Configure a MegaRAID 9560 Controller in UEFI on a Two-Socket Server
This archived guide describes the Broadcom MegaRAID 9560 UEFI configuration utility on a two-socket server. All 154 source image references are retained as English originals. They show separate configuration examples and, near the end, a different older recovery result; they are not one uninterrupted test. Identify the exact controller, array and physical disks, and preserve required data before changing configuration. No hardware, RAID, cache or recovery operation was executed while preparing this article.
1. Inspect the controller and its settings
1.1 Open the UEFI controller interface
Press Delete at the server logo, open Advanced and select the BROADCOM MegaRAID controller entry. The example identifies MegaRAID 9560-16i 8GB and utility 07.19.04.00:
Set Factory Defaults restores controller defaults and is not a routine inspection action. Silence Alarm silences the current audible indication; it does not repair the fault. The initial page shows Optimal, six drives and CacheVault No. Main Menu opens Configuration Management, Controller Management, Virtual Drive Management, Drive Management and Hardware Components.
1.2 Controller Management
| Parameter / state | Meaning |
|---|---|
| Product Name | Controller model; the example is MegaRAID 9560-16i 8GB. |
| Controller Status | Current controller state; Optimal is the source’s normal-state example. |
| Select Boot Device | Controller boot-device preference. Selecting a virtual drive here does not install an OS or prove it is bootable. |
| Package / Firmware Version | Source package 52.19.0-4233 and firmware 5.190.02-3581; use the actual installed version for compatibility and procedures. |
| Advanced Controller Management | Additional functions including events, consistency-check scheduling and Set Factory Defaults. Inspect the intended function; do not restore defaults merely to open this menu. |
| Advanced Controller Properties | Access alarm/boot policies, memory/cache, spare policy and task rates. |
| Parameter / state | Meaning |
|---|---|
| Alarm Control | Enable/disable the audible alarm policy or silence the current alarm as supported. Muting sound is distinct from resolving the underlying fault. |
| Boot Mode | Controller POST error handling: Stop on errors, Pause on errors, Ignore on errors or Safe mode on errors. This is an error-response policy, not the server’s UEFI/Legacy selection. The source shows Pause on errors. |
| ROC Temperature (C) | RAID-on-chip temperature; the captured value is 50 °C, not a recommended universal temperature limit. |
Open Advanced Controller Properties > Cache and Memory to inspect the recorded memory and NVRAM values:
The source reports Memory Size 8192 MB, NVRAM Size 128 KB, Cache Flush Interval 4 and Preserved Cache No. NVRAM stores controller configuration-related information. Preserve cache data if it is needed for an offline/missing virtual drive; Discard Preserved Cache is not a harmless refresh operation.
In Spare, the example selects Emergency Spare = Unconfigured Good and Global Hot Spare:
This permits eligible unconfigured-good drives and global spares to participate in emergency replacement/rebuild under the policy. It does not guarantee that any arbitrary disk is compatible or that rebuild succeeds. Verify capacity, drive type, actual spare policy and current array state.
In Task Rates, review the background-task resource settings:
| Parameter / state | Meaning |
|---|---|
| BGI Rate | Resource allocation for background initialization. |
| CC Rate | Resource allocation for consistency checking. |
| Patrol Read Rate | Resource allocation for patrol reads; also visible in the source screen. |
| Rebuild Rate | Resource allocation for member reconstruction during rebuild. |
| Reconstruction Rate | Resource allocation for supported reconstruction work such as expansion or RAID-level migration. It is not a command to force an offline array online. |
All five captured rates are 30. These are controller task-rate settings relative to competing work, not a promised throughput or a measured CPU-utilization result. The original prose incorrectly describes reconstruction as force-online recovery; the operations must be kept separate.
2. Inspect and identify physical drives
Open Drive Management to list recognized devices, then select the intended disk to inspect its properties and Operation choices:
| Parameter / state | Meaning |
|---|---|
| Unconfigured Good (UG) | A recognized drive eligible for a new configuration or spare assignment, subject to compatibility and existing metadata checks. |
| Unconfigured Bad (UB) | A drive marked bad or with a fault/residual configuration requiring investigation. The label is not proof that initialization will make it healthy. |
| Online | An online member of a configured virtual drive. |
| Hot Spare | A global or dedicated spare; inspect its association rather than assuming a scope from this state alone. |
| Foreign | Configuration metadata that the controller treats as foreign. It may describe a legitimate array to import or residual metadata on an intended reusable disk; decide which before clearing anything. |
Start Locate / Stop Locate control the locator indicator when supported. The online-member example offers Place Drive Offline, while the UG example offers operations including initialization, spare assignment and Make JBOD. Read the actual state and target first. Initialize Drive can modify data and is not a universal prerequisite for every recognized disk or a repair for a physically bad drive. Eligible UG drives are selected directly in the subsequent creation examples.
