Unable to patch ESXi host in vSAN cluster due to unsupported NVMe Boot Controller as cache

Products

• VMware vSAN

Issue/Introduction

• When navigate to vSAN cluster > Updates, an error may display preventing the update below:

 

 

 

• Under vSAN cluster > Configure > vSAN Skyline Health, an error indicates that the NVMe device is uncertified.

 

 

• Ensure vSAN HCL DB is up to date.

• Check whether the device with the specific device IDs (vid, did, svid, ssid) is present in the latest all.json file.

• Checking the vSAN disk group mappings will reveal that the Boot disk controller is being used as an NVMe cache. The output will look similar to this:

# vdq -iH

DiskMapping[0]:

SSD:  t##.NVMe____xxxxxxxxx_xxxxx_Boot_Controller______xxxxxxxxxx
MD:   naa.xxxxxxxxxxxxx

 

 

 

Environment

VMware vSAN 8.x

VMware vSAN 9.x

Cause

This issue occurs because an unsupported NVMe device is configured as the vSAN cache disk. Specifically, the cache disk is running on an NVMe boot disk controller.

vSAN requires NVMe devices to be connected directly to a PCIe slot or directly to the motherboard without an intermediary controller. Because the boot controller topology is unsupported for vSAN cache, the host enters a state that prevents patching operations from completing.

Resolution

To resolve this issue and allow the ESXi host to be patched, you must replace the unsupported configuration.

• Remove the unsupported NVMe boot controller from the vSAN cache tier.
• Reinstall the ESXi host.
• Recreate the vSAN disk group using a supported NVMe disk (directly attached to the PCIe bus or motherboard) or SSD disk for the cache tier.

Additional Information

Check the KB article vSAN support of NVMe devices behind tri-mode controllers for more information.