Fix over-eager ping'ing in logical replication receiver.

tglsfdc · tglsfdc · commit 7156a0eac3af · 2020-09-04T20:20:05.000-04:00
Commit 3f60f69 only partially fixed the broken-status-tracking issue in LogicalRepApplyLoop: we need ping_sent to have the same lifetime as last_recv_timestamp. The effects are much less serious than what that commit fixed, though. AFAICS this would just lead to extra ping requests being sent, once per second until the sender responds. Still, it's a bug, so backpatch to v10 as before. Discussion: https://postgr.es/m/959627.1599248476@sss.pgh.pa.us
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
@@ -1145,6 +1145,7 @@ static void
 LogicalRepApplyLoop(XLogRecPtr last_received)
 {
 	TimestampTz last_recv_timestamp = GetCurrentTimestamp();
+	bool		ping_sent = false;
 
 	/*
 	 * Init the ApplyMessageContext which we clean up after each replication
@@ -1157,14 +1158,14 @@ LogicalRepApplyLoop(XLogRecPtr last_received)
 	/* mark as idle, before starting to loop */
 	pgstat_report_activity(STATE_IDLE, NULL);
 
+	/* This outer loop iterates once per wait. */
 	for (;;)
 	{
 		pgsocket	fd = PGINVALID_SOCKET;
 		int			rc;
 		int			len;
 		char	   *buf = NULL;
 		bool		endofstream = false;
-		bool		ping_sent = false;
 		long		wait_time;
 
 		CHECK_FOR_INTERRUPTS();
@@ -1175,7 +1176,7 @@ LogicalRepApplyLoop(XLogRecPtr last_received)
 
 		if (len != 0)
 		{
-			/* Process the data */
+			/* Loop to process all available data (without blocking). */
 			for (;;)
 			{
 				CHECK_FOR_INTERRUPTS();
@@ -1348,10 +1349,7 @@ LogicalRepApplyLoop(XLogRecPtr last_received)
 					ereport(ERROR,
 							(errmsg("terminating logical replication worker due to timeout")));
 
-				/*
-				 * We didn't receive anything new, for half of receiver
-				 * replication timeout. Ping the server.
-				 */
+				/* Check to see if it's time for a ping. */
 				if (!ping_sent)
 				{
 					timeout = TimestampTzPlusMilliseconds(last_recv_timestamp,
