[Mauiusers] Maui's Suspend and Resume Functionality

Gerson Galang gerson.sapac@gawab.com
Tue, 13 Jul 2004 15:13:42 +0930


Hi,

I just want to know if anyone in the mauiusers list has gotten Maui's 
suspend and resume functionality to work on a cluster running MPI 
(mpiexec) jobs.

What's happening at the moment is that whenever I suspend the job using 
Maui's "mjobctl -s" command, the process on the first node is the only 
one getting suspended and the rest of the nodes still continues to run 
the MPI job. The server tells me that the job has already been suspended 
but if you monitor the compute nodes, you will see that that particular 
job that I submitted still uses up 98% of the nodes' computing power.

Do I need to do any special configuration with my Maui installation to 
get this functionality to work? I'm using Maui 3.2.6 patch 6.

Thanks,
Gerson

-- 
Gerson Galang
Research Programmer

South Australian Partnership for Advanced Computing
School of Physics
The University of Adelaide
Adelaide 5005
SA, AUSTRALIA

Phone:  61 8 8303 3185
Email: gerson.galang@adelaide.edu.au