Query statistics
Note: Supported only in on-premises deployments.
You can view queries recently initiated in the AE system and check their performance. Root and system admins with permission for this page can use it. Query statistics consists of two parts: real-time monitoring and query logs.
Real-time monitoring
Select the monitoring scope, including the time range, project, and cluster used. You can then see information about the queries within the monitoring scope, including each query's initiator, type, display name, duration, and detailed SQL.
About the real-time query volume:
- Active (number of queries): The total number of queries submitted in each 5-second interval. Use it to identify peak periods of query submission.
- Queued (queue length): The number of queries submitted in each 5-second interval that are still in the common queue. Use it to assess query backlog.
Click a record to expand it and view its details. When queries are blocked, if all available information indicates that a query can be canceled, click Cancel Query in the Action column.
To be more responsive to abnormal queries, you can set up monitoring alerts that use Real-time query duration, Scheduled query duration, and Queued as metrics. When a metric exceeds its threshold for N consecutive sampling intervals, system admins with the relevant permissions are notified through a pop-up. If you don't want to receive multiple similar alerts within a short time, select Mute alerts for 1 hour on the alert card.
Query logs
AE collects query activity and reports it to your system so that you can analyze and review past queries. Query logs show data from up to the past 90 days. By combining filters and sorting columns, you can find queries that took a long time or consumed a lot of resources, and view their details. If a query frequently consumes a large amount of query resources, we recommend optimizing it based on how its output is used.
When more than 1000 results are returned for the query period, the most recent 1000 are displayed. To view all results, click Export and view them locally. Note that querying a large volume of query logs can also affect the performance of the current system, so we recommend troubleshooting during off-peak hours. To troubleshoot current system issues during peak hours, use Real-time monitoring.