To request an interactive session on the cluster, we use the salloc command:

salloc --partition=Instruction --time=02:00:00 --mem=1G --tasks=1 --cpus-per-task=1

Parts to note:

• the partition is set to Instruction, don’t change this value while this is being used for course work
• The time in the example is set to 2 hours. You can increase/shrink this as needed. Note however that the Instruction partition has a 12 hour max time limit. If jobs need to run longer, let the admin team know by emailing hummingbird@ucsc.edu
• The amount of RAM being requested is 1 GB per job. This should be more than enough, but can be adjusted
• The number of CPU Cores being request per job is 1. If the program being run handles threading/multicore, feel free to increase this value up to 2, 3 or 4.

After running the above command, you should see a line like this:

alloc: Granted job allocation 390096

This means your job submitted successfully. You can further check the queue to verify this:

squeue -al -u $USER

This should show you your currently running jobs in the cluster queue. Now, to get access to the compute node where your allocation is running, run the command:

ssh $SLURM_NODELIST

This will connect you to your allocation node. You can verify this worked be looking at the command prompt. If you’re on the login node, it looks like:

[rkparson@hb ~]$

but if you’re on your compute node, it will look like:

[rkparson@hbcomp-000 ~]#

(note – your compute node number will be different than this example)

When you are done with your allocation, first run exit once to log out of the allocated compute node, and then run exit a second time to release your allocation back to the pool.

You can run exit one final time to log out of the HB cluster.