From: Tom Lane <tgl@sss.pgh.pa.us>
Date: Fri, 7 Feb 2025 18:41:43 +0000 (-0500)
Subject: Fix pgbench performance issue induced by commit af35fe501.
X-Git-Tag: REL_14_16~11
X-Git-Url: http://git.postgresql.org/gitweb/?a=commitdiff_plain;h=5addea71c7c7693ed9aa071e122038e2dd7011ab;p=postgresql.git

Fix pgbench performance issue induced by commit af35fe501.

Commit af35fe501 caused "pgbench -i" to emit a '\r' character
for each data row loaded (when stderr is a terminal).
That's effectively invisible on-screen, but it causes the
connected terminal program to consume a lot of cycles.
It's even worse if you're connected over ssh, as the data
then has to pass through the ssh tunnel.

Simplest fix is to move the added logic inside the if-tests
that check whether to print a progress line.  We could do
it another way that avoids duplicating these few lines,
but on the whole this seems the most transparent way to
write it.

Like the previous commit, back-patch to all supported versions.

Reported-by: Andres Freund <andres@anarazel.de>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Nathan Bossart <nathandbossart@gmail.com>
Discussion: https://postgr.es/m/4k4drkh7bcmdezq6zbkhp25mnrzpswqi2o75d5uv2eeg3aq6q7@b7kqdmzzwzgb
Backpatch-through: 13
---

diff --git a/src/bin/pgbench/pgbench.c b/src/bin/pgbench/pgbench.c
index 1f6cb0e6038..bef07be3925 100644
--- a/src/bin/pgbench/pgbench.c
+++ b/src/bin/pgbench/pgbench.c
@@ -4312,6 +4312,16 @@ initGenerateDataClientSide(PGconn *con)
 							j, (int64) naccounts * scale,
 							(int) (((int64) j * 100) / (naccounts * (int64) scale)),
 							elapsed_sec, remaining_sec);
+
+			/*
+			 * If the previous progress message is longer than the current
+			 * one, add spaces to the current line to fully overwrite any
+			 * remaining characters from the previous message.
+			 */
+			if (prev_chars > chars)
+				fprintf(stderr, "%*c", prev_chars - chars, ' ');
+			fputc(eol, stderr);
+			prev_chars = chars;
 		}
 		/* let's not call the timing for each row, but only each 100 rows */
 		else if (use_quiet && (j % 100 == 0))
@@ -4326,20 +4336,20 @@ initGenerateDataClientSide(PGconn *con)
 								j, (int64) naccounts * scale,
 								(int) (((int64) j * 100) / (naccounts * (int64) scale)), elapsed_sec, remaining_sec);
 
+				/*
+				 * If the previous progress message is longer than the current
+				 * one, add spaces to the current line to fully overwrite any
+				 * remaining characters from the previous message.
+				 */
+				if (prev_chars > chars)
+					fprintf(stderr, "%*c", prev_chars - chars, ' ');
+				fputc(eol, stderr);
+				prev_chars = chars;
+
 				/* skip to the next interval */
 				log_interval = (int) ceil(elapsed_sec / LOG_STEP_SECONDS);
 			}
 		}
-
-		/*
-		 * If the previous progress message is longer than the current one,
-		 * add spaces to the current line to fully overwrite any remaining
-		 * characters from the previous message.
-		 */
-		if (prev_chars > chars)
-			fprintf(stderr, "%*c", prev_chars - chars, ' ');
-		fputc(eol, stderr);
-		prev_chars = chars;
 	}
 
 	if (eol != '\n')