[GE users] Long delay when submitting large jobs

Rayson Ho raysonho at eseenet.com
Mon Feb 7 20:13:21 GMT 2005


    [ 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. ]

>Perhaps I'm missing something, but isn't the whole point of 
>tight-integration to allow SGE to control the slaves, and thus get 
>accounting information and be able to properly shut them down, 
>etc? It sounds to me like what you're suggesting would break tight-
>integration.

No, it won't break tight integration.

When mpiexec uses TM call to start the parallel slave tasks, it will
contact the remote execds (equ to MOMs in a PBS cluster).

Currently, it's qmaster who contacts the execds, and when we integrate
mpiexec into SGE, mpiexec will do the work. But the net effect is the same,
the tasks are started by SGE.

So the work for us to do is to create a TM library for SGE...

Rayson

---------------------------------------------------------
Get your FREE E-mail account at http://www.eseenet.com !

---------------------------------------------------------------------
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