Useful sbatch Options

Usually, you can parse arbitrary additional options to Python submit scripts (additional to the script’s own options). These options will be parsed to the sbatch command that submits the Slurm job scripts.

This page contains a subset of sbatch options you might consider useful. Refer to the documentation of sbatch for a full list of all options.

You can set defaults for all sbatch options in the Configuration File. This is especially useful for options like --account, --mail-type, --mail-user or --partition.


Charge resources used by this job to the specified account (more details).


Defer the allocation of the job until the specified time (more details).


Set the working directory of the batch script before it is executed (more details).


Specify which features are required by this job (more details). You can use sinfo --format "%12P | %.15f" to see which features are available on which partition.


Number of CPUs per task (more details).


Defer the start of this job until the specified dependencies have been satisfied (more details). If a submit script submits multiple jobs that depend on each other, the given dependency only applies to the top-level job and the sub-level jobs only depend on the top-level job.


Explicitly exclude certain nodes from the resources granted to the job (more details).


Don’t share nodes with other jobs (more details).


Restrict node selection to nodes with at least the specified number of sockets, cores per socket and/or threads per core (more details).


Generic consumable resources (more details).


Submit the job in a held state (more details).


Specify a name for the job allocation (more details).


{“yes”, “no”}

Whether to terminate the job when it has an invalid dependency and thus can never run (more details).



Notify user by email when certain event types occur (more details).


User to receive email notification (more details).


Memory required per node. (more details).


Minimum number of CPUs per node (more details).


Specifies that the batch job should never be requeued under any circumstances (more details).


Number of nodes to allocate (more details).


Request a specific list of nodes (more details).


Number of tasks per node (more details).


The file to which to write the batch script’s standard output (more details).


Request a specific partition for the resource allocation (more details). You can use sinfo --summarize to get a list of all partitions available on your computing cluster.


Return an estimate of when a job would be scheduled to run. No job is actually submitted (more details).


Set a total run time limit (more details). You can use sinfo --summarize to get the maximum allowed run time limits for each partition on your computing cluster.


Set a minimum time limit. If specified, the job may have its --time limit lowered to a value no lower than --time-min if doing so permits the job to begin execution earlier than otherwise possible (more details).