Reputation: 3208
I have a NodeJS app which sends HTTP get requests from various places in the code, some are even dependent (sending a request, waiting for a reply, processing it and based on results sending another request). I need to limit the rate the requests (e.g., 10 requests per hour).
I thought about queuing the requests and then at some central point releasing them in a controlled manner, but got stuck at how to queue the callback functions and their dependent parameters.
Would be happy to hear suggestions how to over come this scenario with minimum restructuring for the app.
Thanks
Upvotes: 14
Views: 19840
Reputation: 16495
You can use this library
By default the express processing is in parallel
var express = require('express');
var app = express();
app.get('/endpoint', function(req, res) {
console.log(new Date())
setTimeout(function () {
res.type('text/plain');
res.send('Hell , its about time!!');
}, 1000)
});
app.listen(process.env.PORT || 8080);
Log
var express = require('express');
var queue = require('express-queue');
var app = express();
app.use(queue({ activeLimit: 1, queuedLimit: -1 }));
app.get('/endpoint', function(req, res) {
console.log(new Date())
setTimeout(function () {
res.type('text/plain');
res.send('Hell , its about time!!');
}, 1000)
});
app.listen(process.env.PORT || 8080);
Log
As you can see the processing is sequential.
Upvotes: 0
Reputation: 43158
I would use Deferreds and return one for every queued request. You can then add succeed/fail callbacks to the deferred promise after it has been queued.
var deferred = queue.add('http://example.com/something');
deferred.fail(function(error) { /* handle failure */ });
deferred.done(function(response) { /* handle response */ });
You can hold a [ url, deferred ]
pairs in your queue, and each time you dequeue a URL you'll also have the Deferred that goes with it, which you can resolve or fail after you process the request.
Upvotes: 0
Reputation: 3101
The Async module has a number of control flow options that could help you. queue
sounds like a good fit, where you can limit concurrency.
Upvotes: 2
Reputation: 18979
I think that you have answered your question already. A central queue that can throttle your requests is the way to go. The only problem here is that the queue has to have the full information of for the request and the callback(s) that should be used. I would abstract this in a QueueableRequest
object that could look something like this:
var QueueableRequest = function(url, params, httpMethod, success, failure){
this.url = url;
this.params = params;
...
}
//Then you can queue your request with
queue.add(new QueueableRequest({
"api.test.com",
{"test": 1},
"GET",
function(data){ console.log('success');},
function(err){ console.log('error');}
}));
Of course this is just sample code that could be much prettier, but I hope you get the picture.
Upvotes: 8