postgres/timescaledb
1
0
Fork 0
mirror of https://github.com/timescale/timescaledb.git synced 2025-05-16 10:33:27 +08:00
Code Issues Packages Projects Releases Wiki Activity
timescaledb/tsl/test/sql/compression.sql
Jan Nidzwetzki f3b3e55693 Add sorted paths for compressed chunks 
This patch adds per-chunk sorted paths for compressed chunks. This has
two advantages:

   (1) Make ChunkAppend possible. If at least one chunk of a hypertable
is uncompressed, PostgreSQL will generate a MergeAppend path in
generate_orderedappend_paths() / create_merge_append_path() due to the
existing pathkeys of the index on the uncompressed chunk. If all chunks
are compressed, no path keys are present and no MergeAppend path is
generated by PostgreSQL. In that case, the ChunkAppend optimization
cannot be used because MergeAppend path can be promoted in
ts_chunk_append_path_create(). Adding a sorted path with pathkeys
makes ChunkAppend possible for these queries.

   (2) Sorting on a per-chunk basis and merging / appending these
results could be faster than sorting the whole input. Especially limit
queries that use an ORDER BY that is compatible with the partitioning of
the hypertable could be inefficiently executed otherwise. For example,
an expensive query plan with a sort node on top of the append node could
be chosen. Due to the sort node at the high level in the query plan and
the missing ChunkAppend node (see (1)), all chunks are decompressed
(instead of only the actually needed ones).

This patch adds sorted versions of the DecompressChunkPaths. This
ensures that pathkeys are present and PostgreSQL can generate a
MergeAppend path that could be promoted into a ChunkAppend path.

Fixes: #4223
2023-10-27 09:13:22 +02:00

1075 lines
45 KiB
PL/PgSQL
Raw Blame History

-- This file and its contents are licensed under the Timescale License.
-- Please see the included NOTICE for copyright information and
-- LICENSE-TIMESCALE for a copy of the license.
SET timescaledb.enable_transparent_decompression to OFF;
\set PREFIX 'EXPLAIN (analyze, verbose, costs off, timing off, summary off)'
\ir include/rand_generator.sql
--test_collation ---
--basic test with count
create table foo (a integer, b integer, c integer, d integer);
select table_name from create_hypertable('foo', 'a', chunk_time_interval=> 10);
create unique index foo_uniq ON foo (a, b);
--note that the "d" order by column is all NULL
insert into foo values( 3 , 16 , 20, NULL);
insert into foo values( 10 , 10 , 20, NULL);
insert into foo values( 20 , 11 , 20, NULL);
insert into foo values( 30 , 12 , 20, NULL);
alter table foo set (timescaledb.compress, timescaledb.compress_segmentby = 'a,b', timescaledb.compress_orderby = 'c desc, d asc nulls last');
--test self-refencing updates
SET timescaledb.enable_transparent_decompression to ON;
update foo set c = 40
where a = (SELECT max(a) FROM foo);
SET timescaledb.enable_transparent_decompression to OFF;
select id, schema_name, table_name, compression_state as compressed, compressed_hypertable_id from
_timescaledb_catalog.hypertable order by id;
select * from _timescaledb_catalog.hypertable_compression order by hypertable_id, attname;
select * from timescaledb_information.compression_settings ORDER BY hypertable_name;
-- TEST2 compress-chunk for the chunks created earlier --
select compress_chunk( '_timescaledb_internal._hyper_1_2_chunk');
select tgname , tgtype, tgenabled , relname
from pg_trigger t, pg_class rel
where t.tgrelid = rel.oid and rel.relname like '_hyper_1_2_chunk' order by tgname;
\x
select * from chunk_compression_stats('foo')
order by chunk_name limit 2;
\x
select compress_chunk( '_timescaledb_internal._hyper_1_1_chunk');
\x
select * from _timescaledb_catalog.compression_chunk_size
order by chunk_id;
\x
select ch1.id, ch1.schema_name, ch1.table_name , ch2.table_name as compress_table
from
_timescaledb_catalog.chunk ch1, _timescaledb_catalog.chunk ch2
where ch1.compressed_chunk_id = ch2.id;
\set ON_ERROR_STOP 0
--cannot recompress the chunk the second time around
select compress_chunk( '_timescaledb_internal._hyper_1_2_chunk');
--TEST2a try DML on a compressed chunk
BEGIN;
insert into foo values( 11 , 10 , 20, 120);
ROLLBACK;
update foo set b =20 where a = 10;
delete from foo where a = 10;
--TEST2b try complex DML on compressed chunk
create table foo_join ( a integer, newval integer);
select table_name from create_hypertable('foo_join', 'a', chunk_time_interval=> 10);
insert into foo_join select generate_series(0,40, 10), 111;
create table foo_join2 ( a integer, newval integer);
select table_name from create_hypertable('foo_join2', 'a', chunk_time_interval=> 10);
insert into foo_join select generate_series(0,40, 10), 222;
update foo
set b = newval
from foo_join where foo.a = foo_join.a;
update foo
set b = newval
from foo_join where foo.a = foo_join.a and foo_join.a > 10;
--here the chunk gets excluded , so succeeds --
update foo
set b = newval
from foo_join where foo.a = foo_join.a and foo.a > 20;
update foo
set b = (select f1.newval from foo_join f1 left join lateral (select newval as newval2 from foo_join2 f2 where f1.a= f2.a ) subq on true limit 1);
--upsert test --
insert into foo values(10, 12, 12, 12)
on conflict( a, b)
do update set b = excluded.b;
SELECT * from foo ORDER BY a,b;
--TEST2c Do DML directly on the chunk.
