After an event such as instantaneous jitter or abnormal connection count occurs in the database, you can initiate a conversation to identify the possible root cause.
When to use
The cluster experiences short-term jitter and you need to retrospectively analyze the cause.
The number of connections or active sessions surges, and you need to confirm the trigger source.
You have received an exception notification and need to quickly assess the impact scope and root cause before handling it.
Prerequisites
You have configured the OCP and OAS connections for the product to which the target cluster belongs.
Procedure
Type
@in the session to select the target cluster and tenant.Describe the exception phenomenon and the time it occurred, for example:
The connection count for this cluster suddenly spiked around 15:20 yesterday. Please help me analyze the cause.You can also click the shortcut Root Cause Analysis of Exceptions on the Create Conversation page.
Press Enter to send the request and wait for the Agent to complete the analysis.
The Agent will make a comprehensive judgment based on information such as the exception details, root cause analysis results, and session snapshots at the time of the exception.
View the root cause analysis conclusion.
The Agent outputs a list of root causes ranked by likelihood and provides verification suggestions for each.
Note
The root cause analysis conclusion is inferred from monitoring and sampled session data and serves as an auxiliary judgment. Before taking any remedial action, verify the root cause according to the Agent's verification suggestions.
What to do next
Expand Tool Output to check the event and session snapshot data referenced in the conclusion. For more information, see View the tool invocation process.
If the root cause indicates execution plan degradation, you can proceed with statistics diagnostics. For more information, see Statistics diagnostics.