JASMIN Slurm

This is a compilation of useful commands for the Jasmin LOTUS cluster. A full documentation is available on the Jasmin SLURM article. (SLURM = Simple Linux Utility for Resource Management)

A list of important commands and how to use them is here

1. Job submission and specification

Choosing which nodes to run on:

To select a host with a specific processor model and memory, add the following SLURM directive in your job script #SBATCH --constraint="<host-group-name>"

For example #SBATCH --constraint="skylake348G"

See here for a specification of available nodes on Jamsin.

Choosing a queue to schedule the job in:

Queues represent a set of pending jobs, lined up in a defined order, and waiting for their opportunity to use resources. The queue is specified in the job script file using SLURM scheduler directive #SBATCH -p <partition=queue_name> where <queue_name> is the name of the queue/partition. If the queue is not selected, SLURM will schedule the job to the queue short-serial by default.

See here for a specification of available queues, their priorities and typical runtimes.

2. Job monitoring

Monitoring commands

The main monitoring commands are sinfo (show available scheduling queues and how busy they are) and squeue (List user's pending and running jobs).

Jasmin Dashboard

On the JASMIN dashboard site you can see how busy the queues are in general, including the number of jobs running and pending, and a detailed view of the load on individual hosts.