Make hash_lookup_unwind.js resilient to burn_in runs

XMLWordPrintableJSON

    • Type: Bug
    • Resolution: Unresolved
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • None
    • Query Execution
    • ALL
    • Hide 
       buildscripts/resmoke.py run --suites=unsplittable_collections_created_on_any_shard_jscore_passthrough --continueOnFailure --jobs=4 --repeatTestsMax=1000 --repeatTestsMin=2 --repeatTestsSecs=600.0 --shuffle --runAllFeatureFlagTests --enableEvergreenApiTestSelection --storageEngineCacheSizeGB=1 jstests/core/query/release_memory/hash_lookup_unwind.js
      Show
      buildscripts/resmoke.py run --suites=unsplittable_collections_created_on_any_shard_jscore_passthrough --continueOnFailure --jobs=4 --repeatTestsMax=1000 --repeatTestsMin=2 --repeatTestsSecs=600.0 --shuffle --runAllFeatureFlagTests --enableEvergreenApiTestSelection --storageEngineCacheSizeGB=1 jstests/core/query/release_memory/hash_lookup_unwind.js
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      jstests/core/query/release_memory/hash_lookup_unwind.js is flaky in the unsplittable_collections_created_on_any_shard_jscore_passthrough suite because it asserts on the global metrics.query.lookup.hashLookupSpills serverStatus counter, which is also incremented by background aggregation hooks in this suite.

      Make the test robust to concurrent $lookup spills by either:

      • relaxing assertions on hashLookupSpills to tolerate background noise, or
      • wrapping the spill-counter assertions in a small retry helper (similar to the $group release_memory test) so they only fail on consistent mismatches.

            Assignee:
            Unassigned
            Reporter:
            Catalin Sumanaru
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

              Created:
              Updated: