A work queue for Node.js producers and consumers that uses Redis.
Example of a client producing work:
var Queue = require('simple-redis-safe-work-queue');
client = Queue.client('send-email');
client.push({to: 'someone@somewhere.com', subject: 'hey there', body: 'yo'});
or, with a callback for when the work is pushed:
client.push({to: 'someone@somewhere.com', subject: 'hey there', body: 'yo'}, function(err) {
if (err) console.error('Error when pushing work into the queue: ', err.stack);
});
You can also provide some options for the work item you're pushing:
var options = {
timeout: 120e3 // 120 seconds
};
client.push({to: 'someone@somewhere.com', subject: 'hey there', body: 'yo'}, options, function(err) {
if (err) console.error('Error when pushing work into the queue: ', err.stack);
});
Client also accepts options as second argument in constructor:
- port: redis port, defaults to 6379
- host: redis host name, defaults to "127.0.0.1"
- password: redis password, defaults to
undefined
- redisOptions: additional redis options, defaults to
undefined
- defaultTimeout: the default worker timeout, in miliseconds. defaults to 60000 (60 seconds)
Client emmits the following events:
emit('before push', work)
- before work is pushedemit('after push', work)
- after work is successfully pushedemit('error', err)
- when a push error happens and a callback wasn't provided
Example of a worker consuming work:
var Queue = require('simple-redis-safe-work-queue');
worker = Queue.worker('send-email', workFunction);
function workFunction(email, cb) {
sendEmail(email, function(err) {
if (err) cb(err);
else {
console.log('email send successfully, calling back with no errors');
cb();
}
});
}
You can pass some options on the third argument of the worker constructor:
- port: redis port (defaults to 6379)
- host: redis host (defaults to "127.0.0.1")
- password: redis password
- redisOptions: any option allowed by the redis client, defaults to
undefined
- maxConcurrency: the maximum pending work units. defaults to 10.
- maxRetries: the maximum number of retries when a work unit errors. defaults to 10.
- popTimeout: the worker pop timeout, after which it retries, in seconds. Defaults to 3 seconds.
- runTimeoutWatchdog: run a timeout watchdog, defaults to
true
- runStalledWatchdog: run a stalled watchdog, defaults to
true
- autoListen: worker listens automatically (otherwise you must call
.listen()
or.fetch()
when ready for the next message), defaults totrue
The .fetch()
method is a non-long-polling version of .listen()
, useful when you want to get a null
response if there are no messages, rather than waiting for one to be pushed. If you're using autoListen: false
then you probably want to use .fetch()
.
A worker emits the following events:
emit('ready')
: when redis client is readyemit('listening')
: when listening for workemit('worker error', err)
: when a worker error occursemit('work done', payload)
: when a worker finishes a piece of workemit('repush', payload)
: when a work unit is being repushed after failureemit('max retries', lastError, payload)
: when the maximum retries has been reached
Redis 2.6 or greater, with Lua scripting enabled.
Clone this repo, enter the repo directory, start a redis server and run:
$ npm test
MIT