blob: feade3ed93df559f01f1f9fceb8a7c579d0ae3e2 [file] [log] [blame]
-- ALTER TABLE ... SET DISTRIBUTED BY
-- This is the main interface for system expansion
\set DATA values(1, 2), (2, 3), (3, 4)
-- Basic sanity tests
set optimizer_print_missing_stats = off;
create table atsdb (i int, j text) distributed by (i);
insert into atsdb :DATA;
analyze atsdb;
-- should fail
alter table atsdb set distributed by ();
alter table atsdb set distributed by (m);
alter table atsdb set distributed by (i, i);
alter table atsdb set distributed by (i, m);
alter table atsdb set distributed by (i);
-- should work
alter table atsdb set distributed randomly;
select localoid::regclass, distkey from gp_distribution_policy where localoid = 'atsdb'::regclass;
-- not possible to correctly verify random distribution
alter table atsdb set distributed by (j);
select localoid::regclass, distkey from gp_distribution_policy where localoid = 'atsdb'::regclass;
-- verify that the data is correctly redistributed by building a fresh
-- table with the same policy
create table ats_test (i int, j text) distributed by (j);
insert into ats_test :DATA;
select gp_segment_id, * from ats_test except
select gp_segment_id, * from atsdb;
drop table ats_test;
alter table atsdb set distributed by (i, j);
select localoid::regclass, distkey from gp_distribution_policy where localoid = 'atsdb'::regclass;
-- verify
create table ats_test (i int, j text) distributed by (i, j);
insert into ats_test :DATA;
select gp_segment_id, * from ats_test except
select gp_segment_id, * from atsdb;
drop table ats_test;
alter table atsdb set distributed by (j, i);
select localoid::regclass, distkey from gp_distribution_policy where localoid = 'atsdb'::regclass;
-- verify
create table ats_test (i int, j text) distributed by (j, i);
insert into ats_test :DATA;
select gp_segment_id, * from ats_test except
select gp_segment_id, * from atsdb;
drop table ats_test;
-- Now make sure indexes work.
create index atsdb_i_idx on atsdb(i);
set enable_seqscan to off;
explain select * from atsdb where i = 1;
select * from atsdb where i = 1;
alter table atsdb set distributed by (i);
explain select * from atsdb where i = 1;
select * from atsdb where i = 1;
drop table atsdb;
-- Now try AO
create table atsdb_ao (i int, j text) with (appendonly=true) distributed by (i);
insert into atsdb_ao select i, (i+1)::text from generate_series(1, 100) i;
insert into atsdb_ao select i, (i+1)::text from generate_series(1, 100) i;
-- check that we're an AO table
select count(*) from pg_appendonly where relid='atsdb_ao'::regclass;
select count(*) from atsdb_ao;
alter table atsdb_ao set distributed by (j);
-- Still AO?
select count(*) from pg_appendonly where relid='atsdb_ao'::regclass;
select count(*) from atsdb_ao;
-- check alter, vacuum analyze, and then alter
delete from atsdb_ao where i = any(array(select generate_series(1,90)));
vacuum analyze atsdb_ao;
alter table atsdb_ao set distributed randomly;
select count(*) from atsdb_ao;
drop table atsdb_ao;
-- Can't redistribute system catalogs
alter table pg_class set distributed by (relname);
alter table pg_class set with(appendonly = true);
alter table pg_class set with(appendonly = true);
-- WITH clause
create table atsdb (i int, j text) distributed by (j);
insert into atsdb select i, i::text from generate_series(1, 10) i;
alter table atsdb set with(appendonly = true);
select relname, segrelid != 0, reloptions from pg_class, pg_appendonly where pg_class.oid =
'atsdb'::regclass and relid = pg_class.oid;
select * from atsdb;
drop table atsdb;
create view distcheck as select relname as rel, attname from
gp_distribution_policy g, pg_attribute p, pg_class c
where g.localoid = p.attrelid and attnum = any(g.distkey) and
c.oid = p.attrelid;
-- dropped columns
create table atsdb (i int, j int, t text, n numeric) distributed by (j);
insert into atsdb select i, i+1, i+2, i+3 from generate_series(1, 100) i;
alter table atsdb drop column i;
select * from atsdb;
alter table atsdb set distributed by (t);
select * from distcheck where rel = 'atsdb';
alter table atsdb drop column n;
alter table atsdb set with(appendonly = true, compresslevel = 3);
select relname, segrelid != 0, reloptions from pg_class, pg_appendonly where pg_class.oid =
'atsdb'::regclass and relid = pg_class.oid;
select * from distcheck where rel = 'atsdb';
select * from atsdb;
alter table atsdb set distributed by (j);
select * from distcheck where rel = 'atsdb';
select relname, segrelid != 0, reloptions from pg_class, pg_appendonly where pg_class.oid =
'atsdb'::regclass and relid = pg_class.oid;
select * from atsdb;
-- validate parameters
alter table atsdb set with (appendonly = ff);
alter table atsdb set with (reorganize = true);
alter table atsdb set with (fgdfgef = asds);
alter table atsdb set with(reorganize = true, reorganize = false) distributed
randomly;
drop table atsdb;
-- check distribution after dropping distribution key column.
create table atsdb (i int, j int, t text, n numeric) distributed by (i, j);
insert into atsdb select i, i+1, i+2, i+3 from generate_series(1, 20) i;
alter table atsdb drop column i;
select * from atsdb;
select * from distcheck where rel = 'atsdb';
drop table atsdb;
-- check DROP TYPE..CASCADE updates distribution policy to random if
-- any a dist key column is dropped for multi key distribution key columns
create domain int_new as int;
create table atsdb (i int_new, j int, t text, n numeric) distributed by (i, j);
insert into atsdb select i, i+1, i+2, i+3 from generate_series(1, 20) i;
drop type int_new cascade;
select * from atsdb;
select * from distcheck where rel = 'atsdb';
drop table atsdb;
-- Check that we correctly cascade for partitioned tables
create table atsdb (i int, j int, k int) distributed by (i) partition by range(k)
(start(1) end(10) every(1));
insert into atsdb select i+2, i+1, i from generate_series(1, 9) i;
select * from distcheck where rel like 'atsdb%';
alter table atsdb set distributed by (j);
select * from distcheck where rel like 'atsdb%';
select * from atsdb order by 1, 2, 3;
alter table atsdb set with(appendonly = true);
select relname, a.blocksize, compresslevel, compresstype, checksum from pg_class c, pg_appendonly a where
relname like 'atsdb%' and c.oid = a.relid order by 1;
select * from atsdb order by 1, 2, 3;
insert into atsdb select i+2, i+1, i from generate_series(1, 9) i;
select * from atsdb order by 1, 2, 3;
drop table atsdb;
-- check distribution correctly cascaded for inherited tables
create table dropColumnCascade (a int, b int, e int);
create table dropColumnCascadeChild (c int) inherits (dropColumnCascade);
create table dropColumnCascadeAnother (d int) inherits (dropColumnCascadeChild);
insert into dropColumnCascadeAnother select i,i,i from generate_series(1,10)i;
alter table dropColumnCascade drop column a;
select * from distcheck where rel like 'dropcolumnicascade%';
drop table dropColumnCascade cascade;
-- check DROP TYPE..CASCADE for dist key type for inherited tables
-- distribution should be set to randomly for base and inherited tables
create domain int_new as int;
create table dropColumnCascade (a int_new, b int, e int);
create table dropColumnCascadeChild (c int) inherits (dropColumnCascade);
create table dropColumnCascadeAnother (d int) inherits (dropColumnCascadeChild);
