cluster-config.yaml

Imds:
  ImdsSupport: v2.0
HeadNode:
  InstanceType: c5.xlarge
  Imds:
    Secured: false
  Ssh:
    KeyName: PclusterKeyPair
  LocalStorage:
    RootVolume:
      VolumeType: gp3
      Size: 512
  Networking:
    SubnetId: subnet-060e9faee76679f67
    AdditionalSecurityGroups:
      - sg-0d2f7c1635c457f03
      - sg-0c87f3cc385a3c147
      - sg-0e39397679e3c4cb5
  Iam:
    AdditionalIamPolicies:
      - Policy: arn:aws:iam::aws:policy/AmazonSSMManagedInstanceCore
      - Policy: arn:aws:iam::aws:policy/SecretsManagerReadWrite
      - Policy: arn:aws:iam::600627354061:policy/FsxPclusterUserPolicy
      - Policy: arn:aws:iam::aws:policy/CloudWatchFullAccess
      - Policy: arn:aws:iam::aws:policy/AWSPriceListServiceFullAccess
      - Policy: arn:aws:iam::aws:policy/AmazonSSMFullAccess
      - Policy: arn:aws:iam::aws:policy/AWSCloudFormationReadOnlyAccess
  CustomActions:
    OnNodeConfigured:
      Sequence:
        - Script: https://raw.githubusercontent.com/aws-samples/aws-parallelcluster-post-install-scripts/main/rest-api/postinstall.sh
        - Script: https://raw.githubusercontent.com/zhaohaom-aws/aws-parallelcluster-monitoring/main/post-install.sh
Scheduling:
  Scheduler: slurm
  SlurmQueues:
    - Name: queue-1
      AllocationStrategy: lowest-price
      ComputeResources:
      - Efa:
          Enabled: true
        InstanceType: trn1.32xlarge
        MaxCount: 1
        MinCount: 0
        Name: kaena-training-0
      - Efa:
          Enabled: true
        InstanceType: trn1.32xlarge
        MaxCount: 1
        MinCount: 0
        Name: kaena-training-1
      - Efa:
          Enabled: true
        InstanceType: trn1.32xlarge
        MaxCount: 1
        MinCount: 0
        Name: kaena-training-2
      - Efa:
          Enabled: true
        InstanceType: trn1.32xlarge
        MaxCount: 1
        MinCount: 0
        Name: kaena-training-3
      ComputeSettings:
        LocalStorage:
          RootVolume:
            VolumeType: gp3
            Size: 512
          EphemeralVolume:
            MountDir: /local_storage
      CustomActions:
        OnNodeConfigured:
          Script: https://raw.githubusercontent.com/zhaohaom-aws/aws-parallelcluster-monitoring/main/post-install.sh
      Networking:
        SubnetIds:
          - subnet-02961b064555b7b26
        PlacementGroup:
          Enabled: false
      Iam:
        AdditionalIamPolicies:
          - Policy: arn:aws:iam::aws:policy/AmazonS3FullAccess
          - Policy: >-
              arn:aws:iam::aws:policy/AmazonElasticContainerRegistryPublicFullAccess
          - Policy: arn:aws:iam::aws:policy/AmazonSSMManagedInstanceCore
          - Policy: arn:aws:iam::600627354061:policy/FsxPclusterUserPolicy
  SlurmSettings:
    QueueUpdateStrategy: TERMINATE
    Database:
      Uri: >-
        slurm-accounting-cluster.cluster-c1s046ccsiof.us-west-2.rds.amazonaws.com:3306
      UserName: clusteradmin
      PasswordSecretArn: >-
        arn:aws:secretsmanager:us-west-2:600627354061:secret:AccountingClusterAdminSecre-cO3gvYHQxo2j-RclylZ
Region: us-west-2
Image:
  Os: alinux2
Tags:
  - Key: parallelcluster-ui
    Value: 'true'
  - Key: Grafana
    Value: 'true'
SharedStorage:
  - Name: FsxLustre0
    StorageType: FsxLustre
    MountDir: /fsx
    FsxLustreSettings:
      FileSystemId: fs-0850ce039c4b8110e