Exit from base backups when shutdown is requested

When the exit waits until the whole backup completes, it may take
a very long time.

In passing, add back an error check in the main loop so we detect
clients that disconnect much earlier if the backup is large.

3 files changed, 23 insertions, 9 deletions

diff --git a/src/backend/replication/basebackup.c b/src/backend/replication/basebackup.c
index 144b17c66b6..7929f855f6d 100644
--- a/src/backend/replication/basebackup.c
+++ b/src/backend/replication/basebackup.c
@@ -308,6 +308,15 @@ sendDir(char *path, int basepathlen, bool sizeonly)
 					strlen(PG_TEMP_FILE_PREFIX)) == 0)
 			continue;
 
+		/*
+		 * Check if the postmaster has signaled us to exit, and abort
+		 * with an error in that case. The error handler further up
+		 * will call do_pg_abort_backup() for us.
+		 */
+		if (walsender_shutdown_requested || walsender_ready_to_stop)
+			ereport(ERROR,
+					(errmsg("shutdown requested, aborting active base backup")));
+
 		snprintf(pathbuf, MAXPGPATH, "%s/%s", path, de->d_name);
 
 		/* Skip postmaster.pid in the data directory */
@@ -462,7 +471,10 @@ sendFile(char *filename, int basepathlen, struct stat * statbuf)
 	while ((cnt = fread(buf, 1, Min(sizeof(buf), statbuf->st_size - len), fp)) > 0)
 	{
 		/* Send the chunk as a CopyData message */
-		pq_putmessage('d', buf, cnt);
+		if (pq_putmessage('d', buf, cnt))
+			ereport(ERROR,
+					(errmsg("base backup could not send data, aborting backup")));
+
 		len += cnt;
 
 		if (len >= statbuf->st_size)
diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c
index cacd577acce..2347567ccd6 100644
--- a/src/backend/replication/walsender.c
+++ b/src/backend/replication/walsender.c
@@ -88,8 +88,8 @@ static XLogRecPtr sentPtr = {0, 0};
 
 /* Flags set by signal handlers for later service in main loop */
 static volatile sig_atomic_t got_SIGHUP = false;
-static volatile sig_atomic_t shutdown_requested = false;
-static volatile sig_atomic_t ready_to_stop = false;
+volatile sig_atomic_t walsender_shutdown_requested = false;
+volatile sig_atomic_t walsender_ready_to_stop = false;
 
 /* Signal handlers */
 static void WalSndSigHupHandler(SIGNAL_ARGS);
@@ -426,16 +426,16 @@ WalSndLoop(void)
 		 * When SIGUSR2 arrives, we send all outstanding logs up to the
 		 * shutdown checkpoint record (i.e., the latest record) and exit.
 		 */
-		if (ready_to_stop)
+		if (walsender_ready_to_stop)
 		{
 			if (!XLogSend(output_message, &caughtup))
 				break;
 			if (caughtup)
-				shutdown_requested = true;
+				walsender_shutdown_requested = true;
 		}
 
 		/* Normal exit from the walsender is here */
-		if (shutdown_requested)
+		if (walsender_shutdown_requested)
 		{
 			/* Inform the standby that XLOG streaming was done */
 			pq_puttextmessage('C', "COPY 0");
@@ -461,7 +461,7 @@ WalSndLoop(void)
 
 			if (!XLogSend(output_message, &caughtup))
 				break;
-			if (caughtup && !got_SIGHUP && !ready_to_stop && !shutdown_requested)
+			if (caughtup && !got_SIGHUP && !walsender_ready_to_stop && !walsender_shutdown_requested)
 			{
 				/*
 				 * XXX: We don't really need the periodic wakeups anymore,
@@ -841,7 +841,7 @@ WalSndSigHupHandler(SIGNAL_ARGS)
 static void
 WalSndShutdownHandler(SIGNAL_ARGS)
 {
-	shutdown_requested = true;
+	walsender_shutdown_requested = true;
 	if (MyWalSnd)
 		SetLatch(&MyWalSnd->latch);
 }
@@ -889,7 +889,7 @@ WalSndXLogSendHandler(SIGNAL_ARGS)
 static void
 WalSndLastCycleHandler(SIGNAL_ARGS)
 {
-	ready_to_stop = true;
+	walsender_ready_to_stop = true;
 	if (MyWalSnd)
 		SetLatch(&MyWalSnd->latch);
 }
diff --git a/src/include/replication/walsender.h b/src/include/replication/walsender.h
index 0b4a143f827..9039b240c44 100644
--- a/src/include/replication/walsender.h
+++ b/src/include/replication/walsender.h
@@ -53,6 +53,8 @@ extern WalSndCtlData *WalSndCtl;
 
 /* global state */
 extern bool am_walsender;
+extern volatile sig_atomic_t walsender_shutdown_requested;
+extern volatile sig_atomic_t walsender_ready_to_stop;
 
 /* user-settable parameters */
 extern int	WalSndDelay;