Eureka and rabbitmq

[michas]

Hi,

I'm new to Eureka and after setting up some examples

with (Java-based) clients + services I'd like to know

how Eureka fits for more complex setups.

I have a couple of (Java-) services which are connected

via a clustered Rabbit-Queue. So each producer/consumer

has to deal with several rabbit-instances.

My first approach would be setting up an Eureka-client as a separate

service on each rabbit-machine so that I could do client-side-

loadbalancing by wrapping rabbit-client with Eureka/Ribbon.

Is this a good approach? If so, how do I handle a crashed

Rabbit-Service while Eureka still sends positive hearbeats?

And if not: What would be a better solution?

Thanks in advance,

michas

[Tomasz Bak]

I think you want a kind of a service we run for zookeeper cluster. Exhibitor (https://github.com/Netflix/exhibitor) registers with

Eureka server, and monitors its co-located zookeeper node.

--
You received this message because you are subscribed to the Google Groups "eureka_netflix" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eureka_netfli...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[michas]

Thanks for your reply!

Although your suggestion may solve my issue I still have doubts

if Eureka in general is a good solution for this kind of use-case (Java + non-Java Services).

Your suggestion implies 2 technologies (Eureka, Zookeeper) doing the same thing.

Is it best (or even good) practice to do it this way or is Eureka maybe not

the best choice for these (heterogenous) scenarios?

Do you have experience (in production) with such setups?

[dl...@netflix.com]

Hi Michas,

From what you describe, it looks like what you are looking for are solutions to two different problems. In one case, you are looking for a solution to manage rabbit service health states, and possibly tie this in with eureka statuses, and in a second case, you are looking for a solution for service discovery and client-side load balancing. We don't really run rabbitmq, but an analogous system is how we run zookeeper internally (where zookeeper servers == rabbitmq servers in the sense that it's something we need to manage server health states).

For server health, we use Netflix/exhibitor which is essentially a sidecar process that monitors the zk process's health. Our exhibitor service is also integrated with eureka client (see example discussions here https://github.com/Netflix/exhibitor/issues/193) where when the exhibitor detects server healthy/unhealthy for the zookeeper process, it will update the eureka status correspondingly.

For service discovery and client side loadbalancing, we use standard eureka/ribbon.