[HIVE-12662] StackOverflowError in HiveSortJoinReduceRule when limit=0

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.0, 2.1.0
• Fix Version/s: 2.0.0
• Assignee: Jesús Camacho Rodríguez
• Reporter: Jesús Camacho Rodríguez
• Created: 12/Dec/15 10:46
• Updated: 27/Feb/24 22:23
• Resolved: 13/Dec/15 11:44

Description

L96 of HiveSortJoinReduceRule, you will see

    // Finally, if we do not reduce the input size, we bail out
    if (RexLiteral.intValue(sortLimit.fetch)
            >= RelMetadataQuery.getRowCount(reducedInput)) {
      return false;
    }

It is using “ RelMetadataQuery.getRowCount” which is always at least 1. This is the problem that we resolved in CALCITE-987.

To confirm this, I just run the q file :

set hive.mapred.mode=nonstrict;
set hive.optimize.limitjointranspose=true;
set hive.optimize.limitjointranspose.reductionpercentage=1f;
set hive.optimize.limitjointranspose.reductiontuples=0;

explain
select *
from src src1 right outer join (
  select *
  from src src2 left outer join src src3
  on src2.value = src3.value) src2
on src1.key = src2.key
limit 0;

And I got

2015-12-11T10:21:04,435 ERROR [c1efb099-f900-46dc-9f74-97af0944a99d main[]]: parse.CalcitePlanner (CalcitePlanner.java:genOPTree(301)) - CBO failed, skipping CBO.
java.lang.RuntimeException: java.lang.StackOverflowError
        at org.apache.hadoop.hive.ql.parse.CalcitePlanner.rethrowCalciteException(CalcitePlanner.java:749) ~[hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.CalcitePlanner.getOptimizedAST(CalcitePlanner.java:645) ~[hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.CalcitePlanner.genOPTree(CalcitePlanner.java:264) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.SemanticAnalyzer.analyzeInternal(SemanticAnalyzer.java:10076) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.CalcitePlanner.analyzeInternal(CalcitePlanner.java:223) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer.analyze(BaseSemanticAnalyzer.java:237) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.ExplainSemanticAnalyzer.analyzeInternal(ExplainSemanticAnalyzer.java:74) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer.analyze(BaseSemanticAnalyzer.java:237) [hive-exec-2.1.0-SNAPSHOT.jar:2.1.0-SNAPSHOT]
        at org.apache.hadoop.hive.ql.Driver.compile(Driver.java:456) [hive-exec-2.1.0-SNAPSHOT.jar:?]
        at org.apache.hadoop.hive.ql.Driver.compile(Driver.java:310) [hive-exec-2.1.0-SNAPSHOT.jar:?]
        at org.apache.hadoop.hive.ql.Driver.compileInternal(Driver.java:1138) [hive-exec-2.1.0-SNAPSHOT.jar:?]
        at org.apache.hadoop.hive.ql.Driver.runInternal(Driver.java:1187) [hive-exec-2.1.0-SNAPSHOT.jar:?]
        at org.apache.hadoop.hive.ql.Driver.run(Driver.java:1063) [hive-exec-2.1.0-SNAPSHOT.jar:?]

via pxiong

Issue Links

• is broken by: HIVE-11684 Implement limit pushdown through outer join in CBO