Fix it now
0x8004230F is VSS_E_UNEXPECTED_PROVIDER_ERROR: the shadow copy provider hit an unexpected error while processing the request. 0x80042308 is VSS_E_OBJECT_NOT_FOUND, the object does not exist. Between them they point at a writer, a provider or a missing storage association – and the requester that reported them is never the fault.
vssadmin list writers
vssadmin list shadowstorage
fsutil usn queryjournal C:
- Read the writer list for any writer whose state is not stable, or whose last error is anything other than none. Note the writer name; that is what tells you which service to restart.
- Restart the service that owns it. Microsoft documents that the VSS service itself hosts the registry, COM+ class registration database, shadow copy optimization and ASR writers, and that a writer belonging to an application is reset by restarting that application. A machine restart resets all of them.
- Check that each volume has somewhere to keep shadow copies. A volume with no association cannot hold one, and 0x80042308 is what a missing object looks like.
- Give it room where it has none:
vssadmin resize shadowstorage /for=C: /on=C: /maxsize=20GB. Microsoft warns this may cause existing shadow copies to disappear, so do it deliberately. - Free space on the source volume itself, then retry the operation that failed.
For Windows Server Backup, Microsoft publishes minimum free space per volume: 50 MB on volumes under 500 MB, 320 MB on volumes over 500 MB, and at least 1 GB recommended on volumes over 1 GB.
If writers report stable and the operation now completes, you are done. If it fails again, the next section explains which of the three participants is refusing.
Why it happens
A shadow copy is a coordinated operation between three kinds of participant. A requester asks for one: a backup product, System Restore, or a built-in Windows component. Writers represent applications that hold data, and each is told to bring its data to a consistent point and hold still. A provider makes the point-in-time copy, either the software provider built into Windows or a hardware provider supplied by a storage array.
Any of those can fail, and the requester only ever sees the resulting code. That is why the same handful of codes appear whether the requester was a backup job or the System Protection dialog, and why the useful work is one level down: find out which writer or provider is unhappy and deal with that.
The codes divide cleanly along those lines. 0x8004230F, VSS_E_UNEXPECTED_PROVIDER_ERROR, is the provider reporting an unexpected error while processing the operation – and Microsoft’s guidance for provider errors is that the provider logs the detail in the event log, so the code is a pointer rather than an answer. 0x80042308, VSS_E_OBJECT_NOT_FOUND, means the object does not exist on the server; Microsoft’s Data Protection Manager error list gives the same value as the data source not being available. 0x800423F3, VSS_E_WRITERERROR_RETRYABLE, is a writer reporting a transient error, and Microsoft’s own advice is that retrying might work, and that if the error persists you should restart the service hosting the writer.
0x80071A91 is the odd one out and the most actionable. It is ERROR_RM_NOT_ACTIVE, and Microsoft documents it as the transaction resource manager being unable to process file operations. Its published causes are a corrupted TxF transaction log on the system volume, KTM or TxF being unable to start or manage file transactions, a previous update leaving the transaction subsystem inconsistent, disk errors blocking access to the log, an antivirus program or file system filter driver blocking transactional operations, and a corrupted component store. Its documented first fix is one command.
A writer is stuck in a failed state
You have this one if vssadmin list writers shows a writer whose state is not stable, or which reports an error from its last attempt.
- Note the writer’s name. If it is the registry, COM+ class registration database, shadow copy optimization or ASR writer, restart the Volume Shadow Copy service, which hosts all four.
- Otherwise restart the service belonging to the application that owns the writer, which is what Microsoft documents for writers hosted in their own process.
- Re-run
vssadmin list writersand confirm the writer is now stable. If restarting the service does not clear it, restart the machine. - If it fails again immediately, read that application’s own logs – a writer that keeps failing is usually reacting to a real fault in the application it represents.
0x800423F3 specifically means retrying may work. It is worth one retry before anything else, because a transient writer error is exactly what it says.
There is no shadow storage association, or it is too small
You have this one if 0x80042308, copies that are created and immediately disappear, or a failure on one specific volume while others work.
- List what exists, then create or resize the association:
vssadmin resize shadowstorage /for=E: /on=E: /maxsize=25GB. - For volumes with a lot of change, place the shadow storage on a different volume by giving a different letter to
/on=. - Confirm there is genuine free space on the source volume too; the association is a limit, not a reservation.
