Fatal error

[simo2777]

Dear all,

I am trying to run code_Saturne, but I got a fatal error.
I run code_Saturne with a virtual machine.

MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode 1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------
--------------------------------------------------------------------------
mpiexec.openmpi has exited due to process rank 0 with PID 58707 on
node xxx-virtual-machine exiting improperly. There are two reasons this could occur:

1. this process did not call "init" before exiting, but others in
the job did. This can cause a job to hang indefinitely while it waits
for all processes to call "init". By rule, if one process calls "init",
then ALL processes must call "init" prior to termination.

2. this process called "init", but exited without calling "finalize".
By rule, all processes that call "init" MUST call "finalize" prior to
exiting or it will be considered an "abnormal termination"

This may have caused other processes in the application to be
terminated by signals sent by mpiexec.openmpi (as reported here).

Could anybody please tell me what is the reason of this error?
Is this because of the mesh?

Under the solution folder I found a error text message:

cs_file.c:1182: Fatal error.

MPI IO error for file: mesh_input
Error type: MPI_ERR_OTHER: known error not in list


Call stack:
1: 0x7f3ba45776c9 <+0x1c06c9> (libsaturne.so.0)
2: 0x7f3ba457b190 <+0x1c4190> (libsaturne.so.0)
3: 0x7f3ba457b380 <cs_file_open+0x150> (libsaturne.so.0)
4: 0x7f3ba458e7c2 <+0x1d77c2> (libsaturne.so.0)
5: 0x7f3ba458f167 <cs_io_initialize+0x87> (libsaturne.so.0)
6: 0x7f3ba4497599 <+0xe0599> (libsaturne.so.0)
7: 0x7f3ba449d8c5 <cs_preprocessor_data_read_mesh+0x285> (libsaturne.so.0)
8: 0x7f3ba4496bd3 <cs_preprocess_mesh+0x123> (libsaturne.so.0)
9: 0x7f3ba4402bb5 <cs_run+0x135> (libsaturne.so.0)
10: 0x7f3ba4402967 <main+0x147> (libsaturne.so.0)
11: 0x7f3ba4013ec5 <__libc_start_main+0xf5> (libc.so.6)
12: 0x4006f9 <> (cs_solver)
End of stack

Thankyou in advence for your help.
Simone

[simo2777]

Good evening,
I changed the parameters from 'Default' to 'Standard I/O, serial' under Calculation mamagement/Performance tuning/Input/output, both for read method and write method.
I don't know if it's the right solution, but in this way code_Saturne works.
Best regards.
Simone

[Yvan Fournier]

Hello,

It seems your MPI IO is buggy on that machine. Your solution is correct.
You could also configure Code_Saturne with the --disable-mpi-io option when installing it so you do not need to change the I/O option for each new computation.

What Linux distribution and MPI library are you using ?

Regards,

Yvan

[simo2777]

Bood morning.
Thank you for your reply.
I'm using the Ubuntu 14.10 64 bit distribuition on a WM-machine.
I installed code_saturne from the repository.
The workstation is a HP Z400.
I got this error only in this machine, I tried the same configuration in a different and more recent workstation, and it works well.
Maybe it's not the cause, but I also installed the last release of Openfoam following the wiki help
(https://openfoamwiki.net/index.php/Inst ... 3.0/Ubuntu), and in the point 5 there is the suggestion to create a new symbolic link for the openmpi to the openfoam mpi. I didn't have the time to investigate if this is the cause of the problem.
However, removing and installing again code_Saturne from the Ubuntu repository doesn't solve the problem.
I will try with the --disable-mpi option.
Best regards.
Simone

[simo2777]

Oh, I forgot to say that the Ubuntu 14.10 mpi version is 1.6.5-8.
Simone

[Yvan Fournier]

Hello,

OK, thanks for the info. Did you use any user subroutines in your case ? A quick test on a virtual machine some time ago on my side seemed to fail, due to the Ubuntu packing mixing compile options from the 32 and 64 bit variants (the upstream Debian package does not have this issue, and the unpackaged code definitely does not).

I did not file a bug report yet (just did a very quick test, and Ubuntu 10.14 is not my primary Linux environment).

Regards,

Yvan