Is there a way to run two agent servers and split request in certain ratio manually, using same SIP configuration? I'd like to test two agents with different settings.

It seems likely that you can run two agent servers and split requests between them using the same SIP configuration. According to the LiveKit documentation, LiveKit's architecture supports deploying multiple agents and distributing load across them. You can use a load balancer to manually control the distribution of requests between the two agents, allowing you to test different settings. If you want more detailed guidance on setting this up, please let me know! Sources: - LiveKit documentation 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.

@millions-hair-24956 Let say I want to 1 worker takes only 1 call at a time, no simultaneous calls, and the other worker takes all the other calls. How do I do it?

It might be possible to achieve this by configuring the

load_fnc

and

load_threshold

parameters in the

WorkerOptions

for each worker. According to the LiveKit documentation, you can set a custom

load_fnc

to control how jobs are distributed to workers. For the worker that should handle only one call at a time, you could set the

load_threshold

to a very low value, ensuring it doesn't accept more than one job. The other worker can have a higher threshold to handle multiple calls. Would you like more detailed instructions on how to configure this? Sources: - LiveKit documentation 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.