Bump version to 10.0.3

(cherry picked from commit 92130ae2a2)

CHANGELOG.md

@@ -1,3 +1,66 @@
+### citus v10.0.3 (March 16, 2021) ###
+
+* Prevents infinite recursion for queries that involve `UNION ALL`
+  below `JOIN`
+
+* Fixes a crash in queries with a modifying `CTE` and a `SELECT`
+  without `FROM`
+
+* Fixes upgrade and downgrade paths for `citus_update_table_statistics`
+
+* Fixes a bug that causes `SELECT` queries to use 2PC unnecessarily
+
+* Fixes a bug that might cause self-deadlocks with
+  `CREATE INDEX` / `REINDEX CONCURRENTLY` commands
+
+* Adds `citus.max_cached_connection_lifetime` GUC to set maximum connection
+  lifetime
+
+* Adds `citus.remote_copy_flush_threshold` GUC that controls
+  per-shard memory usages by `COPY`
+
+* Adds `citus_get_active_worker_nodes` UDF to deprecate
+  `master_get_active_worker_nodes`
+
+* Skips 2PC for readonly connections in a transaction
+
+* Makes sure that local execution starts coordinated transaction
+
+* Removes open temporary file warning when cancelling a query with
+  an open tuple store
+
+* Relaxes the locks when adding an existing node
+
+### citus v10.0.2 (March 3, 2021) ###
+
+* Adds a configure flag to enforce security
+
+* Fixes a bug due to cross join without target list
+
+* Fixes a bug with `UNION ALL` on PG 13
+
+* Fixes a compatibility issue with pg_audit in utility calls
+
+* Fixes insert query with CTEs/sublinks/subqueries etc
+
+* Grants `SELECT` permission on `citus_tables` view to `public`
+
+* Grants `SELECT` permission on columnar metadata tables to `public`
+
+* Improves `citus_update_table_statistics` and provides distributed deadlock
+  detection
+
+* Preserves colocation with procedures in `alter_distributed_table`
+
+* Prevents using `alter_columnar_table_set` and `alter_columnar_table_reset`
+  on a columnar table not owned by the user
+
+* Removes limits around long table names
+
+### citus v10.0.1 (February 19, 2021) ###
+
+* Fixes an issue in creation of `pg_catalog.time_partitions` view
+
 ### citus v10.0.0 (February 16, 2021) ###
 
 * Adds support for per-table option for columnar storage

Makefile.global.in

@@ -86,6 +86,7 @@ endif
 
 # Add options passed to configure or computed therein, to CFLAGS/CPPFLAGS/...
 override CFLAGS += @CFLAGS@ @CITUS_CFLAGS@
+override BITCODE_CFLAGS := $(BITCODE_CFLAGS) @CITUS_BITCODE_CFLAGS@
 ifneq ($(GIT_VERSION),)
     override CFLAGS += -DGIT_VERSION=\"$(GIT_VERSION)\"
 endif

configure

@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Citus 10.0devel.
+# Generated by GNU Autoconf 2.69 for Citus 10.0.3.
 #
 #
 # Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
@@ -579,8 +579,8 @@ MAKEFLAGS=
 # Identity of this package.
 PACKAGE_NAME='Citus'
 PACKAGE_TARNAME='citus'
-PACKAGE_VERSION='10.0devel'
-PACKAGE_STRING='Citus 10.0devel'
+PACKAGE_VERSION='10.0.3'
+PACKAGE_STRING='Citus 10.0.3'
 PACKAGE_BUGREPORT=''
 PACKAGE_URL=''
 
@@ -628,8 +628,10 @@ POSTGRES_BUILDDIR
 POSTGRES_SRCDIR
 CITUS_LDFLAGS
 CITUS_CPPFLAGS
+CITUS_BITCODE_CFLAGS
 CITUS_CFLAGS
 GIT_BIN
+with_security_flags
 with_zstd
 with_lz4
 EGREP
@@ -696,6 +698,7 @@ with_libcurl
 with_reports_hostname
 with_lz4
 with_zstd
+with_security_flags
 '
       ac_precious_vars='build_alias
 host_alias
@@ -1258,7 +1261,7 @@ if test "$ac_init_help" = "long"; then
   # Omit some internal or obsolete options to make the list less imposing.
   # This message is too long to be a string in the A/UX 3.1 sh.
   cat <<_ACEOF
-\`configure' configures Citus 10.0devel to adapt to many kinds of systems.
+\`configure' configures Citus 10.0.3 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1320,7 +1323,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of Citus 10.0devel:";;
+     short | recursive ) echo "Configuration of Citus 10.0.3:";;
    esac
   cat <<\_ACEOF
 
@@ -1342,6 +1345,7 @@ Optional Packages:
                           and update checks
   --without-lz4           do not use lz4
   --without-zstd          do not use zstd
+  --with-security-flags   use security flags
 
 Some influential environment variables:
   PG_CONFIG   Location to find pg_config for target PostgreSQL instalation
@@ -1422,7 +1426,7 @@ fi
 test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
-Citus configure 10.0devel
+Citus configure 10.0.3
 generated by GNU Autoconf 2.69
 
 Copyright (C) 2012 Free Software Foundation, Inc.
@@ -1905,7 +1909,7 @@ cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
-It was created by Citus $as_me 10.0devel, which was
+It was created by Citus $as_me 10.0.3, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
@@ -4346,6 +4350,48 @@ if test x"$citusac_cv_prog_cc_cflags__Werror_return_type" = x"yes"; then
   CITUS_CFLAGS="$CITUS_CFLAGS -Werror=return-type"
 fi
 
+# Security flags
+# Flags taken from: https://liquid.microsoft.com/Web/Object/Read/ms.security/Requirements/Microsoft.Security.SystemsADM.10203#guide
+# We do not enforce the following flag because it is only available on GCC>=8
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC supports -fstack-clash-protection" >&5
+$as_echo_n "checking whether $CC supports -fstack-clash-protection... " >&6; }
+if ${citusac_cv_prog_cc_cflags__fstack_clash_protection+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  citusac_save_CFLAGS=$CFLAGS
+flag=-fstack-clash-protection
+case $flag in -Wno*)
+	 flag=-W$(echo $flag | cut -c 6-)
+esac
+CFLAGS="$citusac_save_CFLAGS $flag"
+ac_save_c_werror_flag=$ac_c_werror_flag
+ac_c_werror_flag=yes
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  citusac_cv_prog_cc_cflags__fstack_clash_protection=yes
+else
+  citusac_cv_prog_cc_cflags__fstack_clash_protection=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+ac_c_werror_flag=$ac_save_c_werror_flag
+CFLAGS="$citusac_save_CFLAGS"
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $citusac_cv_prog_cc_cflags__fstack_clash_protection" >&5
+$as_echo "$citusac_cv_prog_cc_cflags__fstack_clash_protection" >&6; }
+if test x"$citusac_cv_prog_cc_cflags__fstack_clash_protection" = x"yes"; then
+  CITUS_CFLAGS="$CITUS_CFLAGS -fstack-clash-protection"
+fi
+
 
 #
 # --enable-coverage enables generation of code coverage metrics with gcov
@@ -4493,8 +4539,8 @@ if test "$version_num" != '11'; then
 $as_echo "#define HAS_TABLEAM 1" >>confdefs.h
 
 else
-   { $as_echo "$as_me:${as_lineno-$LINENO}: postgres version does not support table access methodds" >&5
-$as_echo "$as_me: postgres version does not support table access methodds" >&6;}
+   { $as_echo "$as_me:${as_lineno-$LINENO}: postgres version does not support table access methods" >&5
+$as_echo "$as_me: postgres version does not support table access methods" >&6;}
 fi;
 
 # Require lz4 & zstd only if we are compiling columnar
@@ -4687,6 +4733,55 @@ fi
 
 fi # test "$HAS_TABLEAM" == 'yes'
 
+
+
+
+
+# Check whether --with-security-flags was given.
+if test "${with_security_flags+set}" = set; then :
+  withval=$with_security_flags;
+  case $withval in
+    yes)
+      :
+      ;;
+    no)
+      :
+      ;;
+    *)
+      as_fn_error $? "no argument expected for --with-security-flags option" "$LINENO" 5
+      ;;
+  esac
+
+else
+  with_security_flags=no
+
+fi
+
+
+
+
+if test "$with_security_flags" = yes; then
+# Flags taken from: https://liquid.microsoft.com/Web/Object/Read/ms.security/Requirements/Microsoft.Security.SystemsADM.10203#guide
+
+# We always want to have some compiler flags for security concerns.
+SECURITY_CFLAGS="-fstack-protector-strong -D_FORTIFY_SOURCE=2 -O2 -z noexecstack -fpic -shared -Wl,-z,relro -Wl,-z,now -Wformat -Wformat-security -Werror=format-security"
+CITUS_CFLAGS="$CITUS_CFLAGS $SECURITY_CFLAGS"
+{ $as_echo "$as_me:${as_lineno-$LINENO}: Blindly added security flags for linker: $SECURITY_CFLAGS" >&5
+$as_echo "$as_me: Blindly added security flags for linker: $SECURITY_CFLAGS" >&6;}
+
+# We always want to have some clang flags for security concerns.
+# This doesn't include "-Wl,-z,relro -Wl,-z,now" on purpuse, because bitcode is not linked.
+# This doesn't include -fsanitize=cfi because it breaks builds on many distros including
+# Debian/Buster, Debian/Stretch, Ubuntu/Bionic, Ubuntu/Xenial and EL7.
+SECURITY_BITCODE_CFLAGS="-fsanitize=safe-stack -fstack-protector-strong -flto -fPIC -Wformat -Wformat-security -Werror=format-security"
+CITUS_BITCODE_CFLAGS="$CITUS_BITCODE_CFLAGS $SECURITY_BITCODE_CFLAGS"
+{ $as_echo "$as_me:${as_lineno-$LINENO}: Blindly added security flags for llvm: $SECURITY_BITCODE_CFLAGS" >&5
+$as_echo "$as_me: Blindly added security flags for llvm: $SECURITY_BITCODE_CFLAGS" >&6;}
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: If you run into issues during linking or bitcode compilation, you can use --without-security-flags." >&5
+$as_echo "$as_me: WARNING: If you run into issues during linking or bitcode compilation, you can use --without-security-flags." >&2;}
+fi
+
 # Check if git is installed, when installed the gitref of the checkout will be baked in the application
 # Extract the first word of "git", so it can be a program name with args.
 set dummy git; ac_word=$2
@@ -4752,6 +4847,8 @@ fi
 
 CITUS_CFLAGS="$CITUS_CFLAGS"
 
+CITUS_BITCODE_CFLAGS="$CITUS_BITCODE_CFLAGS"
+
 CITUS_CPPFLAGS="$CITUS_CPPFLAGS"
 
 CITUS_LDFLAGS="$LIBS $CITUS_LDFLAGS"
@@ -5276,7 +5373,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
-This file was extended by Citus $as_me 10.0devel, which was
+This file was extended by Citus $as_me 10.0.3, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -5338,7 +5435,7 @@ _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
-Citus config.status 10.0devel
+Citus config.status 10.0.3
 configured by $0, generated by GNU Autoconf 2.69,
   with options \\"\$ac_cs_config\\"
 

configure.in

@@ -5,7 +5,7 @@
 # everyone needing autoconf installed, the resulting files are checked
 # into the SCM.
 
-AC_INIT([Citus], [10.0devel])
+AC_INIT([Citus], [10.0.3])
 AC_COPYRIGHT([Copyright (c) Citus Data, Inc.])
 
 # we'll need sed and awk for some of the version commands
@@ -174,6 +174,10 @@ CITUSAC_PROG_CC_CFLAGS_OPT([-Werror=vla])  # visual studio does not support thes
 CITUSAC_PROG_CC_CFLAGS_OPT([-Werror=implicit-int])
 CITUSAC_PROG_CC_CFLAGS_OPT([-Werror=implicit-function-declaration])
 CITUSAC_PROG_CC_CFLAGS_OPT([-Werror=return-type])
+# Security flags
+# Flags taken from: https://liquid.microsoft.com/Web/Object/Read/ms.security/Requirements/Microsoft.Security.SystemsADM.10203#guide
+# We do not enforce the following flag because it is only available on GCC>=8
+CITUSAC_PROG_CC_CFLAGS_OPT([-fstack-clash-protection])
 
 #
 # --enable-coverage enables generation of code coverage metrics with gcov
@@ -216,7 +220,7 @@ if test "$version_num" != '11'; then
    HAS_TABLEAM=yes
    AC_DEFINE([HAS_TABLEAM], 1, [Define to 1 to build with table access method support, pg12 and up])
 else
-   AC_MSG_NOTICE([postgres version does not support table access methodds])
+   AC_MSG_NOTICE([postgres version does not support table access methods])
 fi;
 
 # Require lz4 & zstd only if we are compiling columnar
@@ -261,11 +265,36 @@ if test "$HAS_TABLEAM" == 'yes'; then
 
 fi # test "$HAS_TABLEAM" == 'yes'
 
+
+PGAC_ARG_BOOL(with, security-flags, no,
+              [use security flags])
+AC_SUBST(with_security_flags)
+
+if test "$with_security_flags" = yes; then
+# Flags taken from: https://liquid.microsoft.com/Web/Object/Read/ms.security/Requirements/Microsoft.Security.SystemsADM.10203#guide
+
+# We always want to have some compiler flags for security concerns.
+SECURITY_CFLAGS="-fstack-protector-strong -D_FORTIFY_SOURCE=2 -O2 -z noexecstack -fpic -shared -Wl,-z,relro -Wl,-z,now -Wformat -Wformat-security -Werror=format-security"
+CITUS_CFLAGS="$CITUS_CFLAGS $SECURITY_CFLAGS"
+AC_MSG_NOTICE([Blindly added security flags for linker: $SECURITY_CFLAGS])
+
+# We always want to have some clang flags for security concerns.
+# This doesn't include "-Wl,-z,relro -Wl,-z,now" on purpuse, because bitcode is not linked.
+# This doesn't include -fsanitize=cfi because it breaks builds on many distros including
+# Debian/Buster, Debian/Stretch, Ubuntu/Bionic, Ubuntu/Xenial and EL7.
+SECURITY_BITCODE_CFLAGS="-fsanitize=safe-stack -fstack-protector-strong -flto -fPIC -Wformat -Wformat-security -Werror=format-security"
+CITUS_BITCODE_CFLAGS="$CITUS_BITCODE_CFLAGS $SECURITY_BITCODE_CFLAGS"
+AC_MSG_NOTICE([Blindly added security flags for llvm: $SECURITY_BITCODE_CFLAGS])
+
+AC_MSG_WARN([If you run into issues during linking or bitcode compilation, you can use --without-security-flags.])
+fi
+
 # Check if git is installed, when installed the gitref of the checkout will be baked in the application
 AC_PATH_PROG(GIT_BIN, git)
 AC_CHECK_FILE(.git,[HAS_DOTGIT=yes], [HAS_DOTGIT=])
 
 AC_SUBST(CITUS_CFLAGS, "$CITUS_CFLAGS")
+AC_SUBST(CITUS_BITCODE_CFLAGS, "$CITUS_BITCODE_CFLAGS")
 AC_SUBST(CITUS_CPPFLAGS, "$CITUS_CPPFLAGS")
 AC_SUBST(CITUS_LDFLAGS, "$LIBS $CITUS_LDFLAGS")
 AC_SUBST(POSTGRES_SRCDIR, "$POSTGRES_SRCDIR")

src/backend/columnar/cstore_metadata_tables.c

@@ -1087,7 +1087,11 @@ DatumToBytea(Datum value, Form_pg_attribute attrForm)
 	{
 		if (attrForm->attbyval)
 		{
-			store_att_byval(VARDATA(result), value, attrForm->attlen);
+			Datum tmp;
+			store_att_byval(&tmp, value, attrForm->attlen);
+
+			memcpy_s(VARDATA(result), datumLength + VARHDRSZ,
+					 &tmp, attrForm->attlen);
 		}
 		else
 		{

src/backend/columnar/cstore_tableam.c

@@ -1662,6 +1662,8 @@ alter_columnar_table_set(PG_FUNCTION_ARGS)
 							   quote_identifier(RelationGetRelationName(rel)))));
 	}
 
+	EnsureTableOwner(relationId);
+
 	ColumnarOptions options = { 0 };
 	if (!ReadColumnarOptions(relationId, &options))
 	{
@@ -1769,6 +1771,8 @@ alter_columnar_table_reset(PG_FUNCTION_ARGS)
 							   quote_identifier(RelationGetRelationName(rel)))));
 	}
 
+	EnsureTableOwner(relationId);
+
 	ColumnarOptions options = { 0 };
 	if (!ReadColumnarOptions(relationId, &options))
 	{

src/backend/columnar/sql/columnar--10.0-1--10.0-2.sql

@@ -0,0 +1,5 @@
+/* columnar--10.0-1--10.0-2.sql */
+
+-- grant read access for columnar metadata tables to unprivileged user
+GRANT USAGE ON SCHEMA columnar TO PUBLIC;
+GRANT SELECT ON ALL tables IN SCHEMA columnar TO PUBLIC ;

src/backend/columnar/sql/downgrades/columnar--10.0-2--10.0-1.sql

@@ -0,0 +1,5 @@
+/* columnar--10.0-2--10.0-1.sql */
+
+-- revoke read access for columnar metadata tables from unprivileged user
+REVOKE USAGE ON SCHEMA columnar FROM PUBLIC;
+REVOKE SELECT ON ALL tables IN SCHEMA columnar FROM PUBLIC;

src/backend/distributed/citus.control

@@ -1,6 +1,6 @@
 # Citus extension
 comment = 'Citus distributed database'
-default_version = '10.0-1'
+default_version = '10.0-3'
 module_pathname = '$libdir/citus'
 relocatable = false
 schema = pg_catalog

src/backend/distributed/commands/alter_table.c

@@ -43,12 +43,15 @@
 #include "distributed/listutils.h"
 #include "distributed/local_executor.h"
 #include "distributed/metadata/dependency.h"
+#include "distributed/metadata/distobject.h"
 #include "distributed/metadata_cache.h"
 #include "distributed/metadata_sync.h"
 #include "distributed/multi_executor.h"
 #include "distributed/multi_logical_planner.h"
 #include "distributed/multi_partitioning_utils.h"
 #include "distributed/reference_table_utils.h"
+#include "distributed/relation_access_tracking.h"
+#include "distributed/shard_utils.h"
 #include "distributed/worker_protocol.h"
 #include "distributed/worker_transaction.h"
 #include "executor/spi.h"
@@ -175,6 +178,8 @@ static TableConversionReturn * AlterDistributedTable(TableConversionParameters *
 static TableConversionReturn * AlterTableSetAccessMethod(
 	TableConversionParameters *params);
 static TableConversionReturn * ConvertTable(TableConversionState *con);
+static bool SwitchToSequentialAndLocalExecutionIfShardNameTooLong(char *relationName,
+																  char *longestShardName);
 static void EnsureTableNotReferencing(Oid relationId, char conversionType);
 static void EnsureTableNotReferenced(Oid relationId, char conversionType);
 static void EnsureTableNotForeign(Oid relationId);
@@ -511,6 +516,10 @@ ConvertTable(TableConversionState *con)
 	bool oldEnableLocalReferenceForeignKeys = EnableLocalReferenceForeignKeys;
 	SetLocalEnableLocalReferenceForeignKeys(false);
 
+	/* switch to sequential execution if shard names will be too long */
+	SwitchToSequentialAndLocalExecutionIfRelationNameTooLong(con->relationId,
+															 con->relationName);
+
 	if (con->conversionType == UNDISTRIBUTE_TABLE && con->cascadeViaForeignKeys &&
 		(TableReferencing(con->relationId) || TableReferenced(con->relationId)))
 	{
@@ -673,7 +682,7 @@ ConvertTable(TableConversionState *con)
 		Node *parseTree = ParseTreeNode(tableCreationSql);
 
 		RelayEventExtendNames(parseTree, con->schemaName, con->hashOfName);
-		ProcessUtilityParseTree(parseTree, tableCreationSql, PROCESS_UTILITY_TOPLEVEL,
+		ProcessUtilityParseTree(parseTree, tableCreationSql, PROCESS_UTILITY_QUERY,
 								NULL, None_Receiver, NULL);
 	}
 
@@ -711,6 +720,32 @@ ConvertTable(TableConversionState *con)
 		CreateCitusTableLike(con);
 	}
 
+	/* preserve colocation with procedures/functions */
+	if (con->conversionType == ALTER_DISTRIBUTED_TABLE)
+	{
+		/*
+		 * Updating the colocationId of functions is always desirable for
+		 * the following scenario:
+		 *    we have shardCount or colocateWith change
+		 *    AND  entire co-location group is altered
+		 * The reason for the second condition is because we currently don't
+		 * remember the original table specified in the colocateWith when
+		 * distributing the function. We only remember the colocationId in
+		 * pg_dist_object table.
+		 */
+		if ((!con->shardCountIsNull || con->colocateWith != NULL) &&
+			(con->cascadeToColocated == CASCADE_TO_COLOCATED_YES || list_length(
+				 con->colocatedTableList) == 1) && con->distributionColumn == NULL)
+		{
+			/*
+			 * Update the colocationId from the one of the old relation to the one
+			 * of the new relation for all tuples in citus.pg_dist_object
+			 */
+			UpdateDistributedObjectColocationId(TableColocationId(con->relationId),
+												TableColocationId(con->newRelationId));
+		}
+	}
+
 	ReplaceTable(con->relationId, con->newRelationId, justBeforeDropCommands,
 				 con->suppressNoticeMessages);
 
@@ -728,7 +763,7 @@ ConvertTable(TableConversionState *con)
 		Node *parseTree = ParseTreeNode(attachPartitionCommand);
 
 		ProcessUtilityParseTree(parseTree, attachPartitionCommand,
-								PROCESS_UTILITY_TOPLEVEL,
+								PROCESS_UTILITY_QUERY,
 								NULL, None_Receiver, NULL);
 	}
 
@@ -1134,7 +1169,7 @@ ReplaceTable(Oid sourceId, Oid targetId, List *justBeforeDropCommands,
 	{
 		if (!suppressNoticeMessages)
 		{
-			ereport(NOTICE, (errmsg("Moving the data of %s",
+			ereport(NOTICE, (errmsg("moving the data of %s",
 									quote_qualified_identifier(schemaName, sourceName))));
 		}
 
@@ -1207,7 +1242,7 @@ ReplaceTable(Oid sourceId, Oid targetId, List *justBeforeDropCommands,
 
 	if (!suppressNoticeMessages)
 	{
-		ereport(NOTICE, (errmsg("Dropping the old %s",
+		ereport(NOTICE, (errmsg("dropping the old %s",
 								quote_qualified_identifier(schemaName, sourceName))));
 	}
 
@@ -1218,7 +1253,7 @@ ReplaceTable(Oid sourceId, Oid targetId, List *justBeforeDropCommands,
 
 	if (!suppressNoticeMessages)
 	{
-		ereport(NOTICE, (errmsg("Renaming the new table to %s",
+		ereport(NOTICE, (errmsg("renaming the new table to %s",
 								quote_qualified_identifier(schemaName, sourceName))));
 	}
 
@@ -1572,3 +1607,104 @@ ExecuteQueryViaSPI(char *query, int SPIOK)
 		ereport(ERROR, (errmsg("could not finish SPI connection")));
 	}
 }
+
+
+/*
+ * SwitchToSequentialAndLocalExecutionIfRelationNameTooLong generates the longest shard name
+ * on the shards of a distributed table, and if exceeds the limit switches to sequential and
+ * local execution to prevent self-deadlocks.
+ *
+ * In case of a RENAME, the relation name parameter should store the new table name, so
+ * that the function can generate shard names of the renamed relations
+ */
+void
+SwitchToSequentialAndLocalExecutionIfRelationNameTooLong(Oid relationId,
+														 char *finalRelationName)
+{
+	if (!IsCitusTable(relationId))
+	{
+		return;
+	}
+
+	if (ShardIntervalCount(relationId) == 0)
+	{
+		/*
+		 * Relation has no shards, so we cannot run into "long shard relation
+		 * name" issue.
+		 */
+		return;
+	}
+
+	char *longestShardName = GetLongestShardName(relationId, finalRelationName);
+	bool switchedToSequentialAndLocalExecution =
+		SwitchToSequentialAndLocalExecutionIfShardNameTooLong(finalRelationName,
+															  longestShardName);
+
+	if (switchedToSequentialAndLocalExecution)
+	{
+		return;
+	}
+
+	if (PartitionedTable(relationId))
+	{
+		Oid longestNamePartitionId = PartitionWithLongestNameRelationId(relationId);
+		if (!OidIsValid(longestNamePartitionId))
+		{
+			/* no partitions have been created yet */
+			return;
+		}
+
+		char *longestPartitionName = get_rel_name(longestNamePartitionId);
+		char *longestPartitionShardName = GetLongestShardName(longestNamePartitionId,
+															  longestPartitionName);
+
+		SwitchToSequentialAndLocalExecutionIfShardNameTooLong(longestPartitionName,
+															  longestPartitionShardName);
+	}
+}
+
+
+/*
+ * SwitchToSequentialAndLocalExecutionIfShardNameTooLong switches to sequential and local
+ * execution if the shard name is too long.
+ *
+ * returns true if switched to sequential and local execution.
+ */
+static bool
+SwitchToSequentialAndLocalExecutionIfShardNameTooLong(char *relationName,
+													  char *longestShardName)
+{
+	if (strlen(longestShardName) >= NAMEDATALEN - 1)
+	{
+		if (ParallelQueryExecutedInTransaction())
+		{
+			/*
+			 * If there has already been a parallel query executed, the sequential mode
+			 * would still use the already opened parallel connections to the workers,
+			 * thus contradicting our purpose of using sequential mode.
+			 */
+			ereport(ERROR, (errmsg(
+								"Shard name (%s) for table (%s) is too long and could "
+								"lead to deadlocks when executed in a transaction "
+								"block after a parallel query", longestShardName,
+								relationName),
+							errhint("Try re-running the transaction with "
+									"\"SET LOCAL citus.multi_shard_modify_mode TO "
+									"\'sequential\';\"")));
+		}
+		else
+		{
+			elog(DEBUG1, "the name of the shard (%s) for relation (%s) is too long, "
+						 "switching to sequential and local execution mode to prevent "
+						 "self deadlocks",
+				 longestShardName, relationName);
+
+			SetLocalMultiShardModifyModeToSequential();
+			SetLocalExecutionStatus(LOCAL_EXECUTION_REQUIRED);
+
+			return true;
+		}
+	}
+
+	return false;
+}

src/backend/distributed/commands/cascade_table_operation_for_connected_relations.c

@@ -510,6 +510,6 @@ ExecuteForeignKeyCreateCommand(const char *commandString, bool skip_validation)
 								"command \"%s\"", commandString)));
 	}
 
-	ProcessUtilityParseTree(parseTree, commandString, PROCESS_UTILITY_TOPLEVEL,
+	ProcessUtilityParseTree(parseTree, commandString, PROCESS_UTILITY_QUERY,
 							NULL, None_Receiver, NULL);
 }

src/backend/distributed/commands/index.c

@@ -411,15 +411,16 @@ static char *
 GenerateLongestShardPartitionIndexName(IndexStmt *createIndexStatement)
 {
 	Oid relationId = CreateIndexStmtGetRelationId(createIndexStatement);
-	char *longestPartitionName = LongestPartitionName(relationId);
-	if (longestPartitionName == NULL)
+	Oid longestNamePartitionId = PartitionWithLongestNameRelationId(relationId);
+	if (!OidIsValid(longestNamePartitionId))
 	{
 		/* no partitions have been created yet */
 		return NULL;
 	}
 
-	char *longestPartitionShardName = pstrdup(longestPartitionName);
-	ShardInterval *shardInterval = LoadShardIntervalWithLongestShardName(relationId);
+	char *longestPartitionShardName = get_rel_name(longestNamePartitionId);
+	ShardInterval *shardInterval = LoadShardIntervalWithLongestShardName(
+		longestNamePartitionId);
 	AppendShardIdToName(&longestPartitionShardName, shardInterval->shardId);
 
 	IndexStmt *createLongestShardIndexStmt = copyObject(createIndexStatement);

src/backend/distributed/commands/multi_copy.c

@@ -2244,7 +2244,7 @@ CitusCopyDestReceiverStartup(DestReceiver *dest, int operation,
 	if (cacheEntry->replicationModel == REPLICATION_MODEL_2PC ||
 		MultiShardCommitProtocol == COMMIT_PROTOCOL_2PC)
 	{
-		CoordinatedTransactionUse2PC();
+		CoordinatedTransactionShouldUse2PC();
 	}
 
 	/* define how tuples will be serialised */

src/backend/distributed/commands/rename.c

@@ -109,6 +109,13 @@ PreprocessRenameStmt(Node *node, const char *renameCommand,
 	 */
 	ErrorIfUnsupportedRenameStmt(renameStmt);
 
+	if (renameStmt->renameType == OBJECT_TABLE ||
+		renameStmt->renameType == OBJECT_FOREIGN_TABLE)
+	{
+		SwitchToSequentialAndLocalExecutionIfRelationNameTooLong(tableRelationId,
+																 renameStmt->newname);
+	}
+
 	DDLJob *ddlJob = palloc0(sizeof(DDLJob));
 	ddlJob->targetRelationId = tableRelationId;
 	ddlJob->concurrentIndexCmd = false;

src/backend/distributed/commands/utility_hook.c

@@ -910,6 +910,19 @@ ExecuteDistributedDDLJob(DDLJob *ddlJob)
 		 */
 		if (ddlJob->startNewTransaction)
 		{
+			/*
+			 * If cache is not populated, system catalog lookups will cause
+			 * the xmin of current backend to change. Then the last phase
+			 * of CREATE INDEX CONCURRENTLY, which is in a separate backend,
+			 * will hang waiting for our backend and result in a deadlock.
+			 *
+			 * We populate the cache before starting the next transaction to
+			 * avoid this. Most of the metadata has already been resolved in
+			 * planning phase, we only need to lookup metadata needed for
+			 * connection establishment.
+			 */
+			(void) CurrentDatabaseName();
+
 			CommitTransactionCommand();
 			StartTransactionCommand();
 		}

src/backend/distributed/connection/connection_management.c

@@ -32,6 +32,7 @@
 #include "distributed/shared_connection_stats.h"
 #include "distributed/cancel_utils.h"
 #include "distributed/remote_commands.h"
+#include "distributed/time_constants.h"
 #include "distributed/version_compat.h"
 #include "distributed/worker_log_messages.h"
 #include "mb/pg_wchar.h"
@@ -43,6 +44,7 @@
 
 int NodeConnectionTimeout = 30000;
 int MaxCachedConnectionsPerWorker = 1;
+int MaxCachedConnectionLifetime = 10 * MS_PER_MINUTE;
 
 HTAB *ConnectionHash = NULL;
 HTAB *ConnParamsHash = NULL;
@@ -1288,6 +1290,7 @@ AfterXactHostConnectionHandling(ConnectionHashEntry *entry, bool isCommit)
  * - Connection is forced to close at the end of transaction
  * - Connection is not in OK state
  * - A transaction is still in progress (usually because we are cancelling a distributed transaction)
+ * - A connection reached its maximum lifetime
  */
 static bool
 ShouldShutdownConnection(MultiConnection *connection, const int cachedConnectionCount)
@@ -1303,7 +1306,10 @@ ShouldShutdownConnection(MultiConnection *connection, const int cachedConnection
 		   cachedConnectionCount >= MaxCachedConnectionsPerWorker ||
 		   connection->forceCloseAtTransactionEnd ||
 		   PQstatus(connection->pgConn) != CONNECTION_OK ||
-		   !RemoteTransactionIdle(connection);
+		   !RemoteTransactionIdle(connection) ||
+		   (MaxCachedConnectionLifetime >= 0 &&
+			TimestampDifferenceExceeds(connection->connectionStart, GetCurrentTimestamp(),
+									   MaxCachedConnectionLifetime));
 }
 
 

src/backend/distributed/connection/remote_commands.c

@@ -25,7 +25,11 @@
 #include "utils/palloc.h"
 
 
-#define MAX_PUT_COPY_DATA_BUFFER_SIZE (8 * 1024 * 1024)
+/*
+ * Setting that controls how many bytes of COPY data libpq is allowed to buffer
+ * internally before we force a flush.
+ */
+int RemoteCopyFlushThreshold = 8 * 1024 * 1024;
 
 
 /* GUC, determining whether statements sent to remote nodes are logged */
@@ -620,7 +624,7 @@ PutRemoteCopyData(MultiConnection *connection, const char *buffer, int nbytes)
 	 */
 
 	connection->copyBytesWrittenSinceLastFlush += nbytes;
-	if (connection->copyBytesWrittenSinceLastFlush > MAX_PUT_COPY_DATA_BUFFER_SIZE)
+	if (connection->copyBytesWrittenSinceLastFlush > RemoteCopyFlushThreshold)
 	{
 		connection->copyBytesWrittenSinceLastFlush = 0;
 		return FinishConnectionIO(connection, allowInterrupts);

src/backend/distributed/executor/adaptive_executor.c

@@ -1165,23 +1165,6 @@ DecideTransactionPropertiesForTaskList(RowModifyLevel modLevel, List *taskList,
 		return xactProperties;
 	}
 
-	if (GetCurrentLocalExecutionStatus() == LOCAL_EXECUTION_REQUIRED)
-	{
-		/*
-		 * In case localExecutionHappened, we force the executor to use 2PC.
-		 * The primary motivation is that at this point we're definitely expanding
-		 * the nodes participated in the transaction. And, by re-generating the
-		 * remote task lists during local query execution, we might prevent the adaptive
-		 * executor to kick-in 2PC (or even start coordinated transaction, that's why
-		 * we prefer adding this check here instead of
-		 * Activate2PCIfModifyingTransactionExpandsToNewNode()).
-		 */
-		xactProperties.errorOnAnyFailure = true;
-		xactProperties.useRemoteTransactionBlocks = TRANSACTION_BLOCKS_REQUIRED;
-		xactProperties.requires2PC = true;
-		return xactProperties;
-	}
-
 	if (DistributedExecutionRequiresRollback(taskList))
 	{
 		/* transaction blocks are required if the task list needs to roll back */
@@ -1240,7 +1223,7 @@ StartDistributedExecution(DistributedExecution *execution)
 
 	if (xactProperties->requires2PC)
 	{
-		CoordinatedTransactionUse2PC();
+		CoordinatedTransactionShouldUse2PC();
 	}
 
 	/*
@@ -3197,7 +3180,7 @@ Activate2PCIfModifyingTransactionExpandsToNewNode(WorkerSession *session)
 		 * just opened, which means we're now going to make modifications
 		 * over multiple connections. Activate 2PC!
 		 */
-		CoordinatedTransactionUse2PC();
+		CoordinatedTransactionShouldUse2PC();
 	}
 }
 

src/backend/distributed/executor/citus_custom_scan.c

@@ -342,9 +342,12 @@ CitusBeginModifyScan(CustomScanState *node, EState *estate, int eflags)
 		/*
 		 * At this point, we're about to do the shard pruning for fast-path queries.
 		 * Given that pruning is deferred always for INSERTs, we get here
-		 * !EnableFastPathRouterPlanner  as well.
+		 * !EnableFastPathRouterPlanner  as well. Given that INSERT statements with
+		 * CTEs/sublinks etc are not eligible for fast-path router plan, we get here
+		 * jobQuery->commandType == CMD_INSERT as well.
 		 */
-		Assert(currentPlan->fastPathRouterPlan || !EnableFastPathRouterPlanner);
+		Assert(currentPlan->fastPathRouterPlan || !EnableFastPathRouterPlanner ||
+			   jobQuery->commandType == CMD_INSERT);
 
