A redis-based job queue for Node: job queue is hard, so we make it decent for you.
This project is maintained by bitinn
decent
is a decent Redis job queue for Node.js
There are powerful job queue modules for node.js + redis out there, like kue and bull, to name a few. But powerful API comes at a price, they need complex data structure and redis scripts to achieve features such as delayed job, pause/resume and full text search. And since redis doesn't have traditional transaction, ie. no rollback when one of the command failed, and doesn't trigger error in node-redis driver, things can go south without developers noticing. Plus it's difficult to figure out what really happened due to non-intuitive redis data structure.
To us, the proper answer is to design around this problem, instead of adding more features, we want a job queue that's barebone, fully tested, easy to inspect, and doesn't hide errors from developers.
npm install decent --save
TODO
Create a queue with name
and config redis client connection based on opts
, returns a decent queue instance.
var decent = require('decent');
var queue1 = decent('q1');
var queue2 = decent('q2', {
port: 6379
, host: 'localhost'
, connect_timeout: 5000
});
port
: redis server port, default to 6379
host
: redis server host, default to '127.0.0.1'
blockTimeout
: how long a client should wait for next job (see redis document on blocking command, such as BLPOP), defaults to 60
seconds, 0
to block forever.maxRetry
: how many retries a job can have before being moved to failure queue, defaults to 3
, 0
to disable retry.Create a job on queue using data
as payload and allows job specific opts
, returns a promise that resolve to the created job.
queue.add({ a: 1 }).then(function(job) {
console.log(job.data); // { a: 1 }
});
queue.add({ a: 1, b: 1 }, { retry: 1, timeout: 120 }).then(function(job) {
console.log(job.data); // { a: 1, b: 1 }
console.log(job.retry); // 1
console.log(job.timeout); // 120
});
retry
: set initial retry counter, default to 0
timeout
: set worker timeout in seconds, default to 60
id
: job iddata
: payloadretry
: current retry count for this jobtimeout
: how many seconds a worker can run before it's terminated.queue
: which queue this job currently belongs to.Register a handler function that process jobs, and start processing jobs in queue.
queue.worker(function(job, done) {
setTimeout(function() {
console.log(job.data);
done();
}, 100);
});
Must be called to signal the completion of job processing.
If called with an instance of Error
, then decent
will assume worker failed to process this job.
Fail jobs are moved back to work queue when they are below retry threshold, otherwise they are moved to failure queue.
Returns a promise that resolve to the queue length of specified queue, default to work
queue.
queue.count('work').then(function(count) {
console.log(count); // pending job count
});
queue.count('run').then(function(count) {
console.log(count); // running job count
});
queue.count('fail').then(function(count) {
console.log(count); // failed job count
});
Returns a promise that resolve to the job itself.
queue.get(1).then(function(job) {
console.log(job.id); // 1
});
Returns a promise that will resolve when job is removed from redis (both job data and job queue). Default queue is work
.
Note: remove
does not return the job, use get
then remove
instead.
queue.remove(1, 'run').then(function() {
// job has been removed from redis
});
Instructs queue worker to terminate gracefully on next loop. See events on how to monitor queue.
// ... setup queue and worker
queue.on('queue stop', function() {
console.log('queue stopped gracefully');
});
queue.stop();
Restarts the queue worker loop. See events on how to monitor queue.
// ... setup queue and worker
queue.on('queue start', function() {
console.log('queue restarted');
});
queue.restart();
decent
is an instance of EventEmitter
, so you can use queue.on('event', func)
as usual.
queue.emit('queue start')
: queue loop has started.queue.emit('queue work', job)
: queue worker begin to process a job
.queue.emit('queue ok', job)
: queue worker has processed a job
.queue.emit('queue error', err, job)
: queue worker has failed to processed a job
and thrown err
, will retry later.queue.emit('queue failure', err, job)
: queue worker has failed to processed a job
and thrown err
, retry limit reached.queue.emit('queue exit', err)
: queue loop has terminated due to unhandled err
.queue.emit('queue stop')
: queue loop has stopped gracefully.MIT