Migration

What's included

About this module

Migration lives at /migration and covers everything around the copy rather than the copy itself. A read-only Python collector reads your vCenter on your own network — the credential is read from an environment variable, never sent to Turtini, never written to the output, never printed — and the server assesses every machine against what KubeVirt actually runs. Raw Device Mappings, multi-writer VMDKs, PCI passthrough, Fault Tolerance, end-of-life guests, outstanding snapshots, missing tools: each comes back as a finding with the remediation, and each machine is marked ready, needs work, or blocked.

Then it plans. You give it the maintenance windows you actually have and a measured copy rate — a guessed rate is refused, because every downtime figure derives from that one number — and it fits machines into waves largest first, so anything that will not fit surfaces while there is still time to do something about it. Blocked machines are never scheduled, and the list of what did not fit comes back with the reason for each.

Finally it hands off. An approved wave becomes the Forklift Plan, NetworkMap and StorageMap YAML your cluster already understands, which you apply yourself with kubectl, with your own credentials, in your own change window. Turtini never holds a kubeconfig, never reaches your cluster, and never copies a disk. Every migration it plans is cold: without VDDK there is no changed-block tracking, so a machine is down for the whole copy — the module is built around that fact rather than around hiding it.

What you can do with it

• Inventory a vSphere estate with a read-only collector that never sends your vCenter credential

• See what blocks each machine, and what to do about it, before committing to a date

• Plan waves against your real maintenance windows and a copy rate you measured

• Emit Forklift manifests for one approved wave and apply them yourself