Skip Menu |

This queue is for tickets about the DBI CPAN distribution.

Report information
The Basics
Id: 128214
Status: new
Priority: 0/
Queue: DBI
People
Owner: Nobody in particular
Requestors: KENTNL [...] cpan.org
Cc:
AdminCc:
Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: (no value)

History Show all quoted text
  Fri Jan 11 04:20:41 2019 KENTNL [...] cpan.org - Ticket created
Subject: t/80proxy.t fails on 96-core machine with parallel-harness
On a 96 core'd machine with HARNESS_OPTIONS="-j99", for some reason, 80proxy.t fails tests:

Show quoted text
===(    1790;4   92/130   91/137  87/?  62/?  69/99   73/242  27... )===DBI connect('hostname=127.0.0.1;port=35677;debug=0;dsn=DBI:ExampleP:','',...) failed: Cannot log in to DBI::ProxyServer: Cannot connect: Connection refused at /usr/lib64/perl5/vendor_perl/5.26.2/RPC/PlClient.pm line 70. at /usr/lib64/perl5/vendor_perl/5.26.2/Net/Daemon/Log.pm line 136.
 at t/80proxy.t line 121.
Show quoted text
# failed test 1 at line 121
# failed test 2 at line 126
Use of uninitialized value in concatenation (.) or string at t/80proxy.t line 129.
Show quoted text
# failed test 3 at line 130
Can't call method "begin_work" on unblessed reference at t/80proxy.t line 150.
t/80proxy.t ..................... Dubious, test returned 255 (wstat 65280, 0xff00)
Failed 130/136 subtests


Forcing single job testing makes this problem go away, at the expense of greatly slowing down the execution of the test suite

Full test suite output and perl -V output attached in case any of it is useful.


-- 
- CPAN kentnl@cpan.org
- Gentoo Perl Maintainer kentnl@gentoo.org ( perl@gentoo.org )
Subject: testout.txt

Message body is not shown because it is too large.

  Fri Jan 11 04:36:19 2019 KENTNL [...] cpan.org - Correspondence added
Experimental testing indicates that -j78 or lower has a very low incidence of this test failing, and -j79 or higher has a very high incidence of this test failing.

-- 
- CPAN kentnl@cpan.org
- Gentoo Perl Maintainer kentnl@gentoo.org ( perl@gentoo.org )
  Fri Jan 11 05:28:36 2019 KENTNL [...] cpan.org - Correspondence added
Looks like the resource conflict is between t/80proxy.t and its variant, t/zvg_80proxy.t , due to being based on the same code.

You can reproduce this on a machine with only a handful of cores via:

prove -blr -j3 t/80proxy.t t/zvg_80proxy.t

Just it typically takes a lot more cores for this to happen accidentally.

-- 
- CPAN kentnl@cpan.org
- Gentoo Perl Maintainer kentnl@gentoo.org ( perl@gentoo.org )