On Mon, Jul 31, 2017 at 06:12:54PM +0200, László Böszörményi (GCS) wrote:
> Hi Adrian,
> On Mon, Jul 31, 2017 at 2:31 PM, Adrian Bunk <b...@debian.org> wrote:
> > Source: graphicsmagick
> > Version: 1.3.26-3
> > Severity: serious
> > 2 out of the 4 latest build attempts FTBFS on ppc64el:
> > PASS: Magick++/tests/tests.tap 7 - colorHistogram
> > PASS: Magick++/tests/tests.tap 8 - exceptions
> > PASS: Magick++/tests/tests.tap 9 - montageImages
> > PASS: Magick++/tests/tests.tap 10 - morphImages
> > PASS: Magick++/tests/tests.tap 11 - readWriteBlob
> > PASS: Magick++/tests/tests.tap 12 - readWriteImages
> > E: Build killed with signal TERM after 150 minutes of inactivity
> Seen these, I do follow auto builds.
> > The next test is "Magick++/demo/demos.tap 1 - analyze",
> > this seems to be the hanging one?
> I have no idea. I know three things. It can not be reproduced on the
> porterbox - I've tried the build several times. The build occasionally
> hangs since the Perl transition started - but the mentioned test rules
> are shell scripts.
The problem is not the perl transition:
The first failure was on 2017-07-12
The perl transition started on 2017-07-23
> But the most important thing is that the fail always happened on
> ppc64el-unicamp-01 - the other buildd, ppc64el-osuosl-01 can build it
> without any problem. As such, I think it's the buildd that different
> in some way than the source has any known problem.
> Can you reschedule the build, first on ppc64el-unicamp-01, then on
> ppc64el-osuosl-01 and see the results?
I don't have any powers to reschedule anything.
> Maybe the former is not up to
> date and has some inconsistent package set installed?
chroots on the buildds are regenerated twice per week.
And since graphicsmagick is the only package with this problem,
it is unlikely that the problem is something like that.
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed