Adphi
8505dbd02a
refactor: move qemu to its own package tests: implements end to end tests for the convert command with the following images: alpine:3.17, ubuntu:20.04, debian:11, centos:8 Signed-off-by: Adphi <philippe.adrien.nousse@gmail.com> |
||
---|---|---|
.github/workflows | ||
cmd/d2vm | ||
docs | ||
e2e | ||
examples | ||
pkg | ||
scripts | ||
templates | ||
.dockerignore | ||
.gitignore | ||
.goreleaser.yaml | ||
builder_test.go | ||
builder.go | ||
container_disk.go | ||
convert_options.go | ||
convert.go | ||
docker_image_test.go | ||
docker_image.go | ||
Dockerfile | ||
dockerfile.go | ||
go.mod | ||
go.sum | ||
LICENCE | ||
Makefile | ||
os_release.go | ||
pgp.pub | ||
qemu.sh | ||
README.md | ||
version.go | ||
virtinst |
d2vm (Docker to Virtual Machine)
Build virtual machine image from Docker images
The project is heavily inspired by the article and the work done by iximiuz on docker-to-linux.
Many thanks to him.
Status: alpha
Supported Environments:
Only building Linux Virtual Machine images is supported.
Starting from v0.1.0, d2vm automatically run build and convert commands inside Docker when not running on linux or when running without root privileges.
Note: windows should be working, but is totally untested.
Supported VM Linux distributions:
Working and tested:
- Ubuntu (18.04+)
- Debian (stretch+)
- Alpine
- CentOS (8+)
Unsupported:
- RHEL
The program uses the /etc/os-release
file to discover the Linux distribution and install the Kernel,
if the file is missing, the build cannot succeed.
Obviously, Distroless images are not supported.
Prerequisites
osx
- Docker
- QEMU (optional)
- VirtualBox (optional)
Linux
- Docker
- util-linux
- udev
- parted
- e2fsprogs
- mount
- tar
- extlinux
- qemu-utils
- QEMU (optional)
- VirtualBox (optional)
Getting started
Install
With Docker
Note: this will only work if both the source context (and Dockerfile) and the output directory are somewhere inside the directory where you run the command.
docker pull linkacloud/d2vm:latest
alias d2vm="docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock --privileged -v \$PWD:/d2vm -w /d2vm linkacloud/d2vm:latest"
wich d2vm
d2vm: aliased to docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock --privileged -v $PWD:/d2vm -w /d2vm linkacloud/d2vm:latest
With Homebrew
brew install linka-cloud/tap/d2vm
From release
Download the latest release for your platform from the release page.
Extract the tarball, then move the extracted d2vm binary to somewhere in your $PATH
(/usr/local/bin
for most users).
VERSION=$(git ls-remote --tags https://github.com/linka-cloud/d2vm |cut -d'/' -f 3|tail -n 1)
OS=$(uname -s | tr '[:upper:]' '[:lower:]')
ARCH=$([ "$(uname -m)" = "x86_64" ] && echo "amd64" || echo "arm64")
curl -sL "https://github.com/linka-cloud/d2vm/releases/download/${VERSION}/d2vm_${VERSION}_${OS}_${ARCH}.tar.gz" | tar -xvz d2vm
sudo mv d2vm /usr/local/bin/
From source
Clone the git repository:
git clone https://github.com/linka-cloud/d2vm && cd d2vm
Install using the make, docker and the Go tool chain:
make install
The d2vm binary is installed in the $GOBIN
directory.
which d2vm
/go/bin/d2vm
Generate shell completion
The d2vm program supports shell completion for bash, zsh and fish.
It can be enabled by running the following command:
source <(d2vm completion $(basename $SHELL))
Or you can install the completion file in the shell completion directory by following the instructions:
d2vm completion $(basename $SHELL) --help
Converting an existing Docker Image to VM image:
d2vm convert --help
Convert Docker image to vm image
Usage:
d2vm convert [docker image] [flags]
Flags:
--append-to-cmdline string Extra kernel cmdline arguments to append to the generated one
--boot-size uint Size of the boot partition in MB (default 100)
--force Override output qcow2 image
-h, --help help for convert
--network-manager string Network manager to use for the image: none, netplan, ifupdown
-o, --output string The output image, the extension determine the image format, raw will be used if none. Supported formats: qcow2 qed raw vdi vhd vmdk (default "disk0.qcow2")
-p, --password string Optional root user password
--pull Always pull docker image
--push Push the container disk image to the registry
--raw Just convert the container to virtual machine image without installing anything more
-s, --size string The output image size (default "10G")
--split-boot Split the boot partition from the root partition
-t, --tag string Container disk Docker image tag
Global Flags:
--time string Enable formated timed output, valide formats: 'relative (rel | r)', 'full (f)' (default "none")
-v, --verbose Enable Verbose output
Create an image based on the ubuntu official image:
sudo d2vm convert ubuntu -o ubuntu.qcow2 -p MyP4Ssw0rd
Pulling image ubuntu
Inspecting image ubuntu
No network manager specified, using distribution defaults: netplan
Docker image based on Ubuntu 22.04.1 LTS (Jammy Jellyfish)
Building kernel enabled image
Creating vm image
Creating raw image
Mounting raw image
Creating raw image file system
Copying rootfs to raw image
Setting up rootfs
Installing linux kernel
Unmounting raw image
Writing MBR
Converting to qcow2
You can now run your ubuntu image using the created ubuntu.qcow2
image with qemu:
d2vm run qemu ubuntu.qcow2
SeaBIOS (version 1.13.0-1ubuntu1.1)
iPXE (http://ipxe.org) 00:03.0 CA00 PCI2.10 PnP PMM+BFF8C920+BFECC920 CA00
Booting from Hard Disk...
