[Columnar] 11.0 Cherry-Pick Bugfix for Columnar: options ignored during ALTER TABLE (#6410)

DESCRIPTION: Fixes a bug that prevents retaining columnar table options
after a table-rewrite A fix for this issue: Columnar: options ignored
during ALTER TABLE rewrite #5927
The OID for the temporary table created during ALTER TABLE was not the
same as the original table's OID so the columnar options were not being
applied during rewrite.

The change is that I applied the original table's columnar options to
the new table so that it has the correct options during write. I also
added a test.

cherry-pick from commit f21cbe68f8

src/backend/columnar/columnar_tableam.c

@@ -742,7 +742,9 @@ columnar_tuple_insert(Relation relation, TupleTableSlot *slot, CommandId cid,
 	 */
 	ColumnarWriteState *writeState = columnar_init_write_state(relation,
 															   RelationGetDescr(relation),
+															   slot->tts_tableOid,
 															   GetCurrentSubTransactionId());
+
 	MemoryContext oldContext = MemoryContextSwitchTo(ColumnarWritePerTupleContext(
 														 writeState));
 
@@ -784,8 +786,14 @@ columnar_multi_insert(Relation relation, TupleTableSlot **slots, int ntuples,
 {
 	CheckCitusColumnarVersion(ERROR);
 
+	/*
+	 * The callback to .multi_insert is table_multi_insert() and this is only used for the COPY
+	 * command, so slot[i]->tts_tableoid will always be equal to relation->id. Thus, we can send
+	 * RelationGetRelid(relation) as the tupSlotTableOid
+	 */
 	ColumnarWriteState *writeState = columnar_init_write_state(relation,
 															   RelationGetDescr(relation),
+															   RelationGetRelid(relation),
 															   GetCurrentSubTransactionId());
 
 	ColumnarCheckLogicalReplication(relation);

src/backend/columnar/write_state_management.c

@@ -117,6 +117,7 @@ CleanupWriteStateMap(void *arg)
 
 ColumnarWriteState *
 columnar_init_write_state(Relation relation, TupleDesc tupdesc,
+						  Oid tupSlotRelationId,
 						  SubTransactionId currentSubXid)
 {
 	bool found;
@@ -180,7 +181,16 @@ columnar_init_write_state(Relation relation, TupleDesc tupdesc,
 	MemoryContext oldContext = MemoryContextSwitchTo(WriteStateContext);
 
 	ColumnarOptions columnarOptions = { 0 };
-	ReadColumnarOptions(relation->rd_id, &columnarOptions);
+
+	/*
+	 * In case of a table rewrite, we need to fetch table options based on the
+	 * relation id of the source tuple slot.
+	 *
+	 * For this reason, we always pass tupSlotRelationId here; which should be
+	 * same as the target table if the write operation is not related to a table
+	 * rewrite etc.
+	 */
+	ReadColumnarOptions(tupSlotRelationId, &columnarOptions);
 
 	SubXidWriteState *stackEntry = palloc0(sizeof(SubXidWriteState));
 	stackEntry->writeState = ColumnarBeginWrite(relation->rd_node,

src/include/columnar/columnar.h

@@ -302,6 +302,7 @@ extern Datum columnar_relation_storageid(PG_FUNCTION_ARGS);
 /* write_state_management.c */
 extern ColumnarWriteState * columnar_init_write_state(Relation relation, TupleDesc
 													  tupdesc,
+													  Oid tupSlotRelationId,
 													  SubTransactionId currentSubXid);
 extern void FlushWriteStateForRelfilenode(Oid relfilenode, SubTransactionId
 										  currentSubXid);

src/test/regress/expected/columnar_alter_set_type.out

@@ -50,3 +50,33 @@ SELECT * FROM test_alter_table ORDER BY a;
 (4 rows)
 
 DROP TABLE test_alter_table;
+--  Make sure that the correct table options are used when rewriting the table.
+-- This is reflected by the VACUUM VERBOSE output right after a rewrite showing
+-- that all chunks are compressed with the configured compression algorithm
+-- https://github.com/citusdata/citus/issues/5927
+CREATE TABLE test(i int) USING columnar;
+SELECT alter_columnar_table_set('test', compression => 'lz4');
+ alter_columnar_table_set
+---------------------------------------------------------------------
+
+(1 row)
+
+INSERT INTO test VALUES(1);
+VACUUM VERBOSE test;
+INFO:  statistics for "test":
+storage id: xxxxx
+total file size: 24576, total data size: 6
+compression rate: 0.83x
+total row count: 1, stripe count: 1, average rows per stripe: 1
+chunk count: 1, containing data for dropped columns: 0, lz4 compressed: 1
+
+ALTER TABLE test ALTER COLUMN i TYPE int8;
+VACUUM VERBOSE test;
+INFO:  statistics for "test":
+storage id: xxxxx
+total file size: 24576, total data size: 10
+compression rate: 0.90x
+total row count: 1, stripe count: 1, average rows per stripe: 1
+chunk count: 1, containing data for dropped columns: 0, lz4 compressed: 1
+
+DROP TABLE test;

src/test/regress/sql/columnar_alter_set_type.sql

@@ -29,3 +29,17 @@ ALTER TABLE test_alter_table ALTER COLUMN b TYPE float USING (b::float + 0.5);
 SELECT * FROM test_alter_table ORDER BY a;
 
 DROP TABLE test_alter_table;
+
+-- Make sure that the correct table options are used when rewriting the table.
+-- This is reflected by the VACUUM VERBOSE output right after a rewrite showing
+-- that all chunks are compressed with the configured compression algorithm
+-- https://github.com/citusdata/citus/issues/5927
+CREATE TABLE test(i int) USING columnar;
+SELECT alter_columnar_table_set('test', compression => 'lz4');
+INSERT INTO test VALUES(1);
+VACUUM VERBOSE test;
+
+ALTER TABLE test ALTER COLUMN i TYPE int8;
+VACUUM VERBOSE test;
+
+DROP TABLE test;