Reproducible backup set corruption during concurrent restore ui use / "database is locked"

Support
1

os: windows 11 25h2 26200.8655

duplicati v2.3.0.3 (msi, x64)

backup job: local to ftps, size about 6 gb

reproducer steps:

  • start on “home” tab
  • run backup job
    • job shows “checking uploaded files …” (because configured like this)
  • meanwhile: switch to the “restore backup” tab on the left (i am using legacy ui)
    • i clicked on the select time to restore dropdown and looked at the date/timestamps available
  • switched back to the “home” tab where i was before
  • saw a red popup error box for my backup job “database is locked”

additional observations:

  • i tried to run “check files” offered below from the backup job actions
    • got error “Warning: 2026-07-07 14:20:00 +02 - [Warning-Duplicati.Library.Main.Operation.FilelistProcessor-ActiveUploadsDetected]: Active uploads detected, but no cleanup was performed. Run the “repair” command to clean up the incomplete files”

what i expected to happen:

  • when i accessed the backup job db on the restore ui tab while the job was still in its finishing phase, duplicati should have detected that the job is running and refused to gather this info (from the database) so it would not have been concurrently locked.

what happened instead:

  • the database was openend simulatenously a second time and the backup job running on the home tab errored, left in inconsistent state.
3

i can successfully repair the backup set and run the job again successfully after “duplicati-commandline repair …”

4

I agree, that would be a safer approach, and I actually assumed that was what happened.

Next canary will feature a lock manager that rejects such calls if an operation is already using the database.