This issue occurs in two different scenarios:
Scenario 1:
Prior to VDDK 6.0, the virtual disks of a target virtual machine were hotadded to the proxy virtual machine during VixDiskLib_ConnectEx and hotremoved during VixDiskLib_Disconnect API calls.
Starting from VDDK 6.0 and later, each virtual disk of the target virtual machine is hotadded to proxy during VixDiskLib_Open and hotremoved during VixDiskLib_Close API calls. Each hotadd/hotremove requires the proxy virtual machine to be reconfigured.
If the reconfiguration operation is time consuming, some of the hotadd/hotremove calls may timeout due to lock contention resulting in the errors listed in the symptoms section.
Scenario 2:
If a VDDK 6.0 base process crashes before it has cleaned up its lock files, the disk in question cannot be accessed until the lock is manually removed.