Bug ID 997252
Summary hdf5 make check hung for ppc64le
Classification openSUSE
Product openSUSE Tumbleweed
Version Current
Hardware PowerPC-64
OS Other
Status NEW
Severity Normal
Priority P5 - None
Component Other
Assignee bnc-team-screening@forge.provo.novell.com
Reporter normand@linux.vnet.ibm.com
QA Contact qa-bugs@suse.de
Found By ---
Blocker ---

hdf5 make check hung for ppc64le

on Staging B (1), the hdf5 build package hang on testphdf5 (2) for the many
trials as reported by jobhist output (3)

If I build in a ppc64le guest, I identify the hung as related to mpi blocked on
a select call (4)

=== (1)
(11:14:36) maxlin: michel_mno: thanks. and note that, it's grouped with newer
glibc and an changed openmpi
https://build.opensuse.org/project/show/openSUSE:Factory:Staging:B
requests:
- {author: Andreas_Schwab, id: 417981, package: glibc}
- {author: pluskalm, id: 420140, package: ctags}
- {author: WernerFink, id: 420469, package: emacs}
- {author: dimstar_suse, id: 423801, package: openmpi}
=== (2)
[ 2757s] MPI tests finished with no errors
[ 2757s] ===================================
[ 2757s]
[ 2757s] Finished testing t_mpi
[ 2757s] ============================
[ 2757s] make[4]: Leaving directory
'/home/abuild/rpmbuild/BUILD/hdf5-1.10.0-patch1/openmpi/testpar'
[ 2757s] make[4]: Entering directory
'/home/abuild/rpmbuild/BUILD/hdf5-1.10.0-patch1/openmpi/testpar'
[ 2757s] ============================
[ 2757s] Testing  testphdf5
=== (3)
$osc jobhist  openSUSE:Factory:Staging:B:DVD hdf5 standard ppc64le
...
2016-08-19 18:46:04  hdf5 meta change      succeeded 1h  2m 30s   build89:12
2016-08-23 08:17:27  hdf5 meta change      succeeded 1h  1m 20s   build95:2
2016-08-30 23:38:01  hdf5 meta change      failed    8h 47m 15s   build96:11
2016-08-31 09:08:45  hdf5 new build        failed    8h 48m 32s   build96:8
2016-08-31 19:34:35  hdf5 new build        failed    8h 48m 19s   build96:9
2016-09-01 04:25:45  hdf5 new build        failed    8h 46m 30s   build89:4
2016-09-01 21:57:52  hdf5 new build        failed    8h 47m 33s   build89:6
2016-09-02 19:48:00  hdf5 new build        failed    8h 47m 17s   build89:2
2016-09-04 08:19:54  hdf5 new build        failed    8h 45m 50s   build95:16
2016-09-04 20:47:19  hdf5 new build        failed    8h 45m 54s   build96:16
2016-09-05 07:05:24  hdf5 new build        failed    8h 45m 36s   build96:6  
===
=== (4)
 1811 ? S  0:00  |\_ make check
 1813 ? S  0:00  |    \_ make check-TESTS
 1818 ? S  0:00  |        \_ make build-check-p
 1826 ? S  0:00  |            \_ /bin/sh -c for test in t_mpi testphdf5 t_cache
t_pflush1 t_pflush2 t_pshutdown t_prestart t_init_term t_shapesame dumm
 1924 ? S  0:00  |                \_ make testphdf5.chkexe_ RUNEXEC=mpiexec -n
6
 1925 ? S  0:00  |                    \_ /bin/sh -c if test
"Xtestphdf5.chkexe_" != "X.chkexe_" && test "Xtestphdf5.chkexe_" !=
"Xdummy.chkexe_"; then 
 1927 ? Sl 0:00  |                        \_ mpiexec -n 6 ./testphdf5
===
[michel@tw422trial1:~/work/openSUSE:Factory:Staging:B:DVD/hdf5]
$sudo strace -o /tmp/ptrace_mpi_hdf5.log -f -p 1927
Process 1927 attached with 2 threads
^CProcess 1927 detached
Process 1928 detached
... 
$cat /tmp/ptrace_mpi_hdf5.log
1928  select(12, [10 11], NULL, NULL, {829, 885201} <unfinished ...>
1927  restart_syscall(<... resuming interrupted call ...> <detached ...>
====


You are receiving this mail because: