BUG : Potential inconsistency during global reclustering due to lack of overall transaction

[Hemil36]

Is there an existing issue for this?

• I have searched the existing issues

What happened?

Description
The function cluster_util_face_clusters_sync in backend/app/utils/face_clusters.py performs global face reclustering after 24 hours or 100 unclustered Images . It currently calls several database helper functions ( db_delete_all_clusters, db_insert_clusters_batch, db_update_face_cluster_ids_batch, db_update_metadata, etc.),each of which manages its own connection and transaction.

Because there is no overall transaction encompassing the entire reclustering workflow, the database can be left in a partially updated state if any step fails.

Example scenario:

• Clusters are deleted successfully via db_delete_all_clusters()

• Insertion of new clusters fails due to an exception in db_insert_clusters_batch()

• Result: the system ends up with no clusters and invalid face assignments

Record

• I agree to follow this project's Code of Conduct

Checklist before Submitting

• Have you updated docs for it?
• Have you added unit tests?
• Have you made sure unit tests pass?
• Have you made sure code formatting is correct?
• Do Your changes passes all tests?

[Hemil36]

@rahulharpal1603 I would love to work upon this issue.
ETA : 1 day

[Hemil36]

Hi @rahulharpal1603 ,

I believe The root cause is the repeated DB connection and independent transaction pattern across helpers. The global reclustering partial commit issue is a ripple effect of this core problem.

Using a centralized DB connection/transaction context for multi-step workflows would ensure atomicity and improve maintainability. It would be good if future operations follow this approach. If you agree , we can work on this approach.

[rahulharpal1603]

Sure, the transaction idea is good. I will review the PR soon.