Thanks — I’m responding to your question about the error message: ERROR livekit.agents: process did not exit in time, killing process Short answer: Yes — a slow shutdown webhook (or any long-running shutdown callback) can cause that error. Why: the agent framework waits a fixed grace period for shutdown tasks (including registered shutdown callbacks and webhooks) to complete. If those tasks don’t finish before the timeout, the agent force-kills the process and logs exactly that message. What you can do next • Confirm whether a shutdown webhook or shutdown callback runs when the call ends. If so, measure how long it takes to complete. • If the webhook sometimes takes longer than the default grace period, you have two options: ◦ Make the webhook faster or move long work off the shutdown path (e.g., enqueue work to a background job or retry queue instead of waiting synchronously). ◦ Increase the shutdown timeout so the process has more time to complete cleanup. How to increase the timeout • Use the

shutdown_process_timeout

parameter in the agent server options to raise the allowed shutdown time. Increasing this value will give shutdown hooks/webhooks more time before the process is killed. If you’d like, tell me: • Whether the webhook is synchronous during shutdown (it waits for a response), and • How long the webhook currently takes (rough estimate). I can then suggest a reasonable timeout value and give a short example of moving the webhook call off the shutdown path if you prefer. Sources: Job lifecycle | LiveKit Documentation | Server startup modes | LiveKit Documentation | Collect recording consent with tasks | LiveKit Documentation, with interactive elements