From 5051be86ff02d05751d8908aad3624abd584e944 Mon Sep 17 00:00:00 2001
From: Ahmet Gedemenli <afgedemenli@gmail.com>
Date: Tue, 4 Jul 2023 15:19:07 +0300
Subject: [PATCH] Skip distributed schema insertion into pg_dist_schema, if
 already exists (#7044)

Inserting into `pg_dist_schema` causes unexpected duplicate key errors,
for distributed schemas that already exist. With this commit we skip the
insertion if the schema already exists in `pg_dist_schema`.

The error:
```sql
SET citus.enable_schema_based_sharding TO ON;
CREATE SCHEMA sc2;
CREATE SCHEMA IF NOT EXISTS sc2;
NOTICE:  schema "sc2" already exists, skipping
ERROR:  duplicate key value violates unique constraint "pg_dist_schema_pkey"
DETAIL:  Key (schemaid)=(17294) already exists.
```

fixes: #7042
---
 src/backend/distributed/commands/schema.c     | 31 ++++++++++++-------
 .../expected/schema_based_sharding.out        |  2 ++
 .../regress/sql/schema_based_sharding.sql     |  1 +
 3 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/src/backend/distributed/commands/schema.c b/src/backend/distributed/commands/schema.c
index 966a264d6..6eaacc993 100644
--- a/src/backend/distributed/commands/schema.c
+++ b/src/backend/distributed/commands/schema.c
@@ -103,20 +103,29 @@ PostprocessCreateSchemaStmt(Node *node, const char *queryString)
 		}
 
 		/*
-		 * Register the tenant schema on the coordinator and save the command
-		 * to register it on the workers.
+		 * Skip if the schema is already inserted into pg_dist_schema.
+		 * This could occur when trying to create an already existing schema,
+		 * with IF NOT EXISTS clause.
 		 */
-		int shardCount = 1;
-		int replicationFactor = 1;
-		Oid distributionColumnType = InvalidOid;
-		Oid distributionColumnCollation = InvalidOid;
-		uint32 colocationId = CreateColocationGroup(
-			shardCount, replicationFactor, distributionColumnType,
-			distributionColumnCollation);
+		if (!IsTenantSchema(schemaId))
+		{
+			/*
+			 * Register the tenant schema on the coordinator and save the command
+			 * to register it on the workers.
+			 */
+			int shardCount = 1;
+			int replicationFactor = 1;
+			Oid distributionColumnType = InvalidOid;
+			Oid distributionColumnCollation = InvalidOid;
+			uint32 colocationId = CreateColocationGroup(
+				shardCount, replicationFactor, distributionColumnType,
+				distributionColumnCollation);
 
-		InsertTenantSchemaLocally(schemaId, colocationId);
+			InsertTenantSchemaLocally(schemaId, colocationId);
 
-		commands = lappend(commands, TenantSchemaInsertCommand(schemaId, colocationId));
+			commands = lappend(commands, TenantSchemaInsertCommand(schemaId,
+																   colocationId));
+		}
 	}
 
 	commands = lappend(commands, ENABLE_DDL_PROPAGATION);
diff --git a/src/test/regress/expected/schema_based_sharding.out b/src/test/regress/expected/schema_based_sharding.out
index d7b7b4710..c7e515df4 100644
--- a/src/test/regress/expected/schema_based_sharding.out
+++ b/src/test/regress/expected/schema_based_sharding.out
@@ -50,6 +50,8 @@ SELECT COUNT(*)=0 FROM pg_dist_schema WHERE schemaid::regnamespace::text = 'regu
 
 -- empty tenant
 CREATE SCHEMA "tenant\'_1";
+CREATE SCHEMA IF NOT EXISTS "tenant\'_1";
+NOTICE:  schema "tenant\'_1" already exists, skipping
 -- non-empty tenant
 CREATE SCHEMA "tenant\'_2";
 CREATE TABLE "tenant\'_2".test_table(a int, b text);
diff --git a/src/test/regress/sql/schema_based_sharding.sql b/src/test/regress/sql/schema_based_sharding.sql
index 1e5208332..dd0c20d1c 100644
--- a/src/test/regress/sql/schema_based_sharding.sql
+++ b/src/test/regress/sql/schema_based_sharding.sql
@@ -33,6 +33,7 @@ SELECT COUNT(*)=0 FROM pg_dist_schema WHERE schemaid::regnamespace::text = 'regu
 
 -- empty tenant
 CREATE SCHEMA "tenant\'_1";
+CREATE SCHEMA IF NOT EXISTS "tenant\'_1";
 
 -- non-empty tenant
 CREATE SCHEMA "tenant\'_2";