Defining nodes for publish operations about scala-nsq HOT 3 OPEN

Hello!

Can you quote the documentation for nsqd that it should be running locally?

I wrote this library for my previous company project. We had a cluster running on amazon with few nsqd instances and thousands workers. I wrote custom NSQLookup using amazon api to determine where is nsqlookupd instance, as example. I can not imagine running nsq locally in the cloud.

from scala-nsq.

from the the faq: http://nsq.io/overview/faq.html

We strongly recommend running an nsqd alongside any service(s) that produce messages.

As an example, in our topology, we have a number of PHP servers working as HTTP API endpoints for our mobile clients, and a number of servers which consume from nsq. We have an nsqd running on each of our PHP servers; when we publish from our PHP code, we're doing a POST to 127.0.0.1. We also have a small number of nsqlookupd servers with which the producers are registered. These work very well from an autoscaling perspective as the nsqd instances running on PHP boxes all register with the nsqlookupd as new nodes are spawned.

As the library is, it's not difficult to write a custom NSQLookup to only produce to an nsqd running locally, but this is counter to how it's recommended a system be set up so at least documenting this is probably a good idea. Again, I've no problem writing something up in README and submitting a PR for it; indeed, I can share with you the NSQLookup I'm going to write to fulfill my company's needs if you're open to it.

from scala-nsq.

I didn't notice the letter six months ago when I wrote the library, my mistake.

Maybe better to create fork? I don't want commit breaking changes for already running productions.

from scala-nsq.