Pointing to different Nodes

[Jtomage]

Hi all,

New to elasticsearch I am currently for clarification on pointing to nodes with different roles. The current set up is a 3 Master-data node cluster and looking to possibly scale out. Currently there are 2 applications,

1. that continuously writes to elasticsearch
2. a web application for users to search.

According to this post correct-usage-of-coordinating-role-node I can the application point directly to the coordination node or data node via the IP address / url of elasticsearch node.

Meaning for application

1. the writing application, I create connection pool of urls / ip address of only the data nodes and send the data their.
2. the web app, I can add a coordination node to cluster and use only the ip address / url of the coordination node

Then if an ingest node is added to the cluster, do I add that nodes url / ip address to the connection pool of the data nodes?

Thank you

[rschirin (Robert)]

if your new node (with ingest role) is also a data node, you can add it. otherwise not.

[Jtomage]

Did I understand correctly about pointing the different applications to the different nodes?

Interesting, I do not need to point to an ingest only node from the application as elasticsearch will handle it.

[rschirin (Robert)]

from my personal point of view, you should use coordinating-only role for both activities (search and bulk-indexing actions).

following doc, here, ingest node is used to preprocess docs. if you do not need the preprocessing phase, you can skip it.