This guide shows how to rebuild a degraded RAID 1 array after replacing a failed disk on a Lenovo server with a ServeRAID M5210 controller. The procedure was performed on an Ubuntu 24.04 KVM hypervisor using StorCLI.
In this case, the replacement SSD appeared as UBad (Unconfigured Bad), and the rebuild did not start automatically. After changing the replacement drive to UGood and assigning it as a dedicated hot spare for the existing drive group, the controller started rebuilding the mirror.
The final checks confirmed that both SSDs were online and the RAID 1 array had returned to Optimal.
Download and Install StorCLI
The version used for this procedure was 007.2705.0000.0000. Download the archive from Broadcom:
Extract the main archive and the nested package archive:
Install the Ubuntu package:
The installation places the StorCLI executable at:
All commands below are run from a root shell. If you use a regular account, prefix StorCLI commands with sudo.
Check the RAID Controller
Display the controller overview:
Relevant output from this server:
The controller index is 0. The overview shows one drive group and one virtual drive, both not optimal at this stage. The overall controller health is NdAtn (Need Attention).
Display the configuration and topology of controller 0:
Use this output to confirm the existing RAID level and the drive group number before assigning a replacement disk. The RAID 1 array in this example belongs to Drive Group 0.
Check the Physical Disks
List every physical disk attached to controller 0:
The surviving RAID member is the SSD in 252:0. The replacement SSD is the Patriot P220 512GB in 252:2, which is currently marked UBad. The two 1.817 TB HDDs are unconfigured and are not part of this mirror.
- EID:Slt - enclosure ID and slot number
- DID - controller-assigned device ID
- DG - drive group; a dash means the disk is not assigned to a group
- Onln - Online
- UBad - Unconfigured Bad
- UGood - Unconfigured Good
- Rbld - Rebuild
The replacement drive is addressed as /c0/e252/s2: controller 0, enclosure 252, slot 2. These identifiers are specific to this server; use the values reported by your own controller.
Confirm the Replacement Disk
Before changing the drive state, inspect the replacement disk details:
Confirm its model, serial number, capacity and sector size against the disk that was installed. In this example, both SSDs report 475.905 GB and 512B sectors.
Important: The following changes target the confirmed replacement disk. Rebuilding writes mirror data onto that disk. Changing a drive to UGood does not repair a hardware fault or prove that the disk is healthy.
Change the Replacement Disk from UBad to UGood
Change the replacement drive state to Unconfigured Good:
Check the physical disks again:
The replacement SSD now appears as UGood:
At this point, the disk is available for configuration, but it has not yet joined the RAID 1 mirror.
Assign the Disk as a Dedicated Hot Spare
Assign the replacement disk as a dedicated hot spare for Drive Group 0:
The dgs=0 option dedicates the spare to Drive Group 0. In this degraded RAID 1 configuration, the controller used the replacement SSD and automatically started the rebuild after this command.
The successful hot spare command alone does not confirm that rebuilding has started. Check the rebuild status next.
Monitor the Rebuild Progress
Display the rebuild progress for the replacement disk:
Output captured during the rebuild:
Refresh the output every 10 seconds:
Press Ctrl+C to exit watch. This stops the display refresh; the RAID controller continues rebuilding. The remaining time is an estimate and can change while the server is handling other disk activity.
Verify That Both RAID Members Are Online
After the rebuild finishes, check all physical disks again:
Both SSDs should report Onln and belong to Drive Group 0. The final SSD check on this server was:
The replacement SSD is now an active mirror member. It is no longer an unused standby spare.
Verify the RAID 1 Array Is Optimal
Display the controller topology:
The final topology contained the following RAID 1 and member-drive entries:
This is an excerpt of the relevant topology columns. Optl means Optimal. Together with both disks reporting Onln, it confirms that the controller has restored the RAID 1 mirror.
Check the virtual drive state directly:
The RAID 1 virtual drive should also report Optl.
Confirm the Status in Lenovo IMM
The result can also be checked in the Lenovo Integrated Management Module II web interface:
- Open Server Management and select Local Storage.
- Select Storage RAID Configuration.
- Click Refresh and inspect the volume under the ServeRAID M5210 controller.
The supplied IMM screenshot showed Storage Pool 0 as RAID 1 and the GenericR1_0 volume as Optimal, with a capacity of 475.905 GB. This agreed with the final StorCLI output.
Why the Rebuild Did Not Start Immediately
Installing the replacement disk did not make it an active RAID member. The initial output showed it as UBad. After the state was changed to UGood, it was still unassigned, with a dash in the DG column.
In this case, the rebuild began only after the replacement disk was assigned as a dedicated hot spare for the degraded drive group. The supplied output does not establish why the new disk was initially marked UBad. This procedure documents the recovery that worked on this controller.
Quick Command Reference
For this specific controller and replacement slot, the procedure was:
References
Broadcom StorCLI Reference Manual - drive state, hot spare and rebuild commands.