insert into _timescaledb_internal._hyper_1_2_chunk values(10, 12, 12, 12)
on conflict( a, b)
do update set b = excluded.b + 12;
SELECT * from foo ORDER BY a,b;
update _timescaledb_internal._hyper_1_2_chunk
set b = 12;
delete from _timescaledb_internal._hyper_1_2_chunk;
--TEST2d decompress the chunk and try DML
select decompress_chunk( '_timescaledb_internal._hyper_1_2_chunk');
insert into foo values( 11 , 10 , 20, 120);
update foo set b =20 where a = 10;
select * from _timescaledb_internal._hyper_1_2_chunk order by a,b;
delete from foo where a = 10;
select * from _timescaledb_internal._hyper_1_2_chunk order by a,b;
-- TEST3 check if compress data from views is accurate
CREATE TABLE conditions (
time TIMESTAMPTZ NOT NULL,
location TEXT NOT NULL,
location2 char(10) NOT NULL,
temperature DOUBLE PRECISION NULL,
humidity DOUBLE PRECISION NULL
);
select create_hypertable( 'conditions', 'time', chunk_time_interval=> '31days'::interval);
alter table conditions set (timescaledb.compress, timescaledb.compress_segmentby = 'location', timescaledb.compress_orderby = 'time');
insert into conditions
select generate_series('2018-12-01 00:00'::timestamp, '2018-12-31 00:00'::timestamp, '1 day'), 'POR', 'klick', 55, 75;
insert into conditions
select generate_series('2018-12-01 00:00'::timestamp, '2018-12-31 00:00'::timestamp, '1 day'), 'NYC', 'klick', 55, 75;
select hypertable_id, attname, compression_algorithm_id , al.name
from _timescaledb_catalog.hypertable_compression hc,
_timescaledb_catalog.hypertable ht,
_timescaledb_catalog.compression_algorithm al
where ht.id = hc.hypertable_id and ht.table_name like 'conditions' and al.id = hc.compression_algorithm_id
ORDER BY hypertable_id, attname;
select attname, attstorage, typname from pg_attribute at, pg_class cl , pg_type ty
where cl.oid = at.attrelid and at.attnum > 0
and cl.relname = '_compressed_hypertable_4'
and atttypid = ty.oid
order by at.attnum;
SELECT ch1.schema_name|| '.' || ch1.table_name as "CHUNK_NAME", ch1.id "CHUNK_ID"
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id and ht.table_name like 'conditions'
ORDER BY ch1.id
LIMIT 1 \gset
SELECT count(*) from :CHUNK_NAME;
SELECT count(*) as "ORIGINAL_CHUNK_COUNT" from :CHUNK_NAME \gset
select tableoid::regclass, count(*) from conditions group by tableoid order by tableoid;
select compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id and ht.table_name like 'conditions' ORDER BY ch1.id limit 1;
--test that only one chunk was affected
--note tables with 0 rows will not show up in here.
select tableoid::regclass, count(*) from conditions group by tableoid order by tableoid;
select compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id and ht.table_name like 'conditions' and ch1.compressed_chunk_id IS NULL;
select tableoid::regclass, count(*) from conditions group by tableoid order by tableoid;
select compressed.schema_name|| '.' || compressed.table_name as "COMPRESSED_CHUNK_NAME"
from _timescaledb_catalog.chunk uncompressed, _timescaledb_catalog.chunk compressed
where uncompressed.compressed_chunk_id = compressed.id AND uncompressed.id = :'CHUNK_ID' \gset
SELECT count(*) from :CHUNK_NAME;
SELECT count(*) from :COMPRESSED_CHUNK_NAME;
SELECT sum(_ts_meta_count) from :COMPRESSED_CHUNK_NAME;
SELECT location, _ts_meta_sequence_num from :COMPRESSED_CHUNK_NAME ORDER BY 1,2;
\x
SELECT chunk_id, numrows_pre_compression, numrows_post_compression
FROM _timescaledb_catalog.chunk srcch,
_timescaledb_catalog.compression_chunk_size map,
_timescaledb_catalog.hypertable srcht
WHERE map.chunk_id = srcch.id and srcht.id = srcch.hypertable_id
and srcht.table_name like 'conditions'
order by chunk_id;
select * from chunk_compression_stats('conditions')
order by chunk_name;
select * from hypertable_compression_stats('foo');
select * from hypertable_compression_stats('conditions');
vacuum full foo;
vacuum full conditions;
-- After vacuum, table_bytes is 0, but any associated index/toast storage is not
-- completely reclaimed. Sets it at 8K (page size). So a chunk which has
-- been compressed still incurs an overhead of n * 8KB (for every index + toast table) storage on the original uncompressed chunk.
select pg_size_pretty(table_bytes), pg_size_pretty(index_bytes),
pg_size_pretty(toast_bytes), pg_size_pretty(total_bytes)
from hypertable_detailed_size('conditions');
select * from timescaledb_information.hypertables
where hypertable_name like 'foo' or hypertable_name like 'conditions'
order by hypertable_name;
\x
SELECT decompress_chunk(ch1.schema_name|| '.' || ch1.table_name) AS chunk
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht
WHERE ch1.hypertable_id = ht.id and ht.table_name LIKE 'conditions'
ORDER BY chunk;
SELECT count(*), count(*) = :'ORIGINAL_CHUNK_COUNT' from :CHUNK_NAME;
--check that the compressed chunk is dropped
\set ON_ERROR_STOP 0
SELECT count(*) from :COMPRESSED_CHUNK_NAME;
\set ON_ERROR_STOP 1
--size information is gone too
select count(*)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht,
_timescaledb_catalog.compression_chunk_size map
where ch1.hypertable_id = ht.id and ht.table_name like 'conditions'
and map.chunk_id = ch1.id;
--make sure compressed_chunk_id is reset to NULL
select ch1.compressed_chunk_id IS NULL
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id and ht.table_name like 'conditions';
-- test plans get invalidated when chunks get compressed
SET timescaledb.enable_transparent_decompression TO ON;
CREATE TABLE plan_inval(time timestamptz, device_id int);
SELECT create_hypertable('plan_inval','time');
ALTER TABLE plan_inval SET (timescaledb.compress,timescaledb.compress_orderby='time desc');
-- create 2 chunks
INSERT INTO plan_inval SELECT * FROM (VALUES ('2000-01-01'::timestamptz,1), ('2000-01-07'::timestamptz,1)) v(time,device_id);
SET max_parallel_workers_per_gather to 0;
-- Disable hash aggregation to get a deterministic test output
SET enable_hashagg = OFF;
PREPARE prep_plan AS SELECT count(*) FROM plan_inval;
EXECUTE prep_plan;
EXECUTE prep_plan;
EXECUTE prep_plan;
-- get name of first chunk
