Re: variation of row_number with parallel - Mailing list pgsql-hackers

Hi,

I have observed row_number() is giving different results when query executed in parallel. is this expected w.r.t parallel execution.

CREATE TABLE tbl1 (c1 INT) partition by list (c1);

CREATE TABLE tbl1_p1 partition of tbl1 FOR VALUES IN (10);
CREATE TABLE tbl1_p2 partition of tbl1 FOR VALUES IN (20);
CREATE TABLE tbl1_p3 partition of tbl1 FOR VALUES IN (30);

CREATE TABLE tbl2 (c1 INT, c2 INT,c3 INT) partition by list (c1);
CREATE TABLE tbl2_p1 partition of tbl2 FOR VALUES IN (1);
CREATE TABLE tbl2_p2 partition of tbl2 FOR VALUES IN (2);
CREATE TABLE tbl2_p3 partition of tbl2 FOR VALUES IN (3);
CREATE TABLE tbl2_p4 partition of tbl2 FOR VALUES IN (4);
CREATE TABLE tbl2_p5 partition of tbl2 FOR VALUES IN (5);

INSERT INTO tbl1 VALUES (10),(20),(30);

INSERT INTO tbl2 VALUES (1,100,20),(2,200,10),(3,100,20),(4,100,30),(5,100,10);

postgres=# explain select e.c2, row_number() over () from tbl1 d, tbl2 e where d.c1=e.c3;
                                      QUERY PLAN                                      
---------------------------------------------------------------------------------------
 WindowAgg  (cost=1520.35..12287.73 rows=390150 width=12)
   ->  Merge Join  (cost=1520.35..7410.85 rows=390150 width=4)
         Merge Cond: (d.c1 = e.c3)
         ->  Sort  (cost=638.22..657.35 rows=7650 width=4)
               Sort Key: d.c1
               ->  Append  (cost=0.00..144.75 rows=7650 width=4)
                     ->  Seq Scan on tbl1_p1 d_1  (cost=0.00..35.50 rows=2550 width=4)
                     ->  Seq Scan on tbl1_p2 d_2  (cost=0.00..35.50 rows=2550 width=4)
                     ->  Seq Scan on tbl1_p3 d_3  (cost=0.00..35.50 rows=2550 width=4)
         ->  Sort  (cost=882.13..907.63 rows=10200 width=8)
               Sort Key: e.c3
               ->  Append  (cost=0.00..203.00 rows=10200 width=8)
                     ->  Seq Scan on tbl2_p1 e_1  (cost=0.00..30.40 rows=2040 width=8)
                     ->  Seq Scan on tbl2_p2 e_2  (cost=0.00..30.40 rows=2040 width=8)
                     ->  Seq Scan on tbl2_p3 e_3  (cost=0.00..30.40 rows=2040 width=8)
                     ->  Seq Scan on tbl2_p4 e_4  (cost=0.00..30.40 rows=2040 width=8)
                     ->  Seq Scan on tbl2_p5 e_5  (cost=0.00..30.40 rows=2040 width=8)
(17 rows)

postgres=# select e.c2, row_number() over () from tbl1 d, tbl2 e where d.c1=e.c3;
 c2  | row_number
-----+------------
 200 |          1
 100 |          2
 100 |          3
 100 |          4
 100 |          5
(5 rows)

postgres=#
postgres=# set parallel_setup_cost = 0;
SET
postgres=# set parallel_tuple_cost = 0;
SET
postgres=#
postgres=# explain select e.c2, row_number() over () from tbl1 d, tbl2 e where d.c1=e.c3;
                                              QUERY PLAN                                              
------------------------------------------------------------------------------------------------------
 WindowAgg  (cost=130.75..7521.21 rows=390150 width=12)
   ->  Gather  (cost=130.75..2644.34 rows=390150 width=4)
         Workers Planned: 2
         ->  Parallel Hash Join  (cost=130.75..2644.34 rows=162562 width=4)
               Hash Cond: (e.c3 = d.c1)
               ->  Parallel Append  (cost=0.00..131.25 rows=4250 width=8)
                     ->  Parallel Seq Scan on tbl2_p1 e_1  (cost=0.00..22.00 rows=1200 width=8)
                     ->  Parallel Seq Scan on tbl2_p2 e_2  (cost=0.00..22.00 rows=1200 width=8)
                     ->  Parallel Seq Scan on tbl2_p3 e_3  (cost=0.00..22.00 rows=1200 width=8)
                     ->  Parallel Seq Scan on tbl2_p4 e_4  (cost=0.00..22.00 rows=1200 width=8)
                     ->  Parallel Seq Scan on tbl2_p5 e_5  (cost=0.00..22.00 rows=1200 width=8)
               ->  Parallel Hash  (cost=90.93..90.93 rows=3186 width=4)
                     ->  Parallel Append  (cost=0.00..90.93 rows=3186 width=4)
                           ->  Parallel Seq Scan on tbl1_p1 d_1  (cost=0.00..25.00 rows=1500 width=4)
                           ->  Parallel Seq Scan on tbl1_p2 d_2  (cost=0.00..25.00 rows=1500 width=4)
                           ->  Parallel Seq Scan on tbl1_p3 d_3  (cost=0.00..25.00 rows=1500 width=4)
(16 rows)

postgres=# select e.c2, row_number() over () from tbl1 d, tbl2 e where d.c1=e.c3;
 c2  | row_number
-----+------------
 100 |          1
 100 |          2
 100 |          3
 200 |          4
 100 |          5
(5 rows)

Thanks & Regards,

Rajkumar Raghuwanshi