This topic describes how to query the resource capacity, performance metrics, key parameters, compaction status, and background tasks of a cluster or tenant through dialogue, to assist in capacity planning and troubleshooting.
When to use
- Evaluate CPU/memory/disk usage to determine whether scaling up or unit specification adjustment is necessary.
- Identify the cause of elevated CPU, I/O, or memory usage over a period, and confirm the source of high load.
- Query the current values of cluster/tenant-level parameters to verify consistency with baselines.
- Troubleshoot compaction exceptions such as ongoing compactions, compaction timeouts, or backlog of minor compactions.
- Track the execution status of background tasks (such as major compactions, data completion, partition migration).
Prerequisites
- You have configured an OCP connection for the target cluster's product; to query performance metrics, you need to configure an OAS connection. For details, see Create a product connection.
Procedure
In the session, use
@to select the target cluster (or tenant).Enter your query based on your purpose, for example:
What are the current CPU, memory, and disk usage rates of this cluster? Check the CPU and I/O trends of tenant xxx for the last hour. View the values of the cluster parameters memory_limit and ob_max_thread_number. Is the compaction status of this cluster normal? Are any stuck? What are the major compaction tasks running in the background and what is their progress?Press Enter to send the request and wait for the Agent to return the results.
The Agent will call tools related to cluster/tenant resources, metrics, parameters, compaction, and background tasks.
View the returned results.
- Resource and metric results are rendered as line charts or tables for intuitive trend comparison; for display and download methods of charts, see Data visualization.
- Parameter queries list the parameter name, current value, whether it can be dynamically modified, and its value range.
- Compaction and background task status indicates whether there are exceptions and the duration.
Note
For parameter change operations, OAgent only generates suggestions for parameter check, change, validation, and rollback SQL statements; it does not directly execute changes without approval. For details, see Parameter check and change governance.
What to do next
If metric abnormalities require drilling down to the SQL level, refer to Slow SQL, TopSQL, and abnormal SQL analysis.
If you suspect that the abnormal water level is caused by a major compaction or background task, you can perform root cause analysis in conjunction with the exception. For more information, see Root cause analysis of exceptions.
To turn resource inspection into a periodic task, see Create a scheduled task.