Hi,

I'm having trouble with parallel I/O to a file system mounted with NFS
over an infiniband network.  In my test code, I'm simply writing 1 byte
per process to the same file.  When using two nodes, some bytes are not
written (zero bits in the unwritten bytes).  Usually at least some data
from each node is written---it appears to be all data from one node and
partial from the other.

This used to work fine but broke when the cluster was upgraded from
Debian 8 to Debian 9.  I suspect an issue with NFS and not with
OpenMPI.  However, if anyone can suggest a work-around or ways to get
more information, I would appreciate it.  In the sole case where the
file system is exported with 'sync' and mounted with 'hard,intr', I get
the error:
[node1:14823] mca_sharedfp_individual_file_open: Error during datafile
file open
MPI_ERR_FILE: invalid file
[node2:14593] (same)

----------

Some additional info:
- tested versions 1.8.8, 2.1.1, and 3.0.0 self-compiled and packaged and
vendor-supplied versions.  All have same behavior.
- all write methods (individual or collective) fail similarly.
- exporting the file system to two workstations across ethernet and
running the job across the two workstations seems to work fine.
- on a single node, everything works as expected in all cases.  In the
case described above where I get an error, the error is only observed
with processes on two nodes.
- code follows.

Thanks,
Stephen Guzik

----------

#include <iostream>

#include <mpi.h>

int main(int argc, const char* argv[])
{
  MPI_File fh;
  MPI_Status status;

  int mpierr;
  char mpistr[MPI_MAX_ERROR_STRING];
  int mpilen;
  int numProc;
  int procID;
  MPI_Init(&argc, const_cast<char***>(&argv));
  MPI_Comm_size(MPI_COMM_WORLD, &numProc);
  MPI_Comm_rank(MPI_COMM_WORLD, &procID);

  const int filesize = numProc;
  const int bufsize = filesize/numProc;
  char *buf = new char[bufsize];
  buf[0] = (char)(48 + procID);
  int numChars = bufsize/sizeof(char);

  mpierr = MPI_File_open(MPI_COMM_WORLD, "dataio",
                         MPI_MODE_CREATE | MPI_MODE_WRONLY,
MPI_INFO_NULL, &fh);
  if (mpierr != MPI_SUCCESS)
    {
      MPI_Error_string(mpierr, mpistr, &mpilen);
      std::cout << "Error: " << mpistr << std::endl;
    }
  mpierr = MPI_File_write_at_all(fh, (MPI_Offset)(procID*bufsize), buf,
                                 numChars, MPI_CHAR, &status);
  if (mpierr != MPI_SUCCESS)
    {
      MPI_Error_string(mpierr, mpistr, &mpilen);
      std::cout << "Error: " << mpistr << std::endl;
    }
  mpierr = MPI_File_close(&fh);
  if (mpierr != MPI_SUCCESS)
    {
      MPI_Error_string(mpierr, mpistr, &mpilen);
      std::cout << "Error: " << mpistr << std::endl;
    }

  delete[] buf;
  MPI_Finalize();
  return 0;
}

_______________________________________________
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

Reply via email to