Setting up Elasticsearch in Production

Hello All,

I am setting up ES in production with a different approach. I am actually splitting up the node role( master, data & ingest ). I know the configuration for each of them and this is well defined in the documentation. What makes me confuse is the how these three (master, node & ingest) node talk to each other. I have a total of 9 server ( 3 servers each ). Do i need to specify

discovery.zen.ping.unicast.hosts to have FQDN's of all 9 servers?

Or there is some other way or a best practices document to setup Elasticsearch with this type of approach.

If someone can help me point to a documentation it would be really helpful.

Thanks in advance.

--
Niraj

Just list the masters in there and you will be ok.

Thanks mark for the answer.

But i am bit confused as, if i just list the masters in there, how it is gonna know about client and data nodes.

And similarly how does data and client node know about each other.

--
Niraj

Because each node will contact that list and the master will then it will tell that node about the rest of the cluster.

Now i get it :slight_smile: .

And should i point my load balancer to client nodes instead of master. ?

100% yes :slight_smile:

Thanks Mark it works like a charm..... Appreciate the response and the great work elastic is doing.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.