[Condor-users] Problems Condor and MPI

[OMAIRA GALINDO <maggy128@xxxxxxxxx>]

Hi all

1. I'm working condor and MPI, on some servers that have two quad-core processors each, so condor recognizes about 16 slot per machine, my problem is this when I try to send a job to run in parallel universe not can do that because MPI does not allow more than one daemon running mpd per machine, and I want to take advantage of the processor and each core so that each can carry out part of the job, but it is not possible for such restrictions on MPI.

slot1@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:04
slot2@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:05
slot3@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:06
slot4@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:07
slot5@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:08
slot6@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:09
slot7@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:10
slot8@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle     0.000  2011  0+00:00:03


If anyone else has happened and know how to solver it, I would appreciate very much.

2. I wonder if the checkpoints work with the parallel universe is that I did some tests and it seems not.


Thank you, goodbye

-- 
Omaira Galindo Parra
Estudiante de Ingeniería de Sistemas y Computación
UPTC Tunja