g. Update your cluster

We have successfully built your cluster and ran your first MPI job.

Let’s say you want to change your instance type to a different instance type in your compute fleet for example c6i.32xlarge . In this lab, we will learn how to update your cluster with new instance type.

ParallelCluster UI

Click on the cluster in the parallelcluster ui console and select Edit

Edit Cluster

Click on the Queues tab and click Add Queue

  • Select c6i.32xlarge
  • Enable EFA
  • Enable placement group

Edit Cluster

On the next screen,

  • Stop Compute Fleet and click confirm
  • Dryrun to validate the config
  • Update to apply the update

Edit Cluster

Once the cluster goes into UPDATE_COMPLETE make sure to Start it:

Edit Cluster


vi config.yaml
- Name: compute
    - Name: m5xlarge
      InstanceType: m5.xlarge

You would need to stop the cluster (the compute fleet) using the following command:

pcluster update-compute-fleet --cluster-name hpc --status STOP_REQUESTED

Then run a pcluster update-cluster command

pcluster update-cluster --cluster-name hpc --cluster-configuration config.yaml

Start your cluster again after update process is completed.

pcluster update-compute-fleet --cluster-name hpc --status START_REQUESTED

Confirm Update

After the update has completed let’s log back into the cluster and check to make sure we see the new queue:

[ec2-user@ip-172-31-20-178 ~]$ sinfo
compute*     up   infinite     64  idle~ compute-dy-hpc6a-[1-64]
queue1       up   infinite      8  idle~ queue1-dy-queue1-c6i32xlarge-[1-8]

Now you have a better understanding on how AWS ParallelCluster operates. For more information, see the Configuration section of the AWS ParallelCluster User Guide.