insert into dropColumnCascadeAnother select i,i,i from generate_series(1,10)i;
drop type int_new cascade;
select * from distcheck where rel like 'dropcolumncascade%';
drop table dropColumnCascade cascade;
-- Test corner cases in dropping distkey as inherited columns
create table p1 (f1 int, f2 int);
create table c1 (f1 int not null) inherits(p1);
alter table p1 drop column f1;
-- only p1 is randomly distributed, c1 is still distributed by c1.f1
select * from distcheck where rel in ('p1', 'c1');
alter table c1 drop column f1;
-- both c1 and p1 randomly distributed
select * from distcheck where rel in ('p1', 'c1');
drop table p1 cascade;
create table p1 (f1 int, f2 int);
create table c1 () inherits(p1);
alter table only p1 drop column f1;
-- only p1 is randomly distributed, c1 is still distributed by c1.f1
select * from distcheck where rel in ('p1', 'c1');
drop table p1 cascade;
-- check DROP TYPE..CASCADE for dist key type for inherited tables
-- distribution should be set to randomly for base and inherited tables
create domain int_new as int;
create table p1 (f1 int_new, f2 int);
create table c1 (f1 int_new not null) inherits(p1);
create table p1_inh (f1 int_new, f2 int);
create table c1_inh () inherits(p1_inh);
drop type int_new cascade;
-- all above tables set to randomly distributed
select * from distcheck where rel in ('p1', 'c1');
drop table p1 cascade;
drop table p1_inh cascade;
drop view distcheck;
-- MPP-5452
-- Should succeed
create table atsdb (i int, k int) distributed by (i) partition by range(i) (start (1) end(10)
every(1));
alter table atsdb alter partition for (5) set distributed by (i);
alter table atsdb alter partition for (5) set distributed by (i);
alter table atsdb alter partition for (5) set distributed by (i);
drop table atsdb;
--MPP-5500
CREATE TABLE test_add_drop_rename_column_change_datatype(
text_col text,
bigint_col bigint,
char_vary_col character varying(30),
numeric_col numeric,
int_col int4,
float_col float4,
int_array_col int[],
drop_col numeric,
before_rename_col int4,
change_datatype_col numeric,
a_ts_without timestamp without time zone,
b_ts_with timestamp with time zone,
date_column date) distributed randomly;
insert into test_add_drop_rename_column_change_datatype values ('0_zero', 0, '0_zero', 0, 0, 0, '{0}', 0, 0, 0, '2004-10-19 10:23:54', '2004-10-19 10:23:54+02', '1-1-2000');
insert into test_add_drop_rename_column_change_datatype values ('1_zero', 1, '1_zero', 1, 1, 1, '{1}', 1, 1, 1, '2005-10-19 10:23:54', '2005-10-19 10:23:54+02', '1-1-2001');
insert into test_add_drop_rename_column_change_datatype values ('2_zero', 2, '2_zero', 2, 2, 2, '{2}', 2, 2, 2, '2006-10-19 10:23:54', '2006-10-19 10:23:54+02', '1-1-2002');
ALTER TABLE test_add_drop_rename_column_change_datatype ADD COLUMN added_col character varying(30);
ALTER TABLE test_add_drop_rename_column_change_datatype DROP COLUMN drop_col ;
ALTER TABLE test_add_drop_rename_column_change_datatype RENAME COLUMN before_rename_col TO after_rename_col;
ALTER TABLE test_add_drop_rename_column_change_datatype ALTER COLUMN change_datatype_col TYPE int4;
alter table test_add_drop_rename_column_change_datatype set with(reorganize =
true) distributed randomly;
select * from test_add_drop_rename_column_change_datatype ;
drop table test_add_drop_rename_column_change_datatype ;
-- MPP-5501
-- should run without error
create table atsdb with (appendonly=true) as select * from
generate_series(1,1000);
alter table only atsdb set with(reorganize=true) distributed by (generate_series);
select count(*) from atsdb;
drop table atsdb;
-- MPP-5746
create table mpp5746 (c int[], t text);
insert into mpp5746 select array[i], i from generate_series(1, 100) i;
alter table mpp5746 set with (reorganize=true, appendonly = true);
select * from mpp5746 order by 1;
alter table mpp5746 drop column t;
select * from mpp5746 order by 1;
alter table mpp5746 set with (reorganize=true, appendonly = false);
select * from mpp5746 order by 1;
drop table mpp5746;
-- MPP-5738
create table mpp5738 (a int, b int, c int, d int)
partition by range(d) (start(1) end(10) inclusive every(1));
insert into mpp5738 select i, i+1, i+2, i from generate_series(1, 10) i;
select * from mpp5738;
alter table mpp5738 alter partition for (1) set with (appendonly=true);
select * from mpp5738;
drop table mpp5738;
drop table if exists mpp5754;
CREATE TABLE mpp5754 (
N_NATIONKEY INTEGER,
N_NAME CHAR(25),
N_REGIONKEY INTEGER,
N_COMMENT VARCHAR(152)
) with (appendonly = true, checksum = true)
distributed by (N_NATIONKEY);
copy mpp5754 from stdin with delimiter '|';
0|ALGERIA|0| haggle. carefully final deposits detect slyly agai
\.
