This post provides the steps required to build the Docker daemon for 32-bit Linux platforms.
Background
Software today is driven by speed – move fast, learn fast, and deliver fast! Nirmata, my company, enables software developers to move fast by making it super easy to deploy, operate, and optimize their containerized applications. Nirmata builds on Docker, the most popular container technology around. Docker is commonly used in x64 architectures and with popular Linux distros’ like Ubuntu, RHEL, CoreOS. However, there continues to be a steady demand for deploying container on other platforms and architectures like Yocto, Kali Linux and 32-bit architectures.
At Nirmata, we are working with a customer in the IoT space who required a 32-bit Docker daemon. Although Docker continues to add support for many platforms and architectures, it does not seem like official 32-bit support is coming anytime soon (issue #136). There is an existing blog post, which helps build Docker on 32-bit platform, but a lot has changed in Docker world since then.
In this post, I have documented the steps I used to build and use Docker on 32-bit platforms:
Overview
Docker docs provide a well-detailed instructions set for setting up a dev environment, however, it doesn’t provide any instructions to build Docker daemon for 32-bit platform.
Docker provides a Dockerfile in the root directory of its Git repository for building a Docker dev environment. The Dockerfile builds a base image; which has all the tools and binaries required for building Docker daemon binary. For building a daemon compatible with 32-bit, one can follow the same steps what Docker doc suggests, with the following changes to the Dockerfile.
- Use 32-bit Ubuntu image as the base image
- Use clang-3.6 version
- Install Go for 32 bit platform
Note: These steps are tested on Ubuntu virtual machine with at least 2GB memory, but should work the same on any x86_64 Linux distribution.
Detailed Steps
Here are the complete steps to build a 32-bit Docker daemon from Docker Github branch:
- Install git
$ sudo apt-get install git-all
- Install make
$ sudo apt-get install build-essential
- To install Docker, follow this tutorial:
https://docs.docker.com/engine/installation/ubuntulinux/
- Get Docker Source Code
$ git clone https://github.com/docker/docker.git $ cd docker/ $ mv Dockerfile Dockerfile.backup
- wget the Dockerfile to create 32-bit container
$ wget https://gist.githubusercontent.com/prateekgogia/05f058bafbccc2478fcc/raw/1db60ea471678cfb55185215defcf371f7dcec1d/Dockerfile
- We can now build the Docker build environment. This will create a Docker-dev image with all the required tools. Running this command for the first time will take sometime.
$ make build
- Once the command completes, run ‘Docker images’ and it should show a Docker-dev image. This image has all the tools and binaries required to build Docker daemon. Now, we can build a 32-bit Docker daemon using docker-dev container image
$ make binary
A docker binary for 32-bit platform will be generated in directory – “bundles/latest/binary”.
A 32-bit Docker daemon will only support 32-bit container images. There are a few 32-bit images available on Docker hub. If you are using Go, you can also create a static container with a 32-bit binary to run images on 32-bit Linux Platform.
Summary
It’s a great time to be a software developer, and at Nirmata we are excited to be building a platform that empowers developers by allowing them to focus on their applications, while we help take care of the heavy lifting for applications operations and management.
I hope this brief post helps you with your software adventures! For any further questions feel free to reach out to me at: prateek@nirmata.com.
UPDATE (April/18/2017): There has been lot of updates to Docker build procedure since this post, this was original written with 1.9.* Docker version. I have updated the Dockerfile provided above for the current Docker version 17.04.0-ce.
PS- Latest build procedure doesn’t support OSX cross compilation for Docker and CRIU binary, which are available in upstream Dockerfile in Docker repo.
For more updates and news follow us at:
Sorry, the comment form is closed at this time.