- Check the documented minimums for Windows Server Backup – 50 MB under 500 MB, 320 MB over 500 MB, at least 1 GB recommended over 1 GB – before concluding space is not the issue.
vssadmin resize shadowstorage accepts UNBOUNDED and a percentage as well as a size. Microsoft warns that resizing the association may cause shadow copies to disappear.
The transaction resource manager is not active
You have this one if 0x80071A91, often alongside update failures on the same machine.
- Run
fsutil resource setautoreset true C:\and restart the machine. That is Microsoft’s documented first step for this code. - Follow with
DISM /Online /Cleanup-Image /RestoreHealthandsfc /scannow, because a corrupted component store is one of the published causes. - Run
chkdsk C: /f, since disk errors blocking access to the transaction log are another. - Check whether an antivirus product or file system filter driver is blocking transactional file operations, which Microsoft also names.
A leftover third-party provider is in the way
You have this one if A provider from a backup or storage product that is no longer installed still appears, and the built-in provider is not being used.
- Reinstall the product that owns the provider and uninstall it properly with the vendor’s removal tool, so the provider is deregistered cleanly.
- Where that is impossible, follow the vendor’s documented steps for removing the provider rather than deleting files or registry keys at random.
- Confirm afterwards that only the Microsoft software provider, and any hardware provider you actually use, remain.
The volume or its change journal is damaged
You have this one if Failures cluster on one volume, chkdsk finds problems, or the volume has been through a disorderly shutdown or a storage disconnection.
- Run
chkdsk <letter>: /scanfirst, which is non-disruptive, then a full/fpass in a maintenance window if it finds problems. - Check the change journal’s state with
fsutil usn queryjournal C:. - Where the journal is genuinely damaged and your storage or backup vendor instructs you to recreate it, do so only after reading the warning below.
Full reference
Which service resets which writer
| Writer | How it is reset |
|---|---|
| Registry writer | Restart the Volume Shadow Copy service, which hosts it |
| COM+ class registration database writer | Restart the Volume Shadow Copy service |
| Shadow copy optimization writer | Restart the Volume Shadow Copy service |
| Automated System Recovery (ASR) writer | Restart the Volume Shadow Copy service |
| Any writer hosted by an application | Restart that application, which is what Microsoft documents |
| Any writer, when you are unsure | Restart the machine; writers reset on a restart |
Those four are the ones Microsoft names as hosted in the VSS service. Every other writer belongs to the product that installed it, and the documented advice is to restart that product rather than the shadow copy service – restarting VSS will not touch a writer running in someone else’s process.
The codes, and how far each one gets you
| Code | Published as | Next step |
|---|---|---|
0x8004230F |
VSS_E_UNEXPECTED_PROVIDER_ERROR | The provider logged the detail. Read the Application log entries around the failure |
0x80042308 |
VSS_E_OBJECT_NOT_FOUND: the object does not exist on the server | Check the shadow storage association and, in a backup context, whether the data source is available |
0x800423F3 |
VSS_E_WRITERERROR_RETRYABLE: a transient writer error | Retry once, then restart the service hosting the writer |
0x80071A91 |
ERROR_RM_NOT_ACTIVE | fsutil resource setautoreset true C:\, then restart, then repair the component store |
0x81000101 |
Not published by Microsoft | Treat as a request that did not complete in time; read the log around it |
0x80042301 |
Not published by Microsoft | Treat as a request made in a state the service could not serve; restart the writers and retry |
Space, which is the cause people rule out too early
The software provider needs three things at once: somewhere to store the copy-on-write data, which is what shadow storage is; free space on the source volume itself; and a healthy change journal on NTFS volumes. Starve it of any one of those and creation fails, often with a code that says nothing about space at all. Microsoft’s published minimums for Windows Server Backup are a useful floor: at least 50 MB free on volumes smaller than 500 MB, at least 320 MB on volumes larger than 500 MB, and at least 1 GB recommended on any volume over 1 GB.
Never delete the change journal on a domain controller, a DFS Replication member, or any machine whose backup product depends on it, without confirming the consequences with that product’s vendor. Replication and incremental backup tooling use the journal to know what changed, so recreating it can force a full resynchronisation – and on a replication member the consequences are considerably worse than that. Try fsutil resource setautoreset true C:\ and a component store repair first; those address the documented cause of 0x80071A91 without touching the journal.
