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.
You set seven days of retention and find thirteen restore points on disk, or a repository twice the size you planned. In a forward incremental chain this is how retention is designed to behave. Each incremental depends on the one before it, so the oldest point cannot be deleted alone. The software waits until every point of the old chain is outdated, then deletes the whole chain at once.
The same symptom has other causes: a scheduled full did not run, the job was idle and the minimum of three points applies, or the backups are no longer attached to a job. Count what you have before changing anything.
What the symptom means
Veeam's short-term retention page says that once the allowed number is exceeded the software "removes the earliest restore point from the backup chain". For forward incremental chains the detail page explains why that cannot be done point by point: "If you delete a full backup file, the entire chain of incremental backup files will become useless. Similarly, if you delete any incremental backup file before the point to which you want to roll back, you will not be able to restore VM data (since later incremental backup files depend on earlier incremental backup files). For this reason, if you select the forward incremental backup method, on some days, there will be more restore points on disk than specified by retention policy settings. Veeam Backup & Replication will remove the full backup chain only after the last incremental backup file in the chain becomes outdated."
Three more rules from the same documentation change the count:
- Full backups are your job. "Veeam Backup & Replication does not create full backups automatically based on retention limits." The administrator must schedule synthetic or active fulls if the job is to build separate chains. Veeam deletes an old chain only after a new chain has started, so a scheduled full that never runs leaves the old chain in place.
- Minimum of three points. With retention in days, "the minimum number of retained restore points is 3." Veeam's example: retention of 5 days, a job stopped for 10 days, and three points remain after the next run. Veeam does not state the reason for the minimum; the example shows the effect, which is that a long stoppage does not leave the job with no restore points.
- The day the policy applies is ignored. Veeam says it "keeps the restore points for the N + 1 days, where N is the number of days you specify", and that "the retention period may be longer depending on the specified backup method."
Forever forward incremental chains behave differently. When retention is exceeded, Veeam "reuses empty data blocks in the full backup file" and merges the oldest incremental into the full, so the count stays near the setting. Veeam's Forward Incremental Backup page says: "If the synthetic full backup and active full backups are not enabled, Veeam Backup & Replication will produce a forever forward incremental backup chain." A job with no full scheduled therefore does not pile up points. Points and space grow with no drop when a scheduled full does not run.
Likely causes
| Cause | What you see | How to confirm |
|---|---|---|
| Forward incremental with periodic fulls | Point count swings between the setting and nearly double | Count points through the week; chains start on full-backup days |
| Forever forward incremental (no synthetic or active full scheduled) | Count stays near the setting; this is not a cause of growth | One VBK file that is updated as the oldest increment merges into it, many VIB files |
| A scheduled full did not run (failed or skipped session, or a chained job) | One long chain after weeks, points and space grow with no drop | No new VBK on the scheduled day. Veeam: a chained job whose initial job skips a full-backup day will not create the full |
| Minimum of 3 points | Retention of a few days, but more points left after downtime | Job stopped or failed for longer than the retention |
| Backups no longer processed by a job | Old backups stay, listed as orphaned | Veeam: background retention handles them; imported, copied and some other backups are excluded |
| Deleted machines kept | Data for removed VMs stays | Veeam: without the deleted-items option it "will remain in the backup repository forever" |
| Files locked by another process | Retention waits | Veeam: background retention "waits until the backup file is unlocked" |
| Bareos: pruning keeps the current backup | Job records older than the retention remain | Bareos never deletes the only record of a current backup of a FileSet |
Prove the cause before changing anything
Everything here reads only.
- Read the job's settings: retention in days or points, whether it is forward incremental or forever forward incremental, and the synthetic and active full schedules.
- Count points. The Veeam PowerShell Reference documents:
$backup = Get-VBRBackup -Name "<job name>"
Get-VBRRestorePoint -Backup $backup
- Count chains. List the job folder on the repository. Each VBK file starts a chain. The files in the Veeam documentation are VBK (full) and VIB (incremental).
- Find the newest full. Its date tells you when the next chain started, and when the previous one can go.
