Victoria Markman created DRILL-2967:
---------------------------------------

             Summary: Incompatible types error reported in a "not in" query 
with compatible data types 
                 Key: DRILL-2967
                 URL: https://issues.apache.org/jira/browse/DRILL-2967
             Project: Apache Drill
          Issue Type: Bug
          Components: Query Planning & Optimization
    Affects Versions: 0.9.0
            Reporter: Victoria Markman
            Assignee: Jinfeng Ni


Two tables, parquet files (attached in the bug):
{code}
0: jdbc:drill:schema=dfs> select * from t1;
+------------+------------+------------+
|     a1     |     b1     |     c1     |
+------------+------------+------------+
| 1          | aaaaa      | 2015-01-01 |
| 2          | bbbbb      | 2015-01-02 |
| 3          | ccccc      | 2015-01-03 |
| 4          | null       | 2015-01-04 |
| 5          | eeeee      | 2015-01-05 |
| 6          | fffff      | 2015-01-06 |
| 7          | ggggg      | 2015-01-07 |
| null       | hhhhh      | 2015-01-08 |
| 9          | iiiii      | null       |
| 10         | jjjjj      | 2015-01-10 |
+------------+------------+------------+
10 rows selected (0.119 seconds)

0: jdbc:drill:schema=dfs> select * from t2;
+------------+------------+------------+
|     a2     |     b2     |     c2     |
+------------+------------+------------+
| 0          | zzz        | 2014-12-31 |
| 1          | aaaaa      | 2015-01-01 |
| 2          | bbbbb      | 2015-01-02 |
| 2          | bbbbb      | 2015-01-02 |
| 2          | bbbbb      | 2015-01-02 |
| 3          | ccccc      | 2015-01-03 |
| 4          | ddddd      | 2015-01-04 |
| 5          | eeeee      | 2015-01-05 |
| 6          | fffff      | 2015-01-06 |
| 7          | ggggg      | 2015-01-07 |
| 7          | ggggg      | 2015-01-07 |
| 8          | hhhhh      | 2015-01-08 |
| 9          | iiiii      | 2015-01-09 |
+------------+------------+------------+
13 rows selected (0.116 seconds)
{code}

Disable hash join and set slice_target = 1:

alter session set `planner.enable_hashjoin` = false;
alter session set `planner.slice_target` = 1;

Correct result:
{code}
0: jdbc:drill:schema=dfs> select * from t1 where b1 not in (select b2 from t2);
+------------+------------+------------+
|     a1     |     b1     |     c1     |
+------------+------------+------------+
| 10         | jjjjj      | 2015-01-10 |
+------------+------------+------------+
1 row selected (0.625 seconds)
{code}


Swap tables and you get an error:
{code}
0: jdbc:drill:schema=dfs> select * from t2 where b2 not in (select b1 from t1);
+------------+------------+------------+
|     a1     |     b1     |     c1     |
+------------+------------+------------+
Query failed: SYSTEM ERROR: Join only supports implicit casts between 1. 
Numeric data
 2. Varchar, Varbinary data Left type: INT, Right type: VARCHAR. Add explicit 
casts to avoid this error

Fragment 1:0

[1a83aa50-39aa-452c-91dd-970bf4a8f03d on atsqa4-133.qa.lab:31010]
java.lang.RuntimeException: java.sql.SQLException: Failure while executing 
query.
        at sqlline.SqlLine$IncrementalRows.hasNext(SqlLine.java:2514)
        at sqlline.SqlLine$TableOutputFormat.print(SqlLine.java:2148)
        at sqlline.SqlLine.print(SqlLine.java:1809)
        at sqlline.SqlLine$Commands.execute(SqlLine.java:3766)
        at sqlline.SqlLine$Commands.sql(SqlLine.java:3663)
        at sqlline.SqlLine.dispatch(SqlLine.java:889)
        at sqlline.SqlLine.begin(SqlLine.java:763)
        at sqlline.SqlLine.start(SqlLine.java:498)
        at sqlline.SqlLine.main(SqlLine.java:460)
{code}

