Skip to main content

RayJobConfig

This class represents the configuration for a Ray job. It allows users to specify details such as worker and head node configurations, autoscaling settings, runtime environments, and job termination behavior. This configuration is used to launch and manage Ray clusters for executing jobs.

Attributes​

  • worker_node_config: typing.List[WorkerNodeConfig]

    • Configuration for the worker nodes.

  • head_node_config: typing.Optional[HeadNodeConfig] = None

    • Configuration for the head node.

  • enable_autoscaling: bool = False

    • Whether to enable autoscaling for the job.

  • runtime_env: typing.Optional[dict] = None

    • The runtime environment for the job.

  • address: typing.Optional[str] = None

    • The address of the Ray cluster.

  • shutdown_after_job_finishes: bool = False

    • Whether to shut down the cluster after the job finishes.

  • ttl_seconds_after_finished: typing.Optional[int] = None

    • The time-to-live in seconds after the job finishes.

Constructors​

  • Initializes a RayJobConfig object.

  • Parameters

    • worker_node_config: typing.List[WorkerNodeConfig]
      • Configuration for worker nodes.
    • head_node_config: typing.Optional[HeadNodeConfig]
      • Configuration for the head node. Defaults to None.
    • enable_autoscaling: bool
      • Whether to enable autoscaling. Defaults to False.
    • runtime_env: typing.Optional[dict]
      • Runtime environment configuration. Defaults to None.
    • address: typing.Optional[str]
      • Address of the Ray cluster. Defaults to None.
    • shutdown_after_job_finishes: bool
      • Whether to shut down the cluster after the job finishes. Defaults to False.
    • ttl_seconds_after_finished: typing.Optional[int]
      • Time-to-live in seconds after the job finishes. Defaults to None.