[IMPALA-1737] invalid tuple_idx when combining INSERT INTO with analytic subquery

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: Impala 2.1
• Fix Version/s: Impala 2.2.7, Impala 2.3.0
• Environment: RHEL 6.5 and CENTOS 6.4
• Target Version: Impala 2.2.7, Impala 2.3.0
• Assignee: Alexander Behm
• Reporter: Tristan Stevens
• Created: 04/Feb/15 11:30
• Updated: 20/Dec/15 00:05
• Resolved: 03/Aug/15 19:35

Description

The following query is producing invalid tuple_idx when executing. If the INSRET INTO is changed into a CREATE TABLE AS statement it executes fine.

CREATE TABLE test1  (a bigint, b string, c string)
PARTITIONED BY (d int);

CREATE TABLE test2  (a bigint, b string, c string)
PARTITIONED BY (d int);

INSERT INTO test2
PARTITION(d)
SELECT 
      SUM(a) OVER (
            PARTITION BY b ORDER BY c ROWS UNBOUNDED PRECEDING
            )
       AS a
   ,b
   ,c
   ,d
FROM test1;

Workaround
Force expression materialization via a UNION ALL as follows:

INSERT INTO test2
PARTITION(d)
SELECT 
SUM(a) OVER (
PARTITION BY b ORDER BY c ROWS UNBOUNDED PRECEDING
)
AS a
,b
,c
,d
FROM test1
UNION ALL <--- Adding UNION ALL with the SELECT statement below is the workaround
SELECT NULL, NULL, NULL, NULL from test1 WHERE false;