Make new replication slot test code even less racy

alvherre · alvherre · commit ce413eba4116 · 2021-07-19T17:21:07.000-04:00
Further fix the test code in ead9e51, this time by waiting until the checkpoint has completed before moving on; this ensures that the WAL segment removal has already happened when we create the next slot. Author: Kyotaro Horiguchi <horikyota.ntt@gmail.com> Discussion: https://postgr.es/m/20210719.111318.2042379313472032754.horikyota.ntt@gmail.com
diff --git a/src/test/recovery/t/019_replslot_limit.pl b/src/test/recovery/t/019_replslot_limit.pl
@@ -8,7 +8,7 @@
 use PostgresNode;
 
 use File::Path qw(rmtree);
-use Test::More tests => 15;
+use Test::More tests => 16;
 use Time::HiRes qw(usleep);
 
 $ENV{PGDATABASE} = 'postgres';
@@ -198,6 +198,19 @@
 is($result, "rep1|f|t|lost|",
 	'check that the slot became inactive and the state "lost" persists');
 
+# Wait until current checkpoint ends
+my $checkpoint_ended = 0;
+for (my $i = 0; $i < 10000; $i++)
+{
+	if (find_in_log($node_master, "checkpoint complete: ", $logstart))
+	{
+		$checkpoint_ended = 1;
+		last;
+	}
+	usleep(100_000);
+}
+ok($checkpoint_ended, 'waited for checkpoint to end');
+
 # The invalidated slot shouldn't keep the old-segment horizon back;
 # see bug #17103: https://postgr.es/m/17103-004130e8f27782c9@postgresql.org
 # Test for this by creating a new slot and comparing its restart LSN
