ray.serve.llm.LLMServer.collective_rpc

async LLMServer.collective_rpc(method: str, timeout: float | None = None, args: tuple = (), kwargs: dict | None = None) → list

Execute a collective RPC call on all workers.

This is used for RLHF workflows where a trainer needs to execute methods on all TP/PP workers (e.g., for weight synchronization).

Parameters:

• method – Name of the worker method to execute.

• timeout – Maximum time in seconds to wait for execution.

• args – Positional arguments to pass to the worker method.

• kwargs – Keyword arguments to pass to the worker method.

Returns:

A list containing the results from each worker.