AWS ParallelCluster - changing compute node config after cluster creation

Question

I have created a Slurm cluster on Ubuntu 20.04 using AWS ParallelCluster.

I then made a bunch of manual changes to the head/login node. I want to make a couple of small changes to the compute nodes as well (just install a couple of apt packages).

I know that I should treat the cluster as ephemeral and the way to make changes should be to use pcluster to change the cloudformation stack that generates the cluster. But in this case I am just trying to demo something and I've already invested time in making changes to the head node - I'm looking for a more quick and dirty solution with my existing cluster.

So what I am wondering is, is there any way to modify the compute instances short of recreating this cluster? For example, could I change the AMI that is used to launch these instances, or set/modify the user data or launch template that is run when they start?

Thanks.

Answer

The suggested alternative is to use a [custom bootstrap script](https://docs.aws.amazon.com/parallelcluster/latest/ug/custom-bootstrap-actions-v3.html) to install the required software in your compute nodes.

If you're using ParallelCluster 3.x you can also think to update the AMI of the compute nodes, through the [CustomAmi](https://docs.aws.amazon.com/parallelcluster/latest/ug/Scheduling-v3.html#yaml-Scheduling-SlurmQueues-Image-CustomAmi) configuration parameter (the compute fleet must be [stopped](https://docs.aws.amazon.com/parallelcluster/latest/ug/pcluster.update-compute-fleet-v3.html) to be able to perform this update).

Answer

PCluster use "Launch Template" to start EC2 resource when submitting jobs. You can find a Launch Template name like "-", try to change it.

AWS ParallelCluster - changing compute node config after cluster creation

Relevant content