Managing your cluster
Create an OpenSearch cluster, protect its data, and control how it handles indexing and search workloads.
Set up a cluster
Follow these steps to set up a production cluster.
Install and configure OpenSearch on each node using Docker, a tarball, a package manager, or Kubernetes.
Learn how to specify settings in opensearch.yml, at startup, or using the Cluster Settings API, and whether you can change each setting while the cluster is running.
Choose node types, assign node roles, and configure the nodes in opensearch.yml to form a multi-node cluster.
Configure seed hosts, cluster bootstrapping, and voting so that nodes can find each other and elect a cluster manager.
Set up TLS certificates, users, and roles before you send production traffic to the cluster.
Back up and replicate data
Back up your data to remote storage and keep copies of it on other nodes and clusters.
Snapshots
Back up indexes and cluster state to a repository and restore them to the same or a different cluster.
Remote-backed storage
Send every index transaction to remote storage so that you can restore data after a node or cluster failure.
Control resource usage
When nodes receive more requests than they can complete, OpenSearch can limit, reject, or cancel requests. The following table lists the features that control resource usage.
| Feature | Effect |
|---|---|
| Workload management | Groups search requests into workload groups and limits the CPU and memory that each group can use. |
| Search backpressure | Cancels resource-intensive search requests when a node is under duress. |
| Shard indexing backpressure | Rejects indexing requests on overloaded shards. |
| Concurrency limits | Caps the number of requests that an action processes at the same time. |
| Cluster manager task throttling | Limits the number of tasks that nodes can submit to the cluster manager. |
Tune performance
To increase indexing throughput, see Tuning for indexing speed. To run indexing and search on different nodes, see Separating index and search workloads.
Manage scheduled jobs
OpenSearch plugins such as Index State Management, snapshot management, and anomaly detection run tasks on a schedule. To view the state of these tasks and tune how they are scheduled, see Scheduled jobs.
Monitor your cluster
OpenSearch can generate metrics and traces about its own operation and export them using OpenTelemetry, and it can report the slowest and most resource-intensive queries running in your cluster. For more information, see Monitoring your cluster.