Deploy and Host OpenVidu Meet on Railway

OpenVidu Meet is a self-hosted video calling app, an open source alternative to Zoom and Google Meet. Create a room, share the link, and people join from a browser with no account or install. It has screen sharing, chat, recording and a REST API for adding calls to your own product.

About Hosting OpenVidu Meet

Five services and a bucket. Meet is the app and API. LiveKit carries the audio and video. Egress makes recordings. MongoDB and Redis hold rooms and state. Recordings go to the bucket.

Calls run over TCP, through one port. Railway has no inbound UDP, which is what video calls normally use, and it's why video servers rarely work here. This template sends all media through a TCP proxy and sets up the address for you. Upstream's server doesn't work this way, so the template uses stock LiveKit.
Recording works on first deploy. Storage is already connected to the bucket, and files can be played or downloaded from the Meet console.
Nothing is open to strangers. The admin password and API key are generated for your deploy, and creating rooms requires one of them.
Sized for a small team. All five services together use about 410 MB at idle.

What to expect from TCP. It suits small meetings on normal connections. In testing, a two-person call ran at 720p and a four-person call at 360p per tile, both with no lost packets, on one good connection. Larger meetings and poor connections haven't been tested. On a poor connection a lost packet holds up everything behind it, so video stutters more than it would over UDP. Networks that only allow ports 80 and 443 can't connect, because the media port is a high-numbered one chosen by Railway. Some corporate and guest networks are like this.

Common Use Cases

A private meeting room for a team or a family, with no per-seat fees
Video calls inside your own app, using the REST API and embeddable links
Recording interviews, classes and consultations to storage you control
Client calls that shouldn't run through a third-party meeting service

Dependencies for OpenVidu Meet Hosting

Nothing external. MongoDB, Redis and the bucket are part of the template.

Deployment Dependencies

Implementation Details

First steps

Wait for all five services to finish. Egress takes the longest to build.
Open the Meet service's domain and sign in as admin. The password is MEET_INITIAL_ADMIN_PASSWORD in the Meet service's Variables. Your REST API key is MEET_INITIAL_API_KEY, sent as the X-API-KEY header.
Create a room and share its link. Moderator and speaker links are separate.

The admin password and API key are only read on the first start. Changing them later in Variables has no effect. Change the password under My profile in the Meet console.

Recording. Egress runs a headless browser to compose the video. Recording a four-person call at 720p used about 0.6 CPU cores and 330 MB of memory, on top of what it uses at idle, and ran fine with each service limited to 1 GB. A recording only uses this while it's running. If you don't need recording, you can remove the Egress service.

Bandwidth. Every participant sends and receives through Railway, and video adds up fast. Railway bills outbound traffic, so check your usage after your first real meetings.

Custom domains. After adding one to Meet, update MEET_BASE_URL. After adding one to LiveKit, update LIVEKIT_URL on Meet and LIVEKIT_WS_URL on Egress. Media keeps using the TCP proxy either way.

Not included. Phone dial-in (SIP), which needs UDP, and streaming in from OBS, which needs a separate ingest service. LiveKit's built-in TURN is also off, because it would have to run over the same TCP route anyway.

Keep the TCP proxy. Removing LiveKit's TCP proxy stops all calls. LiveKit reads the proxy's address and port each time it starts, so restart LiveKit if Railway ever changes them.

Licence. OpenVidu Meet Community Edition and LiveKit are Apache-2.0.

Why Deploy OpenVidu Meet on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying OpenVidu Meet on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.