Identify the open handle with elevated Sysinternals Handle before using /x; closed applications do not exclude background services, filter drivers, or system processes from holding files on the volume.
Evidence-first workflow
- Open Command Prompt as administrator. Both CHKDSK and Handle require administrative privileges.^1^ ^2^
- Ensure the command prompt’s current directory is not on the target volume:
cd /d C:\ - Search for handles referencing the secondary drive, substituting its letter:
If that returns nothing useful, broaden the searches:handle.exe -u E:\
Handle performs a case-insensitive substring search against object names and reports the owning process, PID, hexadecimal handle value, object type, and path.^2^handle.exe E: handle.exe -a E: - Stop the identified owner through its supported control path—for example, stop its Windows service, pause its backup job, unmount its application data, or exit the process normally. Then rerun:
Avoidchkdsk E: /fhandle.exe -cas the routine fix. Microsoft warns that forcibly closing an individual handle can cause application or system instability.^2^
Process Explorer is the GUI alternative: run it elevated, use its handle search for E:\, and inspect the matching process. It is specifically designed to find processes with particular handles open.^3^
Do not assume AV, Search, or VSS without an owner
The lock failure establishes only that CHKDSK could not obtain exclusive volume access. An exclusive volume open is blocked when files are already open on that volume.^4^
The available documentation does not establish AV/EDR or Windows Search as the usual owner. Treat those as candidates only if Handle identifies their process or associated service.
VSS has one specifically documented case: a Shadow Copies for Shared Folders source volume cannot be locked while that feature is enabled. That does not establish generic VSS activity as the cause on ordinary client volumes.^1^
For repeated incidents across clients, record the process name, PID, path, owning account, endpoint product/version, and time before stopping anything. That distinguishes a common agent from unrelated per-machine activity.
When forcing a dismount is acceptable
After saving work and stopping known workloads, use:
chkdsk E: /x
/x forces the volume to dismount, invalidates every open handle to it, and already includes /f; chkdsk E: /f /x is therefore redundant.^1^
A forced dismount can produce unexpected behavior in programs whose handles are invalidated, so use it during a maintenance window rather than as the first diagnostic action.^5^
If the owner cannot be stopped safely, schedule an offline check at restart instead of force-dismounting a live workload. CHKDSK documents that when /f cannot lock a drive, Windows can offer to check it during the next restart.^1^
References