ray/release/stress_tests/cluster.yaml

54 lines
1.2 KiB
YAML

cluster_name: ray-stress-tests
min_workers: 100
max_workers: 100
idle_timeout_minutes: 5
provider:
type: aws
region: us-west-2
availability_zone: us-west-2a
cache_stopped_nodes: False
auth:
ssh_user: ubuntu
available_node_types:
cpu_64_ondemand:
node_config:
InstanceType: m4.16xlarge
resources: {"CPU": 64}
min_workers: 0
max_workers: 0
cpu_2_spot:
node_config:
InstanceType: m4.large
InstanceMarketOptions:
MarketType: spot
resources: {"CPU": 2}
min_workers: 100
max_workers: 100
docker:
image: anyscale/ray:nightly
container_name: ray_container
pull_before_run: False
head_node_type: cpu_64_ondemand
worker_default_node_type: cpu_2_spot
setup_commands:
- pip install -U pip
- pip install terminado
- pip install boto3 cython==0.29.0
- pip install -U {{env["RAY_WHEEL"]}}
head_start_ray_commands:
- ray stop
- ulimit -n 65536; ray start --head --port=6379 --autoscaling-config=~/ray_bootstrap_config.yaml
worker_start_ray_commands:
- ray stop
- ulimit -n 65536; ray start --address=$RAY_HEAD_IP:6379 --num-gpus=100