ray/release/long_running_distributed_tests/cluster.yaml

cluster_name: long-running-distributed-tests

idle_timeout_minutes: 15

docker:
    image: anyscale/ray-ml:nightly-gpu
    container_name: ray_container
    pull_before_run: True

provider:
    type: aws
    region: us-west-2
    availability_zone: us-west-2a
    cache_stopped_nodes: False


available_node_types:
    gpu_2_32:
        node_config:
            InstanceType: g3.8xlarge

            BlockDeviceMappings:
                - DeviceName: /dev/sda1
                  Ebs:
                      VolumeSize: 201

            TagSpecifications:
                - ResourceType: "instance"
                  Tags:
                    - Key: anyscale-user
                      Value: '{{env["ANYSCALE_USER"]}}'
                    - Key: anyscale-expiration
                      Value: '{{env["ANSYCALE_EXPIRATION"]}}'

        resources: {"CPU": 32, "GPU": 2}
        min_workers: 3
        max_workers: 3


auth:
    ssh_user: ubuntu

head_node_type: gpu_2_32
worker_default_node_type: gpu_2_32

file_mounts_sync_continuously: false

setup_commands:
   - pip install -U {{env["RAY_WHEEL"]}}