Hi, Andrey!
I started reviewing the TAP tests and for the current master (14ee8e6403001c3788f2622cdcf81a8451502dc2),
src/test/modules/test_slru/t/001_multixact.pl reproduces the problem, but we can do it in a more transparent way.
The test should fail on timeout; otherwise, it would be hard to find the source of the problem.
The current state of 001_multixact.pl just leaves the 'mike' node in a deadlock state and
it looks like a very long test that doesn't finish for vague reasons.
I believe that we should provide more comprehensible feedback to developers
by interrupting the deadlock state with a psql timeout. 15 seconds seems safe
enough to distinguish between slow node operation and deadlock issue.
Here is the modified 001_multixact.pl
----
# Verify thet recorded multi is readble, this call must not hang.
# Also note that all injection points disappeared after server restart.
my $timed_out = 0;
$node->safe_psql(
'postgres',
qq{SELECT test_read_multixact('$multi'::xid);},
timeout => 15,
timed_out => \$timed_out);
ok($timed_out == 0, 'recorded multi is readble');
$node->stop;
done_testing();
----