A healthy result for forward incremental is a sawtooth: the count rises through the week and drops when an old chain is deleted. A steady rise with no drop on a job that has a full scheduled means the scheduled full is not arriving. A job with no full scheduled should stay near the setting, because it is a forever forward chain. A count that stays at three with older data left over points to the minimum, or an orphaned backup.
Fix
Forward incremental, working as designed. Plan for two chains on disk at the peak. Veeam says "on some days, there will be more restore points on disk than specified". Size the repository for the peak, not the retention number. Shortening the chain with more frequent fulls reduces the number of old points but stores more full backups, so measure before choosing.
A scheduled full did not run. Find why the session failed or was skipped and fix that. For chained jobs, make sure the initial job runs on the full-backup days, because Veeam says the chained job otherwise creates no active or synthetic full on those days. If a job has no full scheduled and you want separate chains, scheduling synthetic or active fulls changes the method; read the next step first.
Disruptive step: changing the backup method, for example from forward incremental to forever forward incremental, changes how the chain is built and how retention deletes. Veeam's Switching Between Backup Methods page says it "does not transform the previously created chain" and instead "creates a new backup chain next to the existing one". When you move from forward incremental to forever forward incremental, "When the number of restore points created since the last full backup reaches the retention limit, the old backup chain is deleted." Moving from forever forward to forward incremental deletes the old chain "when the number of restore points in the new chain reaches the retention limit". Keep a copy of the old chain elsewhere if you need it beyond that point. Veeam documents no way to restore a deleted chain; switching back starts yet another new chain. A forever forward chain has a long dependency run and a merge on every cycle. On deduplicating storage appliances Veeam appends data instead of reusing blocks, "the backup chain may consume more disk space on the appliance".
Minimum of three. The setting can be changed in the configuration file on Linux backup servers or by a registry value on Windows backup servers. Veeam directs you to its support for this change. Do not edit it casually: Veeam's example shows the minimum keeping three points after a long stoppage.
Risk: lowering the minimum lets retention delete more restore points after a stoppage. Note the current value first. Putting it back protects only the points that still exist.
Orphaned or deleted-machine data. The deleted-items option on the job removes data for machines that are no longer processed. Veeam says "You must use the retention policy for deleted items data carefully" and recommends 7 days or more "to prevent unwanted data loss". Veeam also states that, with per-machine backup files disabled, the deleted-items retention policy "does not function if you enable synthetic full backups and active full backup", so it may not help on a job that builds synthetic fulls.
Disruptive step: enabling the deleted-items option removes the data of machines that are no longer in the job once the number of days you set has passed. Impact: those machines' restore points are no longer available. There is no way back once the data is removed; set the days high enough, and keep a copy of anything you may still need. For orphaned backups, background retention runs every 24 hours at 00:30, and can be started by hand with the Apply retention option.
Disruptive step: Apply retention deletes outdated restore points and, for orphaned backups, can delete the entire backup. Veeam documents no undo. Confirm the retention settings first, and keep a copy of anything you cannot recreate.
Bareos. Bareos prunes catalogue records, not data. The documentation says the shortest of the three retention periods takes precedence, and that Volume retention is applied separately. In the source for job pruning, a comment states "we never delete the only Job record that contains a current backup of a FileSet." A Volume is recycled only when all its jobs are pruned and the Volume retention has passed: "even if all the File and Job records are pruned from a Volume, the Volume will not be marked Purged until the Volume retention period expires." The Client directives Auto Prune, Job Retention and File Retention are marked deprecated in the Bareos 25 documentation. Check which Pool and Client settings your version actually uses.
Confirm the fix
After the next full backup has finished and enough days have passed for the old chain's last point to be outdated, count the points again. For forward incremental you should see the sawtooth, with the low point near your retention setting. For a skipped full, confirm a new VBK appeared on the scheduled day.
