DISCOUNTHLL function returns an error when value of log2m is not 8

[saurabhlambe]

Example query:

SELECT DISTINCTCOUNTHLL(analytics_session_id_if_mobility_book,16) AS value FROM f_pax_app_sessions WHERE session_start_timestamp_10m >= '2024-01-12 17:00:00.0' AND session_start_timestamp_10m < '2024-01-15 17:00:00.0' HAVING value >= 0 LIMIT 10000000

As per Pinot docs, the DISCOUNTHLL function takes 2 arguments, log2m being optional. When the value of log2m is 8, the query runs correctly, when used a different value, it throws the following error:

[BaseSingleBlockCombineOperator] [pqr-6] Caught exception while merging results blocks (query: QueryContext{_tableName='f_pax_app_sessions_OFFLINE', _subquery=null, _selectExpressions=[distinctcounthll(analytics_session_id,'16')], _distinct=false, _aliasList=[value], _filter=(is_mobility_home_or_mobility_confirmation = '1' AND (session_start_timestamp_10m >= '1707238800000' AND session_start_timestamp_10m < '1707498000000')), _groupByExpressions=null, _havingFilter=distinctcounthll(analytics_session_id,'16') >= '0', _orderByExpressions=null, _limit=10, _offset=0, _queryOptions={responseFormat=sql, groupByMode=sql, timeoutMs=4902}, _expressionOverrideHints={}, _explain=false}) java.lang.IllegalStateException: Cannot merge HyperLogLogs of different sizes
 at org.apache.pinot.shaded.com.google.common.base.Preconditions.checkState(Preconditions.java:512) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.aggregation.function.DistinctCountHLLAggregationFunction.merge(DistinctCountHLLAggregationFunction.java:338) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.aggregation.function.DistinctCountHLLAggregationFunction.merge(DistinctCountHLLAggregationFunction.java:41) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.combine.merger.AggregationResultsBlockMerger.mergeResultsBlocks(AggregationResultsBlockMerger.java:42) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.combine.merger.AggregationResultsBlockMerger.mergeResultsBlocks(AggregationResultsBlockMerger.java:27) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.combine.BaseSingleBlockCombineOperator.mergeResults(BaseSingleBlockCombineOperator.java:146) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.combine.BaseSingleBlockCombineOperator.getNextBlock(BaseSingleBlockCombineOperator.java:62) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.combine.BaseSingleBlockCombineOperator.getNextBlock(BaseSingleBlockCombineOperator.java:45) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.BaseOperator.nextBlock(BaseOperator.java:43) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.InstanceResponseOperator.getCombinedResults(InstanceResponseOperator.java:118) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.InstanceResponseOperator.getNextBlock(InstanceResponseOperator.java:111) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.InstanceResponseOperator.getNextBlock(InstanceResponseOperator.java:39) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.operator.BaseOperator.nextBlock(BaseOperator.java:43) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.plan.GlobalPlanImplV0.execute(GlobalPlanImplV0.java:57) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.executor.ServerQueryExecutorV1Impl.executeInternal(ServerQueryExecutorV1Impl.java:376) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.executor.ServerQueryExecutorV1Impl.executeInternal(ServerQueryExecutorV1Impl.java:252) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.executor.ServerQueryExecutorV1Impl.execute(ServerQueryExecutorV1Impl.java:135) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.executor.QueryExecutor.execute(QueryExecutor.java:59) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.scheduler.QueryScheduler.processQueryAndSerialize(QueryScheduler.java:154) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.core.query.scheduler.QueryScheduler.lambda$createQueryFutureTask$0(QueryScheduler.java:136) ~[pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at java.util.concurrent.FutureTask.run(FutureTask.java:264) [?:?]
 at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:515) [?:?]
 at org.apache.pinot.shaded.com.google.common.util.concurrent.TrustedListenableFutureTask$TrustedFutureInterruptibleTask.runInterruptibly(TrustedListenableFutureTask.java:131) [pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.shaded.com.google.common.util.concurrent.InterruptibleTask.run(InterruptibleTask.java:75) [pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at org.apache.pinot.shaded.com.google.common.util.concurrent.TrustedListenableFutureTask.run(TrustedListenableFutureTask.java:82) [pinot-all-1.0.0-jar-with-dependencies.jar:1.0.0-b6bdf6c9686b286a149d2d1aea4a385ee98f3e79]
 at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1128) [?:?]
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:628) [?:?]
 at java.lang.Thread.run(Thread.java:829) [?:?]

[gortiz]

Can this be replicated in any quickstart? Can you also specify the Pinot version where the error is present and the types and indexes involved in the query/table?

[gortiz]

It seems the issue was a bit more complex than expected. There is a DISTINCT_COUNT_HLL star-tree index on that column. That precalculation was done with the default log2m. Therefore when a query touches a segment whose data HLL is read from the star-tree index and another where the data has been calculated at runtime with a different log2m, the error is thrown.

I think Pinot should detect the discrepancy in the arguments used and do not use the star-tree index in this case. What do you think @Jackie-Jiang ?

[Jackie-Jiang]

Yes. This requires bigger changes. Basically star-tree should keep the extra arguments stored in the metadata, and match the whole aggregation. Currently it only stores the main column and aggregation type, thus causing this problem.

[yashmayya]

Will be fixed by https://github.com/apache/pinot/pull/13835.