Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
ON THIS PAGE
 

Create an OpenSearch Multinode Cluster (Optional)

If you have an extra-large or custom deployment that requires more scale than those referenced in Apstra Flow Scaling Considerations, we recommend that you create an OpenSearch multinode cluster. A multinode cluster is a collection of nodes, that work together as a single unit. A multinode cluster can scale to hundreds of nodes, ensures high availability, and prevents downtime.

Create a Multinode Cluster

To create a multinode OpenSearch cluster, follow these steps:

  1. Configure the node type and associated roles.
    1. From the CLI, run /usr/local/bin/startup.
    2. Choose the node type from the TUI (text-based UI).
      You can choose from Data Node, Cluster Manager Node, or Coordinating Node. For descriptions about the different node types, see Multinode Roles and Configurations.
    3. Select Quit.
      When the TUI closes, the node name and node type are automatically set.
    4. Repeat steps 1 through 3 for each node in your configuration.
  2. Configure the cluster.
    Edit the opensearch.yaml file to set the IP address on all nodes in the cluster. For example:
  3. Restart OpenSearch for the changes to take affect.
  4. Run the following command to verify that cluster was successfully created.

    In the results output, the status has three possible values as described below. In this example, the green signifies that the cluster is healthy.

    • green: All primary and replica shards are active.
    • yellow: All primary shards are active, but some replica shards are unassigned.
    • red: Some primary shards are not active.
Continue to Apply Your License.

Multinode Roles and Configurations

  • Data node

    Stores data and participates in the cluster's indexing and search capabilities. Start with 2+ nodes and add more as data volumes grow.

  • Cluster manager node

    Manages the overall operation of a cluster and keeps track of the cluster state. Use odd numbers (3 or 5) for quorum. Avoid using an even number, which can cause instability.

  • Coordinating (client) node

    Manages search and indexing requests, and helps distribute the load across the cluster. Use 2 to 3 nodes to offload data nodes.

Scaling Considerations

  • Extra large or custom sizing options

    We recommend that you transition to a multinode cluster if you are processing over 1 TB of data or greater than 15,000 FPS. You might also want to consider a multinode cluster if you require OpenSearch to be highly available.

  • Odd versus even number of nodes

    For cluster manager nodes, always maintain an odd number to ensure a quorum. You can configure data, ingest and client nodes based on workload without this restriction.

  • Minimum node counts

    For a production environment, a minimum of three nodes is recommended to ensure fault tolerance (three cluster manager nodes and at least two data nodes).