Troubleshoot S2D health - loss of comms to one of NVMe cache disks

I have an NVMe cache disk reported as retired due to 'lost communication' but the disk is showing as healthy via the vendor's tools. Is this purely a hardware issue or is there some storage spaces direct troubleshooting that can help resolve the issue?

3 node cluster with two NVMe cache disks each

PS C:\Windows\system32> Get-HealthFault

Severity: Critical

Reason : Windows cannot communicate with the drive. It may be loose or disconnected, its connector may have failed, or the drive itself may have failed. Windows automatically retires drives after 15

minutes of lost communication.

Recommendation : Check that the drive is working and properly connected. If communication resumes, Windows will automatically start using the drive again. If the drive has failed, replace it.

Location : hvh0004cl01-003, Storage Scale Unit 'hvh0004cl01-003', HPE, ProLiant DL380 Gen10, CZ20320HTJ, Slot position: 'PCI Slot 14 : Adapter 2'

Description : Physical Disk 'VO001920KWUEB'. Manufacturer 'Not available', Model 'VO001920KWUEB', Part Number 'Not available', Serial Number 'Y950A002TC7R '.