SELECT tableoid::regclass AS "CHUNK_NAME" FROM plan_inval ORDER BY time LIMIT 1
\gset
SELECT compress_chunk(:'CHUNK_NAME');
EXECUTE prep_plan;
EXPLAIN (COSTS OFF) EXECUTE prep_plan;
SET enable_hashagg = ON;
CREATE TABLE test_collation (
time TIMESTAMPTZ NOT NULL,
device_id TEXT COLLATE "C" NULL,
device_id_2 TEXT COLLATE "POSIX" NULL,
val_1 TEXT COLLATE "C" NULL,
val_2 TEXT COLLATE "POSIX" NULL
);
--we want all the data to go into 1 chunk. so use 1 year chunk interval
select create_hypertable( 'test_collation', 'time', chunk_time_interval=> '1 day'::interval);
\set ON_ERROR_STOP 0
--forbid setting collation in compression ORDER BY clause. (parse error is fine)
alter table test_collation set (timescaledb.compress, timescaledb.compress_segmentby='device_id, device_id_2', timescaledb.compress_orderby = 'val_1 COLLATE "POSIX", val2, time');
\set ON_ERROR_STOP 1
alter table test_collation set (timescaledb.compress, timescaledb.compress_segmentby='device_id, device_id_2', timescaledb.compress_orderby = 'val_1, val_2, time');
insert into test_collation
select generate_series('2018-01-01 00:00'::timestamp, '2018-01-10 00:00'::timestamp, '2 hour'), 'device_1', 'device_3', gen_rand_minstd(), gen_rand_minstd();
insert into test_collation
select generate_series('2018-01-01 00:00'::timestamp, '2018-01-10 00:00'::timestamp, '2 hour'), 'device_2', 'device_4', gen_rand_minstd(), gen_rand_minstd();
insert into test_collation
select generate_series('2018-01-01 00:00'::timestamp, '2018-01-10 00:00'::timestamp, '2 hour'), NULL, 'device_5', gen_rand_minstd(), gen_rand_minstd();
--compress 2 chunks
SELECT compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id
and ht.table_name like 'test_collation' ORDER BY ch1.id LIMIT 2;
CREATE OR REPLACE PROCEDURE reindex_compressed_hypertable(hypertable REGCLASS)
AS $$
DECLARE
hyper_id int;
BEGIN
SELECT h.compressed_hypertable_id
INTO hyper_id
FROM _timescaledb_catalog.hypertable h
WHERE h.table_name = hypertable::name;
EXECUTE format('REINDEX TABLE _timescaledb_internal._compressed_hypertable_%s',
hyper_id);
END $$ LANGUAGE plpgsql;
-- reindexing compressed hypertable to update statistics
CALL reindex_compressed_hypertable('test_collation');
--segment bys are pushed down correctly
EXPLAIN (costs off) SELECT * FROM test_collation WHERE device_id < 'a';
EXPLAIN (costs off) SELECT * FROM test_collation WHERE device_id < 'a' COLLATE "POSIX";
\set ON_ERROR_STOP 0
EXPLAIN (costs off) SELECT * FROM test_collation WHERE device_id COLLATE "POSIX" < device_id_2 COLLATE "C";
SELECT device_id < device_id_2 FROM test_collation;
\set ON_ERROR_STOP 1
--segment meta on order bys pushdown
--should work
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_1 < 'a';
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_2 < 'a';
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_1 < 'a' COLLATE "C";
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_2 < 'a' COLLATE "POSIX";
--cannot pushdown when op collation does not match column's collation since min/max used different collation than what op needs
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_1 < 'a' COLLATE "POSIX";
EXPLAIN (costs off) SELECT * FROM test_collation WHERE val_2 < 'a' COLLATE "C";
--test datatypes
CREATE TABLE datatype_test(
time timestamptz NOT NULL,
int2_column int2,
int4_column int4,
int8_column int8,
float4_column float4,
float8_column float8,
date_column date,
timestamp_column timestamp,
timestamptz_column timestamptz,
interval_column interval,
numeric_column numeric,
decimal_column decimal,
text_column text,
char_column char
);
SELECT create_hypertable('datatype_test','time');
ALTER TABLE datatype_test SET (timescaledb.compress);
INSERT INTO datatype_test VALUES ('2000-01-01',2,4,8,4.0,8.0,'2000-01-01','2001-01-01 12:00','2001-01-01 6:00','1 week', 3.41, 4.2, 'text', 'x');
SELECT compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id
and ht.table_name like 'datatype_test' ORDER BY ch1.id;
SELECT
attname, alg.name
FROM _timescaledb_catalog.hypertable ht
INNER JOIN _timescaledb_catalog.hypertable_compression htc ON ht.id=htc.hypertable_id
INNER JOIN _timescaledb_catalog.compression_algorithm alg ON alg.id=htc.compression_algorithm_id
WHERE ht.table_name='datatype_test'
ORDER BY attname;
--TEST try to compress a hypertable that has a continuous aggregate
CREATE TABLE metrics(time timestamptz, device_id int, v1 float, v2 float);
SELECT create_hypertable('metrics','time');
INSERT INTO metrics SELECT generate_series('2000-01-01'::timestamptz,'2000-01-10','1m'),1,0.25,0.75;
-- check expressions in view definition
CREATE MATERIALIZED VIEW cagg_expr WITH (timescaledb.continuous)
AS
SELECT
time_bucket('1d', time) AS time,
'Const'::text AS Const,
4.3::numeric AS "numeric",
first(metrics,time),
CASE WHEN true THEN 'foo' ELSE 'bar' END,
COALESCE(NULL,'coalesce'),
avg(v1) + avg(v2) AS avg1,
avg(v1+v2) AS avg2,
count(*) AS cnt
FROM metrics
GROUP BY 1 WITH NO DATA;
CALL refresh_continuous_aggregate('cagg_expr', NULL, NULL);
SELECT * FROM cagg_expr ORDER BY time LIMIT 5;
ALTER TABLE metrics set(timescaledb.compress);
-- test rescan in compress chunk dml blocker
CREATE TABLE rescan_test(id integer NOT NULL, t timestamptz NOT NULL, val double precision, PRIMARY KEY(id, t));
SELECT create_hypertable('rescan_test', 't', chunk_time_interval => interval '1 day');
-- compression
ALTER TABLE rescan_test SET (timescaledb.compress, timescaledb.compress_segmentby = 'id');
-- INSERT dummy data
INSERT INTO rescan_test SELECT 1, time, random() FROM generate_series('2000-01-01'::timestamptz, '2000-01-05'::timestamptz, '1h'::interval) g(time);
SELECT count(*) FROM rescan_test;
-- compress first chunk
SELECT compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht where ch1.hypertable_id = ht.id
and ht.table_name like 'rescan_test' ORDER BY ch1.id LIMIT 1;
-- count should be equal to count before compression
SELECT count(*) FROM rescan_test;
-- single row update is fine
UPDATE rescan_test SET val = val + 1 WHERE rescan_test.id = 1 AND rescan_test.t = '2000-01-03 00:00:00+00';
