AnalyticDB:View monitoring information of AnalyticDB for MySQL

Prerequisites

Before you begin, ensure that you have:

• An AnalyticDB for MySQL cluster

• Access to the AnalyticDB for MySQL console

View monitoring metrics

The monitoring page supports a maximum time range of two days within the last month. For example, you can select any two-day window from the last month.

1. Log on to the AnalyticDB for MySQL console. In the upper-left corner, select a region. In the left-side navigation pane, click Clusters, find the target cluster, and click its cluster ID.

2. Go to the Monitoring and Alerts page:

   • Enterprise Edition, Basic Edition, or Data Lakehouse Edition: In the left-side navigation pane, choose Cluster Management > Monitoring and Alerts.

   • Data Warehouse Edition: In the left-side navigation pane, click Monitoring and Alerts.

3. On the Monitoring tab, click Standard Views or Custom Views.

   • Standard Views displays common metrics by default. To see all available metrics, click More Metrics.

   • To create a custom view, click More Metrics on the Standard Views tab, select the metrics to display, clear the metrics to hide, and then click Save As to add your selection to the Custom Views tab.

For descriptions of all available metrics, see Metric overview.

FAQ

Why did average CPU utilization increase after I switched a C32 cluster from reserved mode to elastic mode?

In elastic mode, a single node runs with 8 cores instead of the higher core count in reserved mode. BUILD jobs occupy 3 of those cores by default, which raises the average CPU utilization. If this does not affect your workloads, no action is needed. Otherwise, upgrade your cluster or submit a ticket. For more information about BUILD jobs, see BUILD.

Why are the values of the Regular Index and Primary Key Index metrics large?

Large index metric values are typically caused by one of the following:

• Indexes or primary key indexes are defined on a large number of columns.

• Index column values are long, or the total data volume in an index column is large.

• Index columns contain many distinct values, which reduces the compression ratio. For example, an index column with four distinct values (A1, A2, A3, A4) compresses poorly.

• Primary key values are long, or the primary key is a composite primary key spanning multiple columns.

The Monitoring and Alerts page shows a large response time, but there are no slow SQL statements on the Diagnostics and Optimization page. Why?

The total duration shown on the Diagnostics and Optimization page includes queuing time, execution plan duration, and execution duration — but excludes the time spent caching the result set. When a large result set takes extra time to cache, the response time on the monitoring page is higher than what diagnostics reports. To find the slow SQL statements, check the SQL Audit page.

SQL audit and historical query tracing

To trace historical SQL query records, for example to troubleshoot an abnormal query, use the SQL Audit feature.

1. In the left-side navigation pane of the instance details page, click SQL Audit.

2. Select the XIHE SQL Audit or SPARK SQL Audit tab.

3. Set the query conditions (execution status, username, time range, client IP address, database, and operation type), and then click Query.

SQL query history older than 30 days cannot be recovered. ActionTrail records only management-plane API operations (such as creating or deleting a cluster), does not include SQL query records, and retains data for 90 days by default.

For more information, see SQL Audit.

What's next

• Cluster performance optimization

• For more information, see SQL Audit.

API reference

Use the following APIs to query monitoring data programmatically.