 		/*
 		 * We can only now decide which shard to use, so we need to build a new task

src/backend/distributed/executor/local_executor.c

@@ -209,6 +209,19 @@ ExecuteLocalTaskListExtended(List *taskList,
 	Oid *parameterTypes = NULL;
 	uint64 totalRowsProcessed = 0;
 
+	/*
+	 * Even if we are executing local tasks, we still enable
+	 * coordinated transaction. This is because
+	 *  (a) we might be in a transaction, and the next commands may
+	 *      require coordinated transaction
+	 *  (b) we might be executing some tasks locally and the others
+	 *      via remote execution
+	 *
+	 * Also, there is no harm enabling coordinated transaction even if
+	 * we only deal with local tasks in the transaction.
+	 */
+	UseCoordinatedTransaction();
+
 	if (paramListInfo != NULL)
 	{
 		/* not used anywhere, so declare here */
@@ -236,6 +249,17 @@ ExecuteLocalTaskListExtended(List *taskList,
 		{
 			SetLocalExecutionStatus(LOCAL_EXECUTION_REQUIRED);
 		}
+
+		if (!ReadOnlyTask(task->taskType))
+		{
+			/*
+			 * Any modification on the local execution should enable 2PC. If remote
+			 * queries are also ReadOnly, our 2PC logic is smart enough to skip sending
+			 * PREPARE to those connections.
+			 */
+			CoordinatedTransactionShouldUse2PC();
+		}
+
 		LogLocalCommand(task);
 
 		if (isUtilityCommand)
@@ -406,7 +430,7 @@ ExecuteUtilityCommand(const char *taskQueryCommand)
 			 * process utility.
 			 */
 			ProcessUtilityParseTree(taskRawParseTree, taskQueryCommand,
-									PROCESS_UTILITY_TOPLEVEL, NULL, None_Receiver,
+									PROCESS_UTILITY_QUERY, NULL, None_Receiver,
 									NULL);
 		}
 	}

src/backend/distributed/metadata/distobject.c

@@ -373,3 +373,56 @@ GetDistributedObjectAddressList(void)
 
 	return objectAddressList;
 }
+
+
+/*
+ * UpdateDistributedObjectColocationId gets an old and a new colocationId
+ * and updates the colocationId of all tuples in citus.pg_dist_object which
+ * have the old colocationId to the new colocationId.
+ */
+void
+UpdateDistributedObjectColocationId(uint32 oldColocationId,
+									uint32 newColocationId)
+{
+	const bool indexOK = false;
+	ScanKeyData scanKey[1];
+	Relation pgDistObjectRel = table_open(DistObjectRelationId(),
+										  RowExclusiveLock);
+	TupleDesc tupleDescriptor = RelationGetDescr(pgDistObjectRel);
+
+	/* scan pg_dist_object for colocationId equal to old colocationId */
+	ScanKeyInit(&scanKey[0], Anum_pg_dist_object_colocationid,
+				BTEqualStrategyNumber,
+				F_INT4EQ, UInt32GetDatum(oldColocationId));
+
+	SysScanDesc scanDescriptor = systable_beginscan(pgDistObjectRel,
+													InvalidOid,
+													indexOK,
+													NULL, 1, scanKey);
+	HeapTuple heapTuple;
+	while (HeapTupleIsValid(heapTuple = systable_getnext(scanDescriptor)))
+	{
+		Datum values[Natts_pg_dist_object];
+		bool isnull[Natts_pg_dist_object];
+		bool replace[Natts_pg_dist_object];
+
+		memset(replace, 0, sizeof(replace));
+
+		replace[Anum_pg_dist_object_colocationid - 1] = true;
+
+		/* update the colocationId to the new one */
+		values[Anum_pg_dist_object_colocationid - 1] = UInt32GetDatum(newColocationId);
+
+		isnull[Anum_pg_dist_object_colocationid - 1] = false;
+
+		heapTuple = heap_modify_tuple(heapTuple, tupleDescriptor, values, isnull,
+									  replace);
+
+		CatalogTupleUpdate(pgDistObjectRel, &heapTuple->t_self, heapTuple);
+		CitusInvalidateRelcacheByRelid(DistObjectRelationId());
+	}
+
+	systable_endscan(scanDescriptor);
+	table_close(pgDistObjectRel, NoLock);
+	CommandCounterIncrement();
+}

src/backend/distributed/metadata/metadata_utility.c

@@ -79,14 +79,24 @@ static bool DistributedTableSizeOnWorker(WorkerNode *workerNode, Oid relationId,
 										 char *sizeQuery, bool failOnError,
 										 uint64 *tableSize);
 static List * ShardIntervalsOnWorkerGroup(WorkerNode *workerNode, Oid relationId);
-static char * GenerateShardNameAndSizeQueryForShardList(List *shardIntervalList);
-static char * GenerateAllShardNameAndSizeQueryForNode(WorkerNode *workerNode);
-static List * GenerateShardSizesQueryList(List *workerNodeList);
+static char * GenerateShardStatisticsQueryForShardList(List *shardIntervalList, bool
+													   useShardMinMaxQuery);
+static char * GenerateAllShardStatisticsQueryForNode(WorkerNode *workerNode,
+													 List *citusTableIds, bool
+													 useShardMinMaxQuery);
+static List * GenerateShardStatisticsQueryList(List *workerNodeList, List *citusTableIds,
+											   bool useShardMinMaxQuery);
 static void ErrorIfNotSuitableToGetSize(Oid relationId);
 static List * OpenConnectionToNodes(List *workerNodeList);
 static void ReceiveShardNameAndSizeResults(List *connectionList,
 										   Tuplestorestate *tupleStore,
 										   TupleDesc tupleDescriptor);
+static void AppendShardSizeMinMaxQuery(StringInfo selectQuery, uint64 shardId,
+									   ShardInterval *
+									   shardInterval, char *shardName,
+									   char *quotedShardName);
+static void AppendShardSizeQuery(StringInfo selectQuery, ShardInterval *shardInterval,
+								 char *quotedShardName);
 
 /* exports for SQL callable functions */
 PG_FUNCTION_INFO_V1(citus_table_size);
@@ -102,25 +112,16 @@ citus_shard_sizes(PG_FUNCTION_ARGS)
 {
 	CheckCitusVersion(ERROR);
 
-	List *workerNodeList = ActivePrimaryNodeList(NoLock);
+	List *allCitusTableIds = AllCitusTableIds();
 
-	List *shardSizesQueryList = GenerateShardSizesQueryList(workerNodeList);
+	/* we don't need a distributed transaction here */
+	bool useDistributedTransaction = false;
 
-	List *connectionList = OpenConnectionToNodes(workerNodeList);
-	FinishConnectionListEstablishment(connectionList);
-
-
-	/* send commands in parallel */
-	for (int i = 0; i < list_length(connectionList); i++)
-	{
-		MultiConnection *connection = (MultiConnection *) list_nth(connectionList, i);
-		char *shardSizesQuery = (char *) list_nth(shardSizesQueryList, i);
-		int querySent = SendRemoteCommand(connection, shardSizesQuery);
-		if (querySent == 0)
-		{
-			ReportConnectionError(connection, WARNING);
-		}
-	}
+	/* we only want the shard sizes here so useShardMinMaxQuery parameter is false */
+	bool useShardMinMaxQuery = false;
+	List *connectionList = SendShardStatisticsQueriesInParallel(allCitusTableIds,
+																useDistributedTransaction,
+																useShardMinMaxQuery);
 
 	TupleDesc tupleDescriptor = NULL;
 	Tuplestorestate *tupleStore = SetupTuplestore(fcinfo, &tupleDescriptor);
@@ -225,6 +226,59 @@ citus_relation_size(PG_FUNCTION_ARGS)
 }
 
 
+/*
+ * SendShardStatisticsQueriesInParallel generates query lists for obtaining shard
+ * statistics and then sends the commands in parallel by opening connections
+ * to available nodes. It returns the connection list.
+ */
+List *
+SendShardStatisticsQueriesInParallel(List *citusTableIds, bool useDistributedTransaction,
+									 bool
+									 useShardMinMaxQuery)
+{
+	List *workerNodeList = ActivePrimaryNodeList(NoLock);
+
+	List *shardSizesQueryList = GenerateShardStatisticsQueryList(workerNodeList,
+																 citusTableIds,
+																 useShardMinMaxQuery);
+
+	List *connectionList = OpenConnectionToNodes(workerNodeList);
+	FinishConnectionListEstablishment(connectionList);
+
+	if (useDistributedTransaction)
+	{
+		/*
+		 * For now, in the case we want to include shard min and max values, we also
+		 * want to update the entries in pg_dist_placement and pg_dist_shard with the
+		 * latest statistics. In order to detect distributed deadlocks, we assign a
+		 * distributed transaction ID to the current transaction
+		 */
+		UseCoordinatedTransaction();
+	}
+
+	/* send commands in parallel */
+	for (int i = 0; i < list_length(connectionList); i++)
+	{
+		MultiConnection *connection = (MultiConnection *) list_nth(connectionList, i);
+		char *shardSizesQuery = (char *) list_nth(shardSizesQueryList, i);
+
+		if (useDistributedTransaction)
+		{
+			/* run the size query in a distributed transaction */
+			RemoteTransactionBeginIfNecessary(connection);
+		}
+
+		int querySent = SendRemoteCommand(connection, shardSizesQuery);
+
+		if (querySent == 0)
+		{
+			ReportConnectionError(connection, WARNING);
+		}
+	}
+	return connectionList;
+}
+
+
 /*
  * OpenConnectionToNodes opens a single connection per node
  * for the given workerNodeList.
@@ -250,20 +304,25 @@ OpenConnectionToNodes(List *workerNodeList)
 
 
 /*
- * GenerateShardSizesQueryList generates a query per node that
- * will return all shard_name, shard_size pairs from the node.
+ * GenerateShardStatisticsQueryList generates a query per node that will return:
+ * - all shard_name, shard_size pairs from the node (if includeShardMinMax is false)
+ * - all shard_id, shard_minvalue, shard_maxvalue, shard_size quartuples from the node (if true)
  */
 static List *
-GenerateShardSizesQueryList(List *workerNodeList)
+GenerateShardStatisticsQueryList(List *workerNodeList, List *citusTableIds, bool
+								 useShardMinMaxQuery)
 {
-	List *shardSizesQueryList = NIL;
+	List *shardStatisticsQueryList = NIL;
 	WorkerNode *workerNode = NULL;
 	foreach_ptr(workerNode, workerNodeList)
 	{
-		char *shardSizesQuery = GenerateAllShardNameAndSizeQueryForNode(workerNode);
-		shardSizesQueryList = lappend(shardSizesQueryList, shardSizesQuery);
+		char *shardStatisticsQuery = GenerateAllShardStatisticsQueryForNode(workerNode,
+																			citusTableIds,
+																			useShardMinMaxQuery);
+		shardStatisticsQueryList = lappend(shardStatisticsQueryList,
+										   shardStatisticsQuery);
 	}
-	return shardSizesQueryList;
+	return shardStatisticsQueryList;
 }
 
 
@@ -572,37 +631,50 @@ GenerateSizeQueryOnMultiplePlacements(List *shardIntervalList, char *sizeQuery)
 
 
 /*
- * GenerateAllShardNameAndSizeQueryForNode generates a query that returns all
- * shard_name, shard_size pairs for the given node.
+ * GenerateAllShardStatisticsQueryForNode generates a query that returns:
+ * - all shard_name, shard_size pairs for the given node (if useShardMinMaxQuery is false)
+ * - all shard_id, shard_minvalue, shard_maxvalue, shard_size quartuples (if true)
  */
 static char *
-GenerateAllShardNameAndSizeQueryForNode(WorkerNode *workerNode)
+GenerateAllShardStatisticsQueryForNode(WorkerNode *workerNode, List *citusTableIds, bool
+									   useShardMinMaxQuery)
 {
-	List *allCitusTableIds = AllCitusTableIds();
-
-	StringInfo allShardNameAndSizeQuery = makeStringInfo();
+	StringInfo allShardStatisticsQuery = makeStringInfo();
 
 	Oid relationId = InvalidOid;
-	foreach_oid(relationId, allCitusTableIds)
+	foreach_oid(relationId, citusTableIds)
 	{
 		List *shardIntervalsOnNode = ShardIntervalsOnWorkerGroup(workerNode, relationId);
-		char *shardNameAndSizeQuery =
-			GenerateShardNameAndSizeQueryForShardList(shardIntervalsOnNode);
-		appendStringInfoString(allShardNameAndSizeQuery, shardNameAndSizeQuery);
+		char *shardStatisticsQuery =
+			GenerateShardStatisticsQueryForShardList(shardIntervalsOnNode,
+													 useShardMinMaxQuery);
+		appendStringInfoString(allShardStatisticsQuery, shardStatisticsQuery);
 	}
 
 	/* Add a dummy entry so that UNION ALL doesn't complain */
-	appendStringInfo(allShardNameAndSizeQuery, "SELECT NULL::text, 0::bigint;");
-	return allShardNameAndSizeQuery->data;
+	if (useShardMinMaxQuery)
+	{
+		/* 0 for shard_id, NULL for min, NULL for text, 0 for shard_size */
+		appendStringInfo(allShardStatisticsQuery,
+						 "SELECT 0::bigint, NULL::text, NULL::text, 0::bigint;");
+	}
+	else
+	{
+		/* NULL for shard_name, 0 for shard_size */
+		appendStringInfo(allShardStatisticsQuery, "SELECT NULL::text, 0::bigint;");
+	}
+	return allShardStatisticsQuery->data;
 }
 
 
 /*
- * GenerateShardNameAndSizeQueryForShardList generates a SELECT shard_name - shard_size query to get
- * size of multiple tables.
+ * GenerateShardStatisticsQueryForShardList generates one of the two types of queries:
+ * - SELECT shard_name - shard_size (if useShardMinMaxQuery is false)
+ * - SELECT shard_id, shard_minvalue, shard_maxvalue, shard_size (if true)
  */
 static char *
-GenerateShardNameAndSizeQueryForShardList(List *shardIntervalList)
+GenerateShardStatisticsQueryForShardList(List *shardIntervalList, bool
+										 useShardMinMaxQuery)
 {
 	StringInfo selectQuery = makeStringInfo();
 
@@ -618,8 +690,15 @@ GenerateShardNameAndSizeQueryForShardList(List *shardIntervalList)
 		char *shardQualifiedName = quote_qualified_identifier(schemaName, shardName);
 		char *quotedShardName = quote_literal_cstr(shardQualifiedName);
 
-		appendStringInfo(selectQuery, "SELECT %s AS shard_name, ", quotedShardName);
-		appendStringInfo(selectQuery, PG_RELATION_SIZE_FUNCTION, quotedShardName);
+		if (useShardMinMaxQuery)
+		{
+			AppendShardSizeMinMaxQuery(selectQuery, shardId, shardInterval, shardName,
+									   quotedShardName);
+		}
+		else
+		{
+			AppendShardSizeQuery(selectQuery, shardInterval, quotedShardName);
+		}
 		appendStringInfo(selectQuery, " UNION ALL ");
 	}
 
@@ -627,6 +706,54 @@ GenerateShardNameAndSizeQueryForShardList(List *shardIntervalList)
 }
 
 
+/*
+ * AppendShardSizeMinMaxQuery appends a query in the following form to selectQuery
+ * SELECT shard_id, shard_minvalue, shard_maxvalue, shard_size
+ */
+static void
+AppendShardSizeMinMaxQuery(StringInfo selectQuery, uint64 shardId,
+						   ShardInterval *shardInterval, char *shardName,
+						   char *quotedShardName)
+{
+	if (IsCitusTableType(shardInterval->relationId, APPEND_DISTRIBUTED))
+	{
+		/* fill in the partition column name */
+		const uint32 unusedTableId = 1;
+		Var *partitionColumn = PartitionColumn(shardInterval->relationId,
+											   unusedTableId);
+		char *partitionColumnName = get_attname(shardInterval->relationId,
+												partitionColumn->varattno, false);
+		appendStringInfo(selectQuery,
+						 "SELECT " UINT64_FORMAT
+						 " AS shard_id, min(%s)::text AS shard_minvalue, max(%s)::text AS shard_maxvalue, pg_relation_size(%s) AS shard_size FROM %s ",
+						 shardId, partitionColumnName,
+						 partitionColumnName,
+						 quotedShardName, shardName);
+	}
+	else
+	{
+		/* we don't need to update min/max for non-append distributed tables because they don't change */
+		appendStringInfo(selectQuery,
+						 "SELECT " UINT64_FORMAT
+						 " AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size(%s) AS shard_size ",
+						 shardId, quotedShardName);
+	}
+}
+
+
+/*
+ * AppendShardSizeQuery appends a query in the following form to selectQuery
+ * SELECT shard_name, shard_size
+ */
+static void
+AppendShardSizeQuery(StringInfo selectQuery, ShardInterval *shardInterval,
+					 char *quotedShardName)
+{
+	appendStringInfo(selectQuery, "SELECT %s AS shard_name, ", quotedShardName);
+	appendStringInfo(selectQuery, PG_RELATION_SIZE_FUNCTION, quotedShardName);
+}
+
+
 /*
  * ErrorIfNotSuitableToGetSize determines whether the table is suitable to find
  * its' size with internal functions.

src/backend/distributed/metadata/node_metadata.c

@@ -1384,16 +1384,34 @@ AddNodeMetadata(char *nodeName, int32 nodePort,
 	*nodeAlreadyExists = false;
 
 	/*
-	 * Take an exclusive lock on pg_dist_node to serialize node changes.
+	 * Prevent / wait for concurrent modification before checking whether
+	 * the worker already exists in pg_dist_node.
+	 */
+	LockRelationOid(DistNodeRelationId(), RowShareLock);
+
+	WorkerNode *workerNode = FindWorkerNodeAnyCluster(nodeName, nodePort);
+	if (workerNode != NULL)
+	{
+		/* return early without holding locks when the node already exists */
+		*nodeAlreadyExists = true;
+
+		return workerNode->nodeId;
+	}
+
+	/*
+	 * We are going to change pg_dist_node, prevent any concurrent reads that
+	 * are not tolerant to concurrent node addition by taking an exclusive
+	 * lock (conflicts with all but AccessShareLock).
+	 *
 	 * We may want to relax or have more fine-grained locking in the future
 	 * to allow users to add multiple nodes concurrently.
 	 */
 	LockRelationOid(DistNodeRelationId(), ExclusiveLock);
 
-	WorkerNode *workerNode = FindWorkerNodeAnyCluster(nodeName, nodePort);
+	/* recheck in case 2 node additions pass the first check concurrently */
+	workerNode = FindWorkerNodeAnyCluster(nodeName, nodePort);
 	if (workerNode != NULL)
 	{
-		/* fill return data and return */
 		*nodeAlreadyExists = true;
 
 		return workerNode->nodeId;

src/backend/distributed/operations/delete_protocol.c

@@ -332,7 +332,7 @@ DropShards(Oid relationId, char *schemaName, char *relationName,
 	 */
 	if (MultiShardCommitProtocol == COMMIT_PROTOCOL_2PC)
 	{
-		CoordinatedTransactionUse2PC();
+		CoordinatedTransactionShouldUse2PC();
 	}
 
 	List *dropTaskList = DropTaskList(relationId, schemaName, relationName,

src/backend/distributed/operations/node_protocol.c

@@ -85,6 +85,7 @@ PG_FUNCTION_INFO_V1(master_get_table_ddl_events);
 PG_FUNCTION_INFO_V1(master_get_new_shardid);
 PG_FUNCTION_INFO_V1(master_get_new_placementid);
 PG_FUNCTION_INFO_V1(master_get_active_worker_nodes);
+PG_FUNCTION_INFO_V1(citus_get_active_worker_nodes);
 PG_FUNCTION_INFO_V1(master_get_round_robin_candidate_nodes);
 PG_FUNCTION_INFO_V1(master_stage_shard_row);
 PG_FUNCTION_INFO_V1(master_stage_shard_placement_row);
@@ -442,12 +443,12 @@ master_stage_shard_placement_row(PG_FUNCTION_ARGS)
 
 
 /*
- * master_get_active_worker_nodes returns a set of active worker host names and
+ * citus_get_active_worker_nodes returns a set of active worker host names and
  * port numbers in deterministic order. Currently we assume that all worker
  * nodes in pg_dist_node are active.
  */
 Datum
-master_get_active_worker_nodes(PG_FUNCTION_ARGS)
+citus_get_active_worker_nodes(PG_FUNCTION_ARGS)
 {
 	FuncCallContext *functionContext = NULL;
 	uint32 workerNodeIndex = 0;
@@ -512,6 +513,16 @@ master_get_active_worker_nodes(PG_FUNCTION_ARGS)
 }
 
 
+/*
+ * master_get_active_worker_nodes is a wrapper function for old UDF name.
+ */
+Datum
+master_get_active_worker_nodes(PG_FUNCTION_ARGS)
+{
+	return citus_get_active_worker_nodes(fcinfo);
+}
+
+
 /* Finds the relationId from a potentially qualified relation name. */
 Oid
 ResolveRelationId(text *relationName, bool missingOk)

src/backend/distributed/operations/stage_protocol.c

@@ -32,7 +32,9 @@
 #include "distributed/connection_management.h"
 #include "distributed/deparse_shard_query.h"
 #include "distributed/distributed_planner.h"
+#include "distributed/foreign_key_relationship.h"
 #include "distributed/listutils.h"
+#include "distributed/lock_graph.h"
 #include "distributed/multi_client_executor.h"
 #include "distributed/multi_executor.h"
 #include "distributed/metadata_utility.h"
@@ -65,12 +67,22 @@ static List * RelationShardListForShardCreate(ShardInterval *shardInterval);
 static bool WorkerShardStats(ShardPlacement *placement, Oid relationId,
 							 const char *shardName, uint64 *shardSize,
 							 text **shardMinValue, text **shardMaxValue);
+static void UpdateTableStatistics(Oid relationId);
+static void ReceiveAndUpdateShardsSizeAndMinMax(List *connectionList);
+static void UpdateShardSizeAndMinMax(uint64 shardId, ShardInterval *shardInterval, Oid
+									 relationId, List *shardPlacementList, uint64
+									 shardSize, text *shardMinValue,
+									 text *shardMaxValue);
+static bool ProcessShardStatisticsRow(PGresult *result, int64 rowIndex, uint64 *shardId,
+									  text **shardMinValue, text **shardMaxValue,
+									  uint64 *shardSize);
 
 /* exports for SQL callable functions */
 PG_FUNCTION_INFO_V1(master_create_empty_shard);
 PG_FUNCTION_INFO_V1(master_append_table_to_shard);
 PG_FUNCTION_INFO_V1(citus_update_shard_statistics);
 PG_FUNCTION_INFO_V1(master_update_shard_statistics);
+PG_FUNCTION_INFO_V1(citus_update_table_statistics);
 
 
 /*
@@ -361,6 +373,23 @@ citus_update_shard_statistics(PG_FUNCTION_ARGS)
 }
 
 
+/*
+ * citus_update_table_statistics updates metadata (shard size and shard min/max
+ * values) of the shards of the given table
+ */
+Datum
+citus_update_table_statistics(PG_FUNCTION_ARGS)
+{
+	Oid distributedTableId = PG_GETARG_OID(0);
+
+	CheckCitusVersion(ERROR);
+
+	UpdateTableStatistics(distributedTableId);
+
+	PG_RETURN_VOID();
+}
+
+
 /*
  * master_update_shard_statistics is a wrapper function for old UDF name.
  */
@@ -782,7 +811,6 @@ UpdateShardStatistics(int64 shardId)
 {
 	ShardInterval *shardInterval = LoadShardInterval(shardId);
 	Oid relationId = shardInterval->relationId;
-	char storageType = shardInterval->storageType;
 	bool statsOK = false;
 	uint64 shardSize = 0;
 	text *minValue = NULL;
@@ -825,17 +853,166 @@ UpdateShardStatistics(int64 shardId)
 						  errdetail("Setting shard statistics to NULL")));
 	}
 
-	/* make sure we don't process cancel signals */
-	HOLD_INTERRUPTS();
+	UpdateShardSizeAndMinMax(shardId, shardInterval, relationId, shardPlacementList,
+							 shardSize, minValue, maxValue);
+	return shardSize;
+}
 
-	/* update metadata for each shard placement we appended to */
+
+/*
+ * UpdateTableStatistics updates metadata (shard size and shard min/max values)
+ * of the shards of the given table. Follows a similar logic to citus_shard_sizes function.
+ */
+static void
+UpdateTableStatistics(Oid relationId)
+{
+	List *citusTableIds = NIL;
+	citusTableIds = lappend_oid(citusTableIds, relationId);
+
+	/* we want to use a distributed transaction here to detect distributed deadlocks */
+	bool useDistributedTransaction = true;
+
+	/* we also want shard min/max values for append distributed tables */
+	bool useShardMinMaxQuery = true;
+
+	List *connectionList = SendShardStatisticsQueriesInParallel(citusTableIds,
+																useDistributedTransaction,
+																useShardMinMaxQuery);
+
+	ReceiveAndUpdateShardsSizeAndMinMax(connectionList);
+}
+
+
+/*
+ * ReceiveAndUpdateShardsSizeAndMinMax receives shard id, size
+ * and min max results from the given connection list, and updates
+ * respective entries in pg_dist_placement and pg_dist_shard
+ */
+static void
+ReceiveAndUpdateShardsSizeAndMinMax(List *connectionList)
+{
+	/*
+	 * From the connection list, we will not get all the shards, but
+	 * all the placements. We use a hash table to remember already visited shard ids
+	 * since we update all the different placements of a shard id at once.
+	 */
+	HTAB *alreadyVisitedShardPlacements = CreateOidVisitedHashSet();
+
+	MultiConnection *connection = NULL;
+	foreach_ptr(connection, connectionList)
+	{
+		if (PQstatus(connection->pgConn) != CONNECTION_OK)
+		{
+			continue;
+		}
+
+		bool raiseInterrupts = true;
+		PGresult *result = GetRemoteCommandResult(connection, raiseInterrupts);
+		if (!IsResponseOK(result))
+		{
+			ReportResultError(connection, result, WARNING);
+			continue;
+		}
+
+		int64 rowCount = PQntuples(result);
+		int64 colCount = PQnfields(result);
+
+		/* Although it is not expected */
+		if (colCount != UPDATE_SHARD_STATISTICS_COLUMN_COUNT)
+		{
+			ereport(WARNING, (errmsg("unexpected number of columns from "
+									 "citus_update_table_statistics")));
+			continue;
+		}
+
+		for (int64 rowIndex = 0; rowIndex < rowCount; rowIndex++)
+		{
+			uint64 shardId = 0;
+			text *shardMinValue = NULL;
+			text *shardMaxValue = NULL;
+			uint64 shardSize = 0;
+
+			if (!ProcessShardStatisticsRow(result, rowIndex, &shardId, &shardMinValue,
+										   &shardMaxValue, &shardSize))
+			{
+				/* this row has no valid shard statistics */
+				continue;
+			}
+
+			if (OidVisited(alreadyVisitedShardPlacements, shardId))
+			{
+				/* We have already updated this placement list */
+				continue;
+			}
+
+			VisitOid(alreadyVisitedShardPlacements, shardId);
+
+			ShardInterval *shardInterval = LoadShardInterval(shardId);
+			Oid relationId = shardInterval->relationId;
+			List *shardPlacementList = ActiveShardPlacementList(shardId);
+
+			UpdateShardSizeAndMinMax(shardId, shardInterval, relationId,
+									 shardPlacementList, shardSize, shardMinValue,
+									 shardMaxValue);
+		}
+		PQclear(result);
+		ForgetResults(connection);
+	}
+	hash_destroy(alreadyVisitedShardPlacements);
+}
+
+
+/*
+ * ProcessShardStatisticsRow processes a row of shard statistics of the input PGresult
+ * - it returns true if this row belongs to a valid shard
+ * - it returns false if this row has no valid shard statistics (shardId = INVALID_SHARD_ID)
+ */
+static bool
+ProcessShardStatisticsRow(PGresult *result, int64 rowIndex, uint64 *shardId,
+						  text **shardMinValue, text **shardMaxValue, uint64 *shardSize)
+{
+	*shardId = ParseIntField(result, rowIndex, 0);
+
+	/* check for the dummy entries we put so that UNION ALL wouldn't complain */
+	if (*shardId == INVALID_SHARD_ID)
+	{
+		/* this row has no valid shard statistics */
+		return false;
+	}
+
+	char *minValueResult = PQgetvalue(result, rowIndex, 1);
+	char *maxValueResult = PQgetvalue(result, rowIndex, 2);
+	*shardMinValue = cstring_to_text(minValueResult);
+	*shardMaxValue = cstring_to_text(maxValueResult);
+	*shardSize = ParseIntField(result, rowIndex, 3);
+	return true;
+}
+
+
+/*
+ * UpdateShardSizeAndMinMax updates the shardlength (shard size) of the given
+ * shard and its placements in pg_dist_placement, and updates the shard min value
+ * and shard max value of the given shard in pg_dist_shard if the relationId belongs
+ * to an append-distributed table
+ */
+static void
+UpdateShardSizeAndMinMax(uint64 shardId, ShardInterval *shardInterval, Oid relationId,
+						 List *shardPlacementList, uint64 shardSize, text *shardMinValue,
+						 text *shardMaxValue)
+{
+	char storageType = shardInterval->storageType;
+
+	ShardPlacement *placement = NULL;
+
+	/* update metadata for each shard placement */
 	foreach_ptr(placement, shardPlacementList)
 	{
 		uint64 placementId = placement->placementId;
 		int32 groupId = placement->groupId;
 
 		DeleteShardPlacementRow(placementId);
-		InsertShardPlacementRow(shardId, placementId, SHARD_STATE_ACTIVE, shardSize,
+		InsertShardPlacementRow(shardId, placementId, SHARD_STATE_ACTIVE,
+								shardSize,
 								groupId);
 	}
 
@@ -843,18 +1020,9 @@ UpdateShardStatistics(int64 shardId)
 	if (IsCitusTableType(relationId, APPEND_DISTRIBUTED))
 	{
 		DeleteShardRow(shardId);
-		InsertShardRow(relationId, shardId, storageType, minValue, maxValue);
+		InsertShardRow(relationId, shardId, storageType, shardMinValue,
+					   shardMaxValue);
 	}
-
-	if (QueryCancelPending)
-	{
-		ereport(WARNING, (errmsg("cancel requests are ignored during metadata update")));
-		QueryCancelPending = false;
-	}
-
-	RESUME_INTERRUPTS();
-
-	return shardSize;
 }
 
 

src/backend/distributed/planner/deparse_shard_query.c

@@ -38,8 +38,9 @@
 #include "utils/rel.h"
 #include "utils/syscache.h"
 
-static void UpdateTaskQueryString(Query *query, Oid distributedTableId,
-								  RangeTblEntry *valuesRTE, Task *task);
+
+static void AddInsertAliasIfNeeded(Query *query);
+static void UpdateTaskQueryString(Query *query, Task *task);
 static bool ReplaceRelationConstraintByShardConstraint(List *relationShardList,
 													   OnConflictExpr *onConflict);
 static RelationShard * FindRelationShard(Oid inputRelationId, List *relationShardList);
@@ -57,27 +58,43 @@ RebuildQueryStrings(Job *workerJob)
 {
 	Query *originalQuery = workerJob->jobQuery;
 	List *taskList = workerJob->taskList;
-	Oid relationId = ((RangeTblEntry *) linitial(originalQuery->rtable))->relid;
-	RangeTblEntry *valuesRTE = ExtractDistributedInsertValuesRTE(originalQuery);
-
 	Task *task = NULL;
 
+	if (originalQuery->commandType == CMD_INSERT)
+	{
+		AddInsertAliasIfNeeded(originalQuery);
+	}
+
 	foreach_ptr(task, taskList)
 	{
 		Query *query = originalQuery;
 
-		if (UpdateOrDeleteQuery(query) && list_length(taskList) > 1)
+		/*
+		 * Copy the query if there are multiple tasks. If there is a single
+		 * task, we scribble on the original query to avoid the copying
+		 * overhead.
+		 */
+		if (list_length(taskList) > 1)
 		{
 			query = copyObject(originalQuery);
 		}
+
+		if (UpdateOrDeleteQuery(query))
+		{
+			/*
+			 * For UPDATE and DELETE queries, we may have subqueries and joins, so
+			 * we use relation shard list to update shard names and call
+			 * pg_get_query_def() directly.
+			 */
+			List *relationShardList = task->relationShardList;
+			UpdateRelationToShardNames((Node *) query, relationShardList);
+		}
 		else if (query->commandType == CMD_INSERT && task->modifyWithSubquery)
 		{
 			/* for INSERT..SELECT, adjust shard names in SELECT part */
 			List *relationShardList = task->relationShardList;
 			ShardInterval *shardInterval = LoadShardInterval(task->anchorShardId);
 
-			query = copyObject(originalQuery);
-
 			RangeTblEntry *copiedInsertRte = ExtractResultRelationRTEOrError(query);
 			RangeTblEntry *copiedSubqueryRte = ExtractSelectRangeTableEntry(query);
 			Query *copiedSubquery = copiedSubqueryRte->subquery;
@@ -90,29 +107,18 @@ RebuildQueryStrings(Job *workerJob)
 
 			ReorderInsertSelectTargetLists(query, copiedInsertRte, copiedSubqueryRte);
 
-			/* setting an alias simplifies deparsing of RETURNING */
-			if (copiedInsertRte->alias == NULL)
-			{
-				Alias *alias = makeAlias(CITUS_TABLE_ALIAS, NIL);
-				copiedInsertRte->alias = alias;
-			}
-
 			UpdateRelationToShardNames((Node *) copiedSubquery, relationShardList);
 		}
-		else if (query->commandType == CMD_INSERT && (query->onConflict != NULL ||
-													  valuesRTE != NULL))
+
+		if (query->commandType == CMD_INSERT)
 		{
+			RangeTblEntry *modifiedRelationRTE = linitial(originalQuery->rtable);
+
 			/*
-			 * Always an alias in UPSERTs and multi-row INSERTs to avoid
-			 * deparsing issues (e.g. RETURNING might reference the original
-			 * table name, which has been replaced by a shard name).
+			 * We store the modified relaiton ID in the task so we can lazily call
+			 * deparse_shard_query when the string is needed
 			 */
-			RangeTblEntry *rangeTableEntry = linitial(query->rtable);
-			if (rangeTableEntry->alias == NULL)
-			{
-				Alias *alias = makeAlias(CITUS_TABLE_ALIAS, NIL);
-				rangeTableEntry->alias = alias;
-			}
+			task->anchorDistributedTableId = modifiedRelationRTE->relid;
 		}
 
 		bool isQueryObjectOrText = GetTaskQueryType(task) == TASK_QUERY_TEXT ||
@@ -122,7 +128,7 @@ RebuildQueryStrings(Job *workerJob)
 								? "(null)"
 								: ApplyLogRedaction(TaskQueryString(task)))));
 
-		UpdateTaskQueryString(query, relationId, valuesRTE, task);
+		UpdateTaskQueryString(query, task);
 
 		/*
 		 * If parameters were resolved in the job query, then they are now also
@@ -137,53 +143,68 @@ RebuildQueryStrings(Job *workerJob)
 