-- multi row update via WHERE is fine
UPDATE rescan_test SET val = val + 1 WHERE rescan_test.id = 1 AND rescan_test.t > '2000-01-03 00:00:00+00';
-- single row update with FROM is allowed if no compressed chunks are hit
UPDATE rescan_test SET val = tmp.val
FROM (SELECT x.id, x.t, x.val FROM unnest(array[(1, '2000-01-03 00:00:00+00', 2.045)]::rescan_test[]) AS x) AS tmp
WHERE rescan_test.id = tmp.id AND rescan_test.t = tmp.t AND rescan_test.t >= '2000-01-03';
-- single row update with FROM is blocked
\set ON_ERROR_STOP 0
UPDATE rescan_test SET val = tmp.val
FROM (SELECT x.id, x.t, x.val FROM unnest(array[(1, '2000-01-03 00:00:00+00', 2.045)]::rescan_test[]) AS x) AS tmp
WHERE rescan_test.id = tmp.id AND rescan_test.t = tmp.t;
-- bulk row update with FROM is blocked
UPDATE rescan_test SET val = tmp.val
FROM (SELECT x.id, x.t, x.val FROM unnest(array[(1, '2000-01-03 00:00:00+00', 2.045), (1, '2000-01-03 01:00:00+00', 8.045)]::rescan_test[]) AS x) AS tmp
WHERE rescan_test.id = tmp.id AND rescan_test.t = tmp.t;
\set ON_ERROR_STOP 1
-- Test FK constraint drop and recreate during compression and decompression on a chunk
CREATE TABLE meta (device_id INT PRIMARY KEY);
CREATE TABLE hyper(
time INT NOT NULL,
device_id INT REFERENCES meta(device_id) ON DELETE CASCADE ON UPDATE CASCADE,
val INT);
SELECT * FROM create_hypertable('hyper', 'time', chunk_time_interval => 10);
ALTER TABLE hyper SET (
timescaledb.compress,
timescaledb.compress_orderby = 'time',
timescaledb.compress_segmentby = 'device_id');
INSERT INTO meta VALUES (1), (2), (3), (4), (5);
INSERT INTO hyper VALUES (1, 1, 1), (2, 2, 1), (3, 3, 1), (10, 3, 2), (11, 4, 2), (11, 5, 2);
SELECT ch1.table_name AS "CHUNK_NAME", ch1.schema_name|| '.' || ch1.table_name AS "CHUNK_FULL_NAME"
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht
WHERE ch1.hypertable_id = ht.id AND ht.table_name LIKE 'hyper'
ORDER BY ch1.id LIMIT 1 \gset
SELECT constraint_schema, constraint_name, table_schema, table_name, constraint_type
FROM information_schema.table_constraints
WHERE table_name = :'CHUNK_NAME' AND constraint_type = 'FOREIGN KEY'
ORDER BY constraint_name;
SELECT compress_chunk(:'CHUNK_FULL_NAME');
SELECT constraint_schema, constraint_name, table_schema, table_name, constraint_type
FROM information_schema.table_constraints
WHERE table_name = :'CHUNK_NAME' AND constraint_type = 'FOREIGN KEY'
ORDER BY constraint_name;
-- Delete data from compressed chunk directly fails
\set ON_ERROR_STOP 0
DELETE FROM hyper WHERE device_id = 3;
\set ON_ERROR_STOP 0
-- Delete data from FK-referenced table deletes data from compressed chunk
SELECT * FROM hyper ORDER BY time, device_id;
DELETE FROM meta WHERE device_id = 3;
SELECT * FROM hyper ORDER BY time, device_id;
SELECT decompress_chunk(:'CHUNK_FULL_NAME');
SELECT constraint_schema, constraint_name, table_schema, table_name, constraint_type
FROM information_schema.table_constraints
WHERE table_name = :'CHUNK_NAME' AND constraint_type = 'FOREIGN KEY'
ORDER BY constraint_name;
-- create hypertable with 2 chunks
CREATE TABLE ht5(time TIMESTAMPTZ NOT NULL);
SELECT create_hypertable('ht5','time');
INSERT INTO ht5 SELECT '2000-01-01'::TIMESTAMPTZ;
INSERT INTO ht5 SELECT '2001-01-01'::TIMESTAMPTZ;
-- compressed chunk stats should not show dropped chunks
ALTER TABLE ht5 SET (timescaledb.compress);
SELECT compress_chunk(i) FROM show_chunks('ht5') i;
SELECT drop_chunks('ht5', newer_than => '2000-01-01'::TIMESTAMPTZ);
select chunk_name from chunk_compression_stats('ht5')
order by chunk_name;
-- Test enabling compression for a table with compound foreign key
-- (Issue https://github.com/timescale/timescaledb/issues/2000)
CREATE TABLE table2(col1 INT, col2 int, primary key (col1,col2));
CREATE TABLE table1(col1 INT NOT NULL, col2 INT);
ALTER TABLE table1 ADD CONSTRAINT fk_table1 FOREIGN KEY (col1,col2) REFERENCES table2(col1,col2);
SELECT create_hypertable('table1','col1', chunk_time_interval => 10);
-- Trying to list an incomplete set of fields of the compound key (should fail with a nice message)
ALTER TABLE table1 SET (timescaledb.compress, timescaledb.compress_segmentby = 'col1');
-- Listing all fields of the compound key should succeed:
ALTER TABLE table1 SET (timescaledb.compress, timescaledb.compress_segmentby = 'col1,col2');
SELECT * FROM timescaledb_information.compression_settings ORDER BY hypertable_name;
-- test delete/update on non-compressed tables involving hypertables with compression
CREATE TABLE uncompressed_ht (
time timestamptz NOT NULL,
value double precision,
series_id integer
);
SELECT table_name FROM create_hypertable ('uncompressed_ht', 'time');
INSERT INTO uncompressed_ht
VALUES ('2020-04-20 01:01', 100, 1), ('2020-05-20 01:01', 100, 1), ('2020-04-20 01:01', 200, 2);
CREATE TABLE compressed_ht (
time timestamptz NOT NULL,
value double precision,
series_id integer
);
SELECT table_name FROM create_hypertable ('compressed_ht', 'time');
ALTER TABLE compressed_ht SET (timescaledb.compress);
INSERT INTO compressed_ht
VALUES ('2020-04-20 01:01', 100, 1), ('2020-05-20 01:01', 100, 1);
SELECT compress_chunk (ch1.schema_name || '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1,
_timescaledb_catalog.hypertable ht
WHERE ch1.hypertable_id = ht.id
AND ht.table_name LIKE 'compressed_ht'
ORDER BY ch1.id;
BEGIN;
WITH compressed AS (
SELECT series_id
FROM compressed_ht
WHERE time >= '2020-04-17 17:14:24.161989+00'
)
DELETE FROM uncompressed_ht
WHERE series_id IN (SELECT series_id FROM compressed);
ROLLBACK;
-- test delete inside CTE
WITH compressed AS (
DELETE FROM compressed_ht RETURNING series_id
)
SELECT * FROM uncompressed_ht
WHERE series_id IN (SELECT series_id FROM compressed);
-- test update inside CTE is blocked
WITH compressed AS (
UPDATE compressed_ht SET value = 0.2 RETURNING *
)
SELECT * FROM uncompressed_ht
WHERE series_id IN (SELECT series_id FROM compressed);
DROP TABLE compressed_ht;
DROP TABLE uncompressed_ht;
-- Test that pg_stats and pg_class stats for uncompressed chunks are correctly updated after compression.
