Goal:

The goal of this Epic is to implement the direct-access data path that allows the Red Hat OpenStack VMware Migration Toolkit to read VMDK files directly from a mounted NFS directory. By enabling the conversion host to bypass network-based VDDK streaming, we aim to eliminate "data hairpinning" (moving data in and out of the host unnecessarily) and reduce the migration storage footprint by removing the need for intermediate disk copies.

Acceptance Criteria:

• Direct NBDKit Configuration: The migration engine must be capable of dynamically switching NBDKit to the file plugin when the import_workloads_local_disk_path variable is detected.

• Path Resolution Logic: The toolkit must successfully map VMware inventory metadata to physical file paths on the NFS mount using the convention: {}{{ import_workloads_local_disk_path }}/{{ vm_folder_name }}/{{ vmdk_filename }}{}.

• Zero-Copy Verification: Migration logs must confirm that disk data is being pulled directly from the local mount point, with a measurable reduction in traffic on the ESXi management network (Port 902).

• Storage Efficiency: The migration process must demonstrate that no temporary duplicate copy of the VMDK is created on the conversion host’s primary filesystem during the transfer to OpenStack Cinder.

• Error Handling: If the specified local path is unavailable or the VMDK is not found in the expected directory, the toolkit must provide a specific error message (e.g., "NFS Mount Point Unreachable" or "Disk File Missing") rather than a generic migration failure.

Open questions:

<-- Capture any open questions and resolutions relating to the goal/acceptance criteria - remove these notes before saving -->

• Any additional details, questions or decisions that need to be made/addressed
• …