Investigate whether certain scenarios are still relevant to primary driven index builds

XMLWordPrintableJSON

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

      Of particular note are:

      jstests/noPassthrough/index_builds/index_abort_stepdown_prepare.js
      jstests/noPassthrough/index_builds/index_build_stepdown_prepared_transaction_intent_shared.js
      jstests/noPassthrough/index_builds/index_stepdown_prepare_conflict.js
      jstests/noPassthrough/index_builds/index_stepdown_during_scan.js

      These tests become racy in the primary driven suite (sometimes the index commits and sometimes not). We should evaluate whether this non-deterministic behavior is a correctness issue or a test interaction with primary driven index builds.

      Also, it may be worth investigating whether these tests even probe the advertised deadlock in a primary driven index build scenario. If not, we may want to keep them disabled.

      There are also a host of tests such as:

      jstests/noPassthrough/index_builds/index_failover_resolved_key_errors.js
      jstests/noPassthrough/index_builds/index_stepdown_abort_prepare_conflict.js

      Whose descriptions outline scenarios that may simply not apply to primary driven index builds. These tests might also pass (effectively becoming no-ops) when primary driven index builds get closer to feature parity with simultaneous.

            Assignee:
            Unassigned
            Reporter:
            Malik Endsley
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated: