[gmx-users] The right way of killing Gromacs

[Erik Lindahl]

Hi,

mdrun finishes on the next step when receiving TERM, and on the next  
multiple of nstxout when receiving USR1.

On some systems, the USR1/SIGHUP is only sent to the parent MPI  
program, not the actual mdrun process, and then there isn't much we  
can do.

In your case, try altering the signals in src/kernel/md.c and see if  
you get it working if you change both signals to cause an exit on the  
next step.

Cheers,

Erik


On Sep 15, 2006, at 11:33 AM, kobi wrote:

> Hello,
>
> the runtime of our cluster is unfortunately limited. Before mdrun  
> is terminated a USR1 signal is
> send, but this doesn't cause mdrun to writeout the current frame.  
> Therefore most of the trajectory
> is lost. I'd like to write a skrip which converts the USR1 sinal to  
> some other signal. The question
> is, which signal would cause mdrun to write out all data and  
> terminate it afterwards?
>
> best regards
> Jan Neumann
> _______________________________________________
> 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