Skip to content

Deploying Red Hat Enterprise Linux or CentOS

This is a guide which walks through the process of deploying either Red Hat Enterprise Linux (RHEL) or CentOS from an operating system image or a Docker image.

Using an Operating System Image

RedHat provides both RHEL and CoreOS images in the RedHat provide RHEL Operating System images in the qcow2 format.

The CentOS images are available the cloud-images site.

RHEL images require a Red Hat Account in order to download, and are available at (login required):

A qcow2 filesystem image which is a full disk image including partition tables, partitions filled with filesystems and the files, and importantly, a boot loader at the beginning of the disk image. It will need to be converted to a raw filesystem image in order to use it.

Converting Image

In order to use this image, it needs to be converted into a raw filesystem. In order to do the conversion, install the qemu-img CLI tool.

apt-get install -y qemu-utils

Then, use the tool to convert the image into a raw filesystem. This example uses one of the CentOS images.

qemu-img convert  ./CentOS-8-GenericCloud-8.3.2011-20201204.2.x86_64.qcow2 -O raw ./CentOS-8-GenericCloud-8.3.2011-20201204.2.x86_64.raw

Optional - You can compress this raw image to save on both local disk space and network bandwidth when deploying the image.

gzip ./CentOS-8-GenericCloud-8.3.2011-20201204.2.x86_64.raw

Move the raw image to a locally accessible web server. To simplify, you can place the image in the Tinkerbell sandbox webroot, which allows access to the image at the IP address of the tink-server.

mv ./CentOS-8-GenericCloud-8.3.2011-20201204.2.x86_64.raw.gz ./sandbox/deploy/state/webroot

Fedora CoreOS

CentOS 8 is the last release and will be going EOL at the end of 2021, but following the acquisition of CoreOS by Red Hat, they distribute an additional operating system called Fedora CoreOS. It is available at Get Fedora, and distributed in both raw and qcow2 format.

fedora-coreos-33.20210217.3.0-metal.x86_64.qcow2.xz
fedora-coreos-33.20210217.3.0-metal.x86_64.raw.xz

Both images come with compressed with the xz compression format. You can decompress these image with the xz command.

xz -d <file.xz>

The raw disk image contains a full partition table (including OS and Swap partition) and boot loader for our Fedora CoreOS system, and can be used without converting it first.

The .qcow2.xz image is a full disk image including partition tables, partitions filled with filesystems and the files, and importantly, a boot loader at the beginning of the disk image. It will need to be converted to a raw filesystem image in order to use it, like RHEL and CentOS.

Creating the Template

The template uses actions from the Artifact Hub.

  • image2disk - to write the OS image to a block device.
  • kexec - to kexec into our newly provisioned operating system.

Important: Don't forget to pull, tag and push quay.io/tinkerbell-actions/image2disk:v1.0.0 prior to using it.

The example template uses the CentOS images, but you can modify it for other the other distributions such as RHEL or Fedora CoreOS.

version: '0.1'
name: CentOS_Deployment
global_timeout: 1800
tasks:
  - name: os-installation
    worker: '{{.device_1}}'
    volumes:
      - '/dev:/dev'
      - '/dev/console:/dev/console'
      - '/lib/firmware:/lib/firmware:ro'
    actions:
      - name: stream-image
        image: 'quay.io/tinkerbell-actions/image2disk:v1.0.0'
        timeout: 600
        environment:
          DEST_DISK: /dev/sda
          IMG_URL: >-
            http://192.168.1.1:8080/CentOS-8-GenericCloud-8.3.2011-20201204.2.x86_64.raw.gz
          COMPRESSED: true
      - name: kexec
        image: 'quay.io/tinkerbell-actions/kexec:v1.0.0'
        timeout: 90
        pid: host
        environment:
          BLOCK_DEVICE: /dev/sda1
          FS_TYPE: ext4

Using a Docker Image for CentOS

We can easily make use of the official docker images to generate a root filesystem for use when deploying with Tinkerbell.

Downloading the CentOS Image

TMPRFS=$(docker container create centos:8)
docker export $TMPRFS > centos_rootfs.tar
docker rm $TMPRFS

Optional - You can compress this filesystem archive to save on both local disk space and network bandwidth when deploying the image.

gzip ./centos_rootfs.tar

Move the raw image to a locally accessible web server. To simplify, you can place the image in the Tinkerbell sandbox webroot, which allows access to the image at the IP address of the tink-server.

mv ./centos_rootfs.tar.gz ./sandbox/deploy/state/webroot

Creating the CentOS Template

The template makes use of the actions from the artifact hub.

  • rootio - to partition our disk and make filesystems.
  • archive2disk - to write the OS image to a block device.
  • cexec - to run commands inside (chroot) our newly provisioned operating system.
  • kexec - to kexec into our newly provisioned operating system.
version: '0.1'
name: debian_bullseye_provisioning
global_timeout: 1800
tasks:
  - name: os-installation
    worker: '{{.device_1}}'
    volumes:
      - '/dev:/dev'
      - '/dev/console:/dev/console'
      - '/lib/firmware:/lib/firmware:ro'
    actions:
      actions:
        - name: disk-wipe-partition
          image: 'quay.io/tinkerbell-actions/rootio:v1.0.0'
          timeout: 90
          command:
            - partition
          environment:
            MIRROR_HOST: 192.168.1.2
        - name: format
          image: 'quay.io/tinkerbell-actions/rootio:v1.0.0'
          timeout: 90
          command:
            - format
          environment:
            MIRROR_HOST: 192.168.1.2
        - name: expand-filesystem-to-root
          image: 'quay.io/tinkerbell-actions/archive2disk:v1.0.0'
          timeout: 90
          environment:
            ARCHIVE_URL: 'http://192.168.1.1:8080/centos_rootfs.tar.gz'
            ARCHIVE_TYPE: targz
            DEST_DISK: /dev/sda3
            FS_TYPE: ext4
            DEST_PATH: /
        - name: install-grub-bootloader
          image: 'quay.io/tinkerbell-actions/cexec:v1.0.0'
          timeout: 90
          environment:
            BLOCK_DEVICE: /dev/sda3
            FS_TYPE: ext4
            CHROOT: 'y'
            CMD_LINE: grub-install --root-directory=/boot /dev/sda
        - name: kexec
          image: 'quay.io/tinkerbell-actions/kexec:v1.0.0'
          timeout: 600
          environment:
            BLOCK_DEVICE: /dev/sda3
            FS_TYPE: ext4

Using a Docker Image for Red Hat Enterprise Linux

We can easily make use of the official docker images to generate a root filesystem for use when deploying with Tinkerbell.

Download the RHEL Image

TMPRFS=$(docker container create registry.access.redhat.com/rhel7:latest)
docker export $TMPRFS > rhel_rootfs.tar
docker rm $TMPRFS

Optional - You can compress this filesystem archive to save on both local disk space and network bandwidth when deploying the image.

gzip ./rhel_rootfs.tar

Move the raw image to a locally accessible web server. To simplify, you can place the image in the Tinkerbell sandbox webroot, which allows access to the image at the IP address of the tink-server.

mv ./rhel_rootfs.tar.gz ./sandbox/deploy/state/webroot

Creating the RHEL Template

The template makes use of the actions from the artifact hub.

  • rootio - to partition our disk and make filesystems.
  • archive2disk - to write the OS image to a block device.
  • cexec - to run commands inside (chroot) our newly provisioned operating system.
  • kexec - to kexec into our newly provisioned operating system.
version: '0.1'
name: debian_bullseye_provisioning
global_timeout: 1800
tasks:
  - name: os-installation
    worker: '{{.device_1}}'
    volumes:
      - '/dev:/dev'
      - '/dev/console:/dev/console'
      - '/lib/firmware:/lib/firmware:ro'
    actions:
      actions:
        - name: disk-wipe-partition
          image: 'quay.io/tinkerbell-actions/rootio:v1.0.0'
          timeout: 90
          command:
            - partition
          environment:
            MIRROR_HOST: 192.168.1.2
        - name: format
          image: 'quay.io/tinkerbell-actions/rootio:v1.0.0'
          timeout: 90
          command:
            - format
          environment:
            MIRROR_HOST: 192.168.1.2
        - name: expand-filesystem-to-root
          image: 'quay.io/tinkerbell-actions/archive2disk:v1.0.0'
          timeout: 90
          environment:
            ARCHIVE_URL: 'http://192.168.1.1:8080/rhel_rootfs.tar.gz'
            ARCHIVE_TYPE: targz
            DEST_DISK: /dev/sda3
            FS_TYPE: ext4
            DEST_PATH: /
        - name: install-EPEL-repo
          image: 'quay.io/tinkerbell-actions/cexec:v1.0.0'
          timeout: 90
          environment:
            BLOCK_DEVICE: /dev/sda3
            FS_TYPE: ext4
            CHROOT: 'y'
            CMD_LINE: >-
              curl -O
              https://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm;
              yum install ./epel-release-latest-7.noarch.rpm; yum install grub2
        - name: install-grub-bootloader
          image: 'quay.io/tinkerbell-actions/cexec:v1.0.0'
          timeout: 90
          environment:
            BLOCK_DEVICE: /dev/sda3
            FS_TYPE: ext4
            CHROOT: 'y'
            CMD_LINE: grub-install --root-directory=/boot /dev/sda
        - name: kexec
          image: 'quay.io/tinkerbell-actions/kexec:v1.0.0'
          timeout: 600
          environment:
            BLOCK_DEVICE: /dev/sda3
            FS_TYPE: ext4