Skip to content

Est. 2011ยทMicrosoft Partner ID 7033487ยทDelivery under 3 minยทSupport 7 days a week

Your vault is empty.

License Error Event ID 25

Event ID 25 and 0x80780119: shadow storage runs out and snapshots vanish

11 min read Updated October 5, 2026 Windows Server: RDS, Hyper-V & Clustering

Fix it now

The shadow copies were deleted because the shadow copy storage area could not hold what the volume was changing. Read the current association and its maximum, then either give it more room or take the competing write load out of the window.

Run these in an elevated Command Prompt. Change the volume letters and the size to match your server

vssadmin list shadowstorage /for=C:
vssadmin list shadows
vssadmin resize shadowstorage /for=C: /on=C: /maxsize=50GB
  1. Note the Maximum Shadow Copy Storage Space in the first output. Microsoft’s documented first move where snapshots are being deleted is to increase it, at least doubling the current value.
  2. If the volume itself has no room, put the diff area on a volume that has: the association names both volumes, so /for= is the volume being snapped and /on= is where its diff area lives.
  3. Look at what is rewriting the volume during the window. Bulk classification, deduplication, defragmentation, large restores and mass deletes all inflate diff area usage enormously.
  4. Move those jobs clear of the backup window and re-run, watching vssadmin list shadowstorage while it runs.

Microsoft’s published warning is that resizing the storage association may cause shadow copies to disappear. Check what depends on the current restore points before you change the size.

If the peak used size stays comfortably below the maximum through a full backup, you are done. The next section explains why the diff area fills at the rate it does.

Why it happens

A shadow copy is not a copy. It works by copy-on-write: when a block on the volume is about to be overwritten, the original contents are first written into the shadow copy storage area – the diff area – so the snapshot can still present the old version. The diff area therefore grows in proportion to how much of the volume changes while the snapshot exists, not to how much data the volume holds.

That is why a quiet file server needs very little and a busy database volume can exhaust a large area in minutes. Write throughput matters as well as size, because the diff area has to be written before the original write may proceed. The area can live on any local volume, but Microsoft’s requirement is that it must be an NTFS volume with enough space to store it.

When either constraint is hit, the service has one option: discard shadow copies on that volume. Microsoft’s own wording for the gentler version of this is that the oldest shadow copy of the volume was deleted to keep disk space usage for shadow copies below the user defined limit. From the point of view of a backup in progress, it makes no difference which version you get – the snapshot it was reading from stops existing and the job fails.

The maximum is simply too small

You have this one if The used size sits at or very near the maximum at the moment of failure.

  1. Read the current figures: vssadmin list shadowstorage /for=X:
  2. At least double the maximum, which is Microsoft’s documented remedy: vssadmin resize shadowstorage /for=X: /on=X: /maxsize=<size>
  3. Re-run the backup and watch the used figure to confirm it now has headroom.
  4. Re-check after a month, because the change rate grows with the workload.

The maximum accepts a size, a percentage, or UNBOUNDED. On a volume that also holds data, an unbounded diff area lets snapshots consume everything, which trades one failure for a worse one.

Something is rewriting the whole volume while the snapshot exists

You have this one if Failures line up with classification, deduplication, defragmentation, archive moves or bulk deletions.

  1. List the scheduled maintenance that touches the volume and compare it against the backup window.
  2. Move optimisation and defragmentation jobs off that window entirely.
  3. Where a job rewrites files rather than editing them – file classification writing properties into Office documents is Microsoft’s own example – run it in stages rather than across the whole volume at once.
  4. Re-run and confirm the peak diff area usage during the window drops.

Defragmenting a volume with active shadow copies moves large amounts of data and can fill a diff area on its own, even though no file contents changed.

The diff area volume itself has no room to grow into

You have this one if The volume is nearly full and the limit is not the binding constraint.

  1. Free space on the volume first; it is the cheapest fix and often the only one needed.
  2. Where another volume has room and reasonable performance, relocate the association to it.
  3. Set monitoring on free space for both volumes so the next occurrence is a warning rather than a failed backup.

Snapshot enumeration is slow because of stale device entries

You have this one if Shadow copies disappear on a server with plenty of space, sometimes with the storage volume reported as not present.

  1. Preview what would be removed: devnodeclean /n
  2. Remove the phantom VolumeSnapshot entries: devnodeclean
  3. Confirm the tool only deletes phantom entries – valid shadow copies are left alone.
  4. Schedule it to run periodically on servers that create and delete many snapshots.

Microsoft documents this for volsnap event 32, where enumeration takes so long that the snapshots are deleted while the volume comes online.

Full reference

Which constraint is actually binding

Observation Constraint
Used size at the maximum before the failure The configured limit is too low
Free space on the diff area volume reaching zero The volume is too full, whatever the limit says
Failure only during optimisation, classification or bulk change Write churn, not sizing
Failure on slow storage with space to spare The diff area cannot be written fast enough
Snapshots vanishing with the storage volume reported absent Phantom device entries slowing enumeration

The commands, and what each is allowed to do

Command Effect
vssadmin list shadowstorage /for=X: Shows the association for that volume with used, allocated and maximum sizes
vssadmin list shadows Lists the existing shadow copies
vssadmin resize shadowstorage /for=X: /on=Y: /maxsize=900MB Resizes the maximum for an existing association. Both switches are required
vssadmin resize shadowstorage /for=C: /on=C: /maxsize=20% The same, expressed as a percentage of the volume
vssadmin resize shadowstorage /for=C: /on=D: /maxsize=UNBOUNDED No limit on storage space for that association
vssadmin delete shadows Deletes volume shadow copies

MaxSizeSpec must be 1 MB or greater and carries a unit – KB, MB, GB, TB, PB or EB. With no unit it is interpreted as bytes, which is a quiet way to set a limit far smaller than you intended. There is a matching lower bound: the minimum amount of storage space is governed by the MinDiffAreaFileSize registry value.

