mpi version failed when meshes' alignment like this

[GoogleCodeExporter]

Please complete the following lines...

Application Version:5.4.0
SVN Revision Number:4442
Compile Date:2009-07-23
Operating System:CentOS Linux 64bit/MPICH2

Describe details of the issue below:

The simplified version generated this message:

[fds@fds1 fds_test]$ mpirun -np 3 fds5_mpi_db test.fds
Process   0 of   2 is running on fds1
Process   2 of   2 is running on fds3
Process   1 of   2 is running on fds4
Mesh   1 is assigned to Process   0
Mesh   2 is assigned to Process   1
Mesh   3 is assigned to Process   2
rank 2 in job 5  fds1_53958   caused collective abort of all ranks
  exit status of rank 2: killed by signal 9

Before simplified,it will generate message like this:(similar to issue 487)

forrtl: severe (408): fort: (2): Subscript #1 of the array REAL_SEND_PKG4
has value 5378 which is greater than the upper bound of 5377

Image              PC                Routine            Line        Source
fds5_mpi_db        000000000190C0EE  Unknown               Unknown  Unknown
fds5_mpi_db        000000000190B080  Unknown               Unknown  Unknown
fds5_mpi_db        00000000018C2126  Unknown               Unknown  Unknown
fds5_mpi_db        0000000001876619  Unknown               Unknown  Unknown
fds5_mpi_db        0000000001875878  Unknown               Unknown  Unknown
fds5_mpi_db        000000000178FABF  fds_.mesh_exchang        1593 
main_mpi.f90
fds5_mpi_db        00000000016D6335  MAIN__                    334 
main_mpi.f90
fds5_mpi_db        0000000000404802  Unknown               Unknown  Unknown
libc.so.6          000000313521D8A4  Unknown               Unknown  Unknown
fds5_mpi_db        0000000000404729  Unknown               Unknown  Unknown
rank 3 in job 9  fds1_53958   caused collective abort of all ranks
  exit status of rank 3: killed by signal 9

Original issue reported on code.google.com by foolstone@gmail.com on 24 Jul 2009 at 7:42

Attachments:

• test.fds

[GoogleCodeExporter]

My 32 bit mpi debug version worked. I'll try 64 bit and let you know.

mpirun -np 3 ~/FDS_Re*/FDS_C*/mpi_intel_linux_32_db/fds5_mpi_intel_linux_32_db 
test.fds
Process   0 of   2 is running on fire61
Process   1 of   2 is running on fire62
Process   2 of   2 is running on fire63
Mesh   1 is assigned to Process   0
Mesh   2 is assigned to Process   1
Mesh   3 is assigned to Process   2

 Fire Dynamics Simulator

 Compilation Date : Wed, 22 Jul 2009
 Version          : 5.4.0 Parallel

 No OpenMP-Version

 SVN Revision No. : 4442

 Job TITLE        :
 Job ID string    : test

 Time Step:      1,    Simulation Time:      0.25 s
 Time Step:      2,    Simulation Time:      0.50 s
 Time Step:      3,    Simulation Time:      0.75 s
 Time Step:      4,    Simulation Time:      1.00 s
 Time Step:      5,    Simulation Time:      1.25 s
 Time Step:      6,    Simulation Time:      1.50 s
 Time Step:      7,    Simulation Time:      1.75 s
 Time Step:      8,    Simulation Time:      2.00 s
 Time Step:      9,    Simulation Time:      2.25 s
 Time Step:     10,    Simulation Time:      2.50 s
 Time Step:     20,    Simulation Time:      5.00 s
 Time Step:     30,    Simulation Time:      7.50 s
 Time Step:     40,    Simulation Time:     10.00 s

STOP: FDS completed successfully

Original comment by mcgra...@gmail.com on 24 Jul 2009 at 12:01

• Changed state: Started

[GoogleCodeExporter]

My 64 bit MPI debug version also worked. I use the Intel Fortran compiler, 
RedHat 
Fedora Core Linux, LAM/MPI. I do not have your same configuration, thus, it 
would be 
impossible to debug your case.

Original comment by mcgra...@gmail.com on 24 Jul 2009 at 5:51

• Changed state: WontFix