If you are processing a large number of files to/from a consumer-level Cloud service (excluding Amazon S3, Microsoft Azure, Google Storage, Backblaze B2, OpenStack, OVH), there is a chance that Out-of-memory errors may be encountered. The error message will be logged as such:


Failed to scan files: Out of memory


You may get this out of memory error when working with any of the consumer-level Cloud services as listed below:


- Box

- Dropbox

- Office 365 (OneDrive for Business and SharePoint)

- OneDrive (Personal)

- Google Drive

- SugarSync
- pCloud
- Citrix ShareFile


The simple solution is to switch to the 64-bit version of SyncBackPro. This lets SyncBack use all available memory instead of being restricted to just 2GB (as is the case with the 32-bit version).


SyncBackPro V12 uses less memory for large Azure and S3 downloads. If you are using an old version of SyncBackPro, and are having issues with memory usage when downloading very large files from Azure or S3 cloud services, then consider upgrading to V12.


Technical Reason


There is a special type of cloud data that needs to be stored in memory. This data include pointers, special-ID, para-ID, etc. which are interlinked together and relate to how the Cloud service handles and processes cloud-based data files.


Due to the way these cloud file details interrelate with one another, SyncBackPro has to store them in memory until the entire batch of files in-scope has completed processing and it can't simply transfer them over to disk storage to free up memory.


As a result, if there are a large number of files to be processed, SyncBackPro will run out of memory eventually. If you are using the 32-bit version of SyncBackPro then it can only use 2GB of free memory in the PC system, regardless of how much RAM is available.


If you encounter such errors during a profile run, we recommend splitting up your data into several profiles and then placing them in a Group so that each profile is processed sequentially. Source path and file selections on each separate profile may need to be tweaked, with each profile pointing to a sub-set so that the data can be backed up in batches.

Alternatively, consider switching to the 64-bit version, which can use all available free memory.