{
  "compilerOptions": {
    "baseUrl": ".",
    "paths": {
      "@redis/client": ["src/redis-client/source"],
      "@redis/client/*": ["src/redis-client/source/*"]
    },
    "moduleResolution": "node16",
    "target": "ES2020",
    "lib": ["ES2020"],
    "strict": true
  }
}

3. Ensure "esModuleInterop": true is set - the source uses mixed default/named exports.

4. Set environment variables as needed:

REDIS_URL=redis://localhost:6379
# For TLS:
REDIS_URL=rediss://user:pass@host:6380

5. In your application entry point, instantiate and connect the client before use (see examples below). The client is not connected until .connect() resolves.

Public API

RedisClientOptions

import { createClient } from '@redis/client';

interface RedisClientOptions<M, F, S, RESP, TYPE_MAPPING, SocketOptions> {
  url?: string;                          // redis[s]://[[user][:pass]@][host][:port][/db]
  socket?: RedisSocketOptions;           // TCP/TLS/Unix socket config
  username?: string;
  password?: string;
  credentialsProvider?: CredentialsProvider | StreamingCredentialsProvider;
  // ... modules, functions, scripts, resp version, type mapping
}

Use RedisClientOptions when constructing a client via createClient(). Supply url for simple setups; use socket for TLS, Unix sockets, or keep-alive tuning. If both credentialsProvider and username/password are set, credentialsProvider wins.

OpenTelemetry.init

import { OpenTelemetry } from '@redis/client'; // re-exported from opentelemetry/index.ts

OpenTelemetry.init(config: ObservabilityConfig): void

Singleton initializer for node-redis metrics. Call once at application startup before creating any Redis clients you want to observe. Throws OpenTelemetryError if called twice or if @opentelemetry/api is not installed. Supports metric groups: "pubsub", "connection-basic", "resiliency".

TokenManager / StreamingCredentialsProvider

import {
  TokenManager,
  TokenManagerConfig,
  StreamingCredentialsProvider,
  StreamingCredentialsListener,
  RetryPolicy
} from '@redis/client'; // re-exported from authx/index.ts

TokenManager wraps an IdentityProvider and continuously refreshes tokens, emitting updates to registered TokenStreamListeners. Use StreamingCredentialsProvider to supply dynamically-rotating credentials (e.g. Entra ID tokens) to a RedisClient without reconnecting. Set a RetryPolicy to control back-off on token fetch failures.

RedisClusterOptions

import { createCluster } from '@redis/client';

interface RedisClusterOptions<M, F, S, RESP, TYPE_MAPPING> {
  rootNodes: Array<RedisClusterClientOptions>; // at least 3 nodes recommended
  defaults?: Partial<RedisClusterClientOptions>;
  // modules, functions, scripts, resp, typeMapping
}

Used with createCluster(). rootNodes seeds topology discovery; defaults applies global settings (TLS, credentials, timeouts) to every shard client. The cluster client handles slot routing, ASKING redirects, and re-subscription automatically.

Working examples

Scenario - Basic standalone client with error handling

A minimal Express service connecting to Redis, setting and getting a key, then cleanly disconnecting on shutdown.

import { createClient } from '@redis/client';

const client = createClient({
  url: process.env.REDIS_URL ?? 'redis://localhost:6379',
  socket: {
    reconnectStrategy: (retries) => Math.min(retries * 50, 2000),
    connectTimeout: 5000,
  },
});

client.on('error', (err) => console.error('Redis error:', err));
client.on('reconnecting', () => console.log('Redis reconnecting...'));

await client.connect();

await client.set('session:abc', JSON.stringify({ userId: 42 }), { EX: 3600 });
const raw = await client.get('session:abc');
const session = raw ? JSON.parse(raw) : null;
console.log(session); // { userId: 42 }

process.on('SIGTERM', async () => {
  await client.quit();
  process.exit(0);
});

Scenario - Cluster client with pipelined multi/exec transaction

Connect to a Redis Cluster, perform an atomic counter increment inside a MULTI/EXEC block.

import { createCluster } from '@redis/client';

const cluster = createCluster({
  rootNodes: [
    { socket: { host: '10.0.0.1', port: 6379 } },
    { socket: { host: '10.0.0.2', port: 6379 } },
    { socket: { host: '10.0.0.3', port: 6379 } },
  ],
  defaults: {
    socket: { tls: false },
  },
});

cluster.on('error', (err) => console.error('Cluster error:', err));
await cluster.connect();

// MULTI/EXEC on a single key (same slot)
const results = await cluster.multi()
  .incr('counter:visits')
  .expire('counter:visits', 86400)
  .exec();

console.log('INCR result:', results[0]); // e.g. 1
console.log('EXPIRE result:', results[1]); // 1

await cluster.disconnect();

Scenario - OpenTelemetry metrics instrumentation

Bootstrap metrics before creating clients so that connection and command latency are tracked.

import { metrics } from '@opentelemetry/api';
import {
  MeterProvider,
  PeriodicExportingMetricReader,
  ConsoleMetricExporter,
} from '@opentelemetry/sdk-metrics';
import { OpenTelemetry, createClient } from '@redis/client';

// 1. Set up OTel SDK
const provider = new MeterProvider({
  readers: [
    new PeriodicExportingMetricReader({
      exporter: new ConsoleMetricExporter(),
      exportIntervalMillis: 10_000,
    }),
  ],
});
metrics.setGlobalMeterProvider(provider);

// 2. Init node-redis OTel - must happen before createClient
OpenTelemetry.init({
  metrics: {
    enabled: true,
    enabledMetricGroups: ['connection-basic', 'resiliency'],
  },
});

// 3. Create and use client normally
const client = createClient({ url: 'redis://localhost:6379' });
client.on('error', console.error);
await client.connect();

await client.ping();
await client.quit();

Scenario - Streaming credentials provider for token-based auth

Use StreamingCredentialsProvider (from authx/) to rotate credentials dynamically without reconnecting.

import { createClient, StreamingCredentialsProvider, StreamingCredentialsListener } from '@redis/client';

// Minimal custom streaming provider
class MyCredentialsProvider implements StreamingCredentialsProvider {
  subscribe(listener: StreamingCredentialsListener): { dispose: () => void } {
    // Push initial credentials
    listener.onNext({ username: 'myuser', password: 'initialpass' });

    const interval = setInterval(() => {
      // Rotate credentials from your identity provider
      listener.onNext({ username: 'myuser', password: fetchNewToken() });
    }, 55_000);

    return { dispose: () => clearInterval(interval) };
  }
}

function fetchNewToken(): string {
  return 'rotated-token-' + Date.now();
}

const client = createClient({
  url: 'redis://localhost:6379',
  credentialsProvider: new MyCredentialsProvider(),
});

client.on('error', console.error);
await client.connect();
await client.set('secure-key', 'value');
await client.quit();

File walkthrough

• RESP/ - Protocol-level type definitions: Command, CommandArguments, TypeMapping, RespVersions, reply types. The foundation every other module builds on.
• authx/ - Credentials and token lifecycle: CredentialsProvider, StreamingCredentialsProvider, TokenManager, IdentityProvider, Token, Disposable. Used by the client for re-authentication without reconnect.
• client/ - The RedisClient class and all its internals: socket management (socket.ts), command queue with back-pressure (commands-queue.ts), pub/sub channel management (pub-sub.ts), client-side caching (cache.ts), connection pool (pool.ts), legacy callback mode (legacy-mode.ts), OTel tracing hooks (tracing.ts).
• cluster/ - RedisCluster client: slot map (cluster-slots.ts), multi-command for cluster (multi-command.ts), cluster index with createCluster.
• commands/ - One file per Redis command. Each exports a typed transformArguments and transformReply pair consumed by commander.ts.
• opentelemetry/ - OpenTelemetry singleton, ClientRegistry, metrics definitions. Optional; only activates when initialized.
• sentinel/ - Sentinel-based HA client for automatic primary discovery and failover.
• utils/ - Internal helpers shared across modules.
• commander.ts - Attaches command definitions to client instances; handles argument prefixes for scripts and functions.
• errors.ts - Typed error hierarchy: ClientClosedError, ClientOfflineError, WatchError, DisconnectsClientError, ErrorReply.
• lua-script.ts - Defines Lua scripts as first-class Redis script objects with SHA caching.
• multi-command.ts - Base MULTI/EXEC queue logic shared by standalone and cluster multi-command classes.
• single-entry-cache.ts - Tiny single-slot cache used internally for repeated argument parsing results.
• test-utils.ts - Test infrastructure helpers; do not import in production code.

Common pitfalls and fixes

• Calling commands before connect() resolves: all commands queue and will reject with ClientOfflineError if the client is never connected; always await client.connect() before issuing commands.
• OpenTelemetry.init called more than once: it is a strict singleton and throws OpenTelemetryError on a second call; guard with a module-level flag or call only in your app entry point.
• credentialsProvider ignored when username/password also set: credentialsProvider takes precedence - remove username/password from options when using a provider.
• Cluster MULTI/EXEC across different hash slots: all keys in a transaction must map to the same slot; use hash tags {tag} in key names to force co-location.
• ESM/CJS interop: the source uses import/export; if your project is CommonJS set "esModuleInterop": true and "allowSyntheticDefaultImports": true in tsconfig.json.
• @opentelemetry/api peer dep missing at runtime: the OTel module does a dynamic require; if the package is absent it throws OpenTelemetryError - install @opentelemetry/api before calling OpenTelemetry.init.

Vibe-coding prompt template

I have dropped the node-redis core client source into `src/redis-client/source/`
and have a USAGE.md at `src/redis-client/USAGE.md`.

The upstream package is `@redis/client` (part of the `node-redis` monorepo).

Please help me integrate this into my existing Node.js/TypeScript project step by step:

1. Read USAGE.md and the file excerpts for `client/index.ts`, `cluster/index.ts`,
   `authx/index.ts`, and `opentelemetry/index.ts` to understand the exported API.

2. Add the necessary `npm install` commands for runtime dependencies.

3. Wire up a `RedisClient` singleton (using `createClient`) with proper error handling
   and graceful shutdown in my Express app entry point (`src/app.ts`).

4. If I need cluster support, create a `RedisCluster` using `createCluster` with my
   node list from environment variables.

5. If I need OpenTelemetry, call `OpenTelemetry.init` before any client is created.

6. Show me how to wrap common operations (GET/SET with TTL, MULTI/EXEC, pub/sub
   subscribe) using the real exports from `source/client/index.ts`.

7. Point out any TypeScript config changes needed (`tsconfig.json` paths, module
   resolution) for this source to compile correctly alongside my existing code.

Use only the real exported symbols documented in USAGE.md. Do not invent APIs.

License and credit

node-redis is released under the MIT License. See source/LICENSE if present, or refer to the upstream repository. Credit: the Redis open-source community and contributors at https://github.com/redis/node-redis.

This product’s unique contribution is clean room implementation of valuable code block into AVCP compatible standard.

完全なインストールガイドと統合プロンプトは購入後に解放されます。

ioredis Full-Featured Redis Client for Node.js

Automation, Utilities & Developer Tools

無料