Redis State Adapter

The Redis state adapter (chat_sdk-state-redis) stores all state data in Redis, making it suitable for multi-process and production deployments.

Installation

# Gemfile
gem "chat_sdk-state-redis"

Configuration

require "chat_sdk-state-redis"

# From URL
state = ChatSDK::State::Redis.new(url: "redis://localhost:6379")

# From environment variable (falls back to redis://localhost:6379)
state = ChatSDK::State::Redis.new

# With a pre-existing RedisClient instance
client = RedisClient.config(url: "redis://localhost:6379").new_client
state = ChatSDK::State::Redis.new(client: client)

Environment Variables

Key Namespacing

All Redis keys are namespaced under chat_sdk::

When to Use

Use Redis state when:

• Running multiple processes (Puma workers, multiple dynos)
• You need state to survive process restarts
• You are deploying to production

Atomic Operations

All Redis operations are atomic:

• Locks use SET NX PX for acquisition and Lua scripts for safe release (only the lock owner can release).
• set_if_absent uses SET NX to prevent race conditions.
• TTL-based expiration is handled by Redis natively.

Clearing State

state = ChatSDK::State::Redis.new
state.clear  # Deletes all chat_sdk:* keys

Use with caution in production -- this removes all subscriptions, locks, and stored state.

Direct Client Access

state = ChatSDK::State::Redis.new
redis_client = state.client  # RedisClient instance