A process wrapper for Bee-Queue workers.
npm install --save @compwright/worker-bee
Single-threaded usage:
const Queue = require('bee-queue');
const Worker = require('@compwright/worker-bee');
const queue = new Queue('test-queue');
worker = new Worker(queue);
worker.start(async (job) => {
job.reportProgress(100);
console.log(`Job ${job.id} completed`);
});
Pressing CTRL+C or sending a SIGINT or SIGTERM signal will cause the worker to gracefully shut down and exit.
Bee-Queue supports workers in multiple processes or servers. Here is a multi-process worker example using the throng process manager:
const throng = require('throng');
const Queue = require('bee-queue');
const Worker = require('@compwright/worker-bee');
function timeout(ms) {
return new Promise(resolve => setTimeout(resolve, ms));
}
// throng by default forks one worker per CPU
throng(pid => {
console.log(`Worker ${pid} started`);
const queue = new Queue('test-queue');
const logger = {
log: (...args) => console.log(`[${pid}]`, ...args),
error: (...args) => console.error(`[${pid}]`, ...args),
};
// handle six jobs concurrently per worker process
worker = new Worker(queue, { concurrency: 6, logger });
worker.start(async (job) => {
for (var i = 0; i < 5; i++) {
job.reportProgress(100 * i / 5);
}
console.log(`Job ${job.id} completed`);
});
});
Throng will relay the shutdown signal to each worker process to tell them to gracefully shut down and exit.
Returns a new worker instance initialized with the given options.
Options:
queue
- A Bee-Queue queue instanceshutdownTimeout
(optional) - time to wait in milliseconds for worker processing to stop before terminating (seeQueue.close()
)stalledCheckInterval
(optional) - how often to check for stalled jobs in milliseconds (seeQueue.checkStalledJobs()
)concurrency
(optional) - how many jobs to process concurrently within the worker thread (seeQueue.process()
)logger
(optional) - logger instance to use instead ofconsole
(must be interface-compatible withconsole
)
Listen for jobs to process.
Options:
callbackFn
- a function to execute the job. Receives thejob
as the first argument.
Stop listening for jobs to process.
Called automatically on SIGINT or SIGTERM signals (such as CTRL+C), and if an uncaught exception occurs.
Options:
error
- an error object to associate with the shutdown. If passed, causes a non-zero exit code.
Copyright (c) 2019
Licensed under the MIT license.