Migrate the operating system to an SSD

⚠Known limitation

This guided migration has not been run end-to-end on real hardware for the current release. Testing needs a node that boots from its microSD card while an NVMe disk is already attached, and no such machine was available during this release's validation. The run is deferred until one is. Treat the steps below as accurate to what the wizard is built to do, not as a flow that's been watched completing on a physical node. The wizard's own safeguards — the eligibility check, the untouched microSD until you explicitly retire it, and Abort — are your safety net if something doesn't go as described.

What this gets you

A node’s operating system moved from its memory card onto a faster, more reliable NVMe SSD, with the rest of that SSD turned into a cluster storage pool in the same guided flow.

Before you start

• A Raspberry Pi 5 node with an NVMe SSD attached via the M.2 HAT+.
• The node’s operating system still on its microSD card (this flow is for moving it, not for a node already on NVMe).
• The SSD has nothing on it you need. Migrating erases it.

The steps

1. Open Storage → Migrate OS to SSD (or the launcher in the node’s Maintenance screen, under its Raspberry Pi section). Buddy checks the node qualifies: the right hardware, an SSD big enough for both the OS and a pool, firmware that can boot from NVMe, and the OS still on the microSD. If something’s missing, it tells you exactly what — e.g. “update the firmware first” — and the action stays blocked until you fix it.

   

   

   The migration wizard's eligibility check step

2. Review the planned layout — firmware, a sized root partition, and a reserved partition for a future pool — and the pool capacity the cluster will gain. Confirm the data-loss warning.

   Migrating erases the SSD. The microSD is left untouched throughout: it stays your fallback until you explicitly retire it in the last step.

3. Watch the live migration log. Buddy copies the OS to the SSD, points the node’s boot order at the NVMe disk (with the microSD kept as a fallback), and reboots. The node comes back booted from the NVMe, which the panel verifies.

4. Once verified, the reserved partition on the SSD appears as a preparable device. Turn it into a normal cluster pool exactly as in Prepare a disk and create a storage.

5. Power the node off, physically remove the microSD, power it back on, then choose “I removed the microSD — verify & finish.” Buddy re-checks that the node boots from the NVMe alone before finishing.

   Until you complete this step, the microSD remains a complete rollback: if anything has gone wrong, the node still boots from it. If a migration fails, or you abort before this step, the node simply keeps booting from the microSD; nothing on the card is touched until you retire it.

How you know it worked

The node’s Maintenance screen shows it booted from the NVMe, and — once you’ve retired the microSD — the card is confirmed removed and the migration marked finished.

What can go wrong

• “Migrate OS to SSD” is missing or blocked
• The node didn’t come back booted from the NVMe
• Boot order didn’t change / it still boots from the microSD
• “Retire microSD” is refused
• I want to undo a migration

See these symptoms in Troubleshooting.