[HIVE-15872] The PERCENTILE_APPROX UDAF does not work with empty set

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.1.2, 2.2.0
• Component/s: UDF
• Flags: Patch
• Assignee: Chaozhong Yang
• Reporter: Chaozhong Yang
• Created: 10/Feb/17 02:20
• Updated: 25/Jul/17 20:05
• Resolved: 14/Feb/17 00:19

Description

1. Original SQL:

select
    percentile_approx(
        column0,
        array(0.50, 0.70, 0.90, 0.95, 0.99)
    )
from
    my_table
where
    date = '20170207'
    and column1 = 'value1'
    and column2 = 'value2'
    and column3 = 'value3'
    and column4 = 'value4'
    and column5 = 'value5'

2. Exception StackTrace:

Error: java.lang.RuntimeException: org.apache.hadoop.hive.ql.metadata.HiveException: Hive Runtime Error while processing row (tag=0) {"key":{},"value":{"_col0":[0.0,10000.0]}} at org.apache.hadoop.hive.ql.exec.mr.ExecReducer.reduce(ExecReducer.java:256) at org.apache.hadoop.mapred.ReduceTask.runOldReducer(ReduceTask.java:453) at org.apache.hadoop.mapred.ReduceTask.run(ReduceTask.java:401) at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:163) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:422) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1671) at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:158) Caused by: org.apache.hadoop.hive.ql.metadata.HiveException: Hive Runtime Error while processing row (tag=0) {"key":{},"value":{"_col0":[0.0,10000.0]}} at org.apache.hadoop.hive.ql.exec.mr.ExecReducer.reduce(ExecReducer.java:244) ... 7 more Caused by: org.apache.hadoop.hive.ql.metadata.HiveException: java.lang.IndexOutOfBoundsException: Index: 2, Size: 2 at org.apache.hadoop.hive.ql.exec.GroupByOperator.process(GroupByOperator.java:766) at org.apache.hadoop.hive.ql.exec.mr.ExecReducer.reduce(ExecReducer.java:235) ... 7 more Caused by: java.lang.IndexOutOfBoundsException: Index: 2, Size: 2 at java.util.ArrayList.rangeCheck(ArrayList.java:653) at java.util.ArrayList.get(ArrayList.java:429) at org.apache.hadoop.hive.ql.udf.generic.NumericHistogram.merge(NumericHistogram.java:134) at org.apache.hadoop.hive.ql.udf.generic.GenericUDAFPercentileApprox$GenericUDAFPercentileApproxEvaluator.merge(GenericUDAFPercentileApprox.java:318) at org.apache.hadoop.hive.ql.udf.generic.GenericUDAFEvaluator.aggregate(GenericUDAFEvaluator.java:188) at org.apache.hadoop.hive.ql.exec.GroupByOperator.updateAggregations(GroupByOperator.java:612) at org.apache.hadoop.hive.ql.exec.GroupByOperator.processAggr(GroupByOperator.java:851) at org.apache.hadoop.hive.ql.exec.GroupByOperator.processKey(GroupByOperator.java:695) at org.apache.hadoop.hive.ql.exec.GroupByOperator.process(GroupByOperator.java:761) ... 8 more

3. review data:

select
    column0
from
    my_table
where
    date = '20170207'
    and column1 = 'value1'
    and column2 = 'value2'
    and column3 = 'value3'
    and column4 = 'value4'
    and column5 = 'value5'

After run this sql, we found the result is NULL.

4. what's the meaning of [0.0, 10000.0] in stacktrace?

In GenericUDAFPercentileApproxEvaluator, the method `merge` should process an ArrayList which name is partialHistogram. Normally, the basic structure of partialHistogram is [npercentiles, percentile0, percentile1..., nbins, bin0.x, bin0.y, bin1.x, bin1.y,...]. However, if we process NULL(empty set) column values, the partialHistoram will only contains [npercentiles(0), nbins(10000)]. That's the reason why the stacktrace shows a strange row data: {"key":{},"value":{"_col0":[0.0,10000.0]}}

Before we call histogram#merge (on-line hisgoram algorithm from paper: http://www.jmlr.org/papers/volume11/ben-haim10a/ben-haim10a.pdf ), the partialHistogram should remove elements which store percentiles like `partialHistogram.subList(0, nquantiles+1).clear();`. In the case of empty set, GenericUDAFPercentileApproxEvaluator will not remove percentiles. Consequently, NumericHistogram will merge a list which contains only 2 elements([0, 10000.0]) and throws IndexOutOfBoundsException.