SYSLINUX 6.04 EDD 20191223 Copyright (C) 1994-2015 H. Peter Anvin et al
Now booting the kernel from SYSLINUX...
Loading /boot/vmlinuz... ok
Loading /boot/initrd.img...ok
[ 0.000000] Linux version 5.4.0-109-generic (buildd@ubuntu) (gcc version 9)
[ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz ro root=UUID=b117d206-b8
[ 0.000000] KERNEL supported cpus:
[ 0.000000] Intel GenuineIntel
[ 0.000000] AMD AuthenticAMD
[ 0.000000] Hygon HygonGenuine
[ 0.000000] Centaur CentaurHauls
[ 0.000000] zhaoxin Shanghai
...
Welcome to Ubuntu 20.04.4 LTS!
[ 3.610631] systemd[1]: Set hostname to <localhost>.
[ 3.838984] systemd[1]: Created slice system-getty.slice.
[ OK ] Created slice system-getty.slice.
[ 3.845038] systemd[1]: Created slice system-modprobe.slice.
[ OK ] Created slice system-modprobe.slice.
[ 3.852054] systemd[1]: Created slice system-serial\x2dgetty.slice.
[ OK ] Created slice system-serial\x2dgetty.slice.
...
Ubuntu 20.04.4 LTS localhost ttyS0
localhost login:
Log in using the root user and the password configured at build time.
localhost login: root
Password:
Welcome to Ubuntu 20.04.4 LTS (GNU/Linux 5.4.0-109-generic x86_64)
* Documentation: https://help.ubuntu.com
* Management: https://landscape.canonical.com
* Support: https://ubuntu.com/advantage
This system has been minimized by removing packages and content that are
not required on a system that users do not log into.
To restore this content, you can run the 'unminimize' command.
The programs included with the Ubuntu system are free software;
the exact distribution terms for each program are described in the
individual files in /usr/share/doc/*/copyright.
Ubuntu comes with ABSOLUTELY NO WARRANTY, to the extent permitted by
applicable law.
root@localhost:~#
Type poweroff
to shut down the vm.
Building a VM Image from a Dockerfile
The example directory contains very minimalistic examples:
cd examples
ubuntu.Dockerfile :
FROM ubuntu
RUN apt update && apt install -y openssh-server && \
echo "PermitRootLogin yes" >> /etc/ssh/sshd_config
Build the vm image:
The build command take most of its flags and arguments from the docker build command.
d2vm build --help
Build a vm image from Dockerfile
Usage:
d2vm build [context directory] [flags]
Flags:
--append-to-cmdline string Extra kernel cmdline arguments to append to the generated one
--boot-size uint Size of the boot partition in MB (default 100)
--build-arg stringArray Set build-time variables
-f, --file string Name of the Dockerfile
--force Override output qcow2 image
-h, --help help for build
--network-manager string Network manager to use for the image: none, netplan, ifupdown
-o, --output string The output image, the extension determine the image format, raw will be used if none. Supported formats: qcow2 qed raw vdi vhd vmdk (default "disk0.qcow2")
-p, --password string Optional root user password
--push Push the container disk image to the registry
--raw Just convert the container to virtual machine image without installing anything more
-s, --size string The output image size (default "10G")
--split-boot Split the boot partition from the root partition
-t, --tag string Container disk Docker image tag
Global Flags:
--time string Enable formated timed output, valide formats: 'relative (rel | r)', 'full (f)' (default "none")
-v, --verbose Enable Verbose output
sudo d2vm build -p MyP4Ssw0rd -f ubuntu.Dockerfile -o ubuntu.qcow2 .
Or if you want to create a VirtualBox image:
sudo d2vm build -p MyP4Ssw0rd -f ubuntu.Dockerfile -o ubuntu.vdi .
KubeVirt Container Disk Images
Using the --tag
flag with the build
and convert
commands, you can create a
Container Disk Image for KubeVirt.
The --push
flag will push the image to the registry.
Complete example
A complete example setting up a ZSH workstation is available in the examples/full directory.
Internal Dockerfile templates
You can find the Dockerfiles used to install the Kernel in the templates directory.
TODO / Questions:
- Create service from
ENTRYPOINT
CMD
WORKDIR
andENV
instructions ? - Inject Image
ENV
variables into.bashrc
or other service environment file ? - Use image layers to create rootfs instead of container ?
Acknowledgments
The run commands are adapted from linuxkit.