All Products
Search
Document Center

OpenSearch:Instance metrics

Last Updated:May 28, 2025

Procedure

On the details page of an OpenSearch Vector Search Edition instance, choose Metric Monitoring > Instance Metrics in the left-side navigation pane. On the page that appears, view the monitoring metrics of the instance. You can select a time range from the drop-down list to view the detailed monitoring data of the instance within the specified period of time. If you access the Instance Metrics page for the first time, you must complete the third-party application authorization. For more information, see Authorize RAM users to view instance monitoring metrics.

You can view the monitoring metrics of the instance, including the metrics of a Query Result Searcher (QRS) worker, query metrics of a Searcher worker, common metrics of a Searcher worker, and real-time data synchronization metrics of a Searcher worker.

Metrics

QRS

Metric

Description

qps

The number of queries that the QRS worker processes per second.

emptyQps

The number of queries that the QRS worker processes but returns empty results per second.

errorQps

The number of queries that the QRS worker fails to process per second due to errors such as syntax errors and query timeout errors.

parseQueryLatency

The time consumed by the QRS worker to parse query statements.

formatLatency

The time consumed by the QRS worker to format query results.

processLatency

The time consumed by the QRS worker to process queries.

rowCount

The number of entries in the result.

totalMergeTime

The time consumed by the QRS worker to merge query results in multiple shards.

totalSortTime

The time consumed by the QRS worker to sort query results.

queryLength

The length of query statements.

resultLen

The length of query results.

procMemUsedRatio

The memory usage of the QRS worker.

procMemRss

The size of memory used by the QRS worker.

procCpu

The CPU utilization of the QRS worker.

SEARCHER-QUERY

Metric

Description

qps

The number of queries that the Searcher worker processes per second.

totalLatency

The time consumed by the Searcher worker to process queries.

fetchFieldsLatency

The time consumed by the Searcher worker to obtain the recall results.

vectorSeekCount

The number of documents that are sought during the vector-based search.

vectorRecallRatio

The recall rate of the vector-based search.

vectorSeekLatency

The time consumed to seek documents during the vector-based search.

SEARCHER-NORMAL

Metric

Description

procCpu

The CPU utilization of the Searcher worker.

procMemRss

The size of memory used by the Searcher worker.

procMemUsedRatio

The memory usage of the Searcher worker.

diskUse

The size of disk space used by the Searcher worker.

diskUsedRatio

The disk usage of the Searcher worker.

partitionDocCount

The number of documents that are stored on the Searcher worker.

SegmentCount

The number of segments on the Searcher worker.

deletedDocCount

The number of documents that are deleted from the Searcher worker.

oldInMemorySegmentMemoryUse

The size of memory used by unreleased memory segments.

partitionMemoryQuotaUse

The memory quota used by Searcher worker indexes.

incIndexMemoryUse

The size of memory used by incremental indexes.

workerStatus

The state of the Searcher worker. Valid values: 1 and 0. A value of 1 indicates that the Searcher worker is in a normal state. A value of 0 indicates that the Searcher worker is in an abnormal state.

requestPoolWaitTime

The period of time for which a query waits in the query pool before the query is processed by the Searcher worker.

SEARCHER-RT-BUILD

Metric

Description

addQps

The number of queries that the Searcher worker processes per second when documents are added by calling API operations.

deleteQPS

The number of queries that the Searcher worker processes per second when documents are deleted by calling API operations.

buildQps

The number of queries that the Searcher worker processes per second for real-time indexing.

rtIndexMemoryUse

The real-time size of memory used by the Searcher worker.

end2EndLatency_default

The latency of data push by calling API operations.

dumpSegmentLatency

The time consumed to dump real-time data segments on the Searcher worker when incremental indexes are applied to the Searcher worker.

processedDocSize

The size of the processed documents.

rtIndexMemoryUseRatio

The real-time memory usage of the Searcher worker.