Command availability, which is not what you might expect
Microsoft’s current vssadmin command reference documents four subcommands – delete shadows, list shadows, list writers and resize shadowstorage – and lists all four as available on both client and server editions. Other verbs the tool accepts, including list providers, list shadowstorage and create shadow, do not appear in that reference. They work on the machines most administrators run them on, but if you are writing a procedure for someone else to follow, build it on the documented four and treat the rest as convenience.
Where the real story is written down
- The Application event log, source VSS. The entries in the seconds before the failure usually identify the actual fault, and provider errors in particular are documented as being logged there rather than returned.
- The application’s own log, for any writer that keeps failing after its service is restarted.
- The System log, for disk and controller errors on the volume being copied.
- The storage array’s log, where a hardware provider is in use – a hardware provider vetoing a snapshot is an array problem, not a Windows one.
- The backup product’s job log, which often records the writer name and failure reason more legibly than Windows does.
Failures that only happen during the backup window
That pattern usually means the volume is too full or too busy rather than broken. The software provider needs working room on the volume being copied, not only in shadow storage, and a provider that times out under load reports a service error rather than a storage one. Measure free space and write latency during the window itself; a measurement taken at lunchtime tells you nothing about what happens at two in the morning.
Every code this article covers
| Code | What it points at | Source |
|---|---|---|
0x8004230F |
VSS_E_UNEXPECTED_PROVIDER_ERROR: the shadow copy provider had an unexpected error while trying to process the specified operation. The provider logs the detail in the event log | Microsoft Learn |
0x80042308 |
VSS_E_OBJECT_NOT_FOUND: the object does not exist on the server. In Microsoft’s DPM error list the same value is described as the data source not being available | Microsoft Learn |
0x800423F3 |
VSS_E_WRITERERROR_RETRYABLE: the writer experienced a transient error, and retrying might succeed. If it persists, restart the service hosting the writer | Microsoft Learn |
0x80071A91 |
ERROR_RM_NOT_ACTIVE: the transaction resource manager cannot process file operations. Documented causes include a corrupted TxF log, disk errors blocking it, a filter driver blocking transactional operations and a corrupt component store | Microsoft Learn |
0x81000101 |
Seen when a shadow copy request does not complete. Microsoft publishes no meaning for this value, so read the Application log entries around the failure rather than the code | not published by the vendor |
0x80042301 |
Returned when the shadow copy service is asked for something in a state it cannot serve. Microsoft publishes no meaning tied to this value, so treat it functionally and reset the writers before retrying | not published by the vendor |
Confirm the fix worked
vssadmin list writersshows every writer in a stable state with no last error.- The operation that originally failed – a backup job, a restore point, a snapshot – completes.
- Each volume that needs one has a shadow storage association with room in it, and the source volume has free space above the documented minimums.
- If 0x80071A91 was involved, the machine has been restarted after
fsutil resource setautoreset true C:\and the component store repairs report clean. - No new VSS errors appear in the Application event log over the following days, including during a backup window.
Questions people ask about this
Do I need third-party backup software to fix this?
No. Every step here uses tools built into Windows, and the fixes are configuration, space and service state. A backup product does not repair a broken shadow copy service; it will fail the same way until the underlying fault is dealt with.
Why do backups fail while restore points still work?
They are different requesters using the same plumbing, and they engage different writers. A backup product typically involves far more writers than System Restore, so one failed application writer breaks backups while restore points keep working.
What does 0x80071A91 have to do with shadow copies?
It is a different subsystem that gets in the way of the same operations. Microsoft documents it as the transaction resource manager being unable to process file operations, with a corrupted TxF log, disk errors, a filter driver or a corrupt component store as causes. The documented first step is fsutil resource setautoreset true C:\ followed by a restart.
Should I recreate the USN journal?
Only on the advice of the vendor whose product depends on it, and never casually on a domain controller or a DFS Replication member. Replication and incremental backup use the journal to know what changed, so recreating it can force a full resynchronisation. Deal with the documented causes first.
Will adding disk space fix it?
Often, and it is the cheapest thing to try. The provider needs free space on the source volume and a large enough shadow storage allocation, and both being marginal produces intermittent failures that look far more mysterious than they are. Microsoft’s published minimums for Windows Server Backup are a reasonable floor to measure against.
