Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Skip to content

Commit 77d4164

Browse files
ildusildus
committed
Merge branch 'rel_future_beta' of github.com:postgrespro/pg_pathman into rel_future_beta
2 parents a8c78ff + 11f6d4a commit 77d4164

22 files changed

+754
-267
lines changed

‎Makefile

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -35,6 +35,7 @@ REGRESS = pathman_basic \
3535
pathman_inserts \
3636
pathman_interval \
3737
pathman_join_clause \
38+
pathman_lateral \
3839
pathman_only \
3940
pathman_permissions \
4041
pathman_rowmarks \

‎README.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -68,7 +68,7 @@ Modify the **`shared_preload_libraries`** parameter in `postgresql.conf` as foll
6868
```
6969
shared_preload_libraries = 'pg_pathman'
7070
```
71-
> **Important:** `pg_pathman` may have conflicts with some other extensions which uses the same hook functions. For example, `pg_pathman` uses `ProcessUtility_hook` hook to handle COPY queries for partitioned tables. And it could sometimes interfere with `pg_stat_statements` extension which uses the same hook. In this case try to list libraries in certain order: `shared_preload_libraries = 'pg_pathman, pg_stat_statements'`
71+
> **Important:** `pg_pathman` may cause conflicts with some other extensions that use the same hook functions. For example, `pg_pathman` uses `ProcessUtility_hook` to handle COPY queries for partitioned tables, which means it may interfere with `pg_stat_statements` from time to time. In this case, try listing libraries in certain order: `shared_preload_libraries = 'pg_stat_statements, pg_pathman'`.
7272
7373
It is essential to restart the PostgreSQL instance. After that, execute the following query in psql:
7474
```plpgsql

‎expected/pathman_join_clause.out

Lines changed: 16 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -6,40 +6,41 @@ CREATE SCHEMA test;
66
* Test push down a join clause into child nodes of append
77
*/
88
/* create test tables */
9-
CREATE TABLE fk (
9+
CREATE TABLE test.fk (
1010
id1 INT NOT NULL,
1111
id2 INT NOT NULL,
1212
start_key INT,
1313
end_key INT,
1414
PRIMARY KEY (id1, id2));
15-
CREATE TABLE mytbl (
15+
CREATE TABLE test.mytbl (
1616
id1 INT NOT NULL,
1717
id2 INT NOT NULL,
1818
key INT NOT NULL,
19-
CONSTRAINT fk_fk FOREIGN KEY (id1, id2) REFERENCES fk(id1, id2),
19+
CONSTRAINT fk_fk FOREIGN KEY (id1, id2) REFERENCES test.fk(id1, id2),
2020
PRIMARY KEY (id1, key));
21-
SELECT pathman.create_hash_partitions('mytbl', 'id1', 8);
21+
SELECT pathman.create_hash_partitions('test.mytbl', 'id1', 8);
2222
create_hash_partitions
2323
------------------------
2424
8
2525
(1 row)
2626

2727
/* ...fill out with test data */
28-
INSERT INTO fk VALUES (1, 1);
29-
INSERT INTO mytbl VALUES (1, 1, 5), (1,1,6);
28+
INSERT INTO test.fk VALUES (1, 1);
29+
INSERT INTO test.mytbl VALUES (1, 1, 5), (1,1,6);
3030
/* gather statistics on test tables to have deterministic plans */
31-
ANALYZE fk;
32-
ANALYZE mytbl;
31+
ANALYZE test.fk;
32+
ANALYZE test.mytbl;
3333
/* run test queries */
3434
EXPLAIN (COSTS OFF) /* test plan */
3535
SELECT m.tableoid::regclass, id1, id2, key, start_key, end_key
36-
FROM mytbl m JOIN fk USING(id1, id2)
36+
FROM test.mytbl m JOIN test.fk USING(id1, id2)
3737
WHERE NOT key <@ int4range(6, end_key);
3838
QUERY PLAN
3939
------------------------------------------------------------------------------------
4040
Nested Loop
4141
-> Seq Scan on fk
4242
-> Custom Scan (RuntimeAppend)
43+
Prune by: (fk.id1 = m.id1)
4344
-> Bitmap Heap Scan on mytbl_0 m
4445
Recheck Cond: (id1 = fk.id1)
4546
Filter: ((fk.id2 = id2) AND (NOT (key <@ int4range(6, fk.end_key))))
@@ -80,18 +81,18 @@ SELECT m.tableoid::regclass, id1, id2, key, start_key, end_key
8081
Filter: ((fk.id2 = id2) AND (NOT (key <@ int4range(6, fk.end_key))))
8182
-> Bitmap Index Scan on mytbl_7_pkey
