Detect potential corruption as part of recovery/rollback to stable

XMLWordPrintableJSON

    • Type: Task
    • Resolution: Unresolved
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • None
    • None
    • None

      As part of WT-5786, we began omitting data files that aren't large enough to have a descriptor block (clearly corrupted) in rollback to stable. We need to do this for repair_unfinished_indexes.js so that MongoD is able to startup successfully in repair mode (checking the database with verify and then potentially running salvage). Since we iterate over all data files in recovery and potentially run rollback to stable on them, we have more opportunities to detect corruption so we should do something to signal loudly to MongoDB that there is something wrong without someone explicitly knowing that they should run repair (which is what the test is simulating).

            Assignee:
            [DO NOT USE] Backlog - Storage Engines Team
            Reporter:
            Alex Cameron (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated: