View Search Node Performance Metrics

The release of Search and Vector Search introduced mongot node visibility in Ops Manager for self-managed deployments running on the MongoDB Enterprise Kubernetes Operator. Ops Manager collects performance data from these nodes and displays it in the mongot Details view. To open the Details view for a mongot node, click the node from the Deployment view. Ops Manager scrapes mongot metrics every 60 seconds from each node's Prometheus endpoint.

Prerequisites

Before you view mongot node data in Ops Manager, confirm the following prerequisites:

Run Ops Manager 8.0.25 or later. Cloud Manager doesn't support mongot node monitoring.
Deploy mongot 1.70.1 or later using the MongoDB Enterprise Kubernetes Operator 1.9 or later.
Reference an Enterprise MongoDB deployment or an external mongod deployment in the MongoDBSearch resource. Ops Manager doesn't support MongoDBSearch resources that reference MongoDBCommunity.
Enable the mongot Prometheus endpoint. The MongoDB Enterprise Kubernetes Operator enables this endpoint by default on port 9946 when you deploy the MongoDBSearch resource.

When the MongoDBSearch resource points to a MongoDB deployment in the same Kubernetes cluster, the MongoDB Enterprise Kubernetes Operator forwards metrics to Ops Manager automatically. When it points to an external mongod, you must configure the spec.observability.metricsForwarder settings manually. Allow network connectivity between the Kubernetes Operator metrics collector and Ops Manager. To learn more, see MongoDBSearch Resource Specification.

Limitations

The following limitations apply to mongot node monitoring in Ops Manager:

Ops Manager monitors mongot nodes deployed through the MongoDB Enterprise Kubernetes Operator only. Ops Manager doesn't monitor mongot deployments outside Kubernetes.
Ops Manager provides read-only observability for mongot nodes. Use the MongoDB Enterprise Kubernetes Operator to manage the mongot node lifecycle.
Ops Manager collects mongot metrics only. Ops Manager doesn't collect mongot logs or traces.
Ops Manager doesn't provide Search index management or query tooling.
Ops Manager stores mongot metrics at 1-minute granularity. Requests for finer granularity return 1-minute data.

Health Status

Ops Manager computes a health status for each mongot node from its most recently received metrics. Ops Manager displays the status as a badge next to the node.

Status	Badge Color	Condition
Healthy	Green	At least one index manager is in a steady state, or the node has no indexes.
Degraded	Amber	At least one index manager has failed.
Stale	Red	Ops Manager hasn't received metrics from the node in over 120 seconds, or the node has indexes but none has reached a steady state.
Unknown	Grey	Ops Manager hasn't received any metrics from the node.

Available Metric Charts

The mongot Details view provides the following metric charts. To learn about chart controls and display options, see View MongoDB Processes.

Metric	Description
CPU Usage	Displays normalized CPU usage percentage for the `mongot` process.
JVM Memory	Displays JVM heap memory used and maximum heap capacity for the `mongot` process. Use this chart to identify memory pressure that could affect search performance.
JVM GC Pause	Displays the maximum JVM garbage collection pause duration, in seconds. Frequent or long pauses can affect search query latency.
Disk Space Used	Displays the percentage of disk space used by the `mongot` process.
Index Size	Displays the combined size of all search indexes on the `mongot` process, in bytes.
Index Documents	Displays the total number of documents across all search indexes on the `mongot` process.
Index Segments	Displays the total number of Lucene segments across all search indexes on the `mongot` process.
Search Query Throughput	Displays the total rate and failed rate of Search queries, in operations per second.
Search Query Latency	Displays the average and maximum latency of Search commands, in milliseconds.
Vector Search Throughput	Displays the total rate and failed rate of Vector Search queries, in operations per second.
Vector Search Latency	Displays the average and maximum latency of Vector Search commands, in milliseconds.
Index Manager Health	Displays the number of index managers in each of the following states: `STEADY_STATE`, `FAILED`, and `FAILED_EXCEEDED`.

Learn More

To configure metric collection for mongot nodes, see MongoDBSearch Resource Specification.
To learn about monitoring mongot nodes in Kubernetes, see Monitor Your Deployment.
To review all available metrics in Ops Manager, see Review Available Metrics.
To learn about monitoring self-managed mongot deployments, see Monitor mongot Deployment.

Back

MongoDB Processes

Real Time Metrics