1
0
mirror of https://github.com/postgres/postgres.git synced 2025-04-25 21:42:33 +03:00
Masahiko Sawada 072ee847ad Skip logical decoding of already-aborted transactions.
Previously, transaction aborts were detected concurrently only during
system catalog scans while replaying a transaction in streaming mode.

This commit adds an additional CLOG lookup to check the transaction
status, allowing the logical decoding to skip changes also when it
doesn't touch system catalogs, if the transaction is already
aborted. This optimization enhances logical decoding performance,
especially for large transactions that have already been rolled back,
as it avoids unnecessary disk or network I/O.

To avoid potential slowdowns caused by frequent CLOG lookups for small
transactions (most of which commit), the CLOG lookup is performed only
for large transactions before eviction. The performance benchmark
results showed there is not noticeable performance regression due to
CLOG lookups.

Reviewed-by: Amit Kapila, Peter Smith, Vignesh C, Ajin Cherian
Reviewed-by: Dilip Kumar, Andres Freund
Discussion: https://postgr.es/m/CAD21AoDht9Pz_DFv_R2LqBTBbO4eGrpa9Vojmt5z5sEx3XwD7A@mail.gmail.com
2025-02-12 16:31:34 -08:00

184 lines
7.3 KiB
Plaintext

-- predictability
SET synchronous_commit = on;
SELECT 'init' FROM
pg_create_logical_replication_slot('regression_slot_stats1', 'test_decoding') s1,
pg_create_logical_replication_slot('regression_slot_stats2', 'test_decoding') s2,
pg_create_logical_replication_slot('regression_slot_stats3', 'test_decoding') s3;
?column?
----------
init
(1 row)
CREATE TABLE stats_test(data text);
-- non-spilled xact
SET logical_decoding_work_mem to '64MB';
INSERT INTO stats_test values(1);
SELECT count(*) FROM pg_logical_slot_get_changes('regression_slot_stats1', NULL, NULL, 'skip-empty-xacts', '1');
count
-------
3
(1 row)
SELECT count(*) FROM pg_logical_slot_get_changes('regression_slot_stats2', NULL, NULL, 'skip-empty-xacts', '1');
count
-------
3
(1 row)
SELECT count(*) FROM pg_logical_slot_get_changes('regression_slot_stats3', NULL, NULL, 'skip-empty-xacts', '1');
count
-------
3
(1 row)
SELECT pg_stat_force_next_flush();
pg_stat_force_next_flush
--------------------------
(1 row)
SELECT slot_name, spill_txns = 0 AS spill_txns, spill_count = 0 AS spill_count, total_txns > 0 AS total_txns, total_bytes > 0 AS total_bytes FROM pg_stat_replication_slots ORDER BY slot_name;
slot_name | spill_txns | spill_count | total_txns | total_bytes
------------------------+------------+-------------+------------+-------------
regression_slot_stats1 | t | t | t | t
regression_slot_stats2 | t | t | t | t
regression_slot_stats3 | t | t | t | t
(3 rows)
RESET logical_decoding_work_mem;
-- reset stats for one slot, others should be unaffected
SELECT pg_stat_reset_replication_slot('regression_slot_stats1');
pg_stat_reset_replication_slot
--------------------------------
(1 row)
SELECT slot_name, spill_txns = 0 AS spill_txns, spill_count = 0 AS spill_count, total_txns > 0 AS total_txns, total_bytes > 0 AS total_bytes FROM pg_stat_replication_slots ORDER BY slot_name;
slot_name | spill_txns | spill_count | total_txns | total_bytes
------------------------+------------+-------------+------------+-------------
regression_slot_stats1 | t | t | f | f
regression_slot_stats2 | t | t | t | t
regression_slot_stats3 | t | t | t | t
(3 rows)
-- reset stats for all slots
SELECT pg_stat_reset_replication_slot(NULL);
pg_stat_reset_replication_slot
--------------------------------
(1 row)
SELECT slot_name, spill_txns = 0 AS spill_txns, spill_count = 0 AS spill_count, total_txns > 0 AS total_txns, total_bytes > 0 AS total_bytes FROM pg_stat_replication_slots ORDER BY slot_name;
slot_name | spill_txns | spill_count | total_txns | total_bytes
------------------------+------------+-------------+------------+-------------
regression_slot_stats1 | t | t | f | f
regression_slot_stats2 | t | t | f | f
regression_slot_stats3 | t | t | f | f
(3 rows)
