Rex
Rex copied to clipboard
read buffer tests on FreeBSD are bailing out sporadically
Describe the bug
Every once in a while, read buffer tests on FreeBSD are bailing out sporadically, like in this CPAN Testers report.
See #954 also, which is similar, or it's still the same even.
These two test results seems to be generated on the same testing box (same deps, rex version, perl version, OS version, architecture, uname output, etc.):
How to reproduce it
Since it seems sporadic, I'm not exactly sure about reproduction and we would probably need help from FreeBSD users.
Steps to reproduce the behavior:
- run
t/read_buffers.ttests (probably many times until it happens)
Shortest code example that demonstrates the bug:
$ prove -l t/read_buffers.t
or perhaps
$ perl t/read_buffers.t
Expected behavior
t/read_buffers.t test should reliably finish successfully on FreeBSD systems.
Circumstances
- Rex version: 1.10.0 (probably older ones too)
- Perl version: 5.30.2 (probably older ones too)
- OS running rex: FreeBSD 13.0 (probably older ones too)
- OS managed by rex: -
Debug log
Not a debug log, but test fails with:
Bailout called. Further testing stopped: Reading from buffer timed out
FAILED--Further testing stopped: Reading from buffer timed out
*** Error code 255
At over 20k iterations here with zero failures.
>uname -a
FreeBSD vixen42.vulpes.vvelox.net 12.1-STABLE FreeBSD 12.1-STABLE r359295 vixen42 amd64
use POSIX qw(fmod);
my $loop_int=0;
while ( 1 ){
if ( fmod( $loop_int, 100 ) eq 0 ){
print "loop ".$loop_int."\n";
}
my $test_out=`perl t/read_buffers.t`;
if ( $? != 0 ) {
print "errored on loop ".$loop_int."...\n".$test_out;
}
$loop_int++;
}
Added a pair of "FAIL and PASS" results from CPAN Testers to the opening post which seems to be coming from the same testing box.