I had this issue too. I did figure out it was related to another issue we had about a month ago. Last month we were applying some Windows updates to our vCenter and SQL servers. After rebooting the servers, we had one of our 5 VM host refused to reconnect to vCenter. All the VMs were still running but the host just wasn't communicating with vCenter. One article I read said to bounce the hostd and xpxa services. After that was done, the hostd daemon kept crashing. This made it impossible to connect to the host via the vSphere client. Fortunately, I had already turned on SSH before all of that. What caused the hostd daemon to crash was a full ramdisk. That was caused by a log that was consuming the ramdisk entirely. That log was for CIMPLE which is used by many hardware manufactures to establish communication between vSphere and CIM sensors. Truncating the log to free up space on the ramdisk fixed the issue.
Fast forward to yesterday, I was having the issue with unmounting a LUN from vSphere which returned the "cannot read the host cache configuration because some other process is modifying it at this time". Two of my VM Hosts were the issue. After reading you comment that rebooting the host resolved the issue, I remembered that would clear the ramdisk. I check on the ramdisk on both of those Hosts and they were full. I truncated that log and I was able to unmounts and delete that LUN.