On Mon, Mar 29, 2010 at 5:36 PM, Abhishek Gupta <[email protected]> wrote: > Let me ask my question in more detail. > We want only a certain group to use node 1-7 and so if there is any job > submitted, it should first look into remaining nodes, and then in case > all the remaining nodes( other than 1-7 ) are in use, jobs can run on > 1-7 by the other users too. > nodes 1-7 are also different from rest of the nodes in terms of > processors and other resources. > Could anyone suggest me know what would be the right approach to achieve > that?
if the nodes are significantly different, then i would just create two separate queues for each set of nodes and set ACLs so that one group can only submit to one, while the other can submit to both (or only the other part). everything more complex is a waste of your time. KISS! i would just run this simple setup for a while, collect usage data and then decide based on that whether there are ways and a need to improve the scheduling and utilization. axel. > Thanks, > Abhi. -- Dr. Axel Kohlmeyer [email protected] http://sites.google.com/site/akohlmey/ Institute for Computational Molecular Science Temple University, Philadelphia PA, USA. _______________________________________________ mauiusers mailing list [email protected] http://www.supercluster.org/mailman/listinfo/mauiusers
