Troubleshooting

This section addresses common issues that may occur during Intel® VROC setup, RAID configuration, or runtime operation on Linux*. Each issue includes likely causes and corresponding solutions to help you quickly restore system functionality.

Issue: RAID volumes do not appear after reboot

• Possible Cause 1: The configuration file /etc/mdadm.conf is missing or incomplete.

  • Solution: Rebuild the configuration file and rescan all arrays:

mdadm --examine --scan > /etc/mdadm.conf
mdadm --assemble --scan

• Possible Cause 2: The mdadm service is not enabled at startup.

  • Solution: Enable the service to ensure automatic RAID assembly on boot:

# systemctl enable mdmonitor.service
# systemctl start mdmonitor.service

Issue: RAID creation fails or member disks are not detected

• Possible Cause 1: Intel VMD driver is not loaded or initialized.

  • Solution: Verify the VMD module is active and visible to the system:

# lspci | grep -i vmd
# ls /sys/class/vmd/

If missing, load the driver manually:

# modprobe vmd

• Possible Cause 2: NVMe devices are not properly enumerated under Intel® VROC control.

  • Solution: Confirm that BIOS VMD configuration is enabled for the corresponding PCIe ports.

Pro Tip: Reboot into BIOS → Advanced → Intel VMD Technology → Enable all NVMe ports under VROC control.

Issue: RAID volume rebuild speed is extremely slow

• Possible Cause: The default synchronization speed is limited to 200MB/s.

  • Solution: Increase the resync speed limit for NVMe-based arrays:

# cat /sys/block/md0/md/sync_speed_max
# echo 5000000 > /sys/block/md0/md/sync_speed_max

Issue: LED indicators are not functioning or show incorrect states

• Possible Cause: The ledmon daemon is not running.

  • Solution: Download the correct storage driver from the server manufacturer's support website. During the OS installation (this is common in Windows Server), look for an option to Load driver. Provide the driver files via a separate USB drive or another virtual media slot to allow the installer to detect the drives. For Linux, you may need a more recent kernel version or a specific distribution that includes the required driver.

Issue: Installation fails midway with an error

• Possible Cause 1: Corrupted installation media.

  • Solution: Restart and verify the LED monitoring service:

# systemctl restart ledmon.service
# systemctl status ledmon.service

• Possible Cause 2: The server’s backplane is not supported for LED control.

  • Solution: Intel® VROC LED Management supports only NVMe drives connected via compatible backplanes (SGPIO/SES-2). Drives attached directly to the motherboard (M.2) or through PCIe adapters are not LED-controllable.

Issue: RAID migration or Online Capacity Expansion (OCE) fails

• Possible Cause1: Insufficient spare drives or available capacity.

  • Solution: Verify that the target RAID level’s minimum drive count is satisfied and that sufficient unallocated space exists on each member disk.

• Possible Cause2: Bitmap or PPL feature is still enabled.

  • Solution: Disable bitmap or PPL before executing OCE or RAID migration:

# mdadm --grow --consistency-policy=resync /dev/md/volume

Issue: mdadm monitoring alerts are not triggered

• Possible Cause: Missing or misconfigured email address in /etc/mdadm.conf.

  • Solution: Add the following line to enable email-based alerts and restart mdmonitor:

# systemctl restart mdmonitor.service

Issue: mdadm reports “device busy” or cannot stop an array

• Possible Cause: The RAID volume is still mounted or in use by another process.

  • Solution: Unmount the filesystem and reattempt stopping the array:

# umount /mnt/data
# mdadm --stop /dev/md/volume

If needed, check open file handles:

# lsof | grep /dev/md/volume

Issue: RAID volume appears degraded after unexpected power loss

• Possible Cause: Incomplete write operation or missing parity sync.

  • Solution: Check the array status and trigger manual resync if necessary:

# cat /proc/mdstat
# mdadm --detail /dev/md/volume
# mdadm --assemble --scan

Issue: RAID volume not found in Linux installer

• Possible Cause: The installation kernel lacks the Intel VROC driver.

  • Solution:

    • Use a Linux distribution with Intel VROC inbox support.

    • If not available, load Intel’s out-of-box driver package during OS installation.

    • Reboot and confirm that the array is recognized via lsblk or mdadm --detail.

Final Tip: Always maintain an up-to-date backup of /etc/mdadm.conf and RAID metadata before performing reshape, resize, or migration operations.