ICC Home Privacy and Legal Notice LC User Documents Banner


SLURM Reference Manual

Scheduler Types

The system administrator for each machine can configure SLURM to invoke any of several alternative local job schedulers. You can discover which scheduler SLURM currently invokes on any machine by executing

scontrol show config | grep SchedulerType

where the returned string will have one of these values:

(default) is a first-in-first-out scheduler. SLURM executes jobs strictly in the order that they were submitted (for each resource partition). Even if resources become available to start a specific job, SLURM will wait until no previously submitted job is pending (which sometimes confuses impatient job submitters).
modifies strict FIFO scheduling to take advantage of resource islands that may appear as earlier jobs complete. SLURM will start later-submitted jobs out of order if resources become available and if doing so does not delay the expected execution time of any earlier-submitted job. You increase your job's chances of benefiting from such backfill scheduling if you:
(1) specify reasonable time limits (the default is the same time limit for all jobs in your partition, perhaps much too large), and
(2) avoid requiring or excluding specific nodes by name.
BlueGene/L Note: backfill scheduling never occurs on BlueGene/L because that machine's unusual and complex architecture thwarts reliably finding left over resource islands to fill.
uses the Maui Scheduler, with a sophisticated set of internal scheduling algorithms. This choice can be configured in many ways to optimize job throughput. Details are posted on a support web site at http://supercluster.org/maui.

Navigation Links: [ Document List ] [ HPC Home ] [ Previous ] [ Next ]