 
 /*
- * UpdateTaskQueryString updates the query string stored within the provided
- * Task. If the Task has row values from a multi-row INSERT, those are injected
- * into the provided query (using the provided valuesRTE, which must belong to
- * the query) before deparse occurs (the query's full VALUES list will be
- * restored before this function returns).
+ * AddInsertAliasIfNeeded adds an alias in UPSERTs and multi-row INSERTs to avoid
+ * deparsing issues (e.g. RETURNING might reference the original table name,
+ * which has been replaced by a shard name).
  */
 static void
-UpdateTaskQueryString(Query *query, Oid distributedTableId, RangeTblEntry *valuesRTE,
-					  Task *task)
+AddInsertAliasIfNeeded(Query *query)
+{
+	Assert(query->commandType == CMD_INSERT);
+
+	if (query->onConflict == NULL &&
+		ExtractDistributedInsertValuesRTE(query) == NULL)
+	{
+		/* simple single-row insert does not need an alias */
+		return;
+	}
+
+	RangeTblEntry *rangeTableEntry = linitial(query->rtable);
+	if (rangeTableEntry->alias != NULL)
+	{
+		/* INSERT already has an alias */
+		return;
+	}
+
+	Alias *alias = makeAlias(CITUS_TABLE_ALIAS, NIL);
+	rangeTableEntry->alias = alias;
+}
+
+
+/*
+ * UpdateTaskQueryString updates the query string stored within the provided
+ * Task. If the Task has row values from a multi-row INSERT, those are injected
+ * into the provided query before deparse occurs (the query's full VALUES list
+ * will be restored before this function returns).
+ */
+static void
+UpdateTaskQueryString(Query *query, Task *task)
 {
 	List *oldValuesLists = NIL;
-
-	if (valuesRTE != NULL)
-	{
-		Assert(valuesRTE->rtekind == RTE_VALUES);
-		Assert(task->rowValuesLists != NULL);
-
-		oldValuesLists = valuesRTE->values_lists;
-		valuesRTE->values_lists = task->rowValuesLists;
-	}
-
-	if (query->commandType != CMD_INSERT)
-	{
-		/*
-		 * For UPDATE and DELETE queries, we may have subqueries and joins, so
-		 * we use relation shard list to update shard names and call
-		 * pg_get_query_def() directly.
-		 */
-		List *relationShardList = task->relationShardList;
-		UpdateRelationToShardNames((Node *) query, relationShardList);
-	}
-	else if (ShouldLazyDeparseQuery(task))
-	{
-		/*
-		 * not all insert queries are copied before calling this
-		 * function, so we do it here
-		 */
-		query = copyObject(query);
-	}
+	RangeTblEntry *valuesRTE = NULL;
 
 	if (query->commandType == CMD_INSERT)
 	{
-		/*
-		 * We store this in the task so we can lazily call
-		 * deparse_shard_query when the string is needed
-		 */
-		task->anchorDistributedTableId = distributedTableId;
+		/* extract the VALUES from the INSERT */
+		valuesRTE = ExtractDistributedInsertValuesRTE(query);
+
+		if (valuesRTE != NULL)
+		{
+			Assert(valuesRTE->rtekind == RTE_VALUES);
+			Assert(task->rowValuesLists != NULL);
+
+			oldValuesLists = valuesRTE->values_lists;
+			valuesRTE->values_lists = task->rowValuesLists;
+		}
+
+		if (ShouldLazyDeparseQuery(task))
+		{
+			/*
+			 * not all insert queries are copied before calling this
+			 * function, so we do it here
+			 */
+			query = copyObject(query);
+		}
 	}
 
 	SetTaskQueryIfShouldLazyDeparse(task, query);

src/backend/distributed/planner/distributed_planner.c

@@ -49,6 +49,7 @@
 #include "executor/executor.h"
 #include "nodes/makefuncs.h"
 #include "nodes/nodeFuncs.h"
+#include "nodes/pg_list.h"
 #include "parser/parsetree.h"
 #include "parser/parse_type.h"
 #if PG_VERSION_NUM >= PG_VERSION_12
@@ -98,6 +99,7 @@ static PlannedStmt * FinalizeNonRouterPlan(PlannedStmt *localPlan,
 										   DistributedPlan *distributedPlan,
 										   CustomScan *customScan);
 static PlannedStmt * FinalizeRouterPlan(PlannedStmt *localPlan, CustomScan *customScan);
+static AppendRelInfo * FindTargetAppendRelInfo(PlannerInfo *root, int relationRteIndex);
 static List * makeTargetListFromCustomScanList(List *custom_scan_tlist);
 static List * makeCustomScanTargetlistFromExistingTargetList(List *existingTargetlist);
 static int32 BlessRecordExpressionList(List *exprs);
@@ -124,6 +126,7 @@ static PlannedStmt * PlanFastPathDistributedStmt(DistributedPlanningContext *pla
 static PlannedStmt * PlanDistributedStmt(DistributedPlanningContext *planContext,
 										 int rteIdCounter);
 static RTEListProperties * GetRTEListProperties(List *rangeTableList);
+static List * TranslatedVars(PlannerInfo *root, int relationIndex);
 
 
 /* Distributed planner hook */
@@ -1814,6 +1817,8 @@ multi_relation_restriction_hook(PlannerInfo *root, RelOptInfo *relOptInfo,
 
 	/* see comments on GetVarFromAssignedParam() */
 	relationRestriction->outerPlanParamsList = OuterPlanParamsList(root);
+	relationRestriction->translatedVars = TranslatedVars(root,
+														 relationRestriction->index);
 
 	RelationRestrictionContext *relationRestrictionContext =
 		plannerRestrictionContext->relationRestrictionContext;
@@ -1837,6 +1842,61 @@ multi_relation_restriction_hook(PlannerInfo *root, RelOptInfo *relOptInfo,
 }
 
 
+/*
+ * TranslatedVars deep copies the translated vars for the given relation index
+ * if there is any append rel list.
+ */
+static List *
+TranslatedVars(PlannerInfo *root, int relationIndex)
+{
+	List *translatedVars = NIL;
+
+	if (root->append_rel_list != NIL)
+	{
+		AppendRelInfo *targetAppendRelInfo =
+			FindTargetAppendRelInfo(root, relationIndex);
+		if (targetAppendRelInfo != NULL)
+		{
+			/* postgres deletes translated_vars after pg13, hence we deep copy them here */
+			Node *targetNode = NULL;
+			foreach_ptr(targetNode, targetAppendRelInfo->translated_vars)
+			{
+				translatedVars =
+					lappend(translatedVars, copyObject(targetNode));
+			}
+		}
+	}
+	return translatedVars;
+}
+
+
+/*
+ * FindTargetAppendRelInfo finds the target append rel info for the given
+ * relation rte index.
+ */
+static AppendRelInfo *
+FindTargetAppendRelInfo(PlannerInfo *root, int relationRteIndex)
+{
+	AppendRelInfo *appendRelInfo = NULL;
+
+	/* iterate on the queries that are part of UNION ALL subselects */
+	foreach_ptr(appendRelInfo, root->append_rel_list)
+	{
+		/*
+		 * We're only interested in the child rel that is equal to the
+		 * relation we're investigating. Here we don't need to find the offset
+		 * because postgres adds an offset to child_relid and parent_relid after
+		 * calling multi_relation_restriction_hook.
+		 */
+		if (appendRelInfo->child_relid == relationRteIndex)
+		{
+			return appendRelInfo;
+		}
+	}
+	return NULL;
+}
+
+
 /*
  * AdjustReadIntermediateResultCost adjusts the row count and total cost
  * of a read_intermediate_result call based on the file size.
@@ -2143,6 +2203,33 @@ CreateAndPushPlannerRestrictionContext(void)
 }
 
 
+/*
+ * TranslatedVarsForRteIdentity gets an rteIdentity and returns the
+ * translatedVars that belong to the range table relation. If no
+ * translatedVars found, the function returns NIL;
+ */
+List *
+TranslatedVarsForRteIdentity(int rteIdentity)
+{
+	PlannerRestrictionContext *currentPlannerRestrictionContext =
+		CurrentPlannerRestrictionContext();
+
+	List *relationRestrictionList =
+		currentPlannerRestrictionContext->relationRestrictionContext->
+		relationRestrictionList;
+	RelationRestriction *relationRestriction = NULL;
+	foreach_ptr(relationRestriction, relationRestrictionList)
+	{
+		if (GetRTEIdentity(relationRestriction->rte) == rteIdentity)
+		{
+			return relationRestriction->translatedVars;
+		}
+	}
+
+	return NIL;
+}
+
+
 /*
  * CurrentRestrictionContext returns the most recently added
  * PlannerRestrictionContext from the plannerRestrictionContextList list.

src/backend/distributed/planner/multi_physical_planner.c

@@ -824,7 +824,21 @@ static List *
 QueryTargetList(MultiNode *multiNode)
 {
 	List *projectNodeList = FindNodesOfType(multiNode, T_MultiProject);
-	Assert(list_length(projectNodeList) > 0);
+	if (list_length(projectNodeList) == 0)
+	{
+		/*
+		 * The physical planner assumes that all worker queries would have
+		 * target list entries based on the fact that at least the column
+		 * on the JOINs have to be on the target list. However, there is
+		 * an exception to that if there is a cartesian product join and
+		 * there is no additional target list entries belong to one side
+		 * of the JOIN. Once we support cartesian product join, we should
+		 * remove this error.
+		 */
+		ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+						errmsg("cannot perform distributed planning on this query"),
+						errdetail("Cartesian products are currently unsupported")));
+	}
 
 	MultiProject *topProjectNode = (MultiProject *) linitial(projectNodeList);
 	List *columnList = topProjectNode->columnList;

src/backend/distributed/planner/multi_router_planner.c

@@ -555,6 +555,14 @@ ModifyPartialQuerySupported(Query *queryTree, bool multiShardQuery,
 	{
 		ListCell *cteCell = NULL;
 
+		/* CTEs still not supported for INSERTs. */
+		if (queryTree->commandType == CMD_INSERT)
+		{
+			return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
+								 "Router planner doesn't support common table expressions with INSERT queries.",
+								 NULL, NULL);
+		}
+
 		foreach(cteCell, queryTree->cteList)
 		{
 			CommonTableExpr *cte = (CommonTableExpr *) lfirst(cteCell);
@@ -562,31 +570,22 @@ ModifyPartialQuerySupported(Query *queryTree, bool multiShardQuery,
 
 			if (cteQuery->commandType != CMD_SELECT)
 			{
-				/* Modifying CTEs still not supported for INSERTs & multi shard queries. */
-				if (queryTree->commandType == CMD_INSERT)
-				{
-					return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
-										 "Router planner doesn't support non-select common table expressions with non-select queries.",
-										 NULL, NULL);
-				}
-
+				/* Modifying CTEs still not supported for multi shard queries. */
 				if (multiShardQuery)
 				{
 					return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
 										 "Router planner doesn't support non-select common table expressions with multi shard queries.",
 										 NULL, NULL);
 				}
+				/* Modifying CTEs exclude both INSERT CTEs & INSERT queries. */
+				else if (cteQuery->commandType == CMD_INSERT)
+				{
+					return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
+										 "Router planner doesn't support INSERT common table expressions.",
+										 NULL, NULL);
+				}
 			}
 
-			/* Modifying CTEs exclude both INSERT CTEs & INSERT queries. */
-			if (cteQuery->commandType == CMD_INSERT)
-			{
-				return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
-									 "Router planner doesn't support INSERT common table expressions.",
-									 NULL, NULL);
-			}
-
-
 			if (cteQuery->hasForUpdate &&
 				FindNodeMatchingCheckFunctionInRangeTableList(cteQuery->rtable,
 															  IsReferenceTableRTE))

src/backend/distributed/planner/relation_restriction_equivalence.c

@@ -61,6 +61,8 @@ typedef struct AttributeEquivalenceClass
 {
 	uint32 equivalenceId;
 	List *equivalentAttributes;
+
+	Index unionQueryPartitionKeyIndex;
 } AttributeEquivalenceClass;
 
 /*
@@ -83,7 +85,8 @@ typedef struct AttributeEquivalenceClassMember
 
 
 static bool ContextContainsLocalRelation(RelationRestrictionContext *restrictionContext);
-static Var * FindUnionAllVar(PlannerInfo *root, List *appendRelList, Oid relationOid,
+static int RangeTableOffsetCompat(PlannerInfo *root, AppendRelInfo *appendRelInfo);
+static Var * FindUnionAllVar(PlannerInfo *root, List *translatedVars, Oid relationOid,
 							 Index relationRteIndex, Index *partitionKeyIndex);
 static bool ContainsMultipleDistributedRelations(PlannerRestrictionContext *
 												 plannerRestrictionContext);
@@ -91,11 +94,11 @@ static List * GenerateAttributeEquivalencesForRelationRestrictions(
 	RelationRestrictionContext *restrictionContext);
 static AttributeEquivalenceClass * AttributeEquivalenceClassForEquivalenceClass(
 	EquivalenceClass *plannerEqClass, RelationRestriction *relationRestriction);
-static void AddToAttributeEquivalenceClass(AttributeEquivalenceClass **
+static void AddToAttributeEquivalenceClass(AttributeEquivalenceClass *
 										   attributeEquivalenceClass,
 										   PlannerInfo *root, Var *varToBeAdded);
 static void AddRteSubqueryToAttributeEquivalenceClass(AttributeEquivalenceClass *
-													  *attributeEquivalenceClass,
+													  attributeEquivalenceClass,
 													  RangeTblEntry *
 													  rangeTableEntry,
 													  PlannerInfo *root,
@@ -103,17 +106,17 @@ static void AddRteSubqueryToAttributeEquivalenceClass(AttributeEquivalenceClass
 static Query * GetTargetSubquery(PlannerInfo *root, RangeTblEntry *rangeTableEntry,
 								 Var *varToBeAdded);
 static void AddUnionAllSetOperationsToAttributeEquivalenceClass(
-	AttributeEquivalenceClass **
+	AttributeEquivalenceClass *
 	attributeEquivalenceClass,
 	PlannerInfo *root,
 	Var *varToBeAdded);
-static void AddUnionSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass **
+static void AddUnionSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass *
 															 attributeEquivalenceClass,
 															 PlannerInfo *root,
 															 SetOperationStmt *
 															 setOperation,
 															 Var *varToBeAdded);
-static void AddRteRelationToAttributeEquivalenceClass(AttributeEquivalenceClass **
+static void AddRteRelationToAttributeEquivalenceClass(AttributeEquivalenceClass *
 													  attrEquivalenceClass,
 													  RangeTblEntry *rangeTableEntry,
 													  Var *varToBeAdded);
@@ -141,7 +144,7 @@ static AttributeEquivalenceClass * GenerateEquivalenceClassForRelationRestrictio
 	RelationRestrictionContext
 	*
 	relationRestrictionContext);
-static void ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass **
+static void ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass *
 													  firstClass,
 													  AttributeEquivalenceClass *
 													  secondClass);
@@ -156,9 +159,13 @@ static JoinRestrictionContext * FilterJoinRestrictionContext(
 static bool RangeTableArrayContainsAnyRTEIdentities(RangeTblEntry **rangeTableEntries, int
 													rangeTableArrayLength, Relids
 													queryRteIdentities);
-static int RangeTableOffsetCompat(PlannerInfo *root, AppendRelInfo *appendRelInfo);
 static Relids QueryRteIdentities(Query *queryTree);
 
+#if PG_VERSION_NUM >= PG_VERSION_13
+static int ParentCountPriorToAppendRel(List *appendRelList, AppendRelInfo *appendRelInfo);
+#endif
+
+
 /*
  * AllDistributionKeysInQueryAreEqual returns true if either
  *    (i)  there exists join in the query and all relations joined on their
@@ -249,7 +256,7 @@ SafeToPushdownUnionSubquery(PlannerRestrictionContext *plannerRestrictionContext
 		plannerRestrictionContext->relationRestrictionContext;
 	JoinRestrictionContext *joinRestrictionContext =
 		plannerRestrictionContext->joinRestrictionContext;
-	Index unionQueryPartitionKeyIndex = 0;
+
 	AttributeEquivalenceClass *attributeEquivalence =
 		palloc0(sizeof(AttributeEquivalenceClass));
 	ListCell *relationRestrictionCell = NULL;
@@ -279,7 +286,8 @@ SafeToPushdownUnionSubquery(PlannerRestrictionContext *plannerRestrictionContext
 		 */
 		if (appendRelList != NULL)
 		{
-			varToBeAdded = FindUnionAllVar(relationPlannerRoot, appendRelList,
+			varToBeAdded = FindUnionAllVar(relationPlannerRoot,
+										   relationRestriction->translatedVars,
 										   relationRestriction->relationId,
 										   relationRestriction->index,
 										   &partitionKeyIndex);
@@ -323,17 +331,17 @@ SafeToPushdownUnionSubquery(PlannerRestrictionContext *plannerRestrictionContext
 		 * we check whether all the relations have partition keys in the
 		 * same position.
 		 */
-		if (unionQueryPartitionKeyIndex == InvalidAttrNumber)
+		if (attributeEquivalence->unionQueryPartitionKeyIndex == InvalidAttrNumber)
 		{
-			unionQueryPartitionKeyIndex = partitionKeyIndex;
+			attributeEquivalence->unionQueryPartitionKeyIndex = partitionKeyIndex;
 		}
-		else if (unionQueryPartitionKeyIndex != partitionKeyIndex)
+		else if (attributeEquivalence->unionQueryPartitionKeyIndex != partitionKeyIndex)
 		{
 			continue;
 		}
 
 		Assert(varToBeAdded != NULL);
-		AddToAttributeEquivalenceClass(&attributeEquivalence, relationPlannerRoot,
+		AddToAttributeEquivalenceClass(attributeEquivalence, relationPlannerRoot,
 									   varToBeAdded);
 	}
 
@@ -373,66 +381,74 @@ SafeToPushdownUnionSubquery(PlannerRestrictionContext *plannerRestrictionContext
 }
 
 
+/*
+ * RangeTableOffsetCompat returns the range table offset(in glob->finalrtable) for the appendRelInfo.
+ * For PG < 13 this is a no op.
+ */
+static int
+RangeTableOffsetCompat(PlannerInfo *root, AppendRelInfo *appendRelInfo)
+{
+	#if PG_VERSION_NUM >= PG_VERSION_13
+	int parentCount = ParentCountPriorToAppendRel(root->append_rel_list, appendRelInfo);
+	int skipParentCount = parentCount - 1;
+
+	int i = 1;
+	for (; i < root->simple_rel_array_size; i++)
+	{
+		RangeTblEntry *rte = root->simple_rte_array[i];
+		if (rte->inh)
+		{
+			/*
+			 * We skip the previous parents because we want to find the offset
+			 * for the given append rel info.
+			 */
+			if (skipParentCount > 0)
+			{
+				skipParentCount--;
+				continue;
+			}
+			break;
+		}
+	}
+	int indexInRtable = (i - 1);
+
+	/*
+	 * Postgres adds the global rte array size to parent_relid as an offset.
+	 * Here we do the reverse operation: Commit on postgres side:
+	 * 6ef77cf46e81f45716ec981cb08781d426181378
+	 */
+	int parentRelIndex = appendRelInfo->parent_relid - 1;
+	return parentRelIndex - indexInRtable;
+	#else
+	return 0;
+	#endif
+}
+
+
 /*
  * FindUnionAllVar finds the variable used in union all for the side that has
  * relationRteIndex as its index and the same varattno as the partition key of
  * the given relation with relationOid.
  */
 static Var *
-FindUnionAllVar(PlannerInfo *root, List *appendRelList, Oid relationOid,
+FindUnionAllVar(PlannerInfo *root, List *translatedVars, Oid relationOid,
 				Index relationRteIndex, Index *partitionKeyIndex)
 {
-	ListCell *appendRelCell = NULL;
-	AppendRelInfo *targetAppendRelInfo = NULL;
-	AttrNumber childAttrNumber = 0;
-
-	*partitionKeyIndex = 0;
-
-	/* iterate on the queries that are part of UNION ALL subselects */
-	foreach(appendRelCell, appendRelList)
-	{
-		AppendRelInfo *appendRelInfo = (AppendRelInfo *) lfirst(appendRelCell);
-
-
-		int rtoffset = RangeTableOffsetCompat(root, appendRelInfo);
-
-		/*
-		 * We're only interested in the child rel that is equal to the
-		 * relation we're investigating.
-		 */
-		if (appendRelInfo->child_relid - rtoffset == relationRteIndex)
-		{
-			targetAppendRelInfo = appendRelInfo;
-			break;
-		}
-	}
-
-	if (!targetAppendRelInfo)
+	if (!IsCitusTableType(relationOid, STRICTLY_PARTITIONED_DISTRIBUTED_TABLE))
 	{
+		/* we only care about hash and range partitioned tables */
+		*partitionKeyIndex = 0;
 		return NULL;
 	}
 
 	Var *relationPartitionKey = DistPartitionKeyOrError(relationOid);
 
-	#if PG_VERSION_NUM >= PG_VERSION_13
-	for (; childAttrNumber < targetAppendRelInfo->num_child_cols; childAttrNumber++)
-	{
-		int curAttNo = targetAppendRelInfo->parent_colnos[childAttrNumber];
-		if (curAttNo == relationPartitionKey->varattno)
-		{
-			*partitionKeyIndex = (childAttrNumber + 1);
-			int rtoffset = RangeTableOffsetCompat(root, targetAppendRelInfo);
-			relationPartitionKey->varno = targetAppendRelInfo->child_relid - rtoffset;
-			return relationPartitionKey;
-		}
-	}
-	#else
+	AttrNumber childAttrNumber = 0;
+	*partitionKeyIndex = 0;
 	ListCell *translatedVarCell;
-	List *translaterVars = targetAppendRelInfo->translated_vars;
-	foreach(translatedVarCell, translaterVars)
+	foreach(translatedVarCell, translatedVars)
 	{
 		Node *targetNode = (Node *) lfirst(translatedVarCell);
-
 		childAttrNumber++;
 
 		if (!IsA(targetNode, Var))
@@ -449,7 +465,6 @@ FindUnionAllVar(PlannerInfo *root, List *appendRelList, Oid relationOid,
 			return targetVar;
 		}
 	}
-	#endif
 	return NULL;
 }
 
@@ -580,7 +595,6 @@ GenerateAllAttributeEquivalences(PlannerRestrictionContext *plannerRestrictionCo
 	JoinRestrictionContext *joinRestrictionContext =
 		plannerRestrictionContext->joinRestrictionContext;
 
-
 	/* reset the equivalence id counter per call to prevent overflows */
 	attributeEquivalenceId = 1;
 
@@ -788,14 +802,14 @@ AttributeEquivalenceClassForEquivalenceClass(EquivalenceClass *plannerEqClass,
 										equivalenceParam, &outerNodeRoot);
 			if (expressionVar)
 			{
-				AddToAttributeEquivalenceClass(&attributeEquivalence, outerNodeRoot,
+				AddToAttributeEquivalenceClass(attributeEquivalence, outerNodeRoot,
 											   expressionVar);
 			}
 		}
 		else if (IsA(strippedEquivalenceExpr, Var))
 		{
 			expressionVar = (Var *) strippedEquivalenceExpr;
-			AddToAttributeEquivalenceClass(&attributeEquivalence, plannerInfo,
+			AddToAttributeEquivalenceClass(attributeEquivalence, plannerInfo,
 										   expressionVar);
 		}
 	}
@@ -978,7 +992,7 @@ GenerateCommonEquivalence(List *attributeEquivalenceList,
 			if (AttributeClassContainsAttributeClassMember(attributeEquialanceMember,
 														   commonEquivalenceClass))
 			{
-				ListConcatUniqueAttributeClassMemberLists(&commonEquivalenceClass,
+				ListConcatUniqueAttributeClassMemberLists(commonEquivalenceClass,
 														  currentEquivalenceClass);
 
 				addedEquivalenceIds = bms_add_member(addedEquivalenceIds,
@@ -1058,7 +1072,7 @@ GenerateEquivalenceClassForRelationRestriction(
  * firstClass.
  */
 static void
-ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass **firstClass,
+ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass *firstClass,
 										  AttributeEquivalenceClass *secondClass)
 {
 	ListCell *equivalenceClassMemberCell = NULL;
@@ -1069,13 +1083,13 @@ ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass **firstClass
 		AttributeEquivalenceClassMember *newEqMember =
 			(AttributeEquivalenceClassMember *) lfirst(equivalenceClassMemberCell);
 
-		if (AttributeClassContainsAttributeClassMember(newEqMember, *firstClass))
+		if (AttributeClassContainsAttributeClassMember(newEqMember, firstClass))
 		{
 			continue;
 		}
 
-		(*firstClass)->equivalentAttributes = lappend((*firstClass)->equivalentAttributes,
-													  newEqMember);
+		firstClass->equivalentAttributes = lappend(firstClass->equivalentAttributes,
+												   newEqMember);
 	}
 }
 
@@ -1150,10 +1164,10 @@ GenerateAttributeEquivalencesForJoinRestrictions(JoinRestrictionContext *
 				sizeof(AttributeEquivalenceClass));
 			attributeEquivalence->equivalenceId = attributeEquivalenceId++;
 
-			AddToAttributeEquivalenceClass(&attributeEquivalence,
+			AddToAttributeEquivalenceClass(attributeEquivalence,
 										   joinRestriction->plannerInfo, leftVar);
 
-			AddToAttributeEquivalenceClass(&attributeEquivalence,
+			AddToAttributeEquivalenceClass(attributeEquivalence,
 										   joinRestriction->plannerInfo, rightVar);
 
 			attributeEquivalenceList =
@@ -1194,7 +1208,7 @@ GenerateAttributeEquivalencesForJoinRestrictions(JoinRestrictionContext *
  *               equivalence class
  */
 static void
-AddToAttributeEquivalenceClass(AttributeEquivalenceClass **attributeEquivalenceClass,
+AddToAttributeEquivalenceClass(AttributeEquivalenceClass *attributeEquivalenceClass,
 							   PlannerInfo *root, Var *varToBeAdded)
 {
 	/* punt if it's a whole-row var rather than a plain column reference */
@@ -1233,9 +1247,10 @@ AddToAttributeEquivalenceClass(AttributeEquivalenceClass **attributeEquivalenceC
  */
 static void
 AddRteSubqueryToAttributeEquivalenceClass(AttributeEquivalenceClass
-										  **attributeEquivalenceClass,
+										  *attributeEquivalenceClass,
 										  RangeTblEntry *rangeTableEntry,
-										  PlannerInfo *root, Var *varToBeAdded)
+										  PlannerInfo *root,
+										  Var *varToBeAdded)
 {
 	RelOptInfo *baseRelOptInfo = find_base_rel(root, varToBeAdded->varno);
 	Query *targetSubquery = GetTargetSubquery(root, rangeTableEntry, varToBeAdded);
@@ -1355,7 +1370,7 @@ GetTargetSubquery(PlannerInfo *root, RangeTblEntry *rangeTableEntry, Var *varToB
  * var the given equivalence class.
  */
 static void
-AddUnionAllSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass **
+AddUnionAllSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass *
 													attributeEquivalenceClass,
 													PlannerInfo *root,
 													Var *varToBeAdded)
@@ -1377,41 +1392,101 @@ AddUnionAllSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass **
 			continue;
 		}
 		int rtoffset = RangeTableOffsetCompat(root, appendRelInfo);
+		int childRelId = appendRelInfo->child_relid - rtoffset;
 
-		/* set the varno accordingly for this specific child */
-		varToBeAdded->varno = appendRelInfo->child_relid - rtoffset;
+		if (root->simple_rel_array_size <= childRelId)
+		{
+			/* we prefer to return over an Assert or error to be defensive */
+			return;
+		}
 
-		AddToAttributeEquivalenceClass(attributeEquivalenceClass, root,
-									   varToBeAdded);
-	}
-}
-
-
-/*
- * RangeTableOffsetCompat returns the range table offset(in glob->finalrtable) for the appendRelInfo.
- * For PG < 13 this is a no op.
- */
-static int
-RangeTableOffsetCompat(PlannerInfo *root, AppendRelInfo *appendRelInfo)
-{
-	#if PG_VERSION_NUM >= PG_VERSION_13
-	int i = 1;
-	for (; i < root->simple_rel_array_size; i++)
-	{
-		RangeTblEntry *rte = root->simple_rte_array[i];
+		RangeTblEntry *rte = root->simple_rte_array[childRelId];
 		if (rte->inh)
 		{
-			break;
+			/*
+			 * This code-path may require improvements. If a leaf of a UNION ALL
+			 * (e.g., an entry in appendRelList) itself is another UNION ALL
+			 * (e.g., rte->inh = true), the logic here might get into an infinite
+			 * recursion.
+			 *
+			 * The downside of "continue" here is that certain UNION ALL queries
+			 * that are safe to pushdown may not be pushed down.
+			 */
+			continue;
+		}
+		else if (rte->rtekind == RTE_RELATION)
+		{
+			Index partitionKeyIndex = 0;
+			List *translatedVars = TranslatedVarsForRteIdentity(GetRTEIdentity(rte));
+			Var *varToBeAddedOnUnionAllSubquery =
+				FindUnionAllVar(root, translatedVars, rte->relid, childRelId,
+								&partitionKeyIndex);
+			if (partitionKeyIndex == 0)
+			{
+				/* no partition key on the target list */
+				continue;
+			}
+
+			if (attributeEquivalenceClass->unionQueryPartitionKeyIndex == 0)
+			{
+				/* the first partition key index we found */
+				attributeEquivalenceClass->unionQueryPartitionKeyIndex =
+					partitionKeyIndex;
+			}
+			else if (attributeEquivalenceClass->unionQueryPartitionKeyIndex !=
+					 partitionKeyIndex)
+			{
+				/*
+				 * Partition keys on the leaves of the UNION ALL queries on
+				 * different ordinal positions. We cannot pushdown, so skip.
+				 */
+				continue;
+			}
+
+			if (varToBeAddedOnUnionAllSubquery != NULL)
+			{
+				AddToAttributeEquivalenceClass(attributeEquivalenceClass, root,
+											   varToBeAddedOnUnionAllSubquery);
+			}
+		}
+		else
+		{
+			/* set the varno accordingly for this specific child */
+			varToBeAdded->varno = childRelId;
+
+			AddToAttributeEquivalenceClass(attributeEquivalenceClass, root,
+										   varToBeAdded);
 		}
 	}
-	int indexInRtable = (i - 1);
-	return appendRelInfo->parent_relid - 1 - (indexInRtable);
-	#else
-	return 0;
-	#endif
 }
 
 
+#if PG_VERSION_NUM >= PG_VERSION_13
+
+/*
+ * ParentCountPriorToAppendRel returns the number of parents that come before
+ * the given append rel info.
+ */
+static int
+ParentCountPriorToAppendRel(List *appendRelList, AppendRelInfo *targetAppendRelInfo)
+{
+	int targetParentIndex = targetAppendRelInfo->parent_relid;
+	Bitmapset *parent_ids = NULL;
+	AppendRelInfo *appendRelInfo = NULL;
+	foreach_ptr(appendRelInfo, appendRelList)
+	{
+		int curParentIndex = appendRelInfo->parent_relid;
+		if (curParentIndex <= targetParentIndex)
+		{
+			parent_ids = bms_add_member(parent_ids, curParentIndex);
+		}
+	}
+	return bms_num_members(parent_ids);
+}
+
+
+#endif
+
 /*
  * AddUnionSetOperationsToAttributeEquivalenceClass recursively iterates on all the
  * setOperations and adds each corresponding target entry to the given equivalence
@@ -1422,7 +1497,7 @@ RangeTableOffsetCompat(PlannerInfo *root, AppendRelInfo *appendRelInfo)
  * messages.
  */
 static void
-AddUnionSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass **
+AddUnionSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass *
 												 attributeEquivalenceClass,
 												 PlannerInfo *root,
 												 SetOperationStmt *setOperation,
@@ -1450,7 +1525,7 @@ AddUnionSetOperationsToAttributeEquivalenceClass(AttributeEquivalenceClass **
  * the input rte to be an RTE_RELATION.
  */
 static void
-AddRteRelationToAttributeEquivalenceClass(AttributeEquivalenceClass **
+AddRteRelationToAttributeEquivalenceClass(AttributeEquivalenceClass *
 										  attrEquivalenceClass,
 										  RangeTblEntry *rangeTableEntry,
 										  Var *varToBeAdded)
@@ -1487,8 +1562,8 @@ AddRteRelationToAttributeEquivalenceClass(AttributeEquivalenceClass **
 	attributeEqMember->rteIdentity = GetRTEIdentity(rangeTableEntry);
 	attributeEqMember->relationId = rangeTableEntry->relid;
 
-	(*attrEquivalenceClass)->equivalentAttributes =
-		lappend((*attrEquivalenceClass)->equivalentAttributes,
+	attrEquivalenceClass->equivalentAttributes =
+		lappend(attrEquivalenceClass->equivalentAttributes,
 				attributeEqMember);
 }
 

src/backend/distributed/relay/relay_event_utility.c

@@ -556,30 +556,6 @@ RelayEventExtendNames(Node *parseTree, char *schemaName, uint64 shardId)
 
 				AppendShardIdToName(oldRelationName, shardId);
 				AppendShardIdToName(newRelationName, shardId);
-
-				/*
-				 * PostgreSQL creates array types for each ordinary table, with
-				 * the same name plus a prefix of '_'.
-				 *
-				 * ALTER TABLE ... RENAME TO ... also renames the underlying
-				 * array type, and the DDL is run in parallel connections over
-				 * all the placements and shards at once. Concurrent access
-				 * here deadlocks.
-				 *
-				 * Let's provide an easier to understand error message here
-				 * than the deadlock one.
-				 *
-				 * See also https://github.com/citusdata/citus/issues/1664
-				 */
-				int newRelationNameLength = strlen(*newRelationName);
-				if (newRelationNameLength >= (NAMEDATALEN - 1))
-				{
-					ereport(ERROR,
-							(errcode(ERRCODE_NAME_TOO_LONG),
-							 errmsg(
-								 "shard name %s exceeds %d characters",
-								 *newRelationName, NAMEDATALEN - 1)));
-				}
 			}
 			else if (objectType == OBJECT_COLUMN)
 			{

src/backend/distributed/shared_library_init.c

@@ -701,6 +701,19 @@ RegisterCitusConfigVariables(void)
 		GUC_NO_SHOW_ALL,
 		NoticeIfSubqueryPushdownEnabled, NULL, NULL);
 
+	DefineCustomIntVariable(
+		"citus.remote_copy_flush_threshold",
+		gettext_noop("Sets the threshold for remote copy to be flushed."),
+		gettext_noop("When sending data over remote connections via the COPY protocol, "
+					 "bytes are first buffered internally by libpq. If the number of "
+					 "bytes buffered exceeds the threshold, Citus waits for all the "
+					 "bytes to flush."),
+		&RemoteCopyFlushThreshold,
+		8 * 1024 * 1024, 0, INT_MAX,
+		PGC_USERSET,
+		GUC_UNIT_BYTE | GUC_NO_SHOW_ALL,
+		NULL, NULL, NULL);
+
 	DefineCustomIntVariable(
 		"citus.local_copy_flush_threshold",
 		gettext_noop("Sets the threshold for local copy to be flushed."),
@@ -1238,6 +1251,16 @@ RegisterCitusConfigVariables(void)
 		GUC_STANDARD,
 		NULL, NULL, NULL);
 
