#39 build problem tests.x failed
Hello, when trying to build pFUnit with OpenMPI and gfortran, I get the following error:
[miahw@localhost pFUnit]$ make tests MPI=YES
...
mpif90 -o remote.x RemoteProgram.o MakeNaN.o Test_RobustRunner.o MakeInfinity.o Test_Assert.o BrokenSetUpCase.o Test_TestMethod.o robustTestSuite.o Test_TestResult.o Test_SimpleTestCase.o Test_AssertBasic.o Test_MockRepository.o Test_FixtureTestCase.o Test_AssertComplex.o Test_StringConversionUtilities.o Test_MpiParameterizedTestCase.o Test_MpiContext.o Test_UnixProcess.o Test_MpiException.o BrokenTestCase.o Test_BasicOpenMP.o FixtureTestCase.o Test_Exception.o Test_XmlPrinter.o Test_TestSuite.o Test_MockCall.o Test_AssertReal.o SimpleTestCase.o Test_AssertInteger.o MockListener.o Test_MpiTestCase.o ../source/libpfunit.a -I/home/miahw/pFUnit/include -I/home/miahw/pFUnit/source -g -O0 -fbacktrace -fbounds-check -fcheck=mem -I../include -I../source -L../source -lpfunit -g
make[1]: Leaving directory '/home/miahw/pFUnit/tests'
mpirun -np 4 ./tests/tests.x
.....F.................................................................................................................................................................................................
Time: 0.673 seconds
Failure in: StringConversionUtilities.testIsActive
Location: [<unknown location="">]</unknown>
huh
FAILURES!!!
Tests run: 198, Failures: 1, Errors: 0
ERROR STOP
ERROR STOP
Note: The following floating-point exceptions are signalling: IEEE_INVALID_FLAG
ERROR STOP Encountered 1 or more failures/errors during testing.
Primary job terminated normally, but 1 process returned
a non-zero exit code.. Per user-direction, the job has been aborted.
mpirun detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:
Process name: [[40875,1],1]
Exit code: 1
GNUmakefile:222: recipe for target 'tests' failed
make: *** [tests] Error 1
Any help will be greatly appreciated.
Regards
First, the good news: This test failure is harmless for routine usage of pFunit. So, you should be able proceed with 'make install' and start using the framework.
To help us further troubleshoot this issue, could you provide us with the precise version of gfortran, OpenMPI, and OS that you are using? Our regression tests are not showing any problems with this specific routine.
In case you are wondering, this layer is used to support "robust" running which launches a second copy of the executable. The first copy monitors the second and can then detect when processes hang or die. At this time, there are some subtle race conditions that happen on some systems, but we cannot reliably reproduce them in our environment. Fortunately, you can get similar information, albeit in a less elagant/robust manner, by running with the "-d" command line option which prints each test name as it starts. (Hence my comment that you probably would not be impacted by this test failure.)
Cheers,
Hello Tom,
Thanks for the quick reply. Below are the details:
Do let me know if you require further information.
Hello,
I’ve set up an environment that mimics the details below.
I’ve been building the system different ways, and checking to see if not make distclean’ing in between builds might trigger the fault. In a couple dozen builds I got one runtime error—it was in UnixProcess.F90, the same place miahw’s build hit a fault. It’s not a deterministic thing though. Hard to reproduce. I think in my case pFUnit didn’t understand what the OS communicated to it when we popen’ed a process and got a line in newProcess.
A while back, if MPI was slow to start I might get a fault in the first test, which would go away on subsequent builds. That was about a year ago…
So no joy yet.
The IEEE_INVALID_FLAG miahw found is a curiosity...
miahw: Which branch are you using? Master? Or perhaps you’re working from the tar file (which is the master branch).
Cheers,
Mike
--
Michael Lee Rilee, Ph.D.
Rilee Systems Technologies LLC
Contact: mike@rilee.net; Phone: 703-348-8127; Cell: 240-481-3254; Fax: 703-880-7202
On Jan 5, 2015, at 3:53 PM, miahw wadud@users.sf.net wrote:
Related
Bugs:
#39The contents of the VERSION file has:
pFUnit 3.0.2 master
which was contained in the tar file. Do let me know if you require any further information.
Thanks for verifying that. Is this failure intermittent or consistent?
Sent from my iPhone
Related
Bugs:
#39