-- Note that approximate_row_count pulls from pg_class
CREATE TABLE stattest(time TIMESTAMPTZ NOT NULL, c1 int);
SELECT create_hypertable('stattest', 'time');
INSERT INTO stattest SELECT '2020/02/20 01:00'::TIMESTAMPTZ + ('1 hour'::interval * v), 250 * v FROM generate_series(0,25) v;
SELECT table_name INTO TEMPORARY temptable FROM _timescaledb_catalog.chunk WHERE hypertable_id = (SELECT id FROM _timescaledb_catalog.hypertable WHERE table_name = 'stattest');
\set statchunk '(select table_name from temptable)'
SELECT * FROM pg_stats WHERE tablename = :statchunk;
ALTER TABLE stattest SET (timescaledb.compress);
SELECT approximate_row_count('stattest');
SELECT compress_chunk(c) FROM show_chunks('stattest') c;
SELECT approximate_row_count('stattest');
-- Uncompressed chunk table is empty since we just compressed the chunk and moved everything to compressed chunk table.
-- reltuples is initially -1 on PG14 before VACUUM/ANALYZE was run
SELECT relpages, CASE WHEN reltuples > 0 THEN reltuples ELSE 0 END as reltuples FROM pg_class WHERE relname = :statchunk;
SELECT histogram_bounds FROM pg_stats WHERE tablename = :statchunk AND attname = 'c1';
SELECT compch.table_name as "STAT_COMP_CHUNK_NAME"
FROM _timescaledb_catalog.hypertable ht, _timescaledb_catalog.chunk ch
, _timescaledb_catalog.chunk compch
WHERE ht.table_name = 'stattest' AND ch.hypertable_id = ht.id
AND compch.id = ch.compressed_chunk_id AND ch.compressed_chunk_id > 0 \gset
-- reltuples is initially -1 on PG14 before VACUUM/ANALYZE was run
SELECT relpages, CASE WHEN reltuples > 0 THEN reltuples ELSE 0 END as reltuples FROM pg_class WHERE relname = :'STAT_COMP_CHUNK_NAME';
-- Now verify stats are updated on compressed chunk table when we analyze the hypertable.
ANALYZE stattest;
SELECT histogram_bounds FROM pg_stats WHERE tablename = :statchunk AND attname = 'c1';
-- Unfortunately, the stats on the hypertable won't find any rows to sample from the chunk
SELECT histogram_bounds FROM pg_stats WHERE tablename = 'stattest' AND attname = 'c1';
SELECT relpages, reltuples FROM pg_class WHERE relname = :statchunk;
-- verify that corresponding compressed chunk table stats is updated as well.
SELECT relpages, reltuples FROM pg_class WHERE relname = :'STAT_COMP_CHUNK_NAME';
-- Verify partial chunk stats are handled correctly when analyzing
-- for both uncompressed and compressed chunk tables
INSERT INTO stattest SELECT '2020/02/20 01:00'::TIMESTAMPTZ + ('1 hour'::interval * v), 250 * v FROM generate_series(25,50) v;
ANALYZE stattest;
SELECT histogram_bounds FROM pg_stats WHERE tablename = :statchunk AND attname = 'c1';
-- Hypertable will now see the histogram bounds since we have data in the uncompressed chunk table.
SELECT histogram_bounds FROM pg_stats WHERE tablename = 'stattest' AND attname = 'c1';
-- verify that corresponding uncompressed chunk table stats is updated as well.
SELECT relpages, reltuples FROM pg_class WHERE relname = :statchunk;
-- verify that corresponding compressed chunk table stats have not changed since
-- we didn't compress anything new.
SELECT relpages, reltuples FROM pg_class WHERE relname = :'STAT_COMP_CHUNK_NAME';
-- Verify that decompressing the chunk restores autoanalyze to the hypertable's setting
SELECT reloptions FROM pg_class WHERE relname = :statchunk;
SELECT decompress_chunk(c) FROM show_chunks('stattest') c;
SELECT reloptions FROM pg_class WHERE relname = :statchunk;
SELECT compress_chunk(c) FROM show_chunks('stattest') c;
SELECT reloptions FROM pg_class WHERE relname = :statchunk;
ALTER TABLE stattest SET (autovacuum_enabled = false);
SELECT decompress_chunk(c) FROM show_chunks('stattest') c;
SELECT reloptions FROM pg_class WHERE relname = :statchunk;
-- Verify that even a global analyze works as well, changing message scope here
-- to hide WARNINGs for skipped tables
SET client_min_messages TO ERROR;
ANALYZE;
SET client_min_messages TO NOTICE;
SELECT histogram_bounds FROM pg_stats WHERE tablename = :statchunk and attname = 'c1';
SELECT relpages, reltuples FROM pg_class WHERE relname = :statchunk;
DROP TABLE stattest;
--- Test that analyze on compression internal table updates stats on original chunks
CREATE TABLE stattest2(time TIMESTAMPTZ NOT NULL, c1 int, c2 int);
SELECT create_hypertable('stattest2', 'time', chunk_time_interval=>'1 day'::interval);
ALTER TABLE stattest2 SET (timescaledb.compress, timescaledb.compress_segmentby='c1');
INSERT INTO stattest2 SELECT '2020/06/20 01:00'::TIMESTAMPTZ ,1 , generate_series(1, 200, 1);
INSERT INTO stattest2 SELECT '2020/07/20 01:00'::TIMESTAMPTZ ,1 , generate_series(1, 200, 1);
SELECT compress_chunk(ch1.schema_name|| '.' || ch1.table_name)
FROM _timescaledb_catalog.chunk ch1, _timescaledb_catalog.hypertable ht
WHERE ch1.hypertable_id = ht.id and ht.table_name like 'stattest2'
ORDER BY ch1.id limit 1;
-- reltuples is initially -1 on PG14 before VACUUM/ANALYZE has been run
SELECT relname, CASE WHEN reltuples > 0 THEN reltuples ELSE 0 END AS reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = 'stattest2' AND ch.hypertable_id = ht.id )
order by relname;
\c :TEST_DBNAME :ROLE_SUPERUSER
--overwrite pg_class stats for the compressed chunk.
UPDATE pg_class
SET reltuples = 0, relpages = 0
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch,
_timescaledb_catalog.hypertable ht
WHERE ht.table_name = 'stattest2' AND ch.hypertable_id = ht.id
AND ch.compressed_chunk_id > 0 );
\c :TEST_DBNAME :ROLE_DEFAULT_PERM_USER
-- reltuples is initially -1 on PG14 before VACUUM/ANALYZE has been run
SELECT relname, CASE WHEN reltuples > 0 THEN reltuples ELSE 0 END AS reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = 'stattest2' AND ch.hypertable_id = ht.id )
order by relname;
SELECT '_timescaledb_internal.' || compht.table_name as "STAT_COMP_TABLE",
compht.table_name as "STAT_COMP_TABLE_NAME"
FROM _timescaledb_catalog.hypertable ht, _timescaledb_catalog.hypertable compht
WHERE ht.table_name = 'stattest2' AND ht.compressed_hypertable_id = compht.id \gset
--analyze the compressed table, will update stats for the raw table.
ANALYZE :STAT_COMP_TABLE;
-- reltuples is initially -1 on PG14 before VACUUM/ANALYZE has been run
SELECT relname, CASE WHEN reltuples > 0 THEN reltuples ELSE 0 END AS reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = 'stattest2' AND ch.hypertable_id = ht.id )
ORDER BY relname;
SELECT relname, reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = :'STAT_COMP_TABLE_NAME' AND ch.hypertable_id = ht.id )
ORDER BY relname;
--analyze on stattest2 should not overwrite
ANALYZE stattest2;
SELECT relname, reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = 'stattest2' AND ch.hypertable_id = ht.id )
ORDER BY relname;
SELECT relname, reltuples, relpages, relallvisible FROM pg_class
WHERE relname in ( SELECT ch.table_name FROM
_timescaledb_catalog.chunk ch, _timescaledb_catalog.hypertable ht
WHERE ht.table_name = :'STAT_COMP_TABLE_NAME' AND ch.hypertable_id = ht.id )
ORDER BY relname;
-- analyze on compressed hypertable should restore stats
-- Test approximate_row_count() with compressed hypertable
--
CREATE TABLE approx_count(time timestamptz not null, device int, temp float);
SELECT create_hypertable('approx_count', 'time');
INSERT INTO approx_count SELECT t, (abs(timestamp_hash(t::timestamp)) % 10) + 1, random()*80
FROM generate_series('2018-03-02 1:00'::TIMESTAMPTZ, '2018-03-04 1:00', '1 hour') t;
SELECT count(*) FROM approx_count;
ALTER TABLE approx_count SET (timescaledb.compress, timescaledb.compress_segmentby='device', timescaledb.compress_orderby = 'time DESC');
SELECT approximate_row_count('approx_count');
ANALYZE approx_count;
SELECT approximate_row_count('approx_count');
DROP TABLE approx_count;
--TEST drop_chunks from a compressed hypertable (that has caggs defined).
-- chunk metadata is still retained. verify correct status for chunk
SELECT count(*)
FROM (SELECT compress_chunk(ch) FROM show_chunks('metrics') ch ) q;
SELECT drop_chunks('metrics', older_than=>'1 day'::interval);
SELECT
c.table_name as chunk_name,
c.status as chunk_status, c.dropped, c.compressed_chunk_id as comp_id
FROM _timescaledb_catalog.hypertable h, _timescaledb_catalog.chunk c
WHERE h.id = c.hypertable_id and h.table_name = 'metrics'
ORDER BY 1;
SELECT "time", cnt FROM cagg_expr ORDER BY time LIMIT 5;
--now reload data into the dropped chunks region, then compress
-- then verify chunk status/dropped column
INSERT INTO metrics SELECT generate_series('2000-01-01'::timestamptz,'2000-01-10','1m'),1,0.25,0.75;
SELECT count(*)
FROM (SELECT compress_chunk(ch) FROM show_chunks('metrics') ch) q;
SELECT
c.table_name as chunk_name,
c.status as chunk_status, c.dropped, c.compressed_chunk_id as comp_id
FROM _timescaledb_catalog.hypertable h, _timescaledb_catalog.chunk c
WHERE h.id = c.hypertable_id and h.table_name = 'metrics'
ORDER BY 1;
SELECT count(*) FROM metrics;
-- test sequence number is local to segment by
CREATE TABLE local_seq(time timestamptz, device int);
SELECT table_name FROM create_hypertable('local_seq','time');
ALTER TABLE local_seq SET(timescaledb.compress,timescaledb.compress_segmentby='device');
INSERT INTO local_seq SELECT '2000-01-01',1 FROM generate_series(1,3000);
INSERT INTO local_seq SELECT '2000-01-01',2 FROM generate_series(1,3500);
INSERT INTO local_seq SELECT '2000-01-01',3 FROM generate_series(1,3000);
INSERT INTO local_seq SELECT '2000-01-01',4 FROM generate_series(1,3000);
INSERT INTO local_seq SELECT '2000-01-01', generate_series(5,8);
SELECT compress_chunk(c) FROM show_chunks('local_seq') c;
SELECT
format('%s.%s',chunk.schema_name,chunk.table_name) AS "COMP_CHUNK"
FROM _timescaledb_catalog.hypertable ht
INNER JOIN _timescaledb_catalog.hypertable ht_comp ON ht_comp.id = ht.compressed_hypertable_id
INNER JOIN _timescaledb_catalog.chunk ON chunk.hypertable_id = ht_comp.id
WHERE ht.table_name = 'local_seq' \gset
SELECT device, _ts_meta_sequence_num, _ts_meta_count FROM :COMP_CHUNK ORDER BY 1,2;
-- github issue 4872
-- If subplan of ConstraintAwareAppend is TidRangeScan, then SELECT on
-- hypertable fails with error "invalid child of chunk append: Node (26)"
CREATE TABLE tidrangescan_test(time timestamptz, device_id int, v1 float, v2 float);
SELECT create_hypertable('tidrangescan_test','time');
INSERT INTO tidrangescan_test SELECT generate_series('2000-01-01'::timestamptz,'2000-01-10','1m'),1,0.25,0.75;
CREATE MATERIALIZED VIEW tidrangescan_expr WITH (timescaledb.continuous)
AS
SELECT
time_bucket('1d', time) AS time,
'Const'::text AS Const,
4.3::numeric AS "numeric",
first(tidrangescan_test,time),
CASE WHEN true THEN 'foo' ELSE 'bar' END,
COALESCE(NULL,'coalesce'),
avg(v1) + avg(v2) AS avg1,
avg(v1+v2) AS avg2,
count(*) AS cnt
FROM tidrangescan_test
WHERE ctid < '(1,1)'::tid GROUP BY 1 WITH NO DATA;
CALL refresh_continuous_aggregate('tidrangescan_expr', NULL, NULL);
SET timescaledb.enable_chunk_append to off;
SET enable_indexscan to off;
SELECT time, const, numeric,first, avg1, avg2 FROM tidrangescan_expr ORDER BY time LIMIT 5;
RESET timescaledb.enable_chunk_append;
RESET enable_indexscan;
-- Test the number of allocated parallel workers for decompression
-- Test that a parallel plan is generated
-- with different number of parallel workers
CREATE TABLE f_sensor_data(
time timestamptz NOT NULL,
sensor_id integer NOT NULL,
cpu double precision NULL,
temperature double precision NULL
);
SELECT FROM create_hypertable('f_sensor_data','time');
