How INCR incremental backups works
INCR backups are snapshot-based incremental backups that avoid re-backing up unchanged data. NetBackup uses the following Ops Manager APIs during INCR processing:
File list API: Retrieves the list of files (and sizes) contained in the incremental snapshot for a node. NetBackup uses this to understand what is present in the snapshot before backing it up.
Incremental file diff API: Identifies only the files that changed between the previous snapshot and the current incremental snapshot. NetBackup uses this "diff" to back up only changed data instead of the full snapshot contents.
Result: INCR runs are typically faster and smaller than FULL backups because unchanged data is not re-read and re-transferred.
Under certain timing and workload conditions, incremental (INCR) backups taken using MongoDB Ops Manager with NetBackup 11.2 and later may be incomplete.
This behavior can occur when new data is created very shortly after a snapshot is taken (for example, within a few seconds). The likelihood increases in multi‑node replica set environments that are under heavy load or have a high rate of data change. In such cases, one or more collections may not be included in the incremental backup.
Although the backup job may complete successfully, restores that rely on the affected incremental backups can result in missing or inconsistent data.
If an incremental backup is affected, all subsequent incremental backups that depend on the same backup chain may also be unreliable. There is currently no mechanism to detect this condition during backup execution. The issue can only be identified during restore validation.
Based on current analysis:
This issue applies only to block‑level incremental (INCR) backups introduced in NetBackup 11.2 and later
Full (FULL) backups are not affected Transaction log (TLOG) backups are not affected
The issue is less likely in single‑node deployments
This issue is currently under investigation with MongoDB.
For workaround and guidance ,