+	DefineCustomIntVariable(
+		"citus.max_cached_connection_lifetime",
+		gettext_noop("Sets the maximum lifetime of cached connections to other nodes."),
+		NULL,
+		&MaxCachedConnectionLifetime,
+		10 * MS_PER_MINUTE, -1, INT_MAX,
+		PGC_USERSET,
+		GUC_UNIT_MS | GUC_STANDARD,
+		NULL, NULL, NULL);
+
 	DefineCustomIntVariable(
 		"citus.repartition_join_bucket_count_per_node",
 		gettext_noop("Sets the bucket size for repartition joins per node"),

src/backend/distributed/sql/citus--10.0-1--10.0-2.sql

@@ -0,0 +1,5 @@
+-- citus--10.0-1--10.0-2
+
+#include "../../columnar/sql/columnar--10.0-1--10.0-2.sql"
+
+GRANT SELECT ON public.citus_tables TO public;

src/backend/distributed/sql/citus--10.0-2--10.0-3.sql

@@ -0,0 +1,18 @@
+-- citus--10.0-2--10.0-3
+
+#include "udfs/citus_update_table_statistics/10.0-3.sql"
+
+CREATE OR REPLACE FUNCTION master_update_table_statistics(relation regclass)
+RETURNS VOID
+    LANGUAGE C STRICT
+    AS 'MODULE_PATHNAME', $$citus_update_table_statistics$$;
+COMMENT ON FUNCTION pg_catalog.master_update_table_statistics(regclass)
+	IS 'updates shard statistics of the given table';
+
+CREATE OR REPLACE FUNCTION pg_catalog.citus_get_active_worker_nodes(OUT node_name text, OUT node_port bigint)
+    RETURNS SETOF record
+    LANGUAGE C STRICT ROWS 100
+    AS 'MODULE_PATHNAME', $$citus_get_active_worker_nodes$$;
+COMMENT ON FUNCTION pg_catalog.citus_get_active_worker_nodes()
+    IS 'fetch set of active worker nodes';
+

src/backend/distributed/sql/downgrades/citus--10.0-2--10.0-1.sql

@@ -0,0 +1,4 @@
+/* citus--10.0-2--10.0-1.sql */
+#include "../../../columnar/sql/downgrades/columnar--10.0-2--10.0-1.sql"
+
+REVOKE SELECT ON public.citus_tables FROM public;

src/backend/distributed/sql/downgrades/citus--10.0-3--10.0-2.sql

@@ -0,0 +1,26 @@
+-- citus--10.0-3--10.0-2
+-- this is a downgrade path that will revert the changes made in citus--10.0-2--10.0-3.sql
+
+DROP FUNCTION pg_catalog.citus_update_table_statistics(regclass);
+
+#include "../udfs/citus_update_table_statistics/10.0-1.sql"
+
+CREATE OR REPLACE FUNCTION master_update_table_statistics(relation regclass)
+RETURNS VOID AS $$
+DECLARE
+	colocated_tables regclass[];
+BEGIN
+	SELECT get_colocated_table_array(relation) INTO colocated_tables;
+
+	PERFORM
+		master_update_shard_statistics(shardid)
+	FROM
+		pg_dist_shard
+	WHERE
+		logicalrelid = ANY (colocated_tables);
+END;
+$$ LANGUAGE 'plpgsql';
+COMMENT ON FUNCTION master_update_table_statistics(regclass)
+	IS 'updates shard statistics of the given table and its colocated tables';
+
+DROP FUNCTION pg_catalog.citus_get_active_worker_nodes(OUT text, OUT bigint);

src/backend/distributed/sql/udfs/citus_update_table_statistics/10.0-3.sql

@@ -0,0 +1,6 @@
+CREATE OR REPLACE FUNCTION pg_catalog.citus_update_table_statistics(relation regclass)
+	RETURNS VOID
+    LANGUAGE C STRICT
+    AS 'MODULE_PATHNAME', $$citus_update_table_statistics$$;
+COMMENT ON FUNCTION pg_catalog.citus_update_table_statistics(regclass)
+	IS 'updates shard statistics of the given table';

src/backend/distributed/sql/udfs/citus_update_table_statistics/latest.sql

@@ -1,17 +1,6 @@
-CREATE FUNCTION pg_catalog.citus_update_table_statistics(relation regclass)
-RETURNS VOID AS $$
-DECLARE
-	colocated_tables regclass[];
-BEGIN
-	SELECT get_colocated_table_array(relation) INTO colocated_tables;
-
-	PERFORM
-		master_update_shard_statistics(shardid)
-	FROM
-		pg_dist_shard
-	WHERE
-		logicalrelid = ANY (colocated_tables);
-END;
-$$ LANGUAGE 'plpgsql';
+CREATE OR REPLACE FUNCTION pg_catalog.citus_update_table_statistics(relation regclass)
+	RETURNS VOID
+    LANGUAGE C STRICT
+    AS 'MODULE_PATHNAME', $$citus_update_table_statistics$$;
 COMMENT ON FUNCTION pg_catalog.citus_update_table_statistics(regclass)
-	IS 'updates shard statistics of the given table and its colocated tables';
+	IS 'updates shard statistics of the given table';

src/backend/distributed/sql/udfs/time_partitions/10.0-1.sql

@@ -5,12 +5,13 @@ FROM (
   FROM pg_class c
   JOIN pg_inherits i ON (c.oid = inhrelid)
   JOIN pg_partitioned_table p ON (inhparent = partrelid)
-  JOIN pg_attribute a ON (partrelid = attrelid AND ARRAY[attnum] <@ string_to_array(partattrs::text, ' ')::int2[])
+  JOIN pg_attribute a ON (partrelid = attrelid)
   JOIN pg_type t ON (atttypid = t.oid)
   JOIN pg_namespace tn ON (t.typnamespace = tn.oid)
   LEFT JOIN pg_am am ON (c.relam = am.oid),
   pg_catalog.time_partition_range(c.oid)
   WHERE c.relpartbound IS NOT NULL AND p.partstrat = 'r' AND p.partnatts = 1
+  AND a.attnum = ANY(partattrs::int2[])
 ) partitions
 ORDER BY partrelid::text, lower_bound;
 

src/backend/distributed/sql/udfs/time_partitions/latest.sql

@@ -5,12 +5,13 @@ FROM (
   FROM pg_class c
   JOIN pg_inherits i ON (c.oid = inhrelid)
   JOIN pg_partitioned_table p ON (inhparent = partrelid)
-  JOIN pg_attribute a ON (partrelid = attrelid AND ARRAY[attnum] <@ string_to_array(partattrs::text, ' ')::int2[])
+  JOIN pg_attribute a ON (partrelid = attrelid)
   JOIN pg_type t ON (atttypid = t.oid)
   JOIN pg_namespace tn ON (t.typnamespace = tn.oid)
   LEFT JOIN pg_am am ON (c.relam = am.oid),
   pg_catalog.time_partition_range(c.oid)
   WHERE c.relpartbound IS NOT NULL AND p.partstrat = 'r' AND p.partnatts = 1
+  AND a.attnum = ANY(partattrs::int2[])
 ) partitions
 ORDER BY partrelid::text, lower_bound;
 

src/backend/distributed/test/xact_stats.c

@@ -18,9 +18,14 @@
 #include "miscadmin.h"
 #include "pgstat.h"
 
+#include "distributed/transaction_management.h"
+
+
 static Size MemoryContextTotalSpace(MemoryContext context);
 
 PG_FUNCTION_INFO_V1(top_transaction_context_size);
+PG_FUNCTION_INFO_V1(coordinated_transaction_should_use_2PC);
+
 
 /*
  * top_transaction_context_size returns current size of TopTransactionContext.
@@ -54,3 +59,20 @@ MemoryContextTotalSpace(MemoryContext context)
 
 	return totalSpace;
 }
+
+
+/*
+ * coordinated_transaction_should_use_2PC returns true if the transaction is in a
+ * coordinated transaction and uses 2PC. If the transaction is nott in a
+ * coordinated transaction, the function throws an error.
+ */
+Datum
+coordinated_transaction_should_use_2PC(PG_FUNCTION_ARGS)
+{
+	if (!InCoordinatedTransaction())
+	{
+		ereport(ERROR, (errmsg("The transaction is not a coordinated transaction")));
+	}
+
+	PG_RETURN_BOOL(GetCoordinatedTransactionShouldUse2PC());
+}

src/backend/distributed/transaction/remote_transaction.c

@@ -20,6 +20,7 @@
 #include "distributed/connection_management.h"
 #include "distributed/listutils.h"
 #include "distributed/metadata_cache.h"
+#include "distributed/placement_connection.h"
 #include "distributed/remote_commands.h"
 #include "distributed/remote_transaction.h"
 #include "distributed/transaction_identifier.h"
@@ -782,8 +783,16 @@ CoordinatedRemoteTransactionsPrepare(void)
 			continue;
 		}
 
-		StartRemoteTransactionPrepare(connection);
-		connectionList = lappend(connectionList, connection);
+		/*
+		 * Check if any DML or DDL is executed over the connection on any
+		 * placement/table. If yes, we start preparing the transaction, otherwise
+		 * we skip prepare since the connection didn't perform any write (read-only)
+		 */
+		if (ConnectionModifiedPlacement(connection))
+		{
+			StartRemoteTransactionPrepare(connection);
+			connectionList = lappend(connectionList, connection);
+		}
 	}
 
 	bool raiseInterrupts = true;
@@ -798,6 +807,10 @@ CoordinatedRemoteTransactionsPrepare(void)
 
 		if (transaction->transactionState != REMOTE_TRANS_PREPARING)
 		{
+			/*
+			 * Verify that the connection didn't modify any placement
+			 */
+			Assert(!ConnectionModifiedPlacement(connection));
 			continue;
 		}
 

src/backend/distributed/transaction/transaction_management.c

@@ -96,9 +96,16 @@ MemoryContext CommitContext = NULL;
 /*
  * Should this coordinated transaction use 2PC? Set by
  * CoordinatedTransactionUse2PC(), e.g. if DDL was issued and
- * MultiShardCommitProtocol was set to 2PC.
+ * MultiShardCommitProtocol was set to 2PC. But, even if this
+ * flag is set, the transaction manager is smart enough to only
+ * do 2PC on the remote connections that did a modification.
+ *
+ * As a variable name ShouldCoordinatedTransactionUse2PC could
+ * be improved. We use CoordinatedTransactionShouldUse2PC() as the
+ * public API function, hence couldn't come up with a better name
+ * for the underlying variable at the moment.
  */
-bool CoordinatedTransactionUses2PC = false;
+bool ShouldCoordinatedTransactionUse2PC = false;
 
 /* if disabled, distributed statements in a function may run as separate transactions */
 bool FunctionOpensTransactionBlock = true;
@@ -183,15 +190,29 @@ InCoordinatedTransaction(void)
 
 
 /*
- * CoordinatedTransactionUse2PC() signals that the current coordinated
+ * CoordinatedTransactionShouldUse2PC() signals that the current coordinated
  * transaction should use 2PC to commit.
+ *
+ * Note that even if 2PC is enabled, it is only used for connections that make
+ * modification (DML or DDL).
  */
 void
-CoordinatedTransactionUse2PC(void)
+CoordinatedTransactionShouldUse2PC(void)
 {
 	Assert(InCoordinatedTransaction());
 
-	CoordinatedTransactionUses2PC = true;
+	ShouldCoordinatedTransactionUse2PC = true;
+}
+
+
+/*
+ * GetCoordinatedTransactionShouldUse2PC is a wrapper function to read the value
+ * of CoordinatedTransactionShouldUse2PCFlag.
+ */
+bool
+GetCoordinatedTransactionShouldUse2PC(void)
+{
+	return ShouldCoordinatedTransactionUse2PC;
 }
 
 
@@ -297,28 +318,8 @@ CoordinatedTransactionCallback(XactEvent event, void *arg)
 			/* stop propagating notices from workers, we know the query is failed */
 			DisableWorkerMessagePropagation();
 
-			/*
-			 * FIXME: Add warning for the COORD_TRANS_COMMITTED case. That
-			 * can be reached if this backend fails after the
-			 * XACT_EVENT_PRE_COMMIT state.
-			 */
+			RemoveIntermediateResultsDirectory();
 
-			/*
-			 * Call other parts of citus that need to integrate into
-			 * transaction management. Do so before doing other work, so the
-			 * callbacks still can perform work if needed.
-			 */
-			{
-				/*
-				 * On Windows it's not possible to delete a file before you've closed all
-				 * handles to it (rmdir will return success but not take effect). Since
-				 * we're in an ABORT handler it's very likely that not all handles have
-				 * been closed; force them closed here before running
-				 * RemoveIntermediateResultsDirectory.
-				 */
-				AtEOXact_Files(false);
-				RemoveIntermediateResultsDirectory();
-			}
 			ResetShardPlacementTransactionState();
 
 			/* handles both already prepared and open transactions */
@@ -425,7 +426,7 @@ CoordinatedTransactionCallback(XactEvent event, void *arg)
 			 */
 			MarkFailedShardPlacements();
 
-			if (CoordinatedTransactionUses2PC)
+			if (ShouldCoordinatedTransactionUse2PC)
 			{
 				CoordinatedRemoteTransactionsPrepare();
 				CurrentCoordinatedTransactionState = COORD_TRANS_PREPARED;
@@ -453,7 +454,7 @@ CoordinatedTransactionCallback(XactEvent event, void *arg)
 			 * Check again whether shards/placement successfully
 			 * committed. This handles failure at COMMIT/PREPARE time.
 			 */
-			PostCommitMarkFailedShardPlacements(CoordinatedTransactionUses2PC);
+			PostCommitMarkFailedShardPlacements(ShouldCoordinatedTransactionUse2PC);
 			break;
 		}
 
@@ -485,7 +486,7 @@ ResetGlobalVariables()
 	FreeSavedExplainPlan();
 	dlist_init(&InProgressTransactions);
 	activeSetStmts = NULL;
-	CoordinatedTransactionUses2PC = false;
+	ShouldCoordinatedTransactionUse2PC = false;
 	TransactionModifiedNodeMetadata = false;
 	MetadataSyncOnCommit = false;
 	ResetWorkerErrorIndication();

src/backend/distributed/transaction/worker_transaction.c

@@ -96,7 +96,7 @@ SendCommandToWorkerAsUser(const char *nodeName, int32 nodePort, const char *node
 	uint32 connectionFlags = 0;
 
 	UseCoordinatedTransaction();
-	CoordinatedTransactionUse2PC();
+	CoordinatedTransactionShouldUse2PC();
 
 	MultiConnection *transactionConnection = GetNodeUserDatabaseConnection(
 		connectionFlags, nodeName,
@@ -404,7 +404,7 @@ SendCommandToWorkersParamsInternal(TargetWorkerSet targetWorkerSet, const char *
 	List *workerNodeList = TargetWorkerSetNodeList(targetWorkerSet, ShareLock);
 
 	UseCoordinatedTransaction();
-	CoordinatedTransactionUse2PC();
+	CoordinatedTransactionShouldUse2PC();
 
 	/* open connections in parallel */
 	WorkerNode *workerNode = NULL;

src/backend/distributed/utils/foreign_key_relationship.c

@@ -100,9 +100,6 @@ static ForeignConstraintRelationshipNode * CreateOrFindNode(HTAB *adjacencyLists
 															relid);
 static List * GetConnectedListHelper(ForeignConstraintRelationshipNode *node,
 									 bool isReferencing);
-static HTAB * CreateOidVisitedHashSet(void);
-static bool OidVisited(HTAB *oidVisitedMap, Oid oid);
-static void VisitOid(HTAB *oidVisitedMap, Oid oid);
 static List * GetForeignConstraintRelationshipHelper(Oid relationId, bool isReferencing);
 
 
@@ -442,7 +439,7 @@ GetConnectedListHelper(ForeignConstraintRelationshipNode *node, bool isReferenci
  * As hash_create allocates memory in heap, callers are responsible to call
  * hash_destroy when appropriate.
  */
-static HTAB *
+HTAB *
 CreateOidVisitedHashSet(void)
 {
 	HASHCTL info = { 0 };
@@ -464,7 +461,7 @@ CreateOidVisitedHashSet(void)
 /*
  * OidVisited returns true if given oid is visited according to given oid hash-set.
  */
-static bool
+bool
 OidVisited(HTAB *oidVisitedMap, Oid oid)
 {
 	bool found = false;
@@ -476,7 +473,7 @@ OidVisited(HTAB *oidVisitedMap, Oid oid)
 /*
  * VisitOid sets given oid as visited in given hash-set.
  */
-static void
+void
 VisitOid(HTAB *oidVisitedMap, Oid oid)
 {
 	bool found = false;

src/backend/distributed/utils/multi_partitioning_utils.c

@@ -548,13 +548,14 @@ PartitionParentOid(Oid partitionOid)
 
 
 /*
- * LongestPartitionName is a uitility function that returns the partition
- * name which is the longest in terms of number of characters.
+ * PartitionWithLongestNameRelationId is a utility function that returns the
+ * oid of the partition table that has the longest name in terms of number of
+ * characters.
  */
-char *
-LongestPartitionName(Oid parentRelationId)
+Oid
+PartitionWithLongestNameRelationId(Oid parentRelationId)
 {
-	char *longestName = NULL;
+	Oid longestNamePartitionId = InvalidOid;
 	int longestNameLength = 0;
 	List *partitionList = PartitionList(parentRelationId);
 
@@ -565,12 +566,12 @@ LongestPartitionName(Oid parentRelationId)
 		int partitionNameLength = strnlen(partitionName, NAMEDATALEN);
 		if (partitionNameLength > longestNameLength)
 		{
-			longestName = partitionName;
+			longestNamePartitionId = partitionRelationId;
 			longestNameLength = partitionNameLength;
 		}
 	}
 
-	return longestName;
+	return longestNamePartitionId;
 }
 
 

src/backend/distributed/utils/role.c

@@ -41,7 +41,7 @@ alter_role_if_exists(PG_FUNCTION_ARGS)
 
 	Node *parseTree = ParseTreeNode(utilityQuery);
 
-	ProcessUtilityParseTree(parseTree, utilityQuery, PROCESS_UTILITY_TOPLEVEL, NULL,
+	ProcessUtilityParseTree(parseTree, utilityQuery, PROCESS_UTILITY_QUERY, NULL,
 							None_Receiver, NULL);
 
 	PG_RETURN_BOOL(true);
@@ -98,7 +98,7 @@ worker_create_or_alter_role(PG_FUNCTION_ARGS)
 
 		ProcessUtilityParseTree(parseTree,
 								createRoleUtilityQuery,
-								PROCESS_UTILITY_TOPLEVEL,
+								PROCESS_UTILITY_QUERY,
 								NULL,
 								None_Receiver, NULL);
 
@@ -126,7 +126,7 @@ worker_create_or_alter_role(PG_FUNCTION_ARGS)
 
 		ProcessUtilityParseTree(parseTree,
 								alterRoleUtilityQuery,
-								PROCESS_UTILITY_TOPLEVEL,
+								PROCESS_UTILITY_QUERY,
 								NULL,
 								None_Receiver, NULL);
 

src/backend/distributed/utils/shard_utils.c

@@ -12,6 +12,7 @@
 #include "postgres.h"
 
 #include "utils/lsyscache.h"
+#include "distributed/metadata_utility.h"
 #include "distributed/relay_utility.h"
 #include "distributed/shard_utils.h"
 
@@ -36,3 +37,21 @@ GetTableLocalShardOid(Oid citusTableOid, uint64 shardId)
 
 	return shardRelationOid;
 }
+
+
+/*
+ * GetLongestShardName is a utility function that returns the name of the shard of a
+ * table that has the longest name in terms of number of characters.
+ *
+ * Both the Oid and name of the table are required so we can create longest shard names
+ * after a RENAME.
+ */
+char *
+GetLongestShardName(Oid citusTableOid, char *finalRelationName)
+{
+	char *longestShardName = pstrdup(finalRelationName);
+	ShardInterval *shardInterval = LoadShardIntervalWithLongestShardName(citusTableOid);
+	AppendShardIdToName(&longestShardName, shardInterval->shardId);
+
+	return longestShardName;
+}

src/backend/distributed/worker/worker_create_or_replace.c

@@ -111,12 +111,12 @@ worker_create_or_replace_object(PG_FUNCTION_ARGS)
 		RenameStmt *renameStmt = CreateRenameStatement(&address, newName);
 		const char *sqlRenameStmt = DeparseTreeNode((Node *) renameStmt);
 		ProcessUtilityParseTree((Node *) renameStmt, sqlRenameStmt,
-								PROCESS_UTILITY_TOPLEVEL,
+								PROCESS_UTILITY_QUERY,
 								NULL, None_Receiver, NULL);
 	}
 
 	/* apply create statement locally */
-	ProcessUtilityParseTree(parseTree, sqlStatement, PROCESS_UTILITY_TOPLEVEL, NULL,
+	ProcessUtilityParseTree(parseTree, sqlStatement, PROCESS_UTILITY_QUERY, NULL,
 							None_Receiver, NULL);
 
 	/* type has been created */

src/backend/distributed/worker/worker_data_fetch_protocol.c

@@ -396,7 +396,7 @@ worker_apply_shard_ddl_command(PG_FUNCTION_ARGS)
 
 	/* extend names in ddl command and apply extended command */
 	RelayEventExtendNames(ddlCommandNode, schemaName, shardId);
-	ProcessUtilityParseTree(ddlCommandNode, ddlCommand, PROCESS_UTILITY_TOPLEVEL, NULL,
+	ProcessUtilityParseTree(ddlCommandNode, ddlCommand, PROCESS_UTILITY_QUERY, NULL,
 							None_Receiver, NULL);
 
 	PG_RETURN_VOID();
@@ -428,7 +428,7 @@ worker_apply_inter_shard_ddl_command(PG_FUNCTION_ARGS)
 	RelayEventExtendNamesForInterShardCommands(ddlCommandNode, leftShardId,
 											   leftShardSchemaName, rightShardId,
 											   rightShardSchemaName);
-	ProcessUtilityParseTree(ddlCommandNode, ddlCommand, PROCESS_UTILITY_TOPLEVEL, NULL,
+	ProcessUtilityParseTree(ddlCommandNode, ddlCommand, PROCESS_UTILITY_QUERY, NULL,
 							None_Receiver, NULL);
 
 	PG_RETURN_VOID();
@@ -461,7 +461,7 @@ worker_apply_sequence_command(PG_FUNCTION_ARGS)
 	}
 
 	/* run the CREATE SEQUENCE command */
-	ProcessUtilityParseTree(commandNode, commandString, PROCESS_UTILITY_TOPLEVEL, NULL,
+	ProcessUtilityParseTree(commandNode, commandString, PROCESS_UTILITY_QUERY, NULL,
 							None_Receiver, NULL);
 	CommandCounterIncrement();
 
@@ -669,7 +669,7 @@ worker_append_table_to_shard(PG_FUNCTION_ARGS)
 	SetUserIdAndSecContext(CitusExtensionOwner(), SECURITY_LOCAL_USERID_CHANGE);
 
 	ProcessUtilityParseTree((Node *) localCopyCommand, queryString->data,
-							PROCESS_UTILITY_TOPLEVEL, NULL, None_Receiver, NULL);
+							PROCESS_UTILITY_QUERY, NULL, None_Receiver, NULL);
 
 	SetUserIdAndSecContext(savedUserId, savedSecurityContext);
 
@@ -782,7 +782,7 @@ AlterSequenceMinMax(Oid sequenceId, char *schemaName, char *sequenceName,
 
 		/* since the command is an AlterSeqStmt, a dummy command string works fine */
 		ProcessUtilityParseTree((Node *) alterSequenceStatement, dummyString,
-								PROCESS_UTILITY_TOPLEVEL, NULL, None_Receiver, NULL);
+								PROCESS_UTILITY_QUERY, NULL, None_Receiver, NULL);
 	}
 }
 

src/include/distributed/commands.h

@@ -24,6 +24,10 @@
 /* controlled via GUC, should be accessed via EnableLocalReferenceForeignKeys() */
 extern bool EnableLocalReferenceForeignKeys;
 
+extern void SwitchToSequentialAndLocalExecutionIfRelationNameTooLong(Oid relationId,
+																	 char *
+																	 finalRelationName);
+
 
 /*
  * DistributeObjectOps specifies handlers for node/object type pairs.

src/include/distributed/connection_management.h

@@ -200,6 +200,9 @@ extern int NodeConnectionTimeout;
 /* maximum number of connections to cache per worker per session */
 extern int MaxCachedConnectionsPerWorker;
 
+/* maximum lifetime of connections in miliseconds */
+extern int MaxCachedConnectionLifetime;
+
 /* parameters used for outbound connections */
 extern char *NodeConninfo;
 

src/include/distributed/distributed_planner.h

@@ -67,6 +67,9 @@ typedef struct RelationRestriction
 
 	/* list of RootPlanParams for all outer nodes */
 	List *outerPlanParamsList;
+
+	/* list of translated vars, this is copied from postgres since it gets deleted on postgres*/
+	List *translatedVars;
 } RelationRestriction;
 
 typedef struct JoinRestrictionContext
@@ -219,9 +222,9 @@ extern PlannedStmt * distributed_planner(Query *parse,
 #define LOCAL_TABLE_SUBQUERY_CTE_HINT \
 	"Use CTE's or subqueries to select from local tables and use them in joins"
 
-
 extern List * ExtractRangeTableEntryList(Query *query);
 extern bool NeedsDistributedPlanning(Query *query);
+extern List * TranslatedVarsForRteIdentity(int rteIdentity);
 extern struct DistributedPlan * GetDistributedPlan(CustomScan *node);
 extern void multi_relation_restriction_hook(PlannerInfo *root, RelOptInfo *relOptInfo,
 											Index restrictionIndex, RangeTblEntry *rte);

src/include/distributed/foreign_key_relationship.h

@@ -22,5 +22,8 @@ extern List * ReferencingRelationIdList(Oid relationId);
 extern void SetForeignConstraintRelationshipGraphInvalid(void);
 extern bool IsForeignConstraintRelationshipGraphValid(void);
 extern void ClearForeignConstraintRelationshipGraphContext(void);
+extern HTAB * CreateOidVisitedHashSet(void);
+extern bool OidVisited(HTAB *oidVisitedMap, Oid oid);
+extern void VisitOid(HTAB *oidVisitedMap, Oid oid);
 
 #endif

src/include/distributed/metadata/distobject.h

@@ -27,5 +27,6 @@ extern bool IsObjectAddressOwnedByExtension(const ObjectAddress *target,
 											ObjectAddress *extensionAddress);
 
 extern List * GetDistributedObjectAddressList(void);
-
+extern void UpdateDistributedObjectColocationId(uint32 oldColocationId, uint32
+												newColocationId);
 #endif /* CITUS_METADATA_DISTOBJECT_H */

src/include/distributed/metadata_utility.h

@@ -36,6 +36,7 @@
 #define CSTORE_TABLE_SIZE_FUNCTION "cstore_table_size(%s)"
 
 #define SHARD_SIZES_COLUMN_COUNT 2
+#define UPDATE_SHARD_STATISTICS_COLUMN_COUNT 4
 
 /* In-memory representation of a typed tuple in pg_dist_shard. */
 typedef struct ShardInterval
@@ -206,7 +207,6 @@ extern StringInfo GenerateSizeQueryOnMultiplePlacements(List *shardIntervalList,
 extern List * RemoveCoordinatorPlacementIfNotSingleNode(List *placementList);
 extern ShardPlacement * ShardPlacementOnGroup(uint64 shardId, int groupId);
 
-
 /* Function declarations to modify shard and shard placement data */
 extern void InsertShardRow(Oid relationId, uint64 shardId, char storageType,
 						   text *shardMinValue, text *shardMaxValue);
@@ -264,5 +264,8 @@ extern ShardInterval * DeformedDistShardTupleToShardInterval(Datum *datumArray,
 															 int32 intervalTypeMod);
 extern void GetIntervalTypeInfo(char partitionMethod, Var *partitionColumn,
 								Oid *intervalTypeId, int32 *intervalTypeMod);
+extern List * SendShardStatisticsQueriesInParallel(List *citusTableIds, bool
+												   useDistributedTransaction, bool
+												   useShardMinMaxQuery);
 
 #endif   /* METADATA_UTILITY_H */

src/include/distributed/multi_partitioning_utils.h

@@ -19,7 +19,7 @@ extern bool PartitionTableNoLock(Oid relationId);
 extern bool IsChildTable(Oid relationId);
 extern bool IsParentTable(Oid relationId);
 extern Oid PartitionParentOid(Oid partitionOid);
-extern char * LongestPartitionName(Oid parentRelationId);
+extern Oid PartitionWithLongestNameRelationId(Oid parentRelationId);
 extern List * PartitionList(Oid parentRelationId);
 extern char * GenerateDetachPartitionCommand(Oid partitionTableId);
 extern char * GenerateAttachShardPartitionCommand(ShardInterval *shardInterval);

src/include/distributed/relation_restriction_equivalence.h

@@ -33,7 +33,6 @@ extern List * GenerateAllAttributeEquivalences(PlannerRestrictionContext *
 											   plannerRestrictionContext);
 extern uint32 UniqueRelationCount(RelationRestrictionContext *restrictionContext,
 								  CitusTableType tableType);
-
 extern List * DistributedRelationIdList(Query *query);
 extern PlannerRestrictionContext * FilterPlannerRestrictionForQuery(
 	PlannerRestrictionContext *plannerRestrictionContext,

src/include/distributed/remote_commands.h

@@ -21,6 +21,9 @@
 /* GUC, determining whether statements sent to remote nodes are logged */
 extern bool LogRemoteCommands;
 
+/* GUC that determines the number of bytes after which remote COPY is flushed */
+extern int RemoteCopyFlushThreshold;
+
 
 /* simple helpers */
 extern bool IsResponseOK(PGresult *result);

src/include/distributed/shard_utils.h

@@ -14,5 +14,6 @@
 #include "postgres.h"
 
 extern Oid GetTableLocalShardOid(Oid citusTableOid, uint64 shardId);
+extern char * GetLongestShardName(Oid citusTableOid, char *finalRelationName);
 
 #endif /* SHARD_UTILS_H */

src/include/distributed/transaction_management.h

@@ -111,7 +111,8 @@ extern bool TransactionModifiedNodeMetadata;
  */
 extern void UseCoordinatedTransaction(void);
 extern bool InCoordinatedTransaction(void);
-extern void CoordinatedTransactionUse2PC(void);
+extern void CoordinatedTransactionShouldUse2PC(void);
+extern bool GetCoordinatedTransactionShouldUse2PC(void);
 extern bool IsMultiStatementTransaction(void);
 extern void EnsureDistributedTransactionId(void);
 

src/test/regress/bin/normalize.sed

@@ -191,8 +191,8 @@ s/relation with OID [0-9]+ does not exist/relation with OID XXXX does not exist/
 # ignore JIT related messages
 /^DEBUG:  probing availability of JIT.*/d
 /^DEBUG:  provider not available, disabling JIT for current session.*/d
-
-
+/^DEBUG:  time to inline:.*/d
+/^DEBUG:  successfully loaded JIT.*/d
 
 # ignore timing statistics for VACUUM VERBOSE
 /CPU: user: .*s, system: .*s, elapsed: .*s/d
@@ -223,3 +223,11 @@ s/^(ERROR:  child table is missing constraint "\w+)_([0-9])+"/\1_xxxxxx"/g
         s/.*//g
     }
 }
+
+# normalize long table shard name errors for alter_table_set_access_method and alter_distributed_table
+s/^(ERROR:  child table is missing constraint "\w+)_([0-9])+"/\1_xxxxxx"/g
+s/^(DEBUG:  the name of the shard \(abcde_01234567890123456789012345678901234567890_f7ff6612)_([0-9])+/\1_xxxxxx/g
+
+# normalize long index name errors for multi_index_statements
+s/^(ERROR:  The index name \(test_index_creation1_p2020_09_26)_([0-9])+_(tenant_id_timeperiod_idx)/\1_xxxxxx_\3/g
+s/^(DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26)_([0-9])+_(tenant_id_timeperiod_idx)/\1_xxxxxx_\3/g

src/test/regress/expected/alter_distributed_table.out

@@ -53,9 +53,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering distribution column
 SELECT alter_distributed_table('dist_table', distribution_column := 'b');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -82,9 +82,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count
 SELECT alter_distributed_table('dist_table', shard_count := 6);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -111,9 +111,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering colocation, note that shard count will also change
 SELECT alter_distributed_table('dist_table', colocate_with := 'alter_distributed_table.colocation_table');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -139,13 +139,13 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count with cascading, note that the colocation will be kept
 SELECT alter_distributed_table('dist_table', shard_count := 8, cascade_to_colocated := true);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
 NOTICE:  creating a new table for alter_distributed_table.colocation_table
-NOTICE:  Moving the data of alter_distributed_table.colocation_table
-NOTICE:  Dropping the old alter_distributed_table.colocation_table
-NOTICE:  Renaming the new table to alter_distributed_table.colocation_table
+NOTICE:  moving the data of alter_distributed_table.colocation_table
+NOTICE:  dropping the old alter_distributed_table.colocation_table
+NOTICE:  renaming the new table to alter_distributed_table.colocation_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -171,9 +171,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count without cascading, note that the colocation will be broken
 SELECT alter_distributed_table('dist_table', shard_count := 10, cascade_to_colocated := false);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -261,16 +261,16 @@ SELECT * FROM partitioned_table_6_10 ORDER BY 1, 2;
 SELECT alter_distributed_table('partitioned_table', shard_count := 10, distribution_column := 'a');
 NOTICE:  converting the partitions of alter_distributed_table.partitioned_table
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table_1_5
-NOTICE:  Moving the data of alter_distributed_table.partitioned_table_1_5
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table_1_5
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table_1_5
+NOTICE:  moving the data of alter_distributed_table.partitioned_table_1_5
+NOTICE:  dropping the old alter_distributed_table.partitioned_table_1_5
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table_1_5
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table_6_10
-NOTICE:  Moving the data of alter_distributed_table.partitioned_table_6_10
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table_6_10
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table_6_10
+NOTICE:  moving the data of alter_distributed_table.partitioned_table_6_10
+NOTICE:  dropping the old alter_distributed_table.partitioned_table_6_10
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table_6_10
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table
+NOTICE:  dropping the old alter_distributed_table.partitioned_table
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -658,12 +658,12 @@ ALTER TABLE par_table ATTACH PARTITION par_table_1 FOR VALUES FROM (1) TO (5);
 SELECT alter_distributed_table('par_table', distribution_column:='b', colocate_with:='col_table');
 NOTICE:  converting the partitions of alter_distributed_table.par_table
 NOTICE:  creating a new table for alter_distributed_table.par_table_1
-NOTICE:  Moving the data of alter_distributed_table.par_table_1
-NOTICE:  Dropping the old alter_distributed_table.par_table_1
-NOTICE:  Renaming the new table to alter_distributed_table.par_table_1
+NOTICE:  moving the data of alter_distributed_table.par_table_1
+NOTICE:  dropping the old alter_distributed_table.par_table_1
+NOTICE:  renaming the new table to alter_distributed_table.par_table_1
 NOTICE:  creating a new table for alter_distributed_table.par_table
-NOTICE:  Dropping the old alter_distributed_table.par_table
-NOTICE:  Renaming the new table to alter_distributed_table.par_table
+NOTICE:  dropping the old alter_distributed_table.par_table
+NOTICE:  renaming the new table to alter_distributed_table.par_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -685,9 +685,9 @@ HINT:  check citus_tables view to see current properties of the table
 -- first colocate the tables, then try to re-colococate
 SELECT alter_distributed_table('dist_table', colocate_with := 'colocation_table');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -701,9 +701,9 @@ HINT:  check citus_tables view to see current properties of the table
 SELECT alter_distributed_table('dist_table', distribution_column:='b', shard_count:=4, cascade_to_colocated:=false);
 NOTICE:  table is already distributed by b
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -712,9 +712,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 SELECT alter_distributed_table('dist_table', shard_count:=4, colocate_with:='colocation_table_2');
 NOTICE:  shard count of the table is already 4
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -723,9 +723,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 SELECT alter_distributed_table('dist_table', colocate_with:='colocation_table_2', distribution_column:='a');
 NOTICE:  table is already colocated with colocation_table_2
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -750,9 +750,9 @@ HINT:  cascade_to_colocated := false will break the current colocation, cascade_
 -- test changing shard count of a non-colocated table without cascade_to_colocated, shouldn't error
 SELECT alter_distributed_table('dist_table', colocate_with := 'none');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -760,9 +760,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 
 SELECT alter_distributed_table('dist_table', shard_count := 14);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -820,11 +820,11 @@ INSERT INTO mat_view_test VALUES (1,1), (2,2);
 CREATE MATERIALIZED VIEW mat_view AS SELECT * FROM mat_view_test;
 SELECT alter_distributed_table('mat_view_test', shard_count := 5, cascade_to_colocated := false);
 NOTICE:  creating a new table for alter_distributed_table.mat_view_test
-NOTICE:  Moving the data of alter_distributed_table.mat_view_test
-NOTICE:  Dropping the old alter_distributed_table.mat_view_test
+NOTICE:  moving the data of alter_distributed_table.mat_view_test
+NOTICE:  dropping the old alter_distributed_table.mat_view_test
 NOTICE:  drop cascades to materialized view mat_view
 CONTEXT:  SQL statement "DROP TABLE alter_distributed_table.mat_view_test CASCADE"
-NOTICE:  Renaming the new table to alter_distributed_table.mat_view_test
+NOTICE:  renaming the new table to alter_distributed_table.mat_view_test
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -837,5 +837,85 @@ SELECT * FROM mat_view ORDER BY a;
  2 | 2
 (2 rows)
 
