Querying Prometheus metrics
Amazon OpenSearch Service allows you to query your Prometheus data using PromQL (Prometheus Query Language) directly from the Observability interface. When you execute a PromQL query against the Prometheus data source, OpenSearch Service passes the query directly to your workspace API for execution.
Running a PromQL query
To run a query:
-
Open your OpenSearch UI application and observability workspace.
-
Navigate to Observability and select Discover Metrics.
-
In the data source drop-down, select your Prometheus data source.
-
Enter your PromQL query in the query bar.
For example, to find the average per-second CPU usage over a 5-minute window for a specific pod:
avg(rate(container_cpu_usage_seconds_total{pod="payment-service-pod"}[5m])) by (pod)
Note
Set your time picker to a narrow, relevant window (for example, the last 1 hour) to optimize API performance and prevent timeouts.
Visualizing metrics in dashboards
You can add PromQL-driven metric visualizations to your existing observability dashboards to correlate them with your log and trace data.
-
Navigate to Discover Metrics, select your Prometheus workspace from the data source drop-down, and run your PromQL query.
-
Use the visualization tab in Discover Metrics to create a visualization and define your visualization type.
-
Save the visualization to your dashboard.
Note
Metric visualizations can only be added from Discover Metrics. Visualizations found on the visualizations tab are only optimized for logs.
