Before diving into Elasticsearch and searching and aggregating data, you first need to create an Elasticsearch cluster.
Elasticsearch can operate as a single-node or multi-node cluster. The steps to configure both are, in general, quite similar. This page demonstrates how to create and configure a multi-node cluster, but with only a few minor adjustments, you can follow the same steps to create a single-node cluster.
To create and deploy an Elasticsearch cluster according to your requirements, it’s important to understand how node discovery and cluster formation work and what settings govern them.
There are many ways that you can design a cluster. The following illustration shows a basic architecture.
This is a four-node cluster that has one dedicated master node, one dedicated coordinating node, and two data nodes that are master-eligible and also used for ingesting data.
The following table provides brief descriptions of the node types.
| ||Manages the overall operation of a cluster and keeps track of the cluster state.|
| ||Available for election to become a new master node.|
| ||Stores and searches data.|
| ||Preprocesses data before storing it in a cluster.|
| ||Routes search and aggregation requests.|
This page demonstrates how to work with the different node types. It assumes that you have a four-node cluster similar to the preceding illustration.
Table of contents
- Step 1: Name a cluster
- Step 2: Set node attributes for each node in a cluster
- Step 3: Bind a cluster to specific IP addresses
- Step 4: Configure discovery hosts for a cluster
- Step 5: Start the cluster
- Next steps
Before you get started, you must install and configure Elasticsearch on all of your nodes. For information about the available options, see Install and Configure.
After you are done, use SSH to connect to each node, and then open the
You can set all configurations for your cluster in this file.
Step 1: Name a cluster
Specify a unique name for the cluster. If you don’t specify a cluster name, it’s set to
elasticsearch by default. Setting a descriptive cluster name is important, especially if you want to run multiple clusters inside a single network.
To specify the cluster name, change the following line:
Make the same change on all the nodes to make sure that they’ll join to form a cluster.
Step 2: Set node attributes for each node in a cluster
After you name the cluster, set node attributes for each node in your cluster.
Give your master node a name. If you don’t specify a name, Elasticsearch assigns a machine-generated name that makes the node difficult to monitor and troubleshoot.
You can also explicitly specify that this node is a master node. This is already true by default, but adding it makes it easier to identify the master node:
Then make the node a dedicated master that won’t perform double-duty as a data node:
Specify that this node will not be used for ingesting data:
Change the name of two nodes to
You can make them master-eligible data nodes that will also be used for ingesting data:
node.master: true node.data: true node.ingest: true
You can also specify any other attributes that you’d like to set for the data nodes.
Change the name of the coordinating node to
Every node is a coordinating node by default, so to make this node a dedicated coordinating node, set
node.master: false node.data: false node.ingest: false
Step 3: Bind a cluster to specific IP addresses
network_host defines the IP address that’s used to bind the node. By default, Elasticsearch listens on a local host. You can also use
_site_ to bind to any loopback or site-local address, whether IPv4 or IPv6:
network.host: [_local_, _site_]
Make sure to configure these settings on all of your nodes.
Step 4: Configure discovery hosts for a cluster
Now that you’ve configured the network hosts, you need to configure the discovery hosts.
Zen Discovery is the built-in, default mechanism that uses unicast to find other nodes in the cluster.
You can generally just add all of your master-eligible nodes to the
discovery.seed_hosts array. When a node starts up, it finds the other master-eligible nodes, determines which one is the master, and asks to join the cluster.
For example, for
odfe-master the line looks something like this:
discovery.seed_hosts: ["<private IP of odfe-d1>", "<private IP of odfe-d2>", "<private IP of odfe-c1>"]
Step 5: Start the cluster
After you set the configurations, start Elasticsearch on all nodes.
sudo systemctl start elasticsearch.service
Then go to the logs file to see the formation of the cluster:
Perform the following
_cat query on any node to see all the nodes formed as a cluster:
curl -XGET https://<private-ip>:9200/_cat/nodes?v -u admin:admin --insecure
ip heap.percent ram.percent cpu load_1m load_5m load_15m node.role master name x.x.x.x 13 61 0 0.02 0.04 0.05 mi * odfe-master x.x.x.x 16 60 0 0.06 0.05 0.05 md - odfe-d1 x.x.x.x 34 38 0 0.12 0.07 0.06 md - odfe-d2 x.x.x.x 23 38 0 0.12 0.07 0.06 md - odfe-c1
To better understand and monitor your cluster, use the cat API.
To set up security, see Security - Configuration.