Fwd: [gmx-users] mdrun on several clusternodes using PME gromacs 3.3.1

[Carsten Kutzner]

Hi Joern,

I would guess that's the old GROMACS+mpich-1/ch_p4 problem. I have also
encountered it a number of times. Runs fine on one or two nodes, but
fails with a p4_error on more nodes.

What version of mpich are you using? If it is mpich-1.2.x, try  mpich-2,
this solved the problem for me. You could also use LAM or OpenMPI, which
(as mpich-2) runs fine with GROMACS.

Carsten


Joern Lenz wrote:
> thank you lars and sorry guys, 
> i forgot to let you know about another outputfile concerning the problem
> discussed above:
> lars: i submitted my jobs with an mdrun_mpi command to our cluster with:
> "qsub run_the_simulation.sh" and in the sh script the command is 
> 
> grompp -np $NODES -n $1_ion -f pr_mdrun01 -c $1_b4pr_mdrun01 -r 
> $1_b4pr_mdrun01 -p $1 -o $1_pr_mdrun01 2>&1 > output.grompp_pr_mdrun01
> 
> /opt/mpich/ch-p4/bin/mpirun -machinefile ~/machines.lst -np $NODES 
> -v /.../${CONFIGGUESS}/bin/mdrun_mpi -np $NODES -s $1_pr_mdrun01.tpr -o 
> $1_pr_mdrun01 -c $1_pr_mdrun01 -v -e $1_pr_mdrun01 2>&1 > 
> output.mdrun_pr_mdrun01
> 
> where $NODES is 16 and CONFIGGUESS guesses the architecture of the system (32 
> or 64 bit)
> 
> i dont know if nohup and/or & will change anything about the problem.
> 
> SOME FURTHER INFORMATION OF THE PROBLEM:
> 
> running /../i686-pc-linux-gnu/bin/mdrun_mpi on 2 LINUX ch_p4 processors
> Created /../PI13506
> /home/blah
> p0_13586: (5921.985420) net_recv failed for fd = 6
> p0_13586:  p4_error: net_recv read, errno = : 104
> 
> 
> perhaps this gives you further information.
> i tried nearly everything and also discussed the problem with our
> system-administrators... no ideas to solve ...
> the thing only works if I run it local, but this is very time-consuming
> 
> please help me with suggestions or i will die unhappy
> thank you so much
> joern
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

-- 
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics Department
Am Fassberg 11
37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/research/dep/grubmueller/
http://www.gwdg.de/~ckutzne