Re: Bug#787258: FTBFS on mipsel: test-suite failures

James Cowgill <[email protected]> Tue, 02 Jun 2015 10:16:59 +0100

Newsgroups	gmane.linux.debian.ports.mips
Message-ID	<[email protected]>

Hi,

On Mon, 2015-06-01 at 16:41 +0200, Martin Pitt wrote:
> Martin Pitt [2015-06-01  8:02 +0200]:
> > Since this doesn't happen on our mipsel porter box, and I don't have
> > any other mipsel hardware to check this on, it seems this needs some
> > help from the mipsel buildd maintainers. It would be really helpful if
> > our buildd and the porter box were more alike, to be able to reproduce
> > such issues?
> 
> Today's upload had no test failures:
> 
>   https://buildd.debian.org/status/fetch.php?pkg=systemd&arch=mipsel&ver=220-3&stamp=1433167263
> 
> This was built on mipsel-aql-01. Looking into the history on
> 
>   https://buildd.debian.org/status/logs.php?pkg=systemd&arch=mipsel
> 
> it seems that mipsel-aql-01 never failed, while mipsel-manda-01
> recently always failed, and eberlin is somewhat random. I guess one
> shouldn't read too much into that short history, presumably they are
> all a bit flaky. But do you know anything fundamental (kernel version
> etc.) where the porter box (eder) is different from these buildds?

I tried building systemd on a few different machines and couldn't get it
to fail on any of them. Then I looked at what packages were installed on
the buildds. On the machines it fails on, sysvinit-core is installed and
on mipsel-aql-01 (which passed the tests) systemd-sysv is installed.

I then found this:
https://bitbucket.org/Tarnyko/uselessd/issue/7/uselessd-7-test-strv-fails-on-slackware

Presumably systemd has never been installed within the mipsel-manda-01
buildd chroot and so no machine-id has ever been generated for it (which
the tests rely on).

Thanks,
James

signature.asc (application/pgp-signature, 819 B)