select * from mpp5754 order by n_nationkey;
alter table mpp5754 set distributed randomly;
select count(*) from mpp5754;
alter table mpp5754 set distributed by (n_nationkey);
select * from mpp5754 order by n_nationkey;
drop table mpp5754;
-- MPP-5918
create role atsdb;
create table owner_test(i int, toast text) distributed randomly;
alter table owner_test owner to atsdb;
alter table owner_test set with (reorganize = true) distributed by (i);
-- verify, atsdb should own all three
select a.relname,
x.rolname as relowner,
y.rolname as toastowner,
z.rolname as toastidxowner
from pg_class a
inner join pg_class b on b.oid = a.reltoastrelid
inner join pg_index ti on ti.indrelid = b.oid
inner join pg_class c on c.oid = ti.indexrelid,
pg_authid x, pg_authid y, pg_authid z
where a.relname='owner_test'
and x.oid = a.relowner
and y.oid = b.relowner
and z.oid = c.relowner;
-- MPP-9663 - Check that the ownership is consistent on the segments as well
select a.relname,
x.rolname as relowner,
y.rolname as toastowner,
z.rolname as toastidxowner
from gp_dist_random('pg_class') a
inner join gp_dist_random('pg_class') b on b.oid = a.reltoastrelid
inner join pg_index ti on ti.indrelid = b.oid
inner join gp_dist_random('pg_class') c on c.oid = ti.indexrelid,
pg_authid x, pg_authid y, pg_authid z
where a.relname='owner_test'
and x.oid = a.relowner
and y.oid = b.relowner
and z.oid = c.relowner
and a.gp_segment_id = 0
and b.gp_segment_id = 0
and c.gp_segment_id = 0;
-- MPP-9663 - The code path is different when the table has dropped columns
alter table owner_test add column d text;
alter table owner_test drop column d;
alter table owner_test set with (reorganize = true) distributed by (i);
select a.relname,
x.rolname as relowner,
y.rolname as toastowner,
z.rolname as toastidxowner
from gp_dist_random('pg_class') a
inner join gp_dist_random('pg_class') b on b.oid = a.reltoastrelid
inner join pg_index ti on ti.indrelid = b.oid
inner join gp_dist_random('pg_class') c on c.oid = ti.indexrelid,
pg_authid x, pg_authid y, pg_authid z
where a.relname='owner_test'
and x.oid = a.relowner
and y.oid = b.relowner
and z.oid = c.relowner
and a.gp_segment_id = 0
and b.gp_segment_id = 0
and c.gp_segment_id = 0;
drop table owner_test;
drop role atsdb;
-- MPP-6332
create table abc (a int, b int, c int) distributed by (a);
Alter table abc set distributed randomly;
Alter table abc set with (reorganize=false) distributed randomly;
drop table abc;
-- MPP-18660: duplicate entry in gp_distribution_policy
set enable_indexscan=on;
set enable_seqscan=off;
drop table if exists distrib_index_test;
create table distrib_index_test (a int, b text) distributed by (a);
select count(*) from gp_distribution_policy
where localoid in (select oid from pg_class where relname='distrib_index_test');
begin;
drop table distrib_index_test;
rollback;
select count(*) from gp_distribution_policy
where localoid in (select oid from pg_class where relname='distrib_index_test');
reset enable_indexscan;
reset enable_seqscan;
drop table distrib_index_test;
-- alter partitioned table crash
-- Alter partitioned table set distributed by will crash when:
-- 1. reorganize = false.
