Guide).
|
||||
|
||||
## Build and test
|
||||
# Installation
|
||||
|
||||
The following sections detail how to obtain, build, test and deploy the GPU device plugin.
|
||||
The following sections detail how to obtain, build, deploy and test the GPU device plugin.
|
||||
|
||||
### Getting the source code:
|
||||
Examples are provided showing how to deploy the plugin either using a DaemonSet or by hand on a per-node basis.
|
||||
|
||||
```
|
||||
$ mkdir -p $GOPATH/src/github.com/intel/
|
||||
$ cd $GOPATH/src/github.com/intel/
|
||||
$ git clone https://github.com/intel/intel-device-plugins-for-kubernetes.git
|
||||
## Getting the source code
|
||||
|
||||
> **Note:** It is presumed you have a valid and configured [golang](https://golang.org/) environment
|
||||
> that meets the minimum required version.
|
||||
|
||||
```bash
|
||||
$ go get -d -u github.com/intel/intel-device-plugins-for-kubernetes
|
||||
```
|
||||
|
||||
### Verify kubelet socket exists in /var/lib/kubelet/device-plugins/ directory:
|
||||
```
|
||||
## Verify node kubelet config
|
||||
|
||||
Every node that will be running the gpu plugin must have the
|
||||
[kubelet device-plugins](https://kubernetes.io/docs/concepts/extend-kubernetes/compute-storage-net/device-plugins/)
|
||||
configured. For each node, check that the kubelet device plugin socket exists:
|
||||
|
||||
```bash
|
||||
$ ls /var/lib/kubelet/device-plugins/kubelet.sock
|
||||
/var/lib/kubelet/device-plugins/kubelet.sock
|
||||
```
|
||||
|
||||
### Deploy GPU device plugin as host process for development purposes
|
||||
## Deploying as a DaemonSet
|
||||
|
||||
#### Build GPU device plugin:
|
||||
To deploy the gpu plugin as a daemonset, you first need to build a container image for the
|
||||
plugin and ensure that is visible to your nodes.
|
||||
|
||||
### Build the plugin image
|
||||
|
||||
The following will use `docker` to build a local container image called
|
||||
`intel/intel-gpu-plugin` with the tag `devel`.
|
||||
|
||||
The image build tool can be changed from the default `docker` by setting the `BUILDER` argument
|
||||
to the [`Makefile`](Makefile).
|
||||
|
||||
```bash
|
||||
$ cd $GOPATH/src/github.com/intel/intel-device-plugins-for-kubernetes
|
||||
$ make intel-gpu-plugin
|
||||
...
|
||||
Successfully tagged intel/intel-gpu-plugin:devel
|
||||
```
|
||||
|
||||
### Deploy plugin DaemonSet
|
||||
|
||||
You can then use the example DaemonSet YAML file provided to deploy the plugin.
|
||||
|
||||
```bash
|
||||
$ kubectl create -f ./deployments/gpu_plugin/gpu_plugin.yaml
|
||||
daemonset.apps/intel-gpu-plugin created
|
||||
```
|
||||
|
||||
> **Note**: It is also possible to run the GPU device plugin using a non-root user. To do this,
|
||||
the nodes' DAC rules must be configured to device plugin socket creation and kubelet registration.
|
||||
Furthermore, the deployments `securityContext` must be configured with appropriate `runAsUser/runAsGroup`.
|
||||
|
||||
## Deploy by hand
|
||||
|
||||
For development purposes, it is sometimes convenient to deploy the plugin 'by hand' on a node.
|
||||
In this case, you do not need to build the complete container image, and can build just the plugin.
|
||||
|
||||
### Build the plugin
|
||||
|
||||
First we build the plugin:
|
||||
|
||||
```bash
|
||||
$ cd $GOPATH/src/github.com/intel/intel-device-plugins-for-kubernetes
|
||||
$ make gpu_plugin
|
||||
```
|
||||
|
||||
#### Run GPU device plugin as administrator:
|
||||
```
|
||||
### Run the plugin as administrator
|
||||
|
||||
Now we can run the plugin directly on the node:
|
||||
|
||||
```bash
|
||||
$ sudo $GOPATH/src/github.com/intel/intel-device-plugins-for-kubernetes/cmd/gpu_plugin/gpu_plugin
|
||||
device-plugin start server at: /var/lib/kubelet/device-plugins/gpu.intel.com-i915.sock
|
||||
device-plugin registered
|
||||
```
|
||||
|
||||
### Deploy GPU device plugin as a DaemonSet:
|
||||
## Verify plugin registration
|
||||
|
||||
#### Build plugin image
|
||||
```
|
||||
$ make intel-gpu-plugin
|
||||
You can verify the plugin has been registered with the expected nodes by searching for the relevant
|
||||
resource allocation status on the nodes:
|
||||
|
||||
```bash
|
||||
$ kubectl get nodes -o=jsonpath="{range .items[*]}{.metadata.name}{'\n'}{' i915: '}{.status.allocatable.gpu\.intel\.com/i915}{'\n'}"
|
||||
master
|
||||
i915: 1
|
||||
```
|
||||
|
||||
#### Create plugin DaemonSet
|
||||
```
|
||||
$ kubectl create -f ./deployments/gpu_plugin/gpu_plugin.yaml
|
||||
daemonset.apps/intel-gpu-plugin created
|
||||
```
|
||||
## Testing the plugin
|
||||
|
||||
**Note**: It is also possible to run the GPU device plugin using a non-root user. To do this,
|
||||
the nodes' DAC rules must be configured to device plugin socket creation and kubelet registration.
|
||||
Furthermore, the deployments `securityContext` must be configured with appropriate `runAsUser/runAsGroup`.
|
||||
|
||||
### Verify GPU device plugin is registered on master:
|
||||
```
|
||||
$ kubectl describe node <node name> | grep gpu.intel.com
|
||||
gpu.intel.com/i915: 1
|
||||
gpu.intel.com/i915: 1
|
||||
```
|
||||
|
||||
### Test GPU device plugin:
|
||||
We can test the plugin is working by deploying the provided example OpenCL image with FFT offload enabled.
|
||||
|
||||
1. Build a Docker image with an example program offloading FFT computations to GPU:
|
||||
```
|
||||
$ cd demo
|
||||
$ ./build-image.sh ubuntu-demo-opencl
|
||||
```
|
||||
|
||||
This command produces a Docker image named `ubuntu-demo-opencl`.
|
||||
```bash
|
||||
$ cd demo
|
||||
$ ./build-image.sh ubuntu-demo-opencl
|
||||
...
|
||||
Successfully tagged ubuntu-demo-opencl:devel
|
||||
```
|
||||
|
||||
2. Create a pod running unit tests off the local Docker image:
|
||||
```
|
||||
$ kubectl apply -f demo/intelgpu-job.yaml
|
||||
```
|
||||
1. Create a job running unit tests off the local Docker image:
|
||||
|
||||
3. Review the pod's logs:
|
||||
```
|
||||
$ kubectl logs intelgpu-demo-job-xxxx
|
||||
```
|
||||
```bash
|
||||
$ cd $GOPATH/src/github.com/intel/intel-device-plugins-for-kubernetes
|
||||
$ kubectl apply -f demo/intelgpu-job.yaml
|
||||
job.batch/intelgpu-demo-job created
|
||||
```
|
||||
|
||||
1. Review the job's logs:
|
||||
|
||||
```bash
|
||||
$ kubectl get pods | fgrep intelgpu
|
||||
# substitute the 'xxxxx' below for the pod name listed in the above
|
||||
$ kubectl logs intelgpu-demo-job-xxxxx
|
||||
+ WORK_DIR=/root/6-1/fft
|
||||
+ cd /root/6-1/fft
|
||||
+ ./fft
|
||||
+ uprightdiff --format json output.pgm /expected.pgm diff.pgm
|
||||
+ cat diff.json
|
||||
+ jq .modifiedArea
|
||||
+ MODIFIED_AREA=0
|
||||
+ [ 0 -gt 10 ]
|
||||
+ echo Success
|
||||
Success
|
||||
```
|
||||
|
||||
If the pod did not successfully launch, possibly because it could not obtain the gpu
|
||||
resource, it will be stuck in the `Pending` status:
|
||||
|
||||
```bash
|
||||
$ kubectl get pods
|
||||
NAME READY STATUS RESTARTS AGE
|
||||
intelgpu-demo-job-xxxxx 0/1 Pending 0 8s
|
||||
```
|
||||
|
||||
This can be verified by checking the Events of the pod:
|
||||
|
||||
```bash
|
||||
$ kubectl describe pod intelgpu-demo-job-xxxxx
|
||||
...
|
||||
Events:
|
||||
Type Reason Age From Message
|
||||
---- ------ ---- ---- -------
|
||||
Warning FailedScheduling <unknown> default-scheduler 0/1 nodes are available: 1 Insufficient gpu.intel.com/i915.
|
||||
```
|
||||
|
||||
Loading…
Reference in New Issue
Block a user