[GE users] rsh zombies when using mpich2 -- johnny layne

Johnny Layne laynejg at vcu.edu
Wed Aug 1 15:53:25 BST 2007


hi reuti,
>
> to see the dependencies you can use:
>
> ps f -eo pid,ppid,pgrp,command --cols=120
>
> So the processgroup would be the third column, and you could check 
> this way, whether all are the same for the started application.
Well, I'm trying to think down the road when I might want something I 
can easily parse to check process group IDs. Just playing with it at the 
moment.
>
> Why are you using a wrapper scxript for the program? I mean, instead 
> of generating and executing the wrapper script, the line:
>
> ~/script.$port
>
> could just be replaced with:
>
> /usr/global/vasp-p/mpichvasptest/vasp.4.6/vasp
>
> Did you encounter any error with it? - Reuti
Yes, I tried just doing everything from the directory that I was in by 
depending on "#$ -cwd" but then it seems that the mpiexec wants to begin 
work from my home directory, so I generate the little script to cd to 
the working directory & then launch VASP. Otherwise I get VASP errors 
complaining about not finding files etc., so I guess at that point the 
work is in the hands of mpiexec & then VASP. Perhaps there's another 
argument I need to pass to mpiexec, and I'm totally missing something.

Well, oops! Just looked at the man page for this mpiexec, and I see 
there's a "-wdir" argument you can use. I'll try that & see how I like 
it. Though hrmm, I seem to recall trying that before.

OK back to hunting down the mysterious zombie rsh's. If it ever happens 
again...
johnny

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list