-- 2. table have index.
-- 3. partition table have "with" option.
drop index if exists distrib_part_test_idx;
drop table if exists distrib_part_test;
CREATE TABLE distrib_part_test
(
col1 int,
col2 decimal,
col3 text,
col4 bool
)
distributed by (col1)
partition by list(col2)
(
partition part1 values(1,2,3,4,5,6,7,8,9,10) WITH (appendonly=false )
);
create index distrib_part_test_idx on distrib_part_test(col1);
ALTER TABLE public.distrib_part_test SET with (reorganize=false) DISTRIBUTED RANDOMLY;
-- MPP-23801
--
-- ALTER TABLE set distribution key should check compatible with unique index.
-- case 1
CREATE TABLE t_dist1(col1 INTEGER, col2 INTEGER, CONSTRAINT pk_t_dist1 PRIMARY KEY(col2)) DISTRIBUTED BY(col2);
ALTER TABLE t_dist1 SET DISTRIBUTED BY(col1);
-- case 2
CREATE TABLE t_dist2(col1 INTEGER, col2 INTEGER, col3 INTEGER, col4 INTEGER) DISTRIBUTED BY(col1);
CREATE UNIQUE INDEX idx1_t_dist2 ON t_dist2(col1, col2);
CREATE UNIQUE INDEX idx2_t_dist2 ON t_dist2(col1, col2, col3);
CREATE UNIQUE INDEX idx3_t_dist2 ON t_dist2(col1, col2, col4);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col1);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col2);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col1, col2);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col1, col2, col3);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col3);
ALTER TABLE t_dist2 SET DISTRIBUTED BY(col4);
-- Altering distribution policy for temp tables
create temp table atsdb (c1 int, c2 int) distributed randomly;
select * from atsdb;
alter table atsdb set distributed by (c1);
select * from atsdb;
alter table atsdb set distributed by (c2);
select * from atsdb;
--
-- ALTER TABLE SET DATA TYPE tests, where the column is part of the
-- distribution key.
--
CREATE TABLE distpol_typechange (i int2) DISTRIBUTED BY (i);
INSERT INTO distpol_typechange values (123);
ALTER TABLE distpol_typechange ALTER COLUMN i SET DATA TYPE int4;
DROP TABLE distpol_typechange;
CREATE TABLE distpol_typechange (p text) DISTRIBUTED BY (p);
-- This should throw an error, you can't change the datatype of a distribution
-- key column.
INSERT INTO distpol_typechange VALUES ('(1,1)');
ALTER TABLE distpol_typechange ALTER COLUMN p TYPE point USING p::point;
-- unless it's completely empty! But 'point' doesn't have hash a opclass,
-- so it cannot be part of the distribution key. We silently turn the
-- table randomly distributed.
TRUNCATE distpol_typechange;
ALTER TABLE distpol_typechange ALTER COLUMN p TYPE point USING p::point;
select policytype, distkey, distclass from gp_distribution_policy where localoid='distpol_typechange'::regclass;
-- Similar case, but with CREATE UNIQUE INDEX, rather than ALTER TABLE.
-- Creating a unique index on a completely empty table automatically updates
-- the distribution key to match the unique index. (This allows the common
-- case, where no DISTRIBUTED BY was given explicitly, and the system just
-- picked the first column, which isn't compatible with the unique index
-- that's created later, to work.) But not if the unique column doesn't
-- have a hash opclass!
