Background:

Currently, KubeVirt uses noop init containers to fetch and report the digest of containerDisk and kernel boot images. This ensures that during live migration, the VM uses the exact same files from the original container disk/kernel boot image on the target pod, even if the image tag has changed.

With Kubernetes ImageVolumeWithDigest feature, the volume status exposes the image digest. This allows KubeVirt to avoid the workaround.

Motivation:

• Simplify virt-launcher pod manifests by removing unnecessary init containers.
• Reduce CPU and memory usage for virt-launcher pods.
• Improve virt-launcher startup time by eliminating these init containers.
• Simplify maintenance of virt-controller code.
• Align live migration with native Kubernetes functionality, improving reliability.

Goals:

• Remove noop init containers used for digest fetching.
• Use the digest from the ImageVolume volume status for live migration.
• Maintain backward compatibility for existing VMs until the new approach is fully adopted.

Non Goals

• No changes to VM or VMI API fields.
• No additional functionality beyond digest reporting and live migration support.{}