Scaling elasticsearch


(Sunil Chaudhari) #1

Hi,
I want to attach one more node to running cluster.
Can i do it directly by starting one more instance with pre-requisite setting in ES config?

What more I need to do apart from cluster.name ?
What is impact on existing indexes?

br,
Sunil


(David Pilato) #2

Can i do it directly by starting one more instance with pre-requisite setting in ES config?

Not sure I understood the question.

What more I need to do apart from cluster.name ?

Set the discovery unicast list of nodes that are already running so they can discover each others.

What is impact on existing indexes?

Allocation of missing replicas if any and rebalancing of shards if needed...


(Sunil Chaudhari) #3

I mean, I will start ES with same cluster.name on another server in the same network.

discovery.zen.ping.unicast.hosts: ["ip-master","ip-data-node-1","ip-data-node-2"]

is this setting you are talking about?

Is re-balancing of shards required? How to do that?


(David Pilato) #4

is this setting you are talking about?

Yes

Is re-balancing of shards required? How to do that?

It's automatic. Don't worry about that.


(Sunil Chaudhari) #5

Hi,
Thanks.

Does it slows down the cluster when additional node joined?
I mean, may be because of re-balancing?

br,
Sunil


(David Pilato) #6

Not that much. Nothing you can notice unless you are already under a heavy load.

IIRC recovery process use at most 20mb/s bandwidth.


(system) #7

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