CREATE TABLE tstab (i int4, t tsvector) distributed by (i);
CREATE UNIQUE INDEX tstab_idx ON tstab(t);
INSERT INTO tstab VALUES (1, 'foo');
-- ALTER TABLE SET DISTRIBUTED RANDOMLY should not work on a table
-- that has a primary key or unique index.
CREATE TABLE alter_table_with_primary_key (a int primary key);
ALTER TABLE alter_table_with_primary_key SET DISTRIBUTED RANDOMLY;
CREATE TABLE alter_table_with_unique_index (a int unique);
ALTER TABLE alter_table_with_unique_index SET DISTRIBUTED RANDOMLY;
-- Enable reorg partition leaf table
create table reorg_leaf (a int, b int, c int) distributed by (c)
partition by range(a)
subpartition by range (b)
subpartition template
(start(0) end (10) every (5))
(partition p0 start (0) end (5),
partition p1 start (5) end (10));
insert into reorg_leaf select i, i, i from generate_series(0, 9) i;
select *, gp_segment_id from reorg_leaf_1_prt_p0;
-- fail: cannot change the distribution key of one partition
alter table reorg_leaf_1_prt_p0 set with (reorganize=true) distributed by(b);
-- distribution key is already 'c', so this is allowed
alter table reorg_leaf_1_prt_p0 set with (reorganize=true) distributed by(c);
alter table reorg_leaf_1_prt_p0 set with (reorganize=true);
-- same with a leaf partition
alter table reorg_leaf_1_prt_p0_2_prt_1 set with (reorganize=true) distributed by(b);
alter table reorg_leaf_1_prt_p0_2_prt_1 set with (reorganize=true) distributed by(c);
select *, gp_segment_id from reorg_leaf_1_prt_p0;
alter table reorg_leaf_1_prt_p0_2_prt_1 set with (reorganize=true);
select *, gp_segment_id from reorg_leaf_1_prt_p0;
--
-- Test case for GUC gp_force_random_redistribution.
-- Manually toggle the GUC should control the behavior of redistribution for randomly-distributed tables.
-- But REORGANIZE=true should redistribute no matter what.
--
-- this only affects postgres planner;
set optimizer = false;
-- check the distribution difference between 't1' and 't2' after executing 'query_string'
-- return true if data distribution changed, otherwise false.
-- Note: in extremely rare cases, even after 't2' being randomly-distributed from 't1', they could still have the
-- exact same distribution. So let the tables have a reasonably large number of rows to reduce that possibility.
CREATE OR REPLACE FUNCTION check_redistributed(query_string text, t1 text, t2 text)
RETURNS BOOLEAN AS
$$
DECLARE
before_query TEXT;
after_query TEXT;
comparison_query TEXT;
comparison_count INT;
BEGIN
-- Prepare the query strings
before_query := format('SELECT gp_segment_id as segid, count(*) AS tupcount FROM %I GROUP BY gp_segment_id', t1);
after_query := format('SELECT gp_segment_id as segid, count(*) AS tupcount FROM %I GROUP BY gp_segment_id', t2);
comparison_query := format('SELECT COUNT(*) FROM ((TABLE %I EXCEPT TABLE %I) UNION ALL (TABLE %I EXCEPT TABLE %I))q', 'distribution1', 'distribution2', 'distribution2', 'distribution1');
-- Create temp tables to store the result
EXECUTE format('CREATE TEMP TABLE distribution1 AS %s DISTRIBUTED REPLICATED', before_query);
-- Execute provided query string
EXECUTE query_string;
EXECUTE format('CREATE TEMP TABLE distribution2 AS %s DISTRIBUTED REPLICATED', after_query);
-- Compare the tables using EXCEPT clause
EXECUTE comparison_query INTO comparison_count;
-- Drop temp tables
EXECUTE 'DROP TABLE distribution1';
EXECUTE 'DROP TABLE distribution2';
-- If count is greater than zero, then there's a difference
RETURN comparison_count > 0;
END;
$$
LANGUAGE plpgsql;
-- CO table builds temp table first instead of doing CTAS during REORGANIZE=true
create table t_reorganize(a int, b int) using ao_column distributed by (a);
