Revert "Bump up max_wal_senders, to fix src/test/walrep/ tests."

This reverts commit f38f024d7d314ae8ac100d862307a228b212c70e.

src/backend/postmaster/postmaster.c

@@ -1142,12 +1142,8 @@ PostmasterMain(int argc, char *argv[])
 	/*
 	 * This value of max_wal_senders will be inherited by all the child processes
 	 * through fork(). This value is used by XLogIsNeeded().
-	 *
-	 * In a normal setup, each node only has one replica. But have some headroom,
-	 * for other tools that want to use a replication connection, and for hung
-	 * connections, etc.
 	 */
-	max_wal_senders = 5;
+	max_wal_senders = 1;
 
 	if ( GpIdentity.numsegments < 0 )
 	{

src/backend/replication/gp_replication.c

@@ -26,12 +26,8 @@
  * Check the WalSndCtl to obtain if mirror is up or down, if the wal sender is
  * in streaming, and if synchronous replication is enabled or not.
  */
-void
-GetMirrorStatus(FtsResponse *response)
+void GetMirrorStatus(FtsResponse *response)
 {
-	int			i;
-	pg_time_t	most_recent_marked_pid_zero_at_time;
-
 	response->IsMirrorUp = false;
 	response->IsInSync = false;
 	response->RequestRetry = false;
@@ -40,23 +36,31 @@ GetMirrorStatus(FtsResponse *response)
 	 * Greenplum currently supports only ONE mirror per primary.
 	 * If there are more mirrors, this logic in this function need to be revised.
 	 */
-	Assert(max_wal_senders >= 1);
+	Assert(max_wal_senders == 1);
 
 	LWLockAcquire(SyncRepLock, LW_SHARED);
 
-	most_recent_marked_pid_zero_at_time = 0;
-	for (i = 0; i < max_wal_senders; i++)
+	for (int i = 0; i < max_wal_senders; i++)
 	{
 		/* use volatile pointer to prevent code rearrangement */
 		volatile WalSnd *walsnd = &WalSndCtl->walsnds[i];
 
 		if (walsnd->pid == 0)
 		{
-			pg_time_t	marked_pid_zero_at_time = walsnd->marked_pid_zero_at_time;
-
-			Assert(marked_pid_zero_at_time);
-			if (most_recent_marked_pid_zero_at_time > marked_pid_zero_at_time)
-				most_recent_marked_pid_zero_at_time = marked_pid_zero_at_time;
+			Assert(walsnd->marked_pid_zero_at_time);
+			pg_time_t delta = ((pg_time_t) time(NULL)) - walsnd->marked_pid_zero_at_time;
+			/*
+			 * Report mirror as down, only if it didn't connect for below
+			 * grace period to primary. This helps to avoid marking mirror
+			 * down unnecessarily when restarting primary or due to small n/w
+			 * glitch. During this period, request FTS to probe again.
+			 */
+			if (delta < FTS_MARKING_MIRROR_DOWN_GRACE_PERIOD)
+			{
+				elog(LOG,
+					 "requesting fts retry as mirror didn't connect yet but in grace period " INT64_FORMAT, delta);
+				response->RequestRetry = true;
+			}
 		}
 		else
 		{
@@ -70,24 +74,6 @@ GetMirrorStatus(FtsResponse *response)
 		}
 	}
 
-	if (i == max_wal_senders)
-	{
-		pg_time_t delta = ((pg_time_t) time(NULL)) - most_recent_marked_pid_zero_at_time;
-
-		/*
-		 * Report mirror as down, only if it didn't connect for below
-		 * grace period to primary. This helps to avoid marking mirror
-		 * down unnecessarily when restarting primary or due to small n/w
-		 * glitch. During this period, request FTS to probe again.
-		 */
-		if (delta < FTS_MARKING_MIRROR_DOWN_GRACE_PERIOD)
-		{
-			elog(LOG,
-				 "requesting fts retry as mirror didn't connect yet but in grace period " INT64_FORMAT, delta);
-			response->RequestRetry = true;
-		}
-	}
-
 	response->IsSyncRepEnabled = WalSndCtl->sync_standbys_defined;
 
 	LWLockRelease(SyncRepLock);