$merge into legacy timeseries collection in sharded database fails due to StaleConfig

XMLWordPrintableJSON

    • Type: Bug
    • Resolution: Won't Do
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • Query Integration
    • ALL
    • Hide

      Two tests are marked with TODO comments for this ticket in timeseries_merge_target_collection_sharded.js They are currently disabled for legacy timeseries but will fail if enabled and run with the CreateViewlessTimeseriesCollections feature flag disabled. 

      The error will look something like this:

      "StaleConfig{ ns: \"timeseries_merge_sharded_DB.system.buckets.unshardedColl\", vReceived:

      { e: ObjectId('000000000000000000000000'), t: Timestamp(0, 0), v: Timestamp(0, 0) }

      , vWanted: { e: ObjectId('68e403fac1dae8f887ecf9ba'), t: Timestamp(1759773690, 14), v: Timestamp(1, 0) }, shardId: \"timeseries_merge_sharded-rs0\" }: Exceeded maximum number of 10 retries attempting 'MongosProcessInterface::fieldsHaveSupportingUniqueIndex' :: caused by :: timestamp mismatch detected for timeseries_merge_sharded_DB.system.buckets.unshardedColl"}}

      Show
      Two tests are marked with TODO comments for this ticket in timeseries_merge_target_collection_sharded.js They are currently disabled for legacy timeseries but will fail if enabled and run with the CreateViewlessTimeseriesCollections feature flag disabled .  The error will look something like this: "StaleConfig{ ns: \"timeseries_merge_sharded_DB.system.buckets.unshardedColl\", vReceived: { e: ObjectId('000000000000000000000000'), t: Timestamp(0, 0), v: Timestamp(0, 0) } , vWanted: { e: ObjectId('68e403fac1dae8f887ecf9ba'), t: Timestamp(1759773690, 14), v: Timestamp(1, 0) }, shardId: \"timeseries_merge_sharded-rs0\" }: Exceeded maximum number of 10 retries attempting 'MongosProcessInterface::fieldsHaveSupportingUniqueIndex' :: caused by :: timestamp mismatch detected for timeseries_merge_sharded_DB.system.buckets.unshardedColl"}}
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      When $merge is run in a sharded cluster with a legacy timeseries collection that is either a sharded collection or a tracked unsharded collection and with an index specified in the "on" field of the $merge stage, it fails due to exceeding the maximum number of retries when running the ListIndexes command. This operation is expected to fail anways, but the failure should be due to the timeseries collection lacking a unique index as required by $merge.

       

            Assignee:
            Unassigned
            Reporter:
            Natalie Hill
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: