🧠Redis Configuration
Last updated
Last updated
soketi is optimized to work in multi-node or multi-process environments like Kubernetes or PM2. However, as you may expect, the main concern when scaling horizontally is how to make the nodes communicate with each other. To be able to scale horizontally and efficiently enable node-to-node or process-to-process communication, soketi leverages a Redis connection.
You can configure soketi to run using Redis as a primary adapter for all typically local-persistent data using the ADAPTER_DRIVER
environment variable:
Adapters like Redis are used for horizontal scaling only. Configuring and running Redis when running soketi using the local
adapter will have no effect.
These are the Redis variables to configure the adapter for Redis. For the rest of them, like host, port, and clustering support, read Redis Configuration.
Name | Default | Possible values | Description |
---|---|---|---|
SOKETI_ADAPTER_DRIVER
local
redis
, nats
, local
, cluster
The adapter driver to use to store and retrieve each app and channel's persistent data.
SOKETI_ADAPTER_REDIS_PREFIX
''
Any string
The Redis adapter's Pub / Sub channels prefix.
SOKETI_ADAPTER_REDIS_CLUSTER_MODE
false
true
, false
Whether the client should be initialized for Redis Cluster. You have to specify the DB_REDIS_CLUSTER_NODES
value.