Neato Polyester Material

    TypeScript icon, indicating that this package has built-in type declarations

    2.1.2 • Public • Published


    Build Status

    // Connecting
    const Jackd = require('jackd')
    const beanstalkd = new Jackd()
    await beanstalkd.connect()
    // Producing
    await beanstalkd.put('Hello!')
    // Consuming
    const job = await beanstalkd.reserve() // => { id: '1', payload: 'Hello!' }
    // ...process the job... then:
    await beanstalkd.delete(


    $ npm i jackd

    Version 2.x fixes a critical bug

    ⚠️ If you're using jackd in production, you should upgrade jackd to version 2.x. Read more here.


    Most beanstalkd clients don't support promises (fivebeans, nodestalker) and the ones that do have too many dependencies (node-beanstalkd-client). We wanted to make a package that has:

    • A concise and easy to use API
    • Native promise support
    • No dependencies
    • Protocol accuracy/completeness


    beanstalkd is a simple and blazing fast work queue. Producers connected through TCP sockets (by default on port 11300) send in jobs to be processed at a later time by a consumer.

    If you don't have experience using beanstalkd, it's a good idea to read the beanstalkd protocol before using this library.

    Connecting and disconnecting

    const Jackd = require('jackd')
    const beanstalkd = new Jackd()
    await beanstalkd.connect() // Connects to localhost:11300
    await beanstalkd.connect({ host, port })
    await beanstalkd.disconnect() // You can also use beanstalkd.quit; it's an alias


    Adding jobs into a tube

    You can add jobs to a tube by using the put command, which accepts a payload and returns a job ID.

    beanstalkd job payloads are byte arrays. Passing in a Buffer will send the payload as-is.

    // This is a byte array of a UTF-8 encoded string
    const jobId = await beanstalkd.put(Buffer.from('my long running job'))

    You can also pass in a String or an Object and jackd will automatically convert these values into byte arrays.

    const jobId = await beanstalkd.put('my long running job') // Buffer.from(string)
    const jobId = await beanstalkd.put({ foo: 'bar' }) // Buffer.from(JSON.stringify(object))

    All jobs sent to beanstalkd have a priority, a delay, and TTR (time-to-run) specification. By default, all jobs are published with 0 priority, 0 delay, and 60 TTR, which means consumers will have 60 seconds to finish the job after reservation. You can override these defaults:

    await beanstalkd.put(
      { foo: 'bar' },
        delay: 2, // Two second delay
        priority: 10,
        ttr: 600 // Ten minute delay

    Jobs with lower priorities are handled first. Refer to the protocol specs for more information on job options.

    Using different tubes

    All jobs by default are added to the default tube. You can change where you produce jobs with the use command.

    const tubeName = await beanstalkd.use('awesome-tube') // => 'awesome-tube'
    await beanstalkd.put({ foo: 'bar' })


    Reserving a job

    Consumers reserve jobs from tubes. Using await on a reserve command is a blocking operation and execution will stop until a job has been reserved.

    const { id, payload } = await beanstalkd.reserve() // wait until job incoming
    console.log({ id, payload }) // => { id: '1', payload: Buffer }

    jackd will return the payload as-is. This means you'll have to handle the encoding yourself. For instance, if you sent in an Object, you'll need to first convert the Buffer to a String and then parse the JSON.

    const { id, payload } = await beanstalkd.reserve()
    const object = JSON.parse(payload.toString())

    If you passed in a String, you'll need to convert the incoming Buffer to a string.

    const { id, payload } = await beanstalkd.reserve()
    const message = payload.toString()

    Performing job operations (delete/bury/touch/release)

    Once you've reserved a job, there are several operations you can perform on it. The most common operation will be deleting the job after the consumer is finished processing it.

    await beanstalkd.delete(id)

    Consumers can also give up their reservation by releasing the job. You'll usually want to release the job if an error occurred on the consumer and you want to put it back in the queue immediately.

    // Release immediately with high priority (0) and no delay (0)
    await beanstalkd.release(id)
    // You can also specify the priority and the delay
    await beanstalkd.release(id,
      priority: 10
      delay: 10

    However, you may want to bury the job to be processed later under certain conditions, such as a recurring error or a job that can't be processed. Buried jobs will not be processed until they are kicked.

    await beanstalkd.bury(id)
    // ... some time later ...
    await beanstalkd.kickJob(id)

    You'll notice that the kick operation is suffixed by Job. This is because there is a kick command in beanstalkd which will kick a certain number of jobs back into the tube.

    await beanstalkd.kick(10) // 10 buried jobs will be moved to a ready state

    Consumers will sometimes need additional time to run jobs. You can touch those jobs to let beanstalkd know you're still processing them.

    await beanstalkd.touch(id)

    Watching on multiple tubes

    By default, all consumers will watch the default tube only. Consumers can elect what tubes they want to watch.

    const numberOfTubesWatched = await'my-special-tube')
    // => 2

    Consumers can also ignore tubes.

    const numberOfTubesWatched = await beanstalkd.ignore('default')
    // => 1

    Be aware that attempting to ignore the only tube being watched will return an error.

    Executing YAML commands

    beanstalkd has a number of commands that return YAML payloads. These commands mostly return statistics regarding the current beanstalkd instance. jackd, on purpose, does not ship with a YAML parser. This is to:

    • Avoid dependencies
    • Stay close to the protocol spec
    • Let callers decide how to parse YAML

    jackd has full support for all commands, so you can expect to find these YAML commands in the API.

    const stats = await beanstalkd.stats()
    /* =>
    current-jobs-urgent: 0
    current-jobs-ready: 0
    current-jobs-reserved: 0
    current-jobs-delayed: 0
    current-jobs-buried: 0

    You can then pipe this result through a YAML parser to get the actual contents of the YAML file.

    const YAML = require('yaml')
    const stats = await beanstalkd.executeMultiPartCommand('stats\r\n')
    const { 'total-jobs': totalJobs } = YAML.parse(stats)
    // => 0

    Worker pattern

    You may be looking to design a process that does nothing else but consume jobs. You can accomplish this with one jackd client using async/await. Here's an example implementation.

    /* consumer.js */
    const Jackd = require('jackd')
    const beanstalkd = new Jackd()
    async function start() {
      // Might want to do some error handling around connections
      await beanstalkd.connect()
      while (true) {
        try {
          const { id, payload } = await beanstalkd.reserve()
          /* ... process job here ... */
          await beanstalkd.delete(id)
        } catch (err) {
          // Log error somehow





    npm i jackd

    DownloadsWeekly Downloads






    Unpacked Size

    113 kB

    Total Files


    Last publish


    • dimgl