Re: [PATCH 13/13] gnu: Add beets.

[Leo Famulari]

On Fri, Mar 25, 2016 at 12:04:53PM -0400, Mark H Weaver wrote:
> Leo Famulari <address@hidden> writes:
> 
> > On Thu, Mar 03, 2016 at 08:26:39PM -0500, Leo Famulari wrote:
> >> * gnu/packages/music.scm (beets): New variable.
> >
> > I did a final cleaning pass incorporating reviewers' suggestions,
> > satisfying the linter, and updating anything that had received an upstream
> > update since I started.
> >
> > Good thing, because I had forgotten that I'd packaged an old version of
> > python-responses with the intention of updating it later! [0] This involved
> > packaging python-cookies, as well.
> >
> > Pushed as 3bf429af2fd!
> 
> Hydra failed to build this.
> 
>   http://hydra.gnu.org/build/1090043
>   http://hydra.gnu.org/build/1098131
> 
> Can you take a look?

Weird :(

Since beets is basically an interactive database, my understanding is
that the test suite tests a large variety of database operations.

I'd built it on two x86_64 machines at least half a dozen times while
testing the package, without any failures like this.

I just rebuilt successfully on those machines, and also cross-built for
i686-linux. One of the machine's CPUs was fully loaded throughout
(although no real I/O contention), so I don't think that CPU load could
cause this failure.

Maybe there is some problem caused by I/O load or a lack of RAM? I can
try again if you can tell me how to simulate heavy I/O and low RAM
conditions.

Otherwise, I'm stumped. The build logs don't have any errors messages at
all; the messages just stop and hydra reports failure. I didn't
experience this locally.

Should we just try to rebuild it again on hydra?