Seeing some odd behavior occasionally when agents are connecting to cloud and being dispatched to rooms.... 1. Sometimes agents will join a room, but will never be initialized. I never see the on_enter hook being called, they just sit in the room indefinitely. Anyone experience similar issues? a. I'm using explicit dispatches - It seems there is some kind of caching going on as once it happens it keeps happening with subsequent dispatches until I change the agent-name param. b. Example Room Session ID: RM_vFxSbNDLAMKT c. On agents 1.0.15

Hi team, recently 'Connection success' has been dipping below 90%. For a very long time it was basically 100%, is there anything I can do from our side?

Hey livekit team I have deployed my python voicepipeline agent in digital ocean as a worker but it taking my 100% cpu usage what is the reason behind this behaviour have made 2 containers and both of them are maxed out and these are not being used I haven't joined any room or anything just idle and it's taking all the cpu resources ??Can somebody help me in identifying this problem and resolving it .. Slack Conversation

Hello cloud team, we are getting this issue from time to time:

The room connection was not established within 10 seconds after calling job_entry. This may indicate that job_ctx.connect() was not called

. We notice that sometimes the

context.connect()

task can take minutes (after multiple retries) disturbing our LiveKit calls, as the participant won't wait for minutes to start a call.

Hi team, I have a simple agent and when i trigger a call using

lk sip participant create sipParticipant.json

sometimes nothing happens I get the call and once i answer the answer call in my phone it is full silent this happens randomly

2025-04-29 21:44:34,976 - INFO livekit.agents - see tracing information at <http://localhost:52421/debug> 
2025-04-29 21:44:35,155 - INFO livekit.agents - registered worker {"id": "AW_oAn7dMW3ySDE", "url": "<wss://test1-ieaurmkt.livekit.cloud>", "region": "India", "protocol": 15}
2025-04-29 21:45:12,445 - INFO livekit.agents - shutting down worker {"id": "AW_oAn7dMW3ySDE"}
2025-04-29 21:45:13,406 - DEBUG asyncio - Using selector: KqueueSelector 
2025-04-29 21:45:13,410 - INFO livekit.agents - starting worker {"version": "1.0.11", "rtc-version": "1.0.3"}
2025-04-29 21:45:13,412 - INFO livekit.agents - see tracing information at <http://localhost:55281/debug> 
2025-04-29 21:45:13,578 - INFO livekit.agents - registered worker {"id": "AW_om7UTLWSbm6S", "url": "<wss://test1-ieaurmkt.livekit.cloud>", "region": "India", "protocol": 15}
2025-04-29 21:45:15,461 - INFO livekit.agents - received job request {"job_id": "AJ_BVX6pSPMVRCC", "dispatch_id": "", "room_name": "test12d3347jkd7ehdjm8dd", "agent_name": "", "resuming": false}
2025-04-29 21:45:16,115 - INFO livekit.agents - initializing job process {"pid": 53194}
2025-04-29 21:45:16,115 - INFO livekit.agents - job process initialized {"pid": 53194}
2025-04-29 21:45:16,115 - DEBUG asyncio - Using selector: KqueueSelector {"pid": 53194, "job_id": "AJ_BVX6pSPMVRCC"}
on_enter called
2025-04-29 21:45:23,979 - DEBUG livekit.agents - http_session(): creating a new httpclient ctx {"pid": 53194, "job_id": "AJ_BVX6pSPMVRCC"}
2025-04-29 21:45:24,006 - DEBUG livekit.agents - start reading stream {"participant": "sachin-112e3", "source": "SOURCE_MICROPHONE", "pid": 53194, "job_id": "AJ_BVX6pSPMVRCC"}

and even after 20sec still nothing happens , I’m using twilio as the provider has anyone experienced any similar issues ?

Hello Team I'm testing our STT with livekit for this i'm using two script from livekit 1. publish_wave.py # read a local WAV file and publish it to a room 2. transcriber.py # print the transcription Please check the attach scripts. for the details. But the problem is each time I run this test I get different transcript for a same WAV file. Can anyone tell me whet i'm dong wrong. If you need any other information regarding this please let me know Thanks CC @nice-fish-21757

HI Everyone every one I am facing one issue , Only one person can talk to my agaent in my room i want multiple users at a time what to do ?/

once again asking if it is possible to use our own egress server with livekit cloud?

Could anyone help diagnose an issue with RM_R4vf5ujAzSag we had multiple participants confirmed as joined and showin in events as joined, but they never received connection confirmation on the client side, the Participant timeline only shows as 1 part, connection latency is showing as 1,495ms & also multiple participants are showing as "Unknown" region. Any help appreciated to try and stop this from happening again in the future

Can we get an update on the nature of the current

major outage

? https://status.livekit.io/

For agents in a production setting, I understand worker availability is our own responsibility (e.g. monitoring, autoscaling, etc.) Using Livekit Cloud, I'm wondering if there is anyway we setup observability for when jobs are unable to be dispatched to a worker for whatever reason. Livekit Webhooks would seem a natural mechanism to get such notification but it only sends room events. I'm hoping to have alerting / monitoring for both sides of the equation, on both the job producer (Livekit Cloud) side and the job consumer (agent worker) side. Any insights would be appreciate.

Is there an issue with deleting rooms in 1.0.18? I am invoking the hangup function as per the documentation and it only removes my agent instead of disconnecting all participants

👋

hi,. i am getting this below error when connecting to a room can anyone help on what might it be ------------below is the error ---------- LivekitRoomContext.jsx:26 Unknown DataChannel error on lossy {room: '67713c96ebf81223510643ba', roomID: 'RM_54zRkj8v8s3E', participant: 'Josephine#01', pID: 'PA_3CrEqCHNbmv8', event: RTCErrorEvent} event RTCErrorEvent {isTrusted: true, error: OperationError: User-Initiated Abort, reason=, type: 'error', target: RTCDataChannel, currentTarget: RTCDataChannel, …} pID : "PA_3CrEqCHNbmv8" participant : "Josephine#01" room : "67713c96ebf81223510643ba" roomID : "RM_54zRkj8v8s3E" [[Prototype]] : Object Promise.then(anonymous) @ LivekitRoomContext.jsx:26 (anonymous) @ LivekitRoomContext.jsx:88 XMLHttpRequest.send usePostMethod @ CRUDApiHandler.js:258 await in usePostMethod onPressAcceptCall @ ChatscreenCallComp.jsx:203 onClick

Hey team, I am trying to have separate livekit setups for separate environments . So the prod livekit deployment should only serve the prod users while the dev should only serve the dev users. What is the proper way to achieve this ? I came across this message for some context - https://livekit-users.slack.com/archives/C048FRL1N2C/p1745429930336079 So, is creating multiple projects for multiple environments the right approach? Also, how do we set up an organization, or do we need to contact your team for that? Thanks!

anyone experiencing delays and distortion withcloud this morning?

Hello LiveKit Community! I'm working on a project using ESP32 to connect with LiveKit, and I've run into an issue with WebRTC DataChannel communication. My current setup: - ESP32 client - Python SDK for the agent server - Audio transmission is working correctly What's working: - I can successfully send data FROM the agent server TO the ESP32 using the

publish_data

function - The ESP32 successfully receives this data The problem: - When I try to send data FROM the ESP32 TO the LiveKit server, I receive a SACK (acknowledgment), but the data never triggers the

on_datareceived

callback on my agent I've verified that my data is being properly encoded with protobuf before sending from the ESP32, and I can decode it successfully for testing. Has anyone encountered a similar issue with one-way DataChannel communication? Could there be something I'm missing in how the DataChannel needs to be configured for bidirectional data flow with ESP32? Any suggestions or insights would be greatly appreciated!

Hi 🙂 I just created a new account via an email and project on cloud and the members page is broken. in the console it throws:

TypeError: t.user.name[0] is undefined
    Ember 2
    NextJS 23
<anonymous code>:1:145535
TypeError: t.user.name[0] is undefined
    Ember 2
    NextJS 23

Happens on

<https://cloud.livekit.io/projects/><project_id>/settings/members

Also it seems like the wss url now ends on .livekit:78000 and not livekit.cloud. Is this another UI bug? That url doesnt seem to be routable

im trying to deploy my agent to AWS (ECS), im getting the following error: aiohttp.client_exceptions.WSServerHandshakeError: 401, message='Invalid response status', url='wss://XXX.livekit.cloud/agent, it works locally and I double-checked that the key & secret are configured correctly, any one encountered a similar situation ?

This message was deleted.

Hey everyone, I am using this dockerfile to deploy the liveKit agent on an AWS EC2 Ubuntu instance, while building the image it gets stuck in this step - RUN python main.py download-files, without throwing any error Has someone faced same issue and know how to resolve it?, thanks

Hey 👋 Is is just me or does sign in via GitLab Not work anymore? Can‘t reach the dashboard so hope you can help me?

Hi, is there a way to have livekit use a custom domain we own for the url? i.e. removing the livekit.cloud here and using our own domain:

wss://{tenant}.livekit.cloud

Is there an issue with livekit cloud? My worker has been throwing load issues, but my container memory load is not near the worker threshold. Looked at Grafana logs to verify and didn't get much insight from this github issue .

Entirely removing / modifying any of my webhook destinations still results in the webhooks being sent to pre-modification/deletion urls. I figure there's some time delay before changes are displayed but i'm getting scenarios where I change my webhook urls from A to B and an hour later i'm still getting 50/50 being set to A or B.

Hi all, is there any api to check consumption & billing, we want to setup an alert if usage consumption has passed a certain threshold ?

Hey All, Quite Urgent, I can't login through google into the livekit cloud dashboard. Please advise. cc: @refined-appointment-81829, sorry for tagging directly, but we need to login as soon as possible. Thanks 🙏

I am also unable to login and update payment method

Impacting production usage