3. Configure virtual drives, spares and JBOD
3.1 Create RAID 5 and review the shared creation fields
Open Configuration Management > Create Virtual Drive, select RAID 5 and review Select Drives From. Unconfigured Capacity builds from eligible unconfigured disks; Free Capacity allocates unused space in an existing drive group. It does not add a new physical member or create a new independent group by itself. Select Drives opens disk selection:
Enable only the intended members and Apply Changes. Review the virtual-drive parameters before Save Configuration:
| Parameter / state | Meaning |
|---|---|
| Virtual Drive Name | Optional descriptive logical-drive name. |
| Virtual Drive Size / Unit | Intended capacity, with the displayed unit. The maximum is offered by default; check against the actual members and layout. |
| Strip Size | Per-disk strip/chunk size used by the layout; choose for the supported workload requirements. The source examples show 256 KB. |
| Read Ahead | Prefetch anticipated read data into controller cache. It does not mean pre-writing anticipated user data, as the source prose suggests. |
| Write Through | Acknowledge a write after the drive subsystem receives the transaction. This is a write-acknowledgment policy; it does not simply disable all controller caching or read-ahead. |
| Write Back | Acknowledge after the controller cache receives the write, with the supported protection/fallback behavior. Check the actual CacheVault/backup state; a normal Write Back policy can fall back to Write Through when protection is unavailable. |
| Always Write Back | Continue write-back behavior even without healthy backup protection. A sudden power loss can lose acknowledged cached writes. The source selects it in some examples; this article preserves the screenshots but does not recommend that selection. |
| I/O / Access / Drive Cache | Separate policy fields. The captured examples use Direct, Read/Write and Unchanged. These are not interchangeable with the controller write-acknowledgment policy. |
| Initialization / Emulation | The source shows Disable Background Initialization No, Default Initialization No and Emulation Default. Creation success or Optimal alone does not prove every background task or data test completed. |
The final source settings/result show RAID 5 virtual drive 239, 3.637 TB, Optimal. The confirmation warns that creation permanently deletes data on the associated drives and that Always Write Back without backup protection risks corruption after power loss. For an intended creation, review those conditions, set Confirm to Enabled, then Yes and acknowledge the result. Do not dismiss that warning just to reproduce the example.
The opening controller page reports CacheVault No, so its later Always Write Back screenshots must not be interpreted as protected-write validation. The official H3C English guide for the same controller family corroborates the write-policy distinctions; it is not evidence that the OEM system in these images has the same platform/firmware.
3.2 Create RAID 0
Use Configuration Management > Create Virtual Drive, select RAID 0, enable the intended disk(s), Apply Changes and review the shared parameters. Save only the intended configuration and complete the destructive Confirm/Yes prompt:
The recorded single-drive RAID 0 example is virtual drive 239, 1.818 TB, Optimal, with the source selecting Always Write Back. RAID 0 has no member-failure redundancy. The success page is a recorded creation result, not a new test here.
3.3 Create RAID 1
Repeat Create Virtual Drive with RAID 1 and the intended two members. Apply the selection, review capacity and supported policies, Save Configuration and complete the intended Confirm/Yes prompt:
The example result is virtual drive 239, RAID 1, 1.818 TB, Optimal. It is a separate case from the previous RAID 0/5 examples even though the virtual-drive number is reused.
3.4 Assign a global hot spare
In Drive Management select an eligible Unconfigured Good disk. Choose Operation > Assign Global Hot Spare Drive, then Go. The source shows an operation-success page. Configuration Management > View Global HotSpare Drives lists the recorded 1.818 TB hot spare:
A global spare can serve compatible eligible drive groups under the controller’s policy; it is not a promise to protect every possible layout or to make a nonredundant RAID 0 recoverable.
3.5 Assign a dedicated hot spare
From the intended eligible disk’s Operation menu, choose Assign Dedicated Hot Spare Drive and Go. Enable the particular drive group to protect and confirm the assignment:
The example selects Drive Group 0, RAID 1, 1.818 TB and ends with the physical drive shown as Hot Spare. View Associated Drive Groups is available to verify the actual association. Dedicated and Global have distinct scopes; this controller’s Hot Spare status alone does not distinguish them.
3.6 Make eligible disks JBOD in the recorded RAID mode
Open Configuration Management > Make JBOD, enable the intended UG disks and OK. The confirmation explicitly warns that existing disk data will be lost; set Confirm to Enabled and Yes only for that intended conversion:
The final list shows six source disks as JBOD / Online. This is per-drive conversion in the recorded workflow, not proof that changing the controller’s personality mode preserves existing arrays or data.
3.7 Create RAID 10 with three two-drive spans
The source explains a minimum of four disks and walks through six. Open Create Virtual Drive and choose RAID 10. For Span 0, Select Drives, enable the intended pair, Apply Changes and acknowledge OK:
Choose Add More Spans. For Span 1, select the next intended pair, Apply Changes and OK:
A four-disk configuration has two two-disk spans. For this six-disk example, Add More Spans again and select the third pair for Span 2, Apply Changes and OK:
The source uses slots 00/01, 02/03 and 04/05 for the three pairs. Review all spans and shared parameters, Save Configuration and complete the intended destructive confirmation:
The final source list reports virtual drive 239, RAID 10, 5.456 TB, Optimal. That is a controller-state record, not a host-data or failure-tolerance test. If the actual utility presents an additional warning, read and resolve it; the source’s loose instruction to skip a prompt is not permission to ignore a real warning.
3.8 Create RAID 50 with two three-drive spans
The source uses the minimum six disks as two RAID 5 spans. Open Create Virtual Drive, choose RAID 50, and select the first three intended disks for Span 0. Apply Changes:
The source inserts an additional capacity-mismatch warning from a 2021 BIOS example. It states that mixing capacities makes remaining capacity unusable. If this warning actually appears, review the intended members and capacity loss before the Confirm/Yes choice; if it does not appear, no corresponding confirmation step is needed:
Use Add More Spans to create Span 1 and select the other three intended disks. Apply Changes and acknowledge OK:
The completed source layout uses slots 00/01/02 and 03/04/05. Review the intended virtual-drive settings, Save Configuration, enable Confirm, Yes and OK. The next message says virtual-drive creation succeeded and all free configurable space was used:
The final recorded result is virtual drive 239, RAID 50, 7.275 TB, Optimal. The separate mixed-capacity warning does not establish that these six example disks have different usable sizes.
3.9 Delete an intended virtual drive
Open Main Menu > Virtual Drive Management and select the precise virtual drive. Choose Operation > Delete Virtual Drive, Go and review the data-loss confirmation. Enable Confirm and Yes only for the intended destructive deletion:
The source target is RAID 50, 7.275 TB. The subsequent recovery section returns to a separate RAID 1 case; it is not evidence that a deleted RAID 50 volume was rebuilt as RAID 1.
4. Review the single-member RAID 1 recovery case
The source starts with virtual drive 239, RAID 1, 1.818 TB, Optimal and two Online SATA members. After a reported member failure, the logical drive is Degraded:
The source reports an audible alarm at one-second intervals and a steady red drive fault indicator. These are its observations, not a universal LED/alarm contract for every backplane. Correlate the actual fault, member, enclosure/slot and replacement identity before acting.
4.1 Inspect the replacement and any residual foreign metadata
The example replacement first appears as Unconfigured Bad. In that disk’s Operation menu the source selects Make Unconfigured Good and Go:
After the state change, the source list shows Foreign plus Unconfigured Good. This is an example of residual metadata, not a rule that all UB drives should be forced good or that every replacement must become foreign. A physically faulty disk still requires fault handling/replacement. The list and detail crops contain inconsistent slot/identity examples, so do not derive a physical replacement target solely from these screenshots.
The source then opens Configuration Management > Manage Foreign Configuration > Clear Foreign Configuration:
Clear foreign metadata only when it has been verified to be unwanted residual information on the intended reusable replacement. A legitimate foreign array may instead require import to preserve its data. The menu shown clears foreign configurations; check its actual scope and preview all affected foreign sets before confirming. Clearing foreign metadata is not the same as clearing the active controller configuration, and neither is a generic recovery action. The later source inventory shows UG for the intended candidate.
4.2 Associate the intended replacement with the missing member
In the intended UG disk’s Operation menu, the source chooses Replace Missing Drive and Go, then selects the exact degraded drive group/span/row and Replace Drive. Its form names Drive Group 0, Span 0, RAID 1 and Row Index 0:
The source reports Offline for the newly associated member while the other member remains Online. This intermediate state is not recovery completion or permission to select Place Drive Online without reconstructing the missing data.
4.3 Start rebuild and verify the actual final state
On that intended Offline member, the source chooses Operation > Rebuild and Go. The next list shows Rebuilding with Rebuild 0%:
Monitor the real rebuild and the virtual/physical drive states. The source appends these final examples of Optimal and Online:
The two older final pictures do not prove that the preceding 1.818 TB rebuild reached completion. Verify that case’s actual final Optimal/Online status, completed background work and host access separately. The preserved screenshots provide no new recovery or data-integrity test.
References
- Original OEM walkthrough
- Broadcom: MegaRAID 9560-16i product and documentation
- H3C: official English 9540/9560 configuration guide
- H3C: official English 9560 guide, including foreign import/clear distinctions
The official English H3C guides concern the same controller family on a different OEM platform/version. They support terminology and checks and are not exact English copies of the source server’s BIOS. Original English source figures were reused after individual inspection, so no image reconstruction was required for this article.