https://livekit.io logo
Join Slack
Powered by
# self-hosting
  • b

    billowy-analyst-47555

    07/16/2025, 3:11 PM
    hey, just wanna make sure that for deployment, do i need to have the worker (agent) deployed and working & the web gen tokens (server) deployed and working and connect them to whatever deplyed frontend I have, am i missing other thing? also can i use heroku for deploying that backend?
  • s

    stale-psychiatrist-58431

    07/21/2025, 8:23 AM
    I have deployed the LiveKit server on Kubernetes using the official Helm chart. I'm currently facing an issue with pod autoscaling when load is applied to the LiveKit server. I have identified the root cause — it's due to a port conflict. Is there any way, I can rectify this pod autoscaling issue
    n
    • 2
    • 2
  • f

    few-rocket-69583

    07/21/2025, 3:07 PM
    HI All What is the point of
    analyticsservice.go
    ? also following fields are not being set. Is it only for cloud?
    Copy code
    events    rpc.AnalyticsRecorderService_IngestEventsClient
    	stats     rpc.AnalyticsRecorderService_IngestStatsClient
    	nodeRooms rpc.AnalyticsRecorderService_IngestNodeRoomStatesClient
    👀 1
  • c

    cool-piano-89763

    07/24/2025, 2:23 PM
    Hello everyone, I’m setting up a multi-region LiveKit server on a K3s cluster via WireGuard, so the cluster itself is already multi-region. I have a question about the Helm chart: In the chart’s values, there’s a
    livekit.region
    Does this mean every replica I deploy (or scale) will automatically be tagged with that region? Or have I misunderstood? In other words:
    if I install the chart once, does it support multiple regions out of the box, or is each Helm release tied to a single region?
  • a

    astonishing-glass-94804

    07/25/2025, 10:50 AM
    Hey, Has anyone integrated Krisp or any other noise cancellation solution into self hosted livekit?
    p
    • 2
    • 1
  • a

    acceptable-postman-80036

    07/25/2025, 12:54 PM
    Hey folks! Are there any good tutorials for a docker deployment in a gitlab pipeline? The documentation lacks on this topic 😞
    p
    • 2
    • 1
  • r

    rough-sunset-53533

    07/25/2025, 11:41 PM
    We are self hosting via Kubernetes in DigitalOcean and I am interested in what others are using to monitor their deployments. I want to be able to monitor the main media server, egress, and agents. Looking for as much insight into how the system is scaling and overall data as possible.
    e
    • 2
    • 1
  • s

    stale-psychiatrist-58431

    07/28/2025, 8:55 AM
    I have a self-hosted LiveKit deployment on Kubernetes using DigitalOcean. I performed load testing, but observed a packet loss between 30% and 40%. What is the ideal range for packet loss? Also, if codecs are not specified, what are the default codecs used?
    👀 1
  • s

    stale-psychiatrist-58431

    07/28/2025, 1:45 PM
    What should be the best tool for load testing of livekit?
    👀 1
    p
    • 2
    • 3
  • g

    gorgeous-notebook-41273

    07/28/2025, 4:07 PM
    Hey all. I want to get call analytics. on self-hosted Livekit. can someone guide me to do that?
  • c

    colossal-lock-10836

    07/30/2025, 1:34 PM
    I am self hosting livekit server and SIP and creatiing a Azure Openai Realtime deployment. All on premise. What ports do i need to expose from the agent side to the public for the communications?
  • s

    stale-psychiatrist-58431

    07/30/2025, 2:42 PM
    I have deployed self-hosted livekit on a digital ocean kubernetes, The issue is that when pod resources gets increases(as a result of more users in room) , it creates a new pod in a new node due to auto-scaling ..my question is that will my new users coming in a new node will join the same room present in first node?? I saw in docs the room can only be run on a single node. What is the solution for this issue in self hosted version?
    n
    • 2
    • 15
  • h

    helpful-lifeguard-86445

    07/30/2025, 3:35 PM
    Encountered an unexpected error, check the console logs for details: could not establish pc connection
    I get this while joining a room and it creates an alert on my application that I don’t have control over either.
    • 1
    • 1
  • e

    elegant-analyst-53735

    07/30/2025, 9:51 PM
    Hi, quick question - what's the usual way to deal with the room size limit when self hosting LiveKit on Kubernetes? Is there a way to split or replicate a room across multiple nodes?
    n
    f
    • 3
    • 16
  • p

    plain-solstice-34504

    07/31/2025, 12:53 AM
    Hey everyone! 👋 I noticed in the official docs that “LiveKit does not support deployment to serverless and/or private kubernetes clusters because extra NAT layers make WebRTC unhappy.” But I’m curious if anyone here has actually pulled this off: • Environment: Azure AKS (private cluster) • Pods: LiveKit Server, SIP Server, Redis, and LiveKit Agent • Network: Private AKS connected to our SIP trunk provider via Azure ExpressRoute (so no public endpoints at all) Is this fully-private setup even feasible, or does WebRTC inevitably need some public IP exposure?
    n
    • 2
    • 2
  • b

    busy-fish-13948

    07/31/2025, 1:11 PM
    We're deploying out Livekit Server + Egress in EC2 instances following this guide on the docs site. Putting them behind a layer 4 TCP load balancer and we're good to go I suppose?
  • s

