Modern desktop cloud sync services- such as Microsoft OneDrive, Dropbox, and iCloud Drive - rely heavily on "Files On-Demand" or "Smart Sync" features to conserve local storage. They leave lightweight placeholder stubs on your hard drive while keeping the actual file contents in the cloud. However, when an unmanaged backup job scans these directories, it can accidentally trigger a cascading "hydration" process that drains network bandwidth and clogs local disks.

The "Hydration Trap"

When a standard backup scanner attempts to open a cloud-only placeholder file, the host operating system assumes a user or app requires immediate access to the data. In response, the OS automatically downloads the full file payload from the provider down to your local storage.

Running a routine background backup across an unhydrated directory can instantly trigger hundreds of gigabytes in forced downloads. This results in several immediate issues:

• Storage Exhaustion: Local drives quickly fill up as ghost placeholders convert to full-sized files.

  
  

• Bandwidth Saturation: Continuous high-volume downloads saturate home or office network pipelines.

  
  

• API Overhead & Throttling: Massive spikes in read/download requests can trigger rate limits or API fees from cloud providers.

  
  

How --exclude-non-local-files Solves It

Duplicati tackles this issue with the --exclude-non-local-files advanced flag. Rather than blindly opening file streams during a backup run, Duplicati inspects low-level filesystem metadata attributes beforehand.

The flag automatically identifies and filters out files marked with specific cloud-hydration attributes:

• Offline: Identifies files whose primary payload is stored offsite.

  
  

• RecallOnOpen: Indicates that opening the file will force a remote fetch.

  
  

• RecallOnDataAccess: Signals that reading byte streams requires remote retrieval.

  
  

If a file bears these attributes, Duplicati safely skips scanning its content. Files that are fully hydrated and locally present on your drive continue to be backed up normally.

Efficient Offsite Backups Without Data Duplication

Layering local device backups with secondary cloud storage targets (such as Backblaze B2, AWS S3, or Wasabi) provides excellent redundant protection. However, backing up files that already reside safely in your OneDrive or Dropbox cloud creates wasteful, redundant cloud-to-cloud data duplication.

Enabling --exclude-non-local-files creates a clean separation:

• No Unintentional Downloads: Your local backup engine won't pull cloud data down to your machine just to re-encrypt and re-upload it to your backup target.

  
  

• Leaner Backup Footprints: Offsite backup destinations only store truly unique local data, cutting down on secondary storage costs.

  
  

• Faster Scanning: Backup passes finish in a fraction of the time by ignoring remote stubs entirely.

  
  

To enable this protection, add --exclude-non-local-files under Advanced Options in your Duplicati backup job configuration. It ensures your backup pipeline stays lean, local-first, and completely isolated from cloud sync conflicts.