+-- test long table names
+SET client_min_messages TO DEBUG1;
+CREATE TABLE abcde_0123456789012345678901234567890123456789012345678901234567890123456789 (x int, y int);
+NOTICE:  identifier "abcde_0123456789012345678901234567890123456789012345678901234567890123456789" will be truncated to "abcde_012345678901234567890123456789012345678901234567890123456"
+SELECT create_distributed_table('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', 'x');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT alter_distributed_table('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', distribution_column := 'y');
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+NOTICE:  creating a new table for alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+NOTICE:  moving the data of alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  cannot perform distributed INSERT INTO ... SELECT because the partition columns in the source table and subquery do not match
+DETAIL:  The target table's partition column should correspond to a partition column in the subquery.
+CONTEXT:  SQL statement "INSERT INTO alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 (x,y) SELECT x,y FROM alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456"
+DEBUG:  performing repartitioned INSERT ... SELECT
+CONTEXT:  SQL statement "INSERT INTO alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 (x,y) SELECT x,y FROM alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456"
+NOTICE:  dropping the old alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+CONTEXT:  SQL statement "DROP TABLE alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456 CASCADE"
+NOTICE:  renaming the new table to alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+CONTEXT:  SQL statement "ALTER TABLE alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 RENAME TO abcde_012345678901234567890123456789012345678901234567890123456"
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+RESET client_min_messages;
+-- test long partitioned table names
+CREATE TABLE partition_lengths
+(
+    tenant_id integer NOT NULL,
+    timeperiod timestamp without time zone NOT NULL,
+    inserted_utc timestamp without time zone NOT NULL DEFAULT now()
+) PARTITION BY RANGE (timeperiod);
+SELECT create_distributed_table('partition_lengths', 'tenant_id');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE TABLE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890 PARTITION OF partition_lengths FOR VALUES FROM ('2020-09-28 00:00:00') TO ('2020-09-29 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+-- verify alter_distributed_table works with long partition names
+SELECT alter_distributed_table('partition_lengths', shard_count := 29, cascade_to_colocated := false);
+NOTICE:  converting the partitions of alter_distributed_table.partition_lengths
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  moving the data of alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths
+NOTICE:  dropping the old alter_distributed_table.partition_lengths
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+-- test long partition table names
+ALTER TABLE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890 RENAME TO partition_lengths_p2020_09_28;
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+ALTER TABLE partition_lengths RENAME TO partition_lengths_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "partition_lengths_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_123456789012345678901234567890123456789012345"
+-- verify alter_distributed_table works with long partitioned table names
+SELECT alter_distributed_table('partition_lengths_12345678901234567890123456789012345678901234567890', shard_count := 17, cascade_to_colocated := false);
+NOTICE:  converting the partitions of alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  moving the data of alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
 SET client_min_messages TO WARNING;
 DROP SCHEMA alter_distributed_table CASCADE;

src/test/regress/expected/alter_distributed_table_0.out

@@ -53,9 +53,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering distribution column
 SELECT alter_distributed_table('dist_table', distribution_column := 'b');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -82,9 +82,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count
 SELECT alter_distributed_table('dist_table', shard_count := 6);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -111,9 +111,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering colocation, note that shard count will also change
 SELECT alter_distributed_table('dist_table', colocate_with := 'alter_distributed_table.colocation_table');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -139,13 +139,13 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count with cascading, note that the colocation will be kept
 SELECT alter_distributed_table('dist_table', shard_count := 8, cascade_to_colocated := true);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
 NOTICE:  creating a new table for alter_distributed_table.colocation_table
-NOTICE:  Moving the data of alter_distributed_table.colocation_table
-NOTICE:  Dropping the old alter_distributed_table.colocation_table
-NOTICE:  Renaming the new table to alter_distributed_table.colocation_table
+NOTICE:  moving the data of alter_distributed_table.colocation_table
+NOTICE:  dropping the old alter_distributed_table.colocation_table
+NOTICE:  renaming the new table to alter_distributed_table.colocation_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -171,9 +171,9 @@ SELECT STRING_AGG(table_name::text, ', ' ORDER BY 1) AS "Colocation Groups" FROM
 -- test altering shard count without cascading, note that the colocation will be broken
 SELECT alter_distributed_table('dist_table', shard_count := 10, cascade_to_colocated := false);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -261,16 +261,16 @@ SELECT * FROM partitioned_table_6_10 ORDER BY 1, 2;
 SELECT alter_distributed_table('partitioned_table', shard_count := 10, distribution_column := 'a');
 NOTICE:  converting the partitions of alter_distributed_table.partitioned_table
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table_1_5
-NOTICE:  Moving the data of alter_distributed_table.partitioned_table_1_5
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table_1_5
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table_1_5
+NOTICE:  moving the data of alter_distributed_table.partitioned_table_1_5
+NOTICE:  dropping the old alter_distributed_table.partitioned_table_1_5
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table_1_5
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table_6_10
-NOTICE:  Moving the data of alter_distributed_table.partitioned_table_6_10
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table_6_10
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table_6_10
+NOTICE:  moving the data of alter_distributed_table.partitioned_table_6_10
+NOTICE:  dropping the old alter_distributed_table.partitioned_table_6_10
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table_6_10
 NOTICE:  creating a new table for alter_distributed_table.partitioned_table
-NOTICE:  Dropping the old alter_distributed_table.partitioned_table
-NOTICE:  Renaming the new table to alter_distributed_table.partitioned_table
+NOTICE:  dropping the old alter_distributed_table.partitioned_table
+NOTICE:  renaming the new table to alter_distributed_table.partitioned_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -637,12 +637,12 @@ ALTER TABLE par_table ATTACH PARTITION par_table_1 FOR VALUES FROM (1) TO (5);
 SELECT alter_distributed_table('par_table', distribution_column:='b', colocate_with:='col_table');
 NOTICE:  converting the partitions of alter_distributed_table.par_table
 NOTICE:  creating a new table for alter_distributed_table.par_table_1
-NOTICE:  Moving the data of alter_distributed_table.par_table_1
-NOTICE:  Dropping the old alter_distributed_table.par_table_1
-NOTICE:  Renaming the new table to alter_distributed_table.par_table_1
+NOTICE:  moving the data of alter_distributed_table.par_table_1
+NOTICE:  dropping the old alter_distributed_table.par_table_1
+NOTICE:  renaming the new table to alter_distributed_table.par_table_1
 NOTICE:  creating a new table for alter_distributed_table.par_table
-NOTICE:  Dropping the old alter_distributed_table.par_table
-NOTICE:  Renaming the new table to alter_distributed_table.par_table
+NOTICE:  dropping the old alter_distributed_table.par_table
+NOTICE:  renaming the new table to alter_distributed_table.par_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -664,9 +664,9 @@ HINT:  check citus_tables view to see current properties of the table
 -- first colocate the tables, then try to re-colococate
 SELECT alter_distributed_table('dist_table', colocate_with := 'colocation_table');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -680,9 +680,9 @@ HINT:  check citus_tables view to see current properties of the table
 SELECT alter_distributed_table('dist_table', distribution_column:='b', shard_count:=4, cascade_to_colocated:=false);
 NOTICE:  table is already distributed by b
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -691,9 +691,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 SELECT alter_distributed_table('dist_table', shard_count:=4, colocate_with:='colocation_table_2');
 NOTICE:  shard count of the table is already 4
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -702,9 +702,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 SELECT alter_distributed_table('dist_table', colocate_with:='colocation_table_2', distribution_column:='a');
 NOTICE:  table is already colocated with colocation_table_2
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -729,9 +729,9 @@ HINT:  cascade_to_colocated := false will break the current colocation, cascade_
 -- test changing shard count of a non-colocated table without cascade_to_colocated, shouldn't error
 SELECT alter_distributed_table('dist_table', colocate_with := 'none');
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -739,9 +739,9 @@ NOTICE:  Renaming the new table to alter_distributed_table.dist_table
 
 SELECT alter_distributed_table('dist_table', shard_count := 14);
 NOTICE:  creating a new table for alter_distributed_table.dist_table
-NOTICE:  Moving the data of alter_distributed_table.dist_table
-NOTICE:  Dropping the old alter_distributed_table.dist_table
-NOTICE:  Renaming the new table to alter_distributed_table.dist_table
+NOTICE:  moving the data of alter_distributed_table.dist_table
+NOTICE:  dropping the old alter_distributed_table.dist_table
+NOTICE:  renaming the new table to alter_distributed_table.dist_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -799,11 +799,11 @@ INSERT INTO mat_view_test VALUES (1,1), (2,2);
 CREATE MATERIALIZED VIEW mat_view AS SELECT * FROM mat_view_test;
 SELECT alter_distributed_table('mat_view_test', shard_count := 5, cascade_to_colocated := false);
 NOTICE:  creating a new table for alter_distributed_table.mat_view_test
-NOTICE:  Moving the data of alter_distributed_table.mat_view_test
-NOTICE:  Dropping the old alter_distributed_table.mat_view_test
+NOTICE:  moving the data of alter_distributed_table.mat_view_test
+NOTICE:  dropping the old alter_distributed_table.mat_view_test
 NOTICE:  drop cascades to materialized view mat_view
 CONTEXT:  SQL statement "DROP TABLE alter_distributed_table.mat_view_test CASCADE"
-NOTICE:  Renaming the new table to alter_distributed_table.mat_view_test
+NOTICE:  renaming the new table to alter_distributed_table.mat_view_test
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -816,5 +816,85 @@ SELECT * FROM mat_view ORDER BY a;
  2 | 2
 (2 rows)
 
+-- test long table names
+SET client_min_messages TO DEBUG1;
+CREATE TABLE abcde_0123456789012345678901234567890123456789012345678901234567890123456789 (x int, y int);
+NOTICE:  identifier "abcde_0123456789012345678901234567890123456789012345678901234567890123456789" will be truncated to "abcde_012345678901234567890123456789012345678901234567890123456"
+SELECT create_distributed_table('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', 'x');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT alter_distributed_table('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', distribution_column := 'y');
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+NOTICE:  creating a new table for alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+NOTICE:  moving the data of alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  cannot perform distributed INSERT INTO ... SELECT because the partition columns in the source table and subquery do not match
+DETAIL:  The target table's partition column should correspond to a partition column in the subquery.
+CONTEXT:  SQL statement "INSERT INTO alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 (x,y) SELECT x,y FROM alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456"
+DEBUG:  performing repartitioned INSERT ... SELECT
+CONTEXT:  SQL statement "INSERT INTO alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 (x,y) SELECT x,y FROM alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456"
+NOTICE:  dropping the old alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+CONTEXT:  SQL statement "DROP TABLE alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456 CASCADE"
+NOTICE:  renaming the new table to alter_distributed_table.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+CONTEXT:  SQL statement "ALTER TABLE alter_distributed_table.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 RENAME TO abcde_012345678901234567890123456789012345678901234567890123456"
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+RESET client_min_messages;
+-- test long partitioned table names
+CREATE TABLE partition_lengths
+(
+    tenant_id integer NOT NULL,
+    timeperiod timestamp without time zone NOT NULL,
+    inserted_utc timestamp without time zone NOT NULL DEFAULT now()
+) PARTITION BY RANGE (timeperiod);
+SELECT create_distributed_table('partition_lengths', 'tenant_id');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE TABLE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890 PARTITION OF partition_lengths FOR VALUES FROM ('2020-09-28 00:00:00') TO ('2020-09-29 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+-- verify alter_distributed_table works with long partition names
+SELECT alter_distributed_table('partition_lengths', shard_count := 29, cascade_to_colocated := false);
+NOTICE:  converting the partitions of alter_distributed_table.partition_lengths
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  moving the data of alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_p2020_09_28_123456789012345678901234567890123
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths
+NOTICE:  dropping the old alter_distributed_table.partition_lengths
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+-- test long partition table names
+ALTER TABLE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890 RENAME TO partition_lengths_p2020_09_28;
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+ALTER TABLE partition_lengths RENAME TO partition_lengths_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "partition_lengths_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_123456789012345678901234567890123456789012345"
+-- verify alter_distributed_table works with long partitioned table names
+SELECT alter_distributed_table('partition_lengths_12345678901234567890123456789012345678901234567890', shard_count := 17, cascade_to_colocated := false);
+NOTICE:  converting the partitions of alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  moving the data of alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_p2020_09_28
+NOTICE:  creating a new table for alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  dropping the old alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+NOTICE:  renaming the new table to alter_distributed_table.partition_lengths_123456789012345678901234567890123456789012345
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
 SET client_min_messages TO WARNING;
 DROP SCHEMA alter_distributed_table CASCADE;

src/test/regress/expected/alter_table_set_access_method.out

@@ -3,9 +3,9 @@
 CREATE TABLE alter_am_pg_version_table (a INT);
 SELECT alter_table_set_access_method('alter_am_pg_version_table', 'columnar');
 NOTICE:  creating a new table for public.alter_am_pg_version_table
-NOTICE:  Moving the data of public.alter_am_pg_version_table
-NOTICE:  Dropping the old public.alter_am_pg_version_table
-NOTICE:  Renaming the new table to public.alter_am_pg_version_table
+NOTICE:  moving the data of public.alter_am_pg_version_table
+NOTICE:  dropping the old public.alter_am_pg_version_table
+NOTICE:  renaming the new table to public.alter_am_pg_version_table
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -51,9 +51,9 @@ SELECT table_name, access_method FROM public.citus_tables WHERE table_name::text
 
 SELECT alter_table_set_access_method('dist_table', 'columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.dist_table
-NOTICE:  Moving the data of alter_table_set_access_method.dist_table
-NOTICE:  Dropping the old alter_table_set_access_method.dist_table
-NOTICE:  Renaming the new table to alter_table_set_access_method.dist_table
+NOTICE:  moving the data of alter_table_set_access_method.dist_table
+NOTICE:  dropping the old alter_table_set_access_method.dist_table
+NOTICE:  renaming the new table to alter_table_set_access_method.dist_table
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -131,9 +131,9 @@ ERROR:  you cannot alter access method of a partitioned table
 -- test altering the partition's access method
 SELECT alter_table_set_access_method('partitioned_table_1_5', 'columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.partitioned_table_1_5
-NOTICE:  Moving the data of alter_table_set_access_method.partitioned_table_1_5
-NOTICE:  Dropping the old alter_table_set_access_method.partitioned_table_1_5
-NOTICE:  Renaming the new table to alter_table_set_access_method.partitioned_table_1_5
+NOTICE:  moving the data of alter_table_set_access_method.partitioned_table_1_5
+NOTICE:  dropping the old alter_table_set_access_method.partitioned_table_1_5
+NOTICE:  renaming the new table to alter_table_set_access_method.partitioned_table_1_5
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -228,14 +228,14 @@ SELECT event FROM time_partitioned ORDER BY 1;
 CALL alter_old_partitions_set_access_method('time_partitioned', '2021-01-01', 'columnar');
 NOTICE:  converting time_partitioned_d00 with start time Sat Jan 01 00:00:00 2000 and end time Thu Dec 31 00:00:00 2009
 NOTICE:  creating a new table for alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Moving the data of alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Dropping the old alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Renaming the new table to alter_table_set_access_method.time_partitioned_d00
+NOTICE:  moving the data of alter_table_set_access_method.time_partitioned_d00
+NOTICE:  dropping the old alter_table_set_access_method.time_partitioned_d00
+NOTICE:  renaming the new table to alter_table_set_access_method.time_partitioned_d00
 NOTICE:  converting time_partitioned_d10 with start time Fri Jan 01 00:00:00 2010 and end time Tue Dec 31 00:00:00 2019
 NOTICE:  creating a new table for alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Moving the data of alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Dropping the old alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Renaming the new table to alter_table_set_access_method.time_partitioned_d10
+NOTICE:  moving the data of alter_table_set_access_method.time_partitioned_d10
+NOTICE:  dropping the old alter_table_set_access_method.time_partitioned_d10
+NOTICE:  renaming the new table to alter_table_set_access_method.time_partitioned_d10
 SELECT partition, access_method FROM time_partitions WHERE parent_table = 'time_partitioned'::regclass ORDER BY partition::text;
       partition       | access_method
 ---------------------------------------------------------------------
@@ -274,14 +274,14 @@ SELECT event FROM time_partitioned ORDER BY 1;
 CALL alter_old_partitions_set_access_method('time_partitioned', '2021-01-01', 'heap');
 NOTICE:  converting time_partitioned_d00 with start time Sat Jan 01 00:00:00 2000 and end time Thu Dec 31 00:00:00 2009
 NOTICE:  creating a new table for alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Moving the data of alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Dropping the old alter_table_set_access_method.time_partitioned_d00
-NOTICE:  Renaming the new table to alter_table_set_access_method.time_partitioned_d00
+NOTICE:  moving the data of alter_table_set_access_method.time_partitioned_d00
+NOTICE:  dropping the old alter_table_set_access_method.time_partitioned_d00
+NOTICE:  renaming the new table to alter_table_set_access_method.time_partitioned_d00
 NOTICE:  converting time_partitioned_d10 with start time Fri Jan 01 00:00:00 2010 and end time Tue Dec 31 00:00:00 2019
 NOTICE:  creating a new table for alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Moving the data of alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Dropping the old alter_table_set_access_method.time_partitioned_d10
-NOTICE:  Renaming the new table to alter_table_set_access_method.time_partitioned_d10
+NOTICE:  moving the data of alter_table_set_access_method.time_partitioned_d10
+NOTICE:  dropping the old alter_table_set_access_method.time_partitioned_d10
+NOTICE:  renaming the new table to alter_table_set_access_method.time_partitioned_d10
 SELECT partition, access_method FROM time_partitions WHERE parent_table = 'time_partitioned'::regclass ORDER BY partition::text;
       partition       | access_method
 ---------------------------------------------------------------------
@@ -324,9 +324,9 @@ SELECT alter_table_set_access_method('index_table', 'columnar');
 NOTICE:  the index idx1 on table alter_table_set_access_method.index_table will be dropped, because columnar tables cannot have indexes
 NOTICE:  the index idx2 on table alter_table_set_access_method.index_table will be dropped, because columnar tables cannot have indexes
 NOTICE:  creating a new table for alter_table_set_access_method.index_table
-NOTICE:  Moving the data of alter_table_set_access_method.index_table
-NOTICE:  Dropping the old alter_table_set_access_method.index_table
-NOTICE:  Renaming the new table to alter_table_set_access_method.index_table
+NOTICE:  moving the data of alter_table_set_access_method.index_table
+NOTICE:  dropping the old alter_table_set_access_method.index_table
+NOTICE:  renaming the new table to alter_table_set_access_method.index_table
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -395,9 +395,9 @@ NOTICE:  creating a new table for alter_table_set_access_method.table_type_dist
 WARNING:  fake_scan_getnextslot
 CONTEXT:  SQL statement "SELECT EXISTS (SELECT 1 FROM alter_table_set_access_method.table_type_dist_1533505599)"
 WARNING:  fake_scan_getnextslot
-NOTICE:  Moving the data of alter_table_set_access_method.table_type_dist
-NOTICE:  Dropping the old alter_table_set_access_method.table_type_dist
-NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_dist
+NOTICE:  moving the data of alter_table_set_access_method.table_type_dist
+NOTICE:  dropping the old alter_table_set_access_method.table_type_dist
+NOTICE:  renaming the new table to alter_table_set_access_method.table_type_dist
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -408,9 +408,9 @@ NOTICE:  creating a new table for alter_table_set_access_method.table_type_ref
 WARNING:  fake_scan_getnextslot
 CONTEXT:  SQL statement "SELECT EXISTS (SELECT 1 FROM alter_table_set_access_method.table_type_ref_1037855087)"
 WARNING:  fake_scan_getnextslot
-NOTICE:  Moving the data of alter_table_set_access_method.table_type_ref
-NOTICE:  Dropping the old alter_table_set_access_method.table_type_ref
-NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_ref
+NOTICE:  moving the data of alter_table_set_access_method.table_type_ref
+NOTICE:  dropping the old alter_table_set_access_method.table_type_ref
+NOTICE:  renaming the new table to alter_table_set_access_method.table_type_ref
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -418,9 +418,9 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_ref
 
 SELECT alter_table_set_access_method('table_type_pg_local', 'fake_am');
 NOTICE:  creating a new table for alter_table_set_access_method.table_type_pg_local
-NOTICE:  Moving the data of alter_table_set_access_method.table_type_pg_local
-NOTICE:  Dropping the old alter_table_set_access_method.table_type_pg_local
-NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_pg_local
+NOTICE:  moving the data of alter_table_set_access_method.table_type_pg_local
+NOTICE:  dropping the old alter_table_set_access_method.table_type_pg_local
+NOTICE:  renaming the new table to alter_table_set_access_method.table_type_pg_local
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -428,9 +428,9 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_pg_l
 
 SELECT alter_table_set_access_method('table_type_citus_local', 'fake_am');
 NOTICE:  creating a new table for alter_table_set_access_method.table_type_citus_local
-NOTICE:  Moving the data of alter_table_set_access_method.table_type_citus_local
-NOTICE:  Dropping the old alter_table_set_access_method.table_type_citus_local
-NOTICE:  Renaming the new table to alter_table_set_access_method.table_type_citus_local
+NOTICE:  moving the data of alter_table_set_access_method.table_type_citus_local
+NOTICE:  dropping the old alter_table_set_access_method.table_type_citus_local
+NOTICE:  renaming the new table to alter_table_set_access_method.table_type_citus_local
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -459,9 +459,9 @@ create table test_fk_p0 partition of test_fk_p for values from (0) to (10);
 create table test_fk_p1 partition of test_fk_p for values from (10) to (20);
 select alter_table_set_access_method('test_fk_p1', 'columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.test_fk_p1
-NOTICE:  Moving the data of alter_table_set_access_method.test_fk_p1
-NOTICE:  Dropping the old alter_table_set_access_method.test_fk_p1
-NOTICE:  Renaming the new table to alter_table_set_access_method.test_fk_p1
+NOTICE:  moving the data of alter_table_set_access_method.test_fk_p1
+NOTICE:  dropping the old alter_table_set_access_method.test_fk_p1
+NOTICE:  renaming the new table to alter_table_set_access_method.test_fk_p1
 ERROR:  Foreign keys and AFTER ROW triggers are not supported for columnar tables
 HINT:  Consider an AFTER STATEMENT trigger instead.
 CONTEXT:  SQL statement "ALTER TABLE alter_table_set_access_method.test_fk_p ATTACH PARTITION alter_table_set_access_method.test_fk_p1 FOR VALUES FROM (10) TO (20);"
@@ -475,11 +475,11 @@ INSERT INTO mat_view_test VALUES (1), (2);
 CREATE MATERIALIZED VIEW mat_view AS SELECT * FROM mat_view_test;
 SELECT alter_table_set_access_method('mat_view_test','columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.mat_view_test
-NOTICE:  Moving the data of alter_table_set_access_method.mat_view_test
-NOTICE:  Dropping the old alter_table_set_access_method.mat_view_test
+NOTICE:  moving the data of alter_table_set_access_method.mat_view_test
+NOTICE:  dropping the old alter_table_set_access_method.mat_view_test
 NOTICE:  drop cascades to materialized view mat_view
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.mat_view_test CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.mat_view_test
+NOTICE:  renaming the new table to alter_table_set_access_method.mat_view_test
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -519,13 +519,13 @@ create materialized view m_dist as select * from dist;
 create view v_dist as select * from dist;
 select alter_table_set_access_method('local','columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.local
-NOTICE:  Moving the data of alter_table_set_access_method.local
-NOTICE:  Dropping the old alter_table_set_access_method.local
+NOTICE:  moving the data of alter_table_set_access_method.local
+NOTICE:  dropping the old alter_table_set_access_method.local
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_local
 drop cascades to view v_local
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.local CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.local
+NOTICE:  renaming the new table to alter_table_set_access_method.local
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -533,13 +533,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.local
 
 select alter_table_set_access_method('ref','columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.ref
-NOTICE:  Moving the data of alter_table_set_access_method.ref
-NOTICE:  Dropping the old alter_table_set_access_method.ref
+NOTICE:  moving the data of alter_table_set_access_method.ref
+NOTICE:  dropping the old alter_table_set_access_method.ref
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_ref
 drop cascades to view v_ref
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.ref CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.ref
+NOTICE:  renaming the new table to alter_table_set_access_method.ref
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -547,13 +547,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.ref
 
 select alter_table_set_access_method('dist','columnar');
 NOTICE:  creating a new table for alter_table_set_access_method.dist
-NOTICE:  Moving the data of alter_table_set_access_method.dist
-NOTICE:  Dropping the old alter_table_set_access_method.dist
+NOTICE:  moving the data of alter_table_set_access_method.dist
+NOTICE:  dropping the old alter_table_set_access_method.dist
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_dist
 drop cascades to view v_dist
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.dist CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.dist
+NOTICE:  renaming the new table to alter_table_set_access_method.dist
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -561,13 +561,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.dist
 
 SELECT alter_distributed_table('dist', shard_count:=1, cascade_to_colocated:=false);
 NOTICE:  creating a new table for alter_table_set_access_method.dist
-NOTICE:  Moving the data of alter_table_set_access_method.dist
-NOTICE:  Dropping the old alter_table_set_access_method.dist
+NOTICE:  moving the data of alter_table_set_access_method.dist
+NOTICE:  dropping the old alter_table_set_access_method.dist
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_dist
 drop cascades to view v_dist
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.dist CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.dist
+NOTICE:  renaming the new table to alter_table_set_access_method.dist
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -575,13 +575,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.dist
 
 select alter_table_set_access_method('local','heap');
 NOTICE:  creating a new table for alter_table_set_access_method.local
-NOTICE:  Moving the data of alter_table_set_access_method.local
-NOTICE:  Dropping the old alter_table_set_access_method.local
+NOTICE:  moving the data of alter_table_set_access_method.local
+NOTICE:  dropping the old alter_table_set_access_method.local
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_local
 drop cascades to view v_local
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.local CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.local
+NOTICE:  renaming the new table to alter_table_set_access_method.local
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -589,13 +589,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.local
 
 select alter_table_set_access_method('ref','heap');
 NOTICE:  creating a new table for alter_table_set_access_method.ref
-NOTICE:  Moving the data of alter_table_set_access_method.ref
-NOTICE:  Dropping the old alter_table_set_access_method.ref
+NOTICE:  moving the data of alter_table_set_access_method.ref
+NOTICE:  dropping the old alter_table_set_access_method.ref
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_ref
 drop cascades to view v_ref
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.ref CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.ref
+NOTICE:  renaming the new table to alter_table_set_access_method.ref
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -603,13 +603,13 @@ NOTICE:  Renaming the new table to alter_table_set_access_method.ref
 
 select alter_table_set_access_method('dist','heap');
 NOTICE:  creating a new table for alter_table_set_access_method.dist
-NOTICE:  Moving the data of alter_table_set_access_method.dist
-NOTICE:  Dropping the old alter_table_set_access_method.dist
+NOTICE:  moving the data of alter_table_set_access_method.dist
+NOTICE:  dropping the old alter_table_set_access_method.dist
 NOTICE:  drop cascades to 2 other objects
 DETAIL:  drop cascades to materialized view m_dist
 drop cascades to view v_dist
 CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.dist CASCADE"
-NOTICE:  Renaming the new table to alter_table_set_access_method.dist
+NOTICE:  renaming the new table to alter_table_set_access_method.dist
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -681,6 +681,40 @@ CREATE TABLE identity_cols_test (a int, b int generated by default as identity (
 SELECT alter_table_set_access_method('identity_cols_test', 'columnar');
 ERROR:  cannot complete command because relation alter_table_set_access_method.identity_cols_test has identity column
 HINT:  Drop the identity columns and re-try the command
+-- test long table names
+SET client_min_messages TO DEBUG1;
+CREATE TABLE abcde_0123456789012345678901234567890123456789012345678901234567890123456789 (x int, y int);
+NOTICE:  identifier "abcde_0123456789012345678901234567890123456789012345678901234567890123456789" will be truncated to "abcde_012345678901234567890123456789012345678901234567890123456"
+SELECT create_distributed_table('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', 'x');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT alter_table_set_access_method('abcde_0123456789012345678901234567890123456789012345678901234567890123456789', 'columnar');
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+NOTICE:  creating a new table for alter_table_set_access_method.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  pathlist hook for columnar table am
+CONTEXT:  SQL statement "SELECT EXISTS (SELECT 1 FROM alter_table_set_access_method.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162)"
+NOTICE:  moving the data of alter_table_set_access_method.abcde_012345678901234567890123456789012345678901234567890123456
+NOTICE:  dropping the old alter_table_set_access_method.abcde_012345678901234567890123456789012345678901234567890123456
+CONTEXT:  SQL statement "DROP TABLE alter_table_set_access_method.abcde_012345678901234567890123456789012345678901234567890123456 CASCADE"
+NOTICE:  renaming the new table to alter_table_set_access_method.abcde_012345678901234567890123456789012345678901234567890123456
+DEBUG:  the name of the shard (abcde_01234567890123456789012345678901234567890_f7ff6612_xxxxxx) for relation (abcde_012345678901234567890123456789012345678901234567890123456) is too long, switching to sequential and local execution mode to prevent self deadlocks
+CONTEXT:  SQL statement "ALTER TABLE alter_table_set_access_method.abcde_0123456789012345678901234567890123456_f7ff6612_4160710162 RENAME TO abcde_012345678901234567890123456789012345678901234567890123456"
+ alter_table_set_access_method
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT * FROM abcde_0123456789012345678901234567890123456789012345678901234567890123456789;
+NOTICE:  identifier "abcde_0123456789012345678901234567890123456789012345678901234567890123456789" will be truncated to "abcde_012345678901234567890123456789012345678901234567890123456"
+DEBUG:  pathlist hook for columnar table am
+ x | y
+---------------------------------------------------------------------
+(0 rows)
+
+RESET client_min_messages;
 SET client_min_messages TO WARNING;
 DROP SCHEMA alter_table_set_access_method CASCADE;
 SELECT 1 FROM master_remove_node('localhost', :master_port);

src/test/regress/expected/citus_local_tables.out

@@ -351,11 +351,11 @@ ERROR:  Table 'citus_local_table_1' is a local table. Replicating shard of a loc
 BEGIN;
   SELECT undistribute_table('citus_local_table_1');
 NOTICE:  creating a new table for citus_local_tables_test_schema.citus_local_table_1
-NOTICE:  Moving the data of citus_local_tables_test_schema.citus_local_table_1
+NOTICE:  moving the data of citus_local_tables_test_schema.citus_local_table_1
 NOTICE:  executing the command locally: SELECT a FROM citus_local_tables_test_schema.citus_local_table_1_1504027 citus_local_table_1
-NOTICE:  Dropping the old citus_local_tables_test_schema.citus_local_table_1
+NOTICE:  dropping the old citus_local_tables_test_schema.citus_local_table_1
 NOTICE:  executing the command locally: DROP TABLE IF EXISTS citus_local_tables_test_schema.citus_local_table_1_xxxxx CASCADE
-NOTICE:  Renaming the new table to citus_local_tables_test_schema.citus_local_table_1
+NOTICE:  renaming the new table to citus_local_tables_test_schema.citus_local_table_1
  undistribute_table
 ---------------------------------------------------------------------
 

