Adapter

When scaling to multiple Socket.IO servers, you will need to replace the default in-memory adapter by another implementation, so the events are properly routed to all clients.

Besides the in-memory adapter, there are four official implementations:

• the Redis adapter
• the
• the Postgres adapter

There are also several other options which are maintained by the (awesome!) community:

• (e.g. RabbitMQ)
• NATS

You can have access to the adapter instance with:

Starting with , each Adapter instance emits the following events:

• create-room (argument: room)
• delete-room (argument: room)
• join-room (argument: room, id)
• (argument: room, id)

Example:

This may be useful for example in a microservice setup, where all clients connect to the microservice M1, while the microservice M2 uses the emitter to broadcast packets (uni-directional communication).

The emitter also supports the utility methods that were added in socket.io@4.0.0:

• socketsJoin()

• socketsLeave()

• disconnectSockets()

// make all Socket instances disconnect
emitter.disconnectSockets();

// make all Socket instances in the "room1" room disconnect (and discard the low-level connection)
emitter.in("room1").disconnectSockets(true);

// make all Socket instances in the "room1" room of the "admin" namespace disconnect
emitter.of("/admin").in("room1").disconnectSockets();

// this also works with a single socket ID
emitter.of("/admin").in(theSocketId).disconnectSockets();