[Bug 1172555] New: python-control hang on testSiso for PowerPC
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 Bug ID: 1172555 Summary: python-control hang on testSiso for PowerPC Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: Other Assignee: screening-team-bugs@suse.de Reporter: normand@linux.vnet.ibm.com QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Created attachment 838525 --> http://bugzilla.opensuse.org/attachment.cgi?id=838525&action=edit python_control_standard_ppc64le_202006021900.log python-control hang on testSiso for PowerPC as reported in https://build.opensuse.org/package/show/openSUSE:Factory:PowerPC/python-cont... === extract attached python_control_standard_ppc64le_202006021900.log [ 225s] Test h2syn ... ok [ 225s] testHinfsyn (control.tests.robust_test.TestHinf) [ 226s] Test hinfsyn ... ok [ 226s] testSiso (control.tests.robust_test.TestMixsyn) [12229s] mixsyn with SISO system ... qemu-system-ppc64: terminating on signal 15 from pid 31047 () === -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c1 --- Comment #1 from Michel Normand <normand@linux.vnet.ibm.com> --- Created attachment 838526 --> http://bugzilla.opensuse.org/attachment.cgi?id=838526&action=edit python_controls_twppc64le2_passed_build.log If I do a local osc build in a ppc64le guest, then no failure is reported as per attached python_controls_twppc64le2_passed_build.log related testSiso last almost 1000s but all osc build completed with no hang. === extract python_controls_twppc64le2_passed_build.log [ 1256s] testHinfsyn (control.tests.robust_array_test.TestHinf) [ 1262s] Test hinfsyn ... ok [ 1262s] testSiso (control.tests.robust_array_test.TestMixsyn) [ 2153s] mixsyn with SISO system ... ok === I do not know how to investigate such failure. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c2 Benjamin Greiner <code@bnavigator.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |CONFIRMED CC| |normand@linux.vnet.ibm.com Flags| |needinfo?(normand@linux.vne | |t.ibm.com) --- Comment #2 from Benjamin Greiner <code@bnavigator.de> --- I believe for some reason the hinfsyn function is not finding a stabilized controller in time for this test: https://github.com/python-control/python-control/issues/367 Do you know if the tests passed prior to the python-slycot update? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c3 Benjamin Greiner <code@bnavigator.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags| |needinfo? --- Comment #3 from Benjamin Greiner <code@bnavigator.de> --- I tried to bypass the test in https://build.opensuse.org/package/show/home:bnavigator:branches:devel:langu... but the PowerPC platform seems to have bigger problems. The tests of the python-slycot package fail unpredictably with wrong numerical results in ppc64 and ppc64le. Every rerun yields different array mismatches. I had disabled those platforms in the devel package because of this. This affects the python-control package because it makes heavy use of the Slycot routines. I suspect there exist LAPACK library or Fortran compiler problems on PowerPC. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c4 Michel Normand <normand@linux.vnet.ibm.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(normand@linux.vne | |t.ibm.com) | --- Comment #4 from Michel Normand <normand@linux.vnet.ibm.com> --- (In reply to Benjamin Greiner from comment #2)
[CUT] Do you know if the tests passed prior to the python-slycot update?
as per osc jobhist of openSUSE:Factory:PowerPC there were already testSiso hang before python-slycot update. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 Benjamin Greiner <code@bnavigator.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Hardware|Other |PowerPC See Also| |https://bugzilla.opensuse.o | |rg/show_bug.cgi?id=1148173 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c5 Benjamin Greiner <code@bnavigator.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|CONFIRMED |IN_PROGRESS --- Comment #5 from Benjamin Greiner <code@bnavigator.de> --- Possible fix by more reliably building python-slycot and skipping the mixsyn test. https://build.opensuse.org/request/show/817306 https://build.opensuse.org/request/show/817307 Still, it would be really great to have a PowerPC expert who can chime in about the precision issues with that platform. The pyhton-numpy package suffers from this, too. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1172555 http://bugzilla.opensuse.org/show_bug.cgi?id=1172555#c6 Benjamin Greiner <code@bnavigator.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|IN_PROGRESS |RESOLVED Resolution|--- |WORKSFORME --- Comment #6 from Benjamin Greiner <code@bnavigator.de> --- Closing as of the last few weeks the build passes most of the time. If you encounter again and/or find the reason (most likely an underlying library: numpy, scipy, lapack, blas) please reopen or file a new bug. -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠suse.com