Docker Installation On Amazon Linux
In order to setup the nvidia-docker repository for your distribution, follow the instructions below.
- Install Docker-compose On Amazon Linux 2
- Docker Installation On Amazon Linux
- Docker Installation On Amazon Linux Mint
An instance with an attached NVIDIA GPU, such as a P3 or G4dn instance, must have the appropriate NVIDIA driver installed. Depending on the instance type, you can either download a public NVIDIA driver, download a driver from Amazon S3 that is available only to AWS customers, or use an AMI with the driver pre-installed. I’m using Ubuntu 18.04, but the same info applies to most any Linux distribution. Installing Docker on Linux takes just a few commands. More installation info is available in the Docker Documentation. If you already have an older version of Docker, make sure to uninstall it first. Using buildx requires Docker 19.03 and today.
If you feel something is missing or requires additional information, please let us know by filing a new issue.
List of supported distributions:
| OS Name / Version | Identifier | amd64 / x86_64 | ppc64le | arm64 / aarch64 |
|---|---|---|---|---|
| Amazon Linux 1 | amzn1 | ✔ | ||
| Amazon Linux 2 | amzn2 | ✔ | ||
| Amazon Linux 2017.09 | amzn2017.09 | ✔ | ||
| Amazon Linux 2018.03 | amzn2018.03 | ✔ | ||
| Open Suse Leap 15.0 | sles15.0 | ✔ | ||
| Open Suse Leap 15.1 | sles15.1 | ✔ | ||
| Debian Linux 9 | debian9 | ✔ | ||
| Debian Linux 10 | debian10 | ✔ | ||
| Centos 7 | centos7 | ✔ | ✔ | |
| Centos 8 | centos8 | ✔ | ✔ | ✔ |
| RHEL 7.4 | rhel7.4 | ✔ | ✔ | |
| RHEL 7.5 | rhel7.5 | ✔ | ✔ | |
| RHEL 7.6 | rhel7.6 | ✔ | ✔ | |
| RHEL 7.7 | rhel7.7 | ✔ | ✔ | |
| RHEL 7.8 | rhel7.8 | ✔ | ✔ | |
| RHEL 7.9 | rhel7.9 | ✔ | ✔ | |
| RHEL 8.0 | rhel8.0 | ✔ | ✔ | ✔ |
| RHEL 8.1 | rhel8.1 | ✔ | ✔ | ✔ |
| RHEL 8.2 | rhel8.2 | ✔ | ✔ | ✔ |
| RHEL 8.3 | rhel8.3 | ✔ | ✔ | ✔ |
| Ubuntu 16.04 | ubuntu16.04 | ✔ | ✔ | |
| Ubuntu 18.04 | ubuntu18.04 | ✔ | ✔ | ✔ |
| Ubuntu 19.04 | ubuntu19.04 | ✔ | ✔ | ✔ |
| Ubuntu 19.10 | ubuntu19.10 | ✔ | ✔ | ✔ |
| Ubuntu 20.04 | ubuntu20.04 | ✔ | ✔ | ✔ |
Debian-based distributions
For pre-releases, you need to enable the experimental repos of all dependencies:
To later disable the experimental repos of all dependencies, you can run:
RHEL-based distributions
For pre-releases, you need to enable the experimental repos of all dependencies:
To later disable the experimental repos of all dependencies, you can run:
In order to update the nvidia-docker repository key for your distribution, follow the instructions below.
RHEL-based distributions
Amazon Linux (1 and 2)
Be careful to run each instruction one by one!
Debian-based distributions
This blog post is the result of collaboration between Arm and Docker. Special thanks to Jason Andrews @ Arm for creating much of the original content.
Arm and Docker announced a strategic partnership earlier this year to unify software development and deployment across a diverse set of devices, from IoT endpoints to the edge of the network, and into the heart of the data center. Docker has simplified enterprise software development and deployment leading to true multi-platform portability and cost savings on Arm-based cloud instances. Even more exciting is how Docker is changing the way embedded software is being developed and deployed.
Traditionally embedded Linux software applications have been created by cross-compiling and copying files to an embedded target board. There are various methods to automate this process, but it has generally been unchanged since the 1990’s when non-x86 embedded possessors running Linux appeared. Docker stands to make the first significant change to the embedded Linux application developer’s workflow.
This article continues from Building Multi-Arch Images for Arm and x86 with Docker Desktop and shows the same capabilities in Linux. Although Windows and Mac support is great, the majority of software developers targeting embedded Linux systems also do their development work on Linux. The multi-architecture support in Docker also greatly simplifies embedded Linux application development and deployment.
If you are doing software development on x86 Linux machines and want to create Docker images that run on Arm servers or Arm embedded and IoT devices, this article will be helpful to understand the process and the different ways to do it.
Let’s see how to use Docker for Arm software development using the new buildx feature on Linux to create multi-architecture container images and run them. I’m using Ubuntu 18.04, but the same info applies to most any Linux distribution.
Install Docker
Installing Docker on Linux takes just a few commands. More installation info is available in the Docker Documentation.
If you already have an older version of Docker, make sure to uninstall it first. Using buildx requires Docker 19.03 and today the best way to get this is using the test instead of the stable version.
Add the current user to the docker group to avoid needing sudo to run the docker command:
Make sure to log out and back in again. Now test the install with a quick hello-world run.
Use the docker version command to check the running version:
Install buildx for multi-architecture image builds
Install Docker-compose On Amazon Linux 2
There are three options to get buildx on Linux:
- Use buildx directly from the test channel version of Docker
- Download a binary release of buildx and copy it to the $HOME/.docker directory
- Download, build, and install buildx from github.com
Use buildx from Docker test channel
The test version of Docker already has buildx included. The only thing needed is to set the environment variable to enable experimental command line features.
Download a binary release
Another way to get buildx is to download a binary release from github and put in the .docker/cli-plugins directory.
For example, download the buildx for Linux amd64 with a browser from: https://github.com/docker/buildx/releases/tag/v0.2.0
Then copy it to the cli-plugins/ directory (create it first if necessary):
Download, build, and install buildx
Because buildx is a new command and documentation is still catching up, githubis a good place to read more information about how buildx works.
To get buildx from github use the commands:
To confirm buildx is now installed run the help and the version command.
Register Arm executables to run on x64 machines
Install the qemu instruction emulation to register Arm executables to run on the x86 machine. For best results, the latest qemu image should be used. If an older qemu is used some application may not work correctly on the x86 hardware.
To verify the qemu handlers are registered properly, run the following and make sure the first line of the output is “enabled”. Note that the handler registration doesn’t survive a reboot, but could be added to the system start-up scripts.
Create a multi-architecture build instance
Setup a new builder instance to create multi-architecture images.
Try buildx
There are multiple examples of buildx available, but here is a simple one for C programmers! Create a file named hello.c with this code:
Here is a Docker file to build and run it. Let’s get used to using multi-stage Docker files as it will be common for deploying embedded applications. Create a Dockerfile with the following:
Now, use buildx to build for multiple architectures and push to Docker hub.
Use docker login first if needed and substitute your own Hub account.
Run using the sha from the manifest and see the output from uname as armv7l, aarch64, and x86_64:
Next steps
Docker Installation On Amazon Linux
As we have seen, building multi-architecture containers can be created with buildx in the same way as with Docker Desktop for Mac and Windows. Give it a try for yourself and start making the transition to multi-architecture Docker images today.
Docker Installation On Amazon Linux Mint
Further Reading