[gmx-users] Segmentation fault error from mdrun

[Mark Abraham]

On 8/12/2011 7:36 AM, rainy908 wrote:
> Hi,
>
> I encounter the following error when trying to execute mdrun:
>
> # Running Gromacs: read TPR and write output to /gpfs disk
>   $MPIRUN  $MDRUN -v -nice 0 -np $NSLOTS \
>   -s n12_random_50_protein_all.tpr \
>   -o n12_random_50_protein_all.trr \
>   -c n12_random_50_protein_all.gro \
>   -g n12_random_50_protein_all.log \
>   -x n12_random_50_protein_all.xtc \
>   -e n12_random_50_protein_all.edr

You can save yourself some typing with mdrun -deffnm. Also note that 
mdrun -np does nothing for GROMACS 4.x.

>
> Error:
>
> [compute-0-7:12377] Failing at address: 0x7159fd0
> [compute-0-30:07435] [ 1] mdrun [0x761971]
> [compute-0-30:07435] *** End of error message ***
> [compute-0-29:15535] [ 0] /lib64/libpthread.so.0 [0x39df60e7c0]
> [compute-0-29:15535] [ 1] mdrun [0x761d60]
> [compute-0-29:15535] *** End of error message ***
> [compute-1-29:19799] [ 0] /lib64/libpthread.so.0 [0x33aac0e7c0]
> [compute-1-29:19799] [ 1] mdrun [0x762065]
> [compute-1-29:19799] *** End of error message ***
> [compute-0-29:15537] [ 0] /lib64/libpthread.so.0 [0x39df60e7c0]
> [compute-0-29:15537] [ 1] mdrun [0x762065]
> [compute-0-29:15537] *** End of error message ***
> [compute-0-29:15536] [ 0] /lib64/libpthread.so.0 [0x39df60e7c0]
> [compute-0-29:15536] [ 1] mdrun [0x762065]
> [compute-0-29:15536] *** End of error message ***
> [compute-1-31:11981] [ 0] /lib64/libpthread.so.0 [0x374f00e7c0]
> [compute-1-31:11981] [ 1] mdrun [0x761d60]
> [compute-1-31:11981] *** End of error message ***
> [compute-1-31:11982] [ 0] /lib64/libpthread.so.0 [0x374f00e7c0]
> [compute-1-31:11982] [ 1] mdrun [0x761960]
> [compute-1-31:11982] *** End of error message ***
> [compute-0-29:15538] [ 0] /lib64/libpthread.so.0 [0x39df60e7c0]
> [compute-0-29:15538] [ 1] mdrun [0x761960]
> [compute-0-29:15538] *** End of error message ***
> [compute-0-7:12377] [ 0] /lib64/libpthread.so.0 [0x387c60e7c0]
> [compute-0-7:12377] [ 1] mdrun [0x729641]
> [compute-0-7:12377] *** End of error message ***
> [compute-1-29:19796] [ 0] /lib64/libpthread.so.0 [0x33aac0e7c0]
> [compute-1-29:19796] [ 1] mdrun [0x762065]
> [compute-1-29:19796] *** End of error message ***
> [compute-1-31.local][[50630,1],32][btl_tcp_frag.c:216:mca_btl_tcp_frag_recv] mca_btl_tcp_frag_recv: readv failed: Connection reset by peer (104)
> --------------------------------------------------------------------------
> mpirun noticed that process rank 35 with PID 32477 on node compute-1-8.local exited on signal 11 (Segmentation fault).
> --------------------------------------------------------------------------
>
> This is a parallel job that caused segmentation fault on compute-1-8, thus causing the entire job to fail.

You need to look at stderr, stdout (some of which are above) and the 
.log file to find out what GROMACS thought caused the crash. You also 
need to use an mpi-enabled mdrun.

Mark