Redis usage

Redis is the shared infrastructure for sessions, rate limiting, BullMQ queues, and short-lived tokens. The app expects Redis to be running before it boots.

Connection setup

InfrastructureModule registers RedisModule from nestjs-redis-client using the async config from src/config/redis.config.ts.

src/infrastructure/infrastructure.module.ts

RedisModule.registerAsync(redisConfig.asProvider())

The isGlobal: true flag in the config makes the client available globally without re-importing the module.

Redis database layout

The generated project uses different Redis logical databases to keep concerns separated:

DB	Use	Configured in
0	Default Redis connection (sessions, OTP, password reset tokens)	redis.config.ts
2	Throttler storage	security/rate-limiting/rate-limiting.module.ts
3	BullMQ queues	infrastructure/queue/queue.module.ts

Sessions are stored through connect-redis using the same Redis client as the default connection.

Injecting RedisService

Inject RedisService anywhere you need direct Redis access:

import { RedisService } from 'nestjs-redis-client';

constructor(private readonly redisService: RedisService) {}

async storeTemporaryCode(email: string, code: string) {
  await this.redisService.set(`verification:${email}`, code, 600);
}

getClient() returns the underlying ioredis client if you need lower-level commands.

Usage patterns in the starter

• Session store — connect-redis reads and writes session keys via sess:<sessionId>.
• Verification codes — verification:<email> keys with a 600-second TTL.
• Password reset tokens — password-reset:<token> keys with a 600-second TTL.
• WebSocket auth — The gateway reads sess:<sessionId> directly to authenticate socket connections.
• Rate limiting — Throttler stores request counters per IP/route.
• Queues — BullMQ uses Redis for job storage, delayed jobs, and worker coordination.

Scaling

For a single Redis instance, logical databases are enough. If you move to Redis Cluster, logical databases are not supported and you should split concerns by separate clusters or key prefixes. The throttler and BullMQ configurations already accept explicit host/port values, so you can point them to different Redis endpoints when needed.