Skip Menu |

This queue is for tickets about the System-Command CPAN distribution.

Report information
The Basics
Id: 74959
Status: resolved
Priority: 0/
Queue: System-Command

People
Owner: Nobody in particular
Requestors: TODDR [...] cpan.org
Cc: gregoa [...] cpan.org
AdminCc:

Bug Information
Severity: Normal
Broken in: 1.06
Fixed in: 1.07



Subject: Tests fail on CentoOS 6
http://www.cpantesters.org/cpan/report/8dc5d7fc-3fd3-11e1-9d6f-f6dbfa7543f5 [ERROR] [Sun Jan 15 23:48:43 2012] MAKE TEST failed: # Tested System::Command 1.06, Perl 5.010000, /home/cpan/pit/thr-ld/perl-5.10.0/bin/perl t/00-load.t ....... ok t/01-load.t ....... ok t/10-command.t .... ok # DESTROY System::Command=HASH(0xae07c0) # DESTROY System::Command=HASH(0xae0898) # DESTROY System::Command=HASH(0xae8c08) # DESTROY System::Command=HASH(0xae8568) # DESTROY System::Command=HASH(0xb01750) # DESTROY System::Command=HASH(0xae8fb0) # DESTROY System::Command=HASH(0xae95c8) t/15-scope.t ...... ok # $SIG{CHLD} = 'IGNORE' # Failed test 'child was reaped' # at t/20-zombie.t line 82. # Looks like you failed 1 test of 28. t/20-zombie.t ..... Dubious, test returned 1 (wstat 256, 0x100) Failed 1/28 subtests t/90-command.t .... skipped: Test::Command not available t/90-output.t ..... skipped: Test::Output not available t/manifest.t ...... skipped: Author tests not required for installation t/pod-coverage.t .. skipped: Test::Pod::Coverage 1.08 required for testing POD coverage t/pod.t ........... skipped: Test::Pod 1.22 required for testing POD Test Summary Report ------------------- t/20-zombie.t (Wstat: 256 Tests: 28 Failed: 1) Failed test: 25 Non-zero exit status: 1 Files=10, Tests=166, 8 wallclock secs ( 0.05 usr 0.02 sys + 0.40 cusr 0.07 csys = 0.54 CPU) Result: FAIL Failed 1/10 test programs. 1/166 subtests failed. BEGIN failed--compilation aborted at -e line 1. [MSG] [Sun Jan 15 23:48:43 2012] DEFAULT 'proceed_on_test_failure' HANDLER RETURNING 'sub return value'
I'm getting this consistently on my centos6 automated build systems. I cannot re-produce it interactively. I suspect the lack of terminal is causing it to behave unexpectedly. Perhaps some debugging on failures for line 82? If you can suggest something, I'll put it in place and test it. Thanks, Todd
On Mon Feb 13 10:49:33 2012, TODDR wrote: Show quoted text
> I'm getting this consistently on my centos6 automated build systems. I > cannot re-produce it interactively. I suspect the lack of terminal is > causing it to behave unexpectedly. Perhaps some > debugging on failures for line 82? If you can suggest something, I'll > put it in place and test it. > > Thanks, > Todd
I get this too, on CentOS 6.2.
This is off the top of my head, but I thought I'd mention it: I discovered on a related autodie ticket that when I connect via net::openssh to centos6 and then run a perl program that STDIN is not actually an IO::Handle, but instead an IO::Socket. This is probably the Crux of the problem. Todd
I get the same failure with perl 5.14 on ubuntu 11.10: https://gist.github.com/2282848
We're seeing the same on Debian: http://bugs.debian.org/666631 See Niko Tyni's excellent analysis of what's happening in the bug report. Excerpt: "... there was a window of seven milliseconds where the reaped child process could still be signaled ..." Cheers, gregor, Debian Perl Group
On Sun Apr 15 11:38:13 2012, GREGOA wrote: Show quoted text
> We're seeing the same on Debian: > > http://bugs.debian.org/666631 > > See Niko Tyni's excellent analysis of what's happening in the bug
report. Show quoted text
> > Excerpt: > "... there was a window of seven milliseconds where the > reaped child process could still be signaled ..." > >
Thanks everyone for the reports and the descriptions. I've use the debian bug report to avoid the issue (if the window where the reaped child can be signaled exists, the test emits a diagnostic). The fix in included in v1.07. -- BooK
On Sun Apr 15 11:38:13 2012, GREGOA wrote: Show quoted text
> We're seeing the same on Debian: > > http://bugs.debian.org/666631 > > See Niko Tyni's excellent analysis of what's happening in the bug
report. Show quoted text
> > Excerpt: > "... there was a window of seven milliseconds where the > reaped child process could still be signaled ..." > >
Thanks everyone for the reports and the descriptions. I've used the debian bug report to avoid the issue (if the window where the reaped child can be signaled exists, the test emits a diagnostic). The fix in included in v1.07. -- BooK