Scaling ES Cluster with 2 more nodes


Hi there !

I got a 3 node Prod ES cluster for log analytics . All the 3 nodes ( 3 x 1TB ) are configured as both Master & Data . Min Master node is set to 2 .

Now to increase data retention and search performance , we are planning to add 2 more 1 TB nodes to the cluster. Cluster is running on m4 Large ( 16Gig / 4 Core ) in AWS

Got the below questions and i have been reading through materials for answers . Can someone suggest the best

  1. Can we configure the 2 new nodes to be data nodes alone and set the min master node setting to 3 ? Will this setting still ensure availability as Min master node satisfies N/2+1

  2. My current shard setting is 3 . Should I change it to 5 ? or can it be 3 ? Whats the best advisable setting

  3. Is there any sequence of steps to be followed before adding the two nodes . Just want this to be a rolling upgrade and dont want any downtime to users


(Divyanshi Dwivedi) #2
  1. You can add 2 nodes as data node only putting "node.master: false" in elasticsearch.yml.
    By default it is true. As you will be having 3 master nodes, your min master node should be 2. (3/2+1)
  2. You can keep the shards to 3 only. Your number of shards should be according to the rate of data and amount of data.
  3. You can simply add the node into the cluster. There will be some data relocation happening so your cluster will be heavily loaded. But there will be no downtime.

(system) #3

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