To define multiple concurrent connections that an

agent.py

worker can support, you can configure this through the

WorkerOptions

class using load management parameters. Using WorkerOptions for Concurrency Control The

WorkerOptions

class provides parameters to control how many concurrent agent instances a worker can handle:

opts = WorkerOptions(
    entrypoint_fnc=your_entrypoint,
    # Function that reports the current system load (CPU, RAM, etc.)
    load_fnc=your_load_function,
    # Maximum load value - above this, new processes will not spawn
    load_threshold=your_threshold_value,
    # Other options...
)

How Worker Concurrency Works • Load Monitoring: The

load_fnc

parameter allows you to define a function that reports current system load (whether CPU, RAM, or other metrics). • Load Threshold: The

load_threshold

parameter sets the maximum value of your load function—above this threshold, new processes will not spawn. • Automatic Load Management: When a worker experiences high load, it stops taking on new sessions and notifies LiveKit's server. System-Level Concurrency LiveKit's dispatch system is optimized for high concurrency and low latency, supporting hundreds of thousands of new connections per second with dispatch times typically under 150ms. The framework automatically handles load balancing across available workers in the pool. Example Implementation Based on the

agent.py

example, you would configure concurrency when calling `cli.run_app()`:

if __name__ == "__main__":
    cli.run_app(WorkerOptions(
        entrypoint_fnc=entrypoint,
        load_fnc=your_custom_load_function,
        load_threshold=your_desired_threshold
    ))

The specific implementation of

load_fnc

and

load_threshold

values would depend on your system resources and desired concurrency limits.