SELECT set_chunk_time_interval('f_sensor_data', INTERVAL '1 year');
-- Create one chunk manually to ensure, all data is inserted into one chunk
SELECT * FROM _timescaledb_functions.create_chunk('f_sensor_data',' {"time": [181900977000000, 515024000000000]}');
INSERT INTO f_sensor_data
SELECT
time AS time,
sensor_id,
100.0,
36.6
FROM
generate_series('1980-01-01 00:00'::timestamp, '1980-02-28 12:00', INTERVAL '1 day') AS g1(time),
generate_series(1, 1700, 1 ) AS g2(sensor_id)
ORDER BY
time;
ALTER TABLE f_sensor_data SET (timescaledb.compress, timescaledb.compress_segmentby='sensor_id' ,timescaledb.compress_orderby = 'time DESC');
SELECT compress_chunk(i) FROM show_chunks('f_sensor_data') i;
CALL reindex_compressed_hypertable('f_sensor_data');
-- Encourage use of parallel plans
SET parallel_setup_cost = 0;
SET parallel_tuple_cost = 0;
SET min_parallel_table_scan_size TO '0';
\set explain 'EXPLAIN (VERBOSE, COSTS OFF)'
SHOW min_parallel_table_scan_size;
SHOW max_parallel_workers;
SHOW max_parallel_workers_per_gather;
SET max_parallel_workers_per_gather = 4;
SHOW max_parallel_workers_per_gather;
-- We disable enable_parallel_append here to ensure
-- that we create the same query plan in all PG 14.X versions
SET enable_parallel_append = false;
:explain
SELECT sum(cpu) FROM f_sensor_data;
-- Encourage use of Index Scan
SET enable_seqscan = false;
SET enable_indexscan = true;
SET min_parallel_index_scan_size = 0;
SET min_parallel_table_scan_size = 0;
CREATE INDEX ON f_sensor_data (time, sensor_id);
:explain
SELECT * FROM f_sensor_data WHERE sensor_id > 100;
RESET enable_parallel_append;
-- Test for partially compressed chunks
INSERT INTO f_sensor_data
SELECT
time AS time,
sensor_id,
100.0,
36.6
FROM
generate_series('1980-01-01 00:00'::timestamp, '1980-01-30 12:00', INTERVAL '1 day') AS g1(time),
generate_series(1700, 1800, 1 ) AS g2(sensor_id)
ORDER BY
time;
:explain
SELECT sum(cpu) FROM f_sensor_data;
:explain
SELECT * FROM f_sensor_data WHERE sensor_id > 100;
-- Test non-partial paths below append are not executed multiple times
CREATE TABLE ts_device_table(time INTEGER, device INTEGER, location INTEGER, value INTEGER);
CREATE UNIQUE INDEX device_time_idx on ts_device_table(time, device);
SELECT create_hypertable('ts_device_table', 'time', chunk_time_interval => 1000);
INSERT INTO ts_device_table SELECT generate_series(0,999,1), 1, 100, 20;
ALTER TABLE ts_device_table set(timescaledb.compress, timescaledb.compress_segmentby='location', timescaledb.compress_orderby='time');
SELECT compress_chunk(i) AS chunk_name FROM show_chunks('ts_device_table') i \gset
SELECT count(*) FROM ts_device_table;
SELECT count(*) FROM :chunk_name;
INSERT INTO ts_device_table VALUES (1, 1, 100, 100) ON CONFLICT DO NOTHING;
SELECT count(*) FROM :chunk_name;
SET parallel_setup_cost TO '0';
SET parallel_tuple_cost TO '0';
SET min_parallel_table_scan_size TO '8';
SET min_parallel_index_scan_size TO '8';
SET random_page_cost TO '0';
SELECT count(*) FROM :chunk_name;
ANALYZE :chunk_name;
SELECT count(*) FROM :chunk_name;
-- Test that parallel plans are chosen even if partial and small chunks are involved
RESET min_parallel_index_scan_size;
RESET min_parallel_table_scan_size;
CREATE TABLE ht_metrics_partially_compressed(time timestamptz, device int, value float);
SELECT create_hypertable('ht_metrics_partially_compressed','time',create_default_indexes:=false);
ALTER TABLE ht_metrics_partially_compressed SET (timescaledb.compress, timescaledb.compress_segmentby='device');
INSERT INTO ht_metrics_partially_compressed
SELECT time, device, device * 0.1 FROM
generate_series('2020-01-01'::timestamptz,'2020-01-02'::timestamptz, INTERVAL '1 m') g(time),
LATERAL (SELECT generate_series(1,2) AS device) g2;
SELECT compress_chunk(c) FROM show_chunks('ht_metrics_partially_compressed') c;
INSERT INTO ht_metrics_partially_compressed VALUES ('2020-01-01'::timestamptz, 1, 0.1);
:explain
SELECT * FROM ht_metrics_partially_compressed ORDER BY time DESC, device LIMIT 1;
-- Test parameter change on rescan
-- issue 6069
CREATE TABLE IF NOT EXISTS i6069 (
timestamp TIMESTAMP WITHOUT TIME ZONE NOT NULL,
attr_id SMALLINT NOT NULL,
number_val DOUBLE PRECISION DEFAULT NULL
);
SELECT table_name FROM create_hypertable(
'i6069', 'timestamp',
create_default_indexes => FALSE, if_not_exists => TRUE, chunk_time_interval => INTERVAL '1 day'
);
ALTER TABLE i6069 SET (
timescaledb.compress,
timescaledb.compress_segmentby = 'attr_id'
);
INSERT INTO i6069 VALUES('2023-07-01', 1, 1),('2023-07-03', 2, 1),('2023-07-05', 3, 1),
('2023-07-01', 4, 1),('2023-07-03', 5, 1),('2023-07-05', 6, 1),
('2023-07-01', 7, 1),('2023-07-03', 8, 1),('2023-07-05', 9, 1),
('2023-07-01', 10, 1),('2023-07-03', 11, 1),('2023-07-05', 12, 1),
('2023-07-01', 13, 1),('2023-07-03', 14, 1),('2023-07-05', 15, 1),
('2023-07-01', 16, 1),('2023-07-03', 17, 1),('2023-07-05', 18, 1),
('2023-07-01', 19, 1),('2023-07-03', 20, 1),('2023-07-05', 21, 1),
('2023-07-01', 22, 1),('2023-07-03', 23, 1),('2023-07-05', 24, 1),
('2023-07-01', 25, 1),('2023-07-03', 26, 1),('2023-07-05', 27, 1),
('2023-07-01', 28, 1),('2023-07-03', 29, 1),('2023-07-05', 30, 1),
('2023-07-01', 31, 1),('2023-07-03', 32, 1),('2023-07-05', 33, 1),
('2023-07-01', 34, 1),('2023-07-03', 35, 1),('2023-07-05', 36, 1),
('2023-07-01', 37, 1),('2023-07-03', 38, 1),('2023-07-05', 39, 1),