8283
Index Cond: (id1 = fk.id1)
83-
(43 rows)
84+
(44 rows)
8485

8586
/* test joint data */
8687
SELECT m.tableoid::regclass, id1, id2, key, start_key, end_key
87-
FROM mytbl m JOIN fk USING(id1, id2)
88+
FROM test.mytbl m JOIN test.fk USING(id1, id2)
8889
WHERE NOT key <@ int4range(6, end_key);
89-
tableoid | id1 | id2 | key | start_key | end_key
90-
----------+-----+-----+-----+-----------+---------
91-
mytbl_6 | 1 | 1 | 5 | |
90+
tableoid | id1 | id2 | key | start_key | end_key
91+
--------------+-----+-----+-----+-----------+---------
92+
test.mytbl_6 | 1 | 1 | 5 | |
9293
(1 row)
9394

9495
DROP SCHEMA test CASCADE;
96+
NOTICE: drop cascades to 10 other objects
9597
DROP EXTENSION pg_pathman CASCADE;
96-
NOTICE: drop cascades to 8 other objects
9798
DROP SCHEMA pathman CASCADE;

‎expected/pathman_lateral.out

Lines changed: 118 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,118 @@
1+
\set VERBOSITY terse
2+
SET search_path = 'public';
3+
CREATE EXTENSION pg_pathman;
4+
CREATE SCHEMA test_lateral;
5+
/* create table partitioned by HASH */
6+
create table test_lateral.data(id int8 not null);
7+
select create_hash_partitions('test_lateral.data', 'id', 10);
8+
create_hash_partitions
9+
------------------------
10+
10
11+
(1 row)
12+
13+
insert into test_lateral.data select generate_series(1, 10000);
14+
VACUUM ANALYZE;
15+
set enable_hashjoin = off;
16+
set enable_mergejoin = off;
17+
/* all credits go to Ivan Frolkov */
18+
explain (costs off)
19+
select * from
20+
test_lateral.data as t1,
21+
lateral(select * from test_lateral.data as t2 where t2.id > t1.id) t2,
22+
lateral(select * from test_lateral.data as t3 where t3.id = t2.id + t1.id) t3
23+
where t1.id between 1 and 100 and
24+
t2.id between 2 and 299 and
25+
t1.id > t2.id and
26+
exists(select * from test_lateral.data t
27+
where t1.id = t2.id and t.id = t3.id);
28+
QUERY PLAN
29+
--------------------------------------------------------------------------------------------
30+
Nested Loop Semi Join
31+
-> Nested Loop
32+
Join Filter: ((t2.id + t1.id) = t3.id)
33+
-> Append
34+
-> Seq Scan on data_0 t3
35+
-> Seq Scan on data_1 t3_1
36+
-> Seq Scan on data_2 t3_2
37+
-> Seq Scan on data_3 t3_3
38+
-> Seq Scan on data_4 t3_4
39+
-> Seq Scan on data_5 t3_5
40+
-> Seq Scan on data_6 t3_6
41+
-> Seq Scan on data_7 t3_7
42+
-> Seq Scan on data_8 t3_8
43+
-> Seq Scan on data_9 t3_9
44+
-> Materialize
45+
-> Nested Loop
46+
Join Filter: ((t2.id > t1.id) AND (t1.id > t2.id) AND (t1.id = t2.id))
47+
-> Append
48+
-> Seq Scan on data_0 t2
49+
Filter: ((id >= 2) AND (id <= 299))
50+
-> Seq Scan on data_1 t2_1
51+
Filter: ((id >= 2) AND (id <= 299))
52+
-> Seq Scan on data_2 t2_2
53+
Filter: ((id >= 2) AND (id <= 299))
54+
-> Seq Scan on data_3 t2_3
55+
Filter: ((id >= 2) AND (id <= 299))
56+
-> Seq Scan on data_4 t2_4
57+
Filter: ((id >= 2) AND (id <= 299))
58+
-> Seq Scan on data_5 t2_5
59+
Filter: ((id >= 2) AND (id <= 299))
60+
-> Seq Scan on data_6 t2_6
61+
Filter: ((id >= 2) AND (id <= 299))
62+
-> Seq Scan on data_7 t2_7
63+
Filter: ((id >= 2) AND (id <= 299))
64+
-> Seq Scan on data_8 t2_8
65+
Filter: ((id >= 2) AND (id <= 299))
66+
-> Seq Scan on data_9 t2_9
67+
Filter: ((id >= 2) AND (id <= 299))
68+
-> Materialize
69+
-> Append
70+
-> Seq Scan on data_0 t1
71+
Filter: ((id >= 1) AND (id <= 100))
72+
-> Seq Scan on data_1 t1_1
73+
Filter: ((id >= 1) AND (id <= 100))
74+
-> Seq Scan on data_2 t1_2
75+
Filter: ((id >= 1) AND (id <= 100))
76+
-> Seq Scan on data_3 t1_3
77+
Filter: ((id >= 1) AND (id <= 100))
78+
-> Seq Scan on data_4 t1_4
79+
Filter: ((id >= 1) AND (id <= 100))
80+
-> Seq Scan on data_5 t1_5
81+
Filter: ((id >= 1) AND (id <= 100))
82+
-> Seq Scan on data_6 t1_6
83+
Filter: ((id >= 1) AND (id <= 100))
84+
-> Seq Scan on data_7 t1_7
85+
Filter: ((id >= 1) AND (id <= 100))
86+
-> Seq Scan on data_8 t1_8
87+
