About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
This error means NTFS cannot record any more pieces of one file. It is about how many fragments the file is split into, not simply how big it is. Large compressed files and large sparse files are the cases Microsoft's Core team calls "somewhat common", and Veeam reports it on large deduplicated backup files; all of them fragment heavily. Defragmenting does not fix a compressed or sparse file. The lasting fix is to move the data off, reformat the volume with larger NTFS file records, and copy the data back, which erases the volume.
What the error means
Windows reports 0x80070299 with the text "The requested operation could not be completed due to a file system limitation." Microsoft's Core support team explains the mechanism in The Four Stages of NTFS File Growth, Part 2 (displayed date 12 March 2015).
NTFS describes a file's data fragments in file records, and each record is 1 KB by default. A very fragmented file needs many child records, and an attribute list keeps track of them. The post says the attribute list "is created using free space outside the Master File Table" and has "a hard limit of how large it can grow", with an upper limitation of 256 KB. When it is full, "additional file fragments cannot be created", and the post names the error among those it produces, alongside "Insufficient system resources exist to complete the requested service".
Microsoft's knowledge base article A heavily fragmented file in an NTFS volume may not grow beyond a certain size (KB 967351, no date shown, written for Windows Vista, Windows Server 2008 and 2008 R2 and Windows 7) gives the same cause and states that compressed and sparse files reach the limit sooner. A second Microsoft blog quotes the article's figure: the limit is usually between 40 GB and 90 GB for a very fragmented file (File System Limitation, displayed date 6 July 2011).
The Core team post names two situations where this is somewhat common: compressing very large files such as virtual hard disks, and very large SQL snapshots, which are sparse. It adds that you cannot clear this by running defragmentation, because "Sparse and compressed files are going to be fragmented."
Backup repositories hit the same wall. Veeam's knowledge base article Windows Repository with Data Deduplication error: "The requested operation could not be completed due to a file system limitation" (published 4 June 2014, last modified 13 January 2025, listed for Veeam Backup & Replication 9.5 to 13.1) reports the error, with the message "Failed to flush file buffers", when a repository sits on an NTFS volume with Windows Data Deduplication that was formatted with the default 1024-byte file record size. It says large restore points on that volume fail to write.
Likely causes
| Cause | What you see | How to confirm |
|---|---|---|
| Large deduplicated backup files on a volume with 1 KB file records | Backup or copy of one large file fails part-way; small files are fine | fsutil fsinfo ntfsinfo shows 1024 for bytes per file record segment; deduplication is enabled on the volume |
| NTFS compression on the destination folder or volume | A very large VHD or backup file fails to copy after a long run | Check whether compression is enabled on the destination |
| Large sparse files (database snapshots, sparse VHD-style files) | Failure on write or copy of a sparse file | fsutil sparse queryflag <file> reports the sparse flag |
| Very fragmented, long-lived file that keeps growing | Failure after many appends or in-place updates | fsutil file queryextents <file> shows many extents |
| A genuine different limit (path length, free space, quota) | A different error text | Match the error text exactly; this fault has the specific "file system limitation" wording |
Prove the cause before changing anything
Every step here reads only. The commands come from Microsoft's and Veeam's documentation and have not been run.
Check the file record size of the destination volume. Veeam's article tells you to run the following in an administrative command prompt or PowerShell window and check the "Bytes Per FileRecord Segment" value; 1024 is the default:
fsutil fsinfo ntfsinfo <drive>:The fsutil fsinfo reference confirms that
ntfsinfolists NTFS-specific volume information. It does not show the field name, so the field name comes from the Veeam article.Check whether the file is sparse. From the fsutil sparse reference:
fsutil sparse queryflag <filename>Look at how many extents the file has. The fsutil file reference lists
fsutil file queryextents <filename>. On a badly fragmented multi-terabyte file the output will be very long, so redirect it to a file on a different volume.Check compression on the destination folder or volume. The 2011 Microsoft blog found a directory with NTFS compression enabled was the cause of a 150 GB VHD copy failing after about 30 minutes, and reports a figure of about 30 GB as the size above which compressed directories stop accepting files (the post attributes that figure to an MSDN article).
Example (illustrative). BKP01 hosts a backup repository on D: with Windows Data Deduplication. A 3 TB backup file fails with 0x80070299 while smaller files are written without trouble. fsutil fsinfo ntfsinfo D: shows 1024 for bytes per file record segment. That matches the vendor's description, and the cure is a volume with larger records, not a retry.
Fix
Pick the cause that matched.
Compression enabled on the destination. Turn NTFS compression off for the destination. The 2011 post's author disabled it, zipped the file, and copied it without further problems. Disabling compression applies to new files; treat existing very large compressed files as at risk and move them to an uncompressed location.
1 KB file records on a volume that holds very large, fragmented files. The Core team's remedy is a three-step process: "Copy off your files", "Reformat the drive using the switch (Format /L)", and "Copy the files back". File record size is set at format time and cannot be changed afterwards.
- Danger: formatting erases every file on the volume. Microsoft's format reference warns "ALL DATA ON NON-REMOVABLE DISK ... WILL BE LOST". Before you start, have a verified copy of everything, including backups that live only on this volume, and confirm the copy restores.
