Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-9932

Tests needed for crashes during backup restart

    • Type: Icon: Improvement Improvement
    • Resolution: Unresolved
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: Backup
    • Storage Engines
    • StorEng - Refinement Pipeline

      As a followup to WT-9926, we need some automated tests that ensure we have coverage of crashes during backup restart.


      We have been shown that starting up with a WiredTiger.backup file exercise code paths not normally taken, and the consequences of getting it wrong are pretty severe.

      Acceptance Criteria (Definition of Done)

      We need some automated testing that shows that crashes during a startup from a backup (or equivalently, starting up from a full directory copy made while the host system had a backup cursor open) can be full recovered, without loss of metadata or data.

      One approach is to look at the various *_abort programs, and pick one to be used as a model, or maybe one of them could be extended.


      Optionally, we could also use the test/syscall program to verify that the sequence of system calls we are making is sensible.

            backlog-server-storage-engines [DO NOT USE] Backlog - Storage Engines Team
            donald.anderson@mongodb.com Donald Anderson
            0 Vote for this issue
            6 Start watching this issue