Example (illustrative): a daily job on fs01 keeps 7 days with a synthetic full every Sunday. The full is 300 GiB and each incremental 10 GiB, so a chain of one full and six incrementals is 300 + 6 x 10 = 360 GiB. This table is a simplification. Veeam's own worked example shows that the old chain is not removed on the day the new chain first holds the set number of points: Veeam waits for the days to pass after the new full, and the old chain goes on the following run. In the example here the first chain's last point is Saturday of week one, the new chain holds 7 points on Saturday of week two, and the old chain goes on Sunday of week three. At the end of each day in the second week the repository holds:
| Day | Points on disk | Space |
|---|---|---|
| Sunday | 8 | 660 GiB |
| Monday | 9 | 670 GiB |
| Tuesday | 10 | 680 GiB |
| Wednesday | 11 | 690 GiB |
| Thursday | 12 | 700 GiB |
| Friday | 13 | 710 GiB |
| Saturday, the peak | 14 | 720 GiB |
| Next Sunday: old chain removed, new full added | 8 | 660 GiB |
The setting says 7, and the disk holds up to 14 points and twice the space (720 divided by 360). Now suppose the Sunday full never runs for four weeks: the single chain holds one full and 27 incrementals, 300 + 27 x 10 = 570 GiB and 28 points, and nothing is deleted.
What to watch afterwards
- Chart point count and repository space per job. The sawtooth should return to the same low point each cycle.
- Alert if no new full backup file has appeared for more than one cycle.
- Re-check capacity whenever the full schedule, change rate or retention changes.
- Treat the point count as a result of the method, and state the peak when you quote retention to anyone.
Read next
Related articles you may want next.
- Backup & RecoveryGrandfather-father-son retention: counting the restore points a policy really keepsHow to count the restore points a 30 daily, 12 monthly backup policy keeps, where the gaps fall, and what to ask a provider to confirm in writing.Read article
- Backup & RecoverySize Backup Retention Using Measured Storage GrowthBuild a capacity forecast from stored growth, copy formats and retention dependencies, with explicit headroom, recovery needs and uncertainty.Read article
- Backup & RecoveryRetention that outlasts your ability to restoreHow retention periods, backup chains, keys and obsolete formats come apart, and how to test the path you claim to keep.Read article
Risks and limits
These commands have not been run or lab-tested. Veeam 13 uses days for new jobs, while jobs from earlier versions that count restore points keep working. The numbers in the example are arithmetic from stated assumptions, not a measurement. Retention for per-machine chains is counted per machine, and a machine that fails a session keeps its older points longer. Bareos settings and the source comment come from its documentation and source branch, not from a test run.
Do not delete backup files by hand to bring the count down. A manual deletion in the middle of a chain breaks every point that depends on it.
Sources
- Short-Term Retention Policy — Veeam, page updated 2026-05-22, accessed 17 November 2026
- Forward Incremental Backup Retention Policy — Veeam, page updated 2026-01-12, accessed 17 November 2026
- Forever Forward Incremental Backup Retention Policy — Veeam, page updated 2025-08-27, accessed 17 November 2026
- Removal of Restore Points — Veeam, page updated 2026-08-04, accessed 17 November 2026
- Forward Incremental Backup — Veeam, page updated 2026-03-11, accessed 17 November 2026
- Switching Between Backup Methods — Veeam, page updated 2025-08-12, accessed 17 November 2026
- Background Retention — Veeam, page updated 2026-10-07, accessed 17 November 2026
- Retention Policy for Deleted Items — Veeam, page updated 2025-05-30, accessed 17 November 2026
- Synthetic Full Backup and Active Full Backup — Veeam, page updated 2025-03-20, accessed 17 November 2026
- Get-VBRRestorePoint — Veeam, page updated 2026-08-19, accessed 17 November 2026
- Volume Management (pruning directives, recycling algorithm) — Bareos project, footer: Revision November 06, 2026, accessed 17 November 2026
- Director Configuration (Client retention directives) — Bareos project, footer: Revision November 06, 2026, accessed 17 November 2026
- Source file core/src/dird/ua_prune.cc (branch bareos-25) — Bareos project (GitHub), date not shown, accessed 17 November 2026
This article was originally published on .