d. Mount the FSx Lustre Filesystem on the cluster

In this section you will update the cluster created in Lab I and mount the filesystem created earlier in section a. of this lab.

  1. Go to your Cloud9 IDE terminal and run the env_vars if not done already.
source ~/environment/env_vars
  1. Create a post-install script to mount the FSx Lustre Filesystem on the cluster.
export filesystem_id=${FSX_ID}
export filesystem_dns=$(aws fsx --region ${AWS_REGION} describe-file-systems --file-system-ids $filesystem_id --query "FileSystems[0].DNSName" --output text)
export filesystem_mountname=$(aws fsx --region ${AWS_REGION} describe-file-systems --file-system-ids $filesystem_id --query "FileSystems[].LustreConfiguration[].MountName" --output text)
cat > mount-fsx.sh << EOF

sudo mkdir -p /fsx

sudo mount -t lustre -o noatime,flock ${filesystem_dns}@tcp:/${filesystem_mountname} /fsx

  1. Upload the post-install script to the previously created S3 bucket
aws s3 cp mount-fsx.sh s3://${BUCKET_NAME_DATA}
  1. Update cluster configuration file (my-cluster-config.yaml) with the post-install script and update the required policies. Note that we are updating the compute queue and not the head node
export S3PATH=s3://${BUCKET_NAME_DATA}/mount-fsx.sh
yq -i '(.Scheduling.SlurmQueues[0].CustomActions.OnNodeConfigured.Script=env(S3PATH)) |
       (.Scheduling.SlurmQueues[0].Iam.AdditionalIamPolicies[0]={"Policy": "arn:aws:iam::aws:policy/AmazonFSxFullAccess"}) |
       (.Scheduling.SlurmQueues[0].Iam.AdditionalIamPolicies[1]={"Policy": "arn:aws:iam::aws:policy/AmazonS3FullAccess"})' ~/environment/my-cluster-config.yaml
  1. Update the cluster
pcluster update-cluster -n hpc-cluster-lab -c my-cluster-config.yaml --region ${AWS_REGION} --suppress-validators ALL
  1. Wait for the cluster to be updated. You can check the cluster update status as below or monitoring the CloudFormation stack in the AWS Console.
pcluster describe-cluster -n hpc-cluster-lab --query clusterStatus --region ${AWS_REGION}

Once the cluster is updated, you will see a UPDATE_COMPLETE status.

  1. Re-start the compute fleet
pcluster update-compute-fleet -n hpc-cluster-lab --status START_REQUESTED --region ${AWS_REGION}
  1. Check the status of the compute fleet.
pcluster describe-compute-fleet -n hpc-cluster-lab --query status --region ${AWS_REGION}

You should see a RUNNING status after re-starting the compute fleet.

You have now successfully mounted the created Lustre filesystem on the cluster. In the next section, you will monitor the Filesystem and learn more about the HSM capabilities between FSx Lustre and Amazon S3.