Add GUC for segmentwise recompression

Add GUC option to enable or disable segmentwise recompression. If disabled, then a full recompression is done instead when recompression is attempted through `compress_chunk`. If `recompress_chunk_segmentwise` is used when GUC is disabled, then an error is thrown.
timescale · Oct 31, 2024 · 796d690 · 796d690
1 parent 76e3b27
commit 796d690
Show file tree

Hide file tree

Showing 6 changed files with 109 additions and 1 deletion.
diff --git a/.unreleased/pr_7413 b/.unreleased/pr_7413
@@ -0,0 +1 @@
+Implements: #7413: Add GUC for segmentwise recompression.
diff --git a/src/guc.c b/src/guc.c
@@ -139,6 +139,7 @@ TSDLLEXPORT bool ts_guc_enable_compression_indexscan = false;
 TSDLLEXPORT bool ts_guc_enable_bulk_decompression = true;
 TSDLLEXPORT bool ts_guc_auto_sparse_indexes = true;
 bool ts_guc_enable_chunk_skipping = false;
+TSDLLEXPORT bool ts_guc_enable_segmentwise_recompression = true;
 
 /* Enable of disable columnar scans for columnar-oriented storage engines. If
  * disabled, regular sequence scans will be used instead. */
@@ -703,6 +704,17 @@ _guc_init(void)
 							 NULL,
 							 NULL);
 
+	DefineCustomBoolVariable(MAKE_EXTOPTION("enable_segmentwise_recompression"),
+							 "Enable segmentwise recompression functionality",
+							 "Enable segmentwise recompression",
+							 &ts_guc_enable_segmentwise_recompression,
+							 true,
+							 PGC_USERSET,
+							 0,
+							 NULL,
+							 NULL,
+							 NULL);
+
 	/*
 	 * Define the limit on number of invalidation-based refreshes we allow per
 	 * refresh call. If this limit is exceeded, fall back to a single refresh that

diff --git a/src/guc.h b/src/guc.h
@@ -52,6 +52,7 @@ extern bool ts_guc_enable_tss_callbacks;
 extern TSDLLEXPORT bool ts_guc_enable_delete_after_compression;
 extern TSDLLEXPORT bool ts_guc_enable_merge_on_cagg_refresh;
 extern bool ts_guc_enable_chunk_skipping;
+extern TSDLLEXPORT bool ts_guc_enable_segmentwise_recompression;
 
 #ifdef USE_TELEMETRY
 typedef enum TelemetryLevel

diff --git a/tsl/src/compression/api.c b/tsl/src/compression/api.c
@@ -8,6 +8,7 @@
  *  compress and decompress chunks
  */
 #include <postgres.h>
+#include "guc.h"
 #include <access/tableam.h>
 #include <access/xact.h>
 #include <catalog/dependency.h>
@@ -919,12 +920,19 @@ tsl_compress_chunk_wrapper(Chunk *chunk, bool if_not_compressed, bool recompress
 			return uncompressed_chunk_id;
 		}
 
-		if (ts_chunk_is_partial(chunk) && get_compressed_chunk_index_for_recompression(chunk))
+		if (ts_guc_enable_segmentwise_recompression && ts_chunk_is_partial(chunk) &&
+			get_compressed_chunk_index_for_recompression(chunk))
 		{
 			uncompressed_chunk_id = recompress_chunk_segmentwise_impl(chunk);
 		}
 		else
 		{
+			if (!ts_guc_enable_segmentwise_recompression)
+				elog(DEBUG1,
+					 "segmentwise recompression is disabled, performing full recompression on "
+					 "chunk \"%s.%s\"",
+					 NameStr(chunk->fd.schema_name),
+					 NameStr(chunk->fd.table_name));
 			decompress_chunk_impl(chunk, false);
 			compress_chunk_impl(chunk->hypertable_relid, chunk->table_id);
 		}
@@ -1257,6 +1265,14 @@ tsl_recompress_chunk_segmentwise(PG_FUNCTION_ARGS)
 	}
 	else
 	{
+		if (!ts_guc_enable_segmentwise_recompression)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+					 errmsg("segmentwise recompression functionality disabled, "
+							"enable it by first setting "
+							"timescaledb.enable_segmentwise_recompression to on")));
+		}
 		uncompressed_chunk_id = recompress_chunk_segmentwise_impl(chunk);
 	}
 

diff --git a/tsl/test/expected/recompress_chunk_segmentwise.out b/tsl/test/expected/recompress_chunk_segmentwise.out
@@ -593,3 +593,61 @@ select * from :compressed_chunk_name;
               2 | 1 |   | Sat Jan 01 01:00:00 2022 PST | Sat Jan 01 01:00:00 2022 PST | BAAAAneAR/JEAAAAAAAAAAAAAAAAAgAAAAIAAAAAAAAA7gAE7wCP5IgAAATvAI/kh/8= | BAEAAAAAAAAABAAAAAAAAAAEAAAAAQAAAAEAAAAAAAAABAAAAAAAAAAIAAAAAgAAAAEAAAAAAAAAAQAAAAAAAAAB
 (5 rows)
 
