[mpich2-dev] Compiling ROMIO on BG/P system

fjblas fjblas at arcos.inf.uc3m.es
Thu Oct 9 16:34:52 CDT 2008


Hi

Could someone help me with a ROMIO compilation problem?

I am trying to compile ROMIO distribution on a BG/P (ANL). The compilation
process works fine but fails on runtime.

I submit a job that runs the "coll_test" example of ROMIO's test suite. I
have compiled with two different ROMIO 
versions and with two MPI wrappers, gcc and IBM. In both cases, the
compilation works but fails in runtime.

This is my configure string:


./configure --with-file-system=nfs+ufs+testfs+pvfs2
--with-pvfs2=$HOME/pvfs2 --enable-f77 --enable-f90 CC=mpicc F77=mpif77
FC=mpif90m CFLAGS=" -DUSE_MPI_VERSIONS -g -I$HOME/pvfs2/include"
LIBS="-L$HOME/pvfs2/lib -lpvfs2 -L/usr/lib -lcrypto -lrt -lm"



The error log whem I submit the job to the queue:

<Oct 09 15:55:21.972902> FE_MPI (Info) : Starting job 71231
<Oct 09 15:55:22.072584> FE_MPI (Info) : Waiting for job to terminate
<Oct 09 15:55:23.685186> BE_MPI (Info) : IO - Threads initialized
<Oct 09 15:55:23.715849> BE_MPI (Info) : I/O input runner thread terminated
<Oct 09 15:55:33.762575> BE_MPI (Info) : I/O output runner thread
terminated
<Oct 09 15:55:34.098873> BE_MPI (Info) : Job 71231 switched to state
TERMINATED ('T')
<Oct 09 15:55:34.098929> BE_MPI (Info) : Job successfully terminated -
TERMINATED ('T')
<Oct 09 15:55:34.422714> FE_MPI (Info) : Job terminated normally
<Oct 09 15:55:34.422848> FE_MPI (Info) : exit status = (135)
<Oct 09 15:55:34.423016> BE_MPI (Info) : Starting cleanup sequence
<Oct 09 15:55:34.423053> BE_MPI (Info) : cleanupDatabase() - job already
terminated / hasn't been added
<Oct 09 15:55:34.707018> BE_MPI (ERROR): The error message in the job
record is as follows:
<Oct 09 15:55:34.707056> BE_MPI (ERROR):   "killed with signal 7"
<Oct 09 15:55:34.707089> BE_MPI (Info) : Destroying partition
ANL-R00-M1-N10-64


Finally, I show the result of run bgp_stack for one core file:

bgp_stack coll_test core.0

0x0122b1c4
memset
??:0

0x01205078
generic_start_main
darwin-ldouble.c:0

0x01205324
__libc_start_main
??:0

0xfffffffc
??
??:0




Thanks a lot
Javi






More information about the mpich2-dev mailing list