LSI 9361-8i RAID on M6 BMC: JBOD, RAID0/RAID1/RAID5 and Deletion
This archived M6 BMC procedure uses separate examples for viewing storage, configuring JBOD, creating RAID0/RAID1/RAID5 and deleting a logical disk. Match the real controller, firmware, disk identities and retained data before applying a configuration. The RAID examples reuse some disks and logical-disk numbers; they are not a sequence to apply cumulatively to an existing array.
The English illustrations reproduce the original screenshot states. An arrow pointing to a menu option does not mean that option has already been selected. A recorded OK/Optimal state does not establish that initialization, consistency checks, OS access or backup verification have completed. The official BMC manual also notes that storage information is invalid while the server is powered off or starting, and disk identification can be delayed during rebuilding.
1. View the RAID controller and physical drives
Log in to BMC and open Storage → View. Select PCIe2_RAID to list the controller details. The example identifies AVAGO MegaRAID SAS 9361-8i, firmware 4.680.00-8573 and package 24.21.0-0157. JBOD Enable already reads Enable. The fields Drive Count 5 and HD Count 4 are distinct; do not reinterpret them as five pictured physical drives:
Select a Disk_* node for physical-drive details. The original English capture shows Disk_1:4, Device ID17, enclosure252, slot 4, SAS, 3726GB:
2. RAID configuration
2.1 Configure JBOD
The source first selects an intended physical drive, chooses UNCONFIGURED GOOD and saves its firmware state. This is a configuration step for an appropriate available drive, not a recovery instruction to force a failing or active array member to good. The first capture selects Disk_1:6:
Under Configure → Controller, enable JBOD Mode and Save. The pictured controller page has SMART Error Copyback Disabled and JBOD Mode Enabled:
Return to Physical Disk, verify the actual target, choose JBOD and Save. This next source capture selects Disk_1:5, not Disk_1:6. The differing disk numbers are examples; these screenshots do not prove a single state transition on the same drive:
2.2 Create the RAID0 example
Confirm that every intended available member is UNCONFIGURED GOOD. The source’s preliminary physical-drive screen uses Disk_1:6:
Open Configure → Logical Disk and choose RAID0. The menu also lists RAID1, RAID5, RAID6, RAID10, RAID50, RAID60; availability depends on the actual supported configuration:
Choose a write policy appropriate to the controller and its verified cache protection. The source explains Always Write Back as enabling write-back even without a battery; that can expose unwritten cached data to power loss and is not a universal performance recommendation. This screenshot still selects Write Through while its arrow points to Always Write Back in the open menu:
Selected Drive Capacity is a percentage of the capacity available for the intended virtual disk:100%uses the available amount, 10%allocates a smaller portion. Do not treat this as usable capacity equal to the raw sum of members for every RAID level. The source’s one-drive RAID0 form uses Disk_1:4 only, Span Count 1, 64K, Read Write, Read Ahead, Write Through, Direct IO, Unchanged, No Init and100%, then Save:
Return to View → Logical Disk0. Its original English result shows RAID0, Optimal, 3725GB, Write Through, NoInit and Consistency Yes. This is the captured configuration state, not a data-access or backup test:
2.3 Create the RAID1 example
RAID1 requires at least two members here. The separate source form selects Disk_1:4 and Disk_1:5, RAID1, Span Count 1, 64K, Write Through, No Init and100%. Confirm all other policies and the real free members before saving:
In View → Logical Disk0, the source shows RAID1, Optimal, 3725GB, Write Through, NoInit and Consistency No. Optimal alone does not prove mirror initialization or a consistency check has completed:
2.4 Create the RAID5 example
RAID5 requires at least three members here. The source selects Disk_1:5, Disk_1:6, Disk_1:7 with RAID5, 64K and Always Write Back. The Initialization Type menu is open: No Init remains the selected field/blue option, while an arrow points to Quick Init. An arrow is not confirmation that Quick Init was applied:
The final View → Logical Disk1 capture confirms RAID5, Optimal, 7451GB, Always Write Back, NoInit and Consistency No. It does not establish completed initialization or protected write caching:
3. Remove JBOD presentation or delete an array
For the source’s JBOD-removal branch, select the verified physical drive and change its firmware state to UNCONFIGURED GOOD, then Save. The pictured example is Disk_1:6. A state change is not a documented data-preserving migration procedure:
For array deletion, select the intended logical disk under Other Actions, choose Delete Logical Disk and Save only after verifying the real target and that its data may be discarded. The source opens the selector on Logical Disk0, with Logical Disk1 as the alternate. Its five action radio circles are still unchecked and the arrow only identifies Delete Logical Disk; it is not a completed-deletion capture:
4. Hot-spare visibility in this BMC example
The source reports that this BMC interface/firmware cannot configure a hot spare and does not label an existing hot-spare drive on the storage physical-disk page. This is the limitation of the described interface, not proof that the controller lacks hot-spare support. Absence of a BMC label does not establish that a drive is not a spare; verify the actual configuration with the supported controller firmware or CLI tools.