@upstash/lock

[!NOTE]
This project is in the Experimental Stage.

We declare this project experimental to set clear expectations for your usage. There could be known or unknown bugs, the API could evolve, or the project could be discontinued if it does not find community adoption. While we cannot provide professional support for experimental projects, we’d be happy to hear from you if you see value in this project!

@upstash/lock offers a distributed lock and debounce implementation using Upstash Redis.

Please use this lock implementation for efficiency purposes; for example to avoid doing an expensive work more than once or to perform a task mostly once in a best-effort manner. Do not use it to guarantee correctness of your system; such as leader-election or for the tasks requiring exactly once execution.

Upstash Redis uses async replication between replicas, and a lock can be acquired by multiple clients in case of a crash or network partition. Please read the post How to do distributed locking by Martin Kleppman to learn more about the topic.

NPM

npm install @upstash/lock

PNPM

pnpm add @upstash/lock

Bun

bun add @upstash/lock

To see a demo of the lock in action, visit https://lock-upstash.vercel.app

To create the Redis instance, you can use the Redis.fromEnv() method to use an Upstash Redis instance from environment variables. More options can be found here.

import { Lock } from "@upstash/lock";
import { Redis } from "@upstash/redis";

async function handleOperation() {
  const lock = new Lock({
    id: "unique-lock-id",
    redis: Redis.fromEnv(),
  });

  if (await lock.acquire()) {
    // Perform your critical section that requires mutual exclusion
    await criticalSection();
    await lock.release();
  } else {
    // handle lock acquisition failure
  }
}

import { Lock } from "@upstash/lock";
import { Redis } from "@upstash/redis";
import { expensiveWork } from "my-app";

const debouncedFunction = new Debounce({
  id: "unique-function-id",
  redis: Redis.fromEnv(),

  // Wait time of 1 second
  // The debounced function will only be called once per second across all instances
  wait: 1000,

  // Callback function to be debounced
  callback: (arg) => {
    doExpensiveWork(arg);
  },
});

// This example function is called by our app to trigger work we want to only happen once per wait period
async function triggerExpensiveWork(arg: string) {
  // Call the debounced function
  // This will only call the callback function once per wait period
  await debouncedFunction.call(arg)
}

new Lock({
  id: string,
  redis: Redis, // ie. Redis.fromEnv(), new Redis({...})
  lease: number, // default: 10000 ms
  retry: {
    attempts: number, // default: 3
    delay: number, // default: 100 ms
  },
});

Attempts to acquire the lock. Returns true if the lock is acquired, false otherwise.

You can pass a config object to override the default lease and retry options.

async acquire(config?: LockAcquireConfig): Promise<boolean>

Attempts to release the lock. Returns true if the lock is released, false otherwise.

async release(): Promise<boolean>

Attempts to extend the lock lease. Returns true if the lock lease is extended, false otherwise.

async extend(amt: number): Promise<boolean>

Returns whether the lock is ACQUIRED or FREE.

async getStatus(): Promise<LockStatus>

Creates a new debounced function.

new Debounce({
  id: string,
  redis: Redis, // ie. Redis.fromEnv(), new Redis({...})
  wait: number, // default: 1000 ms
  callback: (...arg: any[]) => any // The function to be debounced
});

Calls the debounced function. The function will only be called once per wait period. When called there is a best-effort guarantee that the function will be called once per wait period.

Note: Due to the implementation of the debounce, there is always a delay of wait milliseconds before the function is called (even if the callback is not triggered when you use the call function).

async call(...args: any[]): Promise<void>