On 04/01/2015 05:11 PM, Kevin Fenzi wrote:
On Wed, 01 Apr 2015 17:01:06 -0600
Orion Poplawski <orion@xxxxxxxxxxxxx> wrote:
I'm trying to debug a crash in openmpi on the armv7hl builders:
Open MPI tried to bind a new process, but something went wrong. The
process was killed without launching the target application. Your job
will now abort.
Local host: arm02-builder01
Application name: ./tst_parallel4
Error message: hwloc_set_cpubind returned "Error" for bitmap
"0-3" Location: odls_default_module.c:551
I can't reproduce it on my local armv7hl qemu VM as it appears to be
coupled to the actual topology of the machine.
Anyone have any other suggestions for debugging?
Yes, use arm03-packager00.cloud.fedoraproject.org (or
arm03-packager01.cloud.fedoraproject.org) and try and duplicate in a
mock chroot there? They are both running f21 (the same as the
builders) on the same hardware make/model as the builders.
kevin
Thanks a lot for that. Unfortunately I'm now running into
https://bugzilla.redhat.com/show_bug.cgi?id=1196181
--
Orion Poplawski
Technical Manager 303-415-9701 x222
NWRA/CoRA Division FAX: 303-415-9702
3380 Mitchell Lane orion@xxxxxxxxxxxxx
Boulder, CO 80301 http://www.cora.nwra.com
--
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/devel
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
