How to isolate limiter config when working with horizontal scaling

[michielswaanen]

Description

Hey, first of all great library. It's really saving me a lot of time! 😄

I have a question about how to use Bull with my application that supports horizontal scaling.
Currently my queue initialization code looks like this:

BullModule.registerQueue({
  name: "assets",
  limiter: { 
    // Max 1 jobs every 10  seconds
    max: 1,
    duration: 10000, 
  },
})

When my application has spun up multiple instance (the horizontal scaler) still only processes 1 job every 10 seconds.
I expected that this limiter config wouldn't be shared among the application instances.

Question

Is it possible to make the limiter isolated for each application that connects with Bull/Redis?

Should I look at it differently and create a separate Queue per instance?

[manast]

Bull is a distributed queue system, therefore the limiters are per queue, independent of how many processes you spawn to process the jobs in a given queue. If you need different limiters then you should use a different queue. Also, if you are new to Bull I recommend you to use BullMQ as the rate-limiter there is more robust than the one in Bull: https://docs.bullmq.io/ https://github.com/taskforcesh/bullmq

[michielswaanen]

Hey @manast, thanks for getting back to me. I'm currently using the groupKey to identify different server instances. This seems to work, however, I will take a look at BullMQ!

[manast]

groupKey is not available in BullMQ anymore. But there are more powerful group related functionalities in the Pro version: https://docs.bullmq.io/bullmq-pro/introduction

[michielswaanen]

Thanks I will take look!

[yanyiting]

 收到，谢谢！