[GE users] MPICH and Signals

Reuti reuti at staff.uni-marburg.de
Wed Oct 27 21:44:24 BST 2004

    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Quoting "Brian R. Smith" <brian at cypher.acomp.usf.edu>:

> I did set P4_RSHCOMMAND to /usr/local/sge/mpi/rsh BUT that was AFTER
> setting it to "rsh" yielded no results.  I will try to set it to "rsh"
> again though.  As for my submit script, It works fine.  I have no
> problems with using -nolocal as I do not want processes running on my
> master node.  I can submit jobs similar to an more complex that the
> source you provided.  The problem comes when I want to clean those jobs
> up before they finish themselves.

The -nolocal will mean the master node of the job - hence the first slave node 
in the list of nodes SGE selected. If you get two slots on one machine you 
should get an error message like "Not enough machines for platform: LINUX" - 
because there are no machines left in the list.

Just got your next eMail: can you please make a:

ps f -eo pid,uid,gid,user,pgrp,command --cols=120

during execution of the job on the head node and after the kill?

CU - Reuti

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