Explain plan for the query with an error:
{code}
0: jdbc:drill:schema=dfs> explain plan for select * from t2 where b2 not in 
(select b1 from t1);
+------------+------------+
|    text    |    json    |
+------------+------------+
| 00-00    Screen
00-01      Project(*=[$0])
00-02        UnionExchange
01-01          Project(T27¦¦*=[$0])
01-02            SelectionVectorRemover
01-03              Filter(condition=[NOT(CASE(=($2, 0), false, IS NOT NULL($6), 
true, IS NULL($4), null, <($3, $2), null, false))])
01-04                MergeJoin(condition=[=($4, $5)], joinType=[left])
01-06                  SelectionVectorRemover
01-08                    Sort(sort0=[$4], dir0=[ASC])
01-10                      Project(T27¦¦*=[$0], b2=[$1], $f0=[$2], $f1=[$3], 
b20=[$4])
01-12                        HashToRandomExchange(dist0=[[$4]])
02-01                          UnorderedMuxExchange
04-01                            Project(T27¦¦*=[$0], b2=[$1], $f0=[$2], 
$f1=[$3], b20=[$4], E_X_P_R_H_A_S_H_F_I_E_L_D=[castInt(hash64AsDouble($4))])
04-02                              Project(T27¦¦*=[$0], b2=[$1], $f0=[$2], 
$f1=[$3], b20=[$1])
04-03                                NestedLoopJoin(condition=[true], 
joinType=[inner])
04-05                                  Project(T27¦¦*=[$0], b2=[$1])
04-06                                    Scan(groupscan=[ParquetGroupScan 
[entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/aggregation/t2]], 
selectionRoot=/drill/testdata/aggregation/t2, numFiles=1, columns=[`*`]]])
04-04                                  BroadcastExchange
06-01                                    StreamAgg(group=[{}], 
agg#0=[$SUM0($0)], agg#1=[$SUM0($1)])
06-02                                      UnionExchange
07-01                                        StreamAgg(group=[{}], 
agg#0=[COUNT()], agg#1=[COUNT($0)])
07-02                                          Project(b1=[$0], $f1=[true])
07-03                                            
Scan(groupscan=[ParquetGroupScan [entries=[ReadEntryWithPath 
[path=maprfs:/drill/testdata/aggregation/t1]], 
selectionRoot=/drill/testdata/aggregation/t1, numFiles=1, columns=[`b1`]]])
01-05                  Project(b1=[$0], $f10=[$1])
01-07                    SelectionVectorRemover
01-09                      Sort(sort0=[$0], dir0=[ASC])
01-11                        HashAgg(group=[{0}], agg#0=[MIN($1)])
01-13                          Project(b1=[$0], $f1=[$1])
01-14                            HashToRandomExchange(dist0=[[$0]])
03-01                              UnorderedMuxExchange
05-01                                Project(b1=[$0], $f1=[$1], 
E_X_P_R_H_A_S_H_F_I_E_L_D=[castInt(hash64AsDouble($0))])
05-02                                  HashAgg(group=[{0}], agg#0=[MIN($1)])
05-03                                    Project(b1=[$0], $f1=[true])
05-04                                      Scan(groupscan=[ParquetGroupScan 
[entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/aggregation/t1]], 
selectionRoot=/drill/testdata/aggregation/t1, numFiles=1, columns=[`b1`]]])
{code}

Correct result and correct plan with hash join distributed plan ( 
planner.slice_target = 1)

alter session set `planner.enable_hashjoin` = true;
alter session set `planner.slice_target` = 1;

{code}
0: jdbc:drill:schema=dfs> select * from t2 where b2 not in (select b1 from t1);
+------------+------------+------------+
|     a2     |     b2     |     c2     |
+------------+------------+------------+
+------------+------------+------------+
No rows selected (0.458 seconds)

0: jdbc:drill:schema=dfs> explain plan for select * from t2 where b2 not in 
(select b1 from t1);
+------------+------------+
|    text    |    json    |
+------------+------------+
| 00-00    Screen
00-01      Project(*=[$0])
00-02        Project(T25¦¦*=[$0])
00-03          SelectionVectorRemover
00-04            Filter(condition=[NOT(CASE(=($2, 0), false, IS NOT NULL($6), 
true, IS NULL($4), null, <($3, $2), null, false))])
00-05              HashJoin(condition=[=($4, $5)], joinType=[left])
00-07                Project(T25¦¦*=[$0], b2=[$1], $f0=[$2], $f1=[$3], b20=[$1])
00-09                  NestedLoopJoin(condition=[true], joinType=[inner])
00-11                    Project(T25¦¦*=[$0], b2=[$1])
00-12                      Scan(groupscan=[ParquetGroupScan 
[entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/aggregation/t2]], 
selectionRoot=/drill/testdata/aggregation/t2, numFiles=1, columns=[`*`]]])
00-10                    BroadcastExchange
01-01                      StreamAgg(group=[{}], agg#0=[$SUM0($0)], 
agg#1=[$SUM0($1)])
01-02                        UnionExchange
03-01                          StreamAgg(group=[{}], agg#0=[COUNT()], 
agg#1=[COUNT($0)])
03-02                            Project(b1=[$0], $f1=[true])
03-03                              Scan(groupscan=[ParquetGroupScan 
[entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/aggregation/t1]], 
selectionRoot=/drill/testdata/aggregation/t1, numFiles=1, columns=[`b1`]]])
00-06                Project(b1=[$0], $f10=[$1])
00-08                  BroadcastExchange
02-01                    HashAgg(group=[{0}], agg#0=[MIN($1)])
02-02                      Project(b1=[$0], $f1=[$1])
02-03                        HashToRandomExchange(dist0=[[$0]])
04-01                          UnorderedMuxExchange
05-01                            Project(b1=[$0], $f1=[$1], 
E_X_P_R_H_A_S_H_F_I_E_L_D=[castInt(hash64AsDouble($0))])
05-02                              HashAgg(group=[{0}], agg#0=[MIN($1)])
05-03                                Project(b1=[$0], $f1=[true])
05-04                                  Scan(groupscan=[ParquetGroupScan 
[entries=[ReadEntryWithPath [path=maprfs:/drill/testdata/aggregation/t1]], 
selectionRoot=/drill/testdata/aggregation/t1, numFiles=1, columns=[`b1`]]])
{code}

Same error with the columns of date, time and timestamp types.




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to