On sparc64+ext4, suppress test failures from known WAL read failure.

Buildfarm members kittiwake, tadarida and snapper began to fail
frequently when commits 3cd9c3b921977272e6650a5efbeade4203c4bca2 and
f47ed79cc8a0cfa154dc7f01faaf59822552363f added tests of concurrency, but
the problem was reachable before those commits.  Back-patch to v10 (all
supported versions).

Discussion: https://postgr.es/m/20220116210241[email protected]

diff --git a/contrib/amcheck/t/003_cic_2pc.pl b/contrib/amcheck/t/003_cic_2pc.pl
index c5391869655f5f436d9a5d1d373b105915ac276c..d81734fca784cc160c56b7a1433b82547eabd8b4 100644 (file)
--- a/contrib/amcheck/t/003_cic_2pc.pl
+++ b/contrib/amcheck/t/003_cic_2pc.pl
@@ -11,6 +11,8 @@ use TestLib;
 
 use Test::More tests => 5;
 
+local $TODO = 'filesystem bug' if TestLib::has_wal_read_bug;
+
 my ($node, $result);
 
 #

diff --git a/src/test/perl/TestLib.pm b/src/test/perl/TestLib.pm
index 0b9cd2cd99d75a4a7c5c8aafa5b4970f0753db50..f409a30acaf1b4d30f80a8259c4be67c9b01b622 100644 (file)
--- a/src/test/perl/TestLib.pm
+++ b/src/test/perl/TestLib.pm
@@ -315,6 +315,29 @@ sub perl2host
 
 =pod
 
+=item has_wal_read_bug()
+
+Returns true if $tmp_check is subject to a sparc64+ext4 bug that causes WAL
+readers to see zeros if another process simultaneously wrote the same offsets.
+Consult this in tests that fail frequently on affected configurations.  The
+bug has made streaming standbys fail to advance, reporting corrupt WAL.  It
+has made COMMIT PREPARED fail with "could not read two-phase state from WAL".
+Non-WAL PostgreSQL reads haven't been affected, likely because those readers
+and writers have buffering systems in common.  See
+https://postgr.es/m/[email protected] for details.
+
+=cut
+
+sub has_wal_read_bug
+{
+   return
+        $Config{osname} eq 'linux'
+     && $Config{archname} =~ /^sparc/
+     && !run_log([ qw(df -x ext4), $tmp_check ], '>', '/dev/null', '2>&1');
+}
+
+=pod
+
 =item system_log(@cmd)
 
 Run (via C) the command passed as argument; the return