Remote Mediasoup client

Remote Mediasoup client

This project host a class that allow to connect to servers offering the Remote Mediasoup API, like Remote Mediasoup server, Mediasoup proxy, and Mediasoup cluster. Once connected, it provides the same API that Mediasoup, but using remote instances instead of a local one. This allow to use the same code in both cases, and easily migrate from one to the other, opening the door to scale your Mediasoup based application without having to change your code.

Install

npm install --save @mafalda-sfu/remote-mediasoup-client

API

See API documentation.

Usage

If you already have a Mediasoup based project, to use Remote Mediasoup client on it you’ll only need to do two changes in your project code:

1. Replace the mediasoup import for remote-mediasoup-client:

   - import mediasoup from 'mediasoup'
   + import RemoteMediasoupClient from 'remote-mediasoup-client'
   

2. Create the instance of the RemoteMediasoupClient object, wait for its connected event, and access to the mediasoup property:

   + const remoteMediasoupClient = new RemoteMediasoupClient('ws://example')
   +
   + await once(remoteMediasoupClient, 'connected')
   +
   + const {mediasoup} = remoteMediasoupClient
   

After that, the mediasoup variable will have an object that’s API compatible with the Mediasoup API provided by the Mediasoup package.

Known bugs (AKA “features”)

• On Mediasoup API, close() method (and send() method, and also change of content of appData properties) from Mediasoup objects are synchronous, once you call them, they are done. This is not true on Remote Mediasoup client, since internally they are sending a request to the server, and also it’s possible to have multiple clients connected to the same server calling to them. If you are using only one client, you can assume that the changes will be done, the same as with normal Mediasoup, but if you are using multiple ones (also from the same server application instance), you need to listen to the close event of the objects. This will allows you to keep you application state synchronized with the server state, otherwise you could think one Mediasoup object is alive, without knowing it was closed by another client instance.
• Mediasoup workerBin and Workers pid fields are provided directly by the server we are connected. These can have different values of the ones that would be provided by a local Mediasoup instance, for example having hints about the server that’s running the worker by using a “fake” PID beyond the operating system max limit ones or providing an URL with a domain instead of just only a path, so you must not assume they are relative to the server where your application code is running as would happen with a local Mediasoup instance, or that they are valid ones at all without processing them first.