You are here:
Investigate Notifications
Manage the notifications that you received from monitoring alerts that you created. The system sends a notification to your email inbox whenever agent data for a monitored metric breaches an alert threshold.
Explore Sessions Associated with a Notification
On the Notifications subtab, click a notification name to see its associated sessions. Each row shows relevant session details for this alert, including the session ID, Agent API name, first user message, subagents involved, and other information.
Explore Conversations in a Session
On the Associated Sessions page, click the session ID name to see the conversation flow in a given session. The left side shows the turn-by-turn session log. Click an agent response to see the interaction trace details on the right side.
Snooze Notifications
-
Trigger an alert.
Example: The "High Response Time" alert fires on the Service Agent SDM because the average response time exceeded the 5-second threshold.
-
Observe behavior during the snooze window.
During the snooze window (for example, at the 15-minute mark), performance data can still breach the alert condition but no alert fires. Notifications are suppressed until the snooze expires.
-
Evaluate the condition when the snooze period ends.
When the snooze period ends (for example, after 30 minutes), the alert evaluates the alert condition. If the performance data continues to breach the condition, the alert fires a notification immediately. Otherwise, the alert doesn’t fire a notification.
Debug Alerts in Scale Center
Use the Scale Center to help you debug alerts.
- For this release, you can investigate in Scale Center only agents that invoke Apex.
- Access to Scale Center is enabled by default for all SysAdmins. SysAdmins can optionally assign access to standard users. See Enable Scale Center for Standard Users).
-
On the Sessions page, find a notification and select Scale Center
from the drop-down.
-
Click Submit to start a new analysis.
-
Wait until the status changes to Available.
Refresh the page to check.
-
Click the Analysis Name link to open it.
The Apex Summary Analysis shows the comprehensive performance analysis of overall Apex executions for your org. Agentforce Apex Actions trigger many of these Apex executions. The first section has key takeaways and recommendations on how to optimize Apex execution time. It also shows a view of where the Apex execution time occurs across Apex code, database, and so on.
The next section provides a view of the slowest Apex classes, along with run-time telemetry, organized by Critical, Major, and Minor. You can also optimize some of these Apex classes by fixing Apex AntiPatterns identified by ApexGuru Insights within Scale Center.
The next section shows the top Apex errors that occurred during this spike, along with the count and description of the Apex errors.
Finally, the investigation also shows the top SOQLs triggered by Apex actions, which typically lead to slow overall Apex execution time. It's important to optimize these slow-performing SOQLs to optimize database performance and overall Apex execution times.
Tips for investigating in Scale Center:
- Escalation Rate (%): If the rate is high, look for overall Apex performance, including the slowest Apex classes and Apex errors.
- Agent Response Latency (ms): If latency is high, look for the slowest Apex classes and the most expensive SOQLs in the report. To help with prioritization, focus on the Critical classes first.
- Agent Interaction Error Rate (%): If the rate is high, look for the top Apex errors.
