Is anybody around who can check what's going on with a PPC64 build that appears to be stuck? I noticed that fflas-ffpack has not built successfully on ppc64 recently, so I kicked off a scratch build last night to try to diagnose the problem. It finished fairly quickly on ppc64le, but is still going on ppc64. The last step in the logs is a g++ invocation. I wonder if g++ is inflooping, if the box is so low on memory that it is swapping like crazy, etc.
http://ppc.koji.fedoraproject.org/koji/taskinfo?taskID=3875464
Thanks to anyone who can take a peek at that.
On a related note, the polymake build segfaults on ppc64 (but succeeds on ppc64le). I have access to a ppc64le VM, but not to a big endian machine. Who can give me a shell account with mock capabilities on a ppc64 machine?
Thank you and regards,
On Mon, Nov 14, 2016 at 12:03:33PM -0700, Jerry James wrote:
Is anybody around who can check what's going on with a PPC64 build that appears to be stuck? I noticed that fflas-ffpack has not built successfully on ppc64 recently, so I kicked off a scratch build last night to try to diagnose the problem. It finished fairly quickly on ppc64le, but is still going on ppc64. The last step in the logs is a g++ invocation. I wonder if g++ is inflooping, if the box is so low on memory that it is swapping like crazy, etc.
http://ppc.koji.fedoraproject.org/koji/taskinfo?taskID=3875464
Thanks to anyone who can take a peek at that.
On a related note, the polymake build segfaults on ppc64 (but succeeds on ppc64le). I have access to a ppc64le VM, but not to a big endian machine. Who can give me a shell account with mock capabilities on a ppc64 machine?
puiterwijk possibly.
Zbyszek
Hi,
On Mon, Nov 14, 2016 at 12:03:33PM -0700, Jerry James wrote:
puiterwijk possibly.
Yes. For now, you can just ping me on IRC, or create a ticket on https://pagure.io/fedora-infrastructure to get a PPC64(LE) instance on our Infrastructure cloud.
In the near future, we hope to open up self-service instance creation so that people can spin up their own PPC64(LE) instances without needing to create a ticket, but I'll email the list when that time comes.
Zbyszek
Regards, Patrick Uiterwijk
On Mon, 14 Nov 2016 12:03:33 -0700 Jerry James loganjerry@gmail.com wrote:
Is anybody around who can check what's going on with a PPC64 build that appears to be stuck? I noticed that fflas-ffpack has not built successfully on ppc64 recently, so I kicked off a scratch build last night to try to diagnose the problem. It finished fairly quickly on ppc64le, but is still going on ppc64. The last step in the logs is a g++ invocation. I wonder if g++ is inflooping, if the box is so low on memory that it is swapping like crazy, etc.
http://ppc.koji.fedoraproject.org/koji/taskinfo?taskID=3875464
Thanks to anyone who can take a peek at that.
Looks like it's in tests?
├─kojid,14284 /usr/sbin/kojid --fg --force-lock --verbose │ ├─kojid,23608 /usr/sbin/kojid --fg --force-lock --verbose │ └─kojid,32678 /usr/sbin/kojid --fg --force-lock --verbose │ └─mock,532 -tt /usr/sbin/mock -r koji/f26-build-1023296-763894 --no-clean --target ppc64 --rebuild... │ └─rpmbuild,10547 -bb --target ppc64 --nodeps /builddir/build/SPECS/fflas-ffpack.spec │ └─sh,15499 -e /var/tmp/rpm-tmp.zdaiHn │ └─make,15501 check │ └─sh,15502 -c... │ └─sh,15571 -c... │ └─make,15572 check │ └─sh,15573 -c... │ └─make,15586 check-am │ └─make,15587 check-TESTS │ └─sh,15594 -c... │ └─make,15603 test-suite.log... │ └─sh,15638 -c... │ └─sh,15642 ../build-aux/test-driver --test-name ... │ └─test-lu,15643
kojibui+ 15643 99.9 0.1 24192 12160 ? R 03:47 978:58 ./test-lu
strace doesnt show anything at all...
Hope that helps some.
kevin
On Mon, 14 Nov 2016 12:03:33 -0700 Jerry James loganjerry@gmail.com wrote:
Is anybody around who can check what's going on with a PPC64 build that appears to be stuck? I noticed that fflas-ffpack has not built successfully on ppc64 recently, so I kicked off a scratch build last night to try to diagnose the problem. It finished fairly quickly on ppc64le, but is still going on ppc64. The last step in the logs is a g++ invocation. I wonder if g++ is inflooping, if the box is so low on memory that it is swapping like crazy, etc.
http://ppc.koji.fedoraproject.org/koji/taskinfo?taskID=3875464
Thanks to anyone who can take a peek at that.
On a related note, the polymake build segfaults on ppc64 (but succeeds on ppc64le). I have access to a ppc64le VM, but not to a big endian machine. Who can give me a shell account with mock capabilities on a ppc64 machine?
the fflas-ffpack is tracked upstream in https://github.com/linbox-team/fflas-ffpack/issues/45
Dan
On Mon, 14 Nov 2016 12:03:33 -0700 Jerry James loganjerry@gmail.com wrote:
Is anybody around who can check what's going on with a PPC64 build that appears to be stuck? I noticed that fflas-ffpack has not built successfully on ppc64 recently, so I kicked off a scratch build last night to try to diagnose the problem. It finished fairly quickly on ppc64le, but is still going on ppc64. The last step in the logs is a g++ invocation. I wonder if g++ is inflooping, if the box is so low on memory that it is swapping like crazy, etc.
http://ppc.koji.fedoraproject.org/koji/taskinfo?taskID=3875464
Thanks to anyone who can take a peek at that.
On a related note, the polymake build segfaults on ppc64 (but succeeds on ppc64le). I have access to a ppc64le VM, but not to a big endian machine. Who can give me a shell account with mock capabilities on a ppc64 machine?
and for polymake - it segfaults the same way on s390x (http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=2398862), likely a big endian issue
Dan
(Two weeks pass...)
On Mon, Nov 14, 2016 at 1:17 PM, Dan Horák dan@danny.cz wrote:
and for polymake - it segfaults the same way on s390x (http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=2398862), likely a big endian issue
Thank you Dan, Kevin, Patrick, and Zbigniew. I did not intend to drop out of this email thread, but shortly after I sent this message, the motherboard on my home computer suffered some kind of catastrophic failure. Couldn't even get to the BIOS setup. For me, Black Friday this year was a great day, as it was the day my new computer equipment arrived. After a weekend getting everything set back up, I'm ready to dive back into this issue. Thanks again and regards,
-
Dan Horák -
Jerry James -
Kevin Fenzi -
Patrick Uiterwijk -
Zbigniew Jędrzejewski-Szmek