• Type: Icon: Task Task
    • Resolution: Fixed
    • Priority: Icon: Major - P3 Major - P3
    • 5.3.0
    • Affects Version/s: None
    • Component/s: None
    • None
    • Fully Compatible
    • Execution Team 2022-02-21

      WT-8601 fixes a WT bug where timestamped writes to logged tables (potentially) do not provide repeatable reads across a restart. This is because the WT log does not preserve timestamp information, thus writes recovered from the log can be read by a reader with any read_timestamp input.

      MDB did rely on using read_timestamp for the oplog, but not in a way where the WT bug could manifest (anything the journal recovers is either rewritten by MDB or no longer a hole where the read_timestamp is meaningful).

      WT's change for 8601 isn't to fix the bug per-se, but rather to drop timestamp information completely on writes to logged tables. Thus the behavior is consistent and an application can never successfully rely on it existing.

      This ticket is to transform MDBs usage of read_timestamp on the oplog into the server itself.

            Assignee:
            daniel.gottlieb@mongodb.com Daniel Gottlieb (Inactive)
            Reporter:
            daniel.gottlieb@mongodb.com Daniel Gottlieb (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated:
              Resolved: