Migration — what it does, and what it deliberately does not

Migration plans a move from vSphere to OpenShift Virtualization and hands off the manifests. It does not move the data.

That boundary is the design, not a gap:

Copying a disk is solved. virt-v2v and the converter you are already licensed to run do it, and they do it better than a reimplementation would. What has no good home is everything around the copy — which machines can move at all, what has to change first, how long each one is down, which of them fit in Saturday's window, who agreed to that, and what evidence exists afterwards. That is what this module is.

So Turtini never holds a kubeconfig, never reaches your cluster, and never touches a VM. The Handoff tab produces a YAML file. You apply it, with your own credentials, in your own change window, watching your own logs.

Every migration here is COLD:

Warm migration needs Changed Block Tracking, CBT needs VDDK, and VDDK is the proprietary VMware component that is often the exact thing you cannot get. Without it there is no pre-copy, so a machine is down for the entire copy and downtime is simply copy time.

The module is built around that fact rather than around hiding it. Every downtime figure you see is a full-copy figure, `warm: false` is written into every manifest it emits, and it is not a setting anyone can flip — a plan that claimed warm would fail at cutover, after the window opened and the machine was already down.

The four tabs are the four steps, in order:

Collect gets your estate in. Assessment says what each machine needs. Plan fits them into the windows you actually have. Handoff writes the manifests for one approved wave. You cannot usefully do them out of order, which is why the tabs are a sequence rather than alphabetical.