src/test/regress/expected/citus_update_table_statistics.out

@@ -0,0 +1,190 @@
+--
+-- citus_update_table_statistics.sql
+--
+-- Test citus_update_table_statistics function on both
+-- hash and append distributed tables
+-- This function updates shardlength, shardminvalue and shardmaxvalue
+--
+SET citus.next_shard_id TO 981000;
+SET citus.next_placement_id TO 982000;
+SET citus.shard_count TO 8;
+SET citus.shard_replication_factor TO 2;
+-- test with a hash-distributed table
+-- here we update only shardlength, not shardminvalue and shardmaxvalue
+CREATE TABLE test_table_statistics_hash (id int);
+SELECT create_distributed_table('test_table_statistics_hash', 'id');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+-- populate table
+INSERT INTO test_table_statistics_hash SELECT i FROM generate_series(0, 10000)i;
+-- originally shardlength (size of the shard) is zero
+SELECT
+    ds.logicalrelid::regclass::text AS tablename,
+    ds.shardid AS shardid,
+    dsp.placementid AS placementid,
+    shard_name(ds.logicalrelid, ds.shardid) AS shardname,
+    ds.shardminvalue AS shardminvalue,
+    ds.shardmaxvalue AS shardmaxvalue
+FROM pg_dist_shard ds JOIN pg_dist_shard_placement dsp USING (shardid)
+WHERE ds.logicalrelid::regclass::text in ('test_table_statistics_hash') AND dsp.shardlength = 0
+ORDER BY 2, 3;
+         tablename          | shardid | placementid |             shardname             | shardminvalue | shardmaxvalue
+---------------------------------------------------------------------
+ test_table_statistics_hash |  981000 |      982000 | test_table_statistics_hash_981000 | -2147483648   | -1610612737
+ test_table_statistics_hash |  981000 |      982001 | test_table_statistics_hash_981000 | -2147483648   | -1610612737
+ test_table_statistics_hash |  981001 |      982002 | test_table_statistics_hash_981001 | -1610612736   | -1073741825
+ test_table_statistics_hash |  981001 |      982003 | test_table_statistics_hash_981001 | -1610612736   | -1073741825
+ test_table_statistics_hash |  981002 |      982004 | test_table_statistics_hash_981002 | -1073741824   | -536870913
+ test_table_statistics_hash |  981002 |      982005 | test_table_statistics_hash_981002 | -1073741824   | -536870913
+ test_table_statistics_hash |  981003 |      982006 | test_table_statistics_hash_981003 | -536870912    | -1
+ test_table_statistics_hash |  981003 |      982007 | test_table_statistics_hash_981003 | -536870912    | -1
+ test_table_statistics_hash |  981004 |      982008 | test_table_statistics_hash_981004 | 0             | 536870911
+ test_table_statistics_hash |  981004 |      982009 | test_table_statistics_hash_981004 | 0             | 536870911
+ test_table_statistics_hash |  981005 |      982010 | test_table_statistics_hash_981005 | 536870912     | 1073741823
+ test_table_statistics_hash |  981005 |      982011 | test_table_statistics_hash_981005 | 536870912     | 1073741823
+ test_table_statistics_hash |  981006 |      982012 | test_table_statistics_hash_981006 | 1073741824    | 1610612735
+ test_table_statistics_hash |  981006 |      982013 | test_table_statistics_hash_981006 | 1073741824    | 1610612735
+ test_table_statistics_hash |  981007 |      982014 | test_table_statistics_hash_981007 | 1610612736    | 2147483647
+ test_table_statistics_hash |  981007 |      982015 | test_table_statistics_hash_981007 | 1610612736    | 2147483647
+(16 rows)
+
+-- setting this to on in order to verify that we use a distributed transaction id
+-- to run the size queries from different connections
+-- this is going to help detect deadlocks
+SET citus.log_remote_commands TO ON;
+-- setting this to sequential in order to have a deterministic order
+-- in the output of citus.log_remote_commands
+SET citus.multi_shard_modify_mode TO sequential;
+-- update table statistics and then check that shardlength has changed
+-- but shardminvalue and shardmaxvalue stay the same because this is
+-- a hash distributed table
+SELECT citus_update_table_statistics('test_table_statistics_hash');
+NOTICE:  issuing BEGIN TRANSACTION ISOLATION LEVEL READ COMMITTED;SELECT assign_distributed_transaction_id(xx, xx, 'xxxxxxx');
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing SELECT 981000 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981000') AS shard_size  UNION ALL SELECT 981001 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981001') AS shard_size  UNION ALL SELECT 981002 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981002') AS shard_size  UNION ALL SELECT 981003 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981003') AS shard_size  UNION ALL SELECT 981004 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981004') AS shard_size  UNION ALL SELECT 981005 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981005') AS shard_size  UNION ALL SELECT 981006 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981006') AS shard_size  UNION ALL SELECT 981007 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981007') AS shard_size  UNION ALL SELECT 0::bigint, NULL::text, NULL::text, 0::bigint;
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing BEGIN TRANSACTION ISOLATION LEVEL READ COMMITTED;SELECT assign_distributed_transaction_id(xx, xx, 'xxxxxxx');
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing SELECT 981000 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981000') AS shard_size  UNION ALL SELECT 981001 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981001') AS shard_size  UNION ALL SELECT 981002 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981002') AS shard_size  UNION ALL SELECT 981003 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981003') AS shard_size  UNION ALL SELECT 981004 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981004') AS shard_size  UNION ALL SELECT 981005 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981005') AS shard_size  UNION ALL SELECT 981006 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981006') AS shard_size  UNION ALL SELECT 981007 AS shard_id, NULL::text AS shard_minvalue, NULL::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_hash_981007') AS shard_size  UNION ALL SELECT 0::bigint, NULL::text, NULL::text, 0::bigint;
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+ citus_update_table_statistics
+---------------------------------------------------------------------
+
+(1 row)
+
+RESET citus.log_remote_commands;
+RESET citus.multi_shard_modify_mode;
+SELECT
+    ds.logicalrelid::regclass::text AS tablename,
+    ds.shardid AS shardid,
+    dsp.placementid AS placementid,
+    shard_name(ds.logicalrelid, ds.shardid) AS shardname,
+    ds.shardminvalue as shardminvalue,
+    ds.shardmaxvalue as shardmaxvalue
+FROM pg_dist_shard ds JOIN pg_dist_shard_placement dsp USING (shardid)
+WHERE ds.logicalrelid::regclass::text in ('test_table_statistics_hash') AND dsp.shardlength > 0
+ORDER BY 2, 3;
+         tablename          | shardid | placementid |             shardname             | shardminvalue | shardmaxvalue
+---------------------------------------------------------------------
+ test_table_statistics_hash |  981000 |      982000 | test_table_statistics_hash_981000 | -2147483648   | -1610612737
+ test_table_statistics_hash |  981000 |      982001 | test_table_statistics_hash_981000 | -2147483648   | -1610612737
+ test_table_statistics_hash |  981001 |      982002 | test_table_statistics_hash_981001 | -1610612736   | -1073741825
+ test_table_statistics_hash |  981001 |      982003 | test_table_statistics_hash_981001 | -1610612736   | -1073741825
+ test_table_statistics_hash |  981002 |      982004 | test_table_statistics_hash_981002 | -1073741824   | -536870913
+ test_table_statistics_hash |  981002 |      982005 | test_table_statistics_hash_981002 | -1073741824   | -536870913
+ test_table_statistics_hash |  981003 |      982006 | test_table_statistics_hash_981003 | -536870912    | -1
+ test_table_statistics_hash |  981003 |      982007 | test_table_statistics_hash_981003 | -536870912    | -1
+ test_table_statistics_hash |  981004 |      982008 | test_table_statistics_hash_981004 | 0             | 536870911
+ test_table_statistics_hash |  981004 |      982009 | test_table_statistics_hash_981004 | 0             | 536870911
+ test_table_statistics_hash |  981005 |      982010 | test_table_statistics_hash_981005 | 536870912     | 1073741823
+ test_table_statistics_hash |  981005 |      982011 | test_table_statistics_hash_981005 | 536870912     | 1073741823
+ test_table_statistics_hash |  981006 |      982012 | test_table_statistics_hash_981006 | 1073741824    | 1610612735
+ test_table_statistics_hash |  981006 |      982013 | test_table_statistics_hash_981006 | 1073741824    | 1610612735
+ test_table_statistics_hash |  981007 |      982014 | test_table_statistics_hash_981007 | 1610612736    | 2147483647
+ test_table_statistics_hash |  981007 |      982015 | test_table_statistics_hash_981007 | 1610612736    | 2147483647
+(16 rows)
+
+-- check with an append-distributed table
+-- here we update shardlength, shardminvalue and shardmaxvalue
+CREATE TABLE test_table_statistics_append (id int);
+SELECT create_distributed_table('test_table_statistics_append', 'id', 'append');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+COPY test_table_statistics_append FROM PROGRAM 'echo 0 && echo 1 && echo 2 && echo 3' WITH CSV;
+COPY test_table_statistics_append FROM PROGRAM 'echo 4 && echo 5 && echo 6 && echo 7' WITH CSV;
+-- originally shardminvalue and shardmaxvalue will be 0,3 and 4, 7
+SELECT
+    ds.logicalrelid::regclass::text AS tablename,
+    ds.shardid AS shardid,
+    dsp.placementid AS placementid,
+    shard_name(ds.logicalrelid, ds.shardid) AS shardname,
+    ds.shardminvalue as shardminvalue,
+    ds.shardmaxvalue as shardmaxvalue
+FROM pg_dist_shard ds JOIN pg_dist_shard_placement dsp USING (shardid)
+WHERE ds.logicalrelid::regclass::text in ('test_table_statistics_append')
+ORDER BY 2, 3;
+          tablename           | shardid | placementid |              shardname              | shardminvalue | shardmaxvalue
+---------------------------------------------------------------------
+ test_table_statistics_append |  981008 |      982016 | test_table_statistics_append_981008 | 0             | 3
+ test_table_statistics_append |  981008 |      982017 | test_table_statistics_append_981008 | 0             | 3
+ test_table_statistics_append |  981009 |      982018 | test_table_statistics_append_981009 | 4             | 7
+ test_table_statistics_append |  981009 |      982019 | test_table_statistics_append_981009 | 4             | 7
+(4 rows)
+
+-- delete some data to change shardminvalues of a shards
+DELETE FROM test_table_statistics_append WHERE id = 0 OR id = 4;
+SET citus.log_remote_commands TO ON;
+SET citus.multi_shard_modify_mode TO sequential;
+-- update table statistics and then check that shardminvalue has changed
+-- shardlength (shardsize) is still 8192 since there is very few data
+SELECT citus_update_table_statistics('test_table_statistics_append');
+NOTICE:  issuing BEGIN TRANSACTION ISOLATION LEVEL READ COMMITTED;SELECT assign_distributed_transaction_id(xx, xx, 'xxxxxxx');
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing SELECT 981008 AS shard_id, min(id)::text AS shard_minvalue, max(id)::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_append_981008') AS shard_size FROM test_table_statistics_append_981008  UNION ALL SELECT 981009 AS shard_id, min(id)::text AS shard_minvalue, max(id)::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_append_981009') AS shard_size FROM test_table_statistics_append_981009  UNION ALL SELECT 0::bigint, NULL::text, NULL::text, 0::bigint;
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing BEGIN TRANSACTION ISOLATION LEVEL READ COMMITTED;SELECT assign_distributed_transaction_id(xx, xx, 'xxxxxxx');
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing SELECT 981008 AS shard_id, min(id)::text AS shard_minvalue, max(id)::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_append_981008') AS shard_size FROM test_table_statistics_append_981008  UNION ALL SELECT 981009 AS shard_id, min(id)::text AS shard_minvalue, max(id)::text AS shard_maxvalue, pg_relation_size('public.test_table_statistics_append_981009') AS shard_size FROM test_table_statistics_append_981009  UNION ALL SELECT 0::bigint, NULL::text, NULL::text, 0::bigint;
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+ citus_update_table_statistics
+---------------------------------------------------------------------
+
+(1 row)
+
+RESET citus.log_remote_commands;
+RESET citus.multi_shard_modify_mode;
+SELECT
+    ds.logicalrelid::regclass::text AS tablename,
+    ds.shardid AS shardid,
+    dsp.placementid AS placementid,
+    shard_name(ds.logicalrelid, ds.shardid) AS shardname,
+    ds.shardminvalue as shardminvalue,
+    ds.shardmaxvalue as shardmaxvalue
+FROM pg_dist_shard ds JOIN pg_dist_shard_placement dsp USING (shardid)
+WHERE ds.logicalrelid::regclass::text in ('test_table_statistics_append')
+ORDER BY 2, 3;
+          tablename           | shardid | placementid |              shardname              | shardminvalue | shardmaxvalue
+---------------------------------------------------------------------
+ test_table_statistics_append |  981008 |      982016 | test_table_statistics_append_981008 | 1             | 3
+ test_table_statistics_append |  981008 |      982017 | test_table_statistics_append_981008 | 1             | 3
+ test_table_statistics_append |  981009 |      982018 | test_table_statistics_append_981009 | 5             | 7
+ test_table_statistics_append |  981009 |      982019 | test_table_statistics_append_981009 | 5             | 7
+(4 rows)
+
+DROP TABLE test_table_statistics_hash, test_table_statistics_append;
+ALTER SYSTEM RESET citus.shard_count;
+ALTER SYSTEM RESET citus.shard_replication_factor;

src/test/regress/expected/columnar_citus_integration.out

@@ -250,9 +250,9 @@ $cmd$);
 -- verify undistribute works
 SELECT undistribute_table('table_option');
 NOTICE:  creating a new table for columnar_citus_integration.table_option
-NOTICE:  Moving the data of columnar_citus_integration.table_option
-NOTICE:  Dropping the old columnar_citus_integration.table_option
-NOTICE:  Renaming the new table to columnar_citus_integration.table_option
+NOTICE:  moving the data of columnar_citus_integration.table_option
+NOTICE:  dropping the old columnar_citus_integration.table_option
+NOTICE:  renaming the new table to columnar_citus_integration.table_option
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -569,9 +569,9 @@ $cmd$);
 -- verify undistribute works
 SELECT undistribute_table('table_option');
 NOTICE:  creating a new table for columnar_citus_integration.table_option
-NOTICE:  Moving the data of columnar_citus_integration.table_option
-NOTICE:  Dropping the old columnar_citus_integration.table_option
-NOTICE:  Renaming the new table to columnar_citus_integration.table_option
+NOTICE:  moving the data of columnar_citus_integration.table_option
+NOTICE:  dropping the old columnar_citus_integration.table_option
+NOTICE:  renaming the new table to columnar_citus_integration.table_option
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -808,9 +808,9 @@ $cmd$);
 -- verify undistribute works
 SELECT undistribute_table('table_option_reference');
 NOTICE:  creating a new table for columnar_citus_integration.table_option_reference
-NOTICE:  Moving the data of columnar_citus_integration.table_option_reference
-NOTICE:  Dropping the old columnar_citus_integration.table_option_reference
-NOTICE:  Renaming the new table to columnar_citus_integration.table_option_reference
+NOTICE:  moving the data of columnar_citus_integration.table_option_reference
+NOTICE:  dropping the old columnar_citus_integration.table_option_reference
+NOTICE:  renaming the new table to columnar_citus_integration.table_option_reference
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -1041,9 +1041,9 @@ $cmd$);
 -- verify undistribute works
 SELECT undistribute_table('table_option_citus_local');
 NOTICE:  creating a new table for columnar_citus_integration.table_option_citus_local
-NOTICE:  Moving the data of columnar_citus_integration.table_option_citus_local
-NOTICE:  Dropping the old columnar_citus_integration.table_option_citus_local
-NOTICE:  Renaming the new table to columnar_citus_integration.table_option_citus_local
+NOTICE:  moving the data of columnar_citus_integration.table_option_citus_local
+NOTICE:  dropping the old columnar_citus_integration.table_option_citus_local
+NOTICE:  renaming the new table to columnar_citus_integration.table_option_citus_local
  undistribute_table
 ---------------------------------------------------------------------
 

src/test/regress/expected/columnar_partitioning.out

@@ -46,9 +46,9 @@ SELECT count(*), sum(i), min(i), max(i) FROM parent;
 -- set older partitions as columnar
 SELECT alter_table_set_access_method('p0','columnar');
 NOTICE:  creating a new table for public.p0
-NOTICE:  Moving the data of public.p0
-NOTICE:  Dropping the old public.p0
-NOTICE:  Renaming the new table to public.p0
+NOTICE:  moving the data of public.p0
+NOTICE:  dropping the old public.p0
+NOTICE:  renaming the new table to public.p0
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -56,9 +56,9 @@ NOTICE:  Renaming the new table to public.p0
 
 SELECT alter_table_set_access_method('p1','columnar');
 NOTICE:  creating a new table for public.p1
-NOTICE:  Moving the data of public.p1
-NOTICE:  Dropping the old public.p1
-NOTICE:  Renaming the new table to public.p1
+NOTICE:  moving the data of public.p1
+NOTICE:  dropping the old public.p1
+NOTICE:  renaming the new table to public.p1
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -66,9 +66,9 @@ NOTICE:  Renaming the new table to public.p1
 
 SELECT alter_table_set_access_method('p3','columnar');
 NOTICE:  creating a new table for public.p3
-NOTICE:  Moving the data of public.p3
-NOTICE:  Dropping the old public.p3
-NOTICE:  Renaming the new table to public.p3
+NOTICE:  moving the data of public.p3
+NOTICE:  dropping the old public.p3
+NOTICE:  renaming the new table to public.p3
  alter_table_set_access_method
 ---------------------------------------------------------------------
 

src/test/regress/expected/columnar_partitioning_1.out

@@ -46,9 +46,9 @@ SELECT count(*), sum(i), min(i), max(i) FROM parent;
 -- set older partitions as columnar
 SELECT alter_table_set_access_method('p0','columnar');
 NOTICE:  creating a new table for public.p0
-NOTICE:  Moving the data of public.p0
-NOTICE:  Dropping the old public.p0
-NOTICE:  Renaming the new table to public.p0
+NOTICE:  moving the data of public.p0
+NOTICE:  dropping the old public.p0
+NOTICE:  renaming the new table to public.p0
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -56,9 +56,9 @@ NOTICE:  Renaming the new table to public.p0
 
 SELECT alter_table_set_access_method('p1','columnar');
 NOTICE:  creating a new table for public.p1
-NOTICE:  Moving the data of public.p1
-NOTICE:  Dropping the old public.p1
-NOTICE:  Renaming the new table to public.p1
+NOTICE:  moving the data of public.p1
+NOTICE:  dropping the old public.p1
+NOTICE:  renaming the new table to public.p1
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -66,9 +66,9 @@ NOTICE:  Renaming the new table to public.p1
 
 SELECT alter_table_set_access_method('p3','columnar');
 NOTICE:  creating a new table for public.p3
-NOTICE:  Moving the data of public.p3
-NOTICE:  Dropping the old public.p3
-NOTICE:  Renaming the new table to public.p3
+NOTICE:  moving the data of public.p3
+NOTICE:  dropping the old public.p3
+NOTICE:  renaming the new table to public.p3
  alter_table_set_access_method
 ---------------------------------------------------------------------
 

src/test/regress/expected/coordinator_shouldhaveshards.out

@@ -718,9 +718,9 @@ SELECT conrelid::regclass::text AS "Referencing Table", pg_get_constraintdef(oid
 
 SELECT alter_distributed_table('adt_table', distribution_column:='b', colocate_with:='none');
 NOTICE:  creating a new table for coordinator_shouldhaveshards.adt_table
-NOTICE:  Moving the data of coordinator_shouldhaveshards.adt_table
-NOTICE:  Dropping the old coordinator_shouldhaveshards.adt_table
-NOTICE:  Renaming the new table to coordinator_shouldhaveshards.adt_table
+NOTICE:  moving the data of coordinator_shouldhaveshards.adt_table
+NOTICE:  dropping the old coordinator_shouldhaveshards.adt_table
+NOTICE:  renaming the new table to coordinator_shouldhaveshards.adt_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -895,6 +895,215 @@ NOTICE:  executing the command locally: SELECT count(*) AS count FROM (SELECT in
      0
 (1 row)
 
+-- a helper function which return true if the coordinated
+-- trannsaction uses 2PC
+CREATE OR REPLACE FUNCTION coordinated_transaction_should_use_2PC()
+RETURNS BOOL LANGUAGE C STRICT VOLATILE AS 'citus',
+$$coordinated_transaction_should_use_2PC$$;
+-- a local SELECT followed by remote SELECTs
+-- does not trigger 2PC
+BEGIN;
+  SELECT y FROM test WHERE x = 1;
+NOTICE:  executing the command locally: SELECT y FROM coordinator_shouldhaveshards.test_1503000 test WHERE (x OPERATOR(pg_catalog.=) 1)
+ y
+---------------------------------------------------------------------
+(0 rows)
+
+  WITH cte_1 AS (SELECT y FROM test WHERE x = 1 LIMIT 5) SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  WITH cte_1 as (SELECT * FROM test LIMIT 5) SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ f
+(1 row)
+
+COMMIT;
+-- remote SELECTs followed by local SELECTs
+-- does not trigger 2PC
+BEGIN;
+  SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  WITH cte_1 as (SELECT * FROM test LIMIT 5) SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  SELECT y FROM test WHERE x = 1;
+NOTICE:  executing the command locally: SELECT y FROM coordinator_shouldhaveshards.test_1503000 test WHERE (x OPERATOR(pg_catalog.=) 1)
+ y
+---------------------------------------------------------------------
+(0 rows)
+
+  WITH cte_1 AS (SELECT y FROM test WHERE x = 1 LIMIT 5) SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ f
+(1 row)
+
+COMMIT;
+-- a local SELECT followed by a remote Modify
+-- triggers 2PC
+BEGIN;
+  SELECT y FROM test WHERE x = 1;
+NOTICE:  executing the command locally: SELECT y FROM coordinator_shouldhaveshards.test_1503000 test WHERE (x OPERATOR(pg_catalog.=) 1)
+ y
+---------------------------------------------------------------------
+(0 rows)
+
+  UPDATE test SET y = y +1;
+NOTICE:  executing the command locally: UPDATE coordinator_shouldhaveshards.test_1503000 test SET y = (y OPERATOR(pg_catalog.+) 1)
+NOTICE:  executing the command locally: UPDATE coordinator_shouldhaveshards.test_1503003 test SET y = (y OPERATOR(pg_catalog.+) 1)
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+COMMIT;
+-- a local modify followed by a remote SELECT
+-- triggers 2PC
+BEGIN;
+  INSERT INTO test VALUES (1,1);
+NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.test_1503000 (x, y) VALUES (1, 1)
+  SELECT count(*) FROM test;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503003 test WHERE true
+ count
+---------------------------------------------------------------------
+     1
+(1 row)
+
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+COMMIT;
+-- a local modify followed by a remote MODIFY
+-- triggers 2PC
+BEGIN;
+  INSERT INTO test VALUES (1,1);
+NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.test_1503000 (x, y) VALUES (1, 1)
+  UPDATE test SET y = y +1;
+NOTICE:  executing the command locally: UPDATE coordinator_shouldhaveshards.test_1503000 test SET y = (y OPERATOR(pg_catalog.+) 1)
+NOTICE:  executing the command locally: UPDATE coordinator_shouldhaveshards.test_1503003 test SET y = (y OPERATOR(pg_catalog.+) 1)
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+COMMIT;
+-- a local modify followed by a remote single shard MODIFY
+-- triggers 2PC
+BEGIN;
+  INSERT INTO test VALUES (1,1);
+NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.test_1503000 (x, y) VALUES (1, 1)
+  INSERT INTO test VALUES (3,3);
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+COMMIT;
+-- a remote single shard modify followed by a local single
+-- shard MODIFY triggers 2PC
+BEGIN;
+  INSERT INTO test VALUES (3,3);
+  INSERT INTO test VALUES (1,1);
+NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.test_1503000 (x, y) VALUES (1, 1)
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+COMMIT;
+-- a remote single shard select followed by a local single
+-- shard MODIFY triggers 2PC. But, the transaction manager
+-- is smart enough to skip sending 2PC as the remote
+-- command is read only
+BEGIN;
+  SELECT count(*) FROM test WHERE x = 3;
+ count
+---------------------------------------------------------------------
+     2
+(1 row)
+
+  INSERT INTO test VALUES (1,1);
+NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.test_1503000 (x, y) VALUES (1, 1)
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+  SET LOCAL citus.log_remote_commands TO ON;
+COMMIT;
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
+-- a local single shard select followed by a remote single
+-- shard modify does not trigger 2PC
+BEGIN;
+  SELECT count(*) FROM test WHERE x = 1;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM coordinator_shouldhaveshards.test_1503000 test WHERE (x OPERATOR(pg_catalog.=) 1)
+ count
+---------------------------------------------------------------------
+     5
+(1 row)
+
+  INSERT INTO test VALUES (3,3);
+  SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ f
+(1 row)
+
+  SET LOCAL citus.log_remote_commands TO ON;
+COMMIT;
+NOTICE:  issuing COMMIT
+DETAIL:  on server postgres@localhost:xxxxx connectionId: xxxxxxx
 RESET client_min_messages;
 \set VERBOSITY terse
 DROP TABLE ref_table;
@@ -906,7 +1115,7 @@ DROP TABLE ref;
 NOTICE:  executing the command locally: DROP TABLE IF EXISTS coordinator_shouldhaveshards.ref_xxxxx CASCADE
 DROP TABLE test_append_table;
 DROP SCHEMA coordinator_shouldhaveshards CASCADE;
-NOTICE:  drop cascades to 19 other objects
+NOTICE:  drop cascades to 20 other objects
 SELECT 1 FROM master_set_node_property('localhost', :master_port, 'shouldhaveshards', false);
  ?column?
 ---------------------------------------------------------------------

src/test/regress/expected/cross_join.out

@@ -251,3 +251,48 @@ ORDER BY 1,2,3,4 LIMIT 5;
        1 | Wed Nov 22 22:51:43.132261 2017 |       4 |       0 |       3 |
 (5 rows)
 
+-- we don't support cross JOINs between distributed tables
+-- and without target list entries
+CREATE TABLE dist1(c0 int);
+CREATE TABLE dist2(c0 int);
+CREATE TABLE dist3(c0 int , c1 int);
+CREATE TABLE dist4(c0 int , c1 int);
+SELECT create_distributed_table('dist1', 'c0');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT create_distributed_table('dist2', 'c0');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT create_distributed_table('dist3', 'c1');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT create_distributed_table('dist4', 'c1');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT dist2.c0 FROM dist1, dist3, dist4, dist2 WHERE (dist3.c0) IN (dist4.c0);
+ERROR:  cannot perform distributed planning on this query
+DETAIL:  Cartesian products are currently unsupported
+SELECT 1 FROM dist3, dist4, dist2 WHERE (dist3.c0) IN (dist4.c0);
+ERROR:  cannot perform distributed planning on this query
+DETAIL:  Cartesian products are currently unsupported
+SELECT  FROM dist3, dist4, dist2 WHERE (dist3.c0) IN (dist4.c0);
+ERROR:  cannot perform distributed planning on this query
+DETAIL:  Cartesian products are currently unsupported
+SELECT dist2.c0 FROM dist3, dist4, dist2 WHERE (dist3.c0) IN (dist4.c0);
+ERROR:  cannot perform distributed planning on this query
+DETAIL:  Cartesian products are currently unsupported
+SELECT dist2.* FROM dist3, dist4, dist2 WHERE (dist3.c0) IN (dist4.c0);
+ERROR:  cannot perform distributed planning on this query
+DETAIL:  Cartesian products are currently unsupported

src/test/regress/expected/distributed_collations.out

@@ -73,9 +73,9 @@ SELECT create_distributed_table('test_collate_pushed_down_aggregate', 'a');
 
 SELECT alter_distributed_table('test_collate_pushed_down_aggregate', shard_count := 2, cascade_to_colocated:=false);
 NOTICE:  creating a new table for collation_tests.test_collate_pushed_down_aggregate
-NOTICE:  Moving the data of collation_tests.test_collate_pushed_down_aggregate
-NOTICE:  Dropping the old collation_tests.test_collate_pushed_down_aggregate
-NOTICE:  Renaming the new table to collation_tests.test_collate_pushed_down_aggregate
+NOTICE:  moving the data of collation_tests.test_collate_pushed_down_aggregate
+NOTICE:  dropping the old collation_tests.test_collate_pushed_down_aggregate
+NOTICE:  renaming the new table to collation_tests.test_collate_pushed_down_aggregate
  alter_distributed_table
 ---------------------------------------------------------------------
 

src/test/regress/expected/multi_drop_extension.out

@@ -49,6 +49,32 @@ BEGIN;
   drop schema public cascade;
 ROLLBACK;
 CREATE EXTENSION citus;
+CREATE SCHEMA test_schema;
+SET search_path TO test_schema;
+CREATE TABLE ref(x int, y int);
+SELECT create_reference_table('ref');
+ create_reference_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE INDEX CONCURRENTLY ref_concurrent_idx_x ON ref(x);
+CREATE INDEX CONCURRENTLY ref_concurrent_idx_y ON ref(x);
+SELECT substring(current_Setting('server_version'), '\d+')::int > 11 AS server_version_above_eleven
+\gset
+\if :server_version_above_eleven
+REINDEX INDEX CONCURRENTLY ref_concurrent_idx_x;
+REINDEX INDEX CONCURRENTLY ref_concurrent_idx_y;
+REINDEX TABLE CONCURRENTLY ref;
+REINDEX SCHEMA CONCURRENTLY test_schema;
+\endif
+SET search_path TO public;
+\set VERBOSITY TERSE
+DROP SCHEMA test_schema CASCADE;
+NOTICE:  drop cascades to 2 other objects
+DROP EXTENSION citus CASCADE;
+\set VERBOSITY DEFAULT
+CREATE EXTENSION citus;
 -- this function is dropped in Citus10, added here for tests
 CREATE OR REPLACE FUNCTION pg_catalog.master_create_worker_shards(table_name text, shard_count integer,
                                                                   replication_factor integer DEFAULT 2)

src/test/regress/expected/multi_extension.out

@@ -515,12 +515,45 @@ SELECT * FROM print_extension_changes();
                                                                                  | view time_partitions
 (67 rows)
 
+-- Test downgrade to 10.0-1 from 10.0-2
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+ALTER EXTENSION citus UPDATE TO '10.0-1';
+-- Should be empty result since upgrade+downgrade should be a no-op
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Snapshot of state at 10.0-2
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Test downgrade to 10.0-2 from 10.0-3
+ALTER EXTENSION citus UPDATE TO '10.0-3';
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+-- Should be empty result since upgrade+downgrade should be a no-op
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Snapshot of state at 10.0-3
+ALTER EXTENSION citus UPDATE TO '10.0-3';
+SELECT * FROM print_extension_changes();
+ previous_object |              current_object
+---------------------------------------------------------------------
+                 | function citus_get_active_worker_nodes()
+(1 row)
+
 DROP TABLE prev_objects, extension_diff;
 -- show running version
 SHOW citus.version;
  citus.version
 ---------------------------------------------------------------------
- 10.0devel
+ 10.0.3
 (1 row)
 
 -- ensure no unexpected objects were created outside pg_catalog

src/test/regress/expected/multi_extension_0.out

@@ -511,12 +511,45 @@ SELECT * FROM print_extension_changes();
                                                                                  | view time_partitions
 (63 rows)
 
+-- Test downgrade to 10.0-1 from 10.0-2
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+ALTER EXTENSION citus UPDATE TO '10.0-1';
+-- Should be empty result since upgrade+downgrade should be a no-op
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Snapshot of state at 10.0-2
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Test downgrade to 10.0-2 from 10.0-3
+ALTER EXTENSION citus UPDATE TO '10.0-3';
+ALTER EXTENSION citus UPDATE TO '10.0-2';
+-- Should be empty result since upgrade+downgrade should be a no-op
+SELECT * FROM print_extension_changes();
+ previous_object | current_object
+---------------------------------------------------------------------
+(0 rows)
+
+-- Snapshot of state at 10.0-3
+ALTER EXTENSION citus UPDATE TO '10.0-3';
+SELECT * FROM print_extension_changes();
+ previous_object |              current_object
+---------------------------------------------------------------------
+                 | function citus_get_active_worker_nodes()
+(1 row)
+
 DROP TABLE prev_objects, extension_diff;
 -- show running version
 SHOW citus.version;
  citus.version
 ---------------------------------------------------------------------
- 10.0devel
+ 10.0.3
 (1 row)
 
 -- ensure no unexpected objects were created outside pg_catalog

src/test/regress/expected/multi_generate_ddl_commands.out

@@ -9,7 +9,7 @@ CREATE TABLE simple_table (
 	id bigint
 );
 SELECT master_get_table_ddl_events('simple_table');
-                          master_get_table_ddl_events
+                               master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.simple_table (first_name text, last_name text, id bigint)
  ALTER TABLE public.simple_table OWNER TO postgres
@@ -21,7 +21,7 @@ CREATE TABLE not_null_table (
 	id bigint not null
 );
 SELECT master_get_table_ddl_events('not_null_table');
-                    master_get_table_ddl_events
+                          master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.not_null_table (city text, id bigint NOT NULL)
  ALTER TABLE public.not_null_table OWNER TO postgres
@@ -34,7 +34,7 @@ CREATE TABLE column_constraint_table (
 	age int CONSTRAINT non_negative_age CHECK (age >= 0)
 );
 SELECT master_get_table_ddl_events('column_constraint_table');
-                                                        master_get_table_ddl_events
+                                                              master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.column_constraint_table (first_name text, last_name text, age integer, CONSTRAINT non_negative_age CHECK ((age >= 0)))
  ALTER TABLE public.column_constraint_table OWNER TO postgres
@@ -48,7 +48,7 @@ CREATE TABLE table_constraint_table (
 	CONSTRAINT bids_ordered CHECK (min_bid > max_bid)
 );
 SELECT master_get_table_ddl_events('table_constraint_table');
-                                                                       master_get_table_ddl_events
+                                                                             master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.table_constraint_table (bid_item_id bigint, min_bid numeric NOT NULL, max_bid numeric NOT NULL, CONSTRAINT bids_ordered CHECK ((min_bid > max_bid)))
  ALTER TABLE public.table_constraint_table OWNER TO postgres
@@ -67,7 +67,7 @@ SELECT create_distributed_table('check_constraint_table_1', 'id');
 (1 row)
 
 SELECT master_get_table_ddl_events('check_constraint_table_1');
-                                                 master_get_table_ddl_events
+                                                      master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.check_constraint_table_1 (id integer, b boolean, CONSTRAINT check_constraint_table_1_b_check CHECK (b))
  ALTER TABLE public.check_constraint_table_1 OWNER TO postgres
@@ -84,7 +84,7 @@ SELECT create_distributed_table('check_constraint_table_2', 'id');
 (1 row)
 
 SELECT master_get_table_ddl_events('check_constraint_table_2');
-                                            master_get_table_ddl_events
+                                                 master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.check_constraint_table_2 (id integer, CONSTRAINT check_constraint_table_2_check CHECK (true))
  ALTER TABLE public.check_constraint_table_2 OWNER TO postgres
@@ -96,7 +96,7 @@ CREATE TABLE default_value_table (
 	price decimal default 0.00
 );
 SELECT master_get_table_ddl_events('default_value_table');
-                           master_get_table_ddl_events
+                                master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.default_value_table (name text, price numeric DEFAULT 0.00)
  ALTER TABLE public.default_value_table OWNER TO postgres
@@ -109,7 +109,7 @@ CREATE TABLE pkey_table (
 	id bigint PRIMARY KEY
 );
 SELECT master_get_table_ddl_events('pkey_table');
-                             master_get_table_ddl_events
+                                   master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.pkey_table (first_name text, last_name text, id bigint NOT NULL)
  ALTER TABLE public.pkey_table OWNER TO postgres
@@ -122,7 +122,7 @@ CREATE TABLE unique_table (
 	username text UNIQUE not null
 );
 SELECT master_get_table_ddl_events('unique_table');
-                                master_get_table_ddl_events
+                                  master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.unique_table (user_id bigint NOT NULL, username text NOT NULL)
  ALTER TABLE public.unique_table OWNER TO postgres
@@ -137,7 +137,7 @@ CREATE TABLE clustered_table (
 CREATE INDEX clustered_time_idx ON clustered_table (received_at);
 CLUSTER clustered_table USING clustered_time_idx;
 SELECT master_get_table_ddl_events('clustered_table');
-                                        master_get_table_ddl_events
+                                              master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE TABLE public.clustered_table (data json NOT NULL, received_at timestamp without time zone NOT NULL)
  ALTER TABLE public.clustered_table OWNER TO postgres
@@ -178,30 +178,33 @@ NOTICE:  foreign-data wrapper "fake_fdw" does not have an extension defined
 
 (1 row)
 
-ALTER FOREIGN TABLE foreign_table rename to renamed_foreign_table;
-ALTER FOREIGN TABLE renamed_foreign_table rename full_name to rename_name;
-ALTER FOREIGN TABLE renamed_foreign_table alter rename_name type char(8);
+ALTER FOREIGN TABLE foreign_table rename to renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890" will be truncated to "renamed_foreign_table_with_long_name_12345678901234567890123456"
+ALTER FOREIGN TABLE renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890 rename full_name to rename_name;
+NOTICE:  identifier "renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890" will be truncated to "renamed_foreign_table_with_long_name_12345678901234567890123456"
+ALTER FOREIGN TABLE renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890 alter rename_name type char(8);
+NOTICE:  identifier "renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890" will be truncated to "renamed_foreign_table_with_long_name_12345678901234567890123456"
 \c - - :public_worker_1_host :worker_1_port
 select table_name, column_name, data_type
 from information_schema.columns
 where table_schema='public' and table_name like 'renamed_foreign_table_%' and column_name <> 'id'
 order by table_name;
-          table_name          | column_name | data_type
+                           table_name                            | column_name | data_type
 ---------------------------------------------------------------------
- renamed_foreign_table_610008 | rename_name | character
- renamed_foreign_table_610009 | rename_name | character
- renamed_foreign_table_610010 | rename_name | character
- renamed_foreign_table_610011 | rename_name | character
+ renamed_foreign_table_with_long_name_1234567890_6a8dd6f8_610008 | rename_name | character
+ renamed_foreign_table_with_long_name_1234567890_6a8dd6f8_610009 | rename_name | character
+ renamed_foreign_table_with_long_name_1234567890_6a8dd6f8_610010 | rename_name | character
+ renamed_foreign_table_with_long_name_1234567890_6a8dd6f8_610011 | rename_name | character
 (4 rows)
 
