lllll

Articles

Each job is submitted to a specific queue (the default one is bioinfo.q).

The scheduler attemps to divide the cluster resources equally among all users. If, for example, one user has many jobs already queued (and running) and then another user queues new jobs, those new jobs will rise in priority above some of the first user's jobs until the number of running jobs is approximately shared equally between the two users.  

The following table details the various queues and their caracteristics

Queues

Queue name

Purpose

Available node 

Node caracteristics

bioinfo.q

General use & MPI

node 0 to 2

node 8 to 11 

2 six-core intel 2.4 Ghz, 48 Go RAM

longjob.q

For longer and numerous jobs

> 1 day each, > 10 jobs

node 0 to 1

2 six-core intel 2.4 Ghz, 48 Go RAM

bigmem.q

For job with heavy memory usage requiring

node 3

2 six-core intel 2.4 Ghz, 96 Go RAM 

highmem.q

For job with heavy memory usage requiring

node 4, 7

2 six-core intel 2.4 Ghz, 144 Go RAM

  r900.q Restricted to Galaxy job  node 5   4 quad-core intel 2.93 Ghz, 32 Go RAM
    
 
 

 

 

 

Copyright © 2019 IRD Bioinformatics. All Rights Reserved.
Joomla! is Free Software released under the GNU General Public License.