Docs

ContextSearch

Agentic Search

View monitoring

ContextSearch integrates with CloudMonitor, allowing you to view various monitoring metrics for the Agentic Search scenario directly in the ContextSearch console or CloudMonitor console, and analyze the scenario's running state in real time. This topic describes how to view monitoring.

Prerequisites

An Agentic Search scenario has been created and initialized. For more information, see Create a scenario and Initialize an Agentic Search scenario.

Procedure

Log on to the ContextSearch console.

In the top navigation bar, select the target project and region.

In the left navigation bar, select My Scene, then click the target Agentic Search scenario.

Click the Performance Monitoring tab at the top of the page.

Under the Monitoring Overview tab, you can view the scene’s resource metrics and business metrics.

tip

By default, monitoring information of the last day is displayed. You can view the monitoring data of the last month. You can use the time window on the page to define a custom time range for viewing, accurate to the minute, with a maximum interval of 14 days. ContextSearch also offers quick options for 1 hour, 3 hours, 6 hours, 12 hours, 1 day, 3 days, and 7 days.

If you check chart linkage, when hovering over any monitoring chart to view the specific value of that item at a certain time, the exact values of other items under the current monitoring category at that time will also display synchronously in their corresponding charts.

Resource metrics

Business metrics

Metric name	MetricName	Unit	Subnamespace	Description
CPU usage	ai_node_cpu_usage	%	ai	CPU usage rate at the node where the scene is located.
Memory usage	ai_node_memory_usage	%	ai	Memory usage of the node where the scenario is located.
Disk usage	ai_node_disk_usage	%	ai	Disk usage rate of the node where the scene is located.
GPU usage	ai_node_gpu_usage	%	ai	GPU usage of the node where the scenario is located.
GRAM usage	ai_node_gram_usage	%	ai	GRAM usage rate at the scenario node.

Last updated: 2026.09.09 17:28:36

Metric name	MetricName	Unit	Subnamespace	Description
QPS	ai_node_qps	-	ai	Requests processed per second at the node where the scene is located.
Average response time.	ai_node_mean_reponse_ms	ms	ai	Average response time for requests handled by the node where the scene is located.
Request success rate	ai_node_success_request_percent	%	ai	Request success rate of the node where the scenario is located.
TPS	ai_node_tps	-	ai	The number of Tokens processed per second by the node where the scene is located.