Jobs
Please do not run jobs on the login node
Running jobs directly on the login node can have an adverse affect on other users of the cluster.
All jobs must run on the compute nodes.
Access to compute nodes is only available via the queuing system.
Jobs should be submitted either by requesting an interactive session, or by writing a script to run it in a batch mode. A script is a file that contains all the commands to run the job. When you submit the job, it is then placed in a queueing system which will allocate the job when the requested resources become available.
Please select the relevant option on the left hand menu for more details about the different types of jobs.
Monitoring your jobs
Submit a job script called job.sub
sbatch <job.sub>
Show a list of your running jobs
squeue -u <userid>
Show an overview of your jobs
susers -u <userid>
Show an overview of your array jobs
ausers -u <userid>
Show detailed information about a job
squeue -j <JOBID>
Kill a running a job
scancel <JOBID>
Show a list of all running jobs on the cluster
squeue
Show a overview of all running jobs on the cluster
lshosts
Show a overview of all running jobs on a specific queue (choose relevant queue)
lshosts | grep compute-28-128