Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-10785

Improve the documentation for backup:query_id

    • Type: Icon: Documentation Documentation
    • Resolution: Unresolved
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • Storage Engines
    • 3
    • StorEng - Defined Pipeline

      Summary

      WiredTiger keeps track of the two last backup IDs, which can be then queried by opening a cursor for backup:query_id. The database, however, does not guarantee that those backups have indeed successfully completed from the application's (backup tool's) perspective, i.e., that all data finished copying.

      For example, this can happen when we crash in the middle of taking a backup (even before the backup cursor gets closed), while another thread creates a checkpoint in the meantime.

      The documentation should include a caveat that the backups returned from querying backup:query_id may not be complete. This clarification is important, because relying on the assumption that the returned backups are complete can lead to bugs that could result in corrupted backups.

      Acceptance criteria

      The documentation should mention this possibility on pages that mention querying for backup IDs, especially in the Writing WiredTiger applications section, e.g., on pages Cursors and Backups.

            Assignee:
            backlog-server-storage-engines [DO NOT USE] Backlog - Storage Engines Team
            Reporter:
            peter.macko@mongodb.com Peter Macko
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated: