async-lock

What is async-lock?

The async-lock npm package is designed to handle synchronization issues in Node.js applications where asynchronous operations need to be executed in a controlled sequence. It provides a mechanism to ensure that certain sections of code that should not be executed concurrently can be locked to avoid race conditions and other concurrency-related bugs.

What are async-lock's main functionalities?

Locking asynchronous operations

This feature allows you to lock a section of asynchronous operations using a unique key. The lock ensures that only one block of code that uses the same key can execute at a time, preventing race conditions.

const AsyncLock = require('async-lock');
const lock = new AsyncLock();

lock.acquire('key', function(done) {
  // async work
  setTimeout(function() {
    done(null, 'result');
  }, 1000);
}, function(err, result) {
  console.log(result); // prints 'result' after 1000ms
});

Promise support

Async-lock also supports promises, allowing you to use it in modern asynchronous workflows. This feature simplifies working with asynchronous operations by using promises instead of callbacks.

const AsyncLock = require('async-lock');
const lock = new AsyncLock();

lock.acquire('key', () => {
  return new Promise(resolve => {
    setTimeout(() => {
      resolve('result');
    }, 1000);
  });
}).then(result => {
  console.log(result); // prints 'result' after 1000ms
});

Other packages similar to async-lock

mutexify

Mutexify is another npm package that provides a mutual exclusion lock. It is similar to async-lock but differs in its API and internal implementation. Mutexify uses a simpler queue mechanism and does not support timeouts or promise-based APIs, which makes async-lock more versatile for complex applications.

rwlock

RWLock is a package that provides read and write locks, allowing multiple readers or a single writer. This is different from async-lock, which does not distinguish between read and write operations. RWLock is useful in scenarios where the distinction between read and write operations can optimize performance and resource utilization.

README

async-lock

Lock on asynchronous code

• ES6 promise supported
• Multiple keys lock supported
• Timeout supported
• Pending task limit supported
• Domain reentrant supported
• 100% code coverage

Why you need locking on single threaded nodejs?

Nodejs is single threaded, and the code execution is never get interrupted inside an event loop, so locking is unnecessary? This is true ONLY IF your critical section can be executed inside a single event loop. However, if you have any async code inside your critical section (it can be simply triggered by any I/O operation, or timer), your critical logic will across multiple event loops, therefore it's not concurrency safe!

Consider the following code

redis.get('key', function(err, value){
	redis.set('key', value * 2);
});

The above code simply multiply a redis key by 2. However, if two users run concurrency, the execution order may like this

user1: redis.get('key') -> 1
user2: redis.get('key') -> 1
user1: redis.set('key', 1 x 2) -> 2
user2: redis.set('key', 1 x 2) -> 2

Obviously it's not what you expected

With asyncLock, you can easily write your async critical section

lock.acquire('key', function(cb){
	// Concurrency safe
	redis.get('key', function(err, value){
		redis.set('key', value * 2, cb);
	});
}, function(err, ret){
});

Get Started

var AsyncLock = require('async-lock');
var lock = new AsyncLock();

/**
 * @param {String|Array} key 	resource key or keys to lock
 * @param {function} fn 	execute function
 * @param {function} cb 	(optional) callback function, otherwise will return a promise
 * @param {Object} opts 	(optional) options
 */
lock.acquire(key, function(done){
	// async work
	done(err, ret);
}, function(err, ret){
	// lock released
}, opts);

// Promise mode
lock.acquire(key, function(){
	// return value or promise
}, opts).then(function(){
	// lock released
});

Error Handling

// Callback mode
lock.acquire(key, function(done){
	done(new Error('error'));
}, function(err, ret){
	console.log(err.message) // output: error
});

// Promise mode
lock.acquire(key, function(){
	throw new Error('error');
}).catch(function(err){
	console.log(err.message) // output: error
});

Acquire multiple keys

lock.acquire([key1, key2], fn, cb);

Domain reentrant lock

Lock is reentrant in the same domain

var domain = require('domain');
var lock = new AsyncLock({domainReentrant : true});

var d = domain.create();
d.run(function(){
	lock.acquire('key', function(){
		//Enter lock
		return lock.acquire('key', function(){
			//Enter same lock twice
		});
	});
});

Options

// Specify timeout
var lock = new AsyncLock({timeout : 5000});
lock.acquire(key, fn, function(err, ret){
	// timed out error will be returned here if lock not acquired in given time
});

// Set max pending tasks
var lock = new AsyncLock({maxPending : 1000});
lock.acquire(key, fn, function(err, ret){
	// Handle too much pending error
})

// Whether there is any running or pending async function
lock.isBusy();

// Use your own promise library instead of the global Promise variable
var lock = new AsyncLock({Promise : require('bluebird')}); // Bluebird
var lock = new AsyncLock({Promise : require('q')}); // Q

// Add a task to the front of the queue waiting for a given lock
lock.acquire(key, fn1, cb); // runs immediately
lock.acquire(key, fn2, cb); // added to queue
lock.acquire(key, priorityFn, cb, {skipQueue: true}); // jumps queue and runs before fn2

Changelog

See Changelog

Issues

See isse tracker.

License

MIT, see LICENSE