Published limits worth knowing before you design around this

  • The maximum number of software shadow copies for each volume is 512.
  • By default you can maintain only 64 shadow copies for the Shadow Copies of Shared Folders feature.
  • VSS is not supported on volumes larger than 64 TB. Enabling it, creating writable snapshots, enabling shadow copies for shared folders, running a backup with shadow copies enabled, or running chkdsk on such a volume are all outside support.
  • The diff area can be on any local volume, but it must be NTFS and it must have enough space to store it.

Sizing it from evidence rather than habit

There is no universal figure, because the answer depends entirely on how much of the volume changes while a snapshot is open. Measure it: start a backup, watch the used size climb, and note the peak. Then set the maximum comfortably above that peak and re-check after a month, because change rate grows with the workload and a limit set at commissioning is usually the one still in place three years later.

Where the peak is genuinely large, the more durable answer is usually to reduce what changes during the window rather than to keep enlarging the area. A deduplication or classification job that rewrites entire files, moved out of the backup window, can cut the requirement by an order of magnitude at no cost.

Reducing the maximum below current usage deletes shadow copies to fit, and Microsoft’s warning is that resizing the association at all may cause shadow copies to disappear. Read the used figure and confirm nothing depends on the current restore points before shrinking anything.

When a licence is the actual fix

There is nothing to buy here. Resizing shadow storage, moving a diff area and rescheduling an optimisation job are free, and if the maximum was simply set too low you are finished at no cost. A purchase only enters the picture when the volume has no room because one server is doing too many jobs, and the real fix is to move the role onto its own instance with storage sized for it. That instance needs a licence: Windows Server Standard covers two virtual machines plus one Hyper-V host per licence, and Datacenter covers unlimited virtual machines plus one Hyper-V host. Arco supplies Windows Server 2025 Standard and will check what your existing licences already allow before recommending anything new.

Every code this article covers

Code What it points at Source
Event ID 25 Logged by volsnap when shadow copies of a volume are deleted. Microsoft publishes no message text for this ID; read it as the storage area failing to keep up and diagnose from the association’s used and maximum sizes not published by the vendor
Event ID 33 Also from volsnap. Microsoft publishes the matching System log message – the oldest shadow copy of the volume was deleted to keep disk space usage below the user defined limit – but does not attach an event ID to it not published by the vendor
0x80780119 Reported by Windows Backup when a volume does not have the free space the shadow copy requires. Microsoft publishes no documentation page for the code, so treat the storage association figures as the evidence not published by the vendor
0x8004231F A VSS insufficient-storage failure. Microsoft documents the condition under the constant VSS_E_INSUFFICIENT_STORAGE – the system or provider has insufficient storage space, and old or unnecessary persistent shadow copies should be deleted – but does not publish the hexadecimal value alongside it not published by the vendor

Confirm the fix worked

  1. vssadmin list shadowstorage after a full backup shows a peak used size comfortably below the maximum.
  2. No shadow copies were deleted during the run: the backup’s own snapshot survives to the end of the job.
  3. Free space on both the source volume and the diff area volume has headroom after the window.
  4. A complete backup cycle passes with no volsnap events in the System log.
  5. If phantom entries were the cause, devnodeclean /n now previews nothing to remove.

Questions people ask about this

How big should the diff area be?

Big enough to hold everything that changes on the volume while a snapshot exists, with headroom. There is no universal figure. Measure the peak during a real backup window and size from what you see, then re-check in a month.

Is it safe to set the maximum to UNBOUNDED?

On a volume dedicated to hosting diff areas, yes. On a volume that also holds live data you are permitting snapshots to consume all remaining space, which trades a failed backup for a full volume.

Will resizing delete my existing snapshots?

It can. Reducing the maximum below current usage certainly will, and Microsoft warns that resizing an association may cause shadow copies to disappear in any case. Check the used figure and what depends on those restore points first.

How many shadow copies can a volume hold?

512 software shadow copies per volume, and by default 64 for the Shadow Copies of Shared Folders feature. If you are near either number, that is a design question rather than a sizing one.

Does a newer Windows Server release change any of this?

The mechanism is the same. A supported release keeps receiving storage and snapshot fixes and current vendor drivers underneath it, but it does not make an undersized diff area larger.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

Free Fix Event ID 16010 and 20148: Hyper-V storage operations fail and VMs go missing License Error Event ID 1074 and 6008: separating an expired Server evaluation from a crash License Error HTTP 502.5 and 500.30: ASP.NET Core apps fail to start behind IIS Free Fix Event ID 1194 and 1257: cluster name objects fail to register in AD or DNS
โ† Back to Knowledge Base