Ah, Ross raises a good point. I hadn't thought enough about what was
happening here.
What you did was simply start 2 copies of restart.sh, each of which is
launching a 'sander.MPI' executable. Thus, you have 2 sander.MPI
executables each using 1 thread, and each one is trying to write to exactly
the same files. This is why you get the "cannot open for read" and "cannot
open for write", because you have 2 processes clashing with each other
simultaneously. Thus, it's not a permissions issue. See Ross's mail for
suggestions/fixes (I was simply explaining your error messages).
HTH,
Jason
On Wed, May 25, 2011 at 12:42 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
> Hi Bruno,
>
> > The cluster administrator claimed that my ./restart.sh should be a
> > BINARY
> > file because mpirun only accepts binaries.
>
> This is not strictly true in the real sense of the word BUT you should NOT
> be passing a script as an argument to mpirun. From your comment above I
> would conclude that you are trying something like:
>
> mpirun -np 2 ./restart.sh
>
> This will definitely NOT work. The mpirun needs to be inside your
> restart.sh
> script in front of the pmemd.MPI or sander.MPI call.
>
> > I guess it's not the case as it recognizes the commands inside. Do you
> > have
> > any clue about the input files for mpirun? I read the man from mpirun
>
> Sure it recognizes the commands inside but you should probably read up a
> bit
> on what MPI actually is and you will realize what is going on here and why
> this does not make sense. If you just do mpirun -np 2 ./restart.sh it will
> simply run 2 copies of your restart.sh script and this will just end in a
> big steaming pile of FAIL!
>
> Edit your restart.sh and put the mpirun command in front of the sander.MPI
> or pmemd.MPI executable and then run the restart script with ./restart.sh.
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> ---------------------------------------------------------
> | Assistant Research Professor |
> | San Diego Supercomputer Center |
> | Adjunct Assistant Professor |
> | Dept. of Chemistry and Biochemistry |
> | University of California San Diego |
> | NVIDIA Fellow |
> | http://www.rosswalker.co.uk | http://www.wmd-lab.org/ |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> ---------------------------------------------------------
>
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
>
>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 25 2011 - 10:30:03 PDT
