disque

Golang client for Disque, the Persistent Distributed Job Priority Queue.

• Persistent - Jobs can be either in-memory or persisted on disk^[1].
• Distributed - Disque pool. Multiple producers, multiple consumers.
• Job Priority Queue - Multiple queues. Consumers Get() from higher priority queues first.
• Fault tolerant - Jobs must be replicated to N nodes before Add() returns. Consumer must Ack() the job within a specified RetryAfter timeout or the job will be re-queued automatically.

Note: The examples below ignore error handling for readability.

Producer

import (
    "github.com/goware/disque"
)

func main() {
    // Connect to Disque pool.
    jobs, _ := disque.New("127.0.0.1:7711") // Accepts more arguments.
    defer jobs.Close()

    // Enqueue three jobs with different priorities.
    job1, _ := jobs.Add(data1, "high")
    job2, _ := jobs.Add(data2, "low")
    job3, _ := jobs.Add(data3, "urgent")

    // Block until job3 is done.
    jobs.Wait(job3)
}

Consumer (worker)

import (
    "github.com/goware/disque"
)

func main() {
    // Connect to Disque pool.
    jobs, _ := disque.New("127.0.0.1:7711") // Accepts more arguments.
    defer jobs.Close()

    for {
        // Get job from highest priority queue possible. Blocks by default.
        job, _ := jobs.Get("urgent", "high", "low") // Left-to-right priority.

        // Do some hard work with the job data.
        if err := Process(job.Data); err != nil {
            // Failed. Re-queue the job.
            jobs.Nack(job)
        }

        // Acknowledge (dequeue) the job.
        jobs.Ack(job)
    }
}

Default configuration

Config option	Default value	Description
Timeout	0	Block on each operation until it returns.
Replicate	0	Job doesn't need to be replicated before Add() returns.
Delay	0	Job is enqueued immediately.
RetryAfter	0	Don't re-queue job automatically.
TTL	0	Job lives until it's ACKed.
MaxLen	0	Unlimited queue.

Custom configuration

jobs, _ := disque.New("127.0.0.1:7711")

config := disque.Config{
    Timeout:    time.Second,    // Each operation fails after 1s timeout elapses.
    Replicate:  2,              // Replicates job to 2+ nodes before Add() returns.
    Delay:      time.Hour,      // Schedules the job (enqueues after 1h).
    RetryAfter: time.Minute,    // Re-queues the job after 1min of not being ACKed.
    TTL:        24 * time.Hour, // Removes the job from the queue after one day.
    MaxLen:     1000,           // Fails if there are 1000+ jobs in the queue.
}

// Apply globally.
jobs.Use(config)

// Apply to a single operation.
jobs.With(config).Add(data, "queue")

// Apply single option to a single operation.
jobs.Timeout(time.Second).Get("queue", "queue2")
jobs.MaxLen(1000).RetryAfter(time.Minute).Add(data, "queue")
jobs.TTL(24 * time.Hour).Add(data, "queue")

License

Disque is licensed under the MIT License.