9230 SATA M.2 RAID in Legacy BIOS: Ctrl+M Configuration and Rebuild Status
This is an archived Inspur OEM example for a Marvell 9230-based SATA M.2 RAID controller. Match the actual firmware, server boot mode and supported replacement drive; a generic 88SE9230 chip specification does not prove every feature is exposed by this OEM firmware. The source’s screenshots are already in English. Viewing, deleting, creating and rebuilding are separate operations; deleting/recreating a working array is not a routine repair for a failed mirror member.
1. RAID configuration
1.1 View the existing array
The source enters BIOS → Advanced → CSM Configuration:
For its legacy option-ROM workflow, the pictured Boot option filter is Legacy only and the boxed Network/Storage/Video/Other PCI device entries are Legacy. This describes that example; do not blindly change every boot/option-ROM policy on an installed system. Confirm compatibility with the OS boot layout and the required service procedure:
During POST, find the Marvell 88SE92xx adapter message and use Ctrl+M to enter the controller setup. The source POST screen identifies BIOS 1.0.0.1031 and a Functional RAID 1 volume with two configured Intel SSDs:
In Marvell BIOS Setup, use the arrow keys to select the intended virtual disk and inspect its RAID mode, members and status. The detail capture shows New_VD, RAID 1, 223.5GB, 64K stripe and BGA not running:
1.2 Delete an array only when its data may be discarded
The source selects the virtual disk, presses Enter and chooses Delete:
It confirms both Delete Virtual Disk and Delete MBR prompts with the keyboard Y/Yes action. The second prompt concerns the disk’s master boot record; these confirmations are destructive, not informational. Confirm the real target and retained backup before accepting them:
1.3 Create the source RAID 1 example
For two intended available members, select HBA 0: Marvell 0, press Enter and choose Configuration Wizard. The shown controller information identifies device 9230, revision A1, firmware 2.3.24.1008 and Supported Mode RAID 1; the broader chip’s feature list is not an instruction to select an unavailable mode:
Use Space to select the two verified free physical drives, then Enter to continue. The source selection shows two 223.5GB-class SATA SSDs, not two disk identifiers to reuse on another server:
Review the RAID level, capacity, stripe size, initialization choice and name. The source illustrates RAID 1, 223.5GB, 64K, Quick init Yes and Name Default; these are pictured values, not universally correct defaults:
Choose Next and confirm the creation prompt with the source’s Y/Yes action only when the selected members and intended erasure are correct. The next capture shows New_VD and its two members in the topology:
1.4 Passthrough/JBOD
The source says unconfigured drives default to passthrough when RAID 1 is not used. It supplies no separate JBOD-setting screenshot or migration procedure. Verify the actual firmware’s behavior and OS device presentation; do not delete an existing mirror simply to test passthrough.
2. A single member goes offline
The source describes a RAID 1 mirror becoming Degraded while the surviving member continues serving reads/writes, then automatically rebuilding after a supported replacement is fitted. This assumes a usable surviving member and compatible replacement; it does not guarantee access after an additional fault or prove that replacement is hot-swappable. Follow the exact powered-down/hot-swap service instructions for the M.2 assembly.
POST can display Rebuilding. Enter Ctrl+M and select the virtual disk to inspect progress. The final detail screen shows BGA running and a 24% rebuild; it is not a completed/healthy mirror confirmation:
No controller configuration, deletion, installation or replacement was performed during this translation. Verify the real volume, completed rebuild, OS boot and required data access before returning a system to service.