rekrow

Robust distributed worker queue using RabbitMQ to end all the headache

TODO

• [x] Support retry limit
• [x] Support retry backoff
• [ ] Support retry exponential (or custom) backoff
• [ ] Support job routing (using RabbitMQ exchange routing)

Install

npm i -S rekrow

Usage

Dispatcher

const Rekrow = require('rekrow').default;

const rekrow = new Rekrow({
  url: 'amqp://localhost',
  jobName: 'example'
});

rekrow.connect()
  .then(() => {
    rekrow.enqueue({data: 1});
  });

Worker

const Rekrow = require('rekrow').default;

const rekrow = new Rekrow({
  url: 'amqp://localhost',
  jobName: 'example',
  handle(data) {
    console.log(data);
  }
});

rekrow.connect();

API

new Rekrow(opts)

Create a new Rekrow instance

• opts.url:
• opts.jobName:
• opts.handle: (data: Object) => Promise<any>: (optional) specify this handle when you want to handle incoming jobs.
• opts.maxParallelJobCount: number: (optional) how many job to process at the same time. Default to no limit.
• opts.maxRetryCount: number: (optional) how many time a fail job should be retried, e.g. 4 means the job will be processed in a total of 5 times (retried 4 times), before it is dicarded. Default to 0.
• opts.retryWaitTime: number: (optional) how many milliseconds a job should wait before re-queued for retry. Default to 5000.

rekrow.connect(): Promise<void>

connect much be called before doing anything.

rekrow.enqueue(data: Object): boolean

rekrow.close(): Promise<void>

Gracefully close connection to RabbitMQ. Will wait for any all running job to finish. But it will stop accepting new jobs.