Commit0d2cb6b

committed

Make new replication slot test code even less racy

Further fix the test code inead9e51, this time by waiting untilthe checkpoint has completed before moving on; this ensures that theWAL 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

1 parent344dedf commit0d2cb6bCopy full SHA for 0d2cb6b

File tree

+14

-1

lines changed

+14

-1

lines changed

Lines changed: 14 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -11,7 +11,7 @@`
`11`	`11`	`use PostgresNode;`
`12`	`12`
`13`	`13`	`use File::Pathqw(rmtree);`
`14`		`-use Test::Moretests=>$TestLib::windows_os ?15 :19;`
	`14`	`+use Test::Moretests=>$TestLib::windows_os ?16 :20;`
`15`	`15`	`use Time::HiResqw(usleep);`
`16`	`16`
`17`	`17`	`$ENV{PGDATABASE} ='postgres';`
`@@ -201,6 +201,19 @@`
`201`	`201`	`is($result,"rep1\|f\|t\|lost\|",`
`202`	`202`	`'check that the slot became inactive and the state "lost" persists');`
`203`	`203`
	`204`	`+# Wait until current checkpoint ends`
	`205`	`+my$checkpoint_ended = 0;`
	`206`	`+for (my$i = 0;$i < 10000;$i++)`
	`207`	`+{`
	`208`	`+if (find_in_log($node_primary,"checkpoint complete:",$logstart))`
	`209`	`+{`
	`210`	`+$checkpoint_ended = 1;`
	`211`	`+last;`
	`212`	`+}`
	`213`	`+usleep(100_000);`
	`214`	`+}`
	`215`	`+ok($checkpoint_ended,'waited for checkpoint to end');`
	`216`	`+`
`204`	`217`	`# The invalidated slot shouldn't keep the old-segment horizon back;`
`205`	`218`	`# see bug #17103: https://postgr.es/m/17103-004130e8f27782c9@postgresql.org`
`206`	`219`	`# Test for this by creating a new slot and comparing its restart LSN`

Comments

(0)