+--- Test behaviour when enable_segmentwise_recompression GUC if OFF
+CREATE TABLE guc_test(time timestamptz not null, a int, b int, c int);
+SELECT create_hypertable('guc_test', by_range('time', INTERVAL '1 day'));
+ create_hypertable 
+-------------------
+ (18,t)
+(1 row)
+
+ALTER TABLE guc_test set (timescaledb.compress, timescaledb.compress_segmentby = 'a, b');
+NOTICE:  default order by for hypertable "guc_test" is set to ""time" DESC"
+INSERT INTO guc_test VALUES ('2024-10-30 14:04:00.501519-06'::timestamptz, 1, 1, 1);
+SELECT show_chunks as chunk_to_compress FROM show_chunks('guc_test') LIMIT 1 \gset
+SELECT compress_chunk(:'chunk_to_compress');
+              compress_chunk              
+------------------------------------------
+ _timescaledb_internal._hyper_18_20_chunk
+(1 row)
+
+INSERT INTO guc_test VALUES ('2024-10-30 14:14:00.501519-06'::timestamptz, 1, 1, 2);
+-- When GUC is OFF, recompress function should throw an error
+SET timescaledb.enable_segmentwise_recompression TO OFF;
+\set ON_ERROR_STOP 0
+SELECT _timescaledb_functions.recompress_chunk_segmentwise(:'chunk_to_compress');
+ERROR:  segmentwise recompression functionality disabled, enable it by first setting timescaledb.enable_segmentwise_recompression to on
+\set ON_ERROR_STOP 1
+-- When GUC is OFF, entire chunk should be fully uncompressed and compressed instead
+SET client_min_messages TO DEBUG;
+SELECT compress_chunk(:'chunk_to_compress');
+LOG:  statement: SELECT compress_chunk('_timescaledb_internal._hyper_18_20_chunk');
+DEBUG:  capture_ExecutorStart - rtable #0: (null) (relid: 0)
+DEBUG:  segmentwise recompression is disabled, performing full recompression on chunk "_timescaledb_internal._hyper_18_20_chunk"
+DEBUG:  acquiring locks for decompressing "_timescaledb_internal._hyper_18_20_chunk"
+DEBUG:  locks acquired for decompressing "_timescaledb_internal._hyper_18_20_chunk"
+DEBUG:  finished decompressing 1 rows from "compress_hyper_19_21_chunk"
+DEBUG:  drop auto-cascades to type _timescaledb_internal.compress_hyper_19_21_chunk
+DEBUG:  drop auto-cascades to type _timescaledb_internal.compress_hyper_19_21_chunk[]
+DEBUG:  drop auto-cascades to toast table pg_toast.pg_toast_17376
+DEBUG:  drop auto-cascades to index pg_toast.pg_toast_17376_index
+DEBUG:  drop auto-cascades to index _timescaledb_internal.compress_hyper_19_21_chunk_a_b__ts_meta_min_1__ts_meta_max__idx
+DEBUG:  acquiring locks for compressing "_timescaledb_internal._hyper_18_20_chunk"
+DEBUG:  locks acquired for compressing "_timescaledb_internal._hyper_18_20_chunk"
+DEBUG:  building index "pg_toast_17384_index" on table "pg_toast_17384" serially
+DEBUG:  index "pg_toast_17384_index" can safely use deduplication
+DEBUG:  building index "compress_hyper_19_22_chunk_a_b__ts_meta_min_1__ts_meta_max__idx" on table "compress_hyper_19_22_chunk" serially
+DEBUG:  index "compress_hyper_19_22_chunk_a_b__ts_meta_min_1__ts_meta_max__idx" can safely use deduplication
+DEBUG:  adding index compress_hyper_19_22_chunk_a_b__ts_meta_min_1__ts_meta_max__idx ON _timescaledb_internal.compress_hyper_19_22_chunk USING BTREE(a, b, _ts_meta_min_1 DESC, _ts_meta_max_1 DESC, )
+DEBUG:  new compressed chunk "_timescaledb_internal.compress_hyper_19_22_chunk" created
+DEBUG:  using tuplesort to scan rows from "_hyper_18_20_chunk" for compression
+DEBUG:  finished compressing 2 rows from "_hyper_18_20_chunk"
+DEBUG:  building index "_hyper_18_20_chunk_guc_test_time_idx" on table "_hyper_18_20_chunk" serially
+DEBUG:  index "_hyper_18_20_chunk_guc_test_time_idx" can safely use deduplication
+              compress_chunk              
+------------------------------------------
+ _timescaledb_internal._hyper_18_20_chunk
+(1 row)
+
+RESET client_min_messages;
+LOG:  statement: RESET client_min_messages;
diff --git a/tsl/test/sql/recompress_chunk_segmentwise.sql b/tsl/test/sql/recompress_chunk_segmentwise.sql
@@ -290,3 +290,23 @@ select * from :compressed_chunk_name;
 insert into nullseg_many values (:'start_time', 1, NULL, NULL);
 SELECT compress_chunk(:'chunk_to_compress');
 select * from :compressed_chunk_name;
+
+--- Test behaviour when enable_segmentwise_recompression GUC if OFF
+CREATE TABLE guc_test(time timestamptz not null, a int, b int, c int);
+SELECT create_hypertable('guc_test', by_range('time', INTERVAL '1 day'));
+
+ALTER TABLE guc_test set (timescaledb.compress, timescaledb.compress_segmentby = 'a, b');
+INSERT INTO guc_test VALUES ('2024-10-30 14:04:00.501519-06'::timestamptz, 1, 1, 1);
+SELECT show_chunks as chunk_to_compress FROM show_chunks('guc_test') LIMIT 1 \gset
+SELECT compress_chunk(:'chunk_to_compress');
+
+INSERT INTO guc_test VALUES ('2024-10-30 14:14:00.501519-06'::timestamptz, 1, 1, 2);
+-- When GUC is OFF, recompress function should throw an error
+SET timescaledb.enable_segmentwise_recompression TO OFF;
+\set ON_ERROR_STOP 0
+SELECT _timescaledb_functions.recompress_chunk_segmentwise(:'chunk_to_compress');
+\set ON_ERROR_STOP 1
+-- When GUC is OFF, entire chunk should be fully uncompressed and compressed instead
+SET client_min_messages TO DEBUG;
+SELECT compress_chunk(:'chunk_to_compress');
+RESET client_min_messages;