- The reference describes
/Las NTFS only: "By default, a nontiered volume is formatted with small size file records and a tiered volume is formatted with large size file records. /L and /L:enable forces format to use large size file records and /L:disable forces format to use small size file records." - The Core team post says the larger records are 4 KB instead of 1 KB, so "in theory, a file could have FOUR TIMES the number of fragments before running into the same issue". The KB 967351 article also tells you to format with
/Ltogether with/FS:NTFS. - A command takes this shape:
format <drive>: /FS:NTFS /L /Q. The/Qswitch is a quick format; Microsoft advises using it only on previously formatted volumes in good condition.
Veeam says issues "are reduced or eliminated when the NTFS volume is formatted with 'large FRS', which sets the FileRecord Segment value to 4096", and that "it is possible the same error message may occur later if the 4096 Bytes Per FileRecord Segment limit is hit." It attributes this to "advice from Microsoft and reported experiences of other customers". Its listed workarounds are to enable per-machine backup files so each machine's restore points are separate, smaller files, or, if space allows, to stop the backup software, move the restore points off, reformat with the large file record size, and move them back.
Defragmentation. For files that are not compressed or sparse, KB 967351 lists defragmenting as a workaround. For compressed or sparse files it does not resolve the issue.
Newer Windows releases. The Core team post says that in newer versions NTFS stops fragmenting compressed and sparse files before the attribute list reaches 100% of its maximum, and the post itself places this change at Windows 8.1 and Windows Server 2012 R2 ("until everyone gets to Windows 8.1 or Windows Server 2012 R2"). A comment on the post adds that, from about 75% of the limit, Windows stops adding fragmentation from runs of zeros for empty compression units. The Veeam article, last modified in 2025, shows the error still occurs on deduplicated repositories, so do not assume newer Windows avoids it.
Confirm the fix
- Re-run
fsutil fsinfo ntfsinfo <drive>:and confirm the bytes per file record segment value is the larger one (the sources describe 4 KB, which is 4096 bytes). - Write the same large file that failed, from the same job, to the new volume and confirm it completes.
- Restore from that backup file to a scratch location and compare the restored data. A file that wrote successfully is not proof that it restores.
What to watch afterwards
- The same error on a different large file means the limit is still being approached. Veeam notes the 4096-byte case can also recur.
- Keep individual backup files smaller where the backup software allows it; Veeam's per-machine backup files are one example.
- Avoid putting compressed folders under very large VHD or backup files, and keep a note of which volumes have deduplication enabled and which file record size they were formatted with.
Read next
Related articles you may want next.
- Backup & RecoveryBacking up and restoring Data Deduplication volumes: what is really on the diskPlan Windows Data Deduplication recovery around the chunk store, supported backup modes, compatible servers and realistic restore capacity.Read article
- Backup & RecoverySparse files, hard links and reflinks: why a restore can be bigger than the sourceWhy a Linux restore can need more space than the original: sparse files, hard links and reflinks, how tar, rsync and cp handle each, and how to size a restore.Read article
- HostingChoosing a storage tier for each volume: databases, logs, files and archivesWhen a hosting quote prices every disk on the same fast storage, how to decide volume by volume which data needs speed and which can sit on cheaper capacity storage, using Microsoft's SQL Server file placement guidance.Read article
Risks and limits
- These commands have not been run or lab-tested, and the format step destroys data.
- KB 967351 is written for Windows Vista, Windows Server 2008, 2008 R2 and Windows 7, with no date shown. The Core team post is from 2015. Microsoft's pages on this limit do not restate it for Windows Server 2022 or later; the Veeam article is the only current source showing that it still occurs.
- The 40 GB to 90 GB figure and the 30 GB compressed directory figure come from a 2011 blog quoting other documents and are rough, not guaranteed thresholds.
- The Veeam article names the field to check but does not mention
/L; the/Lsyntax is from Microsoft's format reference. Neither source says what happens on ReFS, so this article does not advise it. - The Microsoft format reference describes
/Lfor NTFS only.
Sources
- The Four Stages of NTFS File Growth, Part 2 — Microsoft Core support blog (archived), displayed date 12 March 2015, accessed 16 November 2026
- A heavily fragmented file in an NTFS volume may not grow beyond a certain size (KB 967351) — Microsoft Support, date not shown, accessed 16 November 2026
- File System Limitation — Microsoft blog (archived), displayed date 6 July 2011, accessed 16 November 2026
- Windows Repository with Data Deduplication error: "The requested operation could not be completed due to a file system limitation" (KB 1893) — Veeam, published 4 June 2014, last modified 13 January 2025, accessed 16 November 2026
- format | Microsoft Learn — Microsoft, page dated 2023-09-28, updated 2025-10-22, accessed 16 November 2026
- fsutil fsinfo | Microsoft Learn — Microsoft, page dated 2017-10-16, updated 2026-02-16, accessed 16 November 2026
- fsutil file | Microsoft Learn — Microsoft, page dated 2017-10-16, updated 2026-02-16, accessed 16 November 2026
- fsutil sparse | Microsoft Learn — Microsoft, page dated 2017-10-16, updated 2026-02-16, accessed 16 November 2026
This article was originally published on .