Fork me on GitHub
#onyx
<
2015-10-11
>
yusup09:10:05

Hi, Is there any doc on how to fire up several nodes with Onyx from fresh ?

yusup09:10:44

I have some Clojure experience ,but I am fairly new to distruted computation .

yusup09:10:39

Could anybody show the way ? or I just should RTFC?

lucasbradstreet10:10:59

Hi there. Do you have an application working in local mode yet?

yusup10:10:14

Hi Lucas. Yes . I do.

lucasbradstreet10:10:58

Cool. The main difference is that they need to use a shared ZK server, not a local one.

yusup10:10:28

I am looking through stuff from here.

yusup10:10:58

but this also includes Riemann etc.

lucasbradstreet10:10:14

Much more recommended than what we do in onyx-benchmark

lucasbradstreet10:10:34

It's still a bit of a work in progress but it's all mostly there

yusup10:10:01

Thanks very much.

yusup10:10:14

Awesome work.

lucasbradstreet10:10:50

The idea with onyx nodes is that you need an ns with a main function that requires any nses used by your lifecycles and functions, and then call start-peers using a config pointing at a shared ZK server. Then they'll be ready to go, and you can call submit-job and it'll submit the job data to ZK, where all the peers will pick it up

lucasbradstreet10:10:18

One other thing, if you're using core async for your input tasks, you'll probably need a different input source because input tasks could end up on any peer

lucasbradstreet10:10:25

You're building a crawler? Cool

yusup10:10:10

I have skimmed the doc of Onyx. Now I have basic idea of what each part do. I have to reread the doc and go with the ansible .

lucasbradstreet10:10:54

IIRC, I think one missing piece in the playbook is where you bundle your systems jar up. If you look at what the onyx-template does, it should be pretty easy to get that part up and running

yusup10:10:22

OK. thanks