Hi,
Have a look at the job_submit plugin ;) Le 19/02/2015 15:43, Brian B a écrit :
Greetings, Our cluster is used for a variety of different processes but some are extremely IO intensive and can cause the whole file system to hang if too many are running at once. I created a “bandwidth” GRE that, when properly used solves the problem. Unfortunately not all users are reading the instructions and requesting the proper bandwidth. Is there a way to set a default GRE request? Be it in the partition, user, account, etc. I do not care. Similarly can one make a default Memory request? The only default I have found is DefaultTime. http://slurm.schedmd.com/slurm.conf.html Regards, Brian
-- --- Mehdi Denou International HPC support +336 45 57 66 56
