Bug#444634: libwww-mechanize-shell-perl: build times out

[Lucas Nussbaum]

On 30/09/07 at 22:26 +0300, Niko Tyni wrote:
> On Sun, Sep 30, 2007 at 12:55:16AM +0200, Lucas Nussbaum wrote:
> > Package: libwww-mechanize-shell-perl
> > Version: 0.44-2
> > Severity: serious
> > User: debian-qa at lists.debian.org
> > Usertags: qa-ftbfs-20070928 qa-ftbfs
> > Justification: FTBFS on i386
>  
> > Building your package times out while running some tests: it seems that
> > all tests are successful, but that it never finishes. I'm building with
> > sbuild.
> 
> Hi,
> 
> I can't reproduce this. Both pbuilder and sbuild build run all the tests
> just fine.  The only problem I found is that the netbase package needs to
> be added manually because libwww-perl needs it but doesn't depend on it
> (#424737).

My build chroot has netbase installed.

> Is it reproducible for you? I'll come up with something to turn on
> debugging messages if it is.

quite (see below)

> > t/17-eval-multiline...................ok
> > t/18-browser-autosync.................Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > Useless content call in void context at /usr/share/perl5/LWP/Simple.pm line 150
> > ok
> 
> I don't see these either. I wonder if it's related.

Might be. I'm building on an SMP box, and there might be an issue with
threading.

> > t/embedded-HTML-Display-Win32-IE......ok
> > sbuild received SIGTERM -- shutting down
> > Killing dpkg-buildpackage subprocess 9038
> > ******************************************************************************
> > Finished at 20070929-0742
> 
> Is there a timeout here or was this a manual TERM signal? It could be
> something SIGALRM related, I guess...

There's a script killing builds that have been running for more than 8
hours. that's what happened here (the SIGTERM is caused by the script).

I tried to re-run the build several times.
1) stuck after t/23-check-dumpresponses..............ok
2) again stuck after t/23-check-dumpresponses..............ok
3) succeeded (but I was running another build in another chroot at the
same time)
4) (is the one running at the same time as (3): stuck at
t/24-source-file......................ok 1/6 

So my best guess is that there's some race condition in the tests. Can
you check on an SMP box ?
-- 
| Lucas Nussbaum
| lucas at lucas-nussbaum.net   http://www.lucas-nussbaum.net/ |
| jabber: lucas at nussbaum.fr             GPG: 1024D/023B3F4F |