('2023-07-01', 40, 1),('2023-07-03', 41, 1),('2023-07-05', 42, 1),
('2023-07-01', 43, 1),('2023-07-03', 44, 1),('2023-07-05', 45, 1),
('2023-07-01', 46, 1),('2023-07-03', 47, 1),('2023-07-05', 48, 1),
('2023-07-01', 49, 1),('2023-07-03', 50, 1),('2023-07-05', 51, 1),
('2023-07-01', 52, 1),('2023-07-03', 53, 1),('2023-07-05', 54, 1),
('2023-07-01', 55, 1),('2023-07-03', 56, 1),('2023-07-05', 57, 1),
('2023-07-01', 58, 1),('2023-07-03', 59, 1),('2023-07-05', 60, 1),
('2023-07-01', 61, 1),('2023-07-03', 62, 1),('2023-07-05', 63, 1),
('2023-07-01', 64, 1),('2023-07-03', 65, 1),('2023-07-05', 66, 1),
('2023-07-01', 67, 1),('2023-07-03', 68, 1),('2023-07-05', 69, 1),
('2023-07-01', 70, 1),('2023-07-03', 71, 1),('2023-07-05', 72, 1),
('2023-07-01', 73, 1),('2023-07-03', 74, 1),('2023-07-05', 75, 1),
('2023-07-01', 76, 1),('2023-07-03', 77, 1),('2023-07-05', 78, 1),
('2023-07-01', 79, 1),('2023-07-03', 80, 1),('2023-07-05', 81, 1),
('2023-07-01', 82, 1),('2023-07-03', 83, 1),('2023-07-05', 84, 1),
('2023-07-01', 85, 1),('2023-07-03', 86, 1),('2023-07-05', 87, 1),
('2023-07-01', 88, 1),('2023-07-03', 89, 1),('2023-07-05', 90, 1),
('2023-07-01', 91, 1),('2023-07-03', 92, 1),('2023-07-05', 93, 1),
('2023-07-01', 94, 1),('2023-07-03', 95, 1),('2023-07-05', 96, 1),
('2023-07-01', 97, 1),('2023-07-03', 98, 1),('2023-07-05', 99, 1),
('2023-07-01', 100, 1),('2023-07-03', 101, 1),('2023-07-05', 102, 1),
('2023-07-01', 103, 1),('2023-07-03', 104, 1),('2023-07-05', 105, 1),
('2023-07-01', 106, 1),('2023-07-03', 107, 1),('2023-07-05', 108, 1),
('2023-07-01', 109, 1),('2023-07-03', 110, 1),('2023-07-05', 111, 1),
('2023-07-01', 112, 1),('2023-07-03', 113, 1),('2023-07-05', 114, 1),
('2023-07-01', 115, 1),('2023-07-03', 116, 1),('2023-07-05', 117, 1),
('2023-07-01', 118, 1),('2023-07-03', 119, 1),('2023-07-05', 120, 1);
SELECT compress_chunk(i, if_not_compressed => true) FROM show_chunks('i6069') i;
SET enable_indexscan = ON;
SET enable_seqscan = OFF;
:explain
SELECT * FROM ( VALUES(1),(2),(3),(4),(5),(6),(7),(8),(9),(10) ) AS attr_ids(attr_id)
INNER JOIN LATERAL (
SELECT * FROM i6069
WHERE i6069.attr_id = attr_ids.attr_id AND
timestamp > '2023-06-30' AND timestamp < '2023-07-06'
ORDER BY timestamp desc LIMIT 1 ) a ON true;
SELECT * FROM ( VALUES(1),(2),(3),(4),(5),(6),(7),(8),(9),(10) ) AS attr_ids(attr_id)
INNER JOIN LATERAL (
SELECT * FROM i6069
WHERE i6069.attr_id = attr_ids.attr_id AND
timestamp > '2023-06-30' AND timestamp < '2023-07-06'
ORDER BY timestamp desc LIMIT 1 ) a ON true;
RESET enable_indexscan;
RESET enable_seqscan;
-- When all chunks are compressed and a limit query is performed, only the needed
-- chunks should be accessed
CREATE TABLE sensor_data_compressed (
time timestamptz not null,
sensor_id integer not null,
cpu double precision null,
temperature double precision null);
SELECT FROM create_hypertable('sensor_data_compressed', 'time');
INSERT INTO sensor_data_compressed (time, sensor_id, cpu, temperature)
VALUES
('1980-01-02 00:00:00-00', 1, 3, 12.0),
('1980-01-03 00:00:00-00', 3, 4, 15.0),
('1980-02-04 00:00:00-00', 1, 2, 17.0),
('1980-02-05 00:00:00-00', 3, 6, 11.0),
('1980-03-06 00:00:00-00', 1, 8, 13.0),
('1980-03-07 00:00:00-00', 3, 4, 4.0),
('1980-04-08 00:00:00-00', 1, 2, 1.0),
('1980-04-03 00:00:00-00', 3, 5, 33.0),
('1980-05-02 00:00:00-00', 1, 8, 41.0),
('1980-05-03 00:00:00-00', 3, 4, 22.0),
('1980-06-02 00:00:00-00', 1, 1, 45.0),
('1980-06-03 00:00:00-00', 3, 3, 44.0);
ALTER TABLE sensor_data_compressed SET (timescaledb.compress, timescaledb.compress_segmentby='sensor_id', timescaledb.compress_orderby = 'time DESC');
-- Compress three of the chunks
SELECT compress_chunk(ch) FROM show_chunks('sensor_data_compressed') ch LIMIT 3;
ANALYZE sensor_data_compressed;
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
-- Only the first chunks should be accessed (sorted merge append is enabled)
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
-- Only the first chunks should be accessed (sorted merge append is disabled)
SET timescaledb.enable_decompression_sorted_merge = FALSE;
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
RESET timescaledb.enable_decompression_sorted_merge;
-- Compress the remaining chunks
SELECT compress_chunk(ch, if_not_compressed => true) FROM show_chunks('sensor_data_compressed') ch;
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
-- Only the first chunks should be accessed (sorted merge append is enabled)
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
-- Only the first chunks should be accessed (sorted merge append is disabled)
SET timescaledb.enable_decompression_sorted_merge = FALSE;
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
RESET timescaledb.enable_decompression_sorted_merge;
-- Convert the last chunk into a partially compressed chunk
INSERT INTO sensor_data_compressed (time, sensor_id, cpu, temperature)
VALUES ('1980-01-02 01:00:00-00', 2, 4, 14.0);
-- Only the first chunks should be accessed (sorted merge append is enabled)
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
-- Only the first chunks should be accessed (sorted merge append is disabled)
SET timescaledb.enable_decompression_sorted_merge = FALSE;
:PREFIX
SELECT * FROM sensor_data_compressed ORDER BY time DESC LIMIT 5;
RESET timescaledb.enable_decompression_sorted_merge;
Reference in New Issue View Git Blame Copy Permalink