-- verify accessing/resetting stats for non-existent slot does something reasonable
SELECT * FROM pg_stat_get_replication_slot('do-not-exist');
slot_name | spill_txns | spill_count | spill_bytes | stream_txns | stream_count | stream_bytes | total_txns | total_bytes | stats_reset
--------------+------------+-------------+-------------+-------------+--------------+--------------+------------+-------------+-------------
do-not-exist | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(1 row)
SELECT pg_stat_reset_replication_slot('do-not-exist');
ERROR: replication slot "do-not-exist" does not exist
SELECT * FROM pg_stat_get_replication_slot('do-not-exist');
slot_name | spill_txns | spill_count | spill_bytes | stream_txns | stream_count | stream_bytes | total_txns | total_bytes | stats_reset
--------------+------------+-------------+-------------+-------------+--------------+--------------+------------+-------------+-------------
do-not-exist | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
(1 row)
-- spilling the xact
BEGIN;
INSERT INTO stats_test SELECT 'serialize-topbig--1:'||g.i FROM generate_series(1, 5000) g(i);
COMMIT;
SELECT count(*) FROM pg_logical_slot_peek_changes('regression_slot_stats1', NULL, NULL, 'skip-empty-xacts', '1');
count
-------
5002
(1 row)
-- Check stats. We can't test the exact stats count as that can vary if any
-- background transaction (say by autovacuum) happens in parallel to the main
-- transaction.
SELECT pg_stat_force_next_flush();
pg_stat_force_next_flush
--------------------------
(1 row)
SELECT slot_name, spill_txns > 0 AS spill_txns, spill_count > 0 AS spill_count FROM pg_stat_replication_slots;
slot_name | spill_txns | spill_count
------------------------+------------+-------------
regression_slot_stats1 | t | t
regression_slot_stats2 | f | f
regression_slot_stats3 | f | f
(3 rows)
-- Ensure stats can be repeatedly accessed using the same stats snapshot. See
-- https://postgr.es/m/20210317230447.c7uc4g3vbs4wi32i%40alap3.anarazel.de
BEGIN;
SELECT slot_name FROM pg_stat_replication_slots;
slot_name
------------------------
regression_slot_stats1
regression_slot_stats2
regression_slot_stats3
(3 rows)
SELECT slot_name FROM pg_stat_replication_slots;
slot_name
------------------------
regression_slot_stats1
regression_slot_stats2
regression_slot_stats3
(3 rows)
COMMIT;
SELECT 'init' FROM pg_create_logical_replication_slot('regression_slot_stats4_twophase', 'test_decoding', false, true) s4;
?column?
----------
init
(1 row)
-- The INSERT changes are large enough to be spilled but will not be, because
-- the transaction is aborted. The logical decoding skips collecting further
-- changes too. The transaction is prepared to make sure the decoding processes
-- the aborted transaction.
BEGIN;
INSERT INTO stats_test SELECT 'serialize-toobig--1:'||g.i FROM generate_series(1, 5000) g(i);
PREPARE TRANSACTION 'test1_abort';
ROLLBACK PREPARED 'test1_abort';
SELECT count(*) FROM pg_logical_slot_get_changes('regression_slot_stats4_twophase', NULL, NULL, 'include-xids', '0', 'skip-empty-xacts', '1');
count
-------
1
(1 row)
-- Verify that the decoding doesn't spill already-aborted transaction's changes.
SELECT pg_stat_force_next_flush();
pg_stat_force_next_flush
--------------------------
(1 row)
SELECT slot_name, spill_txns, spill_count FROM pg_stat_replication_slots WHERE slot_name = 'regression_slot_stats4_twophase';
slot_name | spill_txns | spill_count
---------------------------------+------------+-------------
regression_slot_stats4_twophase | 0 | 0
(1 row)
DROP TABLE stats_test;
SELECT pg_drop_replication_slot('regression_slot_stats1'),
pg_drop_replication_slot('regression_slot_stats2'),
pg_drop_replication_slot('regression_slot_stats3'),
pg_drop_replication_slot('regression_slot_stats4_twophase');
pg_drop_replication_slot | pg_drop_replication_slot | pg_drop_replication_slot | pg_drop_replication_slot
--------------------------+--------------------------+--------------------------+--------------------------
| | |
(1 row)