Setup OpenOndemand in AWS ParallelCluster

Hi all,

I am currently new to OpenOndemand and have had a colleague ask about setting it up on an HPC in AWS. The HPC is using the SLURM scheduler, and spins up and spins down compute resources based on demand (pun intended). The login node is always running, but the execute nodes are not. I was wondering if anyone had ever set something like this up, and if so would be willing to provide pointers.

Thanks!
Dennis

Hi and welcome!

Setting up OOD in AWS should be super similar to on premise installation. I’d bet the requirements are a lot like your login node, in that it needs to be able communicate with the slurm controller and LDAP support so uid/gids exist on the machine. I would say those are the 2 big things - making sure it can communicate to Slurm correctly and making sure users correctly exist on the machine.