PGFIO-F-/list-directed write/ error code returned by host...

I have a problem in running my F90 code in an opteron cluster tha consists 36 dual-opteron nodes, with in total 72 AMD opteron processors.

Compiling WITHOUT any flag:
pgf90 run.f -o run.out

I randomly and not always get this error message:

PGFIO/stdio: No such file or directory
PGFIO-F-/list-directed write/unit=1999/error code returned by host stdio - 2.

This has happened with 2 different sources which have run fine on other platforms.
The error seems to kick in at random and referr to different output files.
I’ve found other people in the web with the same problem

http://www.chpc.utah.edu/apps/helpdot/showpr.php?pr=2908

and

http://www.students.ncl.ac.uk/a.j.mee/blog/index.php?paged=3

but no solution!

do you have any suggestion?

thank you,
Chantal

Where is this file located? Is it local on the node or is it on an NFS mount? Does the same error ever occur if you run on a single node?

The error message says that your IO is failing because a signal 2 (INTERRUPT). This plus that fact that it seems to happen randomly suggests a problem with NFS.

  • Mat