 \c - - :master_host :master_port
-SELECT master_get_table_ddl_events('renamed_foreign_table');
+SELECT master_get_table_ddl_events('renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890');
 NOTICE:  foreign-data wrapper "fake_fdw" does not have an extension defined
-                                                                                   master_get_table_ddl_events
+                                                                                                        master_get_table_ddl_events
 ---------------------------------------------------------------------
  CREATE SERVER IF NOT EXISTS fake_fdw_server FOREIGN DATA WRAPPER fake_fdw
- CREATE FOREIGN TABLE public.renamed_foreign_table (id bigint NOT NULL, rename_name character(8) DEFAULT ''::text NOT NULL) SERVER fake_fdw_server OPTIONS (encoding 'utf-8', compression 'true')
- ALTER TABLE public.renamed_foreign_table OWNER TO postgres
+ CREATE FOREIGN TABLE public.renamed_foreign_table_with_long_name_12345678901234567890123456 (id bigint NOT NULL, rename_name character(8) DEFAULT ''::text NOT NULL) SERVER fake_fdw_server OPTIONS (encoding 'utf-8', compression 'true')
+ ALTER TABLE public.renamed_foreign_table_with_long_name_12345678901234567890123456 OWNER TO postgres
 (3 rows)
 
 -- propagating views is not supported
@@ -210,7 +213,8 @@ SELECT master_get_table_ddl_events('local_view');
 ERROR:  local_view is not a regular, foreign or partitioned table
 -- clean up
 DROP VIEW IF EXISTS local_view;
-DROP FOREIGN TABLE IF EXISTS renamed_foreign_table;
+DROP FOREIGN TABLE IF EXISTS renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "renamed_foreign_table_with_long_name_12345678901234567890123456789012345678901234567890" will be truncated to "renamed_foreign_table_with_long_name_12345678901234567890123456"
 \c - - :public_worker_1_host :worker_1_port
 select table_name, column_name, data_type
 from information_schema.columns

src/test/regress/expected/multi_index_statements.out

@@ -357,13 +357,13 @@ SET client_min_messages TO DEBUG1;
 CREATE INDEX ix_test_index_creation2
     ON test_index_creation1 USING btree
     (tenant_id, timeperiod);
-DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_10311_tenant_id_timeperiod_idx
+DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_xxxxxx_tenant_id_timeperiod_idx
 -- same test with schema qualified
 SET search_path TO public;
 CREATE INDEX ix_test_index_creation3
     ON multi_index_statements.test_index_creation1 USING btree
     (tenant_id, timeperiod);
-DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_10311_tenant_id_timeperiod_idx
+DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_xxxxxx_tenant_id_timeperiod_idx
 SET search_path TO multi_index_statements;
 -- we cannot switch to sequential execution
 -- after a parallel query
@@ -377,7 +377,7 @@ BEGIN;
 	CREATE INDEX ix_test_index_creation4
 	ON test_index_creation1 USING btree
 	(tenant_id, timeperiod);
-ERROR:  The index name (test_index_creation1_p2020_09_26_10311_tenant_id_timeperiod_idx) on a shard is too long and could lead to deadlocks when executed in a transaction block after a parallel query
+ERROR:  The index name (test_index_creation1_p2020_09_26_xxxxxx_tenant_id_timeperiod_idx) on a shard is too long and could lead to deadlocks when executed in a transaction block after a parallel query
 HINT:  Try re-running the transaction with "SET LOCAL citus.multi_shard_modify_mode TO 'sequential';"
 ROLLBACK;
 -- try inside a sequential block
@@ -392,7 +392,7 @@ BEGIN;
 	CREATE INDEX ix_test_index_creation4
 	ON test_index_creation1 USING btree
 	(tenant_id, timeperiod);
-DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_10311_tenant_id_timeperiod_idx
+DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_xxxxxx_tenant_id_timeperiod_idx
 ROLLBACK;
 -- should be able to create indexes with INCLUDE/WHERE
 CREATE INDEX ix_test_index_creation5 ON test_index_creation1
@@ -401,7 +401,7 @@ CREATE INDEX ix_test_index_creation5 ON test_index_creation1
 DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_2_tenant_id_timeperiod_field1_idx
 CREATE UNIQUE INDEX ix_test_index_creation6 ON test_index_creation1
 	USING btree(tenant_id, timeperiod);
-DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_10311_tenant_id_timeperiod_idx
+DEBUG:  the index name on the shards of the partition is too long, switching to sequential and local execution mode to prevent self deadlocks: test_index_creation1_p2020_09_26_xxxxxx_tenant_id_timeperiod_idx
 -- should be able to create short named indexes in parallel
 -- as  the table/index name is short
 CREATE INDEX f1

src/test/regress/expected/multi_insert_select.out

@@ -3059,6 +3059,13 @@ DO UPDATE
    SET    user_id = 42
 RETURNING user_id, value_1_agg;
 ERROR:  modifying the partition value of rows is not allowed
+-- test a small citus.remote_copy_flush_threshold
+BEGIN;
+SET LOCAL citus.remote_copy_flush_threshold TO 1;
+INSERT INTO raw_events_first
+SELECT * FROM raw_events_first OFFSET 0
+ON CONFLICT DO NOTHING;
+ABORT;
 -- wrap in a transaction to improve performance
 BEGIN;
 DROP TABLE coerce_events;

src/test/regress/expected/multi_multiuser.out

@@ -256,6 +256,44 @@ SELECT lock_relation_if_exists('test', 'ACCESS SHARE');
 SELECT lock_relation_if_exists('test', 'EXCLUSIVE');
 ERROR:  permission denied for table test
 ABORT;
+-- test creating columnar tables and accessing to columnar metadata tables via unprivileged user
+-- all below 5 commands should throw no permission errors
+-- read columnar metadata table
+SELECT * FROM columnar.stripe;
+ storage_id | stripe_num | file_offset | data_length | column_count | chunk_row_count | row_count | chunk_group_count
+---------------------------------------------------------------------
+(0 rows)
+
+-- alter a columnar setting
+SET columnar.chunk_group_row_limit = 1050;
+DO $proc$
+BEGIN
+IF substring(current_Setting('server_version'), '\d+')::int >= 12 THEN
+  EXECUTE $$
+    -- create columnar table
+    CREATE TABLE columnar_table (a int) USING columnar;
+    -- alter a columnar table that is created by that unprivileged user
+    SELECT alter_columnar_table_set('columnar_table', chunk_group_row_limit => 100);
+    -- and drop it
+    DROP TABLE columnar_table;
+  $$;
+END IF;
+END$proc$;
+-- cannot modify columnar metadata table as unprivileged user
+INSERT INTO columnar.stripe VALUES(99);
+ERROR:  permission denied for table stripe
+-- Cannot drop columnar metadata table as unprivileged user.
+-- Privileged user also cannot drop but with a different error message.
+-- (since citus extension has a dependency to it)
+DROP TABLE columnar.chunk;
+ERROR:  must be owner of table chunk
+-- test whether a read-only user can read from citus_tables view
+SELECT distribution_column FROM citus_tables WHERE table_name = 'test'::regclass;
+ distribution_column
+---------------------------------------------------------------------
+ id
+(1 row)
+
 -- check no permission
 SET ROLE no_access;
 EXECUTE prepare_insert(1);

src/test/regress/expected/multi_mx_alter_distributed_table.out

@@ -1,6 +1,8 @@
 CREATE SCHEMA mx_alter_distributed_table;
 SET search_path TO mx_alter_distributed_table;
 SET citus.shard_replication_factor TO 1;
+ALTER SEQUENCE pg_catalog.pg_dist_colocationid_seq RESTART 1410000;
+SET citus.replication_model TO 'streaming';
 -- test alter_distributed_table UDF
 CREATE TABLE adt_table (a INT, b INT);
 CREATE TABLE adt_col (a INT UNIQUE, b INT);
@@ -80,9 +82,9 @@ SELECT conrelid::regclass::text AS "Referencing Table", pg_get_constraintdef(oid
 
 SELECT alter_distributed_table('adt_table', distribution_column:='b', colocate_with:='none');
 NOTICE:  creating a new table for mx_alter_distributed_table.adt_table
-NOTICE:  Moving the data of mx_alter_distributed_table.adt_table
-NOTICE:  Dropping the old mx_alter_distributed_table.adt_table
-NOTICE:  Renaming the new table to mx_alter_distributed_table.adt_table
+NOTICE:  moving the data of mx_alter_distributed_table.adt_table
+NOTICE:  dropping the old mx_alter_distributed_table.adt_table
+NOTICE:  renaming the new table to mx_alter_distributed_table.adt_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -138,9 +140,9 @@ BEGIN;
 INSERT INTO adt_table SELECT x, x+1 FROM generate_series(1, 1000) x;
 SELECT alter_distributed_table('adt_table', distribution_column:='a');
 NOTICE:  creating a new table for mx_alter_distributed_table.adt_table
-NOTICE:  Moving the data of mx_alter_distributed_table.adt_table
-NOTICE:  Dropping the old mx_alter_distributed_table.adt_table
-NOTICE:  Renaming the new table to mx_alter_distributed_table.adt_table
+NOTICE:  moving the data of mx_alter_distributed_table.adt_table
+NOTICE:  dropping the old mx_alter_distributed_table.adt_table
+NOTICE:  renaming the new table to mx_alter_distributed_table.adt_table
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -159,5 +161,317 @@ SELECT table_name, citus_table_type, distribution_column, shard_count FROM publi
  adt_table  | distributed      | a                   |           6
 (1 row)
 
+-- test procedure colocation is preserved with alter_distributed_table
+CREATE TABLE test_proc_colocation_0 (a float8);
+SELECT create_distributed_table('test_proc_colocation_0', 'a');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE OR REPLACE procedure proc_0(dist_key float8)
+LANGUAGE plpgsql
+AS $$
+DECLARE
+    res INT := 0;
+BEGIN
+    INSERT INTO test_proc_colocation_0 VALUES (dist_key);
+    SELECT count(*) INTO res FROM test_proc_colocation_0;
+    RAISE NOTICE 'Res: %', res;
+    COMMIT;
+END;$$;
+SELECT create_distributed_function('proc_0(float8)', 'dist_key', 'test_proc_colocation_0' );
+ create_distributed_function
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410002
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0');
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410002
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 1
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+-- shardCount is not null && list_length(colocatedTableList) = 1
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 8);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 2
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410003
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0');
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410003
+(1 row)
+
+-- colocatewith is not null && list_length(colocatedTableList) = 1
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 4);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE TABLE test_proc_colocation_1 (a float8);
+SELECT create_distributed_table('test_proc_colocation_1', 'a', colocate_with := 'none');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT alter_distributed_table('test_proc_colocation_0', colocate_with := 'test_proc_colocation_1');
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 3
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410004
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0');
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410004
+(1 row)
+
+-- shardCount is not null && cascade_to_colocated is true
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 8, cascade_to_colocated := true);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_1
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 4
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410003
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0');
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410003
+(1 row)
+
+-- colocatewith is not null && cascade_to_colocated is true
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 4, cascade_to_colocated := true);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_1
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE TABLE test_proc_colocation_2 (a float8);
+SELECT create_distributed_table('test_proc_colocation_2', 'a', colocate_with := 'none');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT alter_distributed_table('test_proc_colocation_0', colocate_with := 'test_proc_colocation_2', cascade_to_colocated := true);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_1
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 5
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410005
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0');
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410005
+(1 row)
+
+-- try a case with more than one procedure
+CREATE OR REPLACE procedure proc_1(dist_key float8)
+LANGUAGE plpgsql
+AS $$
+DECLARE
+    res INT := 0;
+BEGIN
+    INSERT INTO test_proc_colocation_0 VALUES (dist_key);
+    SELECT count(*) INTO res FROM test_proc_colocation_0;
+    RAISE NOTICE 'Res: %', res;
+    COMMIT;
+END;$$;
+SELECT create_distributed_function('proc_1(float8)', 'dist_key', 'test_proc_colocation_0' );
+ create_distributed_function
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410005
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0', 'proc_1') ORDER BY proname;
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410005
+ proc_1  |      1410005
+(2 rows)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 6
+DETAIL:  from localhost:xxxxx
+CALL proc_1(2.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 7
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 8, cascade_to_colocated := true);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_2
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_2
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_2
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_2
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_1
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_1
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SET client_min_messages TO DEBUG1;
+CALL proc_0(1.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 8
+DETAIL:  from localhost:xxxxx
+CALL proc_1(2.0);
+DEBUG:  pushing down the procedure
+NOTICE:  Res: 9
+DETAIL:  from localhost:xxxxx
+RESET client_min_messages;
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410003
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0', 'proc_1') ORDER BY proname;
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410003
+ proc_1  |      1410003
+(2 rows)
+
+-- case which shouldn't preserve colocation for now
+-- shardCount is not null && cascade_to_colocated is false
+SELECT alter_distributed_table('test_proc_colocation_0', shard_count:= 18, cascade_to_colocated := false);
+NOTICE:  creating a new table for mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  moving the data of mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  dropping the old mx_alter_distributed_table.test_proc_colocation_0
+NOTICE:  renaming the new table to mx_alter_distributed_table.test_proc_colocation_0
+ alter_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+SELECT logicalrelid, colocationid FROM pg_dist_partition WHERE logicalrelid::regclass::text IN ('test_proc_colocation_0');
+      logicalrelid      | colocationid
+---------------------------------------------------------------------
+ test_proc_colocation_0 |      1410006
+(1 row)
+
+SELECT proname, colocationid FROM pg_proc JOIN citus.pg_dist_object ON pg_proc.oid = citus.pg_dist_object.objid WHERE proname IN ('proc_0', 'proc_1') ORDER BY proname;
+ proname | colocationid
+---------------------------------------------------------------------
+ proc_0  |      1410003
+ proc_1  |      1410003
+(2 rows)
+
 SET client_min_messages TO WARNING;
 DROP SCHEMA mx_alter_distributed_table CASCADE;

src/test/regress/expected/multi_mx_router_planner.out

@@ -288,6 +288,31 @@ DEBUG:  query has a single distribution column value: 1
 ---------------------------------------------------------------------
 (0 rows)
 
+WITH update_article AS (
+    UPDATE articles_hash_mx SET word_count = 11 WHERE id = 1 AND word_count = 10 RETURNING *
+)
+SELECT coalesce(1,random());
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_1 for CTE update_article: UPDATE public.articles_hash_mx SET word_count = 11 WHERE ((id OPERATOR(pg_catalog.=) 1) AND (word_count OPERATOR(pg_catalog.=) 10)) RETURNING id, author_id, title, word_count
+DEBUG:  Creating router plan
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT COALESCE((1)::double precision, random()) AS "coalesce"
+DEBUG:  Creating router plan
+ coalesce
+---------------------------------------------------------------------
+        1
+(1 row)
+
+WITH update_article AS (
+    UPDATE articles_hash_mx SET word_count = 10 WHERE author_id = 1 AND id = 1 AND word_count = 11 RETURNING *
+)
+SELECT coalesce(1,random());
+DEBUG:  Creating router plan
+DEBUG:  query has a single distribution column value: 1
+ coalesce
+---------------------------------------------------------------------
+        1
+(1 row)
+
 -- recursive CTEs are supported when filtered on partition column
 INSERT INTO company_employees_mx values(1, 1, 0);
 DEBUG:  Creating router plan

src/test/regress/expected/multi_name_lengths.out

@@ -125,13 +125,112 @@ SELECT "Constraint", "Definition" FROM table_checks WHERE relid='public.name_len
 (1 row)
 
 \c - - :master_host :master_port
--- Placeholders for RENAME operations
-\set VERBOSITY TERSE
+-- Rename the table to a too-long name
+SET client_min_messages TO DEBUG1;
+SET citus.force_max_query_parallelization TO ON;
 ALTER TABLE name_lengths RENAME TO name_len_12345678901234567890123456789012345678901234567890;
-ERROR:  shard name name_len_12345678901234567890123456789012345678_fcd8ab6f_xxxxx exceeds 63 characters
+DEBUG:  the name of the shard (name_len_12345678901234567890123456789012345678_fcd8ab6f_xxxxx) for relation (name_len_12345678901234567890123456789012345678901234567890) is too long, switching to sequential and local execution mode to prevent self deadlocks
+SELECT * FROM name_len_12345678901234567890123456789012345678901234567890;
+ col1 | col2 | float_col_12345678901234567890123456789012345678901234567890 | date_col_12345678901234567890123456789012345678901234567890 | int_col_12345678901234567890123456789012345678901234567890
+---------------------------------------------------------------------
+(0 rows)
+
+ALTER TABLE name_len_12345678901234567890123456789012345678901234567890 RENAME TO name_lengths;
+SELECT * FROM name_lengths;
+ col1 | col2 | float_col_12345678901234567890123456789012345678901234567890 | date_col_12345678901234567890123456789012345678901234567890 | int_col_12345678901234567890123456789012345678901234567890
+---------------------------------------------------------------------
+(0 rows)
+
+-- Test renames on zero shard distributed tables
+CREATE TABLE append_zero_shard_table (a int);
+SELECT create_distributed_table('append_zero_shard_table', 'a', 'append');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+ALTER TABLE append_zero_shard_table rename TO append_zero_shard_table_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "append_zero_shard_table_12345678901234567890123456789012345678901234567890" will be truncated to "append_zero_shard_table_123456789012345678901234567890123456789"
+-- Verify that we do not support long renames after parallel queries are executed in transaction block
+BEGIN;
+ALTER TABLE name_lengths rename col1 to new_column_name;
+ALTER TABLE name_lengths RENAME TO name_len_12345678901234567890123456789012345678901234567890;
+ERROR:  Shard name (name_len_12345678901234567890123456789012345678_fcd8ab6f_xxxxx) for table (name_len_12345678901234567890123456789012345678901234567890) is too long and could lead to deadlocks when executed in a transaction block after a parallel query
+HINT:  Try re-running the transaction with "SET LOCAL citus.multi_shard_modify_mode TO 'sequential';"
+ROLLBACK;
+-- The same operation will work when sequential mode is set
+BEGIN;
+SET LOCAL citus.multi_shard_modify_mode TO 'sequential';
+ALTER TABLE name_lengths rename col1 to new_column_name;
+ALTER TABLE name_lengths RENAME TO name_len_12345678901234567890123456789012345678901234567890;
+DEBUG:  the name of the shard (name_len_12345678901234567890123456789012345678_fcd8ab6f_xxxxx) for relation (name_len_12345678901234567890123456789012345678901234567890) is too long, switching to sequential and local execution mode to prevent self deadlocks
+ROLLBACK;
+RESET client_min_messages;
+-- test long partitioned table renames
+SET citus.shard_replication_factor TO 1;
+CREATE TABLE partition_lengths
+(
+    tenant_id integer NOT NULL,
+    timeperiod timestamp without time zone NOT NULL
+) PARTITION BY RANGE (timeperiod);
+SELECT create_distributed_table('partition_lengths', 'tenant_id');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+CREATE TABLE partition_lengths_p2020_09_28 PARTITION OF partition_lengths FOR VALUES FROM ('2020-09-28 00:00:00') TO ('2020-09-29 00:00:00');
+-- verify that we can rename partitioned tables and partitions to too-long names
+ALTER TABLE partition_lengths RENAME TO partition_lengths_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "partition_lengths_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_123456789012345678901234567890123456789012345"
+-- verify that we can rename partitioned tables and partitions with too-long names
+ALTER TABLE partition_lengths_12345678901234567890123456789012345678901234567890 RENAME TO partition_lengths;
+NOTICE:  identifier "partition_lengths_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_123456789012345678901234567890123456789012345"
+-- Placeholders for unsupported operations
+\set VERBOSITY TERSE
+-- renaming distributed table partitions
+ALTER TABLE partition_lengths_p2020_09_28 RENAME TO partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+-- creating or attaching new partitions with long names create deadlocks
+CREATE TABLE partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890 (LIKE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890);
+NOTICE:  identifier "partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_29_123456789012345678901234567890123"
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+ALTER TABLE partition_lengths
+    ATTACH PARTITION partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890
+    FOR VALUES FROM ('2020-09-29 00:00:00') TO ('2020-09-30 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_29_123456789012345678901234567890123"
+ERROR:  canceling the transaction since it was involved in a distributed deadlock
+CREATE TABLE partition_lengths_p2020_09_30_12345678901234567890123456789012345678901234567890
+    PARTITION OF partition_lengths
+    FOR VALUES FROM ('2020-09-30 00:00:00') TO ('2020-10-01 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_30_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_30_123456789012345678901234567890123"
+ERROR:  canceling the transaction since it was involved in a distributed deadlock
+DROP TABLE partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_29_123456789012345678901234567890123"
+-- creating or attaching new partitions with long names work when using sequential shard modify mode
+BEGIN;
+SET LOCAL citus.multi_shard_modify_mode = sequential;
+CREATE TABLE partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890 (LIKE partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890);
+NOTICE:  identifier "partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_29_123456789012345678901234567890123"
+NOTICE:  identifier "partition_lengths_p2020_09_28_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_28_123456789012345678901234567890123"
+ALTER TABLE partition_lengths
+    ATTACH PARTITION partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890
+    FOR VALUES FROM ('2020-09-29 00:00:00') TO ('2020-09-30 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_29_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_29_123456789012345678901234567890123"
+CREATE TABLE partition_lengths_p2020_09_30_12345678901234567890123456789012345678901234567890
+    PARTITION OF partition_lengths
+    FOR VALUES FROM ('2020-09-30 00:00:00') TO ('2020-10-01 00:00:00');
+NOTICE:  identifier "partition_lengths_p2020_09_30_12345678901234567890123456789012345678901234567890" will be truncated to "partition_lengths_p2020_09_30_123456789012345678901234567890123"
+ROLLBACK;
+-- renaming distributed table constraints are not supported
 ALTER TABLE name_lengths RENAME CONSTRAINT unique_12345678901234567890123456789012345678901234567890 TO unique2_12345678901234567890123456789012345678901234567890;
 ERROR:  renaming constraints belonging to distributed tables is currently unsupported
+DROP TABLE partition_lengths CASCADE;
 \set VERBOSITY DEFAULT
+-- Verify that we can create indexes with very long names on zero shard tables.
+CREATE INDEX append_zero_shard_table_idx_12345678901234567890123456789012345678901234567890 ON append_zero_shard_table_12345678901234567890123456789012345678901234567890(a);
+NOTICE:  identifier "append_zero_shard_table_idx_12345678901234567890123456789012345678901234567890" will be truncated to "append_zero_shard_table_idx_12345678901234567890123456789012345"
+NOTICE:  identifier "append_zero_shard_table_12345678901234567890123456789012345678901234567890" will be truncated to "append_zero_shard_table_123456789012345678901234567890123456789"
 -- Verify that CREATE INDEX on already distributed table has proper shard names.
 CREATE INDEX tmp_idx_12345678901234567890123456789012345678901234567890 ON name_lengths(col2);
 \c - - :public_worker_1_host :worker_1_port
@@ -148,15 +247,19 @@ SELECT "relname", "Column", "Type", "Definition" FROM index_attrs WHERE
 -- by the parser/rewriter before further processing, just as in Postgres.
 CREATE INDEX tmp_idx_123456789012345678901234567890123456789012345678901234567890 ON name_lengths(col2);
 NOTICE:  identifier "tmp_idx_123456789012345678901234567890123456789012345678901234567890" will be truncated to "tmp_idx_1234567890123456789012345678901234567890123456789012345"
+-- Verify we can rename indexes with long names
+ALTER INDEX tmp_idx_123456789012345678901234567890123456789012345678901234567890 RENAME TO tmp_idx_newname_123456789012345678901234567890123456789012345678901234567890;
+NOTICE:  identifier "tmp_idx_123456789012345678901234567890123456789012345678901234567890" will be truncated to "tmp_idx_1234567890123456789012345678901234567890123456789012345"
+NOTICE:  identifier "tmp_idx_newname_123456789012345678901234567890123456789012345678901234567890" will be truncated to "tmp_idx_newname_12345678901234567890123456789012345678901234567"
 \c - - :public_worker_1_host :worker_1_port
 SELECT "relname", "Column", "Type", "Definition" FROM index_attrs WHERE
     relname LIKE 'tmp_idx_%' ORDER BY 1 DESC, 2 DESC, 3 DESC, 4 DESC;
                              relname                             | Column |  Type   | Definition
 ---------------------------------------------------------------------
+ tmp_idx_newname_1234567890123456789012345678901_c54e849b_225003 | col2   | integer | col2
+ tmp_idx_newname_1234567890123456789012345678901_c54e849b_225002 | col2   | integer | col2
  tmp_idx_123456789012345678901234567890123456789_5e470afa_225003 | col2   | integer | col2
  tmp_idx_123456789012345678901234567890123456789_5e470afa_225002 | col2   | integer | col2
- tmp_idx_123456789012345678901234567890123456789_599636aa_225003 | col2   | integer | col2
- tmp_idx_123456789012345678901234567890123456789_599636aa_225002 | col2   | integer | col2
 (4 rows)
 
 \c - - :master_host :master_port
@@ -208,14 +311,14 @@ SELECT master_create_worker_shards('sneaky_name_lengths', '2', '2');
 (1 row)
 
 \c - - :public_worker_1_host :worker_1_port
-\di public.sneaky*225006
+\di public.sneaky*225030
                                                     List of relations
  Schema |                              Name                               | Type  |  Owner   |           Table
 ---------------------------------------------------------------------
- public | sneaky_name_lengths_int_col_1234567890123456789_6402d2cd_225006 | index | postgres | sneaky_name_lengths_225006
+ public | sneaky_name_lengths_int_col_1234567890123456789_6402d2cd_225030 | index | postgres | sneaky_name_lengths_225030
 (1 row)
 
-SELECT "Constraint", "Definition" FROM table_checks WHERE relid='public.sneaky_name_lengths_225006'::regclass ORDER BY 1 DESC, 2 DESC;
+SELECT "Constraint", "Definition" FROM table_checks WHERE relid='public.sneaky_name_lengths_225030'::regclass ORDER BY 1 DESC, 2 DESC;
                         Constraint                         |                                  Definition
 ---------------------------------------------------------------------
  checky_12345678901234567890123456789012345678901234567890 | CHECK (int_col_123456789012345678901234567890123456789012345678901234 > 100)
@@ -239,11 +342,11 @@ SELECT create_distributed_table('sneaky_name_lengths', 'col1', 'hash');
 (1 row)
 
 \c - - :public_worker_1_host :worker_1_port
-\di unique*225008
+\di unique*225032
                                                     List of relations
  Schema |                              Name                               | Type  |  Owner   |           Table
 ---------------------------------------------------------------------
- public | unique_1234567890123456789012345678901234567890_a5986f27_225008 | index | postgres | sneaky_name_lengths_225008
+ public | unique_1234567890123456789012345678901234567890_a5986f27_225032 | index | postgres | sneaky_name_lengths_225032
 (1 row)
 
 \c - - :master_host :master_port
@@ -336,3 +439,4 @@ DROP TABLE multi_name_lengths.too_long_12345678901234567890123456789012345678901
 -- Clean up.
 DROP TABLE name_lengths CASCADE;
 DROP TABLE U&"elephant_!0441!043B!043E!043D!0441!043B!043E!043D!0441!043B!043E!043D!0441!043B!043E!043D!0441!043B!043E!043D!0441!043B!043E!043D" UESCAPE '!' CASCADE;
+RESET citus.force_max_query_parallelization;

src/test/regress/expected/multi_router_planner.out

@@ -507,6 +507,45 @@ DEBUG:  Creating router plan
   1 |         1 | arsenous |         10
 (1 row)
 
+WITH update_article AS (
+    UPDATE articles_hash SET word_count = 11 WHERE id = 1 AND word_count = 10 RETURNING *
+)
+SELECT coalesce(1,random());
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_1 for CTE update_article: UPDATE public.articles_hash SET word_count = 11 WHERE ((id OPERATOR(pg_catalog.=) 1) AND (word_count OPERATOR(pg_catalog.=) 10)) RETURNING id, author_id, title, word_count
+DEBUG:  Creating router plan
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT COALESCE((1)::double precision, random()) AS "coalesce"
+DEBUG:  Creating router plan
+ coalesce
+---------------------------------------------------------------------
+        1
+(1 row)
+
+WITH update_article AS (
+    UPDATE articles_hash SET word_count = 10 WHERE author_id = 1 AND id = 1 AND word_count = 11 RETURNING *
+)
+SELECT coalesce(1,random());
+DEBUG:  Creating router plan
+DEBUG:  query has a single distribution column value: 1
+ coalesce
+---------------------------------------------------------------------
+        1
+(1 row)
+
+WITH update_article AS (
+    UPDATE authors_reference SET name = '' WHERE id = 0 RETURNING *
+)
+SELECT coalesce(1,random());
+DEBUG:  cannot router plan modification of a non-distributed table
+DEBUG:  generating subplan XXX_1 for CTE update_article: UPDATE public.authors_reference SET name = ''::character varying WHERE (id OPERATOR(pg_catalog.=) 0) RETURNING name, id
+DEBUG:  Creating router plan
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT COALESCE((1)::double precision, random()) AS "coalesce"
+DEBUG:  Creating router plan
+ coalesce
+---------------------------------------------------------------------
+        1
+(1 row)
+
 WITH delete_article AS (
     DELETE FROM articles_hash WHERE id = 1 AND word_count = 10 RETURNING *
 )

src/test/regress/expected/multi_router_planner_fast_path.out

@@ -41,7 +41,7 @@ SELECT create_distributed_table('articles_hash', 'author_id');
 
 (1 row)
 
-CREATE TABLE authors_reference ( name varchar(20), id bigint );
+CREATE TABLE authors_reference (id int, name text);
 SELECT create_reference_table('authors_reference');
  create_reference_table
 ---------------------------------------------------------------------
@@ -2111,6 +2111,26 @@ DEBUG:  query has a single distribution column value: 4
      0
 (1 row)
 
+-- test INSERT using values from generate_series() and repeat() functions
+INSERT INTO authors_reference (id, name) VALUES (generate_series(1, 10), repeat('Migjeni', 3));
+DEBUG:  Creating router plan
+SELECT * FROM authors_reference ORDER BY 1, 2;
+DEBUG:  Distributed planning for a fast-path router query
+DEBUG:  Creating router plan
+ id |         name
+---------------------------------------------------------------------
+  1 | MigjeniMigjeniMigjeni
+  2 | MigjeniMigjeniMigjeni
+  3 | MigjeniMigjeniMigjeni
+  4 | MigjeniMigjeniMigjeni
+  5 | MigjeniMigjeniMigjeni
+  6 | MigjeniMigjeniMigjeni
+  7 | MigjeniMigjeniMigjeni
+  8 | MigjeniMigjeniMigjeni
+  9 | MigjeniMigjeniMigjeni
+ 10 | MigjeniMigjeniMigjeni
+(10 rows)
+
 SET client_min_messages to 'NOTICE';
 DROP FUNCTION author_articles_max_id();
 DROP FUNCTION author_articles_id_word_count();

src/test/regress/expected/multi_test_helpers.out

@@ -43,6 +43,20 @@ BEGIN
   END LOOP;
   RETURN false;
 END; $$ language plpgsql;
+-- helper function that returns true if output of given explain has "is not null" (case in-sensitive)
+CREATE OR REPLACE FUNCTION explain_has_distributed_subplan(explain_commmand text)
+RETURNS BOOLEAN AS $$
+DECLARE
+  query_plan text;
+BEGIN
+  FOR query_plan IN EXECUTE explain_commmand LOOP
+    IF query_plan ILIKE '%Distributed Subplan %_%'
+    THEN
+        RETURN true;
+    END IF;
+  END LOOP;
+  RETURN false;
+END; $$ language plpgsql;
 -- helper function to quickly run SQL on the whole cluster
 CREATE OR REPLACE FUNCTION run_command_on_coordinator_and_workers(p_sql text)
 RETURNS void LANGUAGE plpgsql AS $$

src/test/regress/expected/multi_transaction_recovery.out

@@ -331,6 +331,111 @@ SELECT count(*) FROM pg_dist_transaction;
      2
 (1 row)
 
+-- check that read-only participants skip prepare
+SET citus.shard_count TO 4;
+CREATE TABLE test_2pcskip (a int);
+SELECT create_distributed_table('test_2pcskip', 'a');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+INSERT INTO test_2pcskip SELECT i FROM generate_series(0, 5)i;
+SELECT recover_prepared_transactions();
+ recover_prepared_transactions
+---------------------------------------------------------------------
+                             0
+(1 row)
+
+-- for the following test, ensure that 6 and 7 go to different shards on different workers
+SELECT count(DISTINCT nodeport) FROM pg_dist_shard_placement WHERE shardid IN (get_shard_id_for_distribution_column('test_2pcskip', 6),get_shard_id_for_distribution_column('test_2pcskip', 7));
+ count
+---------------------------------------------------------------------
+     2
+(1 row)
+
+-- only two of the connections will perform a write (INSERT)
+SET citus.force_max_query_parallelization TO ON;
+BEGIN;
+-- these inserts use two connections
+INSERT INTO test_2pcskip VALUES (6);
+INSERT INTO test_2pcskip VALUES (7);
+-- we know this will use more than two connections
+SELECT count(*) FROM test_2pcskip;
+ count
+---------------------------------------------------------------------
+     8
+(1 row)
+
+COMMIT;
+SELECT count(*) FROM pg_dist_transaction;
+ count
+---------------------------------------------------------------------
+     2
+(1 row)
+
+SELECT recover_prepared_transactions();
+ recover_prepared_transactions
+---------------------------------------------------------------------
+                             0
+(1 row)
+
+-- only two of the connections will perform a write (INSERT)
+BEGIN;
+-- this insert uses two connections
+INSERT INTO test_2pcskip SELECT i FROM generate_series(6, 7)i;
+-- we know this will use more than two connections
+SELECT COUNT(*) FROM test_2pcskip;
+ count
+---------------------------------------------------------------------
+    10
+(1 row)
+
+COMMIT;
+SELECT count(*) FROM pg_dist_transaction;
+ count
+---------------------------------------------------------------------
+     2
+(1 row)
+
+-- check that reads from a reference table don't trigger 2PC
+-- despite repmodel being 2PC
+CREATE TABLE test_reference (b int);
+SELECT create_reference_table('test_reference');
+ create_reference_table
+---------------------------------------------------------------------
+
+(1 row)
+
+INSERT INTO test_reference VALUES(1);
+INSERT INTO test_reference VALUES(2);
+SELECT recover_prepared_transactions();
+ recover_prepared_transactions
+---------------------------------------------------------------------
+                             0
+(1 row)
+
+BEGIN;
+SELECT * FROM test_reference ORDER BY 1;
+ b
+---------------------------------------------------------------------
+ 1
+ 2
+(2 rows)
+
+COMMIT;
+SELECT count(*) FROM pg_dist_transaction;
+ count
+---------------------------------------------------------------------
+     0
+(1 row)
+
+SELECT recover_prepared_transactions();
+ recover_prepared_transactions
+---------------------------------------------------------------------
+                             0
+(1 row)
+
 -- Test whether auto-recovery runs
 ALTER SYSTEM SET citus.recover_2pc_interval TO 10;
 SELECT pg_reload_conf();
@@ -362,6 +467,8 @@ SELECT pg_reload_conf();
 DROP TABLE test_recovery_ref;
 DROP TABLE test_recovery;
 DROP TABLE test_recovery_single;
+DROP TABLE test_2pcskip;
+DROP TABLE test_reference;
 SELECT 1 FROM master_remove_node('localhost', :master_port);
  ?column?
 ---------------------------------------------------------------------

src/test/regress/expected/null_parameters.out

@@ -3,6 +3,7 @@
 CREATE SCHEMA null_parameters;
 SET search_path TO null_parameters;
 SET citus.next_shard_id TO 1680000;
+SET citus.shard_count to 32;
 CREATE TABLE text_dist_column (key text, value text);
 SELECT create_distributed_table('text_dist_column', 'key');
  create_distributed_table

src/test/regress/expected/pg12.out

@@ -472,9 +472,9 @@ SELECT * FROM generated_stored_dist ORDER BY 1,2,3;
 INSERT INTO generated_stored_dist VALUES (1, 'text_1'), (2, 'text_2');
 SELECT alter_distributed_table('generated_stored_dist', shard_count := 5, cascade_to_colocated := false);
 NOTICE:  creating a new table for test_pg12.generated_stored_dist
-NOTICE:  Moving the data of test_pg12.generated_stored_dist
-NOTICE:  Dropping the old test_pg12.generated_stored_dist
-NOTICE:  Renaming the new table to test_pg12.generated_stored_dist
+NOTICE:  moving the data of test_pg12.generated_stored_dist
+NOTICE:  dropping the old test_pg12.generated_stored_dist
+NOTICE:  renaming the new table to test_pg12.generated_stored_dist
  alter_distributed_table
 ---------------------------------------------------------------------
 