Filter: ((id >= 1) AND (id <= 100))
88+
-> Seq Scan on data_9 t1_9
89+
Filter: ((id >= 1) AND (id <= 100))
90+
-> Custom Scan (RuntimeAppend)
91+
Prune by: (t3.id = t.id)
92+
-> Seq Scan on data_0 t
93+
Filter: (t3.id = id)
94+
-> Seq Scan on data_1 t
95+
Filter: (t3.id = id)
96+
-> Seq Scan on data_2 t
97+
Filter: (t3.id = id)
98+
-> Seq Scan on data_3 t
99+
Filter: (t3.id = id)
100+
-> Seq Scan on data_4 t
101+
Filter: (t3.id = id)
102+
-> Seq Scan on data_5 t
103+
Filter: (t3.id = id)
104+
-> Seq Scan on data_6 t
105+
Filter: (t3.id = id)
106+
-> Seq Scan on data_7 t
107+
Filter: (t3.id = id)
108+
-> Seq Scan on data_8 t
109+
Filter: (t3.id = id)
110+
-> Seq Scan on data_9 t
111+
Filter: (t3.id = id)
112+
(82 rows)
113+
114+
set enable_hashjoin = on;
115+
set enable_mergejoin = on;
116+
DROP SCHEMA test_lateral CASCADE;
117+
NOTICE: drop cascades to 11 other objects
118+
DROP EXTENSION pg_pathman;

‎expected/pathman_only.out

Lines changed: 14 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -178,17 +178,28 @@ SELECT * FROM test_only.from_only_test JOIN q1 USING(val);
178178
-> Seq Scan on from_only_test from_only_test_1
179179
-> CTE Scan on q1
180180
-> Custom Scan (RuntimeAppend)
181+
Prune by: (q1.val = from_only_test.val)
181182
-> Seq Scan on from_only_test_1 from_only_test
183+
Filter: (q1.val = val)
182184
-> Seq Scan on from_only_test_2 from_only_test
185+
Filter: (q1.val = val)
183186
-> Seq Scan on from_only_test_3 from_only_test
187+
Filter: (q1.val = val)
184188
-> Seq Scan on from_only_test_4 from_only_test
189+
Filter: (q1.val = val)
185190
-> Seq Scan on from_only_test_5 from_only_test
191+
Filter: (q1.val = val)
186192
-> Seq Scan on from_only_test_6 from_only_test
193+
Filter: (q1.val = val)
187194
-> Seq Scan on from_only_test_7 from_only_test
195+
Filter: (q1.val = val)
188196
-> Seq Scan on from_only_test_8 from_only_test
197+
Filter: (q1.val = val)
189198
-> Seq Scan on from_only_test_9 from_only_test
199+
Filter: (q1.val = val)
190200
-> Seq Scan on from_only_test_10 from_only_test
191-
(15 rows)
201+
Filter: (q1.val = val)
202+
(26 rows)
192203

193204
/* should be OK */
194205
EXPLAIN (COSTS OFF)
@@ -199,6 +210,7 @@ WHERE val = (SELECT val FROM ONLY test_only.from_only_test
199210
QUERY PLAN
200211
-----------------------------------------------------------------
201212
Custom Scan (RuntimeAppend)
213+
Prune by: (from_only_test.val = $0)
202214
InitPlan 1 (returns $0)
203215
-> Limit
204216
-> Sort
@@ -224,7 +236,7 @@ WHERE val = (SELECT val FROM ONLY test_only.from_only_test
224236
Filter: (val = $0)
225237
-> Seq Scan on from_only_test_10 from_only_test
226238
Filter: (val = $0)
227-
(26 rows)
239+
(27 rows)
228240

229241
DROP SCHEMA test_only CASCADE;
230242
NOTICE: drop cascades to 12 other objects

‎expected/pathman_rowmarks.out

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -100,6 +100,7 @@ FOR SHARE;
100100
-> Seq Scan on first_3
101101
-> Seq Scan on first_4
102102
-> Custom Scan (RuntimeAppend)
103+
Prune by: (first.id = $1)
103104
-> Seq Scan on first_0 first
104105
Filter: (id = $1)
105106
-> Seq Scan on first_1 first
@@ -110,7 +111,7 @@ FOR SHARE;
110111
Filter: (id = $1)
111112
-> Seq Scan on first_4 first
112113
Filter: (id = $1)
113-
(23 rows)
114+
(24 rows)
114115

115116
/* A little harder (execution) */
116117
SELECT * FROM rowmarks.first
@@ -142,6 +143,7 @@ FOR SHARE;
142143
Sort Key: second.id
143144
-> Seq Scan on second
144145
-> Custom Scan (RuntimeAppend)
146+
Prune by: (first.id = $1)
145147
-> Seq Scan on first_0 first
146148
Filter: (id = $1)
147149
-> Seq Scan on first_1 first
@@ -152,7 +154,7 @@ FOR SHARE;
152154
Filter: (id = $1)
153155
-> Seq Scan on first_4 first
154156
Filter: (id = $1)
155-
(18 rows)
157+
(19 rows)
156158

157159
/* Two tables (execution) */
158160
SELECT * FROM rowmarks.first

0 commit comments

Comments
 (0)