i have a shutdown_callback method that triggers at call end. and i also have a livekit-egress webhook. What is the flow of execution of both of these. Which will trigger first.

Thanks for reaching out, Kushagra. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

yes, i need the shutdown callback to wait

Got it — you want the

shutdown_callback

to wait for the egress webhook (

egress_ended

) before finishing. That isn’t automatic, but you can coordinate them. Two patterns: • Preferred: Have your webhook handler signal completion and only run final shutdown work from there. a. Webhook receives

egress_ended

→ mark session egress complete (DB flag, cache, or in-memory store). b. Webhook then invokes your cleanup logic (or triggers the agent shutdown flow) so ordering is guaranteed. • Alternative: Have the

shutdown_callback

wait for the webhook signal. a. Webhook handler sets a shared flag/DB record or releases an async event when

egress_ended

arrives. b. In your

shutdown_callback

, poll or await that flag/event before proceeding. c. Ensure the shutdown timeout is long enough (default is 10s); increase

shutdown_process_timeout

if needed. Minimal example (conceptual):

# shared place: in-memory or persistent store keyed by session_id
egress_done = EventStore()

# webhook handler
def handle_egress_webhook(payload):
    session_id = payload["session_id"]
    egress_done.set(session_id)

# shutdown callback
async def shutdown_hook(ctx):
    session_id = ctx.session_id
    await egress_done.wait(session_id, timeout=30)  # wait for webhook signal
    await do_cleanup()

Notes: • Use a persistent store (DB or Redis) if webhook and agent run on different machines/processes. • Make your cleanup idempotent in case of retries or race conditions. • If you control the webhook endpoint, have it return success only after your cleanup has been triggered/queued to reduce race windows. Which pattern fits your architecture — letting the webhook trigger cleanup, or having the shutdown hook wait? Tell me whether webhook and agent run on the same host/process and I’ll give code tailored to your stack. Sources: Job lifecycle | LiveKit Documentation | Why is my Agent insights data missing? | LiveKit, with interactive elements