ExportXMLWordPrintableJSON

    • Type: Task
    • Resolution: Fixed
    • Priority: Major - P3
    • 9.1.0-rc0, 9.1.0-rc1025
    • Affects Version/s: None
    • Component/s: None
    • None
    • Storage Execution
    • Fully Compatible
    • Storage Execution 2026-11-09, Storage Execution 2026-11-23
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      Add local.replset.recordIdsGroupId, the per-node consistency marker for set-wide Replicated RecordIds. A node honors oplog rids iff this collection holds a groupId.

      Details

      • Single-document collection: {_id: "recordIdsGroupId", groupId: UUID, since: {ts: Timestamp, t: NumberLong}}, where since is the foundingOpTime (term and timestamp) of the fencing entry that minted the groupId.
      • All document writes are timestamped and share the fencing entry's commit timestamp (config.image_collection precedent).
      • The collection must be logged=false: add the special case in WiredTigerUtil::useTableLogging (precedent local.replset.minvalid), so rollback-to-stable reverts it together with the data.
      • Clears are collection drops (clearInitialSyncId precedent). No untimestamped document writes.
      • Readable at startup before replication starts.

      Acceptance criteria

      • Create/read/drop API with timestamped writes.
      • useTableLogging special case with tests. In particular, test the recovery mode FCBIS runs after copying files: FCBIS finishes by running startup oplog recovery in standalone mode (recoverFromOplogAsStandalone with duringInitialSync=true). In that mode the table-logging rules differ from both steady-state replica set operation and plain standalone startup (see useTableLogging: journalWritesBecauseStandalone is false specifically when recovering from the oplog as standalone). The groupId table must keep its logged=false setting through that recovery, so that its contents stay revertible by rollback-to-stable together with the data files it was copied with, and validateTableLogging must accept the on-disk log setting afterward.

            Assignee:
            Ernesto Rodriguez Reina
            Reporter:
            Ernesto Rodriguez Reina
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved: