Elasticsearch domains should be configured with at least three dedicated master nodes

How to fix it

• Interrupts service: Turning dedicated master nodes on runs a blue/green deployment, so latency and rejected requests can rise while data copies; apply it at low traffic.

In the console

OpenSearch Service → Domains → the domain → Actions → Edit cluster configuration

Enable dedicated master nodes

Selected

Instance type

Sized for the cluster's node and shard count, per AWS's dedicated master sizing table

Number of master nodes

3; never an even number, and 5 only for extra failover

With the AWS CLI

aws es update-elasticsearch-domain-config --domain-name <domain-name> --elasticsearch-cluster-config DedicatedMasterEnabled=true,DedicatedMasterCount=3,DedicatedMasterType=<instance-type> --region <region>

With Terraform

resource "aws_elasticsearch_domain" "this" {
  domain_name = var.domain_name

  cluster_config {
    instance_type            = var.instance_type
    instance_count           = var.instance_count
    dedicated_master_enabled = true
    dedicated_master_count   = 3
    dedicated_master_type    = var.dedicated_master_type
  }
}