diff options
| author | Andres Freund <andres@anarazel.de> | 2025-04-02 07:57:11 -0400 |
|---|---|---|
| committer | Andres Freund <andres@anarazel.de> | 2025-04-02 07:57:11 -0400 |
| commit | a6285b150ad308fbba3ca832fccd199ba38fac60 (patch) | |
| tree | 27577816ccf68525a197c66be1a0a0e0f9c4275d | |
| parent | 43dca8a11624d02dde2b4bd348d77b7045c0dfbc (diff) | |
| download | postgresql-a6285b150ad308fbba3ca832fccd199ba38fac60.tar.gz postgresql-a6285b150ad308fbba3ca832fccd199ba38fac60.zip | |
tests: Fix incompatibility of test_aio with *_FORCE_RELEASE
The test added in 93bc3d75d8e failed in a build with RELCACHE_FORCE_RELEASE
and CATCACHE_FORCE_RELEASE defined. The test intentionally forgets to exit
batchmode - normally that would trigger an error at the end of the
transaction, which the test verifies. However, with RELCACHE_FORCE_RELEASE
and CATCACHE_FORCE_RELEASE defined, we get other code (output function lookup)
entering batchmode and erroring out because batchmode isn't allowed to be
entered recursively.
Fix that by changing the queries in question to not output any rows. That's
not exactly pretty, but seems to avoid the problem reliably.
Eventually we might want to make RELCACHE_FORCE_RELEASE and
CATCACHE_FORCE_RELEASE GUCs, so we can disable them where necessary - this
isn't the first test having difficulty with those debug options. But that's
for later.
Per buildfarm member prion.
Discussion: https://postgr.es/m/uc62i6vi5gd4bi6wtjj5poadqxolgy55e7ihkmf3mthjegb6zl@zqo7xez7sc2r
| -rw-r--r-- | src/test/modules/test_aio/t/001_aio.pl | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/src/test/modules/test_aio/t/001_aio.pl b/src/test/modules/test_aio/t/001_aio.pl index ff5f2219d0a..c136d8ee8f5 100644 --- a/src/test/modules/test_aio/t/001_aio.pl +++ b/src/test/modules/test_aio/t/001_aio.pl @@ -325,12 +325,18 @@ sub test_batchmode my $psql = $node->background_psql('postgres', on_error_stop => 0); + # In a build with RELCACHE_FORCE_RELEASE and CATCACHE_FORCE_RELEASE, just + # using SELECT batch_start() causes spurious test failures, because the + # lookup of the type information when printing the result tuple also + # starts a batch. The easiest way around is to not print a result tuple. + my $batch_start_sql = qq(SELECT WHERE batch_start() IS NULL); + # leak warning & recovery: implicit xact psql_like( $io_method, $psql, "batch_start() leak & cleanup in implicit xact", - qq(SELECT batch_start()), + $batch_start_sql, qr/^$/, qr/open AIO batch at end/, "$io_method: leaky batch_start() warns"); @@ -340,7 +346,7 @@ sub test_batchmode $io_method, $psql, "batch_start() leak & cleanup in explicit xact", - qq(BEGIN; SELECT batch_start(); COMMIT;), + qq(BEGIN; $batch_start_sql; COMMIT;), qr/^$/, qr/open AIO batch at end/, "$io_method: leaky batch_start() warns"); @@ -353,7 +359,7 @@ sub test_batchmode # it just means it's a bit harder to find buggy code. #psql_like($io_method, $psql, # "batch_start() leak & cleanup after abort", - # qq(BEGIN; SELECT batch_start(); ROLLBACK;), + # qq(BEGIN; $batch_start_sql; ROLLBACK;), # qr/^$/, # qr/open AIO batch at end/, "$io_method: leaky batch_start() warns"); @@ -362,7 +368,7 @@ sub test_batchmode $io_method, $psql, "batch_start(), batch_end() works", - qq(SELECT batch_start() UNION ALL SELECT batch_end()), + qq($batch_start_sql UNION ALL SELECT WHERE batch_end() IS NULL), qr/^$/, qr/^$/, "$io_method: batch_start(), batch_end()"); |