    stale-psychiatrist-58431

    08/04/2025, 11:51 AM
    I am doing a load test on self hosted livekit cluster but the problem is livekit lk cli is supporting 720p video resolution. is there a way so that i can test it with FullHD video?
  • q

    quick-jelly-34727

    08/04/2025, 2:27 PM
    Which is the best platform to deploy voice agent backend
    m
    • 2
    • 2
  • h

    hundreds-battery-20727

    08/04/2025, 6:00 PM
    Hi, is there any document on what this scenario is (https://docs.livekit.io/home/self-hosting/ports-firewall/):
    Copy code
    Port	Default	Config	Exposed	Description
    TURN/TLS	5349	turn.tls_port	when not using LB	(optional) For a distributed setup, use a network load balancer in front of the port. If not using LB, this port needs to be set to 443.
  • h

    hundreds-battery-20727

    08/04/2025, 7:56 PM
    Has anyone compared c7i and c7g or c7gn for livekit server?
  • t

    thankful-market-54220

    08/05/2025, 6:25 AM
    Team, am struggling with self-hosting a voice agent built using google realtime model with video enabled. When i tried connecting it via livekit meet, i am speaking but agent is not replying back. does anyone tried realtime model and self-hosted it?
  • h

    hallowed-jordan-87556

    08/05/2025, 12:28 PM
    why livekit agent shows this kind of infinite error when try to host in AWS ECS, please guide
    r
    • 2
    • 1
  • s

    strong-balloon-74194

    08/05/2025, 4:32 PM
    Hey everyone, does anyone have experience deploying an agent on Render.com? I built the agent in Python and the frontend in React, using a cloud LiveKit server. I generate the token on the frontend and make the agent auto-join. When I deployed the frontend, it works fine. But when I try to connect the agent, I can't pass the metadata and token from the frontend to the backend. The message says the agent is "here", but no data is being sent to configure it. I’m thinking it might be a CORS issue or something else, I’m not sure. And I don't know how to fix it! It’s an urgent matter, so I’d really appreciate any help!
    r
    • 2
    • 3
  • e

    eager-dog-1123

    08/05/2025, 7:46 PM
    Hi everyone! I’m trying to self-host a LiveKit stack (server, SIP server, egress, workers) on AWS EKS with Flux GitOps. I’m running into connection issues and the setup has been challenging. I’d really appreciate help and I’m happy to post a paid task on Upwork (or similar) for someone interested. Current state: • Pods are up and services (LiveKit server, SIP server, egress) can see each other. • ALB and NLB are configured. • Required UDP/TCP ports, including the RTP range, are open. • TURN/ICE appear to be configured. • I’m not behind a corporate NAT; I just want to get a demo running in my EKS cluster. If you’re available, please DM me with your experience and rates - happy to move this to a paid engagement.
  • n

    numerous-megabyte-6153

    08/06/2025, 2:09 PM
    👋 Hello, team! I have been recently playing with livekit. It's fantastic! I am planning to self-host a livekit server in AWS EKS and enable different teams to develop voice enabled agents for different use-cases. The frontend and agent are going to be deployed in the different AWS accounts of each team. My question is regarding the design of the authentication in such a setup. From the docs it's clear that the frontend can work with a token. My question is regarding the agent. Does the agent always need the server API key and API secret to authenticate itself? Is there any other way I can make it on demand and create one credential for each agent. Any help appreciated thanks!
  • s

    stale-psychiatrist-58431

    08/06/2025, 5:06 PM
    Hi, everyone I have question, about the livekit deployed on kubernetes cluster. If I have a node of 16vcpu and 32 Gb ram or more than this resources and most of the resources given to one pod, how much users will easily managed by the one room in livekit pod.
  • m

    magnificent-optician-44475

    08/06/2025, 9:09 PM
    I just updated my livekit sip server to latest version (via docker pull) but it's still showing as 0.0.1 (despite 1.1.1 on the release tags in github) is this just a matter of the internal version not being updated? It looks like from version.go it's still set to 0.0.1 - this is terribly confusing... Can you confirm this?
  • s

    sparse-camera-68912

    08/07/2025, 11:57 AM
    I'm running a one-server livekit installation, but the TLS/TURN doesn't seem to work according to a manual test with
    Copy code
    turnutils_uclient -T -u 2UGaXFnM2VnTwJ0YfKIo4zG6mZ -w GosGFVlqpgLOa9ZADZ4mPhIT <http://call-turn.server.net|call-turn.server.net> -p 443 -S
    Gives me a
    tls_connect: cannot connect
    connecting to port
    5349
    works just fine. (without the
    -S
    ). I'm using the standard config from https://docs.livekit.io/home/self-hosting/vm/
    • 1
    • 1
  • p

    purple-rainbow-1246

    08/07/2025, 3:15 PM
    I am having an issue where a room closes after ~5 minute when the user is speaking continually. It seems to consider the room to be empty. I have since updated the Livekit config
    room:  empty_timeout
    from
    300
    to
    1800
    to avoid the issue. How can I ensure the user is not considered disconnected. Frontend and Server logs never indicated that the user got disconnected and that the room was empty.
  • m

    magnificent-optician-44475

    08/07/2025, 3:48 PM
    @magnificent-art-43333 are there plans for an enterprise support agreement for self hosted environments? It's good to have that option for highly regulated businesses where we need to control the data completely. There are gaps in documentation and support that would be nice to fill and it's a business opportunity for you!