...
For batch jobs, the default location for stdout and stderr is a file named slurm-
<jobid>
.out
in the working directory from which the job was submitted. (<jobid> is replaced by the numeric Slurm job ID.) To change the name of the standard output file, use the --output=<stdoutfile>
option. To change the name of the standard error file, use the --error=<stderrfile>
option. You can use the special token %j
to include the unique job ID in the filename.
...
Table 2. Common options
...
for sbatch
and salloc
...
Expand |
---|
title | Click here to show /hide. |
---|
| Table 2. Common options for sbatch
and salloc
|
Option | Reduced syntax | Purpose |
---|
--account=<project> | -A <project> | Set the project code to which the job is to be charged. A default project is configured for each Pawsey user. | --nodes=<N> | -N <N> | Request N nodes for the job. | --ntasks=<n> | -n <n> | Specify the maximum number oftasks or processes that each job step will run. On supercomputers with exclusive access to nodes, specify a multiple of the total number of cores available on a node for efficient use of resources. | --ntasks-per-node=<nN> |
| Specify the number of tasks per node. | --cpus-per-task=<c> | -c <c> | Specify the number of physical or logical cores per task. | --mem=<size> |
| Specify the real memory required per node. The given value should be an integer. Different units can be specified using the suffix K,M or G. Default units are megabytes. | --mem-per-cpu=<size> |
| Specify the minimum memory required per CPU core. The given value should be an integer. Different units can be specified using the suffix K,M or G. Default units are megabytes. | --exclusive |
| Indicate that all resources from the requested nodes are going to be granted with exclusive access (contrary to the default scheduling which shares the resources of nodes among different jobs: shared access) | --gres=gpu:<nG> |
| Specify the required number of GPUs per node. | --time=<timeLimit> | -t <timeLimit> | Set the wall-clock time limit for the job (hh:mm:ss). If the job exceeds this time limit, it will be subject to termination. | --job-name=<jobName> | -J <jobName> | Set the job name (as it will be displayed by squeue ). This defaults to the name of the batch script. | --output=<stdoutFile> | -o <stdoutFile> | (sbatch only) Set the file name for standard output. Use the token %j to include the job ID. | --error=<stderrFile> | -r <stderrFile> | (sbatch only) Set the file name for standard error. Use the token %j to include the job ID. | --partition=<partition> | -p <partition> | Request an allocation on the specified partition. If this option is not specified, jobs will be submitted to the default partition. | --qos=<qos> | -q <qos> | Request to run the job with a particular Quality of Service (QoS). | --array=<indexList> | -a <indexList> | (sbatch only) Specify an array job with the defined indices. | --dependency=<dependencyList> | -d <dependencyList> | Specify a job dependency. | --mail-type=<eventList> |
| Request an e-mail notification for events in eventlist. Valid event values include BEGIN, END, FAIL and ALL. Multiple values can be specified in a comma-separated list. | --mail-user=<address> |
| Specify an e-mail address for event notifications. | --export=<variables> |
| (sbatch only) Specify which environment variables are propagated to the batch job. Valid only as a command-line option. The recommended value is NONE. | --distribution=<distributionMethod> | -m <distributionMethod> | Specifies the distribution methods of allocation of cores |
|
When running jobs through Slurm, the unix group ownership of new files generated by the job execution is the one given to Slurm using the --account
option (-A
). This is usually your project group, so others in your project can read the files if they and the directory (and relevant parent directories) have the group-read attribute.
...
...