Replacing a Drive
Replacing a failed or aging NVMe drive is a guided, step-by-step flow in the Dashboard. The node stays online throughout and volumes keep serving.
When Replace Is Available​
Replace becomes available on a drive once it reports a fault. A healthy drive shows the button disabled with a tooltip explaining why. A drive that is not part of the storage cluster cannot be replaced from here.
You can reach the flow two ways:
- Hardware → Drives or Hardware → Cluster Map — open the drive's detail dialog and click Replace
- From a critical hardware alert — selecting the alert takes you straight to the affected drive
A drive swapped out physically without using this flow is still detected automatically: storage is reconfigured to use the new drive and restarted on its own. The guided flow is the safer path because it evicts the old drive and waits for rebuild first.
Choose a Method​
| Method | What happens |
|---|---|
| Use an available spare in this node | Data moves to a spare drive already present in the node. Fully automatic. Unavailable if the node has no spare. |
| Physically swap the drive | Runs in stages: you are prompted to pull the faulty drive, then to insert a new one. |
Tick the acknowledgement — the drive will be evicted and replaced, and volumes stay online during the process — then click Start Replacement.
The Physical Swap, Stage by Stage​
- Preparing — the faulty drive is evicted and the system waits for rebuild
- Ready — remove the drive — the dialog names the node, serial, and bay. Pull that drive, then click Continue
- Insert the new drive — seat the replacement in the same bay, then click Continue. The drive is detected live, with no engine restart
- Complete — the new drive is serving in the cluster
While a replacement is running, a banner appears with a View button so you can return to the dialog and continue the operation from any page.
Pull only the drive named in the dialog — it identifies the drive by node, serial number, and bay.
If It Fails​
The dialog reports the failure with its reason. Re-check the drive is properly seated, then start the flow again. If it keeps failing, collect a log bundle and open a ticket from System → Support.