V1 Time Series Bucket Corruption after Unclean Shutdown

XMLWordPrintableJSON

    • Type: Bug
    • Resolution: Works as Designed
    • Priority: Major - P3
    • None
    • Affects Version/s: 7.0.30
    • Component/s: None
    • None
    • Storage Execution
    • ALL
    • Storage Execution 2026-04-27
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      Unclean shutdown on an Atlas node hosting a time-series collection appears to have left two v1 bucket documents in a structurally corrupted state, such that time-series validation reports:

      The indexes in time-series bucket data fields are not consecutively increasing from '0'.

      This corresponds to the kTimeIndexNotIncreasing timeseries validation result in validate_adaptor.cpp and indicates that the data.<timeField> numeric keys in the bucket are no longer the expected 0..N-1 sequence.

      Since this happened when validating after an unclean shutdown, it seems possible that time series writes that were in-flight during a crash left the system in an inconsistent state, even after WT replayed the journal and rolled back to the last good checkpoint.

      There are more details in the linked AF ticket.

            Assignee:
            Malik Endsley
            Reporter:
            Chris Wolff
            Votes:
            1 Vote for this issue
            Watchers:
            6 Start watching this issue

              Created:
              Updated:
              Resolved: