Skip to content

Solution: OS Install Fails - RAID Controller Driver Missing

Triage

  1. Drop to a shell from the Anaconda installer (Ctrl+Alt+F2 or "Troubleshooting" > "Shell"):
  2. lspci -nn | grep -i raid -- note the PCI vendor:device ID
  3. lsmod | grep mega -- check if any MegaRAID module is loaded
  4. dmesg | grep -iE 'mega|raid|sas|mpt' -- look for driver probe messages
  5. ls /dev/sd* -- confirm no block devices present
  6. cat /proc/devices -- confirm no SCSI block devices registered
  7. Identify the exact controller:
  8. PCI ID 1000:10e2 = Broadcom MegaRAID 9560-8i (Aero series)
  9. This controller requires megaraid_sas driver version 07.719 or later
  10. The RHEL 8.8 inbox kernel (4.18.0-477.x) ships megaraid_sas 07.714, which does not recognize the 9560-8i
  11. Confirm the RAID virtual disk is healthy:
  12. Boot into the controller BIOS (Ctrl+L at POST)
  13. Verify the VD is Optimal, not Foreign or Degraded

Root Cause

The Broadcom MegaRAID 9560-8i (Aero family, PCI ID 1000:10e2) is a newer hardware revision that was not supported by the megaraid_sas driver bundled with the RHEL 8.8 installer kernel. The driver present in the ISO (v07.714) does not have the PCI device ID for this controller in its device table. The controller is detected on the PCI bus but no driver binds to it, so no block devices are created.

Fix

  1. Download the Driver Update Disk (DUD):
  2. Go to Lenovo Support > SR650 V2 > Downloads > RHEL 8 Driver Update Disk
  3. Or Broadcom support: search for "MegaRAID 9560 RHEL 8 DUD"
  4. Download the DUD ISO (e.g., lnvgy_dd_megaraid_07.719.03.00-1_rhel8.iso)
  5. Load the DUD during installation:
  6. Mount the DUD ISO as a second virtual media via BMC/iLO
  7. Boot the installer ISO and at the GRUB menu, edit the boot entry:
    • Append inst.dd to the kernel command line
  8. The installer will prompt to select the DUD media; choose the mounted DUD ISO
  9. Select the megaraid_sas driver package and continue
  10. The installer will load the updated driver and rescan; the RAID virtual disk should now appear
  11. Complete the installation normally once disks are visible.
  12. Post-install verification:
  13. lsmod | grep megaraid_sas -- confirm driver is loaded
  14. modinfo megaraid_sas | grep version -- should show 07.719+
  15. lsblk -- confirm the RAID virtual disk is present
  16. Update the build process:
  17. Integrate the DUD into the kickstart/PXE boot workflow for future SR650 V2 installs
  18. Add the RPM to a local repo so kernel updates pull the correct driver version
  19. Alternatively, build a custom ISO with the driver pre-integrated using lorax

Rollback / Safety

  • The DUD only affects the installer environment; if it causes issues, simply reboot without it.
  • Do not mix driver versions: ensure the post-install kernel also has the updated driver, or the system will fail to boot after the first kernel update.
  • If the DUD is not available, switching the controller to JBOD/HBA mode and using mpt3sas is a viable alternative (but loses hardware RAID).

Common Traps

  • Assuming disks are physically bad: The RAID controller BIOS shows the VD as healthy; this is a driver issue, not hardware.
  • Using the wrong DUD version: The DUD must match the RHEL major version and kernel ABI. A RHEL 9 DUD will not work on RHEL 8.
  • Forgetting post-install driver persistence: The DUD loads the driver into the installer ramdisk. The installed system needs the driver RPM in its repo or the initramfs, or it will fail to boot after a kernel update.
  • Not updating the standard build ISO: Every new SR650 V2 will hit this same issue unless the build pipeline is updated.
  • Confusing JBOD and RAID modes: Switching to JBOD mode destroys any existing RAID virtual disk configuration.