fix ddl recovery PGPRO-1699

Author: kelvich

contrib/mmts/multimaster.c

@@ -5028,9 +5028,6 @@ char* MtmGucSerialize(void)
 
 static void MtmProcessDDLCommand(char const* queryString, bool transactional)
 {
-	if (MtmTx.isReplicated)
-		return;
-
 	if (transactional)
 	{
 		char *gucCtx = MtmGucSerialize();
@@ -5462,7 +5459,7 @@ static void MtmProcessUtility(PlannedStmt *pstmt,
 			break;
 	}
 
-	if (!skipCommand && !MtmTx.isReplicated && !MtmDDLStatement)
+	if (!skipCommand && !MtmDDLStatement)
 	{
 		MTM_LOG3("Process DDL statement '%s', MtmTx.isReplicated=%d, "
 				 "MtmIsLogicalReceiver=%d", stmt_string, MtmTx.isReplicated,

contrib/mmts/tests_testgres/ddl.py

@@ -0,0 +1,38 @@
+import unittest
+import subprocess
+import time
+
+from mm_cluster import Cluster
+
+NUM_NODES = 3
+
+class TestDDL(unittest.TestCase):
+
+    @classmethod
+    def setUpClass(cls):
+        cls.cluster = Cluster(NUM_NODES)
+        cls.cluster.print_conninfo()
+        cls.cluster.start().await_online((0,1,2))
+        # cls.cluster.print_conninfo()
+
+    @classmethod
+    def tearDownClass(cls):
+        cls.cluster.stop()
+
+    # Check that recovery properly processes
+    def test_dll_recovery(self):
+        # create table while one node is stopped
+        self.cluster.nodes[2].stop()
+        self.cluster.await_online((0,1))
+        self.cluster.nodes[0].safe_psql(query='create table t(id int primary key)')
+
+        # now if second node didn't store logical message with DDL and third
+        # node will recover from second then it will not receive this
+        # 'create table' (PGPRO-1699)
+        self.cluster.nodes[2].start()
+        self.cluster.await_online((0,1,2))
+        self.cluster.nodes[2].safe_psql(query='insert into t values(42)')
+
+
+if __name__ == '__main__':
+    unittest.main()