if a room is deleted:
```await ctx.api.room.delet...
# ask-aii
important-judge-78504
09/12/2025, 2:00 PMif a room is deleted:
Copy code
await ctx.api.room.delete_room(
DeleteRoomRequest(
room=ctx.room.name,
)
)t
tall-airline-16756
09/12/2025, 2:00 PMThanks for reaching out, Ed. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
09/12/2025, 2:01 PMI understand that when a room is deleted using the API, it might raise some questions. What happens is that the agent session task ends because everyone in the room, including the agent, gets disconnected. This disconnection triggers a
Disconnectedi
important-judge-78504
09/12/2025, 2:25 PMim noticing that i get this error after the shutdown hook runs
process is unresponsive, killing processt
tall-airline-16756
09/12/2025, 2:26 PMIt looks like you're encountering the error
process is unresponsive, killing processshutdown_process_timeoutWorkerOptionsi
important-judge-78504
09/12/2025, 2:32 PMis it necessary to call
ctx.shutdown()t
tall-airline-16756
09/12/2025, 2:33 PMI'm glad you asked! Yes, it’s really important to call
ctx.shutdown()ctx.shutdown()ctx.shutdown()i
important-judge-78504
09/12/2025, 2:39 PMso what im noticing is that the shutdown callback is completing, but the process hangs after that
t
tall-airline-16756
09/12/2025, 2:40 PMIt’s quite common to see a process hanging after the shutdown callback completes, and this usually happens because some asynchronous tasks or resources haven’t been fully cleaned up yet, even though the shutdown callbacks themselves have finished. Here are a few common reasons this might be happening:
• Background tasks that are still running and haven’t been awaited or cancelled during shutdown.
• External resources like file handles or network connections that haven’t been properly closed.
• Child processes or threads that haven’t been terminated yet.
• The
shutdown_process_timeoutshutdown_process_timeoutWorkerOptionsi
important-judge-78504
09/12/2025, 2:53 PMwhy would existing async tasks that havent been cleaned up cause issues? the process killing timeout comes from the process not responding to livekit server pings right? so if a background task was still active, and prevented shutdown, wouldnt it still respond to pings?
t
tall-airline-16756
09/12/2025, 2:53 PMI understand how tricky async tasks can be when it comes to process shutdowns, so let me walk you through it gently. Uncleaned async tasks might cause process killing timeouts because they stop the process from shutting down smoothly and can block responses to LiveKit server pings. Here’s a simple breakdown:
• LiveKit keeps an eye on agent processes by sending regular ping requests and expects quick responses. If the process doesn’t reply within the
ping_timeoutshutdown_process_timeouti
important-judge-78504
09/12/2025, 3:05 PMhow would you recommend making sure all background tasks are cancelled
t
tall-airline-16756
09/12/2025, 3:06 PMTo help you smoothly cancel all background tasks during shutdown, here’s a friendly step-by-step guide you can follow:
1. Keep track of all tasks: It’s a great idea to maintain a collection (like a list or set) of all active
asyncio.Task.cancel()await asyncio.gather(*tasks, return_exceptions=True)utils.aio.cancel_and_wait()aclose()close()SIGINTSIGTERMutils.aio.cancel_and_wait()i
important-judge-78504
09/12/2025, 8:37 PMwhat are the differences between mute_published_track and update_participant? like what if i do this:
if track_pub.kind == 1 and track_pub.source == 2:
await ctx.api.room.mute_published_track(
MuteRoomTrackRequest(
room=ctx.room.name,
identity=participant.identity,
track_sid=track_pub.sid,
muted=not can_speak,
)
)
await ctx.api.room.update_participant(
UpdateParticipantRequest(
room=ctx.room.name,
identity=participant.identity,
permission=ParticipantPermission(
can_subscribe=can_hear,
can_publish=True,
can_publish_data=True,
),
)
will it stop the system from taking action from the users speech? or am i muting and then unmuting?
t
tall-airline-16756
09/12/2025, 8:38 PMThank you for your question! Let's walk through the differences between the
mute_published_trackupdate_participantmute_published_trackupdate_participantcan_publishi
important-judge-78504
09/12/2025, 8:38 PMso that code i sent doesnt undo the mute?
t
tall-airline-16756
09/12/2025, 8:38 PMThank you for sharing your code! I want to help clarify how the mute functionality works here. The
mute_published_trackmutedupdate_participantcan_subscribecan_publishcan_speakFalsei
important-judge-78504
09/12/2025, 9:10 PMis it possible at all the user state change would interfere with an outbound call being made?
t
tall-airline-16756
09/12/2025, 9:11 PMI completely understand how these technical details can feel a bit overwhelming, so let me walk you through it gently. When a user’s state changes—say from "listening" to "away"—this shift happens within the LiveKit Agents framework and is mainly about managing how the agent behaves during a conversation. It’s triggered by things like Voice Activity Detection (VAD), which helps the system understand if someone is actively engaged or not.
For instance, if a user becomes "away" because of silence for a while, the agent might send a friendly check-in message. But don't worry—this doesn’t interfere with the actual SIP call process, like making or keeping the call connected. The call’s progress, with states like "dialing," "ringing," or "answered," is handled separately by the SIP protocol itself, which manages the technical side of things such as INVITE responses and media timeouts.
So, to sum it up, while user state changes guide how the agent interacts during the call, they don’t affect the core setup or connection of your outbound call. I hope this helps clarify things, and please feel free to ask if you want me to explain anything further!
Sources: Events and error handling | LiveKit Docs | SIP Integration | livekit/livekit_composite | DeepWiki | pkg/sip/outbound.go
i
important-judge-78504
09/12/2025, 9:13 PMno i mean, if were in the middle of making an outbound sip call, like adding a new participant to the existing call, and during that the current users state change happens, will that cause problems that could cancel the outbound call being made?
t
tall-airline-16756
09/12/2025, 9:13 PMI completely understand how important it is to have a seamless experience when managing participants during an ongoing outbound SIP call. The great news is that changing a user's state or adding a participant (like bringing in a new person) doesn’t inherently cause any issues or cancel the call in LiveKit. The system is thoughtfully designed to handle these dynamic participant changes smoothly without disrupting your active calls.
For example:
• The
create_sip_participantsip.callStatusactivei
important-judge-78504
09/12/2025, 9:17 PMdoes livekit have any internal hook that does something when a user goes away
t
tall-airline-16756
09/12/2025, 9:18 PMI'm happy to share that LiveKit includes a helpful internal event called
user_state_changedAgentSessionuser_away_timeoutUserStateChangedEvent Copy code
@session.on("user_state_changed")
def on_user_state_changed(ev):
if ev.new_state == "away":
# Take action when user goes awayi
important-judge-78504
09/12/2025, 9:19 PMwould the emission of an event during an outbound call being created cancel or interrupt the outbound call?
t
tall-airline-16756
09/12/2025, 9:20 PMHi there! I wanted to clarify that when a
user_state_changedCreateSIPParticipantuser_state_changeduser_state_changed