Buildah Build is a GitHub Action for building Docker and Kubernetes-compatible images quickly and easily.

Buildah only works on Linux. GitHub's Ubuntu Environments (ubuntu-18.04 and newer) come with buildah installed. If you are not using these environments, or if you want to use a different version, you must first install buildah.

After building your image, use push-to-registry to push the image and make it pullable.

* The containerfiles input was previously dockerfiles. Refer to this issue.

The image and tags inputs can be provided in one of two forms.

At least one tag must always be provided in tags. Multiple tags are separated by whitespace.

Option 1: Provide both image and tags inputs. The image will be built, and then tagged in the form ${image}:${tag} for each tag.

For example:

image: quay.io/my-namespace/my-image
tags: v1 v1.0.0

will create the image and apply two tags: quay.io/my-namespace/my-image:v1 and quay.io/my-namespace/my-image:v1.0.0.

Option 2: Provide only the tags input, including the image name in each tag. The image will be built, and then tagged with each tag. In this case, the image input is ignored.

For example:

# 'image' input is not set
tags: quay.io/my-namespace/my-image:v1 quay.io/my-namespace/my-image:v1.0.0

will also apply two tags: quay.io/my-namespace/my-image:v1 and quay.io/my-namespace/my-image:v1.0.0.

If the tags input does not have image names in the ${name}:${tag} form, then the image input must be set.

image: The name of the image as it was input.
tags: A space-separated list of the tags that were applied to the new image.
image-with-tag: The name of the image, tagged with the first tag.

For example:

image: "spring-image"
tags: "latest ${{ github.sha }}"
image-with-tag: "spring-image:latest"

You can configure the buildah action to build your image using one or more Containerfiles, or none at all.

If you have been building your images with an existing Containerfile, buildah can reuse your Containerfile.

In this case the inputs needed are image and containerfiles. tag is also recommended. If your Containerfile requires ARGs, these can be passed using build-arg.

name: Build Image using Containerfile
on: [push]

jobs:
  build:
    name: Build image
    runs-on: ubuntu-latest

    steps:
    - uses: actions/checkout@v2

    - name: Buildah Action
      uses: redhat-actions/buildah-build@v2
      with:
        image: my-new-image
        tags: v1 ${{ github.sha }}
        containerfiles: |
          ./Containerfile
        build-args: |
          some_arg=some_value

Building without a Containerfile requires additional inputs, that would normally be specified in the Containerfile.

Do not set containerfiles if you are doing a build from scratch. Otherwise those Containerfiles will be used, and the inputs below will be ignored.

• An output image name and usually a tag.
• base-image
  • In a Containerfile, this would be the FROM directive.
• content to copy into the new image
  • In a Containerfile, this would be COPY directives.
• entrypoint so the container knows what command to run.
  • In a Containerfile, this would be the ENTRYPOINT.
• All other optional configuration inputs, such as port, envs, and workdir.

Example of building a Spring Boot Java app image:

name: Build Image
on: [push]

jobs:
  build-image:
    name: Build image without Containerfile
    runs-on: ubuntu-latest

    steps:
    - uses: actions/checkout@v2

    - run: mvn package

    - name: Build Image
      uses: redhat-actions/buildah-build@v2
      with:
        base-image: docker.io/fabric8/java-alpine-openjdk11-jre
        image: my-new-image
        tags: v1
        content: |
          target/spring-petclinic-2.3.0.BUILD-SNAPSHOT.jar
        entrypoint: java -jar spring-petclinic-2.3.0.BUILD-SNAPSHOT.jar
        port: 8080

Refer to the multi-arch example.

Cross-architecture builds from containerfiles containing RUN instructions require qemu-user-static emulation registered in the Linux kernel.

For example, run sudo apt install qemu-user-static on Debian hosts, or sudo dnf install qemu-user-static on Fedora.

You can run a containerized version of the registration if the package does not exist for your distribution:

sudo podman run --rm --privileged docker.io/tonistiigi/binfmt --install all

This registration remains active until the host reboots.

The arch and platform arguments override the Architecture and Platform labels in the output image, respectively. They do not actually affect the architectures and platforms the output image will run on. The image must still be built for the required architecture or platform.

There is a simple example in this issue.

Use the buildah manifest command to bundle images into an image list, so multiple image can be referenced by the same repository tag.

There are examples and explanations of the manifest command in this issue.

This action does not support the manifest command at this time, but there is an issue open.

Refer to the docker/metadata-action example.

If your build references a private image, run podman-login in a step before this action so you can pull the image. For example:

- name: Log in to Red Hat Registry
  uses: redhat-actions/podman-login@v1
  with:
    registry: registry.redhat.io
    username: ${{ secrets.REGISTRY_REDHAT_IO_USER }}
    password: ${{ secrets.REGISTRY_REDHAT_IO_PASSWORD }}