[gpaw-users] Test failure

Ask Hjorth Larsen asklarsen at gmail.com
Mon Aug 12 14:10:58 CEST 2019


Hi,

Am Fr., 9. Aug. 2019 um 20:28 Uhr schrieb Jay Wai via gpaw-users
<gpaw-users at listserv.fysik.dtu.dk>:
>
> Hello all,
>
> I’ve just installed gpaw-19.8.1 with scalapack and fftw into a CentOS machine.
> There were no error or warning messages during the installation, but I have difficulties in figuring out what causes the following problems in after-installation steps:
>
> 1. ‘gpaw test’ run stops in the lcao/lcao_projections.py part showing a weird message ‘killed’ as follows:
> pw/fulldiagk.py                               3.026  OK
> ext_potential/external.py                     2.956  OK
> ext_potential/external_pw.py                  4.194  OK
> lcao/atomic_corrections.py                    0.000  SKIPPED
> vdw/libvdwxc_h2.py                            0.000  SKIPPED
> generic/mixer.py                              2.478  OK
> lcao/lcao_projections.py                 Killed
>
> Is there any case in which gpaw internally kill the test process? Or should I ask the system manager?

GPAW won't kill it, so this must be some other program.

Try reproducing the error using only that file, by finding it (gpaw
test --list | grep lcao_proj) and running it manually with mpirun -np
N gpaw-python thefile.py.

Then see if it fails with or without MPI, and with 1, 2, 4 processes.

Best regards
Ask

>
> 2. ‘gpaw -P 4 test’ stops right away showing following error messages
> gpaw-python: symbol lookup error: /apps/compiler/gcc/7.2.0/lib/libmca_common_verbs.so.7: undefined symbol: ompi_common_verbs_usnic_register_fake_drivers’
> Primary job  O-direction, the job has been aborted.
> mpiexec 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: [[13827,1],0]
>   Exit code:    127
>
> Openmpi 3.x are installed in the system. Does that message something to do with the compile setting of openmpi?
>
> I have struggled with this problem for a few days.
> It would be grateful if someone could help me on this.
> -Jay
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users



More information about the gpaw-users mailing list