creating-an-image-repository-via-skopeo-copy.adoc

Creating a repository by using Skopeo

In some cases, the podman CLI tool is unable to pull certain artifact types, for example, application/x-mlmodel, or other AI/ML artifacts. Attempting to use podman pull with this artifact type results in the following error:

Error: parsing image configuration: unsupported image-specific operation on artifact with type "application/x-mlmodel"

As an alternative, you can use skopeo copy to copy an artifact from one location to your {productname} repository.

Prerequisites

• You have installed the skopeo CLI.

• You have logged in to a source registry (in this example, \registry.redhat.io) and have a valid authentication file (~/.docker/config.json). Alternatively, you can provide credentials by using the --src-username and --src-password parameters when running a command with the skopeo CLI.

• You have logged in to your {productname} repository.

• Depending on the size of your AI/ML artifact, you might have to Prepare your registry to accept large artifacts.

Procedure

• Use the skopeo copy command on an artifact to copy the artifact to your {productname} repository. For example:

  $ sudo skopeo copy --dest-tls-verify=false --all \ (1) (2)
    --src-username <source_username> --src-password <source_password> \ (3)
    --src-authfile ~/.docker/config.json \ (4)
    --dest-username <username> --dest-password <password> \ (5)
    docker://registry.redhat.io/rhelai1/granite-8b-code-instruct:1.4-1739210683 \ (6)
    docker://quay-server.example.com/<namespace>/granite-8b-code-instruct:latest (7)

  1. Optional. --dest-tls-verify=false disables SSL/TLS verification for the destination registry.

  2. Optional. The --all flag optionally copies all image manifests, including multi-architecture images.

  3. Optional. If you are not logged into a registry, you can pass in the source registry credentials with these parameters.

  4. Optional. The path to your Docker authentication file. Typically located at ~/.docker/config.json.

  5. Your {productname} registry username and password.

  6. The source image or artifact from the Red Hat container registry. Ensure that you are logged in to the registry and that you can pull the image.

  7. The URL of your {productname} repository appended with a namespace and the name of the image.

     Example output

     Getting image source signatures
     Checking if image destination supports signatures
     Copying blob 9538fa2b8ad9 done   |
     Copying blob 491ae95f59a2 done   |
     Copying blob 01196d075d77 done   |
     Copying blob e53a4633c992 done   |
     Copying blob c266e9cfa731 done   |
     Copying blob dae0e701d9b2 done   |
     Copying blob 1e227a2c78d8 done   |
     Copying blob 94ff9338861b done   |
     Copying blob 2f2bba45146f done   |
     Copying blob d3b4df07a0ce done   |
     Copying blob f533a8dbb852 done   |
     Copying config 44136fa355 done   |
     Writing manifest to image destination
     Storing signatures

Next steps

• After you have pushed a machine learning artifact to your {productname} repository, you can View tag information by using the UI or View model card information by using the UI.