SYNCHRONIZE CACHE command: ScsiDeviceIO: ####: Cmd(0x45de37490640) 0x35, CmdSN 0x... from world 0 to dev "naa.###############" failed H:0x0 D:0x2 P:0x0vSAN 8.x
vSphere ESXi 8.x
Cisco UCS HXAF240C-M5SX
This issue is caused by a firmware regression in version 24.00.05.00 on certain Cisco UCS storage controllers. The firmware fails to properly process or acknowledge the SYNCHRONIZE CACHE (0x35) SCSI command, leading to command timeouts and the subsequent failure of vSAN to decommission or remove the disk reference from the cluster metadata The vSAN disk appears as Absent in vSAN Disk Management.
According to the Broadcom Compatibility Guide, the supported baseline for this controller requires firmware 20.00.02.00.
To resolve this issue, the storage controller firmware must be aligned with the VMware Compatibility Guide (VCG) recommendations.
Identify the Absent Disk UUID: Run the following command on the affected host to confirm the UUID of the "Unknown" or "Absent" device: esxcli vsan storage list | grep -A 2 "Device: Unknown" Absent or unknown disk shown in vSAN disk group view after disk replacement
Verify Data Status: Confirm the disk does not contain active data components: cmmds-tool find -u [DISK_UUID] -f json. If the output returns {"entries": [ ]}, the disk is a stale reference. The vSAN disk appears as Absent in vSAN Disk Management.
Firmware Remediation: Back-rev or update the Cisco UCS storage controller firmware to the HCL-compliant version 20.00.02.00 (packaged with UCS release 4.2.2(a)). Ensure the lsi_msgpt35 driver is at version 29.00.00.00-1vmw.80320.00.02.01 or higher.
Remove the Disk via CLI: Once the firmware is updated and the host is rebooted, remove the absent disk using: esxcli vsan storage remove -u [DISK_UUID] Using UI we are not able to remove disk. Note: if the vSAN diskgroups are configured with Deduplication and Compression, the entire diskgroup will need to be recreated. See Add or Remove Disks with Deduplication and Compression Enabled