@@ -533,9 +533,9 @@ create table generated_stored_columnar_p0 partition of generated_stored_columnar
 create table generated_stored_columnar_p1 partition of generated_stored_columnar for values from (10) to (20);
 SELECT alter_table_set_access_method('generated_stored_columnar_p0', 'columnar');
 NOTICE:  creating a new table for test_pg12.generated_stored_columnar_p0
-NOTICE:  Moving the data of test_pg12.generated_stored_columnar_p0
-NOTICE:  Dropping the old test_pg12.generated_stored_columnar_p0
-NOTICE:  Renaming the new table to test_pg12.generated_stored_columnar_p0
+NOTICE:  moving the data of test_pg12.generated_stored_columnar_p0
+NOTICE:  dropping the old test_pg12.generated_stored_columnar_p0
+NOTICE:  renaming the new table to test_pg12.generated_stored_columnar_p0
  alter_table_set_access_method
 ---------------------------------------------------------------------
 
@@ -568,9 +568,9 @@ SELECT * FROM generated_stored_ref ORDER BY 1,2,3,4,5;
 BEGIN;
   SELECT undistribute_table('generated_stored_ref');
 NOTICE:  creating a new table for test_pg12.generated_stored_ref
-NOTICE:  Moving the data of test_pg12.generated_stored_ref
-NOTICE:  Dropping the old test_pg12.generated_stored_ref
-NOTICE:  Renaming the new table to test_pg12.generated_stored_ref
+NOTICE:  moving the data of test_pg12.generated_stored_ref
+NOTICE:  dropping the old test_pg12.generated_stored_ref
+NOTICE:  renaming the new table to test_pg12.generated_stored_ref
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -600,9 +600,9 @@ BEGIN;
   -- show that undistribute_table works fine
   SELECT undistribute_table('generated_stored_ref');
 NOTICE:  creating a new table for test_pg12.generated_stored_ref
-NOTICE:  Moving the data of test_pg12.generated_stored_ref
-NOTICE:  Dropping the old test_pg12.generated_stored_ref
-NOTICE:  Renaming the new table to test_pg12.generated_stored_ref
+NOTICE:  moving the data of test_pg12.generated_stored_ref
+NOTICE:  dropping the old test_pg12.generated_stored_ref
+NOTICE:  renaming the new table to test_pg12.generated_stored_ref
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -630,9 +630,9 @@ BEGIN;
   -- show that undistribute_table works fine
   SELECT undistribute_table('generated_stored_ref');
 NOTICE:  creating a new table for test_pg12.generated_stored_ref
-NOTICE:  Moving the data of test_pg12.generated_stored_ref
-NOTICE:  Dropping the old test_pg12.generated_stored_ref
-NOTICE:  Renaming the new table to test_pg12.generated_stored_ref
+NOTICE:  moving the data of test_pg12.generated_stored_ref
+NOTICE:  dropping the old test_pg12.generated_stored_ref
+NOTICE:  renaming the new table to test_pg12.generated_stored_ref
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -650,8 +650,22 @@ SELECT citus_remove_node('localhost', :master_port);
 
 (1 row)
 
+CREATE TABLE superuser_columnar_table (a int) USING columnar;
+CREATE USER read_access;
+NOTICE:  not propagating CREATE ROLE/USER commands to worker nodes
+HINT:  Connect to worker nodes directly to manually create all necessary users and roles.
+SET ROLE read_access;
+-- user shouldn't be able to execute alter_columnar_table_set
+-- or alter_columnar_table_reset for a columnar table that it
+-- doesn't own
+SELECT alter_columnar_table_set('test_pg12.superuser_columnar_table', chunk_group_row_limit => 100);
+ERROR:  permission denied for schema test_pg12
+SELECT alter_columnar_table_reset('test_pg12.superuser_columnar_table');
+ERROR:  permission denied for schema test_pg12
+RESET ROLE;
+DROP USER read_access;
 \set VERBOSITY terse
 drop schema test_pg12 cascade;
-NOTICE:  drop cascades to 15 other objects
+NOTICE:  drop cascades to 16 other objects
 \set VERBOSITY default
 SET citus.shard_replication_factor to 2;

src/test/regress/expected/pg13_with_ties.out

@@ -155,9 +155,9 @@ TRUNCATE with_ties_table_2;
 -- test INSERT SELECTs into distributed table with a different distribution column
 SELECT undistribute_table('with_ties_table_2');
 NOTICE:  creating a new table for public.with_ties_table_2
-NOTICE:  Moving the data of public.with_ties_table_2
-NOTICE:  Dropping the old public.with_ties_table_2
-NOTICE:  Renaming the new table to public.with_ties_table_2
+NOTICE:  moving the data of public.with_ties_table_2
+NOTICE:  dropping the old public.with_ties_table_2
+NOTICE:  renaming the new table to public.with_ties_table_2
  undistribute_table
 ---------------------------------------------------------------------
 

src/test/regress/expected/set_operations.out

@@ -317,6 +317,14 @@ DEBUG:  Router planner cannot handle multi-shard select queries
 
 SELECT * FROM ((SELECT x, y FROM test) UNION ALL (SELECT y, x FROM test)) u ORDER BY 1,2;
 DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_1 for subquery SELECT x, y FROM recursive_union.test
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_2 for subquery SELECT y, x FROM recursive_union.test
+DEBUG:  Creating router plan
+DEBUG:  generating subplan XXX_3 for subquery SELECT intermediate_result.x, intermediate_result.y FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(x integer, y integer) UNION ALL SELECT intermediate_result.y, intermediate_result.x FROM read_intermediate_result('XXX_2'::text, 'binary'::citus_copy_format) intermediate_result(y integer, x integer)
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT x, y FROM (SELECT intermediate_result.x, intermediate_result.y FROM read_intermediate_result('XXX_3'::text, 'binary'::citus_copy_format) intermediate_result(x integer, y integer)) u ORDER BY x, y
+DEBUG:  Creating router plan
  x | y
 ---------------------------------------------------------------------
  1 | 1
@@ -829,6 +837,57 @@ DEBUG:  Router planner cannot handle multi-shard select queries
 DEBUG:  Router planner cannot handle multi-shard select queries
 ERROR:  cannot compute aggregate (distinct)
 DETAIL:  table partitioning is unsuitable for aggregate (distinct)
+/* these are not safe to push down as the partition key index is different */
+SELECT COUNT(*) FROM ((SELECT x,y FROM test) UNION ALL (SELECT y,x FROM test)) u;
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_1 for subquery SELECT x, y FROM recursive_union.test
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  generating subplan XXX_2 for subquery SELECT y, x FROM recursive_union.test
+DEBUG:  Creating router plan
+DEBUG:  generating subplan XXX_3 for subquery SELECT intermediate_result.x, intermediate_result.y FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(x integer, y integer) UNION ALL SELECT intermediate_result.y, intermediate_result.x FROM read_intermediate_result('XXX_2'::text, 'binary'::citus_copy_format) intermediate_result(y integer, x integer)
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT count(*) AS count FROM (SELECT intermediate_result.x, intermediate_result.y FROM read_intermediate_result('XXX_3'::text, 'binary'::citus_copy_format) intermediate_result(x integer, y integer)) u
+DEBUG:  Creating router plan
+ count
+---------------------------------------------------------------------
+     4
+(1 row)
+
+/* this is safe to push down since the partition key index is the same */
+SELECT COUNT(*) FROM (SELECT x,y FROM test UNION ALL SELECT x,y FROM test) foo;
+DEBUG:  Router planner cannot handle multi-shard select queries
+ count
+---------------------------------------------------------------------
+     4
+(1 row)
+
+SELECT COUNT(*) FROM
+    ((SELECT x,y FROM test UNION ALL SELECT x,y FROM test)
+    UNION ALL
+    (SELECT x,y FROM test UNION ALL SELECT x,y FROM test)) foo;
+DEBUG:  Router planner cannot handle multi-shard select queries
+ count
+---------------------------------------------------------------------
+     8
+(1 row)
+
+SELECT COUNT(*)
+FROM
+  (SELECT user_id  AS user_id
+   FROM
+     (SELECT x  AS user_id
+      FROM test
+      UNION ALL SELECT x AS user_id
+      FROM test) AS bar
+   UNION ALL SELECT x  AS user_id
+   FROM test) AS fool LIMIT 1;
+DEBUG:  Router planner cannot handle multi-shard select queries
+DEBUG:  push down of limit count: 1
+ count
+---------------------------------------------------------------------
+     6
+(1 row)
+
 -- one of the leaves is a repartition join
 SET citus.enable_repartition_joins TO ON;
 --  repartition is recursively planned before the set operation

src/test/regress/expected/shared_connection_stats.out

@@ -807,6 +807,7 @@ SELECT pg_sleep(0.1);
 (1 row)
 
 -- cache connections to the nodes
+SET citus.force_max_query_parallelization TO ON;
 SELECT count(*) FROM test;
  count
 ---------------------------------------------------------------------
@@ -823,6 +824,28 @@ BEGIN;
 (0 rows)
 
 COMMIT;
+-- should close all connections
+SET citus.max_cached_connection_lifetime TO '0s';
+SELECT count(*) FROM test;
+ count
+---------------------------------------------------------------------
+   155
+(1 row)
+
+-- show that no connections are cached
+SELECT
+	connection_count_to_node
+FROM
+	citus_remote_connection_stats()
+WHERE
+	port IN (SELECT node_port FROM master_get_active_worker_nodes()) AND
+	database_name = 'regression'
+ORDER BY
+	hostname, port;
+ connection_count_to_node
+---------------------------------------------------------------------
+(0 rows)
+
 -- in case other tests relies on these setting, reset them
 ALTER SYSTEM RESET citus.distributed_deadlock_detection_factor;
 ALTER SYSTEM RESET citus.recover_2pc_interval;

src/test/regress/expected/single_node.out

@@ -289,9 +289,9 @@ COMMIT;
 -- to test citus local tables
 select undistribute_table('upsert_test');
 NOTICE:  creating a new table for single_node.upsert_test
-NOTICE:  Moving the data of single_node.upsert_test
-NOTICE:  Dropping the old single_node.upsert_test
-NOTICE:  Renaming the new table to single_node.upsert_test
+NOTICE:  moving the data of single_node.upsert_test
+NOTICE:  dropping the old single_node.upsert_test
+NOTICE:  renaming the new table to single_node.upsert_test
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -694,6 +694,76 @@ SELECT * FROM collections_list, collections_list_0 WHERE collections_list.key=co
  100 |             0 | 10000 | 100 |             0 | 10000
 (1 row)
 
+-- test hash distribution using INSERT with generate_series() function
+CREATE OR REPLACE FUNCTION part_hashint4_noop(value int4, seed int8)
+RETURNS int8 AS $$
+SELECT value + seed;
+$$ LANGUAGE SQL IMMUTABLE;
+CREATE OPERATOR CLASS part_test_int4_ops
+FOR TYPE int4
+USING HASH AS
+operator 1 =,
+function 2 part_hashint4_noop(int4, int8);
+CREATE TABLE hash_parted (
+	a int,
+  b int
+) PARTITION BY HASH (a part_test_int4_ops);
+CREATE TABLE hpart0 PARTITION OF hash_parted FOR VALUES WITH (modulus 4, remainder 0);
+CREATE TABLE hpart1 PARTITION OF hash_parted FOR VALUES WITH (modulus 4, remainder 1);
+CREATE TABLE hpart2 PARTITION OF hash_parted FOR VALUES WITH (modulus 4, remainder 2);
+CREATE TABLE hpart3 PARTITION OF hash_parted FOR VALUES WITH (modulus 4, remainder 3);
+SELECT create_distributed_table('hash_parted ', 'a');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+INSERT INTO hash_parted VALUES (1, generate_series(1, 10));
+SELECT * FROM hash_parted ORDER BY 1, 2;
+ a | b
+---------------------------------------------------------------------
+ 1 |  1
+ 1 |  2
+ 1 |  3
+ 1 |  4
+ 1 |  5
+ 1 |  6
+ 1 |  7
+ 1 |  8
+ 1 |  9
+ 1 | 10
+(10 rows)
+
+ALTER TABLE hash_parted DETACH PARTITION hpart0;
+ALTER TABLE hash_parted DETACH PARTITION hpart1;
+ALTER TABLE hash_parted DETACH PARTITION hpart2;
+ALTER TABLE hash_parted DETACH PARTITION hpart3;
+-- test range partition without creating partitions and inserting with generate_series()
+-- should error out even in plain PG since no partition of relation "parent_tab" is found for row
+-- in Citus it errors out because it fails to evaluate partition key in insert
+CREATE TABLE parent_tab (id int) PARTITION BY RANGE (id);
+SELECT create_distributed_table('parent_tab', 'id');
+ create_distributed_table
+---------------------------------------------------------------------
+
+(1 row)
+
+INSERT INTO parent_tab VALUES (generate_series(0, 3));
+ERROR:  failed to evaluate partition key in insert
+HINT:  try using constant values for partition column
+-- now it should work
+CREATE TABLE parent_tab_1_2 PARTITION OF parent_tab FOR VALUES FROM (1) to (2);
+ALTER TABLE parent_tab ADD COLUMN b int;
+INSERT INTO parent_tab VALUES (1, generate_series(0, 3));
+SELECT * FROM parent_tab ORDER BY 1, 2;
+ id | b
+---------------------------------------------------------------------
+  1 | 0
+  1 | 1
+  1 | 2
+  1 | 3
+(4 rows)
+
 -- make sure that parallel accesses are good
 SET citus.force_max_query_parallelization TO ON;
 SELECT * FROM test_2 ORDER BY 1 DESC;
@@ -1125,9 +1195,9 @@ RESET citus.task_executor_type;
 ALTER TABLE test DROP CONSTRAINT foreign_key;
 SELECT undistribute_table('test_2');
 NOTICE:  creating a new table for single_node.test_2
-NOTICE:  Moving the data of single_node.test_2
-NOTICE:  Dropping the old single_node.test_2
-NOTICE:  Renaming the new table to single_node.test_2
+NOTICE:  moving the data of single_node.test_2
+NOTICE:  dropping the old single_node.test_2
+NOTICE:  renaming the new table to single_node.test_2
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -1176,28 +1246,28 @@ ALTER TABLE partitioned_table_1 ADD CONSTRAINT fkey_5 FOREIGN KEY (col_1) REFERE
 SELECT undistribute_table('partitioned_table_1', cascade_via_foreign_keys=>true);
 NOTICE:  converting the partitions of single_node.partitioned_table_1
 NOTICE:  creating a new table for single_node.partitioned_table_1_100_200
-NOTICE:  Moving the data of single_node.partitioned_table_1_100_200
-NOTICE:  Dropping the old single_node.partitioned_table_1_100_200
-NOTICE:  Renaming the new table to single_node.partitioned_table_1_100_200
+NOTICE:  moving the data of single_node.partitioned_table_1_100_200
+NOTICE:  dropping the old single_node.partitioned_table_1_100_200
+NOTICE:  renaming the new table to single_node.partitioned_table_1_100_200
 NOTICE:  creating a new table for single_node.partitioned_table_1_200_300
-NOTICE:  Moving the data of single_node.partitioned_table_1_200_300
-NOTICE:  Dropping the old single_node.partitioned_table_1_200_300
-NOTICE:  Renaming the new table to single_node.partitioned_table_1_200_300
+NOTICE:  moving the data of single_node.partitioned_table_1_200_300
+NOTICE:  dropping the old single_node.partitioned_table_1_200_300
+NOTICE:  renaming the new table to single_node.partitioned_table_1_200_300
 NOTICE:  creating a new table for single_node.partitioned_table_1
-NOTICE:  Dropping the old single_node.partitioned_table_1
-NOTICE:  Renaming the new table to single_node.partitioned_table_1
+NOTICE:  dropping the old single_node.partitioned_table_1
+NOTICE:  renaming the new table to single_node.partitioned_table_1
 NOTICE:  creating a new table for single_node.reference_table_1
-NOTICE:  Moving the data of single_node.reference_table_1
-NOTICE:  Dropping the old single_node.reference_table_1
-NOTICE:  Renaming the new table to single_node.reference_table_1
+NOTICE:  moving the data of single_node.reference_table_1
+NOTICE:  dropping the old single_node.reference_table_1
+NOTICE:  renaming the new table to single_node.reference_table_1
 NOTICE:  creating a new table for single_node.distributed_table_1
-NOTICE:  Moving the data of single_node.distributed_table_1
-NOTICE:  Dropping the old single_node.distributed_table_1
-NOTICE:  Renaming the new table to single_node.distributed_table_1
+NOTICE:  moving the data of single_node.distributed_table_1
+NOTICE:  dropping the old single_node.distributed_table_1
+NOTICE:  renaming the new table to single_node.distributed_table_1
 NOTICE:  creating a new table for single_node.citus_local_table_1
-NOTICE:  Moving the data of single_node.citus_local_table_1
-NOTICE:  Dropping the old single_node.citus_local_table_1
-NOTICE:  Renaming the new table to single_node.citus_local_table_1
+NOTICE:  moving the data of single_node.citus_local_table_1
+NOTICE:  dropping the old single_node.citus_local_table_1
+NOTICE:  renaming the new table to single_node.citus_local_table_1
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -1785,6 +1855,142 @@ NOTICE:  executing the command locally: SELECT bool_and((z IS NULL)) AS bool_and
 (1 row)
 
 RESET citus.local_copy_flush_threshold;
+RESET citus.local_copy_flush_threshold;
+CREATE OR REPLACE FUNCTION coordinated_transaction_should_use_2PC()
+RETURNS BOOL LANGUAGE C STRICT VOLATILE AS 'citus',
+$$coordinated_transaction_should_use_2PC$$;
+-- a multi-shard/single-shard select that is failed over to local
+-- execution doesn't start a 2PC
+BEGIN;
+	SELECT count(*) FROM another_schema_table;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM single_node.another_schema_table_90630511 another_schema_table WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM single_node.another_schema_table_90630512 another_schema_table WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM single_node.another_schema_table_90630513 another_schema_table WHERE true
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM single_node.another_schema_table_90630514 another_schema_table WHERE true
+ count
+---------------------------------------------------------------------
+ 10001
+(1 row)
+
+	SELECT count(*) FROM another_schema_table WHERE a = 1;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM single_node.another_schema_table_90630511 another_schema_table WHERE (a OPERATOR(pg_catalog.=) 1)
+ count
+---------------------------------------------------------------------
+     1
+(1 row)
+
+	WITH cte_1 as (SELECT * FROM another_schema_table LIMIT 10)
+		SELECT count(*) FROM cte_1;
+NOTICE:  executing the command locally: SELECT a, b FROM single_node.another_schema_table_90630511 another_schema_table WHERE true LIMIT '10'::bigint
+NOTICE:  executing the command locally: SELECT a, b FROM single_node.another_schema_table_90630512 another_schema_table WHERE true LIMIT '10'::bigint
+NOTICE:  executing the command locally: SELECT a, b FROM single_node.another_schema_table_90630513 another_schema_table WHERE true LIMIT '10'::bigint
+NOTICE:  executing the command locally: SELECT a, b FROM single_node.another_schema_table_90630514 another_schema_table WHERE true LIMIT '10'::bigint
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM (SELECT intermediate_result.a, intermediate_result.b FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(a integer, b integer)) cte_1
+ count
+---------------------------------------------------------------------
+    10
+(1 row)
+
+	WITH cte_1 as (SELECT * FROM another_schema_table WHERE a = 1 LIMIT 10)
+		SELECT count(*) FROM cte_1;
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM (SELECT another_schema_table.a, another_schema_table.b FROM single_node.another_schema_table_90630511 another_schema_table WHERE (another_schema_table.a OPERATOR(pg_catalog.=) 1) LIMIT 10) cte_1
+ count
+---------------------------------------------------------------------
+     1
+(1 row)
+
+	SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ f
+(1 row)
+
+ROLLBACK;
+-- same without a transaction block
+WITH cte_1 AS (SELECT count(*) as cnt FROM another_schema_table LIMIT 1000),
+	 cte_2 AS (SELECT coordinated_transaction_should_use_2PC() as enabled_2pc)
+SELECT cnt, enabled_2pc FROM cte_1, cte_2;
+NOTICE:  executing the command locally: SELECT count(*) AS cnt FROM single_node.another_schema_table_90630511 another_schema_table WHERE true LIMIT '1000'::bigint
+NOTICE:  executing the command locally: SELECT count(*) AS cnt FROM single_node.another_schema_table_90630512 another_schema_table WHERE true LIMIT '1000'::bigint
+NOTICE:  executing the command locally: SELECT count(*) AS cnt FROM single_node.another_schema_table_90630513 another_schema_table WHERE true LIMIT '1000'::bigint
+NOTICE:  executing the command locally: SELECT count(*) AS cnt FROM single_node.another_schema_table_90630514 another_schema_table WHERE true LIMIT '1000'::bigint
+NOTICE:  executing the command locally: SELECT cte_1.cnt, cte_2.enabled_2pc FROM (SELECT intermediate_result.cnt FROM read_intermediate_result('XXX_2'::text, 'binary'::citus_copy_format) intermediate_result(cnt bigint)) cte_1, (SELECT intermediate_result.enabled_2pc FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(enabled_2pc boolean)) cte_2
+  cnt  | enabled_2pc
+---------------------------------------------------------------------
+ 10001 | f
+(1 row)
+
+-- a multi-shard modification that is failed over to local
+-- execution starts a 2PC
+BEGIN;
+	UPDATE another_schema_table SET b = b + 1;
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630511 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1)
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630512 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1)
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630513 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1)
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630514 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1)
+	SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+ROLLBACK;
+-- a multi-shard modification that is failed over to local
+-- execution starts a 2PC
+BEGIN;
+	WITH cte_1 AS (UPDATE another_schema_table SET b = b + 1 RETURNING *)
+		SELECT count(*) FROM cte_1;
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630511 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630512 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630513 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630514 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: SELECT count(*) AS count FROM (SELECT intermediate_result.a, intermediate_result.b FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(a integer, b integer)) cte_1
+ count
+---------------------------------------------------------------------
+ 10001
+(1 row)
+
+	SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+ROLLBACK;
+-- same without transaction block
+WITH cte_1 AS (UPDATE another_schema_table SET b = b + 1 RETURNING *)
+SELECT coordinated_transaction_should_use_2PC();
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630511 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630512 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630513 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630514 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) RETURNING a, b
+NOTICE:  executing the command locally: SELECT single_node.coordinated_transaction_should_use_2pc() AS coordinated_transaction_should_use_2pc
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+-- a single-shard modification that is failed over to local
+-- starts 2PC execution
+BEGIN;
+	UPDATE another_schema_table SET b = b + 1 WHERE a = 1;
+NOTICE:  executing the command locally: UPDATE single_node.another_schema_table_90630511 another_schema_table SET b = (b OPERATOR(pg_catalog.+) 1) WHERE (a OPERATOR(pg_catalog.=) 1)
+	SELECT coordinated_transaction_should_use_2PC();
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
+ROLLBACK;
+-- same without transaction block
+WITH cte_1 AS (UPDATE another_schema_table SET b = b + 1 WHERE a = 1 RETURNING *)
+SELECT coordinated_transaction_should_use_2PC() FROM cte_1;
+NOTICE:  executing the command locally: WITH cte_1 AS (UPDATE single_node.another_schema_table_90630511 another_schema_table SET b = (another_schema_table.b OPERATOR(pg_catalog.+) 1) WHERE (another_schema_table.a OPERATOR(pg_catalog.=) 1) RETURNING another_schema_table.a, another_schema_table.b) SELECT single_node.coordinated_transaction_should_use_2pc() AS coordinated_transaction_should_use_2pc FROM cte_1
+ coordinated_transaction_should_use_2pc
+---------------------------------------------------------------------
+ t
+(1 row)
+
 -- if the local execution is disabled, we cannot failover to
 -- local execution and the queries would fail
 SET citus.enable_local_execution TO  false;

src/test/regress/expected/undistribute_table.out

@@ -35,15 +35,15 @@ SELECT * FROM dist_table ORDER BY 1, 2, 3;
 -- the name->OID conversion happens at parse time.
 SELECT undistribute_table('dist_table'), create_distributed_table('dist_table', 'a');
 NOTICE:  creating a new table for undistribute_table.dist_table
-NOTICE:  Moving the data of undistribute_table.dist_table
-NOTICE:  Dropping the old undistribute_table.dist_table
-NOTICE:  Renaming the new table to undistribute_table.dist_table
+NOTICE:  moving the data of undistribute_table.dist_table
+NOTICE:  dropping the old undistribute_table.dist_table
+NOTICE:  renaming the new table to undistribute_table.dist_table
 ERROR:  relation with OID XXXX does not exist
 SELECT undistribute_table('dist_table');
 NOTICE:  creating a new table for undistribute_table.dist_table
-NOTICE:  Moving the data of undistribute_table.dist_table
-NOTICE:  Dropping the old undistribute_table.dist_table
-NOTICE:  Renaming the new table to undistribute_table.dist_table
+NOTICE:  moving the data of undistribute_table.dist_table
+NOTICE:  dropping the old undistribute_table.dist_table
+NOTICE:  renaming the new table to undistribute_table.dist_table
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -88,9 +88,9 @@ SELECT * FROM pg_indexes WHERE tablename = 'dist_table';
 
 SELECT undistribute_table('dist_table');
 NOTICE:  creating a new table for undistribute_table.dist_table
-NOTICE:  Moving the data of undistribute_table.dist_table
-NOTICE:  Dropping the old undistribute_table.dist_table
-NOTICE:  Renaming the new table to undistribute_table.dist_table
+NOTICE:  moving the data of undistribute_table.dist_table
+NOTICE:  dropping the old undistribute_table.dist_table
+NOTICE:  renaming the new table to undistribute_table.dist_table
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -204,16 +204,16 @@ HINT:  the parent table is "partitioned_table"
 SELECT undistribute_table('partitioned_table');
 NOTICE:  converting the partitions of undistribute_table.partitioned_table
 NOTICE:  creating a new table for undistribute_table.partitioned_table_1_5
-NOTICE:  Moving the data of undistribute_table.partitioned_table_1_5
-NOTICE:  Dropping the old undistribute_table.partitioned_table_1_5
-NOTICE:  Renaming the new table to undistribute_table.partitioned_table_1_5
+NOTICE:  moving the data of undistribute_table.partitioned_table_1_5
+NOTICE:  dropping the old undistribute_table.partitioned_table_1_5
+NOTICE:  renaming the new table to undistribute_table.partitioned_table_1_5
 NOTICE:  creating a new table for undistribute_table.partitioned_table_6_10
-NOTICE:  Moving the data of undistribute_table.partitioned_table_6_10
-NOTICE:  Dropping the old undistribute_table.partitioned_table_6_10
-NOTICE:  Renaming the new table to undistribute_table.partitioned_table_6_10
+NOTICE:  moving the data of undistribute_table.partitioned_table_6_10
+NOTICE:  dropping the old undistribute_table.partitioned_table_6_10
+NOTICE:  renaming the new table to undistribute_table.partitioned_table_6_10
 NOTICE:  creating a new table for undistribute_table.partitioned_table
-NOTICE:  Dropping the old undistribute_table.partitioned_table
-NOTICE:  Renaming the new table to undistribute_table.partitioned_table
+NOTICE:  dropping the old undistribute_table.partitioned_table
+NOTICE:  renaming the new table to undistribute_table.partitioned_table
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -283,9 +283,9 @@ SELECT * FROM seq_table ORDER BY a;
 
 SELECT undistribute_table('seq_table');
 NOTICE:  creating a new table for undistribute_table.seq_table
-NOTICE:  Moving the data of undistribute_table.seq_table
-NOTICE:  Dropping the old undistribute_table.seq_table
-NOTICE:  Renaming the new table to undistribute_table.seq_table
+NOTICE:  moving the data of undistribute_table.seq_table
+NOTICE:  dropping the old undistribute_table.seq_table
+NOTICE:  renaming the new table to undistribute_table.seq_table
  undistribute_table
 ---------------------------------------------------------------------
 
@@ -348,14 +348,14 @@ SELECT * FROM another_schema.undis_view3 ORDER BY 1, 2;
 
 SELECT undistribute_table('view_table');
 NOTICE:  creating a new table for undistribute_table.view_table
-NOTICE:  Moving the data of undistribute_table.view_table
-NOTICE:  Dropping the old undistribute_table.view_table
+NOTICE:  moving the data of undistribute_table.view_table
+NOTICE:  dropping the old undistribute_table.view_table
 NOTICE:  drop cascades to 3 other objects
 DETAIL:  drop cascades to view undis_view1
 drop cascades to view undis_view2
 drop cascades to view another_schema.undis_view3
 CONTEXT:  SQL statement "DROP TABLE undistribute_table.view_table CASCADE"
-NOTICE:  Renaming the new table to undistribute_table.view_table
+NOTICE:  renaming the new table to undistribute_table.view_table
  undistribute_table
 ---------------------------------------------------------------------
 

src/test/regress/expected/upgrade_list_citus_objects.out

@@ -55,6 +55,7 @@ ORDER BY 1;
  function citus_executor_name(integer)
  function citus_extradata_container(internal)
  function citus_finish_pg_upgrade()
+ function citus_get_active_worker_nodes()
  function citus_internal.columnar_ensure_objects_exist()
  function citus_internal.find_groupid_for_node(text,integer)
  function citus_internal.pg_dist_node_trigger_func()
@@ -242,5 +243,5 @@ ORDER BY 1;
  view citus_worker_stat_activity
  view pg_dist_shard_placement
  view time_partitions
-(226 rows)
+(227 rows)
 

src/test/regress/expected/upgrade_list_citus_objects_0.out

@@ -52,6 +52,7 @@ ORDER BY 1;
  function citus_executor_name(integer)
  function citus_extradata_container(internal)
  function citus_finish_pg_upgrade()
+ function citus_get_active_worker_nodes()
  function citus_internal.columnar_ensure_objects_exist()
  function citus_internal.find_groupid_for_node(text,integer)
  function citus_internal.pg_dist_node_trigger_func()
@@ -238,5 +239,5 @@ ORDER BY 1;
  view citus_worker_stat_activity
  view pg_dist_shard_placement
  view time_partitions
-(222 rows)
+(223 rows)
 

src/test/regress/expected/with_modifying.out

@@ -822,6 +822,69 @@ EXECUTE olu(1,ARRAY[1,2],ARRAY[1,2]);
  {1}       | {1}       | {NULL}
 (1 row)
 
+-- test insert query with insert CTE
+WITH insert_cte AS
+       (INSERT INTO with_modifying.modify_table VALUES (23, 7))
+INSERT INTO with_modifying.anchor_table VALUES (1998);
+SELECT * FROM with_modifying.modify_table WHERE id = 23 AND val = 7;
+ id | val
+---------------------------------------------------------------------
+ 23 |   7
+(1 row)
+
+SELECT * FROM with_modifying.anchor_table WHERE id = 1998;
+  id
+---------------------------------------------------------------------
+ 1998
+(1 row)
+
+-- test insert query with multiple CTEs
+WITH select_cte AS (SELECT * FROM with_modifying.anchor_table),
+       modifying_cte AS (INSERT INTO with_modifying.anchor_table SELECT * FROM select_cte)
+INSERT INTO with_modifying.anchor_table VALUES (1995);
+SELECT * FROM with_modifying.anchor_table ORDER BY 1;
+  id
+---------------------------------------------------------------------
+    1
+    1
+    2
+    2
+ 1995
+ 1998
+ 1998
+(7 rows)
+
+-- test with returning
+WITH returning_cte AS (INSERT INTO with_modifying.anchor_table values (1997) RETURNING *)
+INSERT INTO with_modifying.anchor_table VALUES (1996);
+SELECT * FROM with_modifying.anchor_table WHERE id IN (1996, 1997) ORDER BY 1;
+  id
+---------------------------------------------------------------------
+ 1996
+ 1997
+(2 rows)
+
+-- test insert query with select CTE
+WITH select_cte AS
+       (SELECT * FROM with_modifying.modify_table)
+INSERT INTO with_modifying.anchor_table VALUES (1990);
+SELECT * FROM with_modifying.anchor_table WHERE id = 1990;
+  id
+---------------------------------------------------------------------
+ 1990
+(1 row)
+
+-- even if we do multi-row insert, it is not fast path router due to cte
+WITH select_cte AS (SELECT 1 AS col)
+INSERT INTO with_modifying.anchor_table VALUES (1991), (1992);
+SELECT * FROM with_modifying.anchor_table WHERE id IN (1991, 1992) ORDER BY 1;
+  id
+---------------------------------------------------------------------
+ 1991
+ 1992
+(2 rows)
+
+DELETE FROM with_modifying.anchor_table WHERE id IN (1990, 1991, 1992, 1995, 1996, 1997, 1998);
 -- Test with replication factor 2
 SET citus.shard_replication_factor to 2;
 DROP TABLE modify_table;

src/test/regress/multi_schedule

@@ -72,7 +72,7 @@ test: multi_create_fdw
 # ----------
 # Tests for recursive subquery planning
 # ----------
-# NOTE: The next 6 were in parallel originally, but we got "too many
+# NOTE: The next 7 were in parallel originally, but we got "too many
 # connection" errors on CI. Requires investigation before doing them in
 # parallel again.
 test: subquery_basics
@@ -80,6 +80,7 @@ test: subquery_local_tables
 test: subquery_executors
 test: subquery_and_cte
 test: set_operations
+test: union_pushdown
 test: set_operation_and_local_tables
 
 test: subqueries_deep subquery_view subquery_partitioning subqueries_not_supported
@@ -96,6 +97,11 @@ test: tableam
 test: propagate_statistics
 test: pg13_propagate_statistics
 
+# ----------
+# Test for updating table statistics
+# ----------
+test: citus_update_table_statistics
+
 # ----------
 # Miscellaneous tests to check our query planning behavior
 # ----------