[Pw_forum] Soft exit signal has no effect (NSCF jobs will be killed by the queuing system)

Kiss, Ioan kissi at uni-mainz.de
Mon Feb 20 17:21:21 CET 2012

Dear PWSCF users and developers,

I have a question regarding the NSCF calculations with Espresso 4.3.2.
Namely, is it possible to stop the NSCF calculation in a controlled fashion
during a band structure calculation when the code goes over a list of user
defined points in the k-space before generating the spaghetti plot?

In the computer-center the maximum runtime is only 24 hours, and during
such short runtime the NSCF job will not be finished. Thus, the execution of the
code will be killed by the queuing system about 10 minutes after the maximum
runtime was expired.
Due to this the files generated by PWSCF are most often corrupted and I
cannot use them to restart the NSCF calculation. I thought that creating the
$prefix.EXIT file in the run directory, or in the directory where the binary is located,
or defining the maximum runtime with the  max_seconds in the input file should
properly stop the execution of the code.
However, the code does not stop during NSCF calculations independent on the
parallel environment (tested on PBS/OpenMPI, LSF/mvapich, loadleveler/IBM poe)
no matter where I create the $prefix.EXIT file or what kind of value I define as the
max_seconds variable.
For the SCF calculations I have tested the soft exit on the same systems and it
worked flawlessly, pw.x did properly exit after writing out all files necessary for restart.
Therefore, it would be really helpful if someone could tell me what to do to get
a similar behavior in case of NSCF calculations as well.

I would like to thank you in advance for any helpful comment or idea.

Best regards,
Janos Kiss.

  Dr. Janos Kiss      e-mail: kissi at uni-mainz.de
  Johannes Gutenberg-Universitaet
  Institut f. Anorg. u. Analyt. Chemie
  AK Prof. Dr. Claudia Felser
  Staudinger Weg 9 / Raum 01-230
  55128 Mainz/ Germany
  Phone: +49-(0)6131-39-22703
  Fax:     +49-(0)6131-39-26267
  Web:     http://www.superconductivity.de/

More information about the users mailing list