Chapter 7. Creating bootc images from scratch


With bootc images from scratch, you can have control over the underlying image content, and tailor your system environment to your requirements.

You can use the bootc-base-imagectl command to create a bootc image from scratch by using an existing bootc base image as a build environment, providing greater control over the content included in the build process. This process takes the user RPMs as input, so you need to rebuild the image if the RPMs change.

The custom base derives from the base container, and does not automatically consume changes to the default base image unless you make them part of a container pipeline.

You can use the bootc-base-imagectl rechunk subcommand on any bootc container image.

If you want to perform kernel management, you do not need to create a bootc image from scratch. See Managing kernel arguments in bootc systems.

7.1. Using pinned content to build images

To ensure the base image version contains a set of packages at exactly specific versions, for example, defined by a lockfile, or an rpm-md or yum repository, you can use several tools to manage snapshots of rpm-md or yum repository repositories.

With the bootc image from scratch feature, you can configure and override package information in source RPM repositories, while referencing mirrored, pinned, or snapshotted repository content. Consequently, you gain control over package versions and their dependencies.

For example, you might want to gain control over package versions and their dependencies in the following situations:

  • You need to use a specific package version because of strict certification and compliance requirements.
  • You need to use specific software versions to support critical dependencies.

Prerequisites

  • A standard bootc base image.

Procedure

  • The following example creates a bootc image from scratch with pinned content:

    # Begin with a standard bootc base image that serves as a "builder" for our custom image.
    FROM registry.redhat.io/rhel10/rhel-bootc:latest
    # Configure and override source RPM repositories, if necessary. The following step is required when referencing specific content views or target mirrored/snapshotted/pinned versions of content.
    RUN rm -vf /etc/yum.repos.d
    COPY mypinnedcontent.repo /etc/yum.repos
    # Add additional repositories to apply customizations to the image. However, referencing a custom manifest in this step is not currently supported without forking the code.
    # Build the root file system by using the specified repositories and non-RPM content from the "builder" base image.
    # If no repositories are defined, the default build will be used. You can modify the scope of packages in the base image by changing the manifest between the "standard" and "minimal" sets.
    RUN /usr/libexec/bootc-base-imagectl build-rootfs --manifest=standard /target-rootfs
    # Create a new, empty image from scratch.
    FROM scratch
    # Copy the root file system built in the previous step into this image.
    COPY --from=builder /target-rootfs/ /
    # Apply customizations to the image. This syntax uses "heredocs" https://www.docker.com/blog/introduction-to-heredocs-in-dockerfiles/ to pass multi-line arguments in a more readable format.
    RUN <<EORUN
    # Set pipefail to display failures within the heredoc and avoid false-positive successful builds.
    set -xeuo pipefail
    # Install necessary packages, run scripts, etc.
    dnf -y install NetworkManager emacs
    # Remove leftover build artifacts from installing packages in the final built image.
    dnf clean all
    rm /var/{log,cache,lib}/* -rf
    EORUN
    # Define required labels for this bootc image to be recognized as such.
    LABEL containers.bootc 1
    LABEL ostree.bootable 1
    # Optional labels that only apply when running this image as a container. These keep the default entry point running under systemd.
    STOPSIGNAL SIGRTMIN+3
    CMD ["/sbin/init"]
    # Run the bootc linter to avoid encountering certain bugs and maintain content quality. Place this command last in your Containerfile.
    RUN bootc container lint
    Copy to Clipboard

Verification

  1. Save and build your image.

    $ podman build -t quay.io/<namespace>/<image>:<tag> . --cap-add=all --security-opt=label=type:container_runtime_t --device /dev/fuse
    Copy to Clipboard
  2. Build <_image_> image by using the Containerfile in the current directory:

    $ podman build -t quay.io/<namespace>/<image>:<tag> .
    Copy to Clipboard

7.2. Building a base image up from minimal

Previously, you could build just a standard image by using image mode for RHEL. The standard image is roughly a headless server-oriented installation, although you can use it for desktops as well, and includes many opinionated packages for networking, CLI tool, among others.

You now have the option to generate from the standard image a new minimal image which only starts from bootc, kernel, and dnf. This image can then be extended further in a multi-stage build. At the current time the minimal image is not shipped pre-built in the registry.

The base images include the /usr/libexec/bootc-base-imagectl tool that enables you to generate a custom base image. By using the tool, you can build a root file system that is based on the RPM packages that you selected in the base image.

Prerequisites

  • A standard bootc base image.

Procedure

  • The following example creates a custom minimal base image:

    # Begin with a standard bootc base image that is reused as a "builder" for the custom image.
    FROM registry.redhat.io/rhel10/rhel-bootc:latest
    # Configure and override source RPM repositories, if necessary. This step is not required when building up from minimal unless referencing specific content views or target mirrored/snapshotted/pinned versions of content.
    # Add additional repositories to apply customizations to the image. However, referencing a custom manifest in this step is not currently supported without forking the code.
    # Build the root file system by using the specified repositories and non-RPM content from the "builder" base image.
    # If no repositories are defined, the default build will be used. You can modify the scope of packages in the base image by changing the manifest between the "standard" and "minimal" sets.
    RUN /usr/libexec/bootc-base-imagectl build-rootfs --manifest=minimal /target-rootfs
    # Create a new, empty image from scratch.
    FROM scratch
    # Copy the root file system built in the previous step into this image.
    COPY --from=builder /target-rootfs/ /
    # Apply customizations to the image. This syntax uses "heredocs" https://www.docker.com/blog/introduction-to-heredocs-in-dockerfiles/ to pass multi-line arguments in a more readable format.
    RUN <<EORUN
    # Set pipefail to display failures within the heredoc and avoid false-positive successful builds.
    set -xeuo pipefail
    # Install required packages for our custom bootc image.
    # Note that using a minimal manifest means we need to add critical components specific to our use case and environment.
    dnf -y install NetworkManager openssh-server
    # Remove package caches
    dnf clean all
    # Clean up all logs and caches
    rm /var/{log,cache,lib}/* -rf
    # Run the bootc linter to perform build-time verification. Keep this as the last command in your build instructions.
    bootc container lint
    # Close the shell command.
    EORUN
    # Define required labels for this bootc image to be recognized as such.
    LABEL containers.bootc 1
    LABEL ostree.bootable 1
    # Optional labels that only apply when running this image as a container. These keep the default entry point running under systemd.
    STOPSIGNAL SIGRTMIN+3
    CMD ["/sbin/init"]
    Copy to Clipboard

7.3. Building required privileges

Generating a root filesystem from scratch requires the inner build process to use some nested containerization (such as mount namespacing) that are not enabled by default by many container build tools.

Prerequisites

  • In this example using podman, the container-tools meta-package is installed.

Procedure

  • Generate a new root file system, providing these arguments at a minimum to podman build:

    --cap-add=all --security-opt=label=type:container_runtime_t --device /dev/fuse
    Copy to Clipboard

7.4. Generating your bootc images from scratch

Create bootc images from scratch from a custom RHEL bootc default base container image to get a small root content set.

Prerequisites

  • The container-tools metapackage is installed.

Procedure

  • Create a Containerfile. The following is an example:

    # The following example reuses the default base image as a "builder" image. Optionally,  you can use the commented instructions to configure or override the RPM repositories in /etc/yum.repos.d to, for example, refer to pinned versions
    FROM registry.redhat.io/rhel10/rhel-bootc:latest
    # RUN rm -rf /etc/yum.repos.d/*
    # COPY mycustom.repo /etc/yum.repos.d
    RUN /usr/libexec/bootc-base-imagectl build-rootfs --manifest=minimal /target-rootfs
    # Create a new, empty image from scratch.
    FROM scratch
    # Copy the root file system built in the previous step into this image.
    COPY --from=builder /target-rootfs/ /
    # You can make arbitrary changes such as copying the systemd units and other tweaks from the baseconfig container image. This example uses the heredocs syntax, to improve and make it easy to add complex instructions, and install critical components
    RUN <<EORUN
    set -xeuo pipefail
    # Install networking support and SSH which are not in minimal
    dnf -y install NetworkManager openssh-server
    dnf clean all
    rm /var/{log,cache,lib}/* -rf
    bootc container lint
    EORUN
    # This label is required
    LABEL containers.bootc 1
    LABEL ostree.bootable 1
    # These labels are optional but useful if you want to keep the default of running under systemd when run as a container image.
    STOPSIGNAL SIGRTMIN+3
    CMD ["/sbin/init"]
    Copy to Clipboard

Next steps

  • After creating your Containerfile, you get an image with a single tar file large layer. Every change, such as pushing to the registry, pulling for clients, results in copying the single large tar file, and increases the container image size. You can optimize the container image that you created for a smaller version.

7.5. Optimizing container images to a smaller version

You can use the bootc-base-imagectl rechunk subcommand to optimize an input container image into a new image with the same filesystem tree, but split into content-addressed reproducible layers, with precomputed SELinux labeling.

This provides better network efficiency (for both pushes and pulls) since layers that did not change across an image build can be reused without causing a transfer.

The rechunk operation works on an image produced by the default mode of creating new images FROM <rhel-bootc>, but is especially useful in combination with the scratch builds that output only a single large tar layer. Without rechunk every change to the input, for example a kernel update, will result in a new layer including the entire contents of the bootc image. This new layer must then be pushed, stored by registries, and pulled by clients.

The bootc-base-imagectl is shipped as part of the bootc images and is intended to be run inside a container, but requires mapping the host containers-storage into the container to execute.

Prerequisites

  • You have a previously-built base image.

Procedure

  • Run the following command to rechunk your base image.

    $ sudo podman run --rm --privileged -v /var/lib/containers:/var/lib/containers \
          registry.redhat.io/rhel10/rhel-bootc:latest \
          /usr/libexec/bootc-base-imagectl rechunk \
              quay.io/exampleos/rhel-bootc:single \
              quay.io/exampleos/rhel-bootc:chunked
    Copy to Clipboard
Back to top
Red Hat logoGithubredditYoutubeTwitter

Learn

Try, buy, & sell

Communities

About Red Hat Documentation

We help Red Hat users innovate and achieve their goals with our products and services with content they can trust. Explore our recent updates.

Making open source more inclusive

Red Hat is committed to replacing problematic language in our code, documentation, and web properties. For more details, see the Red Hat Blog.

About Red Hat

We deliver hardened solutions that make it easier for enterprises to work across platforms and environments, from the core datacenter to the network edge.

Theme

© 2025 Red Hat