Since vSAN 8.0U2, vLCM has started to manage the lifecycle of the witness node independently and not with the cluster it's associated with. A dedicated witness appliance in a pre-8.0U2 vCenter Server is managed by vLCM as part of the cluster. After the VC upgrade to 8.0U2, you will see this health check warning because the witness is now managed by vLCM as an independent node which is expected behavior. User action needs to be taken to get the witness appliance managed by vLCM again.
The health check is flagged because the cluster is managed by vLCM while the witness node is not.
Select the witness host and then navigate to Updates -> Hosts -> Image to manage the witness host with a single image.