Feature Introduction
Provides multi-dimensional metrics and detailed displays for StarRocks queries. The query list allows quick viewing of the query statement, query start time, query status, query duration, user, number of rows retrieved, total CPU time, total memory consumed, and other detailed metrics. You can also view related SQL, query plan, profile, and execution topology.
Directions
1. Log in to the EMR Console, and click the corresponding StarRocks Cluster ID/Name in the cluster list to enter the cluster details page. 2. On the cluster details page, click Cluster Services, then select Operations > Query Management in the top-right corner of the StarRocks component to view the related query list. Some column headers support filtering or sorting, and multi-dimensional composite filtering is available. You can also customize the query duration to filter for slow queries.
3. Click Operation Operation Column > Details in the far-right column of the query list to view the query statement, query plan, profile, execution DAG, and operator order.
Note:
StarRocks Query Management currently supports only StarRocks V1.4.0 and later. For StarRocks V1.4.0 clusters purchased after September 12, 2023, the query management feature will be enabled by default, and an emr_admin user will be created for system query information collection. For StarRocks V1.4.0 clusters created before September 12, 2023, Submit a Ticket to contact support for assessment and activation of StarRocks Query Management. Before a query is executed, enter set enable_profile=true in the command to print the profile of the current session’s SQL query. Additionally, the collection service will automatically gather the profile of the submitted SQL in that session.
The visualization feature for execution DAG and operator order in StarRocks Query Management is currently supported only in StarRocks V3.1.8 and later versions. This feature requires Profile collection to be enabled.
To ensure stable operation of cluster services, queries with an execution time of less than 200 ms will be downgraded and ignored. Parameters related to the query collection downgrade policy can be modified by Submitting a Ticket.