insert into t_reorganize select 0,i from generate_series(1,1000)i;
select gp_segment_id, count(*) from t_reorganize group by gp_segment_id;
-- firstly, no redistribute
set gp_force_random_redistribution = off;
select check_redistributed('alter table t_reorganize set with (reorganize=true) distributed randomly', 't_reorganize', 't_reorganize');
-- reorganize from randomly to randomly should still redistribute
select check_redistributed('alter table t_reorganize set with (reorganize=true) distributed randomly', 't_reorganize', 't_reorganize');
-- but insert into table won't redistribute
create table t_random (like t_reorganize) distributed randomly;
select check_redistributed('insert into t_random select * from t_reorganize', 't_reorganize', 't_random');
-- but insert into a different distribution policy would still redistribute
create table t_distbya (like t_reorganize) distributed by (a);
select check_redistributed('insert into t_distbya select * from t_reorganize', 't_reorganize', 't_distbya');
-- now force distribute should redistribute in all cases
set gp_force_random_redistribution = on;
select check_redistributed('alter table t_reorganize set with (reorganize=true) distributed randomly', 't_reorganize', 't_reorganize');
select check_redistributed('alter table t_reorganize set with (reorganize=true) distributed randomly', 't_reorganize', 't_reorganize');
create table t_random (like t_reorganize) distributed randomly;
select check_redistributed('insert into t_random select * from t_reorganize', 't_reorganize', 't_random');
create table t_distbya (like t_reorganize) distributed by (a);
select check_redistributed('insert into t_distbya select * from t_reorganize', 't_reorganize', 't_distbya');
reset optimizer;
reset gp_force_random_redistribution;
-- When reorganize=false, we won't reorganize and this shouldn't be affected by the existing reloptions.
CREATE TABLE public.t_reorganize_false (
a integer,
b integer
) with (appendonly=false, autovacuum_enabled=false) DISTRIBUTED BY (a);
-- Insert values which will all be on one segment
INSERT INTO t_reorganize_false VALUES (0, generate_series(1,100));
SELECT gp_segment_id,count(*) from t_reorganize_false GROUP BY 1;
-- Change the distribution policy but because REORGANIZE=false, it should NOT be re-distributed
ALTER TABLE t_reorganize_false SET WITH (REORGANIZE=false) DISTRIBUTED RANDOMLY;
SELECT gp_segment_id,count(*) from t_reorganize_false GROUP BY 1;
DROP TABLE t_reorganize_false;
-- Same rule should apply to partitioned table too
CREATE TABLE public.t_reorganize_false (
a integer,
b integer
)
DISTRIBUTED BY (a) PARTITION BY RANGE(b)
(
PARTITION "00" START (0) END (1000) WITH (tablename='t_reorganize_false_0', appendonly='false', autovacuum_enabled=false),
PARTITION "01" START (1000) END (2000) WITH (tablename='t_reorganize_false_1', appendonly='false', autovacuum_enabled=false),
DEFAULT PARTITION def WITH (tablename='t_reorganize_false_def', appendonly='false', autovacuum_enabled=false)
);
-- Insert values which will all be on one segment
INSERT INTO t_reorganize_false VALUES (0, generate_series(1,100));
SELECT gp_segment_id,count(*) from t_reorganize_false GROUP BY 1;
-- Should NOT be re-distributed
ALTER TABLE t_reorganize_false SET WITH (REORGANIZE=false) DISTRIBUTED RANDOMLY;
SELECT gp_segment_id,count(*) from t_reorganize_false GROUP BY 1;
DROP TABLE t_reorganize_false;
-- Check that AT SET DISTRIBUTED BY cannot be combined with other subcommands
-- on the same table
CREATE TABLE atsdby_multiple(i int, j int);
ALTER TABLE atsdby_multiple SET DISTRIBUTED BY(j), ADD COLUMN k int;
ALTER TABLE atsdby_multiple SET WITH (reorganize=true), ADD COLUMN k int;