OpenStack Guest Images

This documentation covers the steps needed to build OpenStack guests using a tool called packer. In addition, we include documentation which covers how to build the images manually in case you run into issues with packer.

Current Images

We are currently maintaining these images:

Image

x86_64

AARCH64

POWER8

POWER9

AlmaLinux 8

x

x

x

x

AlmaLinux 9

x

x

x

x

CentOS 7

x

x

x

x

CentOS Stream 8

x

x

x

x

CentOS Stream 9

x

x

x

x

Debian 10

x

x

x

x

Debian 11

x

x

x

x

Debian 12

x

x

x

x

Fedora (latest)

x

x

Rocky Linux 8

x

x

x

x

Rocky Linux 9

x

x

x

x

Ubuntu 18.04

x

x

x

x

Ubuntu 20.04

x

x

x

x

Ubuntu 22.04

x

x

x

We are currently only maintaining ppc64le Little endian (LE) images for POWER8/9. IBM is focusing their efforts primarily on LE, so Big endian (BE) support is limited to only Debian Sid (unstable) As all other distributions have removed support for BE.

If there are any other images you would like us to create, please let us know.

Building images with packer

We build all of our images using packer via our packer-templates github repo.

Manual Guest Installation

Note

This is only useful for manually testing some images and is no longer used for building images.

Note

For AARCH64, Please run the following commands on one of the machines directly. Also, make sure you have the AAVMF package installed which contains the UEFI firmware to boot the VMs properly.

  1. Download the DVD or net install ISO for desired distribution. For this example we are using Debian 10:

export DISTRO="debian-12"
export DISTRO_ISO="debian.iso"

  1. Create a 3G qcow2 disk image:

qemu-img create -f qcow2 $DISTRO.qcow2 3G

  1. Boot up the image using kvm manually:

x86 / POWER:

$ /usr/libexec/qemu-kvm -m 2048M -boot strict=on -name $DISTRO -machine type=pseries,accel=kvm \
  -cdrom $DISTRO_ISO -netdev user,id=user.0,hostfwd=tcp::2222-:22 \
  -device virtio-net,netdev=user.0 \
  -drive file=$DISTRO.qcow2,if=virtio,cache=writeback,discard=ignore,format=qcow2 \
  -vnc 0.0.0.0:99

AARCH64:

$ /usr/libexec/qemu-kvm -m 2048M -boot strict=on -name $DISTRO -machine gic-version=3,accel=kvm \
  -cdrom $DISTRO_ISO -netdev user,id=user.0,hostfwd=tcp::2222-:22 \
  -cpu host -monitor none \
  -drive file=$DISTRO.qcow2,if=virtio,cache=writeback,discard=ignore,format=qcow2 \
  -bios /usr/share/AAVMF/AAVMF_CODE.fd
  -vnc 0.0.0.0:99

  1. Connect via VNC to the VM and complete the installation:

$ vncviewer $host:99

  1. Wait for the VM to complete the install.

Uploading Images

Note

We do not recommend uploading images using the GUI interface for any images larger than 1G in size due to limits we have set with Apache. Instead we recommend you install using the CLI tools which will communication directly to the image service (glance).

  1. Install OpenStack CLI packages:

# Debian/Ubuntu
$ apt install python3-openstackclient

# Fedora
$ dnf install python3-openstackclient

# MacOS (Brew)
$ brew install openstackclient

Note

RHEL requires installing the RDO repository to install the python3-openstackclient package. Run dnf search openstack, select the latest release to install and then install the python3-openstackclient package.

This can also be installed using pip by installing the openstackclient meta package.

  1. Download and source openrc file

Login to the OpenStack GUI interface and then on your user on the upper right corner and pressing “OpenStack RC file”. That will download a file which you will then source using your shell environment.

  1. Import image into OpenStack:

Due to our backend storage uses Ceph, we recommend to upload images using the raw disk format. This allows for Copy-on-Write features being used which speeds up VM deployment. If you already have something in qcow2 format, you can easily convert it by doing the following:

$ qemu-img convert -O raw -p $DISTRO.qcow2 $DISTRO.raw

Now upload the image:

$ source openrc
$ openstack image create \
  --file $DISTRO-compressed.raw \
  --disk-format raw \
  --property hw_scsi_model=virtio-scsi \
  --property hw_disk_bus=scsi \
  --property hw_qemu_guest_agent=yes \
  --property os_require_quiesce=yes \
  $DISTRO

Note

The extra properties are optional but do align best with how our backend systems are configured. It will allow for a better user experience if you include those.

However, if you still prefer to use qcow2, make sure you compress it first:

$ qemu-img convert -O qcow2 -cp $DISTRO.qcow2 $DISTRO-compressed.qcow2
$ source openrc
$ openstack image create \
  --file $DISTRO-compressed.qcow2 \
  --disk-format qcow2 \
  --property hw_scsi_model=virtio-scsi \
  --property hw_disk_bus=scsi \
  --property hw_qemu_guest_agent